Root Methods - A Handbook (PDFDrive) PDF

Root Methods: A Handbook
Springer-Verlag Berlin Heidelberg GmbH

A.L. Smit . A.G. Bengough . C. Engels
M. van Noordwijk . S. Pellerin . S.C. van de Geijn
Editors
Root Methods
A Handbook
With 108 Figures, 4 in Color, and 39 Tables
i Springer
Dr. Albert 1. Smit Dr. Meine van Noordwijk
Plant Research International ICRAF, JI CIFOR
(Wageningen UR) PO Box 161, Situ Gede
Postbus 16,6700 AA Wageningen, Sindang Barang Bogor 16680
The Netherlands 16001 Bogor,Indonesia
Dr. A. Glyn Bengough Dr. Sylvain Pellerin

Scottish Crop Research Institute (SCRI) INRA
Invergowie B.P. 81, 71 Avenue Edouard Bourleaux
Dundee, DD2 5DA, United Kingdom 33883 Villenave d'Ornon, France
Prof. Dr. Christof Engels Dr. Siebe C. van de Geijn

University of Bayreuth Plant Research International
Institute of Geosciences (Wageningen UR)
Department of Agroecology Postbus 16,6700 AA Wageningen,
95440 Bayreuth, Germany The Netherlands
Cover illustration: Maize root system (see Fig. 4.3a in Chap.4 by 1. Pages et al.)
This book is the result of a project financed by the European Union: Concerted Action
AIR3-CT 93-0994: The dynamics of rooting patterns in relation to nutrients and water in soils.
Development, standardisation and documentation of methodologies. The contents of the book
is the sole responsibility of the Editors and Authors and does not represent the views of the
Commission or its services.
ISBN 978-3-642-08602-1
Library of Congress Cataloging-in-Publication Data
Root methods: a handbook I A.L. Smit ... [et al.], (eds.). p. cm. Inc1udes bibliographical refer-
ences.
ISBN 978-3-642-08602-1 ISBN 978-3-662-04188-8 (eBook)
DOI 10.1007/978-3-662-04188-8
1. Roots (Botany) 1. Smit,A. L. (Albert L.), 1950-
QK644.R6562000 575.5'4-dc21
This work is subject to copyright. All rights are reserved, whether the whole or part of the mate-
rial is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recita-
tion, broadcasting, reproduction on microfilm or in any other way, and storage in data banks.
Duplication of this publication or parts thereof is permitted only under the provisions of the
German Copyright Law of September 9, 1965, in its current version, and premissions for use
must always be obtained from Springer-Verlag. Violations are liable for prosecution under the
German Copyright Law.
© Springer-Verlag Berlin Heidelberg 2000
Originally published by Springer-Verlag Berlin Heidelberg New York in 2000
Softcover reprint of the hardcover 1st edition 2000
The use of general descriptive names, registered names, trademarks, etc. in this publication does
not imply, even in the absence of a specific statement, that such names are exempt from the rel-
evant protective laws and regulations and therefore free for general use.
Production: PRO EDIT GmbH, Heidelberg, Germany
Cover design: Design & Production GmbH, Heidelberg, Germany
Typesetting: Best-set Typesetter Ud., Hong Kong
Printed on acid free paper SPIN 10688779 31/3130 5432 10
Preface
Why a New Handbook on Root Methods?
Uptake of water and nutrients is a key process in agricultural and natural

ecosystems. Plant roots play a vital role in the supply of resources for growth,
and yet we have a relatively poor understanding of how they function in the
natural soil environment. Dr. B6hm began the Preface of his standard text on
root research methods (B6hm 1979) with the comment:
"Root research under natural jield conditions is a step-child of science. The

reason for this is primarily methodological. The known methods are tedious,
time-consuming and the accuracy of their results is often not very great".
Although improvements have been made recently for some methods, working
on roots is stiH tedious and time-consuming. Compared to some other disci-
plines, a root researcher requires a considerable budget for his or her work. It
is for this reason that in root research careful planning, the choice of appro-
priate methods and also a clear idea of which root characteristic should be
observed in relation to the goal of the research question are fundamental. Con-
sequently, it is recommended to make use of "root experience" developed else-
where as much as possible.
A logical choice is then to produce a handbook in which root methods in
current use are made available to a broader public. Since B6hm's book several
new methods have been developed: e.g. the use of mini-rhizotrons (already
mentioned in his book) has increased considerably and several (time-saving)
new methods of observing roots using mini-rhizotrons have been developed.
However, many older methods have also been revised and are described in
this book.
Scientific Contents
At a time when researchers are often forced to reduce their budgets, root
research is under pressure because of its time-consuming and expensive char-
acter. This book enables the researcher to make use of the expertise and knowl-
edge available worldwide as much as possible, as it gives a broad overview of
an available methods.
VI Preface
Not only sophisticated new methods (image analysis, CT scans etc.) are
described, but also time-saving "quick and (hopefully not too) dirty tricks" to
quantify root growth and root distribution in the soil profile.
Each chapter of the book was written by authorities on the subject. Virtu-
allyall methods which can be used to expose "the hidden half" of the plant (in
reference to a well-known book on root growth and function, Waisel et al. 1996)
are described.
In the introductory chapter the importance of roots in ecology and crop
production is explained (why and what to measure). The following two chap-
ters deal with methods to measure root system properties (anatomy, root hairs,
mycorrhizae) and to measure the effect of environmental interactions (mechan-
ical stress, water, oxygen, temperature) on root growth and functioning, under
both laboratory and field conditions.
A number of chapters focus on the assessment of root distribution and root
dynamics under field conditions. Advice is given on sampling strategy and
statistics, before treating in detail auger sampling, in-growth cores, pinboard
methods, trench profile techniques and core break methods. AIso, the use of
transparent interfaces in a soil profile (mini-rhizotrons and root windows) is
explained in detail. One chapter is devoted to processing the data from root
observations to analyse (fine) root longevity, an important aspect in ecosystem
processes. Relatively new methods described in the book include modelling
(describing root growth and architecture as well as functioning), the use of root
image analysis (IA), computer-assisted tomography (CAT) and magnetic reso-
nance imaging (MRI).
Finally, techniques for the experimental determination of functioning of
roots are described in the second part of the book, e.g. a review on the use of
isotope techniques in root research and methods to assess nutrient acquisition,
water uptake and plant anchorage.
We hope that this book will be a guide for research workers in many areas
of the plant and soil sciences, agriculture, forestry and horticulture, to the
benefit of production systems and natural ecosystems.
Acknowledgement. The production of this handbook was possible as part of a

project sponsored by the European Union (Concerted Action AIR3-CT93-0994
- "RootAction"). The objectives of RootAction were to establish a representative
network of research centres where plant root studies are done in relation to
nutrient and water dynamics of the plant/crop/soil system. In addition, the
objective of RootAction was to standardise and update methods used currently
in root research. At an early stage of the Concerted Action it was considered
important to produce a book in which the work could be made available to the
scientific community worldwide.
Preface VII
More people than the mentioned authors contributed to this book. Several
participants of RootAction, who are specialists on a certain topic, shared their
expertise with the authors. They are gratefully acknowledged. Furthermore, to
guarantee scientific quality, the individual chapters were reviewed by at least
two independent reviewers. Special thanks are extended to these reviewers
(listed on the pages IX, X). They not only thoroughly reviewed the chapters but,
in many cases, also made valuable suggestions to the authors.
We also thank many colleagues and editors of journals for permission to
use photographs and figures. Many thanks also to Springer-Verlag (Germany)
for excellent and efficient support in the final stages of the production of
this book.
Finally, we hope that this handbook will not only be found in libraries of
institutes and universities, but will also have a permanent place on the desks of
many scientists working on roots. This would be an indication that the book is
used as it was intended: as a handbook!
On behalf of the Editors, A.L. Smit

Wageningen, April 2000
References
Bohm W (1979) Methods of studying root systems. Ecological Studies 33. Springer Berlin Hei-
delberg New York
Waisel Y, Amram E, Katkafi U (eds) (1996) Plant Roots. Marcel Dekker, New Yok
Reviewers
We would like to gratefully acknowledge the kind help of the following

referees.
The Editors
Dr. J. Alegre, ICRAF-PERU, Lima, Peru

Dr. R.R. Allmares, Southern Experiment Station, University of Minnesota,
Waseca, USA
Dr. P. W. Barlow, Department of Agricultural Sciences, Bristol, UK
Dr. G.M. Berntson, Harvard University, Cambridge, USA
Dr. W.L. Bland, University of Wisconisn, Department of Soil Sciences,
Madison, USA
Dr. T. Boutton, Department of Rangeland Ecology, Texas A & M University,
Texas, USA
Dr. 1. Bruckler, INRA, Unite de Science du Sol, Avignon Cedex 9, France
Dr. D.T. Clarkson, IACR-Long Ashton Research Station, Department
of Agricultural Sciences, University of Bristol, Bristol, UK
Dr. Mike Coutts, c/o Forestry Commission Northern Research Station,
Penicuik Midlothian, UK
Dr. w.J. Cram, Department of Biological and Nutritional Sciences,
University of Newcastle upon Tyne, Newcastle upon Tyne, UK
Dr. P.R. Darrah, University of Oxford, Department of Plant Sciences,
Oxford, UK
Dr. L.A. Dawson, Macaulay Land Use Research Institute, Aberdeen, UK
Dr. M.C. Drew, Texas A&M University, College Station, Texas, USA
Dr. W.R. Eason, Institute of Grassland and Environmental Research,
Aberystwyth, Dyfed, Wales, UK
Dr. A.H. Fitter, Department of Biology, University of York, York, UK
Dr. o. Gasparikova Institute for Botany, Slovak Academy of Science, Bratislava,
Slovak Republik
Dr. J. Graham, School of Agriculture, Food and Environment, Cranfield
University, Silsoe Bedfordshire, UK
Dr. T. Heaton, NERC Isotope Geosciences Lab, Nottingham, UK
Dr. T.C. Kaspar, USDAIARS National Soil Tilth Lab, Ames, USA
Dr. E.A. Kirkby, The University of Leeds, Department of Biology Leeds, UK
x Reviewers
Dr. M.B. Kirkham, Department of Agronomy, Kansas State University,

Manhattan, Kansas, USA
Dr. F. Lafolie, INRA, Station de Science du Sol, Montfavet, France
Dr. H. Majdi, The Swedish University of Agricultural Sciences, Department
of Ecology and Environmental Research, Uppsala, Sweden
Dr. H. Persson, The Swedish University of Agricultural Science, Department
of Ecology and Environmental Research, Section of Soil Ecology, Uppsala,
Sweden
Dr. H. Rogers, National Soil Dynamics Lab, USDA-ARS, Auburn, USA
Dr. M. Smith, Institute of Hydrology, Wallingford, UK
Dr. P. Stamp, Institut fUr Pflanzenwissenschaften ETH Ziirich, Switzerland
Dr. Judy Tisdall, School of Agriculture, La Trobe University, Victoria, Australia
Dr. E.W. Tollner, Biological and Agricultural Engineering Department, Athens,
USA
Dr. W.B. Voorhees, North Central Soil Conservation Research Laboratory,
Morris, Minnesota, USA
Contents
Chapter 1. Root Characteristics: Why and What to Measure ............. 1

D.Atkinson
Chapter 2. Anatomy and Histology of Roots and Root-Soil Boundary 33

E. de Neergaard, O.B. Lyshede, T.S. Gahoonia, D. Care, and J.E. Hooker
Chapter 3. Control and Measurement of the Physical Environment

in Root Growth Experiments ...................................... 75
W.R. Whalley, J. Lipiec, W. Stfţpniewski, and F. Tardieu
Chapter 4. Modelling Root System Growth and Architecture ........... 113

L. Pages, S. Asseng, S. Pellerin, and A. Diggle
Chapter 5. Sampling Strategies, Scaling and Statistics ................. 147

A.G. Bengough, A. Castrignano, L. Pages, and M. van Noordwijk
Chapter 6. Auger Sampling, Ingrowth Cores and Pinboard Methods ..... 175
M. do Rosario G. Oliveira, M. van Noordwijk, S.R. Gaze, G. Brouwer,
S. Bona, G. Mosca, and K. Hairiah
Chapter 7. Trench Profile Techniques and Core Break Methods ......... 211
M. van Noordwijk, G. Brouwer, F. Meijboom, M. do Rosario G. Oliveira,
and A.G. Bengough
Chapter 8. Root Observations and Measurements at (Transparent)

Interfaces with Soil ............................................. 235
A.L. Smit, E. George, and J. Groenwold
Chapter 9. The Measurement and Analysis of Fine Root Longevity . . . . .. 273

J.E. Hooker, R. Hendrick, and D. Atkinson
Chapter 10. Root Image Analysis and Interpretation .................. 305

W. Richner, M. Liedgens, H. Bfirgi, A. Soldati, and P. Stamp
Chapter 11. Computer-Assisted Tomography

and Magnetic Resonance Imaging ................................. 343
S. Asseng, L.A.G. Aylmore, J.S. MacFall, J.W. Hopmans, and P.J. Gregory
XII Contents
Chapter 12. Isotope Techniques ................................... 365

LJ. Bingham, A.D.M. Glass, H.J. Kronzucker, D. Robinson,
and C.M. Scrimgeour
Chapter 13. Assessing the Ability of Roots for Nutrient Acquisition . . . . .. 403
Ch. Engels, G. Neumann, T.S. Gahoonia, E. George, and M. Schenk
Chapter 14. Water Uptake ........................................ 461

J.E. Fernandez, B.E. Clothier, and M. van Noordwijk
Chapter 15. Modelling Water and Nutrient Uptake ................... 509

P. de Willigen, N.E. Nielsen, N. Claassen, and A.M. Castrignano
Chapter 16. Plant Anchorage 545

A.R. Ennos and S. Pellerin
Appendix: Suppliers ............................................. 567
Subject Index .................................................. 573

Principal Authors
The foHowing list contains the addresses of those authors who had the main responsibility
for the chapters. The addresses of aH contributors to a chapter can be found on the first
page of the contribution.
Chapter 1: Prof. Dr. D. Atkinson, The Scottish Agricultural College (SAC),

West Mains Road, Edinburgh, EG9 3JG, UK, tel. +44 131 5354004,
fax +441315354340, e-mail: d.atkinson@ed.sac.ac.uk
Chapter 2: Dr. Tara S. Gahoonia, The Royal Veterinary and Agricultural

University, Department of Agricultural Sciences, Plant Nutrition and Soil
Fertility Laboratory, Thorvaldsensvej 40, 1871 Frederiksberg C,
Copenhagen, Denmark, tel. +4535283497, fax +4535283460,
e-mail: TSG@KVL.DK
Chapter 3: Dr. W.R. Whalley, Silsoe Research Institute, Wrest Park, Silsoe,
Bedford, MK45 4HS, UK, tel. +44 1525860000, fax +44 1525860156,
e-mail: richard.whalley@bbsrc.ac.uk
Chapter 4: Dr. L. Pages, INRA-Centre d' Avignon, Ecophysiologie

et Horticulture Site Agroparc, Domaine St. Paul, 84914 Avignon Cedex 9,
France, tel. +330432722431, fax +330432722432, e-mail:
loic.pages@avignon.inra.fr
Chapter 5: Dr. A.G. Bengough, Scottish Crop Research Institute (SCRI),

Invergrowie Dundee, DD2 5DA, UK, tel. +44 1382 5627311ext 2527,
fax +44 1382 562426, e-mail: G.Bengough@scri.sari.ac.uk
Chapter 6: Prof. Dr. Maria do Rosârio G. Oliveira, Universidade de Evora,

Apartado 94,7002-554 Evora, Portugal, tel. +351 266760800,
fax +351266711163, e-mail: mrol@uevora.pt
Chapter 7: Dr. M. van Noordwijk, ICRAF, Il CIFOR, P.O. Box 161, Situ Gede,
Sindang Barang Bogor 16680,16001 Bogor, Indonesia, tel. +62251 625415,
fax +62251625416, e-mail: m.van-noordwijk@cgiar.org
Chapter 8: Dr. A.L. Smit, Plant Research International (Wageningen UR),

Postbus 16, 6700AA Wageningen, The Netherlands, tel. +31 317475877,
fax +31 317 423110, e-mail: A.L.Smit@plant.wag-ur.nl
XIV Principal Authors
Chapter 9: Dr. J.E. Hooker, School of Applied Sciences, University

of Glamorgan, Pontypridd, Mid-Glamorgan, CF37 1DL, UK,
tel. +44 1433482453, e-mail: jhooker@glam.ac.uk
Chapter 10: Dr. W. Richner, ETH Ziirich, Institute of Plant Sciences, ETH
Zentrum, LFW A4, 8092 Ziirich, Switzerland, tel. +41 1 6324237,
fax +41 1 632 1143, e-mail: walter.richner@ipw.agrl.ethz.ch
Chapter 11: Dr. S. Asseng, CSIRO Division of Plant Industry, Private Bag
PO, Wembley, WA 6014, Australia, tel. +61 893336615, fax +61 893878991,
e-mail: s.asseng@ccmar.csiro.au
Chapter 12: Prof. Dr. D. Robinson, Department of Plant and Soil Science,
University of Aberdeen, Aberdeen, AB 24 3UU, UK, tel. +44 1224 273 662,
fax +44 1224 272 703, e-mail: david.robinson@abdn.ac.uk
Chapter 13: Prof. Dr. Ch. Engels, University of Bayreuth, Institut fUr
Geowissenschaften, Abt. Agrar6kologie, 95440 Bayreuth, Germany, tel. +49
921/552292, fax +49921/552315, e-mail: christof.engels@uni-bayreuth.de
Chapter 14: Dr. J.E. Fermindez, Inst. de Recur. Natur. y Agrobiol. de Sevilla,
Avenida de Reina Mercedes, No. 10, Aptdo 1052,41080 Sevilla, Spain,
tel. +34 954 624711, fax +34 954 624002, e-mail: jefer@irnase.csic.es
Chapter 15: Dr. P. de Willigen, Alterra, Green World Research, Postbus 47,
6700 AA Wageningen, The Netherlands, tel. +31317474204,
fax +31 317419000, e-mail: p.dewilligen@alterra.wag-ur.nl
Chapter 16: Dr. S. Pellerin, INRA, B.P. 81, 71 Avenue Edouard Bourleaux,
33883 Villenave d'Ornon Cedex, France, tel. +33556843051,
fax +33 556843054, e-mail: pellerin@bordeaux.inra.fr
CHAPTER 1
Root Characteristics: Why and What to Measure

D.Atkinson
SAC, West Mains Road, Edinburgh, EH9 3JG, UK
CONTENTS
1.1 Introduction 2
1.2 Why Study Roots? 2
Ecologica! Significance 2
Resource Capture 2
Soil Microbes 2
Resource Allocation 3
Plant Interactions 3
Soil Structure 3
Anchorage 3
Root Products 3
Basic Biological Information 3
1.2.1 EcologicaI Significance 3
1.2.2 Resource Cap ture 6
1.2.3 SoiI Microbes 7
1.2.4 Resource Allocation 8
1.2.5 Plant Interactions 10
1.2.6 Soil Structure 10
1.2.7 Anchorage 12
1.2.8 Root Products 12
1.2.9 Basic Biological Information 13
1.3 What Should Be Measured? 13
1.3.1 Available Methods 21
IA The Interconversion of Values 22
1.4.1 Methodological Interconversions 22
1.4.2 Spatial and Temporal Interconversions 24
1.5 Functiona! Relationships Between Traits 24
1.6 Appropriate Scales for Measurement 27
1.7 Individual and Populat ion Measurements 27
1.8 Data for Modelling Studies 28
1.9 What Is the Functional Significance of Field Measurements? 28
1.10 Future Trends 29
References 30
A.L. Smit el al. (Eds.l, Rool Methods

2 D. Atkinson
1.1 Introduction
A volume on available methods for the measurement of roots must begin by
questioning why such a volume is needed and thus with questions such as "Why
measure roots?" "What root parameters need to be measured?" and "What is
the functional, physiological, or ecological significance of a particular morpho-
logical assessment?" To provide some perspective to this volume, these key ques-
tions are discussed here. Measurements of root properties are made and used
by those involved in quantifying and managing both agricultural and natural
ecosystems. It is important therefore that methods and definitions can be used
in relation to the full range of species found in a complete range of habitats.
1.2 Why Study Roots?

A study of roots in the field is probably justified only when there is reason to
believe that the amount of below-ground material is likely to be statistically
and functionally different to that which might be predicted by the allocation of
a fixed amount of photosynthate to a below-ground compartment using an allo-
metric model, or where there is a need to achieve basic understanding of the
system (Atkinson 1996). In many situations, however, measurements will be
needed because available data will be inadequate to allow carbon partitioning
or root mass to be estimated. Many studies which have included measurements
of roots, have either been indecisive, i.e. have not produced a clear functional
link between the root measurement and the problem under investigation (were
the root measurements really necessary?) or demonstrate that the root infor-
mation could as easily have been generated from a model. Prior to beginning
any study of roots, it is important to ask why it is being carried out. Despite this
proviso, the main reasons for studying roots are:
Ecological Significance. In many situations, but especially in relation to natural

vegetation, there is little basic relevant information e.g. amount of root, distri-
bution of root weight with depth. This information is needed to answer ques-
tions, such as "Why do particular plant species grow in the places they do?"
Resource Capture. Roots represent the principal means whereby plants extract
resources such as nutrients and water from soil. Current expenditure on irri-
gation systems and fertilisers aUest to the importance of nutrient and water to
crop production. An understanding of roots will help to eliminate wastage and
adverse environmental effects.
Soil Microbes. The root system represents the major pathway for the flow of
carbon to the soi! and to soil organisms, especially those in the rhizosphere. As
1 Root Characteristics: Why and What to Measure 3
rhizosphere organisms are responsible for many key processes, such as N immo-
bilisation, NH4 oxidation, denitrification, and root nodulation, the supply of
resources to the soil is potentially critical to the evaluation of soil carbon
budgets. In addition, there is an increasing body of information on the effects
of plant species on soil microbial composition. This has gained additional
emphasis as a consequence of the debate about the impact of raised atmos-
pheric levels of CO 2 and nitrous oxide.
Resource Allocation. Information on the relative allocation of resources be1ow-

ground, above-ground, and to different types of root and mycorrhizas tells us
about the coupling of the plant to its environment.
Plant Interactions. Roots represent one of the key means whereby plants of the
same and other species interact. These interactions are now being seen, both in
relation to temperate and tropical multi-crop systems, as means of improving
the efficiency of resource use.
Soil Structure. The root and its associated microflora have a major effect upon
soil structure and the stability of aggregates. The input of organic matter to
the soil which they represent will influence key soil properties such as cation
exchange capacity.
Anchorage. Roots are essential for plant stability and anchorage. While this is
particularly important for tree crops, it has significant economic implications
for many fie1d crops, e.g. cereals.
Root Products. Roots may be used as an energy source in tropical production

systems. They may also be a source of pharmaceutical compounds or of food
additives and flavourings.
Basic Biological Information. To obtain basic information on a part of the plant

which consumes a significant proportion of total resources and which is of
physiological and developmental interest in its own right.
These topics are reviewed briefly to provided a background to the later discus-
sion of methods.
1.2.1 Ecological Significance
The ecological significance of roots is poorly understood (Atkinson 1991).

Although observation of most habitats or groups of habitats indicate that they
4 D. Atkinson
have species which are both characteristic of and different between habitats, we
are still unaware of the cycle of interactions between soil characteristics and
the amount of root. It is clear that in most plants roots exist to absorb nutri-
ents and water and to anchor the plant to its substrate. "Roots are natural
selection's design solution to the problems of obtaining these resources from
a heterogeneous, porous, semi-compressible medium containing solid, liquid
and gaseous phases" (Robinson 1991). Against this basic requirement, plants
have evolved a series of strategies to solve the problem of maintaining an
upright stance; at least for their reproductive structures, and to cap ture suffi-
cient assimilates to reproduce and transfer their genes to the next generat ion.
This variety of strategies alIows for different genotypes to co-exist and for the
acquisition of resources from different soils. These adaptations occur at several
levels (Robinson 1991) and are illustrated in Fig. 1.1. Although activity at sub-
celIular, celIular, organ and whole plant levels must have a molecular basis, it
seems that such higher level functions as morphological variation probably
account for much of the variation in the response to soil factors. Atkinson
(l990a) reviewed variation in morphological and physiological traits and found
that while a morphological feature such as root length density could vary
between species and habitats by a factor of lO3, physiological attributes nor-
malIy varied only by around lOl. This emphasises the need for the types of
values which are measured in field studies of roots.
Developments in soil microbiology (see also Sect. 1.2.2), especialIy in the use
of molecular methods to identify organisms and their effects, indicate the asso-
ciation between specific microbes and individual species. This suggests that
roots may be critical to the establishment of microbial populations in soil which
then influence fac tors such as soil structure and nutrient transformations, which
go on to influence vegetation development. This and other ecological processes
suggest that in natural ecosystems a range of measurements, which are wider
than commonly made in crop situations, will be necessary. In a crop situation,
plants principalIy interact with other plants with whom they share the whole or
most of their genotype. AlI individuals are relatively similar and so the purpose
of studies tends to be the prediction of the use of soil water or the extraction of
nutrients, which will then inform either a fertiliser or irrigation programme. In
natural ecosystems, a range of genotypes will be present. A range of strategies
will be exhibited with the possibility of particular characteristics leading to
effective resource acquisition in a limiting situation. As roots are not effective in
radiation capture, or the production of reproductive structures, there will be
many situations where resource allocated to the root system, even when needed
to absorb water and nutrients, represents a reduction in total productivity. The
challenge of having sufficient plasticity to cope with variations in environmen-
taI resources, is illustrated in Fig. 1.2.
m
Fig. 1.1. Hierarchies of (al
o (bl
nutrient and water fiux in H"
roots and root systems.

a Transmembrane fiuxes of C'
cations (C-), anions (A -) and
uncharged solutes (5) and
their reiat ion to active A-
proton efftux (modified from
Smith and Smith 1986): H '
m membrane, o outside,
S
i inside. b Symplastic (solid
arrows) and apoplastic
(broken arrows) fiuxes within
C'
and between cells in a
transverse section of a root
cortex (C) and stele (5).
c Fluxes in the soil solution (dl
to and from a single root.
d Fluxes to and from root
systems and within plants
growing in a community.
(After Robinson 1991)
The balan ce, in terms of root system design, between risk and insurance
strategies is poorly understood, but some elements have been discussed by
Grime et al. (1991) for some of the species found in natural vegetation. Grime
(1974) suggested that plant evolutionary strateg ies could be divided into three
trait groups (1) competitors, (2) stress tolerators and (3) ruderals. Competitors
are associated with rapid growth rates and therefore depend upon the ability
to sustain high rates of resource capture. Morphological plasticity is important
here. These species have high rates of reinvestment of the captured resources
in new roots. This type of investment must result in access to significant edaphic
resources. Stress tolerators depend upon their capacity to both capture and to
retain scarce resources. This tends to mean that roots will be long lived with
plasticity expressed by physiological changes. Ruderal species exhibit the
premature development of reproductive structures. In their vegetative phase,
6 D.Atkinson
Resource plentiful Optimal Habitats with acute

habitats Functional resource limitations
Design
Low C allocation High C allocation

(R:S) (R:S)
Risk strategy Insured strategy
Ephemeral structures Perennial structures
Limited microflora Abundant microflora
Fig. 1.2. The range of possible variation in root system optimal function design
adaptive foraging for nutrients will be important. The complexity of the above
attributes indicates the need for a varied range of methods for the study of
natural ecosystems.
1.2.2 Resource Capture
In cultivated situations, but rarely in more natural situations, nutrient supply

and to a lesser extent water supply and crop health are, within limits, guaran-
teed. Breeding programmes for crops have therefore aimed at maximising par-
titioning into grain and, as such, have gone for a higher risk scenario (Fig. 1.2).
In agricultural species, the ability to absorb nutrients from very fertile soils can
be substantial. Spring barley is able to absorb 80% of its total nitrogen content
of 170kgha- 1 in a single month, i.e. at a rate of around 4.5kgha- 1 day-l. This
uptake will occur via the roots; although, in aH situations, uptake will depend
upon both the supply of available nutrients in the rooting medium and upon
the root system. The importance of the root system will, as a consequence, vary.
Where nutrients are abundant and soil conditions are such as to make them
effectively available, then beyond the need for a minimum length of root, root
system characteristics are likely to be unimportant. As nutrient supply reduces
in total, or in the intensity of supply, or as it becomes diluted through a large
volume of soil, as it becomes more varied and complex in terms of its chem-
istry or where spatial distribution is particularly complex, then the root system
and its various properties become critically important. Concerns about the
losses of easily available nutrients and about the environmental impact of
sustainable agricultural systems are resulting in an increase in more extensive
systems of production and in systems, e.g. crop rotations and organic farming
systems, where nutrients are applied in a more measured way and more
complex sources are used (Atkinson 1990b).
Root Characteristics: Why and What to Measure 7
Our understanding of the impact of individual root and root system prop-
erties on nutrient capture is limited, so that it is easier for us to list root system
parameters than to specify the functional and ecological significance of these
parameters. The linkages and interdependence of properties means that a cor-
relation between a property and function may not indicate an unambiguous
functional role for that characteristic, but merely that it is a good surrogate for
a root or root system key characteristic. During the 1970s, considerable efforts
were made to relate nutrient supply to root length (Nye and Tinker 1978). The
ability of plants to absorb nutrients, in a range of glasshouse trials, could be
substantially explained in terms of the intensity of nutrient supply from the soil
and the length of root present in the soil. This led to the assumption that in
relation to nutrient supply, contact between root length and the soil was the
key element in root system design. The correlation between root length and root
mass, itself a complex character, means that true interpretation is far from
simple. In addition, it is clear that evolutionary success has given importance
to other characteristics. A high specific root length (SRL, length per unit mass)
represents an effective use of resources to maximise soil contact. However, SRL
varies between species (Atkinson 1990a). Recorded values for SRL are as low
as 5 m g-l for apple and as high as 750 m g-l in Lolium perenne. In general, SRL
seems lower in dicotyledenous species than in monocotyledonous species. The
amount of root in the soil, root length density (RLD), also varies in the same
way as seems evident for SRL. These properties are also influenced by soil
microbes and will be affected by the ecological strategies discussed above. Unt il
the interaction of these factors is wholly understood, there will be a need for a
suite of measurements to characterise root system performance.
1.2.3 Soil Microbes
Soil microbial characteristics must relate to the morphological and physiolog-

ical properties of the root system. As indicated in Section 1.2.2, a number of
morphological characteristics, which seem counter-intuitive to the simple
exploitation of the soil by the development of a maximum root length are
known to be of significant value. Characteristics such as a large root diameter
and the development of an extensive cortex (Atkinson 1989) will promote inter-
changes and associations between roots and microbes. In the rhizosphere,
interactions occur with soil bacteria, fungal symbionts, such as arbuscular myc-
orrhizal fungi (AMF), fungal pathogens and the wide range of bacteria and
protozoa involved in key nutrient transformations such as N mineralisation
and NH 4 oxidation. The flow of assimilates to these organisms may be as high
as 50% (Coleman and Crossley 1996), while the range of activities carried out
by soil microbes is such that an understanding of carbon flow via roots is crit-
8 D.Atkinson
ical to the understanding of processes as diverse as the stabilisation of soil

aggregates to nitrogen mineralisation (Coleman and Crossley 1996). In addi-
tion to their role as a microbial substrate, roots have direct effects on nutrient
uptake (Atkinson 1990a).
Recent studies have highlighted the importance of feedback between
processes occurring at vegetation, individual plant, root and molecular levels.
A key question in relation to soH microbes is the extent to which the amount of
microbial material and the species present are a function of the plant species
present or of independent soil microbial factors of which the flow of carbon
from roots is but one. Assessments of the likely impact of an increasing con-
centration of COl in the atmosphere on plant growth and biological activity in
the soH, is a dominant theme in many current research programmes. A number
of studies e.g. Curtis et al. (1995) have indicated that increased COl can lead to
increased root growth and to altered soH microbial activity. The importance
of the soH component within the global carbon budget emphasises the need for
"root" inputs to be firmly based. Advances in molecular genetics are indicating
the extent of biodiversity in soH microbes (Coleman and Crossley 1996). The
implication of apparent "excess" species diversity in soH (functional redun-
dancy), clearly indicates the possibility of substantial plasticity in relation to
ecosystem functioning. Measurement of root features, such as length and diam-
eter, will be important to the resolution of this problem. These factors also relate
to the sensitivity of roots to soH-borne plant diseases (CampbellI989). Research
to date has shown there are interactions between microbes and the host which
control disease. AMF infection, which is dependent upon root properties, has
been shown to influence the level of infection due to pathogens such as Phy-
tophthora fragarae.
1.2.4 Resource Allocation
There is no clear consensus on the functional significance of individual

morphological and kinetic parameters of roots. Atkinson (1990a) reviewed the
potential impact of changes in root diameter, root length, root branching, root
distribution and a range of physiological properties. Re concluded that the level
of variation exhibited in morphological properties was greater than that found
in physiological properties.
A number of relative parameters indicate changes in the adaptation of the
plant to the environment in which it finds itself. Of these, the root/shoot ratio
is perhaps the most commonly measured. A high value of R: S indicates the
movement of substantial amounts of photosynthate below-ground which may
relate to the genotype of the species or to the extent to which growth is being
limited by a restricted supply of soH resources, e.g. nitrogen. Although the
amounts of root and shoot are functionalIy related and can only vary within
certain limits, a range of studies indicate that for alI species a certain minimum
proportion of root is needed to supply a given quantity of leaf or shoot
(Atkinson and FogelI997). HelIriegel (1883) formulated his fundamentallaw of
agriculture, a consequence of which is that every limitat ion to root growth leads
to reduced shoot growth. Brouwer (1962) proposed the concept of a functional
shoot-root equilibrium and showed that, in situations of good supply of water
and mineral nutrients, small root systems can alIow maximum shoot produc-
tion (the risk strategy of Fig. 1.2). Large root systems may result in an increase
in the efficiency of water and nutrient extraction, especialIy when nutrient
supply is patchy in either chemical, spatial or temporal terms (the insurance
strategy of Fig. 1.2). This type of variation is shown in Fig. 1.3 for young trees
of Betula pendula (Lavender 1992) supplied with low or high levels of nitrogen.
With low rates of N, shoot growth was limited but constant, while the root-shoot
ratio varied between 0.45 and 0.85. In contrast, with a high N supply, shoot
growth was substantial, around 5 times that with low N but varied from 3.5 to
6.5 x mean low N growth. The R: S ratio in contrast varied only between 0.3 and
0.5, i.e. to a much smaller extent than under low N conditions. Under conditions
of low N supply, available resources were used to increase the size of the root
0.9
o
0.8 '0 0
o
o
00
00 0
0.1 o o
o
0.6 o
V')
6::: o
0.5 • • •
o • ••
•
•• •
0.4
• • •
• •• •
0.3
• • •
0.2
0.0 0.5 1.0 1.5 2.0 2.5 10 3.5 4.0 4.5 5.0 S.S 6.0 6.S 7.0
Shoot Dry Weight (g)
Fig.1.3. The relationship between R: S and shoot dry weight (g) for four Betula pendula clones
at two levels of nitrogen addition. Clased circles represent high and apen circles low levels of
nitrogen addition. Each point represents one plant. (After Lavender 1992)
10 D.Atkinson
system, while under conditions of good N supply, shoot growth increased in

association with small increases in root. Comparable data for other species
would allow the analysis of comparative strategies.
The relationship between root growth and soil condition is complex
(Gowing et al. 1993). However, it is dear that the root system is non-
homogeneous in its response to drying soil. Responses appear to be under hor-
monal control. Roots seem to be able to monitor water availability in soil and
regulate the whole plant's response. Signals generated by roots near the soil
surface move to both deeper roots and to the shoot. Field measurement of root
parameters will not indicate the mechanisms of co-ordination, but describing
effects may allow mechanisms to be identified.
1.2.5 Plant Interactions
The existence of different communities of plants on different soils, some of

which have species in common, indicates that below-ground interactions influ-
ence both growth and survival. A number of studies have shown the importance
of root-root interactions, both in studies of populations of a single species
(Atkinson et al. 1976) and in mixtures of species (Atkinson 1983; CaldwellI987).
These studies have shown that when plants of the same or of different species
interact, a whole range of root properties are changed in a major way. Com-
petition between apple trees (Atkinson 1978, 1985) influenced radial spread,
distribution with depth, total root length (Fig. 1.4), root length density, the devel-
opment of woody roots, and root survival. Competition between apple trees and
grass swards also influenced many of these but changed root system branching,
the periodicity of new growth,AMF infection and root activity (Atkinson 1983).
To date, studies of inter-specific competition between roots has been limited by
our ability to identify the roots of the different species. Molecular methods seem
likely to aid this, so allowing a functional base for studies of interaction.
1.2.6 Soil Structure
Soil structure influences the growth of roots (Passioura 1991; Tardieu 1994;
Atkinson and Mackie-Dawson 1991). The whole "soil cultivation" industry
depends upon this fact. Although species respond to different extents, a relatively
mild mechanical stress reduced root elongation in wheat by 47% and in maize
by 68% (Goss et al. 1989). Conversely, roots and their associated organisms can
also influence the structure of soils. Miller and Jastrow (1992) have shown that
roots and mycorrhizal hyphae "as a result of a series of mechanisms" are
involved in the creation of water-stable soil aggregates. The formation of these
Root Characteristics: Why and What to Measure 11
Wt
8769
% %
~
31 [
48
32
12 16
3 4
Wt
Wt
10939 547 9
[)
51
33
t
13 84 9 22
Fig. 1.4. Golden Deiicious/M9. The effect of spacing on the form of the root system, the weight
of root and shoot and the % distribution of root weight with depth. (After Atkinson 1978)
aggregates is an important prelude to soil stabilisation and the creation of a

nutrient reserve. As a consequence, roots and mycorrhizae are important in
processes such as the link between the restoration of vegetation and the reestab-
lishment of normal soil processes which are critical to the formation of soil
structure and the redevelopment of nutrient cyeles. Miller and Jastrow (1992)
found that the length of fibrous roots in a prairie restoration chronosequence
gave a 0.69 correlation with water-stable aggregates greater than 2 mm, whilst the
length of extra-matricula fungi hyphae gave a correlation of 0.88 with water-
stable aggregates. Through the process of growth, development and death, roots
and their associated micro-organisms create a network ofbiopores (Smucker et
al. 1993). These pores, when vertically oriented, as a significant proportion are,
allow the preferential ftow of nitrate and some pesticides to ground water. This
links to the growing concern in many countries around potable water quality.
Biopores can also act as preferential pathways for the roots of following crops.
12 D. Atkinson
1.2.7 Anchorage
Anchorage is a critic al feature for alI trees and many other woody perennials.
Listing the features of a root system which provide good anchorage (Coutts
1983) remains far from definitive. McKay and Coutts (1989) identified a series
of components of which the weight of roots to be pulled from the soil, the resis-
tance of roots pulled from the soil (related to root system diameter), and the
soil area exploited by the root system were important. Root system properties
inftuencing the anchorage of herbaceous plants have been discussed by Ennos
et al. (1993). The bonding between root and soil is important in resisting verti-
cal forces such as those occurring during grazing, while the production of a
rigid vertical structure extending to a distance below the soil surface and with
horizontal members at this depth also seemed significant in this situation.
These architectural properties are distinct from those reviewed in Section 1.2.1.
Many of the properties needed for anchorage will differ from those which relate
to nutrient acquisition. A number of key properties need to be measured at a
substantial geographic scale.
1.2.8 Root Products
An early publicat ion (Evelyn 1662), indicates a means of recovering roots for
use as an energy source, firewood (Fig. 1.5). Currently, there is interest in roots
Fig. 1.5. A Discourse.

Recovering roots for use
as firewood. (After Evelyn
1662)
as a source of secondary metabolites which can be extracted and used as phar-

maceuticals, as food additives and as oils for aromatherapy. Storage roots are
also important for the production of raw materials such as sugar and as veg-
etable crops. As food and other compounds differ between different roots, at
different times during the season, development of these uses requires knowl-
edge of phenology and demography in addition to data on root mass.
1.2.9 Basic Biological Information
Most field studies of roots and root systems are carried out for applied or strate-
gic reasons. Information from field studies, e.g. total root length, root length
density and specific root length are needed to validate the output from basic
models and for use in detailed studies of individual root properties. The para-
meters needed for process orientated models of crop growth have recently
been reviewed by Van Noordwijk and Van de Geijn (1996). They identified a need
for parameters related to root-soil geometry, soils and plants. The plant
parameters included both state parameters and physiologically linked charac-
teristics such as hydraulic conductance and the minimum effective nutrient con-
centration at the root surface for uptake to occur. The investment in time to make
many of these measurements asks difficult questions about minimum data sets
and the ability to estimate particular root parameters from other properties. In
addition to their role in resource acquisition, roots have a role in producing hor-
mones such as ABA and in storing carbohydrates and proteins (Hennerty
et al. 1980).
1.3 What Should Be Measured?
In the absence of absolute and unique links between particular root properties
and functional objectives, there will be a need to measure a number of para-
meters to obtain a full picture. In addition, establishing functional and allo-
metric links between different parameters will be important for the use of data
in a wide range of circumstances. In many situations, available time and equip-
ment will limit the measurements made and so the ability to estimate other
parameters will be important. In addition, in other situations, only rough esti-
mates of root presence or function may be needed. There is a need, therefore,
to reconcile root measurement with function, to assess the extent of our knowl-
edge on the interrelationships of root parameters and to ask what information
exists on what can be obtained from rapid root system assessments. It is impor-
tant to begin by defining key root system parameters which are then related
to functional significance and to methods of determination (Table 1.1). These
.....
Table 1.1. The definition of key root system parameters and their usual units ""
Parameter Definition Function/significance Usual means of measurement
(alternatives; unit)
Root length Length of ali root members present. Total root system size. Potential for Monolith methods - soi! coring, needle
(m) absorption of nutrients or water from soi!. boards, etc.
Indicator of bas eline soi! microbial, Rhizotron and mini-rhizotron methods can
especially AMF, activity in soil and of be used but rely on assumptions (Atkinson
microbial functioning, e.g. organic 1985). Profile wall methods (Atkinson and
phosphorus catabolism. Mackie-Dawson 1991) can be used when
Basis for most functional calculations. Large limited precis ion is required.
temporal variation.
Root weight Oven-dry weight of the total root system Total root system size. Monolith methods, especially soil coring,
(mass, standing crop; g) (and attached micro-organisms). Corrections Standing crop of below-ground materia!. but may involve partial or total excavations
may need to be made for attached soi! Amount of assimi!ate moved below-ground. for woody perennials.
minerals.
Root volume The space occupied by the root system.

(mI) (Frequently calculated from mean root
diameter and length or assumed to be equal
to the fresh weight of the root system).
Root number The total number of individual roots Hormone production (e.g. cytokinin) Counts of samples obtained by soi! coring,
(~1 °,2°,3°,4°,5°). potential, meristematic activity, presence in profile walls, rhizotron or mini-rhizotron
volume of soi!. methods. Root tips equal to 50% of alilinks
regardless of system topology. Specifying
whether total number or number of tips is
ţi
important.
Root: shoot ratio The ratio of the dry weight of root to the dry Relative allocation strategy. Used against a Estimated from root and shoot weight. ~
S'
(dimensionless) weight of leaves + stem (branches). bench-mark of plant nutrient status. '"o
:::
Root length density The length of root present in a unit volume Probable limitations to soi! nutrient and Calculated from root length and soi! volume
(intensity; cmcm-3 of soi! to or at a specified depth. This is water exploitation. to which measure applies. O
ormlmt') commonly identified as Lv. O
..=
n
Cumulative root density The length of root present under a unit area ::r
III
(mmcm-2 soi!) of soi! surface to a specified depth. This is ii!
commonly identified as LA' ID
.."
:::!.
Specific root length The length of root associated with a unit Within-root system allocation strategy, Calculated from root length and root ..'"
;\'
(SRL; mg-') of dry weight. (This is not always a good relative importance of soi! exploitation. weight. !Il
indicator of diameter for which it is :e::r
sometimes used.) '<
III
~
Q.
Root fresh weight: The ratio of the wet and dry weights of the
dryweight root system. (FW/DW-l has been used to ::r
III
(dimensionless) give an estimate of volume.) ..:e
O
Root density The length of root associated with a unit 3:
ID
(mmt') volume of root tissue. (Together with SRL III
can be used to estimate root diameter.) c
..'"
ID
Root radius The radius (diameter) of an average Potential for mycorrhizal development, Measured directly or calculated from root
(diameter; mm) individual root; usually assumed to be a regulation of water stress, potential for length and volume.
plain cylinder. (Frequently calculated from apoplast-symplast exchange, growth
length and volume or FW.) potential, influences and responds to soi!
physical condition.
Root of primary structure A root with an intact cortex and where

(white root) the externallayer is usually the epidermis.
(This is the normal state for the roots
of annual species.)
Root of secondary A root with at least some woody tissues and For perennial species the amount of Measured in terms of weight or length.
structure where the externallayer is usually a bark investment in root system infrastructure.
(woody, brown root) ceH. (Many of the roots of perennial species
......
are in this category.) U1
....
Table 1.1. (cont.) '"
Root system branching Relative intensity of soi! exploitation, mean Measured lengths or number of roots
pattern root longevity, soi! exploitation strategy. of different orders.
Primary root The main root axis or a root which emerges

from stern hypocotyl tissue and so includes
both seedling (tap, radical seminal) roots
and adventitious (nodal basal) roots (! 0: first
order).
Secondary root A root emerging from a primary root (2°).

These are sometimes described as a first
order lateral.
Tertiary root A root (second order lateral) emerging from

a secondary root (3°) and bearing
quaternary roots (4°) which may in turn
bear quinary roots (5°). (This is the usual
limit to branching.)
Root longevity The length of time for which an individual Potential for rapid adjustment in root Measured on cohorts, identified using
(d) root either is physically present or alive in length. Plasticity in relation to reduction rhizotrons or minirhizotrons.
the soil. Under soi! conditions where in length. Flux of carbon to rhizosphere,
decomposition is slow, defining whether nutrient release. Indicator of soi!
longevity relates to present or alive is exploitation strategy.
critical. This is most commonly given as a ţJ
mean value or as a range for a population.
This may not necessari!y relate to the length ~
5·
of time for which the root is functional. '"
o
::s
if
Root cohort A group or population of roots which are ~
initiated (or become obvious) during a n
:r
III
short, defined period of time.
ii!
Root production An estimate of the total amount of root Overall potential for soi! exploitation, over Sequential soi! core estimates or a it..
(kgha-1yr-1) produced during a given time period for a specific time periods. The ability to increase combination of baseline standing crop and i-
given land area (commonly a growing season root system length. The ability to exploit rhizotron or mini-rhizotron measurements. ~.
or year) and estimated either from the sum the soi!. :e
of new root production on a weekly basis or ~
from the difference between live root and III
:::1
D.
dead root (necromass) amounts at the
beginning and end of the assessment period.
:e
:r
!!:.
Root growth rate The rate of increase in the length of an (ţ
(mm day-l or g week-1) individual root or the short -term gain in 3:
III
mass. III
Mycorrhizal infection The degree of infection of a root with, Carbon allocation strategy, soi! health, real
'"c
Estimation from stained root samples. ii!
(AMF;%) usually, arbuscular mycorrhizal fungi. effective surface for nutrient uptake.
Infection is usually scored on a presence
rather than intensity basis with scores based
upon the proportion of 2 mm-long root
sections containing evidence (hyphae,
arbusc1es, vesic1es) of fungal presence.
Vertical distribution
-Rooting depth The depth of the deepest root found on Physical stability or anchorage, depth of soi! Soi! core samples. Profile walls. Diagnostic
(m) a plant root system. exploited, potential for resource use and the pits for coarse estimates.
estimation of avai!able nutrient and water
resources.
.....
'-J
,....
00
Table 1.1. (cont.)

-Effective rooting depth A semi quantitative estimate of the use of

(m) the depth of soi! avai!able to the plant. The
depth within which either 90% of roots are
found or for which evidence is avai!able for
the presence of root activity, e.g by the
depletion of soi! water. Where significant
capillary rise occurs this can be greater than
rooting depth. The value of this parameter
can be considered conceptually flawed
because it will vary during a single season
and between seasons depending upon the
demands placed upon soi! resources, e.g
under very dry conditions survival may
make the effective functioning of a small
number of roots at depth critical and so
modify the interpretation of the definition.
Horizontal Distribution
-Root system diameter A semi -quantitative estimate of the area of Stability of anchorage. Profile walls, partial or total excavation.
(m) soi! under which the root system is found or Potential for interaction with other species.
which, at a particular point in time, is Estimation of available soi! nutrients and
ţi
influenced by the activities of the root water.
system and its associated organisms. ~
5'
ti>
o
::s
=
O
O
Root activity Temporal changes in the amount of a soil Root system functioning, uptake per unit Depletion of soi! water, or uptake of isotopes
...n
(varied) resource or evidence of a material being amount of root, km, kmio> exudation of from soi!. ::s-
ili
moved into the above-ground part of the physiologically active molecules, nutrient ii!
1"1
plant. This may also include evidence of storage potential. ID
......
hormone production or of storage and ~.
mobilisation. ;::i'
l'!
Root uptake ability The Michaelis-Menten constant for uptake of
(kmor k min ; ţ1m) a specific nutrient at 50% of the maximum ~
III
rate of nutrient uptake ("optimal" rate) bya :::1
CI.
unit length of root. This is calculated as the ::e
minimum soi! solution nutrient ili
concentration (Cmin) from which a nutrient
...::s-
(ţ
can be absorbed. s:
ID
III
Viable (functional) root The length of root present which is able to 11\
C
length carry out resource capture ar physiological ;;
(m) activity under field conditions. This is
difficult ta measure and sa is frequently
assumed ta be equal ta total (unsuberized)
root length.
Phenology (periodicity) The period(s) during the year when root Functional plasticity, ability to complement Rhizotron and mini-rhizotron methods.
production ar growth is occurring. activities of other species, potential ta use
temporarily avai!able resources.
Rooting volume A semi-quantitive estimate of the total

(m') volume of soi! exploited by the plant root
system.
.....
\O
20 D.Atkinson
indicate that the various direct measures and estimated root parameters will
have a different functional significance. The measurement of a number of the
critical attributes of roots e.g. hormone production, require those measure-
ments to be made under the controlled and commonly homogeneous condi-
tions normally found only under laboratory conditions. These laboratory
measurements need to be linked with field measurements to allow the full sig-
nificance of the laboratory measurement to be interpreted (Atkinson and Fogel
1997). Physiological and other more detailed features of roots, often measured
in the laboratory, but interpreted in field context or used in a field based model
(Van Noordwijk and Van de Geijn 1996) are detailed in Table 1.2. The key ques-
tions in root methodology will thus be joint1y advanced by novel means of mea-
suring root and root system parameters and by increased understanding of the
relationship between key morphological and physiological features of the root
system and root functioning. This will link with issues relating to scale and
effects based at the individual root and population levels. Decisions on what to
measure, thus depend upon:
Table 1.2. Physiological parameters associated with root systems and their significance and
estimation
Parameter Function/significance Usual means

of measurement
Root growth Development of the root system Direct observation of

change in root length
Root viability Ability of root to function; Biochemical tests of
usually in resource acquisition functioning
Root hair density Exploration of soil adjacent to Observation and
root estimation of number
on sample length
Root respiration Contribution to soil respiration Direct on roots in cuvette
and plant carbon budget or microcosm or by
calculation from soil
volume
Root exudation Release of secondary Analytical measurement
metabolites from roots on root -associated soil
or from media in
solution culture
Root hydraulic Ability of water to move within Pressure chamber or heat
conductance a root (thermal) pulse
Minimum useable Absorbing power of root, Experiment with range
nutrient concentration supplying power of soil of dilutions in solution
or solution culture
1. The availability of methods,

2. Our ability to interconnect values from different methods,
3. Understanding of the physiological significance of morphological change
and the functional relationship between these traits,
4. The scale at which measurements need to be made and are required to be
applied,
5. The importance of factors working at the population rather than individual
root level, and
6. The minimum data set of values needed for the production of models involv-
ing root parameters.
These factors are discussed in the folIowing sections.
1.3.1 Available Methods
Harper et al. (1991), reviewing the problems of analysing root behaviour, com-
mented "more energy may have been spent on developing technologies than on
studying roots. Each method so far developed has serious shortcomings ... No
one technique, of the many so far developed, informs us about alI aspects of
root growth and structure". They divided available methods into two groups.
The first containing whole plant excavations, profile walI methods, pinboards,
soil cores, the use of isotope applications to soil, resin embedding techniques
and NMR. This group was able to provide information on root distribution and
on the mass of the standing crop. The second group included rhizotrons, mini-
rhizotrons and in-growth bags. This group alIowed the assessment of changes
with time. Although methods do exist for the assessment of mass and tem-
poral changes, there are fewer methods available to measure either three-
dimensional architecture or demographic parameters. Since 1991, however,
considerable progress has been made in both of these areas. The use of NMR
and tomography, together with the combination of information from soil sec-
tions (Krebs 1995) has developed our ability to see soil and the roots within it
on a 3-D basis. Similarly, methods for viewing and establishing cohorts of roots
have alIowed progress in establishing values for root longevity and the demog-
raphy of root populations. These are discussed further in Section 1.6.1.
Other analyses of our needs in relation to methods have been carried out.
Bohm (1979) described a whole range of methods which could be used to
measure roots in the field. Atkinson and Mackie-Dawson (1991) reviewed
methods of root measurement which could be used to relate roots to the phys-
ical properties of soils. They concluded "There is no single method of root mea-
surement applicable for alI situations ... The principal factors infl.uencing the
choice of methods are likely to be the availability of equipment and facilities,
22 D. Atkinson
the crop and or soil to be investigated and the type of root system effect of inter-
est". When measurements are made on the above-ground parts of plants, a
range of features are recorded. Similarly, with the root system, a range of mea-
surements are required so as to be able to characterise factors such as the rate
and type of root growth, the standing crop and root activity. Mackie-Dawson
and Atkinson (1991), who also reviewed a range of methods, concluded "The
main criterion influencing the selection of a method is probably whether infor-
mation is needed on changes with time or whether spatial data on distribution
at one moment in time will be adequate. Where time is involved, an observa-
tional method will be indicated. Where spatial distribution is the key issue a
direct sampling method will be the best". This dichotomy in objectives and
methods still seems to exist and these two groups appear to be the most diffi-
cult to reconcile and to interconvert (Atkinson and FogelI997). Wherever roots
growing in soil are sampled, in addition to the measurement of the quantity of
root present, there is also a need to determine the soil volume with which this
quantity of roots is associated. In extractive methods, this is simply measuring
the volume of the core containing the roots, although alIowances are needed for
compression dur ing sampling if a complete depth section is taken and later sub-
divided. For samples derived from observation methods, estimation of the
depth of field represents a significant element in the scaling process (Atkinson
and Fogel 1997).
1.4 The Interconversion of Values
Data on root systems is expensive to acquire and so there is a need to make the
best use of available information including deriving new, or related values from
exist ing parameters. The major interconversions of data which seem to occur
are: (1) the estimation of information of one type from measurements of a dif-
ferent type, e.g. root length from root weight, (2) the estimation of information
of the same type but for a different spatial or temporal situation.
1.4.1 Methodologicallnterconversions
Table 1.3 shows some of the more commonly attempted interconversions. Of alI
of those listed, the interconversion of weight and length is the most common.
This interconversion is used to calculate length values in studies of nutrient
inflow from fresh weight which can be estimated non destructively from
volume, and as a means of relating root length values, assessed by observation
methods such as the mini-rhizotron technique, to root weight values obtained
by soil cor ing in method comparison studies and in attempts to derive total
Table 1.3. Interconversion of data assessed from a range of root measurements
Measurements (units) Par am eter to which converted Method
Dry weight (g) Length (m) Regression

Fresh weight (g) Volume (mi) Calculat ion
Length (m), volume (mi) Diameter (mm) Calculation
Length (m), diameter (mm) Volume (mi), surface area (cm') Calculation
Length (m) Number (-) Regression
Length (m), soil volume (cm') Root length density (cm-') Calculation
Length at depth a. (m) Length at depth b. (m) Regression
Length at time a. (m) Length at time b. (m) Regression
Profile wall counts (-) Length (m) Calculation
Length (m), weight (g) Specific root length (m g-l) Calculation
..c
~
c:
Ql
....J
woody
l5o
a: ./'
.,/'
RootWeight
Fig.1.6. Variation in root length with increase in root weight for fine and woody roots of Betula
pendula. Data after Lavender (1992). The length scale for woody roots has been increased by a
factor of 30
carbon budgets. For Betula pendula, Lavender (1992), found that for fine roots
the relationship between length and weight was good but it was poorer for
woody roots (Fig. 1.6). Similarly, the relationship between root length assessed
using the mini-rhizotron method and root length assessed using other
methods, although significant, was limited (Mackie-Dawson and Atkinson
24 D.Atkinson
1991). Methodological interconversions can also be used for indicators of root

system activity. For example, the combination of measurements of water
content and water potential allows estimates to be made of water flow. The
combination of data derived from two methods is often essential. Depletion of
soil water, often used as an indicator of root system activity, requires the use
of a range of instruments e.g. tensiometers plus soil psychrometers so as to
cover the entire range of soil water potentials which are of interest. Similarly,
estimation of root length present at different periods during the year can be
made using a combination of methods such as the profile wal1 plus rhizotron
(Atkinson 1986).
1.4.2 Spatial and Temporal Interconversions
Although predictions of the size of the root system, in terms of weight, can be
made from small numbers of infrequent measurements, many studies require
detailed information about the distribution of root weight or length with depth
(see Table 1.1 for rationale). Difficulties in understanding the vertical growth of
root systems has meant that compromises between the level of sampling at any
one point in time, and the frequency at which field sites are sampled, have to
be made. This has often resulted in inadequate amounts of data being acquired
for each of a number of time periods.
Plant breeding programmes rarely involve any assessments of root system
development; most measurements are made on relatively few occasions. The
issues discussed above therefore influence apparent minimum requirements for
breeding programmes. For both physiological and crop breeding purposes,
there is a dear need to establish whether the depth distribution of a root system
can be evaluated from a basis of selected measurements and whether measure-
ments made at one point in time, on a root system, can be used to estimate the
size and distribution of the root system at a later point in time. A case study,
carried out using 24 different varieties of spring barley, illustrates the possibil-
ities of predicting spatial and temporal data from a given data set. Considera-
tion of these points is important in the context of the use of data from any root
measurement exercise and in respect of minimum data sets required from
particular field experiments.
1.5 Functional Relationships Between Traits
Although root growth at various depths are on some occasions correlated, and
although some future properties can be predicted, by and large the case study
shows that where a detailed understanding of root growth and development is
BOX 1.1. Case Study

A study using barley assessed the extent to which the length of root
present at a greater depth eould be estimated from the length present at
more superficial depths (Table 1.4) and the relationship between the
length of root present on different oeeasions (Table 1.5; Atkinson 1989).
The amount of root present at any one depth was always correlated with
that at the next (greater) depth interval, i.e. root length at 15 em depth
showed a 0.46 correlation with that at 25 em depth. For zones where root
growth was most active at the time of the measurements, eorrelations
were both greater and signifieant for a larger number of greater depths.
This suggests that potential exists to assess root growth at greater depths
Table 1.4. Correlalions between root growth at a number of depths for 33-day-old spring
barley grown in transparent growth units, with root growth at other depths. NS = non-
signifieant
Root growth at depth Root growth al depth (em)

(em)
15 25 35 45 55 65 75
5 0.32 NS 0.36 NS NS NS NS
15 0.46 0.41 0.36 NS NS NS
25 0.56 0.33 NS NS NS
35 0.74 0.61 0.54 0.49
45 0.76 0.57 NS
55 0.72 0.42
65 0.76
Table 1.5. Correlation between root growth at a number of

depths for 33-day-old spring barley grown in transparent growth
units with root growlh at Ihe same depth, but on earlier dates (d)
Depth Depth (em)

(em)
day 19 day26
15 25 15 35 45
35 0.55 0.73 0.36

45 0.48 0.65 NS 0.6 1
55 0.37 0.60 NS 0.58 0.75
65 NS 0.63 NS 0.59 0.74
75 NS 0.63 NS NS 0.54
26 D. Atkinson
through its relationship with that at more superficiallevels (Table IA).

Similarly, growth on one date was correlated with that on later dates. The
closer the dates and the nearer to the principal site of new growth, the
better was the relationship. Growth at 15 cm depth on day 19 of the study
was significant1y related to that at 35 and 45 cm on day 33. This would
seem to suggest the possibility of estimating growth on a later occasion
at a greater depth from earlier measurements (Table 1.5).
required, and where firm data is required in relation to quantitative studies,

then there is likely to be no alternative to adequately replicated measurements
of root length and distribution made on a significant number of occasions.
Where only general relationships are required, then predictive modelling
is likely to be helpful and may avoid the need for large-scale field-based
exercises.
The correlations discussed in Section 1.4.1 suggest the existence of func-
tional links between properties. The correlations between root development
at different depths in the soil and with time (Sect. 1.4.3) are evidence of clear
physiological control during development. The existence of limits to R: S imply
functional links between root and shoot development. The data in Fig. 1.6
suggest that in fine roots, an in crease in assimilate allocation to the root system
increases root length proportionalIy. With woody roots (Fig. 1.6) initialIy an
in crease in allocated resource (dry weight) increased the length of this type of
root. However, beyond a certain root system size, 0.5 g in this study, increased
assimilate, as evidenced by increased weight, did not result in an increased
length, suggesting that assimilate is principalIy used for an increase in root
diameter. McCulIy (1987) reviewed the development of maize roots in the field.
She identified an informat ion gap between laboratory and field studies,
especialIy in relation to the state and condition of root tips which were usualIy
absent from framework roots in the field. Under field conditions the relation-
ships may be very different from those found in the laboratory. Gregory (1983,
1986) found that factors such as temperature could influence both the rate and
form of development. Even for a given genotype, responses could be inconsis-
tent between years. Effects on root growth tended to be different from other
processes. Atkinson (1989, 1990a) assessed variation in a range of root charac-
teristics. Root length, its persistence during a season, its distribution and activ-
ity alI varied. The significance of both these parameters and the patterns of
variation within them varied both within and between seasons as limiting com-
ponents of soil resources var ied. Consequent1y, the optimal root system needed
to deliver these limiting resources changed: in addition, the ideal root systems
for coping with limits to P and water supply are not the same. Improvements to
our understanding of what root systems are realIy like in the field should help
to bridge the gap between current field experiments and so allow the extrapo-
lation of laboratory data (Atkinson and FogeI1997).
1.6 Appropriate Scales for Measurement
Even for measurements made in field studies, the appropriate scale for deter-
minations, and the limitations of scaling data up to field or landscape cause
difficulties (Atkinson and FogelI997). The development of the mini-rhizotron
method, and its use in a wide range of field studies, has drawn attention to
a number of scaling issues. These have been identified by Hendrick and
Pregitzer (1992b) and by Atkinson (1992). With the mini-rhizotron method,
measurement of root length, root morphology or root survival are made using
a window area of under 400 mm 2• This information must be converted from the
2-D basis on which it is measured to a 3-D basis to allow the estimation oflength
per unit volume. This estimate, based on a volume of 1-2ml must be trans-
formed to a volume representative of 104 m 3 for a 1 ha field. Although estimates
with appropriate error terms can be obtained, there remains a need to express
the functional variation and the difficulty of obtaining sufficiently precise
estimates to generate useful field-scale estimates. In addition to the size of the
variance term associated with field-scale estimates derived from small initial
samples, there is also the problem of estimating the length or mass of a large
root system e.g that of a tree from a limited number of small samples. In addi-
tion to spatial variation, there is likely to be pattern in the variat ion within the
root system both with depth and distance from the trunk. Both actual values
and variance are likely to vary along these gradients (Atkinson and Wilson
1980). Attempts to scale estimates of root system size from core samples must
work with these gradients.
1.7 Individual and Population Measurements
Hendrick and Pregitzer (1992a) introduced the concept of the cohort, a group
of roots, to form the basis of demographic assessments, for example of root
longevity. The cohort approach has made the volume of information captured
by the mini-rhizotron method, used to estimate values such as longevity, man-
ageable. The cohort assessment gives a population value for root longevity. The
population is composed of a series of individual roots. Variation in soils suggest
most individual roots will experience a different set of physical, chemical and
microbiological conditions to other roots and that the total response of the plant
will be the sum of the activities of individual roots which may be separately
controlled and which may respond in a different manner to a stress such as
28 D. Atkinson
drought, (Gowing et al. 1993). Hooker et al. (1995), in a study of poplar roots,
found that within a cohort of non-mycorrhizal roots 10% survived only 14 days
and a further 40% only 42 days. The distinguishing features of roots surviving
less than 14 days and greater than 42 days are not known but may be of func-
tional significance. Infection with AMF increased the proportions, given above,
to 20% and 50%. Again the distinguishing features of these roots is not known,
but it seems important to understand at both individual and population levels.
Different studies will require estimates to be based on either individuals or
populations depending upon specific objectives. It will be important, however,
and especially in studies of natural vegetation, to be sure whether values relate
to single individuals or to populations.
1.8 Data for Modelling Studie5
The difficuhy of measuring roots has meant that their growth and production
are not uncommonly estimated as a fixed fraction of net assimilation. The
development of models would benefit from:
1. Improved allometric models to aid estimates of partition, and
2. The specification of a minimum data set for different types of models.
In many carbon allocation models, root values and variations between species
for root values can have a significant impact. In the carbon storage model
developed by Cannell et al. (1992) litter production represented by woody roots
was around 50% of that due to branches. Production due to fine roots exceeded
that for leaves. Par am eter iz ing models with the right root characteristics is
important. A range of root, shoot and soil parameters needed for process
models of crop growth have been specified by Van Noordwjik and Van de
Geijn (1996).
1.9 What 15 the Functional Significance

of Field Mea5urement5?
Table 1.2 attempts to identify some of the more obvious links between root
properties and function. These are expanded in Section 1.4.1 above. The clear
answer to the question "What is the functional significance of, for example, large
root diameter or a particular root length?" is, in general, that we do not know.
We may, however, be able to assign value to features in particular situations e.g.
to deep roots under conditions of drought or to a high root length under con-
ditions of P deficiency. This is partially because the functional significance of
even major attributes has not been definitively determined, but more so because
plants have a range of root features which are designed to cope with more
extreme conditions than they will or may normally experience during their
life, i.e. there is a substantial functional excess capacitance built into root
systems. This in-built functional redundancy, or insurance, c1early confuses the
correspondence between a root property and the quantity of a function,
e.g. N0 3 uptake. Field measurement of a wider range of root characteristics
would give a more complete understanding of functional plasticity and the
ranges of morphological and physiological variation possible and he1p the
development of an understanding of re1ationships between environmental
profiles and roots.
1.10 Future Trends
The concept of"fit for purpose" is now being widely used to ensure that facili-
ties and equipment are designed for specific needs but without unnecessary fea-
tures. Many of the measurements made on and of root systems are not used or
are used in a qualitative (presence or absence) manner. Refining the question
to be asked by c1early specifying the information required will allow the con-
siderable effort required for any root study to be used to maximum effect,
usually by increasing the number of measurements taken. The development of
easily accessible modelling capabilities, computerised decis ion support systems
and data bases to aid the design of sampling strategies should greatly enhance
future root sampling exercises.
The capture of data, either from images of a mini rhizotron camera system
or from systems designed to assess root length from washed root samples, will
deve10p as a result of improved capacity for image analysis (Glasbey and
Horgan 1995). Deve10pments should improve our ability to extract data of
increasing complexity from direct and video images and to distinguish roots at
a range of deve10pment stages, inc1uding decaying, from a wide range of back-
grounds, inc1uding soils. A restricted ability to document root deve10pment
has, to date, aggravated a limited capacity to produce root systems with a
desired set of characteristics on new, and other, cultivars. Developments in
biotechnology now present the possibility of creating transgenic plants with
specific genes controlling particular root or developmental features, or the
ability, through the use of anti-sense genes of turning off particular aspects of
gene expression. This will allow the production of material which can be used
to test hypotheses on root functioning. The developments in rhizo-technology
discussed here will be needed to allow these biotechnological developments to
be realised.
30 D.Atkinson
References
Atkinson D (1978) The use of soil resources in high density planting systems. Acta Hortic 65:
75-90
Atkinson D (1983) The growth, activity and distribution of the fruit tree root system. Plant Soil
71: 23-36
Atkinson D (1985) Spatial and temporal aspects of root distribution as indicated by the use
of a root observation laboratory. In: Fitter AH, Atkinson D, Read DJ, Usber MB (eds)
Ecological interactions in soil. Blackwell, Oxford, pp 43-65
Atkinson D (1986) The nutrient requirements of fruit trees: some current considerations. Adv
Plant Nutr 2: 93-128
Atkinson D (1989) Root growth and activity, current performance and future potential. Aspects
Appl Bio122: 1-13
Atkinson D (1990a) Influence of root system morphology and development on the need for
fertilisers and the efficiency of use. In: Balligar VC (ed) Crops as enhancers of nutrient
use. Academic Press, London, pp 411-451
Atkinson D (1990b) Land: agriculture resource or wildlife reserve: reorganisation in the food
factory. Aberdeen Univ Rev 183: 218-225
Atkinson D (1991) Plant roots: an ecological perception. In: Atkinson D (ed) Plant root growth.
Blackwell, Oxford, pp ix-x
Atkinson D (1992) How long is the life span of a root? Trends Ecol Evol 7: 173-174
Atkinson D (1996) Why study roots. Agrofor UK 7: 22-24
Atkinson D, Fogel R (1997) Roots: measurement, function and dry matter budgets. In: Van
Gardingen PR Foody GM, Curran PJ. (eds) Scaling-up. Cambridge
Atkinson D, Mackie-Dawson LA (1991) Root growth: methods of measurement. In: Smith
KA, Mullins CA (eds) Soil analysis: physical methods. Marcel Dekker, New York, pp
447-509
Atkinson D, Wilson SA (1980) The growth and distribution of fruit tree roots: some conse-
quences for nutrit ion uptake. In: Atkinson D, Jackson JE, Sharples RO, Waller WM (eds)
Mineral nutrition of fruit trees. Butterworth, London, pp 137-150
Atkinson D, Naylor D, Coldrick GA (1976) The effect of tree spacing on the apple root system.
Hortic Res 16: 89-105
Biihm W (1979) Methods of studying root systems. Springer Verlag Berlin, Heidelberg New York
Brouwer R (1962) Nutritive influences on the distribution of dry matter in the plant.
Netherlands Journal of agricultural Science 10: 399-408
Caldwell MM (1987) Competition between root systems in natural communities. In: Gregory PJ,
Lake JV, Rose DA (eds) Root development and function. Cambridge University Press, Cam-
bridge, pp 167-186
Campbell R (1989) Microbial interactions leading to biocontrol of root diseases - reality or
myth. Aspects Appl Biol 22: 165-172
Cannell MGR, Dewar RC, Thornley JHM (1992) Carbon flux and storage in European forests.
In: Teller A, Mathy P, Jeffers JNR (eds) Responses of forest ecosystems to environmental
changes. Elsevier, London, pp 256-271
Coleman DC, Crossley DA (1996) Fundamentals of soi! ecology. Academic Press, San Diego
Coutts MP (1983) Root architecture and tree stability. Plant Soi! 71: 171-188
Curtis PS, Zak DR, Pregitzer KS, Teeri JA (1995) Above and below ground responses of Populus
grandidentata to elevated CO 2 and soi! N availabi!ity. Plant Soil 165: 45-54
Ennos AR, Crook MJ, Grimshaw (1993) A comparitive-study of the anchorage systems of
Himalayan balsam (Impatiens glandulifera) and mature sunflower (Helianthus annuus).
J Exp Bot 44: l33-146
Evelyn J (1662) A discourse of forest trees and the propagation of timber in His Majesty's
dominions. Stobbard, London
Glasbey CA, Horgan GW (1995) Image analysis for the biological sciences. John Wiley,
Chichester
Goss MJ, Barraclough PB, Powell BA (1989) The extent to which physical factors in the rooting
zone limit crop growth. Aspects Appl Biol 22: 173-182
Gowing DJG, Davies WJ, Trejo CL, Jones HG (1993) Xylem-transported chemical signals and the
regulation of plant growth and physiology. Philos Trans R Soc Lond Ser B Biol Sci 341:
41-47
Gregory PJ (1983) Response to temperature in a stand of pearl millet (Pennisetum typhoides S.
and H.) III. Root development. J Exp Bot 34: 744-756
Gregory PJ (1986) Response to temperature in a stand of pearl millet (Pennisetum typhoides S.
and H.) VIII. Root development. J Exp Bot 37: 379-388
Grime JP (1974) Vegetation classification by reference to strategies. Nature 250: 25-31
Grime JP, Cambo BD, Mackey JML, Crick JC (1991) Root plasticity, nitrogen capture and
competitive ability. In: Atkinson D (ed) Plant root growth. Blackwell, Oxford, pp 381-398
Harper JL, Jones M, Sackville-Hamilton NR (1991) The evolution of roots and the problems
of analysing their behaviour. In: Atkinson D (ed) Plant root growth. Blackwell, Oxford,
pp 3-24
Hellriegel H (1883) Beitrage zu den naturwissenschaftlichen Grundlagen des Ackerbaus mit
besonderer Beriicksichtigung der agrikultur-chemischen Methode der Sandkultur: eine
Auswahl von Versuchen ausgefiihrt an der Versuchstation Dahme. Braunschweig: Vieweg,
1883,796 P
Hendrick RL, Pregitzer KS (1992a) The demography of fine roots in a northern hardwood forest.
Ecology 73: 1094-1104
Hendrick RL, Pregitzer KS (1992b) Spatial variation in tree root distribution and growth
associated with mini-rhizotrons. Plant Soil143: 283-288
Hendrick RL, Pregitzer KA (1993) Patterns of fine root mortality in two sugar maple forests.
Nature 361: 59-61
Hennerty MJ, O'Kennedy, BT, Titus JS (1980) Conservation and reutilization ofbark proteins in
apple trees. In: Atkinson D, Jackson JE, Sharples RO, Waller WM (eds) Mineral nutrition of
fruit trees, Butterworth, London, pp 369-377
Hooker JE, Black KE, Perry RL, Atkinson D (1995) Arbuscular mycorrhizal fungi induced
alteration to root longevity of poplar. Plant Soill72: 327-329
Krebs (1995) Biogenetic soil fabrics: plant species, specific microfabrics of top soils and aspects
of their genesis (German) PhD Thesis, University of Hohenheim, Hohenheim
Lavender, EA (1992) Genotypic variation in the root system of Betula pendula. PhD Thesis,
University of Aberdeen, Aberdeen
Mackie-Dawson LA, Atkinson D (1991) Methodology for the study of roots in field experiments
and the interpretation of results. In: Atkinson D (ed) Plant root growth. Scientific
Publications, Oxford, pp 45-47
McCully ME (1987) Selective aspects of the structure and development of field-grown roots with
special reference to maize. In: Gregory PJ, Lake JV, Rose DA (eds) Root development and
function. Cambridge University Press, Cambridge, pp 53-70
32 D. Atkinson: Root Characteristics: Why and What to Measure
McKay H, Coutts MP (1989) Limitations placed on forestry production by the root system.
Aspects Appl Biol 22: 245-254
Miller RM, Jastrow JD (1992) The application of VA mycorrhizae to ecosystem restoration and
reclamation. In: Allen MF (ed) Mycorrhizal functioning. Chapman and Hali, New York, pp
438-467
Nye PH, Tinker PB (1978) Solute movement in the soi! root system. Blackwell, Oxford
Passioura JB (1991) Soil structure and plant-growth. Aust J Soi! Res 29: 717-728
Robinson D (1991) Roots and resource fluxes in plant communities. In: Atkinson D (eds) Plant
root growth. Blackwell, Oxford, pp 103-130
Smith FA, Smith SE (1986) Movement across membranes: physiology and biochemistry. In:
Gianazzi Pearson V, Gianazzi S (eds) Proc Eur Symp on Mycorrhizae. INRA, Paris, pp 75-24
Smucker AJM, Richner W, Snow VO (1995) Bypass flow via root-induced macropores (RIMS) in
subirrigated agriculture. In: Clean Water, Clean Environment for the 21st Century. Kansas
City, MO. pp 255-258
Tardieu F (1994) Growth and functioning of roots and of root systems subjected to soi!
compaction - towards a system with multiple signaling. Soi! Tillage Res 30: 217-243
Van Noordwijk M, Van de Geijn SC (1996) Root shoot and soil parameters required for process-
orientated models of crop growth limited by water or nutrients. Plant Soi! 185: 1-25
CHAPTER 2
Anatomy and Histology of Roots

and Root-Soil Boundary
E. de Neergaardl, O.B. Lyshede 2 , T.S. Gahoonia 3*, D. Care\ and J.E. Hooker 5
1,2 The Royal Veterinary and Agricultural University, Department of Plant Biology,
Thorvaldsensvej 40, 1871 Frederiksberg C, Copenhagen, Denmark
3 The Royal Veterinary and Agricultural University, Department of Agricultural Sciences,
Plant Nutrition and Soil Fertility Laboratory, Thorvaldsensvej 40, 1871 Frederiksberg C,
Copenhagen, Denmark
4 AgResearch, Grasslands Division, Ruakura Agriculture Research Centre, Private Bag 3123,
Hamilton, New Zealand.
5 School of Applied Sciences, University of Glamorgen, Pontypridd, Mid-Glamorgen CF37
lDL, UK
CONTENT S
2.1 Introduction 35
2.2 Root Sampling 35

2.3 General Methods for Studying Root Anatomy
and Histology 36
2.3.1 Whole Mounts 36
2.3.2 Tissue Clearing 37
2.3.2.1 Clearing Agents and Procedures 38
2.3.3 Squash Techniques 40
2.3.4 Tissue Fixation for Light- and Transmission Electron
Microscopy 41
2.3.4.1 Freeze Fixation 41
General freeze methods for Light Microscopy 41
2.3.4.2 Chemical Fixation 42
Warning: Chemical Fixatives Conta in Ingredients that are
Harmful to Human Health 42
Commonly Used Fixatives for Light Microscopy 43
Dehydration 44
Embedding 44
Protocols for Fixation, Dehydration and Embedding of Tissues
for Light Microscopy 45
Formaldehyde Fixation Followed by Dehydration with the
Embedding Medium 46
2.3.5 Sectioning 48
* Principal author
A.L. Smit et al. (Eds.), Root Methods

34 E. de Neergaard et al.
2.3.5.1 Hand Sections 48

2.3.5.2 FreshIy Cut Tissues 48
2.3.5.3 Embedded Tissues 48
2.3.6 Staining 49
2.3.6.1 Stains 51
2.3.6.2 Staining Procedures 51
Aniline Blue 51
Berberine-Aniline Blue 51
Chateliers Solution 51
DAPI (4,6-Diamidino-2-phenyIindole) 52
Evan's Blue 52
FSA (Fuchsin-Safranine-Astra Blue) 52
Neutral Red 52
Phenosafranine 52
Safranine-Fast Green/Safranine-Light Green 52
Sud an III and Sudan IV 52
Sudan Black B 52
Sudan Red 7B 52
Sulphorhodamine G (SRG) 53
Toluidine BIue O 53
Trypan Blue 53
2.3.7 Mounting 53
2.3.7.1 Mounting Media 54
Glycerol and Ethanol or Tween 54
Polyvinylalcohol (PVA) 54
Lactoglycerol 55
2.3.8 Procedures for Transmission Electron Microscopy 55
2.3.9 Procedures for Scanning Electron Microscopy 55
2.4 Root-Soil Boundary 56
2.4.1 Root Soil Interface in Thin Sections 56
2.4.2. Root Hairs 57
2.4.2.1 Quantification 57
2.4.2.2 Vitality 60
2.4.2.3 Nutrient Uptake via Root Hairs 61
2.4.2.4 Root Hairs in Nodulation Process 61
2.4.3 Mycorrhiza 62
2.4.3.1 Arbuscular Mycorrhizas 63
External Mycelium 63
Internal Structures 64
Spores 66
2.4.3.2 Ectomycorrhizas 66
References 69
2 Anatomy and Histology of Roots and Root-Soil Boundary 35
2.1 Introduction
Growing roots undergo many anatomical and morphological changes,

which influence their activity and nutrient uptake processes. Therefore, it is
often necessary to obtain structural information on the inner (anatomy and
histology) and outer (morphology) parts of roots. This chapter gives
an overview of methods to obtain information on anatomical and histo-
logical as well as morphological (root hairs and mycorrhiza) properties of roots.
The methods applied for the study of root anatomy do not, generally, differ
from methods used for the study of plant stems and leaves. Methods can
thus be found in generallaboratory books and manuals (Johansen 1940; Sass
1961; Jensen 1962; Purvis et al. 1964; O'Brien and McCully 1981; Neergaard
1997). Before a root specimen and a thin section of root-soil boundary can
be investigated under a microscope it has to pass along a chain of processes
which include sampling, killing and fixing, embedding, sectioning, and
staining. Details of these processes depend on whether light microscopy (LM),
transmis sion electron microscopy (TEM), or scanning electron microscopy
(SEM) is to be used. For LM and TEM, histochemical or immunological
tests may be applied additionally if the purpose is to demonstrate the presence
of certain compounds in cells or tissues. SEM deviates from the other two
mentioned with the exception of initial fixation steps and will be treated in a
separate section. Squash techniques for chromosome studies, most often
carried out on root tips, are also dealt with separately. Information on the
anatomy of roots can be sought in books on plant anatomy (Esau 1965, 1977;
Guttenberg 1968; Mauseth 1988; Fahn 1990) and root physiology (Luxova and
Ciamporova 1989).
2.2 Root Sampling
Sampling of roots for anatomical studies is more difficult and elaborate than
for stems and leaves, because roots usually grow in soil. Adhering soil
particles must be carefully removed by washing and careful brushing as they
otherwise may interfere during preparation for microscopy. The cleaning
problem may, in some cases, be avoided if plants are grown in a liquid
nutrient medium or in a substrate from which the roots are easily removed.
For anatomical studies, actively growing roots should be sampled. A soil-free
white elongation zone inimediately proximal to the root caps indicates actively
growing roots (McClilly 1995). When sampling the roots of target plants
in fields, the roots of weeds may pose a problem. Roots for anatomical
studies should be processed immediately or stored frozen in liquid nitrogen
to avoid anatomical changes prior to further processing. Root segments can
be frozen in liquid N z in cryo-vials for transportation and further processing.

Time of root sampling is also important, if exudation of water and other
substances from roots in situ is to be examined. Rhizosheaths adhering to roots
may be wet from early morning collection and drier from midday and midafter-
noon collections (McCully 1995), because of transpiration fluctuations during
the day.
For sampling of roots for mycorrhiza or root hair studies, it is necessary to
retain their natural features. A sharp knife with a long blade can be used to take
soil cores with roots to a depth of ca. 10 cm. To minimize damage to root hairs,
hyphal connections between roots and ectomycorrhiza roots should not be
pulled through soil. The soil cores can be wrapped in aluminium foil to prevent
crumbling.
For the isolation of roots from soil cores for mycorrhizal or root hair
studies, the soil cores should be kept soaked in water at least overnight.
Ultrasound treatment (l20W, 47kHz) in a water bath is useful for removing
soil particles adhering to roots, especially for root hair studies (Gahoonia
and Nielsen 1997) and perhaps also for mycorrhizal investigations. Moder-
ate magnification (6x), a black background and a lamp of daylight quality
are helpful to get a real impression of the mycorrhiza colours from the begin-
ning of the washing procedure. A white background masks the natural
colour. Cleared mycorrhiza must not be stored in water for long periods because
hyphae can continue to grow and colour changes can occur. Agerer (l991)
and Brundrett et al. (1996) have reviewed the methods of root sampling for my-
corrhizal studies.
2.3 General Methods for Studying Root

Anatomy and Histology
2.3.1 Whole Mounts
Whole mounts of finer roots and rootlets may be studied directly with a
light microscope at lower magnification (Fig. 2.1). Part of the root, e.g. the
root tip, may be bathed in a drop of water, a clearing agent, or a suitable
stain. It is thus possible to study the root zones comprising root cap, root
meristem, elongation zone, root-hair zone, and branching zone where the
formation of lateral roots occurs. Individual structures in the tissue may be
viewed by focusing at different levels. Mounting in iodine (2 g KI + 1 g Iz + 100
mlHzO) may reveal the presence of statoliths in the root cap.
Staining of whole mount plant tissues can be done using appropriate stains
and procedures (Sect. 2.3.6). Entire plant parts normally do not take up
stain readily and, therefore, it is time-consuming to infiltrate plant tissue with
Fig. 2.1. Lepidium sativum.

Root tip in whole mount.
1 root cap with statolith
starch, 2 dividing zone, 3
elongation zone, 4 separated
root cap cells alongside the
surface. (155 x). (Photo by
Ole B. Lyshede)
a staining solution. The use of ethanol, lactic acid, Of lactoglycerol as solvents

for the stain, will often facilitate infiltration. A brief heating of the spe-
cimen while immersed in the sta in or a vacuum treatment speeds up the
process. Stains dissolved in clearing agents or in the mounting medium may
prove successful.
2.3.2 Tissue Clearing
Clearing of plant tissue in whole mounts facilitates observation of intern al

structures includ ing infect ing fungi. Clearing removes certain pigments
(e.g. tannins and other polyphenols) and results in transparency of even rather
thick tissue parts. Internal structures may thus be observed in the light micro-
scope after mounting in a medium with a sufficiently high refraction index
(e.g. glycerol).
Tissue clearing is a general consequence of chemical fixation. Ethanol
(96%) is effective. It can be used cold, but works faster when the tissue is boiled
in it for 5 min. Ethanol is coagulating and a treated specimen cannot be used
afterwards for studies of fine cell structure. However, the procedure is excellent
for removal of air due to the low surface tension of ethanol.
Cleared specimens for light microscopy are obtained if glycerol or
chloral hydrate is used as the mounting medium. Tannins may be removed
by treatment with mixtures of ethanol and acetic acid unless they have
been fixed in aldehyde. NaOCI or H20 2 can be used for decolorization of
polyphenolics.
Root tissue often requires fairly severe methods for clearing and must be
treated with aggressive chemicals (e.g. alkalis), occasionally at elevated tem-
peratures or for long periods. However, use of a strong alkali over very long
periods leads to degradat ion of the cytoplasm and cellulose. This leaves liule
material for staining. Bevege (1968) recommends autoclaving in ethanol fol-
lowed by autoclaving in 1-5% NaOH for studies ofVA-mycorrhiza. O'Brien and

von Teichman (1974) describe a number of different autoclaving procedures of
plant tissue in ethanol. Gardner (1975) reviewed botanical clearing techniques.
2.3.2.1 C/earing Agents and Procedures
Ethanol and methanol are common clearing agents. Vacuum infiltration or

careful heating speeds up the process.
Lactic acid in an 85% solution is traditionally used alone or as lactophenol
(lactic acid + phenol + glycerol + water). Phenol is often replaced by glycerol in
lactoglycerol, which can be prepared by mixing lactic acid + glycerol + water
(v/v) in ratios 1 + 1 + 1 or 1 + 2 + 1 or 1 + 4 + l.
Higher ratios of lactic acid (14 + 1 + 1) have proved especially useful for
treatment of roots. Pure (100%) lactic acid is also recommended.
Lactophenol is made by replacing half the amount of glycerol with
phenol. The use of phenol should be limited due to its harmful effects on
human health. Lactoglycerol can be used together with stains. Aniline blue (=
coUon blue), dissolved in lactoglycerol is a commonly used stain (0.5-l.0%
solution).
The following clearing agents have proved useful on leaf tissue and should
be tried on roots:
Lactoglycerol-ethanol (1 + 2) thoroughly mixed. Clearing is acceler-
ated by heating in a boiling water bath for 3 min or when left in the solution
overnight.
Acetic acid-ethanol. A mixture 1 + 1 (v/v) is generally used. Alternatively
try this: treat with heated 96% ethanol and transfer the material to acetic acid
+96% ethanol + water (1 + 4 + 5). The method may be combined with staining.
Increased ratios of ethanol are reported in the literature, e.g. acetic acid and
50% ethanol (1 + 17) for clearing of leaf sheaths of wheat (Strausbaugh and
Murrey 1989).
Clarke jixatives consist of mixtures of acetic acid and ethanol (for example
1 + 3). Clarke is used as a fixative of root tips prior to chromosome studies by
squash techniques (Sect. 2.3.3). It may be used as a clearing agent, but is not
suitable for cytological and anatomical investigations.
Acetic acid-glycerol. A 1 + 1 mixture is recommended for clearing very
young roots.
Chloral hydrate is used in aqueous solution, alone or in combination with
other chemicals (Table 2.1). Specimens can be left in the solution for 4-5 days
or heating which removes air from the specimen can shorten clearing. NB!
Chloral hydrate is very injurious to human health and should be used only if
other methods faiI and always in a fume hood using gloves.
Table 2.1. Chloral hydrate solutions
(Dhingra and Sinelair 1995): (Gerlach 1984):

Chloral hydrate 250 g Chloral hydrate 160g
Distilled water 100 mI Distilled water 100ml
Glycerol 50ml
Herr (1971) used the folIowing procedure:
Lactic acid + chloral hydrate + phenol + elove oiI + xylene
2 + 2 + 2 + 2 + 1 (by weight)
NB! Working with xylene should be done in a ventilation hood,

and use of xylene substitutes (e.g. histoelear) is preferable.
Lactic acid-methanol-chloroform (1 + 1 + 1) is occasionally recommended

as a clearing agent for entire leaves (24h). By subsequent staining, fungal
structures in the host tissue are demonstrated. The stain is dissolved in
lactoglycerol, lactophenol, or 95% ethanol and boiled for 2 min. (For staining
procedures see Sect. 2.3.6.) Chloral hydrate may be used as a final treatment
(24h) before observation. (NB: toluidine blue O is destained by chloral
hydrate).
Chloroform can be used in a mixture with methanol (2 + 1, v/v) for clear-
ing of leaf tissue. Kuck et al. (1981) used this solution (1-2h at 20°C) for pre-
treatment of rust-infected leaves before applying fluorescent stains for
demonstration of fungal structures.
Alkalis. KOH or NaOH are commonly used for roots and lignified tissue.
Woody tissue can, for example, be treated with lM KOH e.g.:
1. Treat specimens with boiling lM KOH. If necessary, infiltration can be facil-
itated by vacuum,
2. Neutralise with lM HCI (briefly),
3. Mount with lactoglycerol or stain with staining reagents dissolved in
lactoglyceroI.
Roots of perennial plants for investigation of fungal infection can be treated
with strong alkalis (S. Rosendahl, pers. comm.):
1. Immerse roots in 10% KOH at 70-90°C for 30-60min,
2. Add lM HCI (98mI32% HCI + water until1000ml) for neutralising,
3. Stain with staining solution (e.g. trypan blue) in lactoglycerol at 70-90°C.
An alternative method includes the following treatment between steps 1 and 2:
30 mI 10% H20 2 + 3 mI NH 4 0H + 967 mI water for 40 min at 30°C.
Vegetable oils. The commercial product Histoclear consists of citrus oiI and
vegetable oiI. It is non-toxic and claimed to be an efficient clearing agent unless
used on lipid-rich specimens. It is also used as a dewaxing agent.
2.3.3 Squash Techniques
Studies on chromosome number and morphology are very often carried out
on root tips using squash methods (Box 2.1). Pre-treatment of the roots
with colchicin, 8-hydroxyquinoline, or a-monobromnaphtalene for 3-4h
wiU often make the chromosomes more distinct in the microscope (NB! Longer
pre-treatment causes multiplicat ion of the chromosomes). Sudden cooling of
root tips is an alternative to the chemi cal treatment and may also be useful.
BOX 2 .1. Squash technique by Aceto-orcein :

1. Pre-treatment with colchicin a.O. (if preferred);
2. 1- 2mm of the root tips are excised and fixed in Clarke's fixative
(absolute ethanol/acetic acid, 3: 1) for 10 mini
3. Heat root tips in 1 M HCI at about 60°C tiU soft (5- 10 min);
4. Place root tips in a droplet of Aceto-orcein on a glass slide and mash
with a needle;
5. Drop a large cover slip onto the mashed root tips and heat over a
spirit ftame, not exceeding 60 °C. (Check with back of your hand; it
should be very warm, but should not hurt);
6. Place two pieces of filter paper at opposite edges of the cover slip and
hold them steadily with two fingers;
7. Squash the material by moderate tapping with the wooden end of a
matchstick;
8. Replace filter papers with fresh an tap of the cover slip; squeeze with
your thumb;
9. Examine the material.
The preparation may keep for a few days if the cover slip is framed with nail
polish.
Permanent slides can be made as described in Box 2.2.
BOX 2 .2. Making Permanent Slides:

1. Freeze the squashed material with comprimated CO 2;
2. While frozen, separate slide and cover slip with a sharp knife and let
both pieces faH into absolute ethanol for no more than 15 s;
3. Transfer both to absolute ethanollxylene (1 : 1) and on ta xylene;
4. Make two permanent slides by mounting in Canada balsam or other
suitable mounting medium.
Staining of chromosomes with Acetocarmine, Fuchsin, or other stains can be

carried out with slightly deviating methods (Jensen 1962).
2.3.4 Tissue Fixation for Light-

and Transmission Electron Microscopy
Observation of gross anatomical structxures can take place using whole mounts
(see Sect. 2.3.1), or by hand sectioning of fresh tissue (see Sect. 2.3.5). However,
before detailed studies of the fine structures of tissue can be done in LM or
TEM it must be fixed, embedded, and sectioned. Prior to killing and fixation of
the plant material it must be cut into appropriate pieces depending on the final
investigation. For TEM at least one dimension should not exceed about 2 mm
in order to facilitate a quick penetration of the fixative. The tissue compounds
are immobilized and preserved by the fixation procedure and the purpose of
the subsequent embedding is to stabilise and support aH ceH components at
their original site during sectioning. Two fundamentaHy different methods of
fixation are described: freeze fixation and chemical fixation.
2.3.4. 1 Freeze Fixation
By freeze fixation the conservation of the tissue and immobilization of its com-
ponents take place by low temperature and the supporting medium is the
frozen water content of the tissue. Freeze fixation is generaHy faster than chem-
ical fixation and the risk of chemical alterations of the ceH components is much
lower since aH biologic al processes are rapidly arrested. Thus, changes of dis-
tribution of elements and loss of diffusible ions are kept at a minimum. As a
consequence enzyme- and immunological properties of the tissue are relatively
unaffected. The negative effects are mainly caused by the formation of ice crys-
taIs, which may damage ceH components, especiaHy the membranes. Liquid
nitrogen is therefore extensively used because the freezing is rapid and strongly
diminishes the formation and size of ice crystals. Most plant tissues, including
roots, can be frozen directly from the living state.
Freeze fixation can be used prior to both light and electron microscopy. In
the foHowing, the first procedures described aim at light microscopical work.
Freeze substitution and high pressure freezing can be used for both light and
electron microscopy, depending on the embedding medium used.
General Freeze Methods for Light Microscopy. Most simple freezing of plant
tissue takes place by infiltrating smaH specimens with a proper cryoprotective
substance (Table 2.2) and freezing them on a freeze stage, preferably placed in
a cryostat. This is a cooled chamber in which freeze fixation and sectioning
Table 2.2. Cryoprotective substances
Sucrose solution (saturated solution = sucrose + water (52g + 30 mI, w/v)

Gum arabic (40% aqueous solution + a crystal of thymol)
Gum arabic + sucrose (40% gum arabic solution + saturated sucrose (5 + 3, v/v)
Glycerol
Glycerol-gelatin (= glycerol jelly)
Water agar
Gelatin
polyethylenglycol
Polyvinylpyrrolidone
Dimethylsulphoxide
take place. The frozen tissue adheres to the freeze stage by the frozen cryopro-
tective substance and is supported by more cryoprotectant added gradually on
alI sides prior to sectioning. Knox (1970) recommends tissue infiltration before
freezing with dimethylsulfoxide or glycerol. These chemicals can be used as
anti-frost additives to gelatin used as supporting medium during freezing.
Wood can be saturated with water before freezing, but most tissue types should
be treated with cryoprotective substances.
A number of commercial products are available, marketed under various
trade names.
Glycerol is a good cryoprotective substance, but diffuses very slowly into
plant tissue. A precaution is to let the plant take up glycerol in the living state,
but glycerol is osmoticalIy active and there is a danger of plasmolysis unless
applied in gradualIy increasing concentrations. When infiltrated gelatin is used
as a cryoprotective substance, a graded series of gelatin solutions can be used
for dehydration and as an embedding medium. Dehydration takes place by
using 10%-40% (up to 70%) aqueous gelatin. After partial drying at 37°C for
some hours pieces are cut and dried in a vacuum desiccator.
2.3.4.2 Chemical Fixation
Warning: Chemical Fixatives Contain Ingredients that are Harmful to Human

Health. Chemical fixation aims at protein fixation as a key point. The chemi-
cals used for protein fixation are either gel forming (cross-linking fixatives) or
coagulating (precipitating fixatives). The gel forming fixatives do not (or
only slightly) denature the proteins. Coagulating fixatives act by irreversibly
denaturing agents (Lyon 1991). Gel forming fixatives (formaldehyde, glu-
taraldehyde, croton aldehyde, acrolein, potassium dichromate and osmium
tetroxide) act as stabilizers of the protein configuration by establishing bonds
between polypeptides in the same protein and between neighbouring pro-

tein molecules. These fixatives are recommended for TEM. Coagulating fix-
atives (methanol, ethanol, chromium trioxide, mercury chloride, picric acid,
hydrochloric acid, nitric acid and acetone) denature the proteins by opening the
tertiary structure, displaying the re active groups of the proteins. Coagulating
fixation is in some cases an advantageous method for light microscopy provided
that the protein network formed is not too coarse, but it cannot be used for
TEM (Lyon 1991).
Rapid infiltration with the fixative is important. This may be obtained
with small sizes of tissue, at least in one dimension; the use of vacuum
will enhance the infiltration. The concentration of the fixative, its pH and osmo-
larity, and also the temperature at which it is applied are critical for a good
fixation.
The concentration of the fixative is changed on its way through the tissue,
and at the same time osmotic changes take place in the remaining living parts.
At low temperature the autolytic cell processes are slower, but the diffusion time
of the fixative is longer.
Commonly Used Fixatives for Light Microscopy. Formaldehyde has a high dif-
fusion rate compared with other fixatives and therefore is a rapidly working fix-
ative. Used alone as a formalin solution (1-8% in buffer) it is commonly used
as fixative. Formalin (2-4% aqueous solution, Table 2.3) in combination with
acetic acid and ethanol (FAA) is a widely used fixative for routine anatomic al
studies both for LM and SEM.
Other ratios are found in the literature and should be tried if unsatisfac-
tory results are obtained. The dehydration process following FAA fixation starts
with 70% ethanol. FPA (formalin-propionic acid-alcohol) is made by replacing
acetic acid in FAA with the same amount of propionic acid and is considered a
good fixative.
Glutaraldehyde (2-4% in phosphate-buffer) is used for both light and elec-
tron microscopy. It is used alone or in combination with other fixatives (Sect.
2.3.8). If cytological details are required, aldehyde fixation is strongly recom-
mended; this is especially good for TEM. See O'Brien and McCully (1981) and
Glauert (1975) for fixation methods.
Table 2.3. Formalin-acetic acid-alcohol (FAA)
Percentage formaldehyde: 2% 4%
70% ethanol 90 mi 8Sml
40% formalin Sml lOml
Glacial acetic acid Sml Sml
Clarke jixatives consist of mixtures of ethanol and acetic acid (for example
3 + 1). They can be used for studies of chromosomes (Sect. 2.3.3), but do not
work well for tissue fixation.
CRAF jixatives are mixtures of chromium trioxide, acetic acid and for-
maldehyde in different ratios, depending on the tissue character. CRAF fixatives
have similar properties to FAA (Sass 1961; O'Brien and McCully 1981).
Dehydration. Chemically fixed tissue must be dehydrated prior to embed-

ding. Commonly used dehydration media are ethanol and acetone. Boyde and
Maconnachie (1981) have demonstrated that minimum morphological changes
of the specimen take place if dehydration starts with 70% ethanol or 70%
acetone. It should be kept in mind that many cell constituents are removed from
the tissue during the dehydration. A good fixation procedure is the only way to
keep such losses at a minimum. Acetone should probably be preferred in most
cases.
Propylene oxide is often used as an intermediate step between dehydration
and resin infiltration, especially for TEM, but it is toxic and potentially car-
cinogenic and its use must be strongly limited.
Acetonitrile is recommended since it causes less extraction of cell lipids,
including the phospholipids of membranes. It is miscible with water and with
epoxy resins (Edwards et al. 1992).
Acidified dimethoxypropane (3 drops of O.IM HCI + 25ml dimethoxy-
propane) is used as a dehydration agent in electron microscopy, followed by two
changes of acetone (Carlson et al. 1991) or by a stepwise change directly to the
embedding medium (gradually increasing the concentration to pure embed-
ding medium).
Embedding. Tissue is embedded to stabilize the cell components dur ing sec-
tioning, thereby facilitating the production of thin uniform sections. Most
simple is soaking in water (under vacuum) and embedding in ice or an aqueous
cryoprotective substance (see text on freeze fixation in Sect. 2.3.4.1) using a
freeze microtome or a cryostat. Tissue infiltration with embedding media is nor-
mally done at room temperature because the media are generally highly viscous
and their diffusion into the tissue is slow at low temperatures. Cell walls, crys-
talline substances and storage structures such as starch in roots are difficult to
infiltrate with embedding media and may create special problems. Starch may
be removed with amylase prior to embedding for LM, but not for TEM (Lyshede
1977,1979).
Paraffin wax or different types of resins are common embedding media. Wax
embedding media include paraffin wax (Paraplast) and Steedman's wax. Para-
plast is a commercial product, whereas Steedman's wax is produced by poly-
ethylene glycol400 distearate mixed with l-hexadecanol, 9 + 1 (w/w; Norenberg
and Barrett 1987).
Acrylic resins include the Lowicryls, the London Resins (LR White and LR
Gold), and the methacrylates. The methacrylates and the London Resins are
used for light microscopy.
All these are available as commercial products. Epoxy resins (Epon, Araldite
and Spurr) are the most frequently used embedding media for trans-
mission electron microscopy (TEM). The acrylic resins (Lowicryls and LR-
resins) are used for special purposes for TEM, such as studies involving
immuno-reactions.
Protocols for Fixation, Dehydration and Embedding of Tissues for Light Micro-
scopy. Common procedures are suggested below with reference to only the
most well-known fixatives, dehydration media and embedding media. The pro-
tocols should be modified to suit the user's special situation.
The duration of each single treatment step depends on size and cha-
racter of the material. Since infiltration is based on diffusion, the degree of
permeability of the tissue should be considered. Infiltration is regarded as
complete when the root specimen sinks to the bottom of the infiltration
vial. The suggested time course in the schemes will apply for tissue not exceed-
ing 2 mm in at least one dimension, which means that no distance from a
cut surface is more than 1 mm. In other cases all treatments should be
extended.
BOX 2.3. Formaldehyde Fixation and Subsequent Embedding

in Glycolmetacrylate:
1. Fixation with 2-4% formaldehyde in buffer solution (e.g. a 0 .1 M
phosphate buffer or a cacodylate buffer, pH 6.8), using vacuum
infiltration until the tissue sinks to the bottom of the fixation vial.
Leave the vials in the refrigerator overnight.
2. Wash with buffer three times in the course of 1 h;
3. 10% ethanol, 2h;
4. 20% ethanol, 2 h;
6. 50% ethanol, 2 h (or overnight in refrigerator);
7. 70% ethanol,2h;
8. 70% ethanol, overnight;
9. 96% ethanol,2h;
11. 96% ethanol + in filtration solution, 1 + 1,2 h;
12. Infiltration solution (vacuum, if necessary) overnight in refrigerator;
13. Embedding in plastic or gelatine capsules.
BOX 2.4. FAA Fixation and Subsequent Embedding

in Glycolmetacrylate:
1. Formalin-acetic acid-ethanol (FAA), using vacuum infiltration until
the tissue sinks to the bottom of the fixation via!. Leave the vials in
the refrigerator overnight.
6. 96% ethanol + infiltration solution, 1 + 1, 2h;
7. Infiltration solution (vacuum, if necessary) overnight in refrigerator;
8. Embedding in plastic or gelatine capsules.
The dehydration procedures (in Boxes 2.3 and 2.4) apply if the embedding
medium contains approximately 5% water. Otherwise the dehydration proce-
dure should continue to 100% ethanol followed by 100% ethanol/infiltration
solution (1 + 1, v/v).
Formaldehyde Fixation Followed by Dehydration with the Embedding Medium .

For enzyme histochemistry Gerrits and Zuideveld (1983) suggest the folIow-
ing procedure (Boxes 2.5 and 2.6) based on dehydration with the embed-
ding medium (metacrylate, in this case Technovit 7100) and with alI steps
at 4°C:
BOX 2 .5. For maIdehyde Fixation Followed by Dehydration

with the Embedding Medium:
1. Fix in 4% formaldehyde in a 0.1 M phosphate or cacodylate buffer,
pH 6.8, 2h;
2. Wash with buffer, I8h;
3. 70% Embedding medium 30 mini
4. 90% Embedding medium 30 mini
5. IOO% Embedding medium 2 h;
6. 100% Embedding medium I2h;
7. Embedding.
BOX 2.6. Formaldehyde/Histochoice Fixation Followed by Embedding

in Paraffin Wax:
1. Fixation in 4% formaldehyde in buffer pH 6.8. 0.1 M (alternatively
Histochoice). Use vacuum for infiltration;
2. Formaldehyde (alt. Histochoice) + absolute ethanol, 4 +1;
3. 30% isopropanol;
4. 50% isopropanol;
5. 70% isopropanol;
6. 90% isopropanol;
7. 95% isopropanol;
8. 100% isopropanol, 3 times.;
9. Add paraffin wax at +60 °C, gradually increasing the concentration
of wax to 100%;
10. Embed in paraffin wax (sets at room temperature);
Il. After sectioning, the specimens are collected on glass slides
specially treated for adherence. The wax is removed from the plant
tissue by four times treatment with Histoclear, followed by five times
isopropanol.
BOX 2.7. Formaldehyde Fixation Followed by Embedding with

Steedman's Wax. Baluska et al. (1992) Recommended the Following
Method for Fine Structure Studies of Primary Root Apices, Including
Immuno-Analysis:
1. Fixation in 4% formaldehyde (buffer specified for protecting the
structures of interest);
2. Dehydration in an ethanol series diluted in the same buffer as the
fixative, until absolute ethanol;
3. Infiltration at +37°C with a series of mixed absolute ethanol and
Steedman's wax: 2 + 1, 1 + 1, 1 + 2 (v/v), 2h in each step. (Steedman's
wax: polyethylene glycol400 distearate mixed with l -hexadecanol,
9 + 1 (w/w);
4. Pure wax, changed three times under vacuum to remove ethanol;
5. Embedding. The wax sets at room temperature.
6. After sectioning, the specimens are collected on glass slides,
pretreated for adherence. The wax is removed by ethanol.
2.3.5 5ectioning
2.3.5.1 Hand Sections
Fresh or fixed plant tissue can be sectioned using a razor blade. Small or thin
plant parts can be supported between pieces of elderberry pith (Sambucus
nigra) or different types of roots or tubers (carrot, potato) or other relatively
soft, homogenous materials (soap, cork, foam plastic) to facilitate hand
sectioning.
Lindauer (1972) gives an overview of sectioning methods. Treatment
with ethanol is effective in removing the air from the plant tissue and further-
more it hardens soft tissues. On the other hand, glycerol-ethanol (1 + 1 (v/v)
mixtures can soften hard tissues. Increasing the glycerol ratio augments the
effect.
Woody roots may be sectioned directly in the fresh state by hand or after
clamping in a sliding microtome. To soften the material (under vacuum, if nec-
essary) it may be pre-treated by boiling in water or placed in a mixture of equal
amounts of water, ethanol, and glycerol.
2.3.5.2 Freshly Cut Tissues
Tissue (living or formaldehyde fixed) is mounted with cyanocrylate (superglue)

to slides and cut using a vibratome. This instrument produces 70-,Um-thick sec-
tions. Staining or enzyme treatment can take place before microscopy. This
method is used as a pre-treatment before confocallaser microscopy. The use of
specific fluorescent stains (Blancaflor and Hasenstein 1993,1997) can mark the
object of interest.
2.3.5.3 Embedded Tissues
By embedding, the material is penetrated and surrounded by the embedding

substance (e.g. ice, paraffin, epoxy resins, other plastics). Apart from ice, the
material must be dehydrated prior to penetration of the embedding medium.
The material is passed through an ascending concentration of the dehydrating
agent, e.g. alcohol or acetone, followed by ascending concentrations of
the embedding medium dissolved in the dehydrating agent and ending with
pure medium. Finally the embedding medium is polymerized. For details of
embedding methods, see Sass (1961), Glauert (1975), O'Brien and McCully
(1981). Spurr's re sin is especially popular embedding material for TEM (Spurr
1969).
Embedding Of fresh tissue may be carried out in solid paraffin. This is a fairly
rapid and simple method giving sufficient support to the tissue during the sec-
tioning. The tissue is immersed in molten paraffin and then cooled down to a
temperature slightly above the melting point. After a suitable period of infil-
tration the specimen is cooled down. If necessary the paraffin can be removed
using xylene.
Paraffin infiltrated and embedded material is ready for sectioning imme-
diately after the cooling of the paraffin. Material embedded in plastic can be
sectioned shortly after polymerization. Hard or brittle blocks may be moist-
ened on wet filter paper before sectioning. Warming in an incubator may harden
soft plastic blocks.
The microtome is equipped with a holder(s) for glass knives or metal knives
or adapters for disposable knives. Glass knives are produced on a special knife
maker. They have a more perfect cutting edge, but for most light microscope
work metal knives will provide sections of sufficient quality. In addition, they
allow large sections and in this way one benefits from the advantages of many
plastic embedding media in making large sections of less thickness. Metal
knives are available as disposable knives or as knives which can be resharpened.
Cutting angle, cutting speed, humidity and temperature are critical factors
during sectioning. Generally, most of the necessary information appears in the
instruction sheets of the commercial embedding kits.
During sectioning, the material is compressed and the sections need to be
stretched on a water surface, most conveniently on droplets or on a thin layer
on the glass slide. Stretching is best obtained with pure, distilled water with
maximum surface tension (Gerrits et al. 1987). Low concentrations of additives
are occasionally required (for instance 1% acetone). As temperature affects
surface tension the stretching and the subsequent water evaporation should
preferably take place at room temperature, but often a heating plate is used to
speed up the process.
2.3.6 Staining
Staining increases the contrast of biological material when observed under the
microscope. Furthermore, depending on specificity, the staining performs a
cyto-/histochemical analysis of the tissue components. There exists a wide
range of stains for different purposes (Table 2.4). For TEM, staining with uranyl
acetate followed by lead citrate is routinely used to increase contrast. Selected
stains for special compounds may be used.
Paraffin-embedded material must have the paraffin removed and the tissue
hydrated prior to staining. This is most often done through xylene (xylene/l 00%
Table 2.4. List of common stains
Stains for general anatomicallhistological Stains for vitality

survey 4',6 Diamidino - 2 - phenylindole (DAPI)
Aniline blue Neutral red
Chatellier's solution Phenosafranine
Fuchsin - safranin - astra blue (FSA) Trypan blue
Safranine-fast Green or Evan's blue
Safranine-light Green
Toluidine blue o
Trypan blue
Stains for localization of fungal structures Stains for lipids/suberin

in host tissue (mycorrhiza, root Sudan III
pathogens) Sudan IV
Aniline blue Sudan Black B
Toluidine blue O Sudan Red 7B
Trypan blue
Staining procedure for suberin, lignin and Staining method for evaluation of tissue
callose water contents
Berberine-aniline blue Sulphorhodamine G
ethanoll + 1), followed by a graded ethanol series to water (ethanollOO%, 96%,

70%,50%,30%, water). Steedman's wax is removed from sections by ethanol.
Plastic-embedded material may be stained when it adheres to the slide and is
dry. Hand sections may be stained direct1y on glass slides prior to mounting or
transferred through the stains in small sieves.
A number of pitfalls must be avoided in the interpretat ion of the stained
material. Artefacts commonly occur for several reasons: the staining solution
may vary from time to time, or between manufacturers. The active ingredient
may only work in the predicted way when other substances are present or
absent, under controlled pH, molarity, and temperature.
The quality of the staining agents is critical. Presence of impurities and
the actual percentage of the active substance may vary from one batch to
another. Proper controls will ensure that conclusions are drawn on a safe
basis. Stained and unstained material should always be compared (positive
control). Comparison of unfixed or cryofixed tissue with tissue treated
with chemicals is recommended. It should be noted that different chemical fix-
atives affect the electrostatic charge of the resulting specimen and hence inter-
fere with most staining reactions: Formaldehyde fixation of cytoplasm will
result in an increased affinity to acid stains, whereas Cr0 3 will increase the affin-
ity to basic stains. On the other hand, chromosomes will show increased affin-
ity to acid stains after treatment by either of these two fixatives (formaldehyde
and Cr0 3 ).
Other control measures include inhibition of the assumed process or
extraction of the tissue component in question followed by comparison with an
untreated specimen (negative control).
2.3.6.1 Stains
NB!!: Almost all stains are somewhat harmful to human health. Use a fume hood
and take care when preparing the staining solution and when handling the
stain. Waste should be deposed of in suitable waste containers. The most com-
monly used stains are listed in Table 2.4.
Staining for specific tissue components is relevant for more detailed studies.
More Information concern ing this can be found in Lillie (1977), Clark (1981),
Gahan (1984) and Neergaard (1997).
2.3.6.2 Staining Procedures
Aniline Blue. Aqueous solution, 0.5-1.0%, alternatively dissolved in ethanol,

glycerol or lactoglycerol. Glucans are positive.
As Leuco-Aniline Blue: the stain is dissolved in buffer pH 8.3 or higher, and
the tissue observed using fluorescent light at 365-405 nm. Leuco-Aniline Blue is
used for localisation of callose.
Berberine-Aniline Blue. Aqueous solution of berberine hemisulphate, 0.1%

(w/v). Stain for 1 h, subsequently wash in water. After that, stain in aniline blue
solution (0.5% aniline blue w/v, aqueous) 30min and finally treat for several
minutes with FeCl3 in aqueous 50% glycerol, prepared by adding glycerol to fil-
tered aqueous FeCI3 •
Mount in the last solution. Observe with fluorescence microscopy at 365 nm.
Suberin is blue-white to blue, lignin is bright yellow, callose is blue-white, and Cas-
parian strips in the endo- and exodermis are intense yellow (Brundrett et al. 1988).
Chatelliers Solution. Dissolve 1.1 g aniline sulphate in 70 mI distilled H 2 0 under

heating. Filter this solution into a mixture of 60 mllactic acid and 45 mllactic
acid saturated with Sudan III. Dissolve 1 g of KI in 10 mI distilled H 2 0 and add
10 mI 96% ethanol. Dissolve iodine (I2) into this solution and pour into the lactic
acid mixture. Add 6 mI cone. HCI and shake thoroughly.
Mount in this solution, add cover glass, and boiI over flame. Lignin is yellow,
cutin and suberin are red, starch is black, and protein is yellow. The lactic acid
is a clearing agent. Preparations may keep for about a week.
DAPI (4,6-Diamidino-2-phenylindole). Used as a vital stain (0.002% in buffer)

applied for 30min in the dark, or for fixed material (0.01-0.05%) overnight at
+4°C, and observed at 365 nm indicates DNA, hence it can be used for studies
of vitality. Phenols are positive, too.
Evan's Blue (3,7-Diamino-5-phenylphenazinium chloride) used as vital stain.

Stain in a 0.05-0.25% aqueous solution for 5 min. Living cells remain unstained,
dead cells become blue.
FSA (Fuchsin-Safranine-Astra Blue) is made up by mixing basic fuchsin (0.1 g)

safranin O (OAg), astra blue (1.0 g) and acetic acid (20ml). Subsequently water
is added up to 11. Stain for 1-5 min. Cellulose is blue, lignin, cutin and some
suberins, nuclei and plastids are red, whereas starch is unstained.
Neutral Red is a non-toxic stain used for studies of living cells, for example to
observe plasmolysis. The stain is used as an aqueous solution (0.1 %, w/v) and
stains lipids in vacuoles and certain cell walls.
Phenosafranine is used to distinguish living from dead cells. A 0.1 % aqueous

solution, it is taken up by dead or abnormal cells immediately, while living cells
remain unstained for at least 30 min.
Safranine-Fast Green/Safranine-Light Green. A mixture of 1.5 g safranine O

and 0.5g light green or fast green is dissolved in 1.25ml conc. HeI + 75ml
distilled water + 120 mI 96% ethanol. Tissue sections, after being treated
with ethanol, are stained for 5 min, followed by differentiation with 96%
ethanol 3 x 30 s. Cytoplasm and cellulose walls become green, nuclei, chro-
mosomes, as well as lignified, suberised, and cutinised walls become red.
In the case of fungal infections, hyphae become red whereas host cells are
green.
Sudan III and Sudan IV. Saturated solution in 96% ethanol. Stain in solution
for few minutes. Lipids, cutin and suberin are red.
Sudan Black B. 1% Sudan black B is dissolved in 70% ethanol, the tissue is

stained 1-1.5 h, subsequently washed in 70% ethanol (3 x 15 s), and mounted in
75% glycerol before microscopy. Lipids are positive.
Sudan Red 7E. The stain is dissolved (0.1 % w/w) in polyethylene glycol 400-
glycerol (I + 9). Lipids are positive. It is especially good for demonstrating the
presence of suberin.
Sulphorhodamine G (SRG). The specimen is treated 5 s in a 1 mM freshly pre-

pared and filtered solution of sulphorhodamine G before sectioning. Hand sec-
tions are mounted in paraffin oiI for fluorescence microscopy. Distribution of
induced fluorescence corresponds to water content of the tissue. For details, see
Watt et al. (1996).
Toluidine Blue O. The staining solution is prepared by dissolving 0.05% tolui-

dine blue O in buffer, pH 4.4-6.8. Sections or whole tissues are stained for 5-
10min followed by differentiation using the same buffer until excess stain is
removed. Cellulose and other glucans become red-violet depending on pH, lig-
nified elements and other polyphenols become bluish-green. Fungal hyphae are
often distinguished from the surrounding host tissue because of the difference
in cytoplasmic pH.
Trypan Blue. A 0.01-0.2% aqueous solution is used for root staining. The
method is often used for investigation of mycorrhizas. Suitable for whole
mounts as well as sections. Aiso used as a stain for vitality, as living cells gen-
erally remain unstained.
2.3.7 Mounting
After staining, the sections are ready for mounting and microscopy. Initial
observation is recommended using water (a drop of distilled water before the
cover slip is placed). For permanent slides different embedding media are com-
mercially available.
After staining with water-soluble stains, the use of non-aqueous media has
the advantage of stabilizing the material without leaching the stain. The change
from the hydrophilic staining phase to a hydrophobic mounting medium is done
by drying or by dehydration through an ascending series of ethanol, ethanol-
xylene (1 + 1) ending with 100% xylene before mounting. Often the mounting
medium has to be hardened by heating in an incubator. Xylene-containing media
must be stabilized in this way only under conditions with sufficient ventilation
(fume hood or other arrangement), as the evaporating xylene is hazardous
to health. For plastic-embedded sections drying is sufficient before mounting.
Fluorescence microscopy requires a non-fluorescent mounting medium.
Aqua-mount contains phenol and is thus poisonous. Certain aqueous stains
are destained by Aqua-mount. DPX and Canada balsam contain xylene (must
be handled in well-ventilated places such as fume hoods). Specimens stained
with zincchloride-iodine, Millon's reagent, Sudan IV and J-KJ should not be
mounted in Canada balsam.
2.3.7.1 Mounting Media
Table 2.5. Mounting media
Aqueous mounting media Non-aqueous mounting media

Water Canada balsam
Glycerol DPX
Glycerol-gelatine (= glycerol jeUy) Caedax
Aqua-mount Eukitt
Polyvinylpyrrolidone (PVP) Euparal
Aqueous saturated CaCI,
PVP + CaCI, Non-fluorescent mounting media
Polyvinyl alcohol (PVA) Water
Shear's medium Glycerol
Gelatin-sucrose DPX
Malinol Gurr's fluormount
Gelvatol
Specimens of plant tissue are usually observed in distilled water or 70% ethanol
or 0.1 M CaClz or 0.25 M sucrose. If the mounting medium evaporates too
rapidly, glycerol or lactoglycerol can be used instead. It should be kept in mind
that these chemicals may cause swelling of the tissue.
Glycerol and Ethanol or Tween. This can be used as a mounting medium for
specimens for immediate observation using 50% aqueous glycerol. In order to
facilitate infiltration and remove air from the tissue, a small amount of ethanol
or a drop of Tween 20 is added. Alternatively, an aqueous solution of Tween 20
(1 drop in lOOml water) is often sufficient as a mounting medium.
Polyvinylalcohol (PVA). This is a commonly-used mounting medium (refrac-

tive index 1.39) for permanent mounts (Box 2.8). It can be used as a solvent for
stains and as it hardens by heating, it can be used in a combined tissue treat-
ment including staining, conservation and mounting.
BOX 2.8. Preparat ion of a Polyvinylalcohol Solution

Polyvinyl alcohol 1.66g
Distilled water lOml
Lactic acid lOml
Glycerol ImI
The polyvinyl alcohol should be added very slowly to the water (magnetic
stirrer) and it dissolves in 1-2 h. Lactic acid is then added (vigorous stirring is
necessary), followed by glycerol.
Lactoglycerol. Another mounting medium for extended conservation is 50%

glycerol or 20% lactic acid. The cover glass should be sealed with nail varnish.
Different types of mineral oiI can be used if non-aqueous mounting media
are preferred, for example on desiccated or hydrophobic surfaces. Mineral oiI is
useful especially for plant tissues containing high amounts of pectin substances
or polysaccharide slime or with large water containing cells. Warning: mineral
oiI may affect flavonoid cell components.
Clearing, mounting, and staining may be combined in one procedure.
2.3.8 Procedures for Transmission Electron Microscopy
The preparation for transmission electron microscopy involves aldehyde

fixation; dehydration using ethanol or acetone, and embedding in resins espe-
cially suitable for electron microscopy (epoxy resins or certain acrylic resins,
such as the Lowicryls or the London Resins). Generally, the methods are similar
to what is described for LM, but a number of modifications are necessary. The
samples must be smaller. The values of pH and osmolarity of the solutions are
much more critical. If formaldehyde is used as a fixative it should be freshly pre-
pared from paraformaldehyde by slowly dissolving (in fume hood) the neces-
sary amount of paraform in buffer. Use a magnetic stirrer dur ing care fuI heating
until the solution looks clear. Certain postfixation procedures (staining) are
often required; a 1-2% osmium tetroxide solution is usually in the same buffer
as for fixation. (NB! OS04 is extremely hazardous to human health). An ultra-
microtome is needed for making sections in the range 60-70 nm. Preparation
methods for TEM are described in Robards and Wilson (1993).
2.3.9 Procedures for Scanning Electron Microscopy
Scanning electron microscopy is used for studies of surfaces, including cut sur-
faces. The material may be chemically fixed or cryo fixed.
Chemical fixation may be done with either FAA or glutaraldehyde with or
without postfixation with OS04. Chemical fixation should be followed by"crit-
ical point drying". The material is pas sed through an ascending acetone series
and in the pressure chamber of the CPD-apparatus exchanged with liquid
carbon dioxide. By elevating temperature the pressure is raised and when the
critical point of CO 2 has been reached it passes from liquid to the gaseous phase.
The pressure is slowly released and the material coated with gold or gold/pal-
ladium in a sputter coater prior to observation.
For cryo fixation the material is immersed in nitrogen slush prior to trans-
fer into the cooled microscope (-195°C). After sublimation to remove con-
densed ice from the surface, the specimen is sputtered with gold prior to
investigation. Both fixation methods are suitable for the study of root hair mor-
phology and ultrastructure.
Cryoscanning electron microsocpy has been used to observe the contents
of cortical intercellular space in roots (Watt et al. 1996) in studies concerning
the influence of soil water contents on deposits in root tissue.
2.4 Root-Soil Boundary
2.4.1 Root-Soillnterface in Thin Sections
Roots modify biological activity, structure, water status and mineral constitu-
tion of the surrounding soil. These changes can be observed by in situ micro-
scopic examinat ion of the root-soil boundary. For this purpose, thin sections of
soil with roots in place are needed (Jenny and Grossenbacher 1963; Lund
and Beals 1965; Fitzpatrick 1990). Soil blocks for preparing for thin sections in
situ can be obtained under field conditions by freezing the blocks with liquid
nitrogen. Under laboratory conditions, plant roots can be grown in small tubes
filled with soil. Undisturbed soil blocks are taken out of the tubes and are pas sed
through procedures of fixation, dehydration, embedding and polymerisation
in polyester re sin. These procedures are described in Sections 2.3.4 and 2.3.5,
but Tippkotter et al. (1986) provide very useful information on the procedures
for preparation of thin sections for biological studies. Thin sections of desired
thickness are prepared by grinding or by cutting the blocks with a diamond
saw or microtomes, depending on the details to be observed under the micro-
scope. Use of fluorescence microscopy and staining of the thin sections with
acridine orange (1:1000 in 10% HCI) improve brightness and contrast between
soil and plant material (Altemiiller and van Vliet-Lanoe 1990). Blue light exci-
tation exhibits cell walls of the root tissue as green or yellowish green. Soil mate-
rial becomes orange to reddish, hence contrasts well with the root tissues.
Microorganisms and fungi generally show green colours and are discernible in
contact with soil particles. However, they may be invisible if they are attached
to root cell walls. Trolldenier (1965) also used acridine orange in rhizosphere
studies for producing good quality colour micrographs of bacteria on root
hairs. Many of these thin section methods, however, provide mainly qualitative
information, but they allow in situ observations of the processes occurring close
to roots.
Watteau et al. (1996) described methods to study rhizosphere microsites in

situ. The methods are based on electron energy Loss spectroscopy (Egerton
1986) and allow observations at the ultrastructure scale on:
1. Microaggregation,
2. Polysaccharides of bacterial or root origin,
3. Microporosity,
4. Organo-mineral associations (root/mineral, microorganism/mineral), and
5. Plant-microorganism interactions.
The limitation of these methods is the embedding of the sample in resin,
because it may lead to element migration and collapsing of clay. Use of a
carbon-rich resin may interfere with the detection of carbon in the sample.
Cryo-ultramicrotomy may help to overcome these problems. Cryo-scanning
electron microscopy also allows observation of undisturbed accumulation of
water close to roots taken from the field (McCully 1995).
2.4.2 Root Hairs
Root hairs extend effective root surface area and thereby increase root-soil
contact. The initiation and development of root hairs occurs due to cellular dif-
ferentiation in the root epidermis. In the majority of plant species, any proto-
derm cell has the potential to form a root hair. In a few plant species, root hair
formation occurs only in "trichoblasts". Peterson and Farquhar (1996) have
recently reviewed the cytological and functional aspects of root hairs.
2.4.2.1 Quantification
Root hairs have typically been measured using a microscope and occular
micrometer (Caradus 1979; F6hse and Jungk 1983; Lamont 1983; MacKay and
Barber 1984). For root hair studies, the roots should be disturbed as little as
possible. Root hairs are often measured on roots grown in nutrient solution
culture, because it allows sampling of roots with minimum damage of root
hairs. For studying root hairs of solution-grown roots, nutrient concentrations
close to those found in the soil solution should be used. For soil- or field-grown
roots, the roots bearing root hairs must be cleaned without damaging the root
hairs. Ultrasound treatment is useful for cleaning roots for root hair studies
(Gahoonia and Nielsen 1997; Box 2.9).
BOX 2 .9. Sampling and Cleaning of Soil-Grown Roots

for Root Hair Studies:
1. Take soil cores with intact roots using a sharp knife or auger,
2. Immerse the soil cores in water overnight in the dark at SoC,
3. Remove the floating roots carefully e.g. using a (kitchen) sieve,
4. Place the roots in an ultrasound water bath (e.g. Branson 5200)
and subject them to an ultrasound treatment (120W, 47kHz ) for
S- lOmin. Intermittent treatment by switching off the ultrasound bath
for few seconds and then switching it on aga in facilitates cleaning of
roots.
The treatment with ultrasound removes the soil particles without damaging the
root hairs and leaves roots with root hairs comparable to those grown in nutri-
ent solution culture (Fig. 2.2), facilitating measurements of root hair parame-
ters using image analysis as described below. Roots with longer and denser root
hairs may have to be treated for longer with ultrasound.
Depending on the size of sampled root system, root hairs can be quantified
on whole or sub-sampled root systems. To make sub samples, the root system
is laid out in a film of water in a shallow plastic tray. The length of the root is
measured from the root to shoot junction to the longest root tip, and a mark
placed midway down the root. Excess water is blotted from the tray, and a ca.
l-cm strip of root is cut across the midway line using a scalpel and ruler. The
top and bottom portions of the root are removed and placed into bags for dry
weight determination. The l-cm segments are re-floated in water, and ten seg-
ments are randomly chosen for root hair studies. The surplus segments are
placed with the rest of the root, as are the ten segments after they have been
analysed for root hair length and number.
Root hairs and root parameters can be measured on the roots grown in
nutrient solution (Care 1995) or on roots washed out ofthe soil (Gahoonia and
Nielsen 1997) using image analysis. A typical image analysis system for root hair
studies consists of a microscope connected to a video camera, which is further
interfaced with a computer containing appropriate software and cards. The
quality of the optical system is critical for image clarity, as there is loss of image
resolution from the microscope to the computer. The combinat ion of micro-
scope and image analysis package can be critic al and should be evaluated
beforehand, if possible. Though other combinations may be equally good, Care
(1995) and Gahoonia et al. (1997) found that a Leica Wild M8 stereo microscope
(Leica AG, Heerbrugg, Switzerland) to be the most appropriate when coupled
with the MD30™ image analysis software package (Leading Edge Pty. Ud.,
Australia). Each image requires a minimum of 0.2 Mb computer memory.
Fig.2.2. Root hairs of two spring barley cultivars, Salka and Zita. A Root hairs on roots of Salka
grown in nutrient solution culture. BRoot hairs on roots of Salka grown in a field experiment.
Root hairs were extracted from the soi! using ultrasound treatment (120 W, 47 kHz). CRoot hairs
on roots of Zita grown in nutrient solution culture. D Root hairs on roots of Zita grown in a
field experiment. Root hairs were extracted from the soil using ultrasound treatment (120 W,
47kHz). (Photos by Tara S Gahoonia)
This should be kept in mind when planning experiments since cap turing images
of entire root systems may require a large data storage system.
To make root hairs clearly visible, appropriate magnification and illumina-
tion is needed. The best magnification must be determined for each particular
case, and then all the root hair images captured at this magnification. The best
illumination for root hair resolution is achieved by dark phase transmitted light.
The image analysis system must be calibrated using a micrometer (usually sup-
plied with the microscope) at the same magnification at which root hair images
are captured. The calibration procedure is normally described in the accompa-
nying user manuals.
To determine the length of the root hairs, an image is recalled on the com-
puter screen and individual root hairs are selected and highlighted, using an
"anchor and drop" process with the computer mouse. As each root hair is high-
lighted, its length in pixels or millimetres (depending upon the calibration scale
chosen) and also the number of root hairs is recorded in a data file. An root
hairs visible on each image are measured. The diameter of the root is also mea-
sured. The stored data files can be imported into an Excel spreadsheet.
To calculate the number of root hairs on the fun root hair cylinder, it is nec-
essary to determine the depth of field of the microscope (Box 2.10) and the
diameter of the root because on thin roots a higher proportion of root hairs
will be seen than on thicker roots. It should be noted that changing the micro-
scope magnification and/or the aperture setting will alter the depth of field,
hence an measurements should be done at the same settings that the depth of
field is calculated at. To maintain the focus of as many root hairs as possible,
the centre of focus should be close to the longitudinal axis of the root, and at
the widest part of the root.
BOX 2.10. Measuring Depth of F ield of a Microscope:

1. Place a vernier sca1e micrometer on the focus arm of the microscope,
2. Focus the microscope on root hairs at the magnification of interest,
3. Move the focus up or down and measure (on the vernier scale) how
far it moves before the object is no longer in focus. The distance
moved is equal to the depth of field.
The number of hairs in the fun cylinder is calculated as:
number of hairs observed on unit root x diameter of root

depth of the field
If a root map is done and the total root length measured (for methods see Chap.
10) and the length of root with root hairs is known, then the contribution of
root hairs to total root surface area can be ca1culated. This is done by using the
totallength of root hairs (e.g. by multiplying mean root hair length by the total
number of root hairs) in the equation for the surface area of a cylinder as 2m:l,
where r is the root hair radius and 1is the totallength of root hairs.
2.4.2.2 Vitality
Root hairs are often subjected to biotic and abiotic stresses that affect their
vitality and activity. Various stains (Table 2.4) can be used to test the viability
of root hairs. Oprisko et al. (1990) compared the effectiveness of some stains
for assessing root hair viability and reported that Evan's blue (0.5 g/l) gave great-
est precis ion and least variability. Evan's blue performs basically on the reten-
tion or loss of semi-permeable properties of the plasmalemma (Gaff and
Okong'o-ogola 1971). Live root hairs remain colourless because of exclus ion of
Evan's blue and the dead root hairs are permeable and turn blue. To obtain dead
root hairs (control), the root segments can be boiled in water for 10 min (Fischer
et al. 1985).
2.4.2.3 Nutr;ent Uptake v;a Root Ha;rs
The role of root hairs in nutrient uptake can be assessed with microelectrodes.
Recent work (Jones et al. 1995) using a vibrating microelectrode with Limno-
bium root hairs has shown that Ca2+ is taken up at the root hair tip, and K+ is
taken up over the whole root hair surface. This highlights the importance of
root hair number for Ca2+ and root hair length for K+ uptake.
Gahoonia and Nielsen (1998) described a direct method to measure uptake
of soil phosphorus via root hairs. Only root hairs penetrated a tightly stretched
nylon screen (53 Jlm) glued to the bottom of a PVC tube. The penetrating root
hairs grew for 2 and 4 days in soillabelled with radioisotope phosphorus tracer
32p (185 kBqg-l dry soil) in another PVC tube. Transparent plastic rings of thick-
ness ranging from 0.25 mm to 2.0 mm were inserted between the two PVC tubes.
This provided slit width for microscopic observations in situ, which confirmed
that only root hairs were growing into the 32p labelled soil. The uptake of 32p
from soil via the root hairs was quantified by measuring activity of 32p in the
plant shoot C2p uptake only via root hairs). In some cases, no rings were
inserted between the two PVC tubes so that both root hairs and root surface
could participate in 32p uptake from the labelled soil (total 32p uptake). This
study provided direct evidence on the substantial participation of root hairs in
uptake of phosphorus from soil.
2.4.2.3 Root Ha;rs ;n Nodu/at;on Process
Root hairs are the primary sites of infection with Rhizobium in nitrogen fixa-
tion of legumes. The infection process follows curling and deformation of root
hairs. The rate of root hair deformation can be used to compare the activities
of different Nod factors (nodulation genes of Rhizobium). Heidstra et al. (1994)
adapted the method of Bhuvaneswari and Solheim (1985) to assess root hair
deformation induced by different Nod factors. To perform the assay, sterilised
seeds are germinated for 2 days and transferred to Fâhraeus slides, modified as
small trays, containing 1 mI of Făhraeus medium (Fâhraeus 1957). AlI roots are
examined under a microscope, and those with already deformed root hairs are
discarded. The 1 mI medium is replaced by medium containing Nod factor and
the slides are incubated at 22°e. After 3 h (when maximum deformation

occurs), the deformed root hairs are scored under a microscope as O, 1 and 2,
corresponding to 0-20%, 20-60% or >60% deformed root hairs respectively.
The reliability of this assay lies in the fact that over 90% of the plants, when
tested with different concentrations (ranging from 10-7 to 10-13 M) of a Nod
factor, showed almost identical rates of root hair deformation. Based on these
observations, Heidstra et al. (I994) suggested that by applying dilution series of
Nod Factors, the root hair deformation could be used as a semi-quantitative
assay of activity of Nod factors.
2.4.3 Mycorrhiza
Mycorrhizas are symbiotic associations between plant roots and certain fungi.
They are almost universal, with most vascular plants forming a mycorrhizal
symbiosis with a fungal partner. It is widely recognized that the association nor-
mally brings mutual benefits to partners; the plant receives mineral nutrients
and the fungus receives carbon compounds derived from photosynthesis. Dif-
ferent types of association, involving different plant and fungal groups, have
been identified and organised into the following four major categories:
1. Arbuscular mycorrhizas (AM) - the most widespread, involving many agri-
culturally and ecologically important plant species;
2. Ectomycorrhizas (EM) - involving most forest tree species;
3. Ericoid mycorrhizas - confined to plants in the Ericales;
4. Orchid mycorrhizas - confined to orchidaceous plants.
These divisions may be seen as arbitrary. For example, it is now evident that
although such divisions are useful, in practice many variations exist (e.g. ecto-
mycorrhizas can also be arbutoid, monotropoid and ectendo types: see Smith
and Read [1997] for more details). Individual mycorrhizal associations thus
need to be treated as specific cases and clearly characterized. It is therefore
important, when making measurements of mycorrhizas, to identify the possi-
bIe nature of the association and not to make assumptions based on any pre-
conceptions. Therefore, it is the aim of the methods described here (which deal
only with AM and EM) to provide guidance. Because each circumstance will be
different they are not prescriptive protocols but should be seen as starting
points, being revised where appropriate. Finally, it should be mentioned that
these methods provide a single-point assessment of a mycorrhizal symbiosis
and any assumptions made on the basis of data collected should take into
account both spatial and temporal factors.
It is not possible to see root colonisation with vesicular-arbuscular mycor-
rhizae (VAM) fungi without staining and microscopic observation. However, the
staining process does not require expensive instrumentation.
After washing and acidification of the root sample, fungal structures in the
roots are stained with trypan blue, acid fuchsin, chiorazole black or other com-
pounds (Brundrett et al. 1984; Grace and Stribley 1991). Staining of active fungal
parts only may also be possible (Schaffer and Peterson 1993). After staining, in
most cases the percentage of the total root length colonised by mycorrhizal
fungi is determined (Giovannetti and Mosse 1980).
Spores of mycorrhizal fungi can be isolated from soil by a wet sieving and
decanting technique (Box 2.12). AIso, the length of the external mycorrhizal
mycelium in the soil can be determined after washing the hyphae from soil and
subsequent staining (HameI et al. 1990). In the future, DNA-based methods are
expected to facilitate characterisation and distribution of different fungal pop-
ulations in the soil and in the roots.
Before designing a protocol to assess mycorrhizas it is important to
establish the hypothesis being tested. Is the aim, for example, to establish the
presence or absence of a particular type of mycorrhiza, the extent of the colo ni-
sation or perhaps the function? Each will require a different approach and
procedure.
In this section, only a selection of the major methods is provided. Readers
are encouraged to further read the papers referenced or for a more in-depth
account of methods are referred to Ingleby et al. (1990), Norris et al. (1991a, b),
Sieverding (1991) and Brundrett et al. (1996). Marschner (1995) and Smith and
Read (1997) give a summary of the role of mycorrhizas in plant growth.
2.4.3.1 Arbuscu/ar Mycorrhizas
Arbuscular mycorrhizal fungi (AMF) of the order Glamales colonise and

form mycorrhizas in the roots of a wide variety of plants. They are common
in both annuals and perennials and have a major agricultural and ecological
significance. The mycorrhizas have internal fungal structures (hyphae and
often arbuscles and vesicles) and external to the root, hyphae (often forming
an extensive extern al mycelium) and spores growing from the external
mycelium. In some cases spores can also be found growing within the root.
Modifications to the morphology of the root system can also occur through
effects of colonisation on branching (Hooker et al. 1992; Berta et al. 1995;
for methods see Hooker et al. 1998). Here we confine methods applied to
study internal and extern al fungal structures. Relatively simple methods
can be used to visualize these but quantification is, in both cases, more
difficult.
External Mycelium. In the simplest case, Graham et al. (1982) proposed that the
weight of a plant-soil root baU would be proportional to the quantity of AMF
hyphae that bind the unit together. They thus estimated the relative quantity of
AMF hyphae by weighing the plant roots with and without adhering soil.
However, this view has subsequent1y been considered too simple as many
other factors are also involved in the attachment of soil to plant roots (Kough
and Linderman 1986). The determination of chitin content has also been
used (Bethlenfalvay and Ames 1987). However, because chitin is also present in
other fungi and insects, the use of the technique is limited to only very specific
controlled situations where they can be eliminated. Schiiepp et al. (1987) esti-
mated the extension of AMF hyphae in soil by using a membrane which pre-
vents the passage of roots but allows hyphae to pass. By using receiver plants,
the roots of which became colonized by the hyphae, it was possible for them to
calculate the distance that hyphae travel through the soil as well as the speed of
growth. However, it is likely to be difficult to accurately calculate the latter
because of the sensitivity of the technique to time for colonization of the
receiver root.
Direct methods involve three steps: extraction, detection and measurement.
The first, extracting the hyphae from the soil is arguably the most difficult. A
sample of soil is normally first suspended in solution with water. Sieving
(250 Jlm) is then used to separate the hyphae from the soil particles (Abbot et
al. 1984). Although the concept is simple and straightforward, it is difficult to
apply it in practice and recovery can be poor and uncertain. To increase recov-
ery, Vilarino et al. (1993) immersed a rotating wire in the suspended soil. The
hyphae present became attached to the wire and were subsequently washed off
with water.
Hyphae of AMF must be distinguished from other hyphae in the extracted
sample. The diameter of hyphae has been the most often used criterion for
detection (Abbot and Robson 1985; Bethlenfalvay and Ames 1987). The varia-
tion in morphology of AMF hyphae makes distinguishing AMF on this basis
very difficult and in some cases impossible. Thus, although it may be possible
to distinguish the hyphae of AMF from other fungal hyphae in some very spe-
cific situations, in most it is not and data obtained using such methods should
be interpreted cautiously. Hyphae are often, for example, stained with trypan
blue or acid fuschin before identification. Staining also helps in the measure-
ment of hyphae length with the gridline intersect method (Giovannetti and
Mosse 1980). The future hope lies in the possibility of using immunological or
molecular techniques to routinely distinguish between AMF and non-AMF
hyphae in samples extracted from soils.
Internal Structures. Plants vary in the extent to which it is possible to visualize

AMF in their roots. In some plants such as cucumber the transparency of
the roots makes it possible to observe structures using only a dissection
microscope, without preparation or staining. However, in other species, e.g. tree
roots, it is often very difficult to observe AMF structures within the root, essen-
tially due to the deposition of tannins and the poor transparency of cell walls.
There is thus no prescriptive method for preparing roots in order to observe
AMF.
The simplest methods are mechanically teasing roots open (Redhead 1977)
or staining by immersing roots in solutions of cotton blue (Gallaud 1905) meth-
ylene blue (Daft and Nicolson 1966) or acid fuschin (O'Brien and McNaughton
1928). Improvements were subsequently made to staining by using lactophenol
(Mosse and Hepper 1975), extracting phenolic compounds from roots using
potassium hydroxide (Bevege 1968) and bleaching in hydrogen peroxide
(Philips and Hayman 1970). For many years the most common method applied
was that described by Philips and Hayman (1970), often modified by removing
phenol from the procedures (Kormanik and McGraw 1982). Nowadays the most
commonly used method is a modification of the Philips and Hayman technique
by Koske and Gemma (1989). The procedure essentially involves six steps (Box
2.11). It is best carried out on fresh material, although roots can be stored in
70% ethanol for short periods.
BOX 2.11. Preparation o f Roots for Observing AMF Structure:

1. Heat roots in 2.5% KOH for 3min at 121 °C or 10- 30min at 90°C;
2. Rinse roots in water;
3. Bleach roots, if necessary, for 10 to 30 min in alkaline H20 2 and then
rinse in water;
4. Soak roots in 1 % HCI for 1 to 24 h;
5. Stain roots in a solution of acidic glycerol and trypan blue for 3 min
at 121 °C or 10 to 30min at 90°C;
6. Destain roots and store in acidic glycerol.
Most laboratories will further modify and optimize this technique for specific
situations but it is often useful to begin with this protocol, inspecting roots
at each stage using a microscope. When making assessments, even when
optimized, it is essential to inspect stained roots very closely. A dissection
microscope is frequently used but experience shows that the use of a higher
power compound microscope is frequently necessary. This is partic-
ularly important because not all parts of the AMF are always stained, even when
the method is applied very carefully. Another problem is that it is extremely
difficult to visually differentiate between the hyphae of AMF and other fungi.
Sometimes differentiation is possible because of the presence of arbuscles and
or vesicles. Immunological techniques are available which will permit dif-
ferentiation (Hahn et al. 1994) but they require much preliminary work and
are not appropriate for the screening of large numbers of roots. Quantification
of AMF is obtained by estimation of percentage occupancy of roots, usually
using a gridline intersect method (Giovannetti and Mosse 1980). Measures
of colonization by AMF have usually been made only with alI AMF structures,
i.e. grouped, with no separation into the different structures. However, there
is now a consensus amongst many researchers that a breakdown of coloniza-
tion into different structures is likely to be informative. More detailed studies
of histology are possible folIowing sectioning (Gianinazzi and Gianinazzi-
Pearson 1992). Procedures again need careful adaptation to the tissue being
studied. FinalIy, it should be remembered that although the above methods
inform on the presence of AMF structures they do not identify function.
This may be determined, of course, by bioassay but these data do not provide
any information on localization. Techniques are now available to detect func-
tional enzyme activity, e.g. alkaline phosphatase (Tisserant et al. 1993) and the
more widespread use of this and similar techniques is likely to provide valuable
data.
Spores. The quantification of spores extracted from the soil can be a useful
measure. However, it should be emphasized that extrapolations between spore
numbers (relatively easy to obtain) and the extent of colonisation by AMF or
the extern al mycelium cannot be made, because they are likely to be unreliable
and misleading.
The first step is to obtain a suitable soil sample. This should be taken from
the field using a spade or soil corer, sterilized using 70% ethanol and rinsed in
water. Soil from a pot(s) should be mixed before sampling. The precise number
of samples will depend upon individual circumstances. However, it should be
noted that the distribution of spores can be highly heterogeneous and that the
design of an appropriate sampling regime should be a priority. The final sample
size should be between 50 and 100 g. It is best to extract spores from fresh soils
but dried soils can also be used. A procedure for quantification of spores is given
in Box 2.12.
2.4.3.2 Ectomycorrhizas
Ectomycorrhizal fungi (EMF) colonise the roots of many important forest trees.
Ectomycorrhizas are thus widely distributed in forest ecosystems. Many differ-
ent morphological features are distinguishable. Undoubtedly, the most distinct
structures are the fruiting bodies, which are frequently observable under forest
trees. These structures have been widely studied (see Brundrett et al. 1996 for
methods) for accurate identification of genera and/or species. Here we limit the
description of methods to observation of structures within the root and on its
surface.
2 Anatomyand Histology of Roots and Root-Soil Boundary 67
BOX 2.12. Quantification of Spores:

1. Remove stones and large woody roots from the sample by hand. We
do not recommend using a 2 -mm sieve, as this will also remove fine
roots which are likely to have spores attached.
2. Add the soi! (50- 100g) to approximately 91 distilled water in a
suitable container (a bucket is ideal) and stir by hand; we
recommend the use of plastic gloves. Ensure that alI aggregates are
broken apart by pressing between fingers. This is an important part
of the process and is likely to take at least 10 min.
3. Stir the soiUwater solution well and then following agitation pour
through a ser ies of 750, 200, 100 and 50 ţlm sieves.
4. Collect what is retained on each sieve by back washing with distilled
water (plastic bottles normally used for distilled water in the
laboratory are ideal for this) and pour into a centrifuge tube. Make
up to 50ml volume with distilled water. Prepare one tube for
material retained on each sieve i.e. 750,200, 100 and 50)lm.
5. Carry out steps 6 onwards to each tube,
6. Centrifuge at 2500 rpm (swing out head) for 5 min,
7. Discard the supernatant and any organic material (which should be
floating on the surface),
8. Re-suspend the pellet in 40% sucrose and then centrifuge at
2500rpm for approximately 60s (the precise time will need to be
determined but this is a good guide),
9. Pour the supernatant on to a 50 ţlm filter and remove the sucrose
(this is particularly important if you want to use the spores to
inoculate plants) by washing through a large volume (approximately
11) distilled water,
10. Back wash as before to remove spores from the sieve and collect onto
filter paper (we use nitrocellulose membranes, gridded when we
wish to count),
Il. Carefully inspect sieves at alI stages of the procedure using
a stereomicroscope to ensure no spores are lost. lf necessary the
procedures should be modified.
12. Spores should now be examined under a stereomicroscope and
counted, if appropriate. If the aim is to provide identification then
extreme care must be taken. Excellent advice can be obtained from
the website of La Banque Europeenne des Glomales (BEG) at
http://www.ukc.ac. uklbeg/ or the International Culture Collection
of Arbuscular and Vesicular Arbuscular Mycorrhizal Fungi at
http://invam.caf.wvu.edu/
Ectomycorrhizal fungi frequently cause quite dramatic changes to the mor-

phology of colonised roots. Therefore, one of the most common measurements
of EMF is to count the number of mycorrhizal tips or the length of mycorrhizal
roots in a sample. This is relatively easy to do because the roots are often exten-
sively modified, highly branched and are thicker than non-colonised roots.
However, it is often necessary to spend some time looking closely at the mor-
phology of the root system (examining roots under the microscope) before
accurate measurements can be made. Different EMF can result in very different
modifications, so some experience is necessary for making accurate measure-
ments. Roots are then measured using the gridline intersect method and the
number or length of mycorrhizal tips counted.
Isolated tips (which should be fresh, but can be preserved in FAA or gluter-
aldehyde) can then be mounted on slides in 0.1% cotton blue (w/v) in 10% lac-
tophenol (w/v), for general observations on the extent of colonisation.
Examination of the Hartig Net in particular often provides useful information
as to the identity of the fungus involved (for further details see Ingleby et al.
1990).
Sections of ectomycorrhizal root tips can display mantle layers with dif-
ferent kinds of auto-fluorescence. Thickness of sections and required staining
depend upon the information needed from study. Median longitudinal sections
of ectomycorrhiza are more informative than cross-sections (Agerer 1991)
because:
1. They include the ectomycorrhizal mantle from tip,
2. They clearly show the position of Hartig net initiation zone; and where it is
not fulIy developed,
3. The orientation of the cortical cell is oblique, in cross-sections the two
neighbouring celIs are cut,
4. The shape of the cortical celIs enveloped by a Hartig net is more character-
istic in longitudinal sections than in cross sections.
Normarski differential interference contrast microscopy also gives good results

in anatomical characterization of ectomycorrhiza. It shows definite mantle or
rhizomorph layers without interference from hyphal layers beneath or above
the focused level.
Sections of ectomycorrhiza embedded in LR white resin are reported
to provide good results because of its very low viscosity, providing good
infiltration of both symbionts (Massicotte et al. 1987). Because of its hydrophilic
nature, it can be used in most histochemical procedures. The results with
transmission electron microscopy may not be always satisfactory (Peterson
1991). One of the most useful techniques for demonstrating the Hartig net and
mantle hyphae in LR white-embedded ectomycorrhiza is a simple fluorescence
technique modified from Culling (1974). Observation with Normarski
differential interference contrast optics gives particularly clear images of the

Hartig net.
Nuclear magnetic resonance spectroscopy (NMR) can be applied for spe-
cialized studies of carbon, nitrogen and phosphorus metabolism by mycor-
rhizal fungi by using 13e, 15N and 31 P respectively. Martin (1991) has summarised
the prospects of using NMR in ectomycorrhizal fungi. NMR allows the identi-
fication and quantification of multiple metabolites (amino acids, carbohydrates
and organic acids) from the same spectrum and because of non-invasiveness,
serial measurements of intact mycelium and roots are possible. High opera-
tional costs limit its use for routine investigations.
References
Abbot LK, Robson AD (1985) Formation of external hyphae in soH by four species ofvesicular-
arbuscular mycorrhizal fungi. New Phytol 99: 245-255
Abbot LK, Robson AD, De Boer G (1984) The effect of phosphorus on the formation of hyphae
in soH by the vesicular-arbuscular mycorrhizal fungus Glomus fasciculatum. New Phytol
97:437-446
Agerer R (1991) Characterization of ectomycorrhiza. In: Norris JR, Read DT, Varma AK (eds)
Methods in microbiology voI 23. Academic Press, London, pp 25-75
Altemiiller HJ, van Vliet-Lanoe B (1990) SoH thin section fluorescence microscopy. In:
Douglas LA (ed) SoH micromorphology. Eisevier, Amsterdam
Baluska F, Parker JS, Barlow PW (1992) Specific patterns of cortical and endoplasmatic micro-
tubules associated with ceH growth and tissue differentiation in roots of maize (Zea
mays 1.) J CeH Sci 103: 191-200
Berta G, Trotta A, Fusconi A, Hooker JE, Munro M, Atkinson D, Giovannetti M, Morini S,
Fortuna P, Tisserant B, Gianinazzi-Pearson V, Gianinazzi S (1995) Arbuscular mycorrhizal
induced changes to plant growth and root system morphology in Prunus cerasifera. Tree
Physiol15: 281-294
Bethlenfalvay GJ, Ames RN (1987) Comparison of two methods for quantifying extraradial
mycelium of vesicular mycorrhizal fungi. SoH Sci Soc Am J 51: 834-837
Bevege DI (1968) A rapid technique for clearing tannins and staining intact roots for detection
of mycorrhizas caused by Endogone spp. and some records of infection in Austral-asian
plants. Trans Br Mycol Soc 51: 808-810
Bhuvaneswari TV, Solheim B (1985) Root hair deformation in the white cloverlRhizobium tri-
folii symbiosis. Physiol Plant 63: 25-34
Blancaflor EB, Hasenstein KH (1993) Organisation of cortical microtubules in graviresponding
maize roots. Planta 191: 231-237
Blancaflor EB, Hasenstein KH (1997) The organization of the actin cytoskeleton in vertical and
graviresponding primary roots of maize. Plant Physiol113: 1447-1455
Boyde A, Maconnachie E (1981) Morphological correlations with dimensional change during
SEM specimen preparation. Scanning Electron Microsc IV: 27-34
Brundrett M, Bougher N, Dell B, Grove T, Malajczuk N (1996) Working with mycorrhizas in
forestry and agriculture. Australian Centre for International Agricultural Research,
Canberra. ACIAR Monograph 32,374 pp
Brundrett MC, Piche Y, Peterson RL (1984) A new method for observing the morphology of
vesicular-arbuscular mycorrhizae. Can J Bot 62: 2128-2134
Brundrett MC, Enstone DE, Peterson CA (1988) A berberine-aniline blue staining procedure for
suberin, lignin and callose in plant tissue. Protoplasma 146: 133-142
Caradus JR (1979) Selection for root hair length in white clover (Trifolium repens L). Euphytica
28:489-494
Care D (1995) The effect of aluminium concentration on root hairs in white clover (Trifolium
repens 1.). Plant Soi!I7l: 159-162
Carlson H, Stenram U, Gustafsson M, Jansson HB (1991) Electron microscopy of barley root
infection by the fungal pathogen Bipolaris sorokiniana. Can J Bot 69 (12): 2724-2731
Clark G (ed) (1981) Staining procedures, 4th edn. Williams and Wilkins, Baltimore, 512 pp
Culling DFA (1974) Modern microscopy: elementary theory and practice. Butterworths, London
Daft MJ, Nicolson TH (1966) Effect of Endogone mycorrhiza on plant growth. New Phytol 65:
343-350
Dhingra OD, Sinclair JB (1995) Basic plant pathology methods, 2nd edn. CRC, Boca Raton, 434
pp
Edwards HH, Yeh YY, Tamowski BI, Schonboum GR (1992) Acetonitrile as a substitute for
ethanol/propylene oxide in tissue processing for transmission electron microscopy: com-
parison of fine structure and lipid solubility in mouse liver, kidney and intestine. Microsc
Res Technique 21: 39-50
Egerton RF (1986) Electron energy loss spectroscopy in the electron microscope. Plenum Press,
New York
Esau K (1965) Plant anatomy, 2nd edn. Wiley, New York
Esau K (1977) Anatomy of seed plants, 2nd edn. Wi!ey, New York
Fahn A (1990) Plant anatomy, 4th edn. Pergamon Press, Oxford
Făhraeus G (1957) The infection of clover root hairs by nodule bacteria studied by a simple glass
technique. J Genet Microbiol16: 374-381
Fischer JMC, Peterson CA, Bols NC (1985) A new fiuorescence test for cell vitality using Calco-
fiuor white M2R. Stain Technol60: 69-79
Fitzpatrick EA (1990) Roots in thin sections of soi!s. In: Douglas LA (ed) Soil micromorphol-
ogy, voi 19. Elsevier, Amsterdam, pp 9-23
Fohse D, Jungk A (1983) Infiuence of phosphate and nitrate supply on root hair formation of
rape, spinach and tomato plant. Plant Soil 74: 359-368
Gaff DF, Okong'o-ogola O (197l) The use of non-permeating pigments for testing the survival
of cells. J Exp Bot 22: 756-758
Gahan PB (1984) Plant histochemistry and cytochemistry. An introduction. Academic Press,
London, 301 pp
Gahoonia TS, Nielsen NE (1997) Variation in root hairs of barley cultivars doubled soi! phos-
phorus uptake. Euphytica 98 (3): 177-182
Gahoonia TS, Nielsen NE (1998) Direct evidence on participation of root hairs in phosphorus
C2P) uptake from soi!. Plant Soil198: 147-152
Gahoonia TS, Care D, Nielsen NE (1997) Root hairs and acquisition of phosphorus by wheat and
barley cultivars. Plant Soil191: 181-188
Gallaud 1 (1905) Etudes sur les mycorrhizes endotrophes. Rev Genet Bot XVII: 123-127
Gardner RO (1975) An overview of botanical clearing technique. Stain Technol50 (2): 99-lO5
Gerlach D (1984) Botanische Mikrotechnik, 3rd edn. Thieme, Stuttgart, 311 pp
Gerrits PO, Zuideveld R (1983) The influence of dehydration media and catalyst systems
upon the enzyme activity of tissues embedded in 2-hydroxyethyl methacrylate. An
evaluat ion of the three dehydration media and two catalyst systems. Mikroskopie 40:
321-328
Gerrits PO, van Leeuwen MBM, Boon ME, Kok LP (1987) Floating on a water bath and mount-
ing glycol methacrylate and hydroxypropyl methacrylate sections influence final dimen-
sions. J Microsc 145 (1): 107-113
Gianinazzi S, Gianinazzi -Pearson V (1992) Cytology, histochemistry and immunocytochemistry
as tools for studying structure and function in endomycorrhiza. In: Methods in Microbi-
ology, voi 24. Academic Press, London, pp 109-139
Giovannetti M, Mosse B (1980) An evaluation of techniques for measuring vesicular arbuscular
mycorrhizal infection in roots. New Phytol 84: 489-500
Glauert AM (1975) Practical methods in electron microscopy. Fixation, dehydration and embed-
ding of biological specimens. North Hollandl American Elsevier, Amsterdam
Grace C, Stribley DP (1991) A safer procedure for routine staining of vesicular-arbuscular myc-
orrhizal fungi. Mycol Res 95: 1160-1162
Graham JH, Linderman RG, Menge JA (1982) Development of extern al hyphae by different iso-
lates of mycorrhizal Glomus spp. in relation to root colonizat ion and growth of Troyer cit-
range. New Phytol91: 183-189
Guttenberg Hv (1968) Der primare Bau der Angiospermenwurzel. Handbuch der Pflanzen-
anatomie VIII, 5. Gebriider Borntrager, Berlin
Hahn A, Gianinazzi-Pearson V, Hock B (1994) Characterisation of arbuscular mycorrhizal fungi
by immunochemical methods. In: Gianinazzi S, Schiiepp H (eds) Impact of arbuscular myc-
orrhizas on sustainable agriculture and natural ecosystems. Birkhauser, Basel, pp 25-39
Hamei C, Fyles H, Smith DL (1990) Measurements of development of endomycorrhizal
mycelium using three different vital stains. New Phytoll15: 297-302
Heidstra R, Geurts R, Franssen H, Spaink, van Kammen, Bisseling T (1994) Root hair defor-
mat ion activity of nodulation factors and their fate on Vicia sativa. Plant Physiol 105:
787-797
Herr JM Jr (1971) A new clearing-squash technique for the study of ovule development in
angiosperms. Am J Bot 58: 785-790
Hooker JE, Munro M, Atkinson D (1992) Vesicular- arbuscular fungi induced alterat ion in poplar
root system morphology. Plant Soil145: 207-214
Hooker JE, Berta G, Lingua G, Fusconi A, Sgorbati S (1998) Quantification of AMF induced mod-
ifications to root system architecture and longevity. In: Varma A (ed) Mycorrhizal methods.
Springier, Berlin Heidelberg New York
Ingleby K, Mason PA, Last FT, Fleming LV (1990) Identification of ectomycorrhizas Institute of
Terrestrial Ecology, Publ No 5. HMSO, London
Jenny H, Grossenbacher K (1963) Root-soil boundary zones as seen in the electron microscope.
Soil Sci Soc Am Proc 27: 273-277
Jensen WA (1962) Botanical histochemistry. WH Freeman, San Fransisco
Johansen DA (1940) Plant microtechnique. McGraw-Hill, New York
Jones DL, Shaff JE, Kochian L (1995) Role of calcium and other ions in directing root hair tip
growth in Limnobium stoloniferum. 1. Inhibition of tip growth by aluminium. Planta
197 (4): 672-680
Knox RB (1970) Freeze-sectioning of plant tissues. Stain Technol45: 265-272
Kormanik PP, McGraw AC (1982) Quantification of vesicular-arbuscular mycorrhizae in plant

roots. In: Schenck NC (ed) Methods and principles of mycorrhizal research. American Phy-
topathological Society, St Paul, Minnesota, pp 37-45
Koske RG, Gemma JN (1989) A modified procedure for staining roots to detect VA mycorrhizas.
Mycol Res 92: 486-505
Kough JL, Linderman RG (1986) Monitoring extra-matrical hyphae of a vesicular-arbuscular
mycorrhizal fungus with an immunofluorescence assay and the soil aggregation technique.
Soil Biol Biochem 18: 309-3l3
Kuck KH, Tiburzy R, Hănssler G, Reisener HJ (1981) Visualization of rust haustoria in wheat
leaves by using fluorochromes. Physiol Plant Pathol19: 439-441
Lamont B (1983) Root hair dimensions and surface/volume/weight ratios of roots with the aid
of scanning electron microscopy. Plant Soil 74: 149-152
Lillie RD (1977) H.J. Conn's biological stains, 9th edn. Williams and Wilkins, Baltimore, 692 pp
Lindauer R (1972) Die Technik des Handschnittes. Mikrokosmos 61: 144-151
Lund ZF, Beals HO (1965) A technique for making thin sections of soil with roots in place. Soil
Sci Soc Am Proc 29: 633-635
Luxova M, Ciamporova M (1989) Root structure. In: Kolek J, Kozinka V (eds) Physiology of the
Plant Root System. Kluwer Academic Publishers, Dordrecht, pp 31-81
Lyon H (ed) (1991) Theory and strategy in histochemistry. Springer, Berlin Heidelberg New York
Lyshede OB (1977) A method for removing starch from plant tissue with bacterial amylase.
Mikroskopie (Wien) 33: 241-245
Lyshede OB (1979) Effect ofbacterial amylase on the ultrastructure of potato tuber storage cells.
Mikroskopie 35: 314-318
MacKay AD, Barber SA (1984) Effect of soil moisture and phosphate level on root hair growth
of corn roots. Plant Soil86: 321-331
Marschner H (1995) Mineral nutrition of higher plants. Academic Press, London
Martin FM (1991) Nuclear magnetic resonance studies in ectomycorrhizal fungi. In: Norris JR,
Read DJ, Varma AK (eds) Methods in microbiology, voI 23. Academic Press, London,
pp 121-148
Massicotte HB, Melville LH, Peterson RL (1987) Scanning electron microscopy of ectomycor-
rhizae. Potentials and limitations. Scanning Microsc 1: 1439-1454
Mauseth JD (1988) Plant anatomy. Benjamin Cummings, Menlo Park
McCully M (1995) Water efflux from the surface of field grown grass roots. Observations by cryo-
scanning electron microscopy. Physiol Plant 95: 217-224
Mosse B, Hepper C (1975) Vesicular-arbuscular mycorrhizal infections in root organ cultures.
Physiol Plant Pathol 5: 215-223
Neergaard E de (1997) Methods in botanical histopathology. Danish Institute of Seed Pathology
for Developing Countries, Copenhagen, 216 pp
Norenburg JL, Barrett JM (1987) Steedman's wax embedment and de-embedment for combined
light and scanning electron microscopy. J Electron Microsc Technol6: 35-41
Norris JR, Read DJ, Varma AK (eds) (1991 a) Techniques for the study of mycorrhiza. In: Methods
in microbiology, voI 23. Academic Press, London
Norris JR, Read DJ, Varma AK (eds) (1991 b) Techniques for the study of mycorrhiza. In: Methods
in microbiology, voI 24. Academic Press, London
O'Brien DG, McNaughton EJ (1928) The endotrophic mycorrhiza of strawberries and its sig-
nificance. Research Bulletin 1. The West of Scotland College of Agriculture. Edinburgh,
35 pp
O'Brien TP, McCully M (1981) The study of plant structure. Principles and selected methods.
Termarcarphi Pty, Melbourne, Australia
O'Brien TP, von Teichman 1 (1974) Autoclaving as an aid in the clearing of plant specimens.
Stain Technol49: 175-176
Oprisko MI, Green RL, Beard JB, Gates CE (1990) Vital staining of root hairs in 12 warm season
perennial grasses. Crop Sci 30: 947-950
Peterson LR (1991) Histochemistry of ectomycorrhiza. In: Norris JR, Read DJ, Varma AK (eds)
Methods in microbiology, voi 23. Academic Press, London, pp 107-120
Peterson LR, Farquhar ML (1996) Root hairs: Specialized tubular cells extending root surfaces.
Bot Rev 62: 1-40
Philips JM, Hayman DS (1970) Improved procedures for cleaning roots and staining parasitic
and vesicular-arbuscular mycorrhizal fungi for rapid assessment of infection. Trans Br
Mycol Soc 55: 158-161
Purvis MJ, Collier DC, Walls D (1964) Laboratory techniques in botany. Butterworths, London
Redhead JF (1977) Endotrophic mycorrhizas in Nigeria: species ofthe Endogonaceae and their
distribution. Trans Br Mycol Soc 69: 275-280
Robards AW, Wilson AJ (eds) (1993) Procedures in electron microscopy. John Wi!ey, Chichester
Sass JE (1961) Botanical microtechnique. The Iowa State University Press, Ames
Schaffer GF, Peterson RL (1993) Modifications to clearing methods used in combination with
vital staining of roots colonized with vesicular-arbuscular mycorrhizal fungi. Mycorrhiza
4: 29-35
Schiiepp H, Miller DD, Bodman M (1987) A new technique for monitor ing hyphal growth of
vesicular-arbuscular mycorrhizal fungi through soi!. Trans Br Mycol Soc 89: 429-435
Sieverding E (1991) Vesicular-arbuscular mycorrhizal management in tropical ecosystems Tech-
nical Cooperation, Eschborn, Germany
Smith SE, Read DJ (1997) Mycorrhizal symbiosis. Academic Press, San Diego
Spurr AR (1969) A low-viscosity epoxy resin embedding medium for electron microscopy.
J Ultrastruct Res 26: 31-43
Strausbaugh CA, Murray TD (1989) Use of epidermal cell responses to evaluate resistance of
winter wheat cultivars to Pseudocer cosporella herpatrichoides. Phytopathology 79 (10):
1043-1047
Tippkiitter R, Ritz K, Darbyshire JF (1986) The preparation of thin sections for biological studies.
J Soi! Sci 37: 681-690
Tisserant B, Gianinazzi-Pearson V, Gianinazzi S, Gollotte A (1993) In planta histochemical stain-
ing of fungal alkaline phosphatase activity for analysis of efficient arbuscular mycorrhizal
infections. Mycol Res 97: 245-250
Trolldenier G (1965) Fluoreszenzmikroskopie in der Rhizosphărenforschung. ZEISS-Inf 56:
68-69
Vilarino A, Arines J, Schuepp H (1993) Extraction of vesicular-arbuscular mycorrhizal mycelium
from sand samples. Soil Biol Biochem 25: 99-100
Watt M, van der Weele CM, McCully ME, Canny MJ (1996) Effects on local variations in soi!
moisture on hydrophobic deposits and dye diffusion in corn roots. Bot Acta 109: 492-501
Watteau F, Villemin G, Mansot JL, Ghanbaja J, Touain F (1996) Localization and characterization
of brown cellular substances of beech roots by electron energy loss spectroscopy. Soil Biol
Biochem 28: l327-l332
CHAPTER 3
Control and Measurement of the Physical

Environment in Root Growth Experiments
W.R. Whalleyl, J. Lipiec 2, w. St~pniewski2,3, and F. Tardieu4
1 Silsoe Research Institute, Wrest Park, Silsoe, Bedford MK45 4HS, UK
2 Polish Academy of Sciences, Institute of Agrophysics, D08wiadczalna 4, 20-290 Lublin 27,
Poland
3 Technical University of Lublin, 40 Nadbystrzycka Str. 20-618 Lublin, Poland
4 INRA, Laboratoire d'Ecophysiologie, 34060 Montpellier, France
CONTENTS
3.1 Introduction 76
3.2 Laboratory Control of the Growth Environment 78

3.2.1 Control of Water Stress 78
3.2.1.1 Control of Water Supply 79
3.2.1.2 Control of SoH Matric Potential 80
3.2.1.3 The Use of Osmotic Solutions 81
3.2.1.4 Applying Water Stress to Split Root Systems 85
3.2.2 Control of Mechanical Stress 86
3.2.2.1 Controlling Mechanical Impedance at Non-Limiting
Stress Levels 86
3.2.2.2 Effect of Pot Size 88
3.2.2.3 Complete Impedance 88
3.2.2.4 Use of Split Roots in Relation to Mechanical Impedance 90
3.2.3 Oxygen Stress 91
3.2.3.1 Indicators of SoH Aeration Status 91
3.2.3.2 Controlling Oxygen Stress 93
3.2.3.3 Split Root Techniques in Relation to Oxygen Stress 94
3.3 Field Measurement of the Root Growth Environment 95

3.3.1 Field Assessment of Root Conditions 95
3.3.2 Mechanical Impedance 95
3.3.3 Characterising the Effect of Soil Structure on Root Growth 99
3.3.4 Field Measurement of Aeration Status 100
3.3.5 Measuring and Modifying Soil Temperature in the Field 102
3.3.6 Interpreting the Effects of Combinations of Stresses 103
3.4 Future Trends 105
References 106

76 W.R. Whalley et al.
3.1 Introduction
Analyses of the responses of root growth rate to the plant environment in the
last 10 years have greatly increased our understanding of the main mechanisms
which control root growth and development, and have led to the analysis of root
system architecture. However, we currently lack a global approach to plant
growth in varying environments, including growth of the root system. This
leaves a large, and nearly empty'ecological niche' for studies carried out at the
whole plant level, which analyse root system growth as a function of the envi-
ronment. This chapter provides guidance on how to manipulate the rooting
environment.
Mechanisms which control root elongation at the root tip level have been
analysed in detail by several groups. We have methods to analyse ceH division
rate and expansion spatiallywithin the growing zone of a root (Sharp et al. 1988;
Fraser et al. 1990), thereby allowing us to analyse the effect of environmental
conditions on these processes. Seminal roots are probably the tissues most
frequently studied for responses of cell elongation to temperature, water
deficit and salinity. These studies alllead to the conclusion that the mechanical
properties of ceH walls largely determine the response of root growth to envi-
ronmental conditions such as temperature (Pritchard et al. 1990) or water
deficit (Saab et al. 1992). Proteins involved in the process of elongation are start-
ing to be identified (e.g. Saab et al. 1995; Cosgrove 1993). Turgor of growing
cells, which is the driving force for elongation, tends to be buffered against the
same environmental conditions (Spollen and Sharp 1991 for water deficit,
Pritchard et al. 1990 for temperature) and increases in the case of mechanical
impedance (Clark et al. 1996). Overall, these studies suggest that changes in
elongation rate are a primary process with their own control and can in many
cases be seen as a cause, rather than a consequence, of changes in local con-
centrations of solutes (Sharp et al. 1990), especially in carbohydrates (in the
case of mechanical impedance, see Atwell 1990). These studies, carried out at
the single root level, now need to be extended to the whole plant level, thereby
integrating processes over longer timescales, and taking into account the
branching rate and branch abortion as constitutive mechanisms of root
development.
Analyses and models of root system architecture have been developed, and
now give a reasonable approximation of root branching, root development rate
and 3-dimensional root geometry (e.g. Pellerin and Pages 1994; Jourdan et al.
1995) in cases without appreciable environmental constraints. However, the
effect of environmental changes on the parameters of these models has received
little attention (e.g. Aguirrezabal and Tardieu 1996). Systematic analyses of the
effect of root environment on architecture are crucial for a better prediction of
root system growth under variable and fluctuating environments.
3 Control and Measurement of the Physical Environment 77
This chapter deals with the methods available for these studies at the whole
plant level. They essentially deal with the manipulation of water deficit, me-
chanical stress and oxygen stress under laboratory conditions, and with the
measurement of these variables in the field, where they usually cannot be
manipulated as easily.
Questions arising on control and manipulation of the environment in root
studies:
1. Is the control of root growth local (i.e. depending on the conditions of a few
mm around the root) or integrated (i.e. depending on the conditions expe-
rienced by the plant as a whole)?
This question is crucial for determining the relevant measurements of the
plant environment in growth studies. A root system does not usually sense
uniform conditions over the whole soil profile. Under common water deficits,
some roots experience a severe water deficit, while deeper roots stiH grow in
soil at near field capacity. To what extent do unfavourable conditions sensed
by part of the root system, or by shoots, affect the growth of roots subjected
to favourable conditions? In addition to their role in water and nutrient
uptake, roots act as sensors and can send chemical or hydraulic messages to
the rest of the plant (see Davies et al. 1994). This is certainly the case for
water deficit, during which roots synthesise abscisic acid which is translo-
cated in the plant and affects growth of distant organs. It has been argued
(Tardieu 1994) that this is also the case for mechanical impedance: the
changes in characteristics of the whole root system cannot be accounted for
by local effects alone. It is therefore relevant to consider the effect of local
changes in the root environment on the growth of the whole plant, both roots
and shoots. Split root techniques, in which part of the root system is sub-
jected to a constraint while the rest of it is not, are a powerful tool for inte-
grated analyses, and are described in this chapter for several environmental
conditions. A careful analysis of field experiments, where 'split roots' develop
naturally because of uneven soil depletion, can also be very useful, provided
that accurate measurements of soil physical conditions are carried out in the
whole soil profile.
2. To what extent does root growth depend on shoot development?
Roots depend on shoots for carbohydrate supply. Several studies suggest that
root development can vary appreciably even in constant and favourable soil
conditions, when the availability of carbohydrates to roots varies. This is the
case when the intercepted radiation varies, between days or between growing
seasons (Aguirrezabal et al. 1994), in plants with periodic shoot growth
(Thaler and Pages 1996) or if carbon dioxide concentration changes in the
air (Idso and KimbalI1992). Changes in leaf area and intercepted light are,
therefore, a crucial variable for analysing root growth. This is obviously the
case when air temperature or radiation varies among experiments, but may
also be the case when an adverse condition sensed at the root level alters leaf
growth (e.g. Beemster et al. 1996 for soil compaction; Ben Haj Salah and
Tardieu 1996 for water deficit). In this case, it is essential to consider light
interception as an integrating mechanism at the plant level. Adverse condi-
tions sensed by roots can affect leaf growth, which in turn affects assimilate
availability to roots and root growth.
We have not presented methods for studying light interception in this
chapter as this is a common method in ecophysiology, but we consider it a
crucial measurement to be taken into account in root studies, especially in
the field.
3. What spatial and temporal resolution of environmental conditions is accept-
able in a study of root growth?
Due to the low conductivity of the soil to fluxes of water, heat, oxygen, etc.,
local conditions are often variable. To what extent is it necessary to charac-
terise this variability, and to what extent is it acceptable to measure mean
conditions in a soil volume? Physical processes in the soil are frequent1y non-
linear (e.g. responses of soil water potential and hydraulic conductivity to
soil water content), and local conditions sensed by a part of the root system
can affect the growth of the whole plant. In this case, measuring mean envi-
ronmental conditions may not be adequate, depending on the problem under
study. It is therefore safe, especially in the field, to combine local measure-
ments (e.g. given by tensiometers or electrodes) with more integrated mea-
surements (such as pre-dawn leaf water potential for water deficit). The
location of sensors, in relation to the studied phenomenon, and the choice
of integrated vs. local measurements, are frequent1y causes of success or
failure of analyses. We give suggestions in this chapter for taking appropri-
ate decisions.
3.2 Laboratory Control of the Growth Environment
3.2.1 Control of Water Stress
The control of water stress has been reviewed by Krizek (1985), where the
various options for the experimental control of water stress are described. They
are the regulation of water supply; the control of the matric potential of the soil
water; and the use of osmotica.
Before consideration of these approaches in more detail, it is pertinent to
consider the conclusions of Krizek (1985). He considered that withholding
water was the most natural, but gave results that were difficult to interpret. In
contrast, the use of osmotica gave more precise control of water potential, but
care should be taken to avoid the possibility of toxic side effects. However, since
Krizek (1985) wrote his review, the control of soil matric potential has been
developed further. High molecular weight polyethylene glycol, PEG (e.g. 20000)
is relatively common in current literature, in comparison with the widespread
use of PEG (6000) as described by Krizek (1985).
3.2.1.1 Control of Water Supply
Controlling water supply by either reducing the frequency of watering, with-

holding water altogether or by a controlled irrigation is the simplest method of
applying water stress to plants growing in pots. However, the results depend on
soil type and the size of container, both of which influence the amount of water
available to the plant.
Simplistically, a large pot containing clay soil will result in a gradual
in crease in water stress while a small pot containing sandy soil will result in
a rapid onset of water stress. A significant disadvantage of the method is that
it is not possible to determine the potential at the root-soil interface because
water is continually being extracted.
The choice of soil type or alternative growth medium is an important con-
sideration. Schwaegerle (1983) observed that this water withholding method
should only be used with soils with large unsaturated conductivities and with
plants with low water requirement, if a uniform water content is to be achieved.
It is possible to use other growth media; Sharp et al. (1988) equilibrated ver-
mic ulite at potentials between -0.03 and -1.60 MPa. The water release charac-
teristics of the vermiculite used by Sharp et al. (1988) are shown in Table 3.1.
From this data it can be seen that a small change in water content can lead to
a large change in potential. For example, as the vermiculite dries from 0.15 to
0.08 g g-l the water potential decreases bya factor of approximately 2. Therefore
the uptake of only a small amount of water can have a large effect on matric
potential at the root surface. In the experiment of Sharp et al. (1988) the plants
were in a water-saturated environment so that transpiration was negligible.
Table 3.1. Water release characteristics of vermiculite used by

Sharp et al. (1988) in root growth experiments
Water content gg-l Water potential kPa
3.72 -30
0.6 -200
0.15 -810
0.08 -1600
3.2.1.2 Control of Soil Matrie Potential
The requirement for subjecting plants to water stress in a controlled way whilst
avoiding possible toxic effects of osmotica, has led to the development of
sophisticated methods for controlling soil matric potential. There are two prin-
cipal methods of controlling water stress, but in each case the soil must have a
large hydraulic conductivity in order to be certain that the potential at the soil-
root interface is the same as that of the bulk soil. This requirement restricts the
use of this technique to relatively coarse soils at high water potentials (usually
no lower than -20 kPa), if accurate control of the potential at the soil-root inter-
face is required.
BOX 3.1
An experimental system suitable for controlling matric potential is
shown in Fig. 3.1. This represents a refinement of the system described
by Snow and Tingey (1985) in which the growth medium was maintained
above a fixed water table. In the apparatus of Lipiec et al. (1988) ceramic
tubes are connected to a water reservoir in a closed loop. The water
reservoir is a sealed vessel and can be subjected to a pressure below
atmospheric. The pump circulates water to prevent air collecting in the
ceramic tubes, which would seriously limit the performance of the
apparatus. The water uptake by roots can be estimated directly from the
level in the reservoir, provided the system is initially in equilibrium and
there is no change in water storage. Taps can be closed to replenish the
water reservoir.
~
, . . /10
:: -:. :: 9
._ .............. .
8
Fig. 3.1. An experimental system for controlling matric potential of soil (9). Water in reser-
voir (6) is held under a negative pressure and recirculated using a pump (7) through ceramic
tubes (8) in contact with the soil, which operate in a similar way to tensiometers. The pres-
sure in the reservoir (6) is controlled by an automatic pump and pressure control device (s,
1,2,3,4 and 5). (From Lipiec et al. 1988)
Experimental systems as shown in Fig. 3.1 (Box 3.1) have been found to be
useful for studying root infection by pathogens. Iwama et al. (1994) used a
similar system to study the effect of water potential on the infection of tur-
nips (Brassica rapa L.) with Plasmodiophora brassicae which causes club root
disease. They found that for the soil they used (humic volcanic ash) a potential
greater than a critical value of about -11.2 kPa resulted in the initiation of club
root disease. The critical potential was thought to be related to pore size which
dictates the upper limit of the size of mobile spores.
An analogous experimental system is described by Painter (1966). In this
system the soil is packed between two dialysis membranes which are then
surrounded by PEG (20000). With this approach it is theoretically possible to
extend the range of potentials to below the -85 kPa limit (i.e. tensiometer limit)
with the system described by Lipiec et al. (1988). However, care should be taken
to ensure that steep gradients in water potential do not result in more negative
matric potentials at the root surface.
Hsieh et al. (1972) described how the surface of a soil connected to a water
table could be covered with a fine mesh with openings of 28-30 Jlm, to enable
water uptake by root hairs to be quantified. They grew bean (Phaseolus vulgaris
L.), maize (Zea mays L.) and barley (Hordeum vulgare L.) on the surface of the
mesh with no apparent ill-effects. They used gamma attenuation to obtain water
content profiles with a resolution of 1 mm. Fig. 3.2 shows a time sequence of
potential profiles (estimated from water release curves) following the dis con-
nection of the soil water table. This diagram gives a clear illustration of the
uncertainties of the potential at the soil-root interface when the water-
withholding method is used to apply water stress.
3.2.1.3 The Use of Osmotic Solutions
For convenience, osmotic solutions are widelY used to subject plants to water
stress. The osmotic agents most widely used are polyethylene glycol (PEG)
and mannitol. Caution should always be exercised when osmotic agents are
used, because it is necessary to demonstrate that any effect is caused by
water stress and not toxicity or an other property of the osmoticum. PEG is
available commonly in molecular weights between 200 and 20000. Lawlor
(1970) made a detailed study ofthe problems associated with PEG and manni-
tol solutions and, as far as we are aware, it is stiH the most comprehensive. PEG
has been criticised because of the impurities that certain grades sometimes
contain. Lawlor (1970) concluded that this is not a serious problem, although
Janes (1974) observed that the toxicity of PEG 4000 can differ between differ-
ent batches.
Figure 3.3 shows the effect of root damage on PEGs entry into the root for
PEGs of different molecular weights. When undamaged roots were placed in
o OCI:C~:J:1"":1:JO:00'00::JOC:oaoo)O(ol:""O:::X>ClooC oOCOoa,I()(Or::lOCl:JX>::JO~OOOO",,:;1:a::1()(0OlOCO0)0(:ClOCeOX>OOO:::
::oc:::oe::1()(:C""::X>:o:""ooc J: :00:Screen : o o o o oe::::: :: :: C::::J:
:CXlOO::oc::::a:JI()(O:lOC::lOC:JJO:00:
Days since termination of water supp1y Root hair zone
2 8 11 13 15 18 24
~---""-.,,,,- --- ~
~, 3
())
!=: 4
o
N
·ă
..c: 5
"O
8
E 6
o
~
())
u 7
N
<Il
Ei 8
10
11
O -0.5 -1.0 -1.5 -2.0 -2.5 -3.0

Matric potentia1- MPa
Fig.3.2. Matric potential as a function of distance from the root hair screen during the growth
of maize. Parameter is the time elapsed since the termination of water supply. (From Hsieh
et al. 1972)
PEG at -200kPa, little PEG entry occurred for molecular weights of 1000,4000
and 20000, while in damaged roots, PEG 1000 and 4000 entered the root. At a
potential of -1.0 MPa, even in damaged roots, the entry of PEG 20000 was only
small. Lawlor (1970) observed that PEG entry was proportional to the amount
of damage, which is presumably likely to occur in ageing annual plants sub-
jected to mechanical impedance or in perennial plants due to increasing root
permeability.
Both PEG 200 and mannitol can enter undamaged roots, but they have dif-
ferent effects on the leaf compared with higher molecular weight PEG. Zhu and
Boyer (1992) demonstrated that for giant algal cells, mannitol disrupts the
normal process of cell wall metabolism. The use of mannitol to stress roots has
produced some convincing results (e.g. Pritchard et al. 1990). However, it is
uncertain whether some other metabolic effect of water stress or the effect of
mannitol is being observed.
Mexal et al. (1975) have demonstrated that oxygen solubility is decreased
in PEG and the solubility decreases with increasing PEG concentration and
increasing molecular weight. They considered that this problem was sufficiently
(a) (b)
.:g, 30
'51
:;:
{ii
~
bO
'b;, 20
E-
"Ou
.c.
bO
<l!
h
~ 10
>,
fi<l!
%,
p.,
O
-...
2 3 4 5
,...
6 7
,..fIII
8 O 1 2 3 4 5 6 7 8
1
Days from cutting
Fig.3.3. The effect of cutting the root systems of cotton plants on the entry of different mole-
cular weights from a -0.2 MPa and b -1.0 MPa PEG solutions. Cut roots: open triangles 20000,
open square 4000, open cirele 1000. Uncut roots: elosed triangle 20000, elosed square 4000, elosed
cirele 1000. (From Lawlor 1970)
serious that plants growing in nutrient solution containing PEG could suffer
from oxygen stress. Mexal et al. (1975) suggested that solutions of PEG should
be aerated which might allow oxygen to reach the roots directly in a gaseous
form. Verslues et al. (1998) have made a detailed study of oxygen concentrations
around roots growing in PEG (8000) aerated with gas with an oxygen partial
pressure of 20.8kPa, 30AkPa or 45.9kPa. From measurements of oxygen con-
centration profiles around the surface of roots, they concluded that PEG should
be equilibrated with a gas with a higher than ambient oxygen partial pressure
to give root surface oxygen partial pressures equivalent to roots grown in
moist vermiculite. Root elongation rate was observed to be higher when PEG
was aerated with gases containing supplemental oxygen and this effect was
particularly noticeable at water potentials higher than -800 kPa. By making
comparisons between the elongation rates of roots growing in vermiculite equi-
librated at various potentials and PEG, Verslues et al. (1998) concluded that
similar elongation rates could not be taken as evidence against hypoxia. This
was because the elongation rate of roots growing in vermiculite was lower than
that of those growing in PEG aerated with gas with an ambient oxygen partial
pressure at the same water potential. Thus comparisons between the effects of
water stress induced by PEG and other methods (e.g. equilibrated vermiculite)
must be made with caution. However, Verslues et al. (1998) do conclude that
PEG can be used without the complication of oxygen stress if the solution is
aerated with a gas containing supplemental oxygen. The work of Verslues et al.
(1998) is an important reference for those intending to induce water stress

using PEG.
The water potentials of PEG and other osmotica are commonly measured
with a psychrometer. The calibration of concentration against potential is
dependent both on the temperature of the solution and on the molecular weight
of the particular osmotic agent. There are several published calibrations for PEG
(e.g. Malcolm and Rawlinson 1957; Williams and Shaykewich 1969; Steuter et
al. 1981; Materechera et al. 1992), and a description of the technique for using
a psychrometer is given by Mullins (1991).
Osmotic solutions are commonly used in hydroponic systems. These can
be relatively simple as in the case of Gergely et al. (1980) who grew apple (Malus
e
sylvestris L.) seedlings in 1 beakers containing mixtures of PEG and nutrient
solutions. West et al. (1980) grew ornamental plants in 500 mI containers of
coarse washed sand, which were irrigated with PEG solutions. When plants are
transpiring, care must be taken to ensure that the concentration of PEG
does not change due to water uptake by roots. Materechera et al. (1992)
grew seedlings of several monocotyledons and dicotyledons in PEG 20000, by
mixing sand with PEG and placing the mixture on a sintered glass plate at a
matric potential of -2.5 kPa. They used PEG with osmotic potentials of -0.25,
-0.50 and -1.00 MPa, in addition to water at -2.5 kPa. The matric potential
was applied to control the air-filled porosity of the PEG-sand-air mixture. For
aH the dicotyledon species the root diameter was thicker under -1.00 MPa of
water stress than under -2.5kPa. In contrast, for monocotyledons, the root
diameter was thicker in the sand-water mixture at -2.5 kPa. At -1.00 MPa
the elongation rate of the roots was less than 5% of the elongation rate of
unstressed roots.
Ciamporovâ and Luxovâ (1976) found that at osmotic potentials low
enough to stop elongation in maize, radial enlargement occurred. They
found that exposure to PEG 4000 over 24 h caused reversible effects, while a
48-h exposure resulted in irreversible changes. PEG is widely used in germina-
tion studies. For example, Dahal and Bradford (1994) used PEG (8000) to
study the effect of water stress on tomato (Lycopersicon esculentum L.) seed
germination.
To overcome many of the problems with using PEG, such as possible oxygen
stress (Mexal et al. 1975; Verslues et al. 1998) and reduced P uptake (Emmert
1974), Tingey and Stockwell (1977) proposed the use of a semi-permeable
membrane to separate the elongating roots from the PEG. Tingey and StockweH
(1977) claimed that the use of PEG (20000), combined with a membrane
with a molecular weight exclusion limit of 6000 to 8000, reduced many prob-
lems associated with use of PEG. Roots of beans were grown in vermiculite
sandwiched against the semi-permeable membrane. Table 3.2 shows how the
Table 3.2. Influence of various membrane types and thickness on leaf water potential and the
PEG concentration in the root-vermiculite mass. (From Tingey and StockwellI977)
Membrane combinations' Leaf water potential kPa PEG 20000 content mg/g
Fresh wtb
Single dialysis membrane -770 3.8

Double dialysis membranes -840 2
Triple dialysis membranes -980 0.9
Spectrapor' -790 0.4
• Minimum molecular weight exclusion limits for dialysis membrane and Spectrapor are
12-14000 and 6-8000 respectively.
b Expressed as g fresh wt. of the combined root and vermiculite mixture.
, The PEG was extracted from the root-vermiculite mass of plants grown in the membrane
systems for 7 days at -600 kPa. Each mean is based on seven observations. SE = 0.3 for both leaf
water potential and PEG content.
exclusion limit of the membrane affects PEG content of the root-vermiculite

mass.
3.2.1.4 Applying Water Stress to Split Root Systems
Split root systems for studying the effect of water stress on root and plant
growth can be relatively simple. The work of Coutts (1982) is a good example
of a split root experiment. Coutts (1982) described how the roots of a 2-year-
old sitka spruce (Picea sitchensis L.) seedling were divided between two plastic
tubes, 605 mm tall and 109 mm in diameter. The tubes were packed with sandy
loam soil equilibrated to given matric potentials. An example of the treatment
Coutts used is as follows. Both sides watered to -6kPa (determined using ten-
siometers),one side watered to -6kPa and the second side kept to -60kPa or
both sides kept at -60 kPa.
The vertical variation in soil water potential is usually much greater than
the lateral variation. Sommer (1988) describes how it is possible to isolate
compartments in a soil monolith, with the water potential in each compartment
being controlled by a system similar to that shown in Fig. 3.1 and described
by Lipiec et al. (1988). Water movement between adjacent compartments at
different water potentials was prevented with a 2-3 mm thick wax layer
between compartments. A similar experimental system has been used by
Vetterlein and Marschner (1993) to study hydraulic lift in sandy soil planted
with pearl millet. Vetterlein and Marschner (1993) used two compartments,
separated by a 2 cm perlite layer, to prevent capillary flow of water. The lower
layer was wetted by capillary movement of water and the upper layer was
not watered.
3.2.2 Control of Mechanical Stress
3.2.2.7 Controlling Mechanicallmpedance

at Non-Limiting Stress Levels
When studying the effect of mechanical stress on soil, the principal experi-
mental objective is to change soil strength independendy of soil aeration, water
potential and hydraulic conductivity. Voorhees et al. (1975) show that, when soil
is used as a growth medium for roots, the range of water contents and bulk den-
sities used must be carefully selected to avoid poor aeration.
Table 3.3 gives the elongation rate of pea seedling seminal roots in relation
to bulk density, water potential, air-filled porosity and penetrometer resistance.
Any credible experiment with compressible soil which is designed to study the
Table 3.3. Primary root elongation rates, probe measurements, and porosity characteristics of
a Sverdrup sandy loam. (From Voorhees et al. 1975)
Soi! core Primary root Total point Normal point Volumetric Air
bulk elongation resistance, a p resistance, a p water filled
density rate MPa MPa content porosity
gcm- 3 cmhr- t
a= 5° a= 30° a= 5° a= 30°
Water potential - 33 kPa

1.23 0.062 1.62 1.56 0.39 1.09 0.23 0.31
1.24 0.076 1.09 1.05 0.25 0.74 0.24 0.29
1.25 0.079 1.04 1.00 0.24 0.70 0.22 0.31
1.29 0.080 1.00 0.96 0.23 0.67 0.25 0.26
1.38 0.065 1.50 1.44 0.36 1.01 0.27 0.21
1.45 0.050 2.24 2.15 0.56 1.50 0.31 0.15
1.54 0.032 4.48 4.30 1.18 3.01 0.33 0.08
1.61 0.017 6.76 6.44 1.81 4.54 0.30 0.09
Water potential - 100kPa

1.23 0.073 1.56 1.50 0.38 1.05 0.20 0.34
1.25 0.084 0.99 0.95 0.22 0.66 0.19 0.34
1.29 0.075 1.10 1.06 0.26 0.74 0.20 0.32
1.38 0.055 1.98 1.90 0.49 1.33 0.21 0.27
1.54 0.029 5.10 4.90 1.40 3.43 0.24 0.18
1.61 0.017 7.55 7.25 2.08 5.08 0.27 0.12
effect of mechanical stress on root elongation should include such data.

Voorhees et al. (1975) used two types of penetrometer, both conical, but one
with a 5° semi-angle and one with a 30° semi-angle. As discussed by Bengough
(1991) and Greacen et al. (1968) the use of a 5° penetrometer allows the pres-
sure between the root and soil to be estimated. This is because a sharp pen-
etrometer fails soil in a cylindrical failure zone which is thought to be similar
to that caused by roots. The use of penetrometers to estimate resistance to root
elongation is discussed in detail by Bengough and Mullins (1990, 1991) and
Richards and Greacen (1986).
Lipiec et al. (1988) described how it is possible to pack clay soil to a range
of densities and to use the system for controlling matric potential shown in Fig.
3.1, to ensure adequate aeration and water supply.
Abdalla et al. (1969) described how a triaxial cell can be used to apply
mechanical impedance to roots. In this system, the soil is formed into a cylin-
der and placed inside a thin cylindrical rubber membrane to which an exter-
nal confining pressure is applied. A similar approach is described by Barley
(1962) in which the root was pressed against a porous plate with a pressurized
diaphragm. Gordon et al. (1992) developed a specialised cell which allowed
monitor ing of the impeded root by time-Iapse photography. However, in
general, triaxial cells are a specialist apparatus used in soil mechanics and are
only suitable for small seedling roots. A similar approach is described by Scott-
Russell and Goss (1974). In their system, a cell with ftat sides was used that con-
tained a layer of ballotini a few centimetres thick confined within a ftexible
membrane. The ballotini were bathed in aerated nutrient solution and com-
pres sed by an external confining pressure. As Scott-Russell and Goss (1974)
note, a major advantage of all these pressurized systems is that they allow the
effect of mechanical stress per se to be studied. To relate the extern al confining
stress to the pressure a root needs to exert to deform the ballotini, the reader
should refer to Richards and Greacen (1986) and Bengough and Mullins (1991).
Materechera et al. (1992) used a static load placed on an incompressible
sand to apply a confining stress to the root. Care must be taken to minimise
friction at the sand/pot interface, by lining the pot with a low friction PTFE
plastic, if uniform stresses and repeatable results are to be obtained. This
approach is attractive because it is relatively straightforward to implement,
while providing a similar degree of mechanical stress control to the triaxial
apparatus.
To separate out the effects of mechanical impedance from those of oxygen
stress and water stress, Gardner and Danielson (1964) used wax of various
strengths to form layers of known penetration resistance. The degree of aera-
tion and water potential can thus be controlled independently of the strength
of the impeding layer. This work provides an excellent example of the care nec-
essary to isolate the effects of mechanical impedance. The wax layer method for
generating controlled impedances has been used more recently by Yu et al.

(1995) as a screening technique for rice. In their paper they describe in detail
how to modify the strength of the wax layers.
Recently, new developments have been described by Bengough and
Mackenzie (1994) and Bengough et al. (1994) which allow the root growth force
to be controlled whilst the elongation rate is measured simultaneously. The
system they describe applied an axial force to the root tip and resulted in slowing
of root elongation under surprisingly smallloads. Further research is required
in order to reconcile the low growth pressures that severely reduce root elonga-
tion in the apparatus with the relatively strong soils that roots can deform.
However, such apparatus offers the opportunity to increase the understanding
of the cellular response to mechanical stress. An excellent review of mechanical
impedance and root growth is given by Bengough and Mullins (1990).
3.2.2.2 Effect of Pot Size
When selecting pot size there is an obvious compromise which must be

accepted between having a pot which is Iarge enough to provide unrestricted
root growth and a pot which is small enough from the perspective of practi-
calities. A rule of thumb is that an unimpeded main root axis can grow up to 2
or 3 cm per day. Therefore, for a 1O-day experiment one should allow a pot depth
of 30cm (3cm/day X 10 days), to avoid the main root becoming restricted. In
most experiments, soil strength will prevent roots growing at their maximum
rate, so this is a useful guide.
Examples of pot sizes which have been used successfully are 15 cm diame-
ter and 27 cm long (Veen and Boone 1990) and 8.5 cm diameter and 19 cm long
(Masle 1992). In those experiments maize was grown for up to 9 days (Veen and
Boone 1990) and wheat (Triticum aestivum L.) and barley for up to 20 days
(Masle 1992). For older pIants, Arvidsson and Jokela (1995) grew barley for up
to 35 days in cores 29.5 cm in diameter and 50 cm deep.
Ruff et al. (1987) give an excellent account ofhow restricted rooting volume
can reduce the shoot growth of tomato plants, when grown for several weeks
(up to 10). They grew tomato plants in either a 0,45t' or a 13.5t' rooting
container, which were both well watered with nutrient solution. The plants with
a restricted rooting volume were dwarfed. The reader should also refer to
Marschner (1995, p. 530).
3.2.2.3 Complete Impedance
Measurements of maximum axial growth pressure (O'max), which is the max-

imum force per unit area the root can exert, (O'max), are reported widely
(Gill and Bolt 1955; Eavis et al. 1969; Pfeffer 1893; Misra et al. 1986; Souty and
St~pniewski 1988; Whalley and Dexter 1993; Whalley et al. 1994; Clark et al.
1996). Several different apparatus for measuring a max have been reported, all
of which are "home-made". The principle of the measurement is to allow a
root to elongate into a small blind hole in a piece of ceramic which will com-
pletely impede the root. If the root is anchored securely, the maximum
growth force can be measured using a force transducer attached to the ceramic
(Fig. 3.4). The rate of growth force development is dependent on tempera-
ture (Bengough et al. 1994; Whalley et al. 1994). At 25 ac the roots of pea
approach a maximum growth force in about 24 h. Although growth force
may continue to in crease gradually beyond that point, the value of amax is
stable. The value of a max in pea was not found to be sensitive to temperature
Uptumed glass
beaker
Core
Seedling
Ceramic cone
(C) Plaster of Paris
Water
container
Water
Adjustable
stand '--......
Digital balance
Fig. 3.4. Apparatus for measuring the maximum growth force generated by a seedling root.
(From Whalley and Dexter 1993)
over the range 5 to 30°C (Bengough et al. 1994; WhaHey et al. 1994). A major
Iimitation of this approach of measuring root growth pressure is that it can
only be easily used with seedlings. An important advance is described by
Misra (1997) who measured a max in both lateral and primary roots of pea and
eucalypt (Eucalyptus nitens) seedlings. As far as we are aware, measurements of
a max have not been made on the mature roots of seedIings more than a few
days old.
One of the main benefits of maximum growth pressure measurements has
been to provide data which help explain how individual roots respond to envi-
ronmental stresses. The approach has the advantage of providing a direct esti-
mate of root growth pressure. The most important data have been provided by
experiments in which a max was measured as a function of oxygen, water and
temperature stress. Eavis et al. (1969) have shown that in the case of pea, a max
was not significant1y affected when the oxygen partial pressure was reduced
from 21kPa to 3kPa and was in the range 1.1 to lAMPa. In the case of cotton
(Gossipium hirsutum L.), a max was reduced from l.lMPa to 0.5MPa when the
oxygen partial pressure was reduced from 2lkPa to 3kPa. In experiments on
young maize, Souty and St~pniewski (1988) varied oxygen partial pressures
from Oto 20 kPa either around the whole seedIing or just the radic1e. The values
of amax decreased with oxygen partial pressure in both cases, although the
decrease was less severe when only the radic1e was subjected to oxygen short-
age, probably because oxygen cou1d diffuse inside the root. Eavis et al. (1969)
measured no significant decrease in a max for pea and cotton roots when they
were subjected to matric potentials of -0.03 or -0.01 MPa. Whalley et al. (1998)
extended the range, of water stress for pea down to -0.45 MPa. Over this larger
range, the value of a max almost halved (i.e. from 0.66 MPa to 0.35 MPa).
Clark et al. (1996) combined measurements of maximum growth pressure
with micro-pressure probe measurements of ceH turgor. The micro-pressure
probe technique is described by Pritchard et al. (1990) and it was based on the
use of a micro-pipette which is connected to a pressure transducer. By making
turgor pressure measurements in impeded pea roots, mounted in a speciaHy
adapted device for measuring a max, Clark et al. (1996) were able to demonstrate
that when pea roots become impeded, turgor increased from 0.55 to 0.78MPa
and confining pressure due to the ceH wall tension decreased from 0.55 to
0.26 MPa.
3.2.2.4 Use of Sp/it Roots in Re/ation to Mechanica/lmpedance
As far as we are aware, spIit root systems have not been used widely in relation
to mechanical impedance per se. SpIit root systems have been used to study the
effect of water relations on plant growth. Gowing et al. (1990) spIit the roots of
young apple plants between two pots 300 mm long and 60 mm in diameter and
were able to demonstrate the role of hormones in root-shoot interaction in
plants which were growing in soil, when the strength increased as a conse-
quence of soil drying.
Bar-Yosef and Lambert (1981) used two coaxial acrylic cylinders, in which
e
the roots were split between six 2.55 compartments. A range of soil strengths
were obtained by varying either soil bulk density (by packing) or matric po-
tential (by surface irrigation and suction). The experiment was performed
with Norfolk sandy loam packed to bulk densities between 1.35 and 1.80 g cm3
and matric potentials -3 to -60kPa. The authors found that at penetrometer
resistances greater than 2 MPa, roots continued to grow under a matric poten-
tial of -16 kPa and soil density of 1.85 g cm 3 but not under a matric potential
of -60 kPa and soil density of 1.45 g cm-3• For a given penetrometer resistance,
the root elongation rate was always lower when the matric potential was
more negative.
3.2.3 Oxygen Stress
3.2.3.1 Indicators of Soil Aeration Status
The oxygen concentration in the soil atmosphere is determined by the balance

between gas diffusion and respiration. Methods for measuring gas diffusion
through soil cores in laboratory conditions are well developed (e.g. Ball et al.
1981), but they depend upon well-defined boundary conditions which may be
difficult to achieve in root growth experiments. However, within the strict
context of roots, oxygen flux can be determined by polarographic methods. This
is because the platinum electrode can consume oxygen and simulate a sink of
a size similar to a root tip (e.g. 3 mm long and 1 mm in diameter). Polarographic
methods for estimating oxygen flux are discussed at length by Armstrong and
Wright (1976), Blackwell (1983), Gliflski and St~pniewski (1985) and Armstrong
(1994). According to McIntyre (1970) the process of oxygen reduction at an elec-
trode in the pH range 5 to 12 is:
O2 + 2H 2 0 + 4e- ~ 40W,
and for pH < 5

O2 + 4H+ + 4e- ~ 2H 20.
Thus, for the complete reduction of one oxygen molecule, four electrons are
required. The measurement is made with an electrode made from inert plat-
inum which is negatively polarised in relation to a reference electrode. This ref-
erence electrode can be a saturated colomel electrode or a silverlsilver chloride
electrode. The relationship between the potential at the electrode surface and
the current generated from the reduction process is complex (Fig. 3.5). As the
potential of the electrode is decreased from Oto -0.4 V the current will increase,
until a plateau region is reached. In this region current is proportional to the
rate of oxygen reduction, which is in effect the rate at which oxygen molecules
collide with the electrode (i.e. the oxygen flux at the surface of the electrode).
As the potential is decreased further, the electrode current increases sharply in
response to the reduction of hydrogen ions. The plateau region, in which
the current is proportional to the oxygen flux, is normally between -0.4
and -0.7V.
Platinum electrodes (or any other inert electrode) tend to over-estimate
oxygen flux at the surface of a root, because the oxygen concentration at the
surface of an electrode is zero. In contrast, at the surface of a root, the oxygen
concentration is normally greater than zero.
Measurement of oxygen flux with polarographic methods has several prac-
tical limitations. In unsaturated soils there may be no plateau region, and in
such circumstances the method should not be used. Glinski and Sttypniewski
(1985) note that the method is only valid when the electrode is completely
covered with a water film. This corresponds to a matric potential of greater than
about -0.1 MPa. Armstrong (1994) recommended that polarograms should be
frequently checked if there is any doubt. He noted that naked electrodes are very
prone to plateau shifts, usually at the upper boundary of voltage potential. With
time, carbonates and hydroxides can precipitate on the electrode and impair its
3
11
10
7
;;(
~ 6
C 2
~ 5 Fig.3.5. Current versus applied
:::J
o voltage characteristics of a platinum
4
1 electrode in (1) saturated sandy
:/
3 loam, (2) stirred soil suspension and
2 (3) an unsaturated sandy loam. Note
that no plateau is present in the case
of the unsaturated sandy loam, and
the measurement cannot be used to
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
estimate oxygen flux. From Glinski
Applied voltage (V) and Sttypniewski (1985)
performance. This is sometimes called electrode poisoning (Glmski and

Sttţpniewski 1985), and for reliable results, the electrodes should be inserted
immediately before the measurement is taken.
Redox potential is another useful indicat ion of aeration status in the soil.
This is discussed in detail by Glinski and Sttţpniewski (1985) and it is particu-
larly useful for characterising anoxic conditions, where oxygen flux measure-
ments are of less value. Redox potential can be correlated with various different
biological processes. Examples of such processes are the emergence of oats
(Avena sativa L.) and the elongation rate of roots (Blackwell and Wells 1983;
Sttţpniewski et al. 1991). It is difficult to generalise from such relationships, as
the biochemical reactions are specific to the soil type and conditions. Redox
potential is measured with a high impedance voltmeter and with platinum elec-
trodes of various sizes and shapes. The potential is measured with respect to
a reference electrode of saturated calomel (or silver/silver chloride) and is
expressed relative to that of a saturated hydrogen electrode. Electrodes for mea-
suring redox potential can be left in situ for extended periods and stiH provide
good results. The range of water potentials over which the redox potential mea-
surements are reliable is much wider than in the case of oxygen flux measure-
ments. Thus, redox potential measurements appear less problematic than
oxygen flux measurements. Devices for monitor ing redox electrodes have been
reported (Linegarger et al. 1975; Fltihler et al. 1976; Callebaut et al. 1980;
BlackwellI983).
3.2.3.2 Controlling Oxygen Stress
Oxygen stress can be controlled directly by flooding the soil air space with dif-
ferent gas mixtures. In this way, the effect of particular components of the soil
atmosphere can be studied independently. Examples of this approach are
described by Geisler (1967) who considered the effect of the concentration of
oxygen and of carbon dioxide and also Drew et al. (1979), Konings and Jackson
(1979) and Smith and Robertson (1972) who studied the response of roots to
ethylene (ethene). Sarquis et al. (1991) studied the interaction between ethylene
and mechanical impedance. They grew maize seedlings in a triaxial cell (see
Sect. 3.2.2.1), that was continually flushed with ethylene-free air. Using the same
system, Sarquis et al. (1991) studied the effect of applying ethylene exogenously
by flushing the triaxial cell with air containing l/le e- 1 ethylene at a rate
of 1.5eh-1•
The use of different gases to either flood the pore space of soil or bubble
through a soil suspension results in redox potentials varying from +550 to
-350mV (Sttţpniewski et al. 1991). To obtain a redox potential of 500mV, air
was flushed through a Mississippi sandy loam, flushing the soil with gas con-
taining 10 000 Ilg g-l oxygen in nitrogen, resulting in a redox potential of approx-
imately 400 m V, while oxygen concentrations less than 100 Ilg g-l in nitrogen alI
produced redox potentials between O and 100 m V. When soil suspensions were
used, bubbling with air gave a redox potential of 550 m V, while alI other gases
containing less than 10 000 Ilg g-l oxygen gave redox potentials between -200
and -350 m V. It should be noted that the environment of the soil or soil
suspension can take up to 3 days to attain a reasonably stable value of redox
potential.
One of the best examples of an experiment designed to estimate the values
of oxygen flux that limit root growth is described by Blackwell and Wells (1983).
In this experiment oat plants were established in a 1-m-long perspex tube. The
tubes were then flooded, by lifting the water table slowly, and tilted so that the
elongation of the roots could be observed against the side of the perspex
columns. Oxygen stress was induced due to consumption by microbial activity
and a low diffusion rate due to water logging. Blackwell and Wells (1983) mea-
sured oxygen flux by using both the procedure of Armstrong and Wright (1976)
and also by assuming that a single excitation voltage of 0.65 V could be used
(i.e. assuming that there is no need to check for a plateau in the voltage-current
relationship). Comparison of the results from these two approaches clearly
demonstrates the importance of measuring the voltage-current relationship
(e.g. Fig. 3.5) to establish that the current flow is limited by the oxygen flux. We
recommend Blackwell and WelIs (1983) to readers who intend to make labora-
tory measurements of the value of oxygen flux that limits root growth.
More commonly, indirect approaches are used to modify soil aeration. The
water content or the bulk density can be manipulated to change the air-filled
porosity (e.g. Sttţpniewski and Przywara 1992). Typically, air-filled porosities of
less than 10% are associated with oxygen stress, but this depends upon soil type.
As increased bulk density can result in increased mechanical impedance to root
growth, it is more usual to adjust water content to decrease the air-filled
porosity.
3.2.3.3 Split Root Techniques in Relation to Oxygen Stress
Van Noordwijk and Brouwer (1993) used a split root system in which the roots
were split between two 8 e pots containing nutrient solution. The solution was
either permanently aerated in both pots, permanently non-aerated or the aer-
ation was stopped in half the root system at a certain growth stage approxi-
mately two weeks before measurements of root porosity. The air-filled porosity
of the roots of maize, tomato and gerbera (Gerbera jamesonii) was greater in
the plants which were not aerated. When half the roots were transferred to a
non-aerated nutrient solution, the air-filled porosity of the roots increased in
young plants, but not in older plants. Van Noordwijk and Brouwer (1993) noted
that aeration of the solution involved complex factors such as agitation due to
stirring. They developed a recirculation system in which the oxygen partial
pressure could be varied between 1 and 4kPa. The oxygen partial pressure in
the aerated system stayed close to 20 kPa. Again they found that, for most
species, older roots did not respond when half the root system was transferred
to a decreased oxygen environment.
Air-filled porosity of a root can be measured either with the pycnometer
method (Jensen et al. 1969) or by examination of sections of root under a micro-
scope. Both of these methods are compared by Van Noordwijk and Brouwer
(1988). For maize, these methods compared well. However, Van Noordwijk and
Brouwer (1993) showed that this is not always the case.
StCfpniewski and Bennicelli (1992) and StCfpniewski et al. (1994) have devel-
oped a split root technique based on two rhizotrons filled with soil that can be
flooded with various gases (Fig. 3.6). At the end of the experiment, the soil can
be washed away, with the roots being held on a pin-board. StCfpniewski et al.
(1994) observed compensatory growth in the aerated part of the root system. In
a more complex experiment, W. StCfpniewski and G. Przywara (unpubl.) divided
the root system into four parts and modified aeration indirect1y by adjusting the
water content. Their results are presented in Fig. 3.7. The mass of the root system
decreased as the fraction of the root system, that was flooded, increased.
The rooting environment can also be split into independent horizontal
strata, with thin layers of wax petroleum separating the layers of soil (e.g. Sect.
3.2.2.1). Soft wax can be used to avoid impeding the root system mechanically.
3.3. Field Measurement of the Root Growth Environment
3.3.1 Field Assessment of Root Conditions
Procedures for inspecting soil in the field are described by Batey (1988). We rec-
ommend Batey's handbook (1988) to readers as an excellent practical guide to
field inspection of soil conditions. Table 3.4 (Box 3.2) can be used as a rough
guide to help decide on the probable causes of poor root growth and then on
appropriate measurements which provide data beyond visual observation.
3.3.2 Mechanical Impedance
Mechanical impedance is the most ubiquitous constraint to root growth.

Mechanical impedance is usually measured with a penetrometer, although it can
be estimated from the bulk density of the soil.
Fig. 3.6. Split-root rhizotron (a) exposed and the two (b) halves of the root system of the same
maize plant grown in aerated (air) and anoxic (nitrogen) conditions. (From W. Stlţpniewski and
R. Bennicelli, unpubl.)
0.5
1
0.4
1
bJ)
0.3
'"
<Il
<Il
Ei
oo
1 11
0.2
~
0.1
O 0.25 0.5 0.75

Fraction of the root system under flood conditions
Fig. 3.7. Data from a four part split root system used to study the effect of partial or total flood-
ing on a root system. (From W. St~pniewski and G. Przywara, unpubl.)
The use of penetrometers in relation to root growth is discussed in detail

by Bengough (1991). Bennie (1991) has demonstrated that when relative root
growth is plotted against penetrometer resistance, there is a remarkable coa-
lescence of data for maize, cotton, wheat and groundnuts. A penetrometer resis-
tance of approximately 3 MPa is generally regarded as the upper limit beyond
which root growth is negligible (Boone et al. 1986). This is considerably higher
than maximum root growth pressure values reported in Section 3.2.2.3. This
difference is thought to be because soil-metal friction is much larger than root-
soil friction. In addition, when relatively blunt penetrometers are used (i.e. a
cone angle of 30° which is most commonly the case) the failure mode of the soil
differs between penetrometers and roots. This is discussed at length by Ben-
gough and Mullins (1990, 1991).
Most field penetrometers have large diameters, with cone semi-angles of 30°
and are used either for root studies (Whitely and Dexter 1982; Ehlers et al. 1983)
or for engineering purposes (Campbell and O'Sullivan 1991). Large penetrom-
eter cones are relatively insensitive to cracks and heterogenity in the soil at
the scale of the root tip. In coarse-textured soils significant mechanical
impedance can result from sand particles interlocking to resist displacement
(Panayiotopoulos 1989; Gliflski and Lipiec 1990). These cannot be detected by
penetrometers that are much larger in diameter than sand particles.
Dry bulk density is frequent1y used as a measure of mechanical impedance.
There are several different methods for measuring dry bulk density. The
BOX 3.2 Field Investigation of the Rooting Environment

When inspecting a soil profile, common indicators of a poor rooting
environment are sharp structural discontinuities and the presence of
grey and brown mottles which can be indicative of periodic saturation
and aeration problems. Structural discontinuities are very common in
cultivated soils and they can be responsible for limiting deep rooting.
A high number of cracks, continuous pores and planes of weakness
between peds are important in allowing roots to by-pass regions of high
soi! strength.
Table 3.4. Estimation of the most likely causes of poor root growth in reiat ion to soi! type
and water content. This table is intended to give an indicat ion of the likely physical stresses
that should be investigated to determ ine the cause of poor root growlh. The likely physical
stresses that ca n lead to poor root growth considered in this table are water stress, mechani-
cal impedance, temperature or aeration. If any of these stresses are likely to be problematic
then we have attached a p robability of high (H), medium (M) or low (L) to their severity. This
is only intended as a rough guide. The effects of poor aeration and excessive mechanical resis-
tance may in crease in compacted soil. The effects of poor aeration are often smaller al low soi!
temperatures. We have nOI attached any probability to the likely severity of temperature stress
because it is too difficult to generalize. This table has been adapted from ISa (1998)
Water status Clay content Possible Suggested

causes of poor method of
>17% day <17% day root growth investigation
Wilting point Solid, hard, cannot Light colour, darkens Water stress (H) Water content
be moulded, darkens strong.1y when Mechanical Small cone
strongly when moistened impedance (H) penetrometer
moistened
Intermediate Semi· solid, can be Darkens slightly when Water stress (L) Water content
between field moulded, cracks and wened Mechanical Small cone
capacity and crumbles when rolled impedance (M) penetrometer
wilting point by hand into 3 mm
threads, darkens
slightly when wetted
Field capacity; Plastic, can be roLled Fingers get slightly Poor aeration (L) Large cone
free water absent into 3 mm threads moist, no water escapes Mechanical penetrometer
without cracking, no soil pores upon impedance (M) Redox potential
colour <hange upon compression, no colour
wetting change upon wetting
Free water present, Soi! soft and can be Fingers moisten rapidly Poor aeration (M) Large cone
saturating part of roLled into a thread when soil is worked and Mechanical penetrometer
soil pore space <3 mm diameter visible oUltlow from soil impedance (L) Redox potential
pores upon compression
Temperature Temperature
AII pores full, free Ali pores appear AU pores appear \Vater Poor aeration (H) Large cone
\Vater present water filled and free filled and free water penetrometer
water present present Redox potential
Soil surface Soil surface covered Soi! surface covered by Poor aeration (H) Large cone
covered by water by water water penetrometer,
redox potential
simplest of these is to take a core of soil of known volume and oven dry it to
obtain the mass of dry soil per unit volume. A second group of methods involve
excavation of a known mass of soil. The volume of the hole is then obtained by
filling it with particles that pack to a known density. If high spatial resolution
is required gamma attenuation can be used to measure bulk density. An excel-
lent review of methods for estimating bulk density is given by Campbell and
Henshall (1991).
Direct comparisons of bulk density between different soil types are of little
value. To overcome this problem, Hâkansson (1990) proposed the use of the
degree of compactness. This is defined as the ratio of the actual bulk density to
the bulk density of the wet soil under a static load of 200 kPa.
An alternative index is defined by Bennie (1991) as:
[ Pactual- pminJ,
Pmax -pmin
where P is soil bulk density and the subscripts refer to actual and reference
maximum and minimum densities. The minimum bulk density was estimated
from the mass of unsieved dry soil required to fill a 0.2 e container and the
maximum bulk density was determined with the ASTM Proctor test (Felt 1965).
Using the index of Bennie (1991), the values of <0.5,0.5-0.6,0.6-0.7 and 0.7 cor-
respond to low, medium, high and very high degrees of compaction respectively.
Relative compaction parameters give a useful characterisation of soil compact-
ness in studies of root and crop responses to compaction (van Ouwerkerk and
Soane 1994).
3.3.3 Characterising the Effect of Soil Structure

on Root Growth
Methods for studying the interaction between soil structure and root growth
are perhaps the least well developed. This is because soil structure is a
complicated concept which has no simple definition. Dexter (1988) defines
soil structure as "the spatial heterogenity of the different components
of soil".
To illustrate the care that must be taken when relating soil structure to root
growth, it is pertinent to consider two separate classic laboratory experiments.
Firstly, Voorhees et al. (1971) used carefully prepared mixtures of aggregates,
which differed in density. During the experiments, certain density classes were
labelled with 32p. Voorhees et al. (1971) demonstrated that dense aggregates
restricted root growth more than porous aggregates. In the case of very dense
aggregates, root growth was restricted to the periphery of aggregates, while root
growth within the aggregate was affected by the strength and pore size distri-
but ion of adjacent aggregates.
100 w'R. Whalley et al.
Secondly, Scholefield and Hall (1985) demonstrated that roots were able to
penetrate pores smaller than their nominal thickness. From the work of
Voorhees et al. (I97I) and Scholefield and Hall (I985), it is clear that the rela-
tionship between root growth and soil structure is far from trivial. A further
complication is that the relationship between root structure and soil structure
is dynamic. An excellent recent review is presented by Cresswell and Kirkegaard
(l995).
Ehlers et al. (1983) studied the growth of oat roots in tilled and untilled
soils. They characterised the soil structure in several ways. Penetration resis-
tance profiles were measured weekly with a penetrometer which had a 60° cone
angle, Il mm in diameter, and a relieved shaft. The mean penetration resistance
was determined in 5-cm-deep layers down to 40 cm, then in 10-cm-layers
down to a depth of 60 cm. Bulk density profiles were also measured in two layers
with six samples per layer. Bio-pores (root or earthworm holes) were counted
after excavation of the soil to the required depth in an area 50 cm x 50 cm. Ehlers
et al. (1983) used a butane flame to heat the excavated surface of the soil and
then used suction to lift off the loose dry aggregates, so that the bio-pores
could be observed. Ehlers et al. (I983) demonstrated that oat roots exploited
bio-pores, and as little as 2% of bio-pores could improve rootability of the soil.
A similar study on soybean (Glycine max 1.) roots was described by Wang
et al. (l986).
For more detailed studies, the method of preparing thin polished sections
of soil impregnated with resin is particularly useful (McBratney et al. 1992;
Van Noordwijk et al. 1992; Krebs et al. 1994). An excellent example of a thin
section is shown in Fig. 3.8. Van Noordwijk et al. (I992) note that small thin sec-
tions contain insufficient information for quantative analysis, although qualita-
tively they are extremely useful. In the case of Fig. 3.8, the extension of root
hairs into aggregates is clearly visible. Root soil contact is an important
input for models of oxygen, water and nutrient uptake (De Willigen and Van
Noordwijk 1987).
Methods for studying the inter-relationships between soil structure and
root growth need greater development.
3.3.4 Field Measurements of Aeration Status
The polarographic method for mea sur ing oxygen flux described in Section
3.2.3.1 is also suitable for use in the field (BlackwellI983). Similarly, the mea-
surements of redox potential described in Section 3.2.3.1 can also be used in
the field. Armstrong and Wright (l976) and Blackwell (I983) describe devices
which enable large numbers of probes to be monitored. Their equipment can
be used to measure both redox potential and oxygen flux. Readers should be
Fig.3.8. Thin sections of a soil showing different degrees of soil-root contact (black void, light
grainy structures sand particles, thin lines root cell walls). aTransverse section through two roots
growing on the surface of soil peds. b Longitudinal section of root in void between soi! aggre-
gates, establishing contact with soil via root hairs. c Transverse section of root in void with root
hairs reaching for the soi!. d Longitudinal section of root penetrating through and branching
in the soil matrix. (Photographs courtesy of Dr. M.J. Kooistra, SC-DLO Wageningen, the
Netherlands)
warned that considerable effort is needed to develop a system which can

measure oxygen flux and redox potential and then record the data from a
number of probes.
Blackwell (1983) compared the use of platinum electrodes to measure
oxygen flux and redox potential with methods for gas sampling. The gas sam-
pling methods used either 50 mI (intern al volume) sintered bronze samplers or
5 mI polythene and PVC samplers connected to pipes and then to air-tight taps
through which gas and soil solution could be extracted. Blackwell (1983) cau-
tions that some plastics are gas permeable and that they should not be used.
The samplers were placed at depths of up to 80 cm. Blackwell (1983) used 5 mI
gas-tight syringes to sample the chambers and recommended that approxi-
mately half of the volume of the gas sampler chamber should be discarded
before the sample is analysed for gas concentration in the laboratory. Both gas
and liquid phases were analysed to determine their oxygen concentrat ion. The
difference in scale between the large chambers for gas sampling and the small
platinum electrodes requires soil structure to be considered in interpret ing the
results. Blackwell (1983) suggests that, because the dimensions of the electrodes
are much smaller than those of the soi! structural units, the platinum electrodes
are much more likely to be embedded in the soil peds. In contrast, chambers
for gas sampling wiH tend to sample the gas between the soil structural units.
The gas sampling chamber is a convenient method for obtaining repeated
measurements of oxygen concentration in the field and depending on the
availability of laboratory methods for gas analysis it can be extended to
other gases.
Ball et al. (1994) describe a research apparatus which monitors the diffu-
sion of a radioactive gas from an excavated hole in the soil. The experimental
data were used to obtain estimates of relative diffusivity near the soil surface
using a numerical simulation based on Fick's equation. They noted that there
was poor agreement between field and laboratory tests, which they attributed
to variation in air-filled porosity with depth. This technique appears to be
promis ing, but is stiH being developed. BaU et al. (1994) iHustrates clearly how
gas diffusion depends on bulk density, water content, and soil type.
Gas diffusion measurements in the field are rarely made because they are
labour intensive (Gliftski and StfTpniewski 1985) and the air-filled porosity is
often quoted instead.
3.3.5 Measuring and Modifying Soil Temperature in the Field
Soil temperature can be measured with relatively simple and inexpensive

sensors that are widely available. Buchan (1991) discusses temperature mea-
surement in detail.
Soil temperature can be influenced in several ways. Mulching of the soil
surface with plant residues, irrigation or tillage of the soil are the most common
practices in the field (Voorhees et al. 1981).
Irrespective of other effects, changes in soil temperature are determined
largely by soil water content. In dry soil, the smaU loss of latent heat through
evaporation consumes only a small proportion of the incoming radiation, and
heat is conducted slowly. This means that the surface of a dry soil warms up
quicker than wet soil and provides more favourable conditions for the ger mi-
nation of seeds and initial root growth. Drainage helps the soil warm faster in
the early spring, although dry surfaces cool down more rapidly at night and
consequently have a higher incidence of night frost.
In the study of Kuchenbuch and Barber (1988), nine years of air tempera-
ture data (from Lafayette, Indiana) were analysed in relation to the distribution
of root depth in maize. The temperature was expressed in degree days (one DD
was a day with an average daily air temperature of 1 °e above a base tempera-
ture of 10 0C). At silking, the root length density (RLD) below 30 cm was corre-
lated significantly (r = 0.82) with DD for the 2 weeks following planting. The
RLD in the 0-15 cm layer was significantly correlated with precipitation but not
with DD.
200
·0
.... •
30 l;
N
o'"
~ a
N
... o
,.
150
El
• °0'" • •V'
l;
1:
bJ)
"<il
l;
"'" 20
..ci 0l;
'"
» IOa ° b" 'II
... "
-O
oo •
0l;
" ..!!
oo o
• ... 0
l;
o o
~ ~ la
° '"
• °" "
l;",
50
6'.... ~"
l; ....
° ","
o DT." ~ o ... 'tI
o 500 1000 1500 2000 o 500 1000 1500 2000
Accumulated thennal time, °C days
Fig.3.9. Data for wheat from Barraclough and Leigh (1984), showing the accumulation of total
root dry matter or totallength as a function of accumulated thermal time. Solid symbols, 1980
Crops; open symbols 1981 Crops; circles, Rothamsted early-sown; triangles, Woburn early-sown;
squares, Rothamsted late-sown; inverted triangles, Woburn late-sown
Root growth of winter wheat in England was strongly related to various

DDs, accumulated at different rates by varying sowing dates (Barraclough and
Leigh 1984). An example of the relation between accumulated thermal time and
the root length of winter wheat grown in the UK is shown in Fig. 3.9. The early
differences in root length due to sowing date were maintained throughout the
season.
The growth trajectory of maize nodal roots is affected by soil temperature
until the root is 100 mm long (Tardieu and Pellerin 1991). This root response of
young nodal roots to temperature affects the development and proliferation of
the root system later in the development of the plant.
Kovar et aL (1992) used two different tillage systems to influence soil tem-
perature and water regimes. Although differences in soil temperature and root
length density between the two tillage systems were significant, there was no
significant correlation between soil temperature and RLD. The authors sug-
gested that, for prediction of root distribution from the soil temperature and
water content of the soil, both parameters should be monitored regularly.
Various aspects of soil temperature in relation to rooting are reviewed by
Kaspar and Bland (1992).
3.3.6 Interpreting the Effects of Combinations of Stresses
Root growth in the field is affected by many fac tors which vary in space and
time, and can interact. Establishing which factors are limiting root growth is
very difficult. Figure 3.10 shows how both penetrometer resistance and air-filled
porosity are related to matric potential and dry bulk density. An air-filled
porosity of 10% and a penetrometer resistance of 3 MPa are regarded frequently
as critical for root growth. Figure 3.10 illustrates that as bulk density increases,
the range of matric potential, in which aeration and mechanical impedance do
not limit root growth, becomes narrower. In compacted soils, therefore, rela-
tively small changes in water content (either drying or wetting) can lead to
reduced root growth. This has been demonstrated in field experiments by Grath
and Hăkansson (1994). In areas of compaction (e.g. headlands, wheel
tracks), and in depressions, leaf yellowing occurred. Plant species responded
differently to combinations of stress (Goss et al. 1989). In wheat, the effect
on root elongation of a combination of mechanical impedance and poor aera-
tion could be predicted from measurements of the effects of the stresses act ing
independently. In contrast, for maize the combined effect was greater than that
predicted (see Goss et al. 1989) indicating a significant interaction between
stresses.
The effects of soil compaction, temperature, nitrogen and water supply on
the growth and distribution of roots of winter wheat were studied by Barra-
clough et al. (1991). They found that drought-stressed roots grew deeper than
roots in irrigated soil, and soil porosity or mechanical impedance appeared to
be major factors affecting root distribution.
Dense layers such as ploughpans, claypans or fragipans in soil profiles
restrict root growth by large mechanical impedance, insufficient aeration and
sometimes reduced water availability to roots (Unger et al. 1981). These layers
have large values of bulk density and cone resistance. The thicker the layer and
the nearer it is to the soil surface, the greater the restriction of root growth. The
negative effect of the dense layers can sometimes be alleviated by compensatory
root growth. Barraclough and Weir (1988) showed that when water was not a
-1000
Fig. 3.10. Relationship

-10 between penetration
- -
I~~~::::::- ---
resistance of a penetrometer
and air filled porosity of
10% _- - Air-filled porosity a silty loam soil in relation
_14-______-r_-~7~_o__- .________._----__.
to bulk density and matric
1.25 1.35 1.45 1.55 1.65 potential. (From Lipiec et al.
Bulk density, Mg m,3 1991)
limiting factor in the topsoil and N was added, the pan had a negligible effect
on the grain yield of winter wheat.
In soils with a water table within the root zone, root development may be
affected adversely both by anoxic conditions, and by the lower soil temperature
in cold climates. In hot climates, salt in the water table may limit root growth.
3.4 Future Trends
The techniques that we have described for controlling the rooting environment
in the laboratory are almost alI designed to provide an environment that is con-
stant with time. The exceptions are the withholding of water which provides an
in crease in both mechanical impedance and water stress with time, but in an
uncontrolled way. For a single root axis the approach of Bengough et al. (1994)
provides the possibility of changing the mechanical impedance as a function of
time. There is a need for more development of systems that alIow root growth
to be studied in an environment with controlled temporal variation in physical
stresses, particularly mechanical impedance and water stress. The data obtained
should be related to the root architecture.
The approaches for measuring maximum growth pressure of a single root
axis need to be extended to the whole root system. Misra (1997) has made an
important contribution to this objective. There is also a requirement for
methods which alIow the maximum axial growth pressure of roots in mature
plants to be measured.
Significant progress has been made in our understanding of the response
of roots to water stress by Sharp and co-workers and to mechanical stresses by
Bengough and co-workers, but the interaction between these two stresses is less
well understood. More work is needed to identify improved methodologies for
looking at combinations of these and other stresses.
Laboratory approaches for studying the effect of soil structure on root
development are poorly developed. Approaches are needed to study the inter-
act ion between root architecture and soil structure which can elucidate the
function, both for soil structure and the root system.
There is a need for better integration between field work and laboratory
experiments. Both of these approaches are important, but there is a tendency
for them to be conducted in isolation.
We hope that the approaches that we have described in this chapter will
help to raise the level of awareness of the need to conduct root growth experi-
ments in a carefulIy controlled environment.
Acknowledgments. W.R. WhalIey and J. Lipiec thank the British Council and the
KBN in Warsaw who partly funded this work. We thank Dr. A.G. Bengough, Dr.
M.C. Drew, Dr. J. Tisdall and Dr. W.B. Voorhees for the constructive comments
made on an early manuscript. W.R. Whalley was supported by Silsoe Research
Institute which is grant aided by the Biotechnology and Biological Sciences
Research Council.
References
Abdalla AM, Hettiarachi DRP, Reece AR (1969) Mechanics of root growth in granular media.
J Agric Eng Res 14: 236-248
Aguirrezabal LAN, Tardieu F (1996) An architectural analysis of the elongation of field-grown
sunflower root system. Elements for modelling the effects of temperature and intercepted
radiation. J Exp Bot 47: 411-420
Aguirrezabal LAN, Deleens E, Tardieu F (1994) Root elongation rate is accounted for by inter-
cepted PPFD and source-sink re1ations in field and laboratory-grown sunflower. Plant Cell
Environ 17: 443-450
Armstrong W (1994) Polarographic oxygen electrodes and their use in plant aeration studies.
Proc R Soc Edinb 102B: 511-527
Armstrong W, Wright EJ (1976) A polarographic assembly for multiple sampling of soi! oxygen
flux in the fie1d. J Appl Ecol 13: 849-856
Arvidsson J, Jokela WE (1995) A lysimeter study of soil compaction effects on evapotranspira-
tion in a barley crop. Swed J Agric Sci 25: 109-118
Atwell BJ (1990) The effect of soi! compaction on wheat dur ing early tillering. II. Concentration
of cell constituents. New Phytol115: 37-41
Ball BC, Harris W, Burford JR (1981) A laboratory method to measure gas diffusion and flow in
soi! and other porous materials. J Soil Sci 32: 323-333
Ball BC, Galsbey CA, Robertson EAG (1994) Measurement of soil gas diffusivity in situ. Eur J
Soil Sci 45: 3-13
Barley KP (1962) The effects of mechanical stress on the growth of roots. J Exp Bot 13:
95-110
Barraclough PB, Leigh RA (1984) The growth and activity of winter wheat roots in the fie1d: the
effect of sowing date and soil type on root growth of high -yie1ding crops. J Agric Sci Cam-
bridge 103: 59-74
Barraclough PB, Weir AH (1988) Effects of compacted subsoillayer on root and shoot growth,
water use and nutrient uptake of winter wheat. J Agric Sci Cambridge 110: 207-216
Barraclough PB, Weir AH, Kuhlman H (1991) Factors affecting the growth and distribution of
winter wheat roots under UK field conditions. In: McMichal BL, Persson H (eds) Plant roots
and their environment. Elsevier, Amsterdam, pp 417-419
Bar-Yosef B, Lambert JR (1981) Corn and cotton root growth response to soil impedance and
water potential. Soil Sci Soc Am J 45: 930-935
Batey T (1988) Soil husbandry: a practic al guide to the use and management of soi!s. Aberdeen
University Press, Aberdeen
Beemster GTS, Masle J, Williamson RE, Farquhar GD (1996) Effects of soil resistance to root
penetration on leaf expansion in wheat: kinematic analysis of leaf elongation. J Exp Bot
47: 1663-1678
Bengough AG (1991) The penetrometer in reIat ion to mechanical resistance to root growth. In:
Smith KA, Mullins CE (eds) Soil analysis, physical methods. Marcel Dekker, New York,
pp 431-446
Bengough AG, Mackenzie CJ (1994) Simultaneous measurement of root force and elongation for
seedling pea roots. J Exp Bot 45: 95-102
Bengough AG, Mullins CE (1990) Mechanical impedance to root growth: a review of experi-
mental techniques and root growth responses. J Soi! Sci 41: 341-358
Bengough AG, Mullins CE (1991) Penetrometer resistance, root penetration resistance and root
elongation in two sandy loam soi!s. Plant Soil131: 59-66
Bengough AG, Mackenzie CJ, Elangiwe HE (1994) Biophysics of the growth response of pea roots
to changes in penetration resistance. Plant Soi1167: 135-141
Ben Haj Salah H, Tardieu F (1996) Quantitative analysis of the combined effects of temperature,
evaporative demand and light on leaf elongation rate in well-watered field and laboratory-
grown maize plants. J Exp Bot 304: 1689-1698
Bennie ATP (1991) Growth and mechanical impedance. In: Waisel Y, Eshel A, Katkafi U (eds)
Plant roots. The hidden half. Marcel Dekker, New York, pp 393-414
Blackwell PS (1983) Measurements of aeration in waterlogged soils: some improvements of tech-
niques and their application to experiments using Iysimeters. J Soi! Sci 34: 271-285
Blackwell PS, Wells EA (1983) Limiting oxygen flux densities for oat root extension. Plant Soi!
73: 129-139
Boone FR, van der Werf HMG, Kroesbergen B, ten Hag BA, Boers A (1986) The effect of com-
paction of the arable layer in sandy soi!s on the growth of maize for silage. 1. Critical matric
water potentials in relation to soil aeration and mechanical impedance. Neth J Agric Sci
34: 155-171
Buchan GD (1991) Soil temperature regime. In: Smith KA, Mullins CE (eds) Soi! analysis, phys-
ical methods. Marcel Dekker, New York, pp 551-612
Callebaut F, Balcaen M, Gabriels D, de Boodt M (1980) Data acquisition system for field deter-
minat ion of redox potential, oxygen diffusion rate and soi! electrical resistance. Meded Fac
Landbouwk Rijksuniv Gent 45: 15
Campbell DJ, Henshall JK (1991) Bulk density. In: Smith KA, Mullins CE (eds) Soil analysis, phys-
ical methods. Marcel Dekker, New York, pp 399-429
Campbell DJ, Q'Sullivan MF (1991) The cone penetrometer in relation to trafficability, com-
paction and tillage. In: Smith KA, Mullins CE (eds) Soil analysis, physical methods. Marcel
Dekker, New York, pp 399-429
Ciamporova M, Luxova M (1976) The effect of polyethylene glycol-induced water stress on the
maize root apex. Biol Plant (Praha) 18: 173-178
Clark LJ, Whalley WR, Dexter AR, Barraclough PB, Leigh RA (1996) Turgor pressure is increased
in the apex of pea roots by complete mechanical impedence. Plant Cell Environ 19:
1099-1102
Cosgrove DJ (1993) How do plant cells extend? Plant Physiol102: 1-6
Coutts MP (1982) Growth of Sitka spruce seedlings with roots divided between soils of unequal
matric potential. New Phytol 92: 49-61
Cresswell HP, Kirkegaard SA (1995) Subsoil ameliorat ion by plant roots, the process and the evi-
dence. Aust J Soil Res 33: 221-239
Dahal P, Bradford KJ (1994) Hydrothermal time analysis of tomato seed germination at subop-
timal temperature and reduced water potential. Seed Sci Res 4: 71-80
Davies WJ, Tardieu F, Trejo CL (1994) How do chemical signals work in droughted plants? Plant
Physiol104: 309-314
De Willigen P, van Noordwijk M (1987) Roots, plant production and nutrient use efficiency. PhD
Thesis, Agricultural University of Wageningen, Wageningen The Netherlands
Dexter AR (1988) Advances in characterization of soi! structure. Soi! Tillage Res 11: 199-238
Drew MC, Jackson MB, Gifford S (1979) Ethylene-promoted adventitious rooting and cortica!
air spaces (aerenchyma) in roots may be an adaptive response to fiooding in Zea mays. L.
Planta 147: 83-88
Eavis BE, Radiff LF, Taylor HM (1969) Use of a dead-Ioad technique to determine axial root
growth pressure. Agron J 61: 640-643
Ehlers W, Kopke U, Hesse F, Bohm W (1983) Penetration resistance and root growth of oats in
tilled and untilled loess soil. Soi! Tillage Res 3: 261-275
Emmert FH (1974) Inhibition of phosphorus and water passage across intact roots by polyeth-
ylene glycol and phenylmercuric acetate. Plant Physiol 53: 663-665
Felt EJ (1965) Compactibility. In: Black CA (ed) Methods of soi! ana!ysis, Part 1. American
Society of Agronomy, Madison, Wisconsin, pp 400-412
Fliihler H, Ardakani MS, Szuszkiewicz TE, Stolzy LH (1976) Field measured nitrous oxide con-
centration, redox potentials, oxygen diffusion rates, and oxygen partial pressures in rela-
tion to denitrification. Soil Sci 122: 107-114
Fraser TE, Si!k WK, Rost TL (1990) Effects of low water potential on cortical celllength in
growing regions of maize roots. Plant Physiol 93: 648-651
Gardner HR, Danielson RE (1964) Penetration of wax layers by cotton roots as affected by some
soil physical conditions. Soil Sci Soc Am Proc 28: 457-460
Geisler G. (1967) Interactive effects of CO 2 and O2 in soil on root and top growth ofbarley and
peas. Plant Physiol 42: 305-307
Gergely 1, Korcak RF, Faust M (1980) Polyethylene glycol-induced water stress effects on apple
seedlings. 1. Methodology, water consumption, and dry matter production. J Am Soc Horti
Sci 105 (6): 854-857
Gill WR, Bolt GH (1955) Pfeffer's studies on the root growth pressures exerted by plants. Agron
J 47: 166-168
Glinski J, Lipiec J (1990) Soil physica! conditions and plant roots. CRC Press, Boca Raton, Florida
Glinski J, Sttrpniewski W (1985) Soi! aeration and its role for plants. CRC Press, Boca Raton,
Florida
Gordon DC, Hettiaratchi DRP, Bengough AG, Young IM (1992) Non-destructive analysis of root
growth in porous media. Plant Cell Environ 15: 123-128
Goss MJ, Barraclough PB, Powell BA (1989) The extent to which physical factors in the rooting
zone limit crop growth. Aspects Appl Bio122: 173-181
Gowing DJG, Davis WJ, Jones HG (1990) A positive root sourced signal as an indicator of soil
drying in apple Malus X domestica Borka. J Exp Bot 41: 1535-1540
Grath T, Hăkansson I (1994) A case study on soil compaction and aphanomyces root rot as
causes of uneven pea growth. Swed J Agric Res 24: 165-170
Greacen EL, Farrell DA, Cockroft B (1968) Soil resistance to metal probes and plant roots. Trans
9th Congr Int Soc Soi! Sci 1: 769-779
Hăkansson I (1990) A method for characterizing the state of soi! compactness for the plough
layer. Soi! Tillage Res 16: 105-120
Hsieh nC, Gardner WH, Campbell GS (1972) Experimental control of soil water in the vicinity
of root hairs. Soil Sci Soc Am Proc 36: 418-421
Idso SB, Kimball BA (1992) Seasonal fine-root biomass development of sour orange trees grown
in atmospheres of ambient and elevated CO 2 concentration. Plant CeH Environ 15: 337-341
ISO (1998) Soil quality - determination of redox potential - field method. ISO/CD 11271,
Helsinki
Iwama H, Osozawa S, Ushiroda T, Kubota T (1994) Analysis of soil water matric potential
requirement for infection of turnip with Plasmodiophora brassicae using negative pressure
water circulation technique. Soil Sci Plant Nutr 40: 293-299
Janes BE (1974) The effect of molecular size, concentration in nutrient solution, and exposure
time on the amount and distribution of polyethylene glycol in pepper plants. Plant Physiol
54:226-230
Jensen CR, Luxmoore RJ, van Grundy SD, Stolzy LH (1969) Root air space measurements by a
pycnometer method. Agron J 61: 474-475
Jourdan C, Rey H, Guedon Y (1995) Architectural analysis and modelling of the branching
process of young oil-palm root systems. Plant Soill77: 63-72
Kaspar TC, Bland WL (1992) Soil temperature and root growth. Soil Sci 154: 290-298
Konings H,Jackson MB (1979) A relationship between rates of ethylene production by roots and
the promoting or inhibiting effects of exogenous ethylene and water on root elongation.
Z Pflanzenphysiol 92: 385-391
Kovar JL, Barber SA, Kladivoko EJ, Griffith DR (1992) Characterization of soil temperature, water
content and maize root distribution in two tillage systems. Soil Tillage Res 24: 11-27
Krebs M, Kretzschmar A, Babel V, Chadreuf J, Goulard M (1994) Justifications on distribution
paUerns in soil: basic and relative distributions of roots, channels and cracks. In: Ringrose
AJ, Humphreys S (eds) Soil morphology. Eisevier, Amsterdam
Krizek DT (1985) Methods of inducing water stress in plants. Hortic Sci 20: 1028-1038
Kuchenbuch RO, Barber SA (1988) Significance of temperature and precipitation for maize root
distribution in the field. Plant Soil 106: 9-14
Lawlor DW (1970) Absorption of polyethylene glycols by plants and their effects on plant
growth. New Phytol69: 501-513
Linegarger RS, Whisler FD, Lance JC (1975) A new technique for rapid and continuous mea-
surement of redox potentials. Soil Sci Soc Am Proc 39: 375-377
Lipiec J, Kubota T, Iwama H, Hirose J (1988) Measurement of plant water use under controHed
soil moisture conditions by the negative pressure circulation technique. Soil Sci Plant Nutr
34:417-428
Lipiec J, Hăkansson 1, Tarkiewicz S, Kossowski J (1991) Soil physical properties and growth of
spring barley as related to the degree of compactness of two soils. Soil Tillage Res 19:
307-317
Malcolm GN, Rawlinson JS (1957) The thermodynamic properties of aqueoues solutions of poly-
ethylene glycol, polypropylene glycol and droxane. Trans Faraday Soc 153: 921-931
Marschner H (1995) Mineral nutrition of higher plants. Academic Press, Cambridge
Masle J (1992) Genetic variation in the effects of root impedance on growth and transpiration
rates of wheat and barley. Aust J Plant Physiol 19: 109-125
Materechera SA, Dexter AR, Alston AM, Kirby JM (1992) Growth of seedling roots in response
to external osmotic stress by polyethylene glycol 20000. Plant Soil 143: 85-91
McBratney AB, Moran CJ, Stewart JB, Caule SR, Koppi AJ (1992) Modifications to a method of
rapid assessment of soil macropore structure by image analysis. Geoderma 53: 255-274
McIntyre DS (1970) The platinum microelectrode for soil aeration measurement. Adv Agron 23:
235-283
Mexal J, Fisher JT, Osteryoung J, Reid CPC (1975) Oxygen availability in polyethylene glycol solu-
tions and its implications in plant water relations. Plant Physiol 55: 20-24
Misra RK (1997) Maximum axial growth pressures of the lateral roots of pea and eucalypt. Plant
Soi1188: 161-170
Misra RK, Dexter AR, Alston A (1986) Maximum axial and radial growth pressures of plant roots.
Plant Soil 95: 315-326
Mullins CE (1991) Matric potential In: Smith KA, Mullins CE (eds) Soil analysis, physical
methods Marcel Dekker, New York, pp 75-109
Painter LI (1966) Method of subjecting growing plants to a continuous soil moisture stress.
Agron J 58: 459-460
Panayiotopoulos KP (1989) Packing of sands - a review. Soil Tillage Res 13: 101-121
Pellerin S, Pages L (1994) Evaluation of parameters describing the root system architecture of
field grown maize plants. Plant Soil164: 155-167
Pfeffer W (1893) Druck und Arbeitsleistung durch wachsende Pfianzen. Abh K Săchs Gese der
Wiss 33: 235-474
Pritchard J, Wyn Jones RG, Tomos AD (1990) Measurement of yield threshold and cell wall exten-
sibility of intact wheat roots under different ionic, osmotic and temperature treatments.
J Exp Bot 41: 669-675
Richards BG, Greacen EL (1986) Mechanical stresses on expanding cylindrical root analogues
in granular media. Aust J Soi! Res 24: 393-404
Ruff MS, Krizek DT, Mirecki RM, Inonye DW (1987) Restricted root zone volume: infiuence on
growth and development of tomato. J Am Soc Hortic Sci 112: 763-769
Saab IN, Sharp RE, Pritchard J (1992) Effect of inhibition of abscisic acid accumulation on the
spatial distribution of elongation in the primary root and mesocotyl of maize at low water
potentials. Plant Physiol 99: 26-33
Saab IN, Ho THD, Sharp RE (1995) Translatable RNA populations associated with maintenance
of primary root elongation and inhibition of mesocotyl elongation by abscisic acid in
maize seedlings at low water potentials. Plant Physiol 109: 593-601
Sarquis JI, Jordan WR, Morgan PW (1991) Ethylene evolution from maize (Zea mays L.) seedling
roots and shoots in response to mechanical impedance. Plant Pathol96: 1171-1177
Scholefield D, Hall DM (1985) Constricted growth of grass roots through rigid pores. Plant Soi!
85: 153-162
Schwaegerle KE (1983) A method for maintaining soil moisture availability for potted plants.
Soil Sci Soc Am J 47: 608-610
Scott-Russell R, Goss MJ (1974) Physical aspects of soi! fertility - the response of roots to
mechanical impedance. Neth J Agric Sci 22: 305-318
Sharp RE, Silk WK, Hsiao TC (1988) Growth of the maize primary root at low water potentials.
1. Spatial distribution of expansive growth. Plant Physiol 87: 50-57
Sharp RE, Hsiao TC, Silk WK (1990) Growth of the maize primary root at low water potentials.
II. Role of growth and deposition of hexose and potassium in osmotic adjustment. Plant
Physiol 93: 1337-1346
Smith K, Robertson PD (1972) Effect of ethylene on root extension of cereals. Nature (Lond)
234: 148-149
Snow MD, Tingey DT (1985) Evaluation of a system for the imposition of plant water stress.
Plant Physiol 77: 602-607
Sommer C (1988) Soil compaction and water uptake of plants. Catena Suppl II: 107-111
Souty N, Sttţpniewski W (1988) The infiuence of external oxygen concentration on axial root
growth force of maize radicles. Agronomie 8: 295-300
Spollen WG, Sharp RE (1991) Spatial distribution ofturgor and root growth at lowwater poten-
tials. Plant Physiol 96: 438-443
Sttţpniewski W, Bennicelli R (1992) Effect oflocalized anoxia on maize response. In: Kutschera
L, Hubl E, Lichtenegger E, Persson H, Sobotik M (eds) Root ecology and its practical appli-
cation voI 3. ISRR Symp, Univ Bodenkultur Wien, pp 137-138
Sttţpniewski W, Przywara G (1992) The infiuence of soil oxygen availability on yield and
nutrient uptake (N, P, K, Ca, Mg, Na) by winter rye (Secale cereale). Plant Soi! 143:
267-274
Sttţpniewski W, Pezeshki SR, de Laune RD, Patrick WH Jr (1991) Root studies under variable
redox potential in soi! using laboratory rhizotrons. In: Kutschera L, Hubl E, Lichtenegger
E, Persson H, Sobotik M (eds) Root ecology and its practical application. ISRR Symp, voI
3. Univ Bodenkultur Wien, pp 353-356
Sttţpniewski W, Bennicelli R, Glinski J, Stepniewska Z (1994) Split rhizotron technique for local-
ized anoxia studies. Trans 15th Congr ISSS, Acapulco, Mexico 2b: 237-238
Steuter AA, Mozafor A, Goodwin JR (1981) Water potential of polyethylene glycol. Plant Physiol
67: 64-67
Tardieu F (1994) Growth and functioning of roots or root systems subjected to soi! compaction.
Towards a system with multiple signalling? Soi! Tillage Res 30: 217-283
Tardieu F, Pellerin S (1991) Infiuence of soil temperature during root appearance on the trajec-
tory of nodal roots of field grown maize. Plant Soi! 131: 207-214
Thaler P, Pages L (1996) Periodicity in the development of the root system of young rubber trees
(Hevea brasiliensis Muel): relationship with shoot development. Plant Cell Environ 19:
56-64
Tingey DT, Stockwell C (1977) Semipermeable membrane system for subjecting plants to water
stress. Plant Pathol 60: 58-60
Unger PW, Eck HV, Musick JT (1981) Alleviating plant water stress. In: Arkin GF, Taylor HM
(eds) Modifying the root environment to reduce crop stress. American Society of Agricul-
tural Engineers, St Joseph, MJ, pp 61-66
Van Noordwijk M, Brouwer G (1988) Quantification of air filled root porosity: a comparison of
two methods. Plant Soi! 111: 255-258
Van Noordwijk M, Brouwer G (1993) Gas-filled root porosity in response to temporary low
oxygen supply in different growth stages. Plant Soi1152: 187-199
Van Noordwijk M, Kooistra MJ, Boone FR, Veen BW, Schoonderbeek D (1992) Root-soil contact
of maize, as measured by a thin-section techique. 1. Validity of the method. Plant Soil139:
109-118
Van Ouwerkerk C, Soane BD (1994) Conclusions and recommendations for further research on
soil compaction in crop production. In: Soane BD, van Ouwerkerk C (eds) Soil compaction
in crop production. Elsevier, Amsterdam, pp 627-642
Veen BW, Boone FR (1990) The infiuence of mechanical resistance and soi! water on the growth
of seminal roots of maize. Soil Tillage Res 16: 219-226
Verslues PE, Ober ES, Sharp RE (1998) Root growth and oxygen relations at low water poten-
tials. Impact of oxygen availability in polyethylene glycol solutions. Plant Physiol 116:
1403-1412
Vetterlein D, Marschner H (1993) Use of a microtensiometer technique to study hydraulic lift in
a sandy soil planted with pearl millet (Pennis,etum americanum 1. Leeke) Plant Soil 149:
275-282
Voorhees WB,Amemiya M,Allmaras RR, Larson WE (1971) Some effects of aggregate structure
heterogeneity on root growth. Soil Sci Soc Am Proc 35: 638-643
112 W.R. Whalley et al.: Control and Measurement of the Physical Environment
Voorhees WB, Farrell DA, Larson WE (1975) Soil strength and aeration effects on root elonga-
tion. Soil Sci Soc Am Proc 39: 948-953
Voorhees WB, Allmaras RR, Johnson CI (1981) Alleviating temperature stress. In: Arkin GF,
Taylor HM (eds) Modifying the root environment to reduce crop stress. American Society
of Agricultural Engineers, St Joseph, MJ, pp 217-266
Wang J, Hesketh JD, Woolley JT (1986) Pre-existing channels and soybean rooting patterns. Soil
Sci 141: 432-437
West DW, Merrigan IF, Taylor JA, Collins GM (1980) Growth of ornamental plants irrigated with
nutrient or polyethylene glycol solutions of different osmotic potentials. Plant Soil 86:
99-111
Whalley WR, Dexter AR (1993) The maximum axial growth pressure of roots of spring and
autumn cultivars oflupin. Plant Soil157: 313-318
Whalley WR, Clark LJ, Dexter AR (1994) The temperature dependance of the maximum axial
growth pressure of the roots of pea (Pisum sativum 1.). Plant Soil163: 211-215
Whalley WR, Bengough AG, Dexter AR (1998) Water stress induced by PEG decreases the
maximum growth pressure of the roots of pea seedlings. J Exp Bot 49: 1689-1694
Whiteley GM, Dexter AR (1982) Root development and growth of oilseed, wheat and pea crops
on tilled and untilled soil. Soil Tillage Res 36: 370-393
Williams J, Shaykewich CF (1969) An evaluation of polyethylene glycol PEG 6000 and PEG
20000 in the osmotic control of soil water matric potential. Can J Soil Sci 49: 397-401
Yu L-X, Ray JD, Q'Toole JC, Nguyen HT (1995) Use of wax-petrolatum layers for screening rice
root penetration. Crop Sci 35: 684-687
Zhu GL, Boyer JS (1992) Enlargement of Chara studied with a turgor clamp - growth rate is not
determined by turgor. Plant Physioll00: 2071-2080
CHAPTER4
Modelling Root System Growth and Architecture

L. Pages l , S. Asseng 2, S. Pellerin3 , and A. Diggle 4
1 INRA, Unite de Recherches en Ecophysiologie et Horticulture, Site Agroparc, 84914 Avignon
cedex 9, France
2 CSIRO, Division of Plant Industry, Private Bag, PO, Wembley, Western Australia WA 6014,
Australia
3 INRA, Unite d'Agronomie, Domaine de la Grande Ferrade, BP 81,33883 Villenave d'Ornon
cedex,France
4 Western Australian Department of Agriculture, Baron-Hay Court, South Perth, WA 6151,
Australia
CONTENTS
4.1 lntroduction 114
4.2 Root Depth and Root Length Density Models 115
4.2.1 Root Depth Models 115
4.2.2 Root Length Density Models 117
4.2.3 Root Depth-Root Length Density Models 117
4.2.4 Soi! Impacts on Root Distribution 118
4.2.5 Shoot -Root Interactions 120
4.2.6 Exudation, Mortality, Respiration and Activity 121
4.2.7 Conclusions 122
4.3 Models of the Root System Architecture 123
4.3.1 Describing Branching Patterns 123
4.3.1.1 Statistical Descriptions 123
4.3.1.2 Hydraulic Networks 124
4.3.1.3 Fractals 125
4.3.2 Formalisation of Developmental Rules 126
4.3.2.1 General Rules 126
4.3.2.2 Root Emergence 127
4.3.2.3 Growth Process 127
Elongation Rate 128
Growth Direction 128
4.3.2.4 Branching Process 129
Branch Location 129
Branch Initial Direction 130
4.3.2.5 Mortality and Abscission 131
4.3.3 Parameter Estimation 131
4.3.4 Evaluation of Models of Root System Architecture 133
4.3.5 Utilisation 135

114 L. Pages et al.
4.4 Present and Future Trends 137

4.4.1 Root Depth and Root Length Density Models 137
4.4.2 Models of Root System Architecture 138
References 141
Further Reading 146
4.1 Introduction
For any complex system, and particularly for the root system interacting with
the rest of the plant and the environment, a model can be a helpful tool for syn-
thesising knowledge to produce more global representations and for testing
hypotheses on the interacting mechanisms derived from experimental results.
The models presented here are mainly devoted to a description of the kinetics
of root system development and the spatial arrangement of roots, with differ-
ent viewpoints and varying levels of detail according to different objectives.
They have been separated into two main categories that correspond to differ-
ent ways of consider ing the root system.
In the first category are models of root depth and root length density which
aim to describe the root system relative to the soil volume, mapping the pres-
ence (or absence) of roots and/or their density (e.g. root length or root mass
per soil volume). The root system is considered as an abstract object whose
selected characteristic is the spatial distribution of roots, which is described
like a concentration throughout the soil. These models are generalIy one-
dimensional, and they alIow the simulation of various types of root profiles
(especialIy root length density versus depth). They give a simple representation
of the shape of the plant/crop root system in which alI roots are equivalent and
are not explicitly connected. These models have been developed mainly as part
of dynamic hydraulic, plant/crop or ecosystem models. Their ultimate aim is to
simulate carbon, water and nutrient cyeles.
The second category of models considers the root system as a set of con-
nected axes that grow and branch according to a number of production rules
which formalise morphogenetic rules and local (at a root scale) interactions
with the environment (mainly the soil). Therefore, they simulate the whole
architecture of the system, i.e. not only its shape (spatial arrangement of it com-
ponents) but also its structure and topology. The structure refers to the specific
identification of its components, the roots, and to their relationships (connec-
tions, branching patterns, and morphogenetic gradients). Hence, these models
are more specificalIy devoted to the study of root system structure and mor-
phogenesis. The simplified representation which they give alIows inference on
different aspects of root functioning, which depend on both the shape of the
system (e.g. water and nutrient transport from the soil to the roots), and its
4 Modelling Root System Growth and Architecture 115
structure (e.g. uptake and transport within the root system). These root system
models are sometimes part of more integrated plant growth models.
4.2 Root Depth and Root Length Density Models
Root depth and root length density models simulate the shape of a root system
in terms of root elongation to the depth and root length density distribution
over the soil profile. They are based on the assumption that the root system is
usually in contact with several parts of the soil profile which differ in water, air
and nutrient content, texture, bulk density, organic material and pH, whereas
the soil profile is generally considered to consist of layers, and the conditions
in each layer are treated separately. These root models commonly abstract from
the dynamics of a real root architecture to homogeneously distributed roots in
each of the horizontal soillayers. In some cases, only the single process of down-
ward penetration of the deepest roots or the active rooting front (in terms of
uptake capacity) is simulated.
Root models vary in their complexity both in simulating a root system and
in the response of the roots to their environment and are often part of more
complex crop models. Some of them have been reviewed by Klepper and
Rickman (1990).
A summarised overview of root models is given in Table 4.1.
4.2.1 Root Depth Models
One of the simplest concepts of root modelling is based upon the assumption
that the roots in each centimetre of rooted depth absorb the same amount of
water and nutrients, regardless of the quantitative presence of roots and provided
that alllayers are equally moist. As a result, only the advancing rooting front or
deepest root without root distribution is simulated. Some of these models cal-
culate a root system using a predetermined downward penetration rate to a
maximum depth (Wessolek 1993; P.D. Jamieson pers. comm. 1996). In others,
depth increases as a function of air temperature (T.R. Sinclair pers. comm. 1993)
or phenological growth stage (Chapman et al. 1993).Another example in this cat-
egory of models is the one by Borg and Grimes (1986). They developed a descrip-
tion of the time course of rooting depth as a sine function on a normalised scale
without considering any environmental growth impact.
The root models by Stapper (1984), O'Leary et al. (1985), Groot (1987) and
Penning de Vries et al. (1989) also simulate a growing root system only as an
in cre as ing rooting depth, but they account for some soil constraints. Most of
these models include a restriction of the downward penetration due to
unfavourable soil moisture. Some additionally consider soil temperature
116 1. Pages et al.
Table 4.1. Summarised root depth and root length density models according to crop
specification, with effects on shoot:root partitioning, root depth and root distribution
Author Species Impact on Depth Distribution

shoot:root
Adiku et al. (1995) G O

Addiscott and Whitmore (1987) W TA O,N
Andrew (1987) W CG Rs, CNu
Asseng et al. (1997b) W Cp,CW,CN Cw, O, A, T, Rs O,A,N,Rs
Chapman et al. (1993) Sf Cp
Denison and Loomis (1989) L Cp O,T O,T
Groot (1987) W Cp O,T
Hoogenboom and Huck (1986) G Cw O O, Rs
Jakobsen and Dexter (1987) W CG O, T, Rs O, T, Rs
Johnson and Thornley (1985) Gr Cp,CN
Jones and Kiniry (1986) M CG,CW O, H
Jones et al. (1991) M,Sb,W A,Al,Rs A, Al, Ca, Rs, H
Keating et al. (1998) W CG,CW,CN Cw, O, H O, H
O'Leary et al. (1985) W Cp, TA TA> O
Penning de Vries et al. (1989) G Cw,A, T,Rs O
Ritchie et al. (1985) W CG,CW O, H
Robertson et al. (1993) Sg CG O, H
Skiles et al. (1982) Gr Cp,C G
Spek and Oijen (1988) M Cp,CN
Stapper (1984) W Cp Cp,Cw, O, T
Stol et al. (1993) W Cp 8,T
Van Keulen and Seligman (1987) W Cp,CW,CN Cw,CN,Rs
Voit and Sands (1996) Trees 'CG, 'CN
Williams et al. (1984) M,Sg,W,B, TA TA O, A, T, Al, Rs
O,Sf,Sb,L
Co,Gn,Gr
• C refers here to trees instead of crops.

b Abbreviations used: G general crops, W wheat, B barley, O oat, M maize, Sb soybean, Sg
sorghum, Co cotton, L lucerne, Sf sunfiower, Gr grasses, Gn groundnut, Cp crop phenology, CG
crop biomass growth, Cw crop water deficit, CNcrop nitrogen deficit, CNu general nutrient status
of crop, TA air temperature. And for soil properties: O moisture, A aeration, T temperature, N
nitrogen, Al aluminium, Ca calcium, R, soil strength, H user-defined hospitality.
(Stapper 1984; Groot 1987), soil aeration (Penning de Vries et al. 1989) or soil
resistance (Van Keulen and Seligman 1987) at the deepest penetrated soillayer
for calculating the advancing root. An acceleration of the downward penetration
can result from plant water (Stapper 1984) and plant nitrogen deficit (Van
Keulen and Seligman 1987). However, the model by Penning de Vries et al. (1989)
reduces downward penetration with increasing plant water deficit.
4.2.2 Root Length Density Models
The simplest empirical descriptions of root systems which result in more than
just root depth are models which calculate the distribution of roots with depth
under non-limiting growing conditions. Examples are the models by Page and
Gerwitz (1974), Gerwitz and Page (1974), Monteith et al. (1989) and Nielsen and
Mackenthun (1991). Page and Gerwitz (1974) introduced a mathematical model
where roots grow in a manner analogous to the diffusion movement of solutes
in a liquid medium, by calculating the radial root dis tribut ion start ing from a
point or line source into a infinite medium. Monteith's model describes declin-
ing root length with depth as an inverse square root function of depth, whereas
the model by Gerwitz and Page (1974) assumes an exponential decline in root
length density with depth, and supplies the percentage of roots between the
surface and any soil depth:
where Pi is the percentage of roots contained in a soil depth of z cm and a is a

parameter, such that 1/ a is equal to the depth of soil which contains 63% of the
total root mass.
In another model by Nielsen and Mackenthun (1991) the root intensity of
single trees is calculated (in root mass per soil volume) as a function of the basal
stern are a and exponentially declines with the distance from the tree.
4.2.3 Root Depth-Root Length Density Models
A more complex class of models simulates a root system by exploring the soil in
two separate growth processes: (1) downward penetration of an abstract verti-
cal axis and (2) proliferation of root length in any given soillayer within the pen-
etrated depth. In these models the downward penetration is independently
calculated from the proliferation of roots within the soillayers. This approach is
employed in a range of crop models, for example the CERES-models by Ritchie
et al. (1985) and Jones and Kiniry (1986), the crop models by Hansen (1975),
Andrew (1987), Denison and Loomis (1989), Robertson et al. (1993),Asseng et al.
(1997b) and Keating et al. (1998), and the stand-alone root models by Hoogen-
boom and Huck (1986) and Jones et al. (1991). In "root depth-rootlength density
models", the downward penetration serves as the basis for root distribution over
the soil profile and therefore limits the root distribution with depth at a given
time. Root development over the soil profile is then usually distributed propor-
tional with depth. This characteristic, together with a continuously advancing
rooting depth usually results in a negative exponential function of monotonie
decrease of root distribution with soil depth.
118 1. Pages et al.
In root depth-root length density models, calculation of downward pene-

tration is usually similar to the root depth "only" models. Downward penetra-
tion can be driven by air temperature without being affected by any soil
properties, as for example in the CERES models (Ritchie et al. 1985; Jones and
Kiniry 1986) or as in Addiscott and Whitmore (1987), who extended the root
distribution algorithm of Gerwitz and Page (1974) by incorporating a down-
ward penetration rate as a function of degree-days. However, the downward
penetration rate can also be a constant (Robertson et al. 1993), or additional
unfavourable soil properties can be taken into account (Jones et al. 1991).
Keating et al. (B.A. Keating, pers. comm.) use a user-defined soillayer specific
hospitality factor to account for static soil impacts on the downwards root pen-
etration rate. The carbon supply from the shoot (Denison and Loomis 1989) or
the plant water deficit can also be considered for the downward root penetra-
tion rate (Asseng et al. 1997b; Keating et al. 1998).
The model by Jones et al. (1991) simulates downward penetration in a
similar fashion to proliferation in each layer in that penetration is restricted by
the most limiting constraint. The restriction results from soil strength, alu-
minium toxicity, aeration and coarse fraction at the deepest penetrated soillayer.
In the calculation of downward penetration, other models take into account the
phenological stage of the plant as well as soil strength (Andrew 1987), soil mois-
ture (Asseng et al. 1997b; Keating et al. 1998) or soil temperature at the deepest
rooted layer (O'Learyet al. 1985). The following equation summarises the struc-
ture of the rooting depth calculation in most of these models:
Ract(t) = RpotSzC(t),
where Ract(t) is the actual downward penetration rate at time t, R pot is the poten-
tial downward penetration rate which might be a function of crop development
stage or air temperature, Sz is the soil physical constraint at the deepest rooted
depth z and C(t) is the water or nitrogen deficit status of the crop at time t. Note
that in the models by Ritchie et al. (1985), Addiscott and Whitmore (1987) and
Robertson et al. (1993) soil constraints influence only the root proliferation
within soillayers and not the downward root penetration.
4.2.4 Soil Impacts on Root Distribution
Addiscott and Whitmore (1987) have modified the simple root growth model
by Gerwitz and Page (1974) to be sensitive to its environment by incorporating
a growth factor Fi' This factor considers the effect of varying soil moisture and
mineral soil nitrogen contents on root growth distribution, calculated as:
where Ni is the quantity of mineral nitrogen and ei the water content in layer i.
N and e are the means over the rooted soil profile. Note that the model by
Addiscott and Whitmore (1987) does not simulate root length density develop-
ment. However, it defines a fraction of water and solute that is root accessible
in a layer i. Nevertheless, most root length density models distribute the new
growth over the rooted soil profile by using the same or a similar approach. For
example Jakobsen and Dexter (1987) partition available carbon between layers
according to a potential root growth in each layer, which is driven by a number
of soil fac tors. A slightly different method is used in the model by Asseng et al.
(1997b) where, in a first step, aU carbon for root growth is offered to the topsoil
layer. Only under unfavourable growing conditions in that layer is carbon
pas sed to the next deeper layer and so on, until the layer with the deepest root
is reached and aU remaining carbon is used in that layer. In case of an aeration
deficit in a soillayer reached by the root system, carbon is then given back to
the topsoil for root growth.
Impacts of soil moisture on root growth distribution are taken into account
in most root distribution models, such as those by Huck and Hillel (1983),
Ritchie et al. (1985), Hoogenboom and Huck (1986), Penning de Vries et al.
(1989), Asseng et al. (1997b) and Keating et al. (1998). Other effects on root
growth considered in these models are soil nit ro gen (Asseng et al. 1997b;
Keating et al. 1998), soil temperature (Huck and Hille11983; Groot 1987; Jakob-
sen and Dexter 1987), soil resistance (Jakobsen and Dexter 1987; Van Keulen
and Seligman 1987; Robertson et al. 1993; Asseng et al. 1997b), soil cracks
(Jakobsen and Dexter 1987), soil aeration (Williams et al. 1984; Penning de Vries
et al. 1989; Von pfeil et al. 1992; Asseng et al. 1997a,b), aluminium toxicity
(Williams et al. 1984; Hoogenboom and Huck 1986) and calcium deficiency
(Hoogenboom and Huck 1986). Simulated root length density distributions
affected by different initial soil nitrogen concentrations and its change over
time are shown in Fig. 4.1 using the root model by Asseng et al. (1997b).
The root model by Jones et al. (1991), combines aU of the above growth
impacts, but does not account for soil nutrients and cracks. AdditionaUy, Jones'
model considers the effect of coarse structure in the soil on root proliferation.
The growth impacts are expressed as stress factors, which range from 0.0 (no
growth) to 1.0 (no stress). In cases where more than one property affects a
process, the property with the most unfavourable stress is considered limiting
in the Jones model. Note also that in the Jones model the soil water content does
not directly influence root growth, but indirectly via its effect on soil strength.
Some root models employ an extra root soillayer specific hospitality factor, to
account for static soil impacts on growth dis tribut ion or genotype differences
in root morphology. These factors can be user-defined estimates (Ritchie et al.
1985; Keating et al. 1998) or be defined within the model (Jones et al. 1991;
Robertson et al. 1993). In the root model by Keating et al. (1998), hospitality
120 L. Pages et al.
Root Length Density [cm cm-, Fig. 4.1. Simulated root distribution under
a winter wheat crop in early spring (thin
02468 lines) and 50 days later (thick lines) with an
O+---'---'--'--..L.-......I---'---'----'
initial high (--) and an initiallow
10 nitrogen concentration (......... ) in the
20 soi! profile. (After Asseng et al. 1997b)
E 30
~ 40
~
Q)
50
O 60
70
80
90
factor for root distribution is defined independent1y from the hospitality factor
for root depth penetration. A few root models consider the previous root dis-
tribution in the calculation of new root growth (Hansen 1975; Denison and
Loomis 1989; Jones et al. 1991; Adiku et al. 1995; Asseng et al. 1997b).
A quite different approach in estimating the change of root distribution in
cereals with time is the one introduced by Wessolek and Găth (l990). First, root
distribution is calculated at anthesis on the basis of field capacity and wilting
point of each soillayer. Then root distribution is estimated for pre- and post-
anthesis development stages taking into account the soil type and extreme wet
and dry seasons. While it is not a dynamic model, calculating the changing root
distribution with time in the way that most crop models do, it still allows for
estimating root distribution over a growing season.
4.2.5 Shoot-Root Interactions
One style of root modelling involves the partitioning of carbohydrates to dictate

below-ground biomass. In some models the partitioning coefficient is assumed
to be a constant fraction of the daily carbon produced as for example in the
models by Hansen (l975), Jakobsen and Dexter (l987), and Adiku et al. (l995).
Robertson et al. (l993) assume a constant conversion factor for root growth
from the production of above ground biomass. The models by Stapper (l984)
and Van Keulen and Seligman (l987) have a phenologically controlled shoot:
root ratio for plant internal carbon distribution. Other models, for example
those by Huck and Hillel (l983), Hoogenboom and Huck (l986) and Asseng
et al. (l997b) have a phenologically controlled shoot: root ratio component
which is additionally modified by plant water deficiency. A further extension
used by Johnson and Thornley (l985), Spek and Oijen (l988), Asseng et al.
(l997b) and Keating et al. (l998) is the control ofthe shoot:root ratio by plant
nitrogen deficit. The rate at which carbon is utilised by the shoots and roots
depends upon their masses and the internal concentration of carbon and N
in a model by Thornley (1977). In another model, Reynolds and Thornley
(1982) adjust the shoot: root partitioning of carbohydrates according to the
ratio of carbon: nitrogen in the plant storage pools. An additional parameter is
used which determines the degree of control that the plant exhibits over parti-
tioning. Some of the above approaches to modelling shoot: root carbon
partitioning are relevant to trees, if roots are redefined as only the fine
roots, and shoots as the foliage (Cannell 1985). In a tree model by Voit and
Sands (1996) carbon partitioning to the root system is affected by an internal
N level and foliage biomass. A temperature-dependent partitioning is intro-
duced by O'Leary et al. (1985) as temperature either rises or falIs from an
optimum level.
Under optimal growing conditions, shoot growth is favoured in alI of these
models, but roots become more favoured as the conditions change. The folIow-
ing equation modified after Adiku et al. (1995) generalises the shoot: root inter-
action in a range of these models:
Ll W r (t) = XrC(t)Ll Wg (t),
where Ll W r is the availability of carbon for the root system, Xr is a constant frac-
tion of plant-available carbohydrates partitioned to the root which is modified
by a phenologicalIy driven shoot: root ratio, water or nitrogen deficit or sub-
optimal growth temperature represented by C. Ll Wg is the overall amount of
crop-available carbon for growth at time t.
Note that some of the root models in the literature do not take into account
any shoot: root interactions like, for example, the models by Jones et al. (1991),
Addiscott and Whitmore (1987) and Williams et al. (1984).
Root models, which consider a carbon supply from the shoot, usually trans-
form it into length as a basis for later uptake modelIing. This transformat ion
can be done by using a constant parameter for length: mass ratio, hence any
environmental impact is not taken into account (Huck and Hillel 1983; Ritchie
et al. 1985; Hoogenboom and Huck 1986) or by considering soil strength
(Jakobsen and Dexter 1987; Jones et al. 1991; Asseng et al. 1997b) and other con-
straining soil physical properties such as aluminium toxicity and soil coarse
fraction (Jones et al. 1991).
4.2.6 Exudation, Mortality, Respiration and Activity
Some of the root models with carbon coming from the above-ground crop con-
sider a constant (Ritchie et al. 1985; Denison and Loomis 1989) or a plant devel-
opment dependent (Asseng et al. 1997b) exudation rate which is a percentage
of the carbon supplied to the root system for growth.
122 1. Pages et al.
A mortality rate expres sed as a percentage reduction of the current root

system can be calculated as a constant (Ritchie et al. 1985; Denison and Loomis
1989; Robertson et al. 1993) or as a fraction which changes with the plant phe-
nological stage (Asseng et al. 1997b) and soil aeration (Asseng et al. 1997a,b) or
the intern al N level in trees (Voit and Sands 1996). Others consider soil tem-
perature and available carbohydrate reserve levels (Huck and HiHe11983) or soil
aeration and soil moisture (Hoogenboom and Huck 1986; Jones et al. 1991) in
calculating the mortality rate. The model by Skiles et al. (1982) describes root
mortalityand root respiration as a function of soil temperature and plant water
deficit only.
In the model by Asseng et al. (1997b) age classes with different uptake
capacities are applied to roots grown in a soillayer. As a result, the uptake capac-
ity declines with the advancing age of the roots. Johnson and Thornley (1985)
divide the root structure into four compartments (growing, newly expanded,
medium-aged and senescent). This compartmentalisation allows separate
uptake parameters and respiratory maintenance costs to be inserted for differ-
ent ages of root.
4.2.7 Conclusions
The model by Gerwitz and Page (1974) is often utilised in different crop models
because of its simplicity and low parameter requirement. With a few additional
expansions, such as downward penetration and soil impact fac tors (Addiscott
and Whitmore 1987), it can be sufficient for a range of simulation requirements.
Probably the most commonly employed root model is the one incorporated
in the CERES-models (Ritchie et al. 1985; Jones and Kiniry 1986). This root
model is widely tested as part of its crop model (e.g. Otter-Nacke et al. 1986;
Toure et al. 1995) but is also frequently criticised because of its overestimation
of downward root penetration (e.g. Savin et al. 1995), which might be due to
neglecting soil constraints for the downward penetration. The model by Keating
et al. (1998) has derived its root routines from the CERES-models, but incor-
porates soil water, crop water status and a user-defined hospitality factor for the
downward root penetration. In particular, the introduction of an user-defined
hospitality factor made this model more applicable for a range of soil types with
severe root growth constraints (e.g. Asseng et al. 1998).
Jones et al. (1991) constructed the most comprehensive root length density
model in terms of considering soil constraints. It simulates a root system with
most of the possible soil constraints which occur on agriculturalland and offers
a set of required parameters suitable for a range of crops. To be incorporated
into a crop model, however, it stiH needs to be linked via shoot: root interaction
routines as introduced in other models.
4 Modelling Root System Growth and Architedure 123
With more soil impacts incorporated in root models, as well as simulating

the shoot: root interactions, it becomes more and more difficult for model users
to actually measure the increased number of parameters and to verify the large
range of possible scenario combinations. Therefore, the use or further devel-
opment of root length density models should be closely linked to the target
application of the model, the available knowledge of root-soil and root-shoot
interactions, and the practical possibilities of verification under field condi-
tiOllS. As a consequence, any further root growth modelling exercise should aim
to optimise the balance between complexity and large numbers of parameters
on the one hand and simplicity and low parameter requirements on the other.
4.3 Models of the Root System Architecture
These models consider the root system as a set of connected axes, i.e. a tree in
the mathematical sense, with nodes (vertices) and links (edges). Thus, they
manipulate objects which are more concrete than the root length density
models described in the previous section. Their concepts are closer to those
used by botanists to describe real root systems, and they allow both geometri-
cal and structural descriptions of the root system.
Root architecture has been described with a more or less dynamic per-
spective, depending on the objectives and the experimental methods used, thus
leading to different types of models. This distinction between static models
(aiming to describe root systems as particular branching patterns) and dynamic
models (aiming to generate root systems from developmental rules) is very
apparent in the literature. Nevertheless, bridges exist between these two
approaches, as underlined by Shibusawa (1994) who simulated pseudo-fractal
objects with L-Systems (Prusinkiewicz and Lindenmayer 1990).
4.3.1 Describing Branching Patterns
There are several interesting methods for describing the root systems as par-
ticular branching patterns.
4.3.1.1 Statistical Descriptions
Henderson et al. (1983a) analysed the branching pattern (topology and geome-
try) of the structural root system of 16-year-old trees. They showed that it could
be described by a small number of characteristics (lengths, growth directions,
bends, branching angles, and branching frequencies) whose distributions were
124 L. Pages et al.
unexpectedly regular. This regularity contrasted with the global morphology of

the root systems, which appears in contrast to be irregular.
The simulation of rooting patterns at the same stage, with a stochastic
model based upon these characteristic distributions (Henderson et al. 1983b),
confirmed the regularity of the geometric characteristics that were measured.
Moreover, the simulations allowed definit ion of the possible variations of these
characteristics for obtaining realistic rooting patterns.
4.3. 1.2 Hydraulic Networks
This approach has been developed by Fitter and colleagues, from 1982 (Fitter
1982) onwards (Fitter 1996). The general idea was to characterise the root system
architecture, and especially the topology, using the mathematical tools that had
been developed in hydrology in order to describe drainage networks.
For that purpose, the root system is defined as a set of links [linear portion
between a terminal meristem and a branching point (externallink) or between
two branching points (internallink)], being characterised by a few topological
parameters: its magnitude (number of externallinks); altitude (the number of
links in the longest path between an externallink and the base link); topological
index (altitude divided by expected altitude in the case of random branching).
Two extreme topological types ofbranching patterns can be distinguished (Fig.
4.2): the "herringbone" pattern, in which branching is localised only on a main
axis (high hierarchy), and "dichotomous" pattern, in which branching occurs on
every axis, dividing into equivalent axes (low hierarchy). Between these two
A B
4 --_./
3
4 5
Fig.4.2. Diagrammatic
4
representations of A
4 dichotomous and B
herringbone topologies.
7 Numerals are exterior path
lengths, the number of links
4
8 in the path from each
exterior link to the base link.
4 8 (After Fitter 1991)
extreme types, actual root systems present many intermediate patterns that can
be globally quantified by these cited parameters. This method is potentially
useful for the description of functional properties of root systems. It has been
shown to be particularly suitable for describing excavated root systems on which
it is not always possible to distinguish the developmental branching orders.
4.3. 1.3 Fractals
Fractals are mathematical objects (Mandelbrot 1983) that were developed to

describe natural structures with self-similarities at different observation scales.
For root systems, the existence of self-similarities means that the overall branch-
ing patterns may be divided into sub-parts that have approximately the same
appearance as the overall system, and again these sub-parts may be divided into
smaller parts having the same appearance, and so forth. Fractal models are par-
ticularly relevant for natural objects if their self-similarities are high, that is
when zooming-in does not modify the morphological properties of the object,
and when the number of self-similarities levels is high, that is when the number
oflevels for zooming-in is high. For root systems, the number of self-similarities
is determined by the number of branching orders, so it is associated with both
genotype and environmental influences. Their "quality" depends on how
branches at different levels conform to a common morphological model.
These particular properties of a branching pattern can be evaluated in the
plane, for two-dimensional objects, by looking at the relationship between the
number (N) of contacts of the object under study with the square cells of a grid,
and the side r of these square cells. For fractal objects, the relationship is char-
acterised by the simple equation:
with K being a proportionality coefficient, and D being the frac tai dimension.
Note that a straight line has a fractal dimension of 1, whereas a plane filling
object has a frac tai dimension of 2. According to Tatsumi et al. (1989), D is a
synthetic parameter to quantify the intricacy of shape.
Tatsumi et al. (1989) showed that root systems from different species could
be approximated to fractal objects, with self-similarity within the scale range
0.3-20 mm. Moreover, they estimated fractal dimensions D range from 1,48 to
1.58 depending on the species.
The value of D quantifies an overall space-filling property, and so it may be
very useful for quantifying globally colonisation by the root system. The
problem is that the relationship between D and biological or geometrical para-
meters (e.g.length of roots, branching density, number of branching orders) is
not straightforward.
126 1. Pages et al.
Moreover, the application of this method using images, which are two-
dimensional objects, for evaluating the fractal dimension of 3-dimensional
objects raises a number of questions (Berntson 1994b).
4.3.2 Formalisation of Developmental Rules
Unless previous models devoted to the description of established root systems,

these ones aim to generate theoretical growing root systems with a set of pro-
duction rules (Prusinkiewicz and Lindenmayer 1990) i.e. formal translations of
the observed developmental rules.
4.3.2.1 General Rules
These models are essentialIy single-plant models, simulating each plant sepa-
rately, even though it is stiH possible to create theoretical crops by simulating
and gathering several plants, providing that the parameters would be estimated
in the same field conditions (Pages et al. 1989). They are more or less stochas-
tic, since they are made up of different sub-models, with some of them being
deterministic, and others being stochastic. Consequently, each run gives a dif-
ferent simulated root system.
The existing models (e.g. Pages andAries 1988; Fitter et al. 1991; Clausnitzer
and Hopmans 1994) work in approximately the same way. At a given time, the
virtual root system is coded as a list of linear root segments, which describe and
store the characteristics of these segments: geometric coordinates, time of
origin, and references to topological position within the root structure (e.g.
branching order and connection with other segments; Pages and Aries 1988;
Fitter et al. 1991; Clausnitzer and Hopmans 1994). The development of the root
system is simulated in discrete time steps, by applying three formalised devel-
opmental processes to the existing root system at each time step. These are root
emergence (i.e. generation of new root axes from the shoot system), growth (i.e.
elongation of existing axes), and branching (development of new lateral axes).
AlI developing axes of the structure are modified according to the growth and
branching they experience at the given time step.
One of the important characteristics of these models, compared with root
density models (see previous Sect.), is that they do not assume that growth and
branching processes are uniform among root axes. The simulations may take
both the topological position and the spatial position into account. This alIows
the simulation of different developmental behaviours of the axes according to
their morphogenetic origin. A preliminary axis typology is necessary to distin-
guish axis types with analogous behaviours. These types often correspond to
the branching orders (Diggle 1988; Pages and Aries 1988; Fitter et al. 1991;
Clausnitzer and Hopmans 1994). They may depend also on inter-node of origin
for maize (pages et al. 1989), or they may be the result of a more complex typol-
ogy combining branching order and development process (Pages et al. 1995).
In several recent models (Diggle 1988; Pages et al. 1989; Clausnitzer and
Hopmans 1994), growth and branching are functions of the local conditions as
well. These conditions are either fixed at the beginning of the simulation (Diggle
1988; Pages et al. 1989) or they can be time-dependent (temperature in some
models, like Porter et al. 1986, or Clausnitzer and Hopmans 1994), or be the
output of another combined model (e.g. water potential in Clausnitzer and
Hopmans 1994).
4.3.2.2 Root Emergence
This process generates new root axes, directly originating from the shoot
system. They will be called order 1 axes or axile roots in the following.
By the classification of Cannon (1949), two main types of root systems may
be distinguished, according to the importance of the root emergence process:
primary root systems, and adventitious root systems. Primary root systems
originate entirely from a single root - the radicle - which branches and gives
rise to lateral roots of order 2, order 3, and so ono In this case, the root
emergence process is reduced to emergence of the radicle, some time after
germination. Most roots in such systems originate from the branching process.
In adventitious root systems, several axile roots are emitted from the shoot
system throughout plant development. This type of root system is typical of
grasses. In several different species, it has been shown that the root emergence
process is highly organised in space and time, and requires a specific model-
ling approach. In maize, for example, axile roots appear sequentially on the suc-
cessive basal phytomers from the lowest ones to the highest. The rank of the
phytomer on which the emission occurs at a given time was modelled as a linear
relationship of growing degree days (Pages et al. 1989; Pellerin 1993). In a
similar way, in wheat, each nodal axis pair is initiated after a given thermal time
has elapsed (Diggle 1988). Moreover, a probability of occurrence for each axis
of the pair can be used to simulate variations in the number of such axile roots
(Diggle 1988).
4.3.2.3 Growth Process
To obtain a three-dimensional representation of a root system in which the tra-

jectories of the roots are not completely straight, the models have to consider
two aspects of growth: elongation rate, and growth direction.
128 1. Pages et al.
Elongation Rate. Most, if not alI, of the root architecture models use different
growth rates for different root types. Even in a homogeneous environment,
between-root variation in growth behaviour is high (Fitter 1991, 1996; Waisel
and EsheI1991), and these models typically allow the representation of this bulk
variation.
Moreover, in several models devoted to field simulations (Diggle 1988;
Pages et al. 1989; Clausnitzer and Hopmans 1994), attempts were made to model
elongation rate as a root-type-dependent rate modified according to various soil
factors.
Elongation rate generally decreases as root branching order increases. Axile
roots may have growth rates up to several centimetres per day whereas order 3
branches grow only some millimetres per day (Diggle 1988; Pellerin and Pages
1996). The basic growth rates are constant (linear growth) in several models
(Hackett and Rose 1972; Rose 1983; Porter et al. 1986; Diggle 1988), but more
recently, several authors have suggested different refinements for representing
either continuous, decreasing, or determinate root growth (Pages et al. 1992,
1995; Berntson 1994a). Clausnitzer and Hopmans (1994) have suggested repre-
senting alI types of growth function as series of data points where values
between the points are calculated by linear interpolation.
Stochastic growth models have also been developed to describe the large
variations in growth rate that can occur within root types (Pages et al. 1992).
Stochastic methods were used because part of this variation in growth rate was
considered to have an endogenous origin, which could not be described by a
deterministic model at the scale being considered. In these stochastic growth
models, some of the parameters of the growth functions are drawn at random
for each root from normal or lognormal distributions.
The environmental alteration of the "morphogenetic" component of the
growth rate has been simulated in several models (Diggle 1988; Clausnitzer and
Hopmans 1994) by multiplying it with an impedance factor (from O to 1) cal-
culated from several different soil characteristics. In Diggle's model, the imped-
ance factor (called soil resistance) is specified by the user as a parameter for
each soillayer. In Clausnitzer and Hopmans' model, the impedance factor is the
result of a submodel calculat ion that uses soil water content and bulk density
as inputs.
Growth Direction. The growth direction of the roots is obviously an important

component of root system architecture, because some of the long roots have
very specific behaviours (e.g. vertical or horizontal growth) which strongly
determine the overall shape of the system, including such important charac-
teristics as overall width and depth (Coutts 1989). At a smaller scale, roots gen-
erally exhibit some tortuosity in response to the mechanical constraints which
they experience.
The different aspects which determine the trajectories of roots have been
simulated in architecture models (Diggle 1988; Pages and Aries 1988; Pages
et al. 1989; Berntson 1994a; Clausnitzer and Hopmans 1994) by modifying the
direction of growth of each root tip at each time step. In these models the influ-
ences of the various factors which affect direction are summed to produce a net
directional effect (e.g. by adding mathematical vectors, whose modules are
parameters that are defined for each root type). In one of the most accomplished
descriptions (Clausnitzer and Hopmans 1994), these components include the
direction of the root tip at the previous time step, a random deviation repre-
senting the space-exploring behaviour of the root tip, geotropism along an
angle with the horizontal plane that is user-specified, and the negative soil
strength gradient.
This type of direction model can generate quite realistic representations of
the major trends of actual trajectories, even though the small scale tortuosity
is not very well rendered.
4.3.2.4 Branching Process
The description of the branching process in a root architecture model must

specify the locations and times of appearance of branches on the mother root
as well as their initial direction. Subsequent changes in direction would be cal-
culated by the growth submodel (see previous subsection).
Branch Location. Most of the models presented here have dealt only with
acropetal branching, which is the predominant type of root branching in many
species. This acropetal branching takes place on young parts of the root, and
typically results in a branching front which follows the apex of the mother root
at a more or less definite distance (generally some centimetres). The acropetal
branching process has been formalised in slightly different ways by the various
authors. Some of them (Hackett and Rose 1972; Rose 1983) described it as a
purely temporal process that takes place at given time intervals, and at a given
time lag after emergence of the mother root. Conversely, Lungley (1973) or
Pages and Aries (1988) described it by spatial parameters: an inter-branch dis-
tance, and a minimal distance to the apex of the mother root. According to these
models, branches appear distally from previous branches, at a fixed given inter-
branch distance, until the distance from the last branch to the apex is less than
a threshold distance (the apical non-branching length).
Other authors (Diggle 1988; Pages et al. 1992; Clausnitzer and Hopmans
1994) combined both temporal and spatial hypotheses to model the acropetal
branching, considering that branches appear at given fixed distances from
one to the other (inter-branch distance), and that they can only appear on
130 L. Pages et al.
parts of the mother root having reached a given age (duration of apical non
branching).
Fitter et al. (1991) proposed a stochastic model for the location ofbranches,
defining potential nodes for branching which are created during the growth
process. These potential nodes are given a branching probability, and branch-
ing is tested during a temporal window.
In alI these models, the branching parameters are defined for each root
type, with branching densities which tend to decrease with developmental
branching order. To our knowledge, no models integrate relationships between
root branching characteristics and soil properties.
Even though acropetal branching is very important in root system
development, other types of branch roots may appear out of the acropetal
sequence in some species. On rubber tree for example, Le Roux and Pages
(1994) described "late lateral roots", that appear near the base of the taproot in
zones where density is low, and that have an important fate as main and
perennial lateral roots. The emergence of such roots has been observed and
modelled as a rhythmic process synchronised with the development of shoot
flushes (Pages et al. 1995). This type of branching process can easily be
implemented as an extension of the root emergence process that was described
previously.
Branch Initial Direction. The initial direction of branches is generalIy charac-

terised by two angles between branch roots and their mother roots: a radial
angle (or azimuth), and an insertion angle (or branch angle). Lateral roots gen-
eralIy appear on a defined number of ranks, which face the internal xylem poles.
Nevertheless, the geometrical pattern of root branching is much less dear than
the phylIotaxy for the shoot system.
Berntson (1994a) assumed that lateral roots tend to emerge at successive
adjacent ranks, resulting in spiral sequences, like phylIotaxy. Other models
(Pages and Aries 1988; Fitter et al. 1991) formalised the emergence as a random
sequence on a defined number of ranks. Diggle's branching direction model is
stiH less structured, since branch roots are randomly oriented around the
mother root.
The insertion angle is either fixed (Diggle 1988) or drawn at random from
a normal distribution (Pages et Aries 1988).
Root branching distribution, density, and direction, might also be related
to tortuosity. It is well known, for example, that where a root is buckled, or even
mildly curved, a lateral root wiH form preferentially on its convex side and not
on the concave side.
4.3.2.5 Mortality and Abscission
Root mortality and turnover are important aspects of the development and
function of root systems, especially in perennial plants like trees, but to date
this process has received little attention in root architecture models (Pages
et al. 1995). This is probably because of the difficulties involved in formalising
this mortality process, and also because many of the models have been devoted
to the description of young plants, up to some weeks, in which mortality is not
yet a dominant phenomenon. The difficulties with formalisation are due firstly
to the determinism of mortality, which is complex, and secondly because mor-
phological markers are not reliable and are linked to functional aspects.
In an attempt to include root decay as an aspect of root architecture dynam-
ics in their model on Hevea brasiliensis trees, Pages et al. (1995) considered root
abscission to be the final, and most reliable marker of root decay. Root abscis-
sion is very clear at the base of old perennial axes that become gradually bare.
Decay was modelled using the concept of life expectation. A life expectation
value was attributed to each root according to a normal distribution with para-
meters for each root type. Fine roots are fragile, and have a short life expecta-
tion, whereas some of the big roots are perennial. According to this model, a
root is partially or completely eliminated after a given duration (life expecta-
tion), calculated from the moment when it stops growing. It is completely elim-
inated when it does not support any living branch roots. Otherwise, only its
most dis tai end, beyond the last living branch, is eliminated.
4.3.3 Parameter Estimation
The problem of parameter estimation is a subset of the more general problem

of root observation methods (see other chapters of this book). One major inter-
est of the modelling approach is to drive the observer directly to a selected set
of morphological criteria that are worth observing, because they allow estima-
tion of model parameters. Furthermore, the repetition of identical sub-
structures within the root system, which are clearly identified by the model,
allows interesting reasoning for sampling design. In some species, some para-
meters were shown to be very stable over a wide range of environmental con-
ditions, or easily predicted (e.g. the rate of appearance of adventitious roots on
maize), so that the parameter estimation can focus on parameters which are
more dependant on environmental conditions. The parameters to consider
obviously depend on the model under use, but it is possible to distinguish
between the parameters that need a dynamic observation of the modelled
process (e.g. elongation rate), from those that can be estimated from purely
static observations (e.g. trajectories).
132 1. Pages et al.
In hydraulic networks or fractal models (Fitter 1982,1986,1987,1991,1996;

Tatsumi et al. 1989; Fitter and Stickland 1992; Berntson 1994b), which are
devoted to a more or less static description of the root system, most parame-
ters can be obtained from unique observations of root systems. Moreover, the
measured variables are purely morphological; they do not require expertise in
the underlying developmental processes. They can be measured on detached
parts of root systems.
In estimating parameters for developmental models, expertise is always
required in order to distinguish between root types, i.e. branching orders,
originating phytomers, and so ono These distinctions are relatively easy on
young root systems, but they may be difficult on older ones because radial
growth and root decay can tend to hide some of the morphological markers of
development. Observations of the root system at closely spaced stages of
development can obviously help to abstract the continuity of developmental
processes.
It is worth noting that many of the parameters of these developmental
models can be obtained from purely static observations. Branching densities,
branching angles, or trajectory parameters (Tardieu and Pellerin 1990; Berntson
1994a; Pages and Pellerin 1994; Pellerin and Pages 1996) are examples of such
parameters. Consequent1y, these parameters are quite easy to obtain in field
conditions, even from scattered observations. However, one must take care to
allow for the effects of root decay, which can reduce root density on old parts
of the roots, especially on perennial plants. Parameters which require dynamic
observation are the rate of appearance of roots and their elongation rate. In
grasses, the rate of appearance of adventitious roots on successive internodes
can be easily recorded from regular sampling of plants. Stable relationships
were found with foliar stages on several species, so that this parameter need not
be evaluated in each situation (Klepper et al. 1984; Picard et al. 1985). The
appearance of laterals was found to occur after a constant lag-time of non-
branching behind the apex, whose duration can be estimated in artificial con-
ditions (Pellerin and Tabourel 1995). Estimation of the parameters for the
elongation process, which is particularly important because it drives the kinet-
ies of root system extension, is more difficult. Elongation rates of individual
roots are highly dependant on environmental conditions, and very few satis-
factory methods exist for estimating this parameter under field conditions. The
growth rate parameters can be derived from continuous monitoring of root
systems, either in the field with mini-rhizotrons or in root observation boxes
(see Chap. 8 of this book; Berntson 1992, 1994a; Pages 1992). They can also be
estimated from plant or root populations that are excavated at different stages
of their development. Pellerin and Pages (1994) presented such a study in which
they approximated the growth curve of nodal roots in maize. Pages et al. (1992,
1993) and Tsegaye et al. (1995) have presented general methods for obtaining
the growth characteristics of different types of roots using periodic excavation

of plants.
Global estimation of parameters, by optimising one or more criteria of the
global output of a model (e.g. total numbers or totallength of roots, root maps,
or root profiles), is possible in some situations. Sometimes, however, this is very
difficult, because some aspects of the output may be relatively insensitive to
some parameters, or correlations between parameters may be important. Also,
the accuracy of globally estimated parameters will be very sensitive to bad con-
ditioning of the model. Since the parameters involved tend to have concrete bio-
logical significance, separate estimation seems to be attractive in principle.
Furthermore, individual calibration of parameters allows subsequent validation
using the global outputs, as they have not been used direct1y for calibrating
the model.
4.3.4 Evaluation of Models of Root System Architecture
An essential aspect of modelling, especially concerning dynamic models, is pro-

duction of quantitative outputs that can be compared with actual data, in order
to test and to modify the underlying hypotheses of the models, and thereby
improve them.
Sin ce models of this type are in fact aggregations of several more or less
independent sub-models, which formalise different aspects of development, and
since the objectives of modellers may differ, evaluation of the models can be
done at different levels, using a number of different criteria (e.g. geometry,
topology, developmental kinetics). Moreover, the problem is generally not to
accept or to refute the model, which contains phenomenological parts which
are not really refutable, but to quantify the discrepancies between actual data
and model predictions for criteria of interest (Pages and Pellerin 1996). Given
the potential multiplicity of such criteria, and the multiplicity of the modellers'
objectives, the evaluat ion approach is not easy to formalise, and is still nearly
virgin territory.
Different authors have suggested global and visual validation of their
models, showing pictures of simulated architectures (Fig. 4.3; Diggle 1988; Pages
and Aries 1988; Fitter et al. 1991; Berntson 1994a; Clausnitzer and Hopmans
1994), compared with root system drawings from Kutschera and Lichtenegger
(1992), or with some of their specific observations (Berntson 1994a). This pro-
cedure makes the structural nature of the simulation more tangible, and gives
a global and qualitative feeling about the quality of the model, but it is not very
useful for evaluating or discriminating between hypotheses. The criteria under
evaluation in this sort of comparison are not explicit, and the stochastic nature
of the models raises problems, as each run gives a different pic ture. Graphical
134 L. Pages et al.
Fig.4.3. Projection of
simulated 3·dimensional
root systems. a Maize root
system at the 700 growing
degree days stage (After
Grabarnik et al. 1998, with
kind permission from
Kluwer Academic
Publishers). b Root system of
a 2-year-old rubber tree root
system. Note the abscission
affecting some of the basal
lateral roots. (After Pages
et al. 1995)
1 10CM
aspects of the image, like resolution, colour, or texture rendering are important
criteria for making a picture more or less realistic, and they may hide aspects
of the simulated geometry that the model should represent.
ro evaluate developmental aspects of the models more specifically, Pages et
al. (1992) suggested comparing the observed and simulated kinetics of the
...
o ...
o
ACTUAL SIMULATE O
,,. • t ;-
:,
O
E O
N E N ~ ;. : t
"; '.;
,,, ,,, '.,.-

'.
~ -. ~
rf)
, -'j '
.. ,, (f)
; ... o":,
,,
,,
W W 'J.
- ,,
I : .'
f- ' .1' -. ~ f- -F
<!: <!: ,
Z
O
,,' - ' .
-,-, Z
O
1"" , "- ., ,
15 , 15
.,-,,
,,,
.
'.'
cr: , cr:
O l . ··~·
.;'"
.,,
O "
O O .:.'\.- .' :
, . ~. O O
"'I~ '-. ! ,
O ~ O ~
""" AOw
omR INlII! NT(A
AOW ROW RO"
tr~
..,
O
..,
O
-40 -20 O 20 40 -40 -20 O 20 40
COOR DINATES (em) COORDINATES (em)
Fig. 4.4. Actual and simulated examples of horizontal root maps. Big dots are primary roots,
and small dots are other roots. Note the clustered pattern. (After Pellerin and Pages 1996)
length and number of roots of different branching orders. The stochastic nature
of the model alIowed the simulation of mean relationships as well as confidence
intervals. These could be used to test the hypothesis that observed individuals
could be accepted as members of the set of alI possible runs of the model for a
particular set of parameters.
Geometrical aspects of these models have been tested indirectly using one-
or two-dimensional summaries of global outputs from these models. Porter
et al. (1986) qualitatively compared observed and simulated root length density
profiles, and discussed the possible explanations for the discrepancies. More
recently, Pellerin and Pages (1996) compared observed and simulated maize
root densities and root aggregation on 2-dimensional horizontal maps, and
showed the quality and limits of the geometrical representation produced by a
model of root system architecture (Fig. 4.4).
4.3.5 Utilisation
Root architectural models have only been developed in recent years, so their
utilisation by biologists is stiH extremely limited. Several applications
are, however, expected in the ne ar future, and there are already some pioneer
works.
The most immediate applications of root architectural models are those
directly stemming from the modelling approach: synthesis of scattered pieces
of information about the root systems of several species (Diggle 1988 on wheat;
Pages et al. 1989 on maize), bringing to the fore apparent lack ofknowledge and
suggestion of new hypotheses for further experimentation. Sensitivity analyses
136 1. Pages et al.
will help in studying the relative importance of root parameters on the final
root distribution (what is the consequence of an alteration of one root para-
meter on the final soil colonisation?). Summarising root architecture in a
limited set of parameters provides a basis for comparison of genotypes (Colin-
Belgrand et al. 1989; Lynch and van Beem 1993). At the extreme, fractal branch-
ing models provide relationships between one or two easily measured
parameters (the proximal root diameter for instance, which is close to the plant
stern) and total root length or surface area (Spek and van Noordwijk 1994; Van
Noordwijk et al. 1994; Van Noordwijk and Purnomosidhi 1995).
Important applications of three-dimensional architectural root models lie
in their combination with classical methods for root observation. Spatial
and temporal information about root distribution given by the model may
help in selecting an appropriate sampling strategy. This point is discussed in
Chapter 5. Additionally, simulations may help in investigating both practical
and theoretical problems associated with these classical methods. Using a
model, Pages and Pellerin (1996) have investigated practical problems associ-
ated with the trench profile wall method, namely the effect of the roughness
of the observation face on the rooting profile observed. Bengough et al. (1992)
have used an architectural root model to study the relationships between the
number of root intersecting horizontal and vertical planes and the root length
density.
Another area of application for root architecture models is in studying the
response of the root system to its environment. Root architectural models are
well-adapted for simulat ing interactions of roots with dynamic and spatially
variable soil processes. A pioneer work in this area was proposed by Clausnitzer
and Hopmans (1994) who have developed a model for simultaneous, dynamic
simulation of soil water movement and plant root growth, with the effect of
current soil water content on soil mechanical properties and subsequent
root growth being considered. A similar work was also proposed by Tsegaye et
al. (1995). Important soil features for root growth, such as soil cracks or earth-
worm channels, are unevenly distributed within the soil volume, so that the
geometrical information contained in architectural models is relevant for cal-
culating the probability that roots encounter such features. Architectural
models are also a good support for associating responses of individual roots to
their local environment, and the whole root system response to the percepted
environment, which involves chemical signalling and assimilate distribution
between shoots and roots. Aguirrezabal et al. (1994) have shown that the
response of roots to carbon nutrition depends on their position in the root
system architecture, thus emphasising the idea that architectural models are a
necessary tool for understanding and predicting the root system response to
carbohydrate availability. An attempt at combining these two categories of
processes (local, and general, through carbon distribution) was proposed by
Clausnitzer and Hopmans (1994). An approach of the carbon cost of root

systems, based on an architectural model, was also proposed by Nielsen et al.
(1994) by associating carbon costs for biomass deposition, respiration and exu-
dation to elementary root growth processes. Root architectural models contain
information (for instance the number of growing apices at each time step)
which help in understanding the sink strength of the root system for carbon.
This is of particular interest in the context of studying plant response to atmos-
pheric COz increase.
A fourth important are a of application for architectural models is in the
context of water and nutrient uptake and transport. Nearly aU existing models
for resource cap ture are based on the assumption that roots are verticaUy and
evenly distributed within the soil volume, and that they aU have the same
absorption properties. The information contained in architectural models
about the root system geometry is an appropriate basis for more precise esti-
mates of the volume of soil explored, aUowing for overlaps of depletion zones
between competing roots. Competition between adjacent plants for soil
resources may be studied too. Fitter (1987), Fitter et al. (1991) and Berntson
(1994a) have used architectural models for studying the relative effect of size-
dependent, and architectural, size-independent, properties of root systems on
their potential for resource acquisition. By combining this approach with a
carbon cost approach, Fitter et al. (1991), Nielsen et al. (1994) and Berntson
(1994a) have developed a cost/benefit approach of root systems (volume of soil
explored per unit of volume of tissue required to construct the root system).
Recent progress has also been made in modelling water and nutrient uptake
with the actual, uneven distribution of roots being taken into account
(Bruckler et al. 1991; Lafolie et al. 1991; Tardieu et al. 1992). Until now, these
investigations were performed assuming that aU roots have the same uptake
properties, but the information contained in a root architectural model about
root age and branching order make it possible to relate uptake properties to
these variables. FinaUy, architectural models of the root system provide a basis
for considering transport resistance within the root system, as proposed by
Fitter (1987).
4.4 Present and Future Trends
4.4.1 Root Depth and Root Length Density Models
Root depth and root length density models have reached an advanced state of
development. These models can be used to quantify plant growth effects via the
root system of any soil factor and shoot: root interaction in an empirical way.
They can make direct predictions of readily measurable characteristics of root
138 1. Pages et al.
systems under field conditions. For this reason, the accuracy of their predic-
tions can be verified from straightforward field trials.
Future development of these models is likely to occur in two areas. The first
is in selecting of the particular factors which are most important for any par-
ticular situation being examined, and custom tailoring a root model to deal with
those factors. This effort will be driven by the limitations which are inherent in
increasing the complexity of models. Beyond a certain point, the errors associ-
ated with estimating additional parameters outweigh any possible improve-
ments in precision due to refinement of the structure, as discussed by Passioura
(1996). Because of these limitations, obtaining accurate predictions from
models will depend on skilful selection of the optimal parameter set.
The second area of development of these models is likely to be the con-
tinuing quantification of effects of site and season-specific conditions on
proliferation of roots. Root depth and density models are highly abstract rep-
resentations of root systems and, as a consequence, estimating these effects
can be complicated. The simulation of increased heterogeneity of root distri-
bution deeper in the soil profile represented with one-dimensional models will
be a continuous challenge, in particular when they are the basis for water and
nutrient uptake under supply-limited conditions. It may be that architectural
root models, which concentrate more on mechanism than on measurability, will
prove useful as part of this process.
4.4.2 Models of Root System Architecture
In the case of models of root system architecture, there are a number of areas
where further development or new applications are likely to prove fruitful. This
potential arises from the richness of the information which models of this sort
can store about roots.
Architectural models have complete information about the locations of alI
parts of the root system. For this reason architectural models of root systems
will be increasingly used to simulate interaction of roots with dynamic soi!
processes. Examples have been presented of the use of these models in concert
with models of soil water and soil strength (Clausnitzer and Hopmans 1994).
They can, and no doubt will, also be used with models of the distribution pat-
terns and dynamics of mobile and immobile nutrients, pH, oxygen, and soil
temperature. The three-dimensional nature of the spatial information in these
models will make them particularly relevant to the study of inter-plant compe-
tition for resources.
Furthermore, architectural models store complete information about the
geometry of roots. This information can and, to some extent, already has been
used to examine and add value to various measurement techniques. Many types
of root measurement involve counting intersections between roots and planes.

Simplifying assumptions are then made about root geometry so that root length
density can be inferred. Architectural root models can replace the less sophis-
ticated geometric assumptions currently used in this process. Another area
which could benefit from application of geometric information from these
models is the design of statisticalIy adequate sampling regimes for any type
of measurement. Measurement methods which have been related to model
predictions to date include length measurement by the core break technique
(Bengough et al. 1992), intersection of roots with pit faces (Pellerin and Pages
1996), and time of arrival of roots at transparent walls of containers (Tsegaye
et al. 1995). Other candidate techniques could be core sampling, mini-rhizotrons
and pinboards.
Geometric information can also be used to refine models of movement of
material to and from roots by processes such as diffusion. Such processes
depend on the relationship between distance from roots and the volume of soil
which is within that space. Models which calculate these processes must make
some assumption about this relationship. Baldwin et al. (1973) for example
discuss models of uptake of nutrients assuming either a regular array of roots
or randomly dispersed roots. In fact, of course, roots are not distributed in
either of these ways. The relationship between soil volume and distance from
roots can be estimated for simulated root systems by repeatedly evaluating the
distance between random points in the simulation volume and the nearest part
of the root system. The resulting information can then be represented mathe-
maticalIy in calculations of diffusion by using an appropriate geometric entity
such as a fractal.
A further class of information which is embodied in architectural models
relates to the status of the plant and of alI of its parts, and to their inter-
connection. In plants, the growth and branching behaviour of individual
roots depends not only on the local environmental conditions, but also on the
conditions experienced by alI other parts of the plant at that time and on
the collective state of the plant. This information propagates in plants by
various means. It impacts on the patterns of allocation of assimilate through
time and results in characteristic responses to such stimuli as disease, injury,
and uneven distributions of nutrients and water. Architectural models are
uniquely suited to examining this sort of relationship and will be applied to
studies in this area.
A localised disruption in a root system, as might be caused by some types
of insect or fungal infection, can immediately affect parts which are quite
remote but directly connected. These models can readily be used to address this
type of occurrence.
Architectural root models can also make use of geometric information
relating to other objects in the soil. This capability has been used by Brown and
140 L. Pages et al.
Kulasiri (1994) to detect encounters between growing roots and pieces of

root systems infected by fungi. It could also be used to model effects of
cracks, old root channels, walIs of containers, mini-rhizotrons, or other such
structures.
There are also several examples where architectural models of roots are
combined with other emerging technologies. Considerable advances have
been made in recent times in the architectural modelling of plant tops (i.e.
Prusinkiewicz and Lindenmayer 1990), and it is possible to combine these
models with root models to produce architectural models of whole plants. By
using such a model a unified approach could be taken to calculating growth and
development of plant tops and roots, and partitioning of assimilate within and
between them. A case where the pattern of interconnection between roots
and tops may be important is in the relationship between tillers in wheat and
the growth of their associated nodal roots.
It is also possible to produce architectural root models in what is referred
to as "voxel space" (Greene 1991). Voxel space consists of an array of rectangu-
Iar volume elements which are the three-dimensional equivalent of the pixels
which make up computer displays. The elements, or voxels, are very small (typ-
icalIy less than 1 mm) and each can have any of several states, such as root tip,
root, obstacle, or air. Such a model operates by calculating the state of each voxel
in each time step based on its previous state and the state of its neighbours. The
simulated roots are made up of a collection of adjoining voxels rather than a
collection ofline segments, as is the case in other architectural root models. The
roots can be made to grow in any direction from any point at any time; so elon-
gation, branching and thickening of roots alI occur similarly. Voxel technology
uses large amounts of computer memory, but it could be applied to study inter-
actions between roots and very smalI-scale features of soil. Voxel models are
particularly well suited to visualising complex structures because any voxels or
types of voxel can be defined to be transparent for display purposes, making
elaborate cut-away views possible.
Another class of emerging technology is used for the non-destructive
imaging of volumes (e.g. nuclear magnetic resonance and microtomography).
As discussed in Chapter 11 the resolution achievable with these techniques
is steadily improving and first in-situ images of root systems have been pre-
sented. Such technology will very much facilitate the testing of architectural
models.
Acknowledgments. We thank Drs P. Barlow,A. Fitter, and M. Robertson, for com-

ments on the manuscript and the Grains Research and Development Corpora-
tion for financial support of SA.
References
Addiscott TM, Whitmore AP (1987) Computer simulation of changes in soil mineral nitrogen
and crop nitrogen during autumn, winter and spring. J Agric Sci 109: 141-157
Adiku SGK, Braddock RD, Rose CW (1995) Simulating root growth dynamics of cowpea
under varying soi! conditions. Proc MODSIM95, Int Congr on Modelling and
Simulation, The University of Newcastle, 27-30 November 1995, Newcastle, Australia,
pp 57-61
Aguirrezabal A, Deleens E, Tardieu F (1994) Root elongation rate is accounted for by intercepted
PPFD and source-sink relations in field and laboratory-grown sunflower. Plant Cell
Environ 17: 443-450
Andrew SPS (1987) A mathematical model of root exploration and of grain fill with particular
reference to winter wheat. Fertil Res 11: 267-281
Asseng S, Keating BA, Huth NI, Eastham J (1997a) Simulation of perched water tables in a duplex
soi!. Proc MODSIM 97, Int Congr on Modelling and Simulation, 8-11 December 1997,
Hobart, Tasmania, Australia, pp 538-543
Asseng S, Richter C, Wessolek G (1997b) Modelling root growth of wheat as the linkage between
crop and soi!. Plant Soil190: 267-277
Asseng S, Keating BA, Fillery IPR, Gregory PJ, Bowden JW, Turner NC, Palta JA, Abrecht DG
(1998) Performance of the APSIM wheat model in Western Australia. Field Crops Res 57:
163-179
Baldwin JP, Nye PH, Tinker PB (1973) Uptake of solutes by multiple root systems from soil. III.
A model for calculating the solute uptake by a randomly dispersed root system develop-
ing in a finite volume of soil. Plant Soil38: 621-635
Bengough AG, Mackenzie q, Diggle AJ (1992) Relations between root length densities and
root intersections with horizontal and vertical planes using root growth modelling in 3-
dimensions. Plant Soil145: 245-252
Berntson GM (1992) A computer program for characterizing root system branching patterns.
Plant Soil 140: 145-149
Berntson GM (1994a) Modelling root architecture: are there trade-offs between efficiency and
potential or resource acquisition? New Phytol 127: 483-493
Berntson GM (1994b) Root systems and fractals: how reliable are calculations offractal dimen-
sions? Ann Bot 73: 281-284
Borg H, Grimes DW (1986) Depth development of roots with time: an empirical description.
Am Soc Agric Eng 29: 194-197
Brown TN, Kulasiri D (1994) Simulation of Pinus radiata root system structure for ecosytem
management applications. Simulation 62: 286-294
Bruckler L, Lafolie F, Tardieu F (1991) Modeling root water potential and soil-root water trans-
port: II. Field comparisons. Soil Sci Soc Am J 55: 1213-1220
Cannell MGR (1985) Dry matter partitioning in tree crops. In: Cannell MGR, Jackson JE (eds)
Trees as crop plants. Titus Wilson,Abbots Ripton, Huntington, 160-193
Cannon WA (1949) A tentative classification of root systems. Ecology 30: 542-548
Chapman SC, Hammer GL, Meinke H (1993) A sunflower simulation model: 1. Model develop-
ment. Agron J 85: 725-735
Clausnitzer V, Hopmans JW (1994) Simultaneous modeling of transient three-dimensional root
growth and soil water flow. Plant Soi! 164: 299-314
142 1. Pages et al.
Colin-Belgrand M, Pages L, Dreyer E, Joannes H (1989) Analysis and simulation of the archi-
tecture of a growing root system: application to a comparative study of several tree
seedlings. Ann Sci For 46 (Suppl): 288-293
Coutts MP (1989) Factors affecting the direction of growth of tree roots. Ann Sci For 46 (Suppl):
277-287
Denison RF, Loomis RS (1989) An integrative physiological model of alfalfa growth and devel-
opment. Div of Agricultural and Natural Resources, Oaklands, California, 73pp
Diggle AJ (1988) ROOTMAP - a model in three-dimensional coordinates of the growth and
structure of fibrous root systems. Plant Soill05: 169-178
Fitter AH (1982) Morphometric analysis of root systems: application of the technique and influ-
ence of soil fertility on root system development in two herbaceous species. Plant CeH
Environ 5: 313-322
Fitter AH (1986) The topology and geometry of plant root systems: influence of watering rate
on root system topology in Trifolium pratense. Ann Bot 58: 91-101
Fitter AH (1987) An architectural approach to the comparative ecology of plant root systems.
New Phytol106 (Suppl): 61-77
Fitter AH (1991) Characteristics and functions of root systems. In: Waisel Y, Eshel A, Kafkafi U
(eds) Plant roots, the hidden half. Dekker, New York, pp 3-25
Fitter AH (1996) Characteristics and functions of root systems. In: Waisel Y, Eshel A, Kafkafi U
(eds) Plant roots, the hidden half. 2nd edn. Dekker, New York, pp 1-20
Fitter AH, Stickland TR (1992) Fractal characterization of root system architecture. Funct Ecol
6:632-635
Fitter AH, Stickland TR, Harvey ML, Wilson GW (1991) Architectural analysis of plant root
systems. 1. Architectural correlates of exploitation efficiency. New Phytoll18: 375-382
Gerwitz A, Page ER (1974) An empirical mathematical model to describe plant root systems.
J Appl Ecolll: 773-781
Grabarnik P, Pages L, Bengough AG (1998) Geometrica! properties of simulated root systems:
consequences for length density and intersection density. Plant Soil200, 157-167
Greene N (1991) Detailing tree skeletons with voxel automata. SIGGRAPH '91 Course Notes on
Photorealistic Volume Modeling and Rendering Techniques, Association for Competing
Machinery, New York
Groot JJR (1987) Simulation of nitrogen balance in a system of winter wheat and soil. Simula-
tion Report CABO-TT, 13. Centre for Agrobiological Research, Wageningen
Hackett C, Rose DA (1972) A model of the extension and branching of a seminal root of barley,
and its use in studying relations between root dimensions. I. The model. Aust J Biol Sci 25:
669-679
Hansen GK (1975) A dynamic continuous simulation model of water state and transportation
in the soil-plant-atmosphere system. I. The model and its sensitivity. Acta Agric Scand 25:
129-149
Henderson R, Ford ED, Renshaw E, Deans JD (1983a) Morphology of the structural root system
of sitka spruce. 1. Analysis and quantitative description. Forestry 56: 121-135
Henderson R, Ford ED, Renshaw E (1983b) Morphology of the structural root system of sitka
spruce. 2. Computer simulation of rooting patterns. Forestry 56: 136-153
Hoogenboom G, Huck MG (1986) ROOTSIMU V 4.0.A dynamic simulation of root growth, water
uptake, and biomass partitioning in a soil-plant-atmosphere continuum: update and doc-
umentation. Agronomy and Soils Departmental Series 109. Auburn University, Alabama,
83pp
Huck MG, Hillel D (1983) A model of root growth and water uptake accounting for photosyn-
thesis, respiration, transpiration, and soi! hydraulics. Adv Irrig 2: 273-333
Jakobsen BF, Dexter AR (1987) Effect of soi! structure on wheat root growth, water uptake and
grain yield. A computer simulation model. Soi! Tillage Res 10: 331-345
Johnson IR, Thornley JHM (1985) Dynamic model of the response of a vegetative grass crop to
light, temperature and nitrogen. Plant Cell Environ 8: 485-499
Jones CA, Kiniry JR (1986) CERES-maize: a simulation model of maize growth and develop-
ment. Texas A&M University Press, College Station, Texas
Jones CA, Bland WL, Ritchie JT, Williams JR (1991) Simulation of root growth. In: Hanks J,
Ritchie JT (eds) Modeling plant and soi! systems. Agronomy Monograph 31, American
Society of Agronomy, Inc., Madison, Wisconsin, pp 92-123
Klepper B, Belford RK, Rickman RW (1984) Root and shoot development in winter wheat. Agron
J 76: 117-122
Kutschera L, Lichtenegger E (1992) Wurzelatlas mitteleuropăischer Griinlandpfianzen. Gustav
Fischer, Stuttgart
Lafolie F, Bruckler L, Tardieu F (1991) Modeling root water potential and soi!-root water trans-
port: I. Model presentation. Soi! Sci Soc Am J 55: 1203-1212
Le Roux Y, Pages L (1994) Developpement et polymorphisme racinaires chez de jeunes semis
d'hevea (Hevea brasiliensis). Can J Bot 72: 924-932
Lungley DR (1973) The growth of root systems: a numerical computer simulation model. Plant
Soi! 38: 145-159
Lynch J, Van Beem JJ (1993) Growth and architecture of seedling roots of common bean geno-
types. Crop Sci 33: 1253-1257
Mandelbrot B (1983) The fractal geometry of nature. Freeman, New York
Monteith JL, Huda AKS, Midya D (1989) RESCAP: a resource capture model for sorghum and
pearl millet. In: Virmani SM, Tandon HLS, Alagarswamy G (eds) Modeling the growth and
development of sorghum and pearl millet. ICRISAT Res Bul112: 30-34
Nielsen CHN, Mackenthun G (1991) Die horizontale Variation der Feinwurzelintensităt in
Waldboden in Abhăngigkeit von der Bestockungsdichte. Allg Forst- Jagd Ztg 162: 112-
119
Nielsen KL, Lynch JP, Jablokow AG, Curtis PS (1994) Carbon cost of root systems: an architec-
tural approach. Plant Soi! 165: 161-169
O'Leary GJ, Connor DJ, White DH (1985) A simulation model of the development, growth and
yield of the wheat crop. Agric Syst 17: 1-26
Otter-Nacke S, Godwin DC, Ritchie JT (1986) Testing and validating the CERES-wheat model in
diverse environments. AGRISTARS Yield Model Development. USDA-ARS Temple, Texas
and IFDC Muscle Shoals, Alabama, 147pp
Page ER, Gerwitz A (1974) Mathematical models, based on diffusion equations, to describe root
systems of isolated plants, row crops, and swards. Plant Soi! 41: 243-254
Pages L (1992) Mini-rhizotrons transparents pour l'etude du systeme racinaire de jeunes plantes.
Application Il la caracterisation du developpement racinaire de jeunes chenes (Quercus
robur). Can J Bot 70: 1840-1847
Pages L,Aries F (1988) SARAH: modele de simulation de la croissance, du developpement et de
l'architecture des systemes racinaires. Agronomie 8: 889-896
Pages L, Pellerin S (1994) Evaluation of parameters describing the root system architecture of
field grown maize plants (Zea mays 1.). II. Density, length, and branching of first-order
lateral roots. Plant Soi! 164: 169-176
144 L. Pages et al.
Pages L, Pellerin S (1996) Study of differences between vertical root maps observed in a maize
crop and simulated maps obtained using a model for the three-dimensional architecture
of the root system. Plant Soi! 182: 329-337
Pages L, Jordan MO, Picard D (1989) A simulation model of the three-dimensional architecture
of the maize root system. Plant Soi! 119: 147-154
Pages L, Chadoeuf J, Kervella J (1992) Modelisation stochastique de la croissance et du
developpement du systeme racinaire de jeunes pechers. 1. Estimation et validation du
modele. Agronomie 12: 447-458
Pages L, Kervella J, Chadoeuf J (1993) Development of the root system of young peach trees
(Prunus persica L. Batsch): a morphometrical analysis. Ann Bot 71: 369-375
Pages L, Le Roux Y, Thaler P (1995) Modelisation de l'architecture racinaire. Plant Rech Dev 2:
19-34
Passioura JB (1996) Simulation models: science, snake oii, education or engineering? Agron J 88:
690-694
Pellerin S (1993) Rate of differentiation and emergence of nodal maize roots. Plant Soil 148:
155-161
field-grown maize plants (Zea mays L.). 1. Elongation of seminal and nodal roots and exten-
sion of their branched zone. Plant Soil 164: 155-167
Pellerin S, Pages L (1996) Evaluation in field conditions of a three-dimensional architectural
model of the maize root system: comparison of simulated and observed horizontal root
maps. Plant Soi! 178: 101-112
Pellerin S, Tabourel F (1995) Length of the apical unbranched zone of maize axile roots: its rela-
tionship to root elongation rate. Environ Exp Bot 35: 193-200
Penning de Vries FWT, Jansen DM, Ten Berge HFM, Bakema A (1989) Simulation of ecophysio-
logical processes of growth in several annual crops. Simulat ion Monographs. Pudoc,
Wageningen
Picard D, Jordan MO, Trendel R (1985) Rythme d'apparition des racines primaires du mals (Zea
mays L.). 1. Etude detaillee pour une variete en un lieu donne. Agronomie 5: 667-676
Porter JR, Klepper B, Belford RK (1986) A model (WHTROOT) which synchronizes root growth
and development with shoot development for winter wheat. Plant Soil 92: 133-145
Prusinkiewicz P, Lindenmayer A (1990) The algorithmic beauty of plants. Springer, Berlin Hei-
delberg New York
Reynolds JF, Thornley JHM (1982) A shoot: root partitioning model. Ann Bot 49: 585-597
Ritchie JT, Godwin DC, Otter S (1985) CERES-wheat: a user oriented wheat yield model. Pre-
liminary documentation. AGRISTARS Publ No YM-U3-04442-JSC-18892. Michigan State
University, MI, 252pp
Robertson MJ, Fukai S, Hammer GL, Ludlow MM (1993) Modelling root growth of grain
sorghum using the CERES approach. Field Crops Res 33: 113-130
Rose DA (1983) The description of the growth of root systems. Plant Soi! 75: 405-415
Savin R, Satorre EH, Hall AJ, Slafer GA (1995) Assessing strategies for wheat cropping in the
monsoonal climate of the Pampas using the CERES-wheat simulation model. Field Crops
Res 42: 81-91
Shibusawa S (1994) Modelling the branching growth fractal pattern of the maize root system.
Plant Soil 165: 339-347
Skiles JW, Hanson JD, Parton WJ (1982) Simulation of above- and below-ground carbon and
nitrogen dynamics of Bouteloua graci/is and Agrophyron smithii. In: Lavenroth WK,
Skogerboe Gv, Flug M (eds) Analysis of ecological systems: state of the art in ecological
modelling. Elsevier, Amsterdam pp 467-473
Spek LY, Van Noordwijk M (1994) Proximal root diameter as predictor of total root size for
fractal branching models. II. Numerical model. Plant Soil164: 119-127
Spek L, Van Oijen M (1988) A simulation model of root and shoot growth at different levels of
nitrogen availability. Plant Soil 111: 191-197
Stapper M (1984) SIMTAG: a simulation model ofwheat genotypes. Model documentation. Uni-
versity of New England, Armidale, 108pp
Stol W, van Keulen H, van Kraalingen DWG (1993) The FORTRAN vers ion of the Van Keulen -
Seligman CSMP-spring wheat model. Simulation Report CABO-TT 30, 112, Wageningen
Agricultural University, Wageningen
Tardieu F, Pellerin S (1990) Trajectory of the nodal roots of maize in fields with low mechani-
cal constraints. Plant Soil 124: 39-45
Tardieu F, Bruckler F, Lafolie F (1992) Root clumping may affect the root water potential and
the resistance to soil-root water transport. Plant Soil140: 291-301
Tatsumi J, Yamauchi A, Kono Y (1989) Fractal analysis of plant root systems. Ann Bot 64: 499-503
Thornley JHM (1977) Root:shoot interactions. In: Jennings DM (ed) Integration of activity in
the higher plants. Cambridge University Press, Cambridge, pp 367-389
Toun! A, Major DJ, Lindwall CW (1995) Comparison of five wheat simulation models in south-
ern Alberta. Can J Plant Sci 75: 61-68
Tsegaye T, Mullins CE, Diggle AJ (1995) An experimental procedure for obtaining input para-
meters for the "ROOTMAP" root simulation program for peas (Pisum sativum 1.). Plant
Soill72: 1-16
Van Keulen H, Seligman NG (1987) Simulation of water use, nitrogen nutrition and growth of
a spring wheat crop. Simulation Monographs. Pudoc, Wageningen
Van Noordwijk M, Purnomosidhi P (1995) Root architecture in relation to tree-soil-crop inter-
actions and shoot pruning in agroforestry. Agrofor Syst 30: 161-173
Van Noordwijk M, Spek LY, De Willigen P (1994) Proximal root diameter as predictor of total
root size for fractal branching models.1. Theory. Plant Soil164: 107-117
Voit EO, Sands PJ (1996) Modeling forest growth 1. Canonical approach. Ecol Model 86: 51-
71
Von pfeil E, Hundertmark W, Thies FD, Widmoser P (1992) Calibration of the simulation model
"ceres wheat" under conditions of soils with shallow water table and temperate climate.
Part 1: Limitations in the applicability of the original model and necessary modifications.
Z Pflanzenernăhr Bodenkd 155: 323-326
Waisel Y, Eshel A (1991) Multiform behavior of various constituents of one root system.
In: Waisel Y, Eshel A, Kafkafi U (eds) Plant roots, the hidden half. Dekker, New York,
39-52
Wessolek G (1993) Einfluss von Klimaănderungen auf den Bodenwasserhaushalt (regionale Fall-
studien). Mitt Dtsch Bodenkdl Ges 69: 289-292
Wessolek G, Găth S (1990) Estimation of root density in modelling nutrient requirements. In:
Proc 22th Colloq Int Potash Institute, Soligorsk, USSR
Williams JR, Jones CA, Dyke PT (1984) The EPIC model and its application. Proc Int Symp on
Minimum Data Sets for Agrotechnology Transfer, Patancheru, India, 21-26 March 1983,
pp 111-121
146 L. Pages et al.: Modelling Root System Growth and Architecture
Further Reading
Jones CA, Bland WL, Ritchie JT, Williams JR (1991) Simulation of root growth. In: Hanks J,
Ritchie JT (eds) Modeling plant and soi! systems. Agron Monogr 31: 92-124
Klepper B, Rickman RW (1990) Modeling crop root growth and function.Adv Agron 44: 113-132
Lynch J (1995) Root architecture and plant productivity. Plant Physiol109: 7-13
CHAPTER 5
Sampling Strateg ies, Scaling, and Statistics

A.G. BengoughI, A. Castrignan0 2, L. Pages\ and M. van Noordwijk4
1 Scottish Crop Research Institute (SCRI), Invergrowie, Dundee, DD2 5DA, UK
2 Instituto Sperimentale Agronomico, Via C. Ulsiani 5,70125 Bari, Italy
3 INRA, Unite de Recherches en Ecophysiologie et Horticulture, Site Agroparc, 84914 Avignon
Cedex 9, France
4 ICRAF, JI CI FOR, PO Box 161, Situ Gede, Sindang Barang Bogor 16680, 16001 Bogor,
Indonesia
CONTENTS
5.1 Introduction 147
5.2 Root System Structure and Spatial Distribution 150
5.2.1 Systematic Trends 150
5.2.2 Clustering 152
5.2.3 Anisotropy 152
5.3 Choosing a Measurement Technique 154
5.4 Sampling Design 157
5.4.1 Architecture-Guided Sampling 157
5.4.2 Measuring the Distribution of Root Length Density 158
5.4.3 Sample Location 159
5,4,4 How Big a Sample, and How Many Samples? 160
5.5 New Developments 165
5.5.1 Using Root Growth Models to Improve Sampling Strategies 165
5.5.2 Applications of Geostatistics to Study Spatial Variability 166
5.5.2.1 Semivariograms 167
5.5.2.2 Kriging and Cokriging 169
5.5.2.3 Conditional Simulations 169
5.6 Conclusions 170
References 170
S.l Introduction
Many experiments aim to understand how some aspect of the aerial and soil envi-
ronment of a plant influences its growth. Considerable biomass is often allocated
to the root system, and it is the roots that absorb most nutrients and water

148 A.G. Bengough et al.
(Russell 1977). The property of the root system which is most appro-priate to
measure depends on the objective of the experiment. Root length should be mea-
sured to calculate the inflow rate of water and nutrients (see Chaps. 6, 13, 14). Root
dry mass indicates the carbon allocation to the root system. Branching patterns
and the number and lengths of each class of root, together with the distribution
of root diameters gives a more complete pic ture of root architecture, but requires
a large investment oflabour. Detailed information on root architecture may be of
interest in constructing mathematical models of root growth (see Chap. 4), in
comparing the structure and function of root systems of different species or
genotypes (e.g. Fitter and Stickland 1991), and in evaluating root responses to
environmental conditions (e.g. Robinson 1994).
It is not easy to design an appropriate scheme to adequately sample the root
system, because of the complex and variably branched structure of roots (see
Fig. 5.1), the variability of root distribution in space and, most importantly, the
opaque growing environment, which means the experimenter works continu-
ally"in the dark". Root researchers face the peculiar problem of only being able
to see or sample a small fraction of the root system at any time, using sampling
procedures that are often destructive (e.g. see Chaps. 6 and 7). The design
of sampling schemes for roots that are adequate and appropriate for different
situations has received little study. This is despite the large effort that has been
directed, and sometimes misdirected, towards sampling roots in the field. Many
studies have considered different techniques for measuring roots, but the
problem of where to sample, and how big a sample to collect, is rarely consid-
ered (van Noordwijk et al. 1985).
The "problem" of large variation between replicate samples of roots is
widely known. This has often been seen as a nuisance, making it necessary to
sample many times to obtain an accurate estimate of the mean. However, it is
becoming recognized that the variation itself is worthy of study. This is also
being seen in the soil sciences, where interest is increasing in the study of soil
heterogeneity, and the resources that the root system is trying to exploit (e.g.
Fig. 5.1. Root systems vary widely in architecture and in spatial distribution. Examples are
shown of: a A 54-year-old Austrian pine (Pinus nigra) in sandy soil in the Netherlands - roots
weighed a total of 30 kg and extended about 5 m from the base of the tree. b An excavated hor-
izontal lateral root of a 3-year-old rubber tree (Hevea brasiliensis) in sandy soi! in the Ivory
Coast. c The thick vertical taproot and horizontallateral roots near the base of the trunk of a
3-year-old rubber tree. In deeper horizons the lateral roots are finer and less horizontal. d An
excavated nodal axis of maize (Zea mays L.) at silking, in a loamy clay soi! at Colmar. e Maize
grown for 690 Kd (above base temperature 6 ac; seven visible leaves present) in aeroponics. f and
g Root systems ofbean (Phaseolus vulgaris) and rye-grass plants grown in root boxes for 2 weeks
under glasshouse conditions. Photos kindly supplied by Drs. F. Egas (a), Yannick Le Roux (b,c),
S. Pellerin (d,e), and F. Meijboom (f,g)
5 Sampling Strateg ies, Scaling, and Statistics 149
Fig.5.1a-g
Jackson and CaldwellI993a,b).A plant with a root system that explores an area
of soil with a 50 cm radius may experience a variability in nutrient distribution
equivalent to that across a 100 m2 field plot (Jackson and CaldwellI993b). A root
system that is distributed non-uniformly can extract water and nutrients much
faster from areas where the root length density is large. In contrast, resources
may be largely inaccessible in regions of the soil that are rooted sparsely. The
superposition of soil and root data in space promises to give new insights into
how roots explore the soil environment.
The aim of this chapter is to guide the reader towards designing effective
schemes for sampling roots in the field. Specific statistical analyses appropriate
to particular methods are discussed in the appropriate technique chapter (e.g.
cohort analysis in Chap. 9). We examine the spatial properties of root systems,
and the problems associated with the techniques for measuring them. The
coefficients of variation of root systems of different crops are detailed to
help determine the likely number of samples required. The potential uses of
root growth models in experimental design are described, together with how
geostatistics can be used to study spatial variability.
5.2 Root System Structure and Spatial Distribution
In the following sections, we identify three main properties associated with the
spatial distribution of root systems: (1) systematic trends in the root length
density of root systems - i.e.large scale (several decimetres) gradual variations;
(2) root clustering - variations at a small scale (several centimetres); and (3)
root anisotropy - the non-uniform directional distribution of roots.
5.2.1 Systematic Trends
The form of a root system depends both on the plant genotype and on its
interaction with the environment. Species may be grouped into classes that have
broadly similar forms of root system. Monocotyledons generally form dense
fibrous root systems. The seminal and nodal axes of cereals and other grasses
produce several successive orders of fine lateral roots. The total root length of a
single 16-week-old winter rye plant (Secale cereale) was measured to be more
than 500km long (Dittmer 1937). Dicotyledons form root systems often initially
centred around a tap root from which many lateral roots emerge (e.g. sugarbeet,
Beta vulgaris). Trees may also have a taproot system; a more surface "plateroot"
system consisting of horizontal main root axes; or a "heartroot" system which
consists of more evenly distributed main axes that split into smaller roots (Wilde
1958). Certain species of tree (e.g. Populus tremuloides) grow new shoots from
buds on roots in the surface layers of soil: Such groups of interconnected trees
form very extensive root systems, exploring huge volumes of soil.
The depths and distributions of root systems vary enormously according
to species, climatic zone, and soil type. In a comprehensive review of rooting
depths around the globe, maximum rooting depths varied between <0.3 m in
some tundra species, to 68 m for Boscia albitrunca in the Kalahari desert
(Canadell et al. 1996). Maximum rooting depths averaged about 9m in the
desert, with only about half of that root biomass being in the top 0.3 m (Jackson
et al. 1996). In contrast, tundra vegetation averaged 0.5 m maximum rooting
depth, with 80-90% of root biomass being in the top 0.3 m. When plants are
grouped across biomes by functional group, trees, shrubs and herbaceous plants
had global average rooting depths of 7 m, 5.1 m, and 2.6 m respectively.
The length of root per unit ground surface are a has been reviewed briefly
by Newman (1969). Perennial grasses had the densest root systems, with
360-3400 cm/cm2 of root in the top 10 cm layer of soil. This corresponds to 36
to 340cm/cm 3 volume of soil. Cereals and non-gramineae herbs had smaller
root lengths per unit area than grasses, but generally greater root lengths per
unit area than woody shrubs and trees.
The spatial distribution of roots depends on the spacing and arrangement
of individual plants in an ecosystem (van Noordwijk et al. 1985). In grassland,
the plants are spaced closely in all directions, so that the root length density
varies mainly in one dimension with depth. In cereaI crops planted densely in
rows, the pattern of root length density is essentially two-dimensional. As dis-
tance between plants increases, the pattern becomes three-dimensional. In
agroforestry systems, trees, often arranged in rows, are interplanted with grass
or an arable crop, leading to two interacting root system structures. Large vari-
ations in the height of the soil surface, as occurs across the ridges and furrows
of the potato crop, create even more complex three-dimensional patterns (e.g.
Parker et al. 1991). Natural ecosystems in which plants are randomly spaced
or clumped on an undulating landscape may mean that the plants have to be
treated as isolated individuals or small groups, without any of the symmetry
afforded by agricultural practice.
Root systems originate from seeds, rhizomes or shoots that are near to the
surface of the soil. The root system extends and branches gradually, extending
away from this origin. This creates a pattern of root length density that is ori-
ented centrifugally. Root length density decreases with increasing distance from
the stern base, both in the horizontal and in the vertical directions (Gajri et al.
1994). Gradients in soil characteristics, such as soil-water content, soil strength,
and temperature, interact with the development of the root system (e.g. Tardieu
and Pellerin 1990). Such gradients often restrict the spread of the root system
with depth but, in practice, it is very difficult to separate the effects of genotype
and environment on root distribution.
The gradual decrease of root length density with depth is a feature that is
general to most root systems. A negative exponential function has been used to
model the relationship between root distribution (measured as length density
or root mass) and depth for vegetable crops, cereals, and grasses (Gerwitz
and Page 1974). The exponential model accounted for at least two thirds of the
variation in 70 of 10 1 cases.
The variation in root length density across a crop has been measured by
several authors (see Fig. 5.2; van Noordwijk et al. 1985; Gajri et al. 1994; Pellerin
and Pages 1996). The variation was greatest near the surface for crops grown
in widely spaced rows. The variation was small in established forest stands
(Persson 1978). The horizontal variation in root length density tends to de crease
with increasing depth, because the "fan" shape of the root systems tends to
smooth out fluctuations between neighbouring plants.
5.2.2 Clustering
Large variations in root length density occur across distances of several cen-
timetres because of root clustering or clumping. Roots cluster in space because
of their branched connections with daughter roots necessarily concentrated
close to their mother root (Logsdon and Allmaras 1991). The constraint caused
by the connection of roots, and the consequent aggregation of roots in space
is exacerbated in root systems that have many short branches, with short
link lengths between branches (e.g. Varney et al. 1991). A large percentage of
the total root length is segmented into roots shorter than 10 cm, even in root
systems that explore a total volume of several cubic metres (Lyr and Hoffman
1967). Such short roots are generally much thinner than the main axis, and
can represent an economical way for the plant to increase the surface area
of the root-soil interface, and perhaps exploit a nutrient source before its
competitors.
Clustering of roots can also be induced by changes in the local environ-
ment of the root. Roots grow preferentially in cracks or biopores in compacted
soil, such that the soil space is colonised irregularly (Wang et al. 1986; Tardieu
1988; Logsdon and Linden 1992). Patches of nutrients or residues from previous
crops may also cause local increases in root length density, and this localised
branching has become the subject of many studies (see review by Robinson
1994).
5.2.3 Anisotropy
Anisotropy is the non-uniform directional distribution of roots. It is the

result of both morphogenetic and environmental factors, which interact during
5 5ampling Strateg ies, Scaling, and Statistics 153
'"'E
u 0.8
~
.~ 0.6
c
(])
D
.c 0.4
~
~
O 0.2
li
o
O 2 4 6 8 10 12
Distance from row (cm)
1.5
(b) maize
<?'
E
u
E
.s
~
Cii
c
(])
D
.c
OJ 0.5
c
~
~
O
o
Fig. 5.2. Root length density at different a:
2O-30crn A_A
distanees from rows of erops of a wheat O
~o
(row spaeing 22 em), and b maize (row O 5 10 15 20 25 30

spacing 60em). (Data from Gajri et al. 1994) Distance from row (cm)
the development of the root system. Tropisms, and especially gravitropism,

orient root growth in defined directions, according to the root type: the
orientation is vertical for orthogravitropic roots, inclined at an angle for
plagravitropic roots, and is horizontal for diagravitropic roots (Coutts 1989).
The pattern of root branching also makes some root directions more likely. In
some diarch surface roots of trees, for example, branching takes place only in a
horizontal plane, resulting in a typically anisotropic pattern (Wilson 1964).
Many soil properties (e.g. temperature, water content, strength) and mor-
phological features, are structured in space, and contribute to root anisotropy
(Logsdon and Linden 1992). Compact layers of soil, such as a tillage pan,
can deflect roots horizontally for considerable distances, leading to anisotropy.
This anisotropy is particularly marked when the main tap root of a plant
encounters an impenetrable layer, as illustrated in Bennie (1996), for the tap
root of cotton.
5.3 Choosing a Measurement Technique
The main steps in deciding on an appropriate sampling scheme are diseussed

in the following seetions (see Fig. 5.3). Details of the various root system prop-
erties and measurement teehniques are given in Table 5.1, with referenee to
appropriate ehapters where the teehniques are deseribed.
It is first essential to foeus on a clear researeh question, so that resourees
are used to maximum effeet. Root sampling and measurement is always labour
intensive, and the workload may be deereased by eoneentrating measurements
on particular layers of the soil, or on particular samples. In the time that it takes
to measure aeeurately the root length density in one sample eontaining 5 em
em-3 of root, several samples eontaining 0.5 em em-3 may be measured. To deter-
mine, for example, the volume of soil from which a root system ean extraet water
.
IResearch question?- see chapter 11
Property of root system to be measured?
• accuracy required?
• spatial information required?
Technique appropriate?
• depends on a priori knowledge of root system structure.
(see Table 5-1, and chapters 1,6, 7, and 8 in this book)
Sampling scheme?
• how many samples? (example coefficients ofvariation in Tables 5-2 & 3)
• location of samples? (gradients in field, spatial arrangement of plants,
randomisation)
• when to sample?
Data analysis and statistics?

(see technique chapters 6, 7, and 8)
Interpretation?
• re-assess experimental design, if experiment is to be repeated
Fig. S.3. Flow diagram of steps in deciding on experimental technique and sampling procedure
VI
VI
III
3
'2.
:r
le
Table S.l. Root system properties, measurement techniques, and their associated advantages and problems
la
iil
Property Technique and description Advantages Problems iD
le
iDo
!:!'
Root length density Auger (Chap. 6). Soi! cores sampled. Roots Most accurate method for root Labour intensive. Many (e.g. 30%) 11'1
in cores washed from cores in the lab length density roots lost during washing. Sample !.
size determined by auger :i"
I@
III
:::1
Root end distribution Trench wall and root mapping techniques Relatively fast. Good spatial Many fine roots go undetected. Do
in excavated planes (Chap. 7). Vertical or horizontal planes are resolution Problems converting la
III
excavated in the soil to expose root ends. Root measurements to root length ..!a-
end locations are plotted on transparent sheet, density n'
In
and number of ends per unit are a calculated
Root end distribution Core-break technique (Chap. 7). Root ends Relatively fast Many fine roots go undetected.
in excavated cores are counted on the face of a cut soil core Problems converting
measurements to the root length
density. Sample size determined
byauger
Root end or length Minirhizotron (Chap. 8): roots are observed Non-destructive, in situ. Equipment expensive. Root grow
distribution on cylindrical with an endoscope around a transparent Can monitor temporal preferentially along tube walls.
surface cylinder buried in the soi! variation Poor correlation with root length
density
V1
V1
-
......
V1
0\
Table S.1. (cont.)
Property Technique and description Advantages Problems
Root diameter and Minirhizotron (above) (as above) (as above)

branching patterns
Rhizotron (Chap. 8):
Roots observed at a soi!-glass interface
(as for minirhizotron, above) Expensive facility. Roots growth
preferentially along interface.
Pinboard (Chap. 6): a board containing nails Spatial dis tribut ion of roots Roots disturbed and lost during
arranged in a regular grid is pushed into an in soi!. Interactions between washing
excavated plane. The block of pinned soi! is then species
then washed to expose the roots
Whole root system Excavation and washing (Chap. 6) - with woody Three-dimensional root Much fine root materiallost.
architecture root systems, much of the soil may be washed or distribution Roots moved dur ing excavation
or blown from the root system to expose the
main roots
Root turnover Rhizotron/minirhizotron (Chap. 9) Non -destructive Preferential root growth along
interface. Equipment expensive.
Difficult to determine live and
o>
t>:I
<b
dead roots
~.,:
ag.
~
~
S sampling Strateg ies, scaling, and statistics 157
effectively, it may be more efficient to concentrate on measuring many samples

that are rooted sparsely, than on measuring a few densely rooted samples accu-
rately. Under some circumstances it may be appropriate to avoid measuring root
growth and, instead, measure some function associated with roots: Gregory
et al. (1978) monitored the advancing rooting front in wheat by measuring the
movement of the drying front using a neutron probe.
Measurement of washed samples is very rapid using image analysis
techniques (discussed in Chap. 10), but much time is required to spread root
samples before analysis. Simple visual comparison of washed samples with
standards of known length can provide faster but more approximate measure-
ments of root length, with less preparation of samples than is required normally
for image analysis (see Chap. 6 for visual comparison techniques).
It is impossible to set rigid rules for balancing alI aspects of experimental
design. The best approach depends crucialIy on the particular research ques-
tion, the priorities and resources of the investigator, and a sound knowledge of
the current literature.
5.4 Sampling Design
Two approaches based on different concepts are possible: firstly, we may

excavate a given root system, individual root, or part of a root. We shall call this
"architecture-guided" sampling. Secondly, we may determine the density
distribution of the root system in some volume of soil. We shall consider these
two approaches separately. In both cases, good prior knowledge is essential of
the way that root systems are structured and distributed.
5.4.1 Architecture-Guided Sampling
It is possible to measure the length and branching density of individual roots,

by gradually folIowing and excavating them, starting from the base of the shoot
(e.g. Tardieu and Pellerin 1990; Pages and Pellerin 1994; Fig. 5.1b,d). This type
of sampling requires an appreciation of root type, morphology, and behaviour
(e.g. Waisel and Eshel1991). Various criteria can be used to classify roots during
this sampling procedure (e.g. Le Roux and Pages 1994). Roots are often classi-
fied by some developmental criteria, such as the origin of the roots (seminal,
nodal, branch), or branching order (primary axis; first second or third order
lateral). Roots in some developmental categories continue to appear through-
out the life of the plant. Age is an important additional criterion for classifying
the sub-population, because it affects both root morphology and root physiol-
ogy. For example, the proportion of primary lateral roots of maize that can be
classified as determinate or normal, depends on the age of the parent root to

which they are attached (Varney and McCully 1991). This information can be
was obtained by careful excavation and sampling of individual nodal axes,
followed by histological examinat ion.
Spatial and temporal variation in the soil is superimposed on the endoge-
nous variability between roots. In the field it is often difficult to assess the vari-
ation in soil conditions that a root has experienced during its existence. By
studying the spatial distribution of a root system, it should also be possible to
identify regions of soil of particular importance. For example, regions of root
proliferation, or places roots have failed to penetrate could be selected for
further study.
For trees and other perennials root architecture can also be approached by
a study of fractal branching patterns. The basic assumption is that secondary
thickening of roots is based on the demands for transport and thus at any point
in the branched system transport capacity is proportional to the amount of fine
roots distal to it. The proximal root diameter of a tree root (at the stern base)
may contain enough information to reconstruct the total branched structure -
provided that a number of simplifying assumptions hold, and that a number of
parameters (which probably depend on tree species, but can otherwise be
treated as constants) are known. The basic assumption of a "pipe-stern" model
is that cross-sectional area (or the sum of diameter squares) is conserved during
a branching event. Van Noordwijk et al. (1994) and Spek and van Noordwijk
(1994) relaxed this assumption by introducing a proportionality factor alpha,
but assumed that alpha is independent of current root diameter.A second para-
meter, q, describes the share of the largest "daughter" root in the total sum of
squares of"daughters". To test the assumptions, one has to trace individual roots
and take measurements of diameters before and after branching points (Van
Noordwijk and Purnomosidhi 1995), although this is obviously a labour inten-
sive process. If regression of alpha and q against diameter does not reveal any
dependency, fractal (self-repetitive) models can be used. To re-construct three-
dimensional structures of root systems, additional information on link ("inter
node") length and branching angles is needed. Spek (1997) has since developed
a visualisation routine by using a model originally developed for complex mol-
ecules. The overall prospects for this approach are relatively"quick and not-too-
dirty" statements about tree root distribution, which can, if necessary, be backed
up by detailed destructive sampling, but which can at least be used to specify
sampling schemes in situations where tree roots are important.
5.4.2 Measuring the Distribution of Root Length Density
The root length density can be measured by separating the roots from a soil
sample of known volume. Another method of assessing density relies on count-
ing the number of intersections that roots make with a surface of known area
in the soil. The properties measured, with a brief description of the techniques,
are listed in Table 5.1, together with some advantages and problems associated
with the techniques. Fuller information about each technique is given in the
appropriate chapters of this book as indicated in Table 5.1.
None of the techniques are entirely satisfactory: the auger technique is the
standard method for measuring root length density although, using this tech-
nique, up to a third of the fine root length may be lost dur ing washing. The root-
mapping technique is better for measuring the spatial distribution of individual
roots. It is possible theoretically to calculate the root length density from the
density of root intersections with a plane, although corre1ations obtained from
field data suggest that a large proportion of the fine roots are often overlooked
using the root -mapping techniques, leading to underestimates in the root length
density - this is discussed in Bengough et al. (1992).
5.4.3 Sample Location
When designing any field experiment, it is important to assess the major sources
of variation in data. Standard texts on experimental design discuss these
sources of variation (e.g. Pearce 1983), which can be divided into patterned and
non-patterned sources. Factors, such as slope, soil fertility, soil depth, shading,
and variation in soil texture, are sources of patterned variability, and their
relative importance will vary with weather conditions between seasons. Much
more information is now becoming available on the spatial variation of yield in
arable crops, due to the advent of precision agricultural equipment, which
makes use of global satellite position references (Robert et al. 1996). The yield
data from such systems gives an indication of plant performance, but there is
no fixed relation between yield and root system development. Sources of non-
patterned variability include errors such as loss of root material during washing
and measurement.
When sampling real root systems, it is not possible to recover the whole
root system of individual plants. Special consideration must be given to decid-
ing which part of the root system should be sampled. Root systems of neigh-
bouring plants are often intermingled, although this depends on the species
concerned: Nelson and Allmaras (1969) found that maize roots intermingled
with neighbouring soybean roots to a much greater extent than with neigh-
bouring roots of the same species. It is possible to define a "unit soil area" for
crop plants that are spaced regularly (Fig. SA, after van Noordwijk et al. 1985).
The soil below each unit area is expected to contain a total root length equal to
the mean root length per plant. Many of the roots within the unit area may
be10ng to neighbouring plants but, similarly, an equal number of roots from the
plant may have extended outside the area.
Row midpoint
• • •
• • •
• •
Unit soil area for
le a single plant
• .1-_._---_.
•
--- ---
i • Each unit area con be
split into four symmetrical
• • • areas
• •
Fig. 5.4. Plan view showing unit area for single plants, represented by black circles, in a row
crop. By symmetry, each unit area consists of four smaller representative areas. (After van
Noordwijk et al. 1985)
The unit area can be divided into four equal parts which are equivalent,
because of symmetry. The quarter area represents the fundamental unit in
which the distribution of root length density should be studied. The most
appropriate sampling schemes have been modelled for a range of crops (Fig.
5.5; after van Noordwijk et al. 1985). The assumptions of root distribution that
underlie the choice of these sampling schemes are discussed in detail in van
Noordwijk et al. (1985). Systematic trends in the root length density can bias
the estimates of mean root length density, if the way that the quarter area is
sampled over-represents either the dense or sparsely rooted areas (Fig. 5.5).
Simply averaging the root dry weight in the row and the inter-row of cereals can
overestimate the total root dry weight by as much as 30%.
5.4.4 How Big a Sample, and How Many Samples?
Clustering of roots, and the associated variability in root length density is an

important characteristic of the root system (Tardieu 1988; Logsdon and
Allmaras 1991). The size of the sample taken using an auger determines
the minimum scale on which the variat ion of the root length density will be
detected. Clustering increases the variance between samples and so, to achieve
a given precision, more samples are required.
The ave rage root length density measured for two populations can be com-
pared using a T-test, provided that the root length density data are normally
5 Sampling Strategies, Scaling, and Statistics 161
Grassland - alilocations
valid equally
C}XJJ)
Anyscale
Cereai row crops - central auger
has double the weighting of outslde
samples (20 cm inter-row distance)
O lOcm, row midpoint

Sugar beet (50 cm infer-row distance,
30 cm within row)
o ~-L__~-L__~~
25cm, row midpoint
Potato - most practical scheme
samples top, middle and inter-row,
at and between planfs (74 cm inter-row
distance)
Fig. 5.5. Plan view of locations where samples should be taken in grassland and in crops of
cereals, sugar beet and potato. Black circles represent the plants, open circles represent sampling
locations using a cylindrical auger. (After van Noordwijk et al. 1985)
distributed, and that the variance for the two data sets is similar (see Box 5.1).
The number of replicates required to give a good chance of finding a signifi-
cant difference between treatments is discussed in Box 5.1 for a T-test.
Approaches to estimating sample number for more complex experimental
designs, for example involving blocking, are discussed in Mace (1964, especially
Chap. 3) and Cochrane and Cox (1957, especially Chap. 2).
If the assumptions associated with parametric statistics are invalid (e.g. the
quantity is not normally distributed), it may be possible to transform the data
so that the transformed values can be tested. Failing this, or for rank or root
count data, it is necessary to use non-parametric statistical tests, such as the Chi
squared test. These tests can be relatively simple to perform, although they use
less information than the parametric tests, and so are weaker.
In planning an experiment it is necessary to have some prior estimate of
the degree of variation that is expected. In Tables 5.2 and 5.3 the coefficients of
variation are listed for measurements made using the auger technique from
crops grown under a variety of conditions. In the crops studied the coefficient
BOX 5.1. Number of Samples Required for T-test

The number of replicates that is required to give a 50% chance of
detecting the difference between two means is shown in Fig. 5.6. More
replicates are required to detect small differences between treatments, or
for populations with a large coefficient of variation. For example, at least
25 replicates are needed to detect a difference of 22% between two
means, if the coefficient of variation is 40%. To detect a difference of just
10% between two means requires more than 120 replicates, if the
coefficient of variation is 40%. More details of T-tests are given in
standard statistical texts (e.g. Sanders 1990)
180
Coefficient of variation
25 33 40 50 60 75
100
0L ----=-----=-----=----=
-----=-
----:=-~~~~
5 10 15 20 25 30 35 40 45 50
Difference belween means (%)
Fig. 5.6. Plot of the number of replicates required to give a 50% chance of distinguishing
between two means in a two-sided I-Iesl at 95% significance level. (After van Noordwijk el
al. 1985)
of variation increased with depth of sampling (Tables 5.2 and 5.3). The reason
for this may be partly associated with the sparse rooting of these deeper layers,
with relatively few main axes present, surrounded by associated clusters of
lateral branch roots (Grabarnik et al. 1998). Another contributing factor is
changes in the soil structure with depth - in compacted subsoils, roots may be
confined to continuous cracks and biopores in the soil to a much greater extent
(e~g. Ehlers et al. 1983).
The coefficients of variation for root dry weight in auger samples from
grassland are typically between 30 and 50% (Table 5.3). There is considerable
variation between the coefficient of variation of root dry weights measured in
Table 5.2. a Coefficients of variation for root length density from Kueke et al. (1995). Samples
were taken from fields of sugar beet, wheat and rye, using an auger (6.5em diameter by 15em
long) from soils of different texture. b Coeffieients of variation for the root length density of
maize. Samples taken using an auger (5.1 em diameter by 15 em deep)
a
Depth (em) Coeffieients of variation (%)
Sanda Loama Clat
0-15 12 27 21
15-30 28 8 54
30-45 76 12 57
45-60 129 60 41
60-75 74 71 69
75-90 113 35 60
a Sugar beet followed by winter wheat.

b Sugar beet followed by rye.
Depth (em) Coefficients of variation (%)
45 Days after emergenee 59 Days after emergenee
15 42 30
30 42 22
45 33 49
60 71 61
75 54 45
90 69 53
different studies - they average 38% for studies j to 1 in Table 5.3, but 59% for
mI and nI. Data from samples taken in the row should not be pooled with data
from samples taken between the rows, as the root samples represent distinct
populations. This is shown by the increased variance when the data is pooled
(see columns headed m3 and n3 in Table 5.3). The root weight is large in the
surface layer of soil immediately around the stern base. The ratio of the mean
root dry weight in the row to that between rows is generally bigger than for the
ratio of root length densities. This is because the root mass per unit length
is normally greatest in the topsoil at the base of the stern. The coefficient of
variation for root length density of cereals is often between 30 and 70%
(Tables 5.2a,b), depending on the particular study.
......
0\
H>-
Table 5.3. Coeffieients of variation of root dry weight in auger samples of grassland and eereals.' (After van Noordwijk et al. 1985)
Depth (em) Grassland Cereals
Ref a b e d e f g h k mI m2 m3 nI n2 n3
Samples 100 100 20 20 20 20 20 50 50 25 25 20 4 4 8 47 32 103
Diameter 7 4 7 7 7 7 7 4 4 7 7 7 7 7 7 105 36 70
0-5 34 30 51 45 41 29 43 30 43 41 89 100 36 31 33
5-10 33 38 34 36 33 48 29 37 83 56 73 42 44 43
10-20 30 45 44 30 55 36 50 34 27 47 40 32 37 63
20-30 36 41 40 38 75 43 56 31 28 64 65 59
30-40 35 41 55 38 53 35 38 36 37 43 45 69 46
40-50 35 41 52 31 48 47 59 46 31 35 29 78 45
50-60 44 44 54 54 76 51 53 39 35 50 54 46 62
60-70 100 56 39 56 47 48 106
70-80 85 75 46 53 53 43 125
80-90 76 61 63 72
90-100 76 44 50 54
a The key to column headings is as follows: a, b homogeneous grassland (1949); e, d young grassland (at Gilze, 1966); e, f same fields as e and d, 4 years
later (June 1970); g established grassland (1976); h, i established grassland: root dryweight and root eounts (estimates from Sehuurman and Knot 1957); :>-
j, k oats: root dry weight and root eounts (estimates); 1 winter wheat on eraeking day soil (Biddinghuizen, May and June 1977); m, n spring wheat on 0
t:C
(!)
day loam and sandy loam, respeetively (Ulrum 1957); mI, nI samples in row; m2, n2 samples between rows; m3, n3 equal number of row and between i:I
C/Q
row samples eombined. o
C/Q
'::r"
~
~
5 sampling Strateg ies, scaling, and statistics 165
5.5 New Developments
5.5.1 Using Root Growth Models

to Improve 5ampling 5trategies
Schemes for sampling root systems must be designed carefully to obtain

good estimates of the total dry weight or root length density in a layer of soil
below a crop. Schemes used traditionally for cereal crops, for example, can
result in a bias of up to 30% in the total root dry weight (van Noordwijk et al.
1985). Experiments that compare different sampling schemes are labour
intensive and subject to experimental errors that are difficult to quantify and
are associated with the excavation, washing and measurement of roots. An
alternative approach, which is complementary to experimental comparisons, is
to use a model of root distribution based on existing knowledge. Examples
are becoming more common in the literature of using models to evaluate
the sampling procedure for root systems, and this is a promis ing area in
experimental design.
The types of model used range from simple exponential functions describ-
ing root distribution (van Noordwijk et al. 1985) to simulations of biomass
fluctuation with time (Singh et al. 1984), and sophisticated simulations of root
architecture in three-dimensional space (Nielsen et al. 1997; Pages and
Bengough 1997; Grabarnik et al. 1998). The basic approach using these models
is the same, although the sophistication and applications of the models may
vary: the model is used to generate a theoretical root distribution in space or
time. This root distribution is then sampled or measured in a particular way, to
simulate some sampling procedure. The simulation can be repeated many times
very rapidly, as it is performed numerically on a computer. The sensitivity
of sampling schemes to changes in the root distribution can be quantified by
systematically varying model-input parameters.
The effect of sample timing and frequency on estimates of root biomass
production was investigated using a model of root biomass for prairie grasses
(Singh et al. 1984). The model was used to simulate variations in biomass with
time, and the effects of sample timing, variability, and replication were studied.
The estimate of biomass production was compared using two methods of cal-
culation: firstly, from the difference between maximum and minimum biomass
values and, secondly, from the summation of increments in biomass across the
season. Estimates of biomass production were between 1.8 and 7 times greater
than the actual production, suggesting that existing sampling schemes were
inadequate. The use of a model allowed the effects of sample timing and
replication to be investigated in more detail than is practical in most field
experiments, although the details of the study were controversial (Lauenroth
et al. 1986; Vogt et al. 1986).
Relations between root length density and root intersections with planes
have been investigated with models that simulate root architecture in three-
dimensions (Bengough et al. 1992). The use of minirhizotron tubes to measure
rooting depth in a maize crop has been simulated using a three-dimensional
model of root architecture of a small plot of 51 maize plants (Pages and
Bengough 1997). It was shown that the maximum rooting depth measured using
minirhizotrons was very variable and could strongly underestimate the true
rooting depth. Underestimation was a particular problem if the tube radius
was smaller than 3 cm, and when the tube was close to the vertical. In this
simulation study, however, the interactions between the growing root and the
tube were not modelIed. Some aspects of root growth and orientation along-
side the minirhizotron tube may be modified by the presence of the tube itself,
but there was insufficient detailed information on root-tube interaction to
enable a realistic model of the interaction to be developed.
The fractal dimensions of two-dimensional projections of root systems
have been compared with the three-dimensional fractal dimension of simulated
bean root architecture (Nielsen et al. 1997). The model showed that the three-
dimensional fractal dimension differed from that in two-dimensions, suggest-
ing that the washing and flattening of the root system is not an acceptable way
of measuring the frac taI dimension of root systems in situ. The use of root
intersection data with horizontal and vertical planes was found to give accurate
estimates of the fractal dimension in three-dimensions, suggesting that this is
a much better way of characterising root distribution in soil.
5.5.2 Applications of Geostatistics to Study Spatial Variability
Geostatistics is a relatively new technique that can be used to study the spatial
variation of roots in a particular depth layer across a field site. It can also be
used to study spatial variation in root systems with depth, although any sys-
tematic trends in vertical root distribution must first be subtracted. A major
disadvantage of geostatistical techniques is that large numbers of samples
(typicalIy >100, but sometimes 300 or more; Jackson and Caldwell 1993b) are
required, each at a known location, and at a range of separations.
The traditional statistical tests used in biology as sume that the data in the
populations being tested have the same distribution, and that each datum is
independent of alI other data. Roots form interconnected branched structures
that may be correlated spatialIy and temporalIy: the presence of a root in a
particular volume is often more likely if a root is present in a neighbouring
volume, or was present when the volume was sampled previously. The scale
and timing at which the sampling is performed determine whether such spatial
and temporal correlations are present.
Geostatistics is a branch of applied statistics that can be used to detect,

model and estimate spatial patterns. It was developed originally for mining and
geology, but has been used more recent1y in the plant and soil sciences (e.g.
Castrignano and Lopez 1988; Castrignano et al. 1994; Bourgault et al. 1997). In
this chapter we give a very brief introduction to some geostatistical techniques.
Concise introductory works that use examples from soil science and agronomy
are those byVieira et al. (1983), Burgess and Webster (1980) and Webster (1985)
while Rossie et al. (1992) is a comprehensive review of geostatistical tools for
ecology. Only a few studies have applied geostatistical techniques to root
systems (Aiken 1992; Jackson and CaldwelII993a,b). The techniques ofvariog-
raphy and kriging are potentialIy very relevant: Variography models spatial
dependence of variables, whilst kriging interpolates between the measured
locations.
5.5.2.1 Semivariograms
Geostatistical techniques quantify spatial correlations and can be used to detect

clustering of roots. Semivariance is a measure of the variability between pairs
of observations and, in semivariograms, is plotted against the separation
between those observations (Box 5.2).
Semivariograms have been used to study the spatial distribution of soil
properties and roots (Jackson and Caldwell 1993a,b). A positive increase in
semivariance with distance (Fig. 5.8) indicates clustering. The separation
distance at which semivariance approaches a constant value is called the range,
and is the length of spatial correlation, corresponding roughly to the dimen-
sion of clustering. Roots of an individual plant may experience very different
conditions due to the heterogeneity of soH conditions. The distribution of
nutrients (nitrate, ammonium, phosphate and potassium), has been studied
on a scale of 0.1-10 m around perennial Artemisia and Pseudoroegneria plants
(Jackson and Caldwell 1993b): nitrate and ammonium contents varied by
between two and three orders of magnitude within the 120 m 2 plot, with
phosphate and potassium showing smaller variation. Clustering of these
nutrients occurred on a scale of <O.5m (corresponding to the range), with no
further increase in variance at scales greater than 1 m.
The semivariance at a separation equal to the range is known as the
sill, and is an estimate of the population variance for observations separated
by distances greater than the range. The semivariance extrapolated to zero
separation is called the nugget, and represents alI unaccounted for spatial
variability at distances smaller than the smallest sampling distance.
The nugget includes both variation due to measurement errors and spatial
variat ion that occurs over distances shorter than the sample spacing. The
BOX 5.2. Semivariance and Semivariograms

A semivariogram is constructed by plotting the semivariance as a
function of the distance separating pairs of observations (Fig. 5.7).
Semivariance is defined as
1 ~) 2
y(h) = - () 2..[R(i)-R(i + h)] l (5.1)
2N h j~1
where y(h) is the semivariance at separation distance (h), N(h) is the

number of paired observations separated by distance h, R(i) is some root
variable (e.g. root number, length density or dry weight) at location i,
and R(i + h) is the same variable separated from i by a distance h (Vieira
et al. 1983)
VI
- range -
--~----------------~
-_._-_._ -------- ----_ . _. -.- ----- --- --

nugget
• Fig. 5.7. Semivariogram
Distance showing idealised behaviour
0.6
Ql
U
C
l\j 0.4 , _fi.-... -- ~- - - - ... ----.----.----. ---.• ---.---.
'-
l\j
>
°E
Ql
(f)
0.2
• rootmass
0L-~-'--~-'----'---_--1 Fig. 5.8. Semivariogram from root
O 2 4 6 8 mass data measured by Jackson
Separation distance (m) and Caldwell (l993a)
difference between the semivariance at the sill and that at the nugget represents
the variance that can be modelled as a spatial dependence, based on the
available sampling grid. A small nugget/sill variance ratio indicates spatial
correlation.
When values of semivariance are being compiled into a semivariogram, it
is assumed that there are no systematic trends in the quantity measured (Aiken
et al. 1991). If such trends exist, functions characterising these trends must first
be subtracted from the data. For example, vertical trends must be removed
before generat ing a variogram of root intersection density with minirhizotrons
or washed root length density with depth. Semivariograms constructed from
data containing trends show in cre as ing semivariance with increasing distance
between observations.
5.5.2.2 Kriging and Cokriging
Kriging is an interpolation procedure that uses the semivariogram to estimate

values at locations where measurements were not made. The values estimated
[Z*(x o) 1using kriging are linear functions of the known values [Z(Xi) 1measured
at N other locations:
N
Z*(x )= I [A.Z(x.)], (5.2)
o i-1 1 1
where the weight fac tors (Ai) are chosen such that the mean value of the dif-
ference [Z*(x o) - Z(xo) 1is zero, and the variance of the difference is minimised.
The semivariogram determines the values of Ai and the relative weight of obser-
vations decreases with distance from the interpolation points.
Cokriging can be used where two or more variables are correlated in space.
Measurements of one variable may be used to produce predictions of how a
second variable is distributed. For example, very hard regions of the soi! may
be rooted sparsely, and so the root length density would be correlated nega-
tively with penetration resistance. Measurement of one variable may be used to
produce predictions of how a second variable is distributed.
5.5.2.3 ConditionalSimulations
Kriging is a procedure for estimating optimal values at intermediate points

on a map. However, no interpolation procedure can produce information that
has not been surveyed. The real spatial variation in the quantity studied
includes an extra component of variation that is not accounted for in the kriged
values.
Conditional simulations show realistic images of the variability of the

quantity studied. By running many of these simulations, the variance and mean
values can be mapped and this map can provide a reliable basis for calculating
root biomass, deciding on root sampling strategies, or for simulating water or
nutrient uptake. Further details of the theory of conditional simulation are
given in Matheron (1973).
5.6 Conclusions
Root systems are complex branched structures that vary in space and time. The
structure of the root system must be appreciated before it can be studied in the
field. It is important to consider the aims of any experiment carefully when
choosing an appropriate measurement technique from the many available, as
root sampling is very labour intensive. Auger sampling gives the most reliable
estimate of root length density, but losses of about 30% can occur during
washing procedures (see Chap. 6). Abias of about 30% for cereals can result in
estimates of root length density, per plant or per layer of soil, made using
traditional sampling schemes. The scientific basis of where to sample roots is
still understood relatively poorly, and there is a need for new experimental and
theoretical studies of root distribution to address this problem. Modelling is
starting to become a useful aid in investigating the consequences of different
temporal and spatial sampling schemes. Geostatistics is also providing new
ways of studying spatial heterogeneity in roots, although large numbers of
samples (e.g. 100 plus) are often required to provide sufficient data for the
analyses.
Acknowledgements. We thank Drs Allmaras and Berntson for their valuable

comments on an earlier draft of this manuscript. AGB and LP thank NATO for
a collaborative research grant. The Scottish Crop Research Institute receives
grant-in-aid from the Scottish Executive Rural Affairs Department.
References
Aiken RM (1992) Functional relations of root distributions with the flux and uptake of water
and nitrate. PhD Dissertation. Michigan State University, East Lansing
Aiken RM, Jawson MD, Grahammer K, Polymenopoulos AD (1991) Positional, spatially corre-
lated and random components of variability in carbon dioxide efflux. J Environ Qual 20:
301-308
Bengough AG, MacKenzie q, Diggle AJ (1992) Relations between root length densities and root
intersections with horizontal and vertical planes using root growth modelling in 3-
dimensions. Plant Soil145: 245-252
Bennie ATP (1996) Growth and mechanical impedance. In: Waisel Y, Eshel A, Kafkafi U (eds)
Plant roots, the hidden half. Marcel Dekker, New York pp 453-470
Bourgault G, Journel AG, Rhoades JD, Corwin DL, Lesch SM (1997) Geostatistical analysis of a
soil salinity data set. Adv Agron 58: 241-292
Burgess TM, Webster R (1980) Optimal interpolation and isarithmic mapping soil properties.1.
The semivariogram and punctual kriging. J Soil Sci 31: 315-331
Canadell J, Jackson RB, Ehleringer JR, Mooney HA, Sala OE, Schulze ED (1996) Maximum
rooting depth ofvegetation types at the global-scale. Oecologia 108: 583-595
Castrignano A, Lopez G (1988) La variebilita delle proprieta del suolo e la sua applicazione nelle
ricerche agronomiche. Proc Conf Italian Soc of Soil Science, Verona, 1988 pp 714-728
Castrignano A, De Giongio D, Stelluti M, Rizzo V (1994) A geostatistical approach to character-
ize spatial variability of yield in a durum wheat submitted to four tillage treatments. Proc
13th Int Conf Int Soil Tillage Research Organisation, Aalborg, 1994 pp 997-1004
Cochran WG, Cox GM (1957) Experimental designs 2nd edn. John Wiley New York
Coutts MP (1989) Factors affecting growth direction of tree roots. Ann Sci For 46: 277-287
Dittmer HJ (1937) A quantitative study of the roots and root hairs of a winter rye plant (Secale
cereale). Am J Bot 24: 417-420
Ehlers W, Kopke U, Hesse F, Bohm W (1983) Penetration resistance and root-growth of oats in
tilled and untilled loess soil. Soil Tillage Res 3: 261-275
Fitter AH, Stickland TR (1991) Architectural analysis of plant root systems 2. Infiuence of nutri-
ent supply on architecture in contrasting plant species. New Phytol118: 383-389
Gajri PR, Arora VK, Kumar K (1994) A procedure for determining average root length density
in row crops by single-site augering. Plant Soil160: 41-47
J Appl Ecol11: 773-781
Grabarnik P, Pages L, Bengough AG (1998) Geometric properties of simulated maize root
systems: consequences for length density and intersection density. Plant Soil200: 157-167
Gregory PJ, McGowan M, Biscoe PV (1978) Water relations of winter wheat. 2. Soil water rela-
tions. J Agric Sci 91: 103-116
Jackson RB, Caldwell MM (1993a) Geostatistical patterns of soil heterogeneity around individ-
ual perennial plants. J Eco181: 683-692
Jackson RB, Caldwell MM (1993b) The scale of nutrient heterogeneity around individual plants
and its quantification with geostatistics. Ecology 74: 612-614
Jackson RB, Canadell J, Ehleringer JR, Mooney HA, Sala OE, Schulze ED (1996) A global analy-
sis of root distributions for terrestrial biomes. Oecologia 108: 389-411
Kucke M, Schmid H, Spiess A (1995) A comparison of four methods for measuring roots of field
crops in three contrasting soils. Plant Soill72: 63-71
Lauenroth WK, Hunt HW, Swift DM, Singh JS (1986) Overestimation of net root production - a
real or imaginary problem - reply. Ecology 67: 580-582
Le Roux Y, Pages L (1994) Developpement et polymorphisme racinaire chez de jeunes semis
d'ht!vea (Hevea brasiliensis). Can J Bot 72: 924-932
Logsdon SD, Allmaras RR (1991) Maize and soybean clustering as indicated by root mapping.
Plant Soil131: 169-176
Logsdon SD, Linden DR (1992) Interactions of earthworms with soil physical conditions infiu-
encing plant growth. Soil Sci 154: 330-336
Lynch JP, Nielsen KL, Davis RD, Jablokow AG (1997) Simroot: modelling and visualization of
root systems. Plant Soil 188: 139-151
Lyr H, Hoffmann G (1967) Growth rates and growth periodicity of tree roots. Int Rev For Res
2: 181-236
Mace AE (1964) Sample-size determination. Reinhold Publishing, New York
Matheron G (1973) The intrinsic random functions and their applications. Adv Appl Probabil
5:438-468
Nelson WW, Allmaras RR (1969) An improved monolith method for excavating and describing
roots. Agron J 61: 751-754
Newman EI (1969) Resistance to water fiow in soil and plant. J Appl Eco16: 1-12
Nielsen KL, Lynch JP, Weiss HN (1997) Fractal geometry of bean root systems: correlations
between spatial and fractal dimension. Am J Bot 84: 26-33
Pages L, Bengough AG (1997) Modelling minirhizotron observations to test experimental pro-
cedures. Plant Soil189: 81-89
Pages L, Pellerin S (1994) Evaluation of parameters describing the root system architecture of
field-grown maize plants (Zea mays L.). II. Density, length, and branching of first-order
lateral roots. Plant Soil164: 169-176
Parker q, Carr MKV, Jarvis NJ, Puplampu BO, Lee VH (1991) An evaluation of the minirhizotron
technique for estimating root distribution in potatoes. J Agric Sci 116: 341-350
Pearce SC (1983) The agricultural field experiment: a statistical examination of theory and prac-
tice. John Wiley, New York
field-grown maize plants (Zea mays L.). 1. Elongation of seminal and nodal roots and exten-
sion of their branched zone. Plant Soil164: 155-167
Pellerin S, Pages L (1996) Evaluation of in-field conditions of a 3-dimensional architectural
model of the maize root-system - comparison of simulated and observed horizontal root
maps. Plant Soil178: 101-112
Persson H (1978) Root dynamics in a young Scots pine stand in central Sweden. Oikos 30: 508-519
Robert PC, Rust RH, Larson WE (1996) Precision agriculture. Proc 3rd Int Conf on precision
agriculture, Minneapolis, 1996
Robinson D (1994) The responses of plants to non-uniform supplies of nutrients. New Phytol
127:635-674
Rossi RE, Mulla DJ, Journel AG, Franz EH (1992) Geostatistical tools for modeling and inter-
preting ecological spatial dependence. Ecol Monogr 62: 277-314
Russell RS (1977) Plant root systems. Their function and interaction with the soil. McGraw Hill,
Maidenhead
Sanders DH (1990) Statistics: a fresh approach (4th edn). McGraw-Hill, New York
Schuurman JJ, Knot L (1957) Het schatten van hoeveelheden wortels in voor wortelonderzoek
genomen monsters. Versl Landbouwk Onderz 63: 31
Singh JS, Lauenroth WK, Hunt HW, Swift DM (1984) Bias and random errors in estimators of
net root production - a simulation approach. Ecology 65: 1760-1764
Spek LY (1997) Generation and visualization of root-like structures in a three-dimensional
space. Plant Soil197: 9-18
Spek LY, Van Noordwijk M (1994) Proximal root diameters as predictors of total root system
size for fractal branching models. II. Numerical model. Plant Soil164: 119-128
Tardieu F (1988) Analysis of the spatial variability of maize root density II. Distances between
roots. Plant Soill07: 267-272
Tardieu F, Pellerin S (1990) Trajectory of the nodal roots of maize in fields with low mechani-
cal constraints. Plant Soil124: 39-45
Van Noordwijk M, Purnomosidhi P (1995) Root architecture in relation to tree-soil-crop inter-

actions and shoot pruning in agroforestry. Agrofor Syst 30: 161-173
Van Noordwijk M, Floris J, de Jager A (1985) Sampling schemes for estimating root density dis-
tribution in cropped fields. Neth J Agric Sci 33: 241-262
Van Noordwijk M, Spek LY, De Willigen P (1994) Proximal root diameters as predictors of total
root system size for fractal branching models. 1. Theory. Plant Soi1164: 107-118
Varney GT, McCully ME (1991) The branch roots of Zea. II. Developmentalloss of the apical
meristem in field-grown roots. New Phytol118: 535-546
Varney GT, Canny MJ, Wang XL, McCully ME (1991) The branch roots of Zea. 1. First order
branches, their number, sizes and division into classes. Ann Bot 67: 357-364
Vieira SR, Hatfield JL, Nielsen DR, Biggar JW (1983) Geostatistical theory and application to
variability of some agronomical properties. Hilgardia 51: (3)
Vogt KA, Grier CC, Gower ST, Sprugel DG, Vogt DJ (1986) Overestimation of net root produc-
tion - a real or imaginary problem? Ecology 67: 577-579
Waisel Y, Eshel A (1991) Multiform behavior of various constituents of one root system. In:
Waisel Y, Eshel A, Kafkafi U (eds) Plant roots, the hidden half. Marcel Dekker, New York
pp 3-24
Wang J, Hesketh JS, Wooley JT (1986) Preexisting channels and soybean rooting patterns. Soil
Sci 141: 432-437
Webster R (1985) Quantitative spatial analysis of soi! in the field. Adv Soi! Sci 3: 1-70
Wilde SA (1958) Forest soils: their protection and relation to si!viculture. Roland Press, New
York
Wilson BF (1964) Structure and growth of woody roots of Acer rubrum L. Harv For Pap 11: 14
CHAPTER6
Auger Sampling, Ingrowth Cores

and Pinboard Methods
Maria do Rosario G. OliveiraI, M. van Noordwijk2, S.R. Gaze\ G. Brouwer\
S. Bonas, G. Mosca s, and K. Hairiah 6
1 University of Evora, Apartado 94,7002-554 Evora, Portugal
2 ICRAF, JI CIFOR, PO Box 161, Situ Gede, Sindang Barang Bogor 16680, 16001Bogor,
Indonesia
3 University of Reading, PO Box 233, Reading, RG6 6DW, United Kingdom
4 Plant Research International, Postbus 16, 6700 AA Wageningen, The Netherlands
5 Universitit di Padova Instituto di Agronomica, Dipt. di Agronomia Ambientale e Produzioni
Vegetali, Agripolis - Via Romea, 16, 35020 Legnaro (Padova), Italy
6 University Brawijaya, Faculty of Agriculture, Department of Soi! Science, Jl. Veteran,
65145 Malang, Indonesia
CONTENTS
6.2 Methods of Root Sampling 177
6.2.1 Auger Sampling 177
6.2.1.1 Hand Sampling 179
6.2.1.2 Mechanised Techniques 179
6.2.1.3 Core Diameter 180
6.2.1.4 Sampling Strategy in the Field 180
Position of Sampling 180
Depth of Sampling 181
Number of Replications 181
6.2.1.5 Special Techniques for Tree Roots 182
6.2.1.6 Drawbacks 182
6.2.2 Ingrowth Cores 183
6.2.2.1 Concept 183
6.2.2.3 Results and Perspectives 185
6.2.3 Pinboard Method 186
6.2.3.1 Concept 186
6.2.3.2 Procedure 187
6.2.3.4 Results and Perspectives 190
6.3 Procedures for Root Washing 191
6.3.1 Hand Washing 191

176 M.R.G. Oliveira et al.
Soil Cores 191

Washing Pinboard Samples 192
6.3.2 Automatic Washing 192
6.3.3 Chemical Dispersing Agents 194
6.3.4 Errors 194
6.3.4.1 Loss of Fine Roots 194
6.3.4.2 Loss of Dry Weight 195
6.3.4.3 Change in Chemical Composition 196
6.3.4.4 Presence of Debris and Soil Particles 196
6.3.4.5 Presence of Dead Roots 197
6.3.4.6 Operator Differences 197
6.3.5 Time Investment 198
6.4 Methods of Storage 198
6.4.1 Storage Before Washing 198
6.4.2 Storage After Washing 199
Cold Storage 199
Freezing 199
Drying 199
Chemicals 200
6.4.3 Effect of Storage Methods on Root Properties 200
6.5 Root Quantification 200

6.5.1 Root Weight 201
6.5.2 Root Length 201
6.5.2.1 Direct Measurement 201
6.5.2.2 Line Intersect Method 202
6.5.2.3 Visual Estimation Method 203
6.5.3 Root Diameter and Surface Area 205
6.5.4 Presentation of Root Data 205
6.6 Conclusions and Perspectives 206
References 206
6.1 Introduction
This chapter oudines those methods for assessing root systems structure and
function in the field which are based on washing roots free from the soil in which
they grew. Some of these methods are included in previous reviews (Kolesnikov
1971; B6hm 1979). The methods are either disruptive or totallydestructive to the
root system being studied and to the immediate environment (Taylor et al. 199I).
6 Auger Sampling, Ingrowth Cores and Pinboard Methods 177
The spatial variability of root systems requires that a large number of repli-
cates be sampled to obtain reasonably accurate estimates of root parameters.
Intensive, destructive sampling disrupts the remaining roots, which may in turn
affect results in the future. Consequent1y, due to the large variability of results,
it is not possible, in practice, to estimate root turnover from frequent sampling
schemes (Van Noordwijk 1993). Nevertheless, soil cores from auger sampling
or pinboards often yield the best quantitative informat ion on root system
biomass and root length per volume of soil (Caldwell and Virginia 1991). These
methods are therefore the basis for calibrating other techniques, including those
for estimating root turnover.
Monolith samples can be taken using a pinboard, which enables the com-
plete root system to be observed after washing the soil away. In contrast, augered
samples are taken at discrete locations in relation to the horizontal and vertical
distance in the soil profile from the plant (stratified sampling). Auger sampling
provides a soil-root sample of a limited volume from the root zone.
The ingrowth core technique (Steen 1984; Fabiăo et al. 1985) is based on
mesh bags filled with root-free soil, placed at different depths and removed at
prescribed intervals. Differences between soil conditions inside the bags and in
the surrounding undisturbed soil, such as soil strength and mineral nutrition
status, can change the pattern and quantity of roots. However this method is
particularly suitable for recording root response to localised fertiliser applica-
tion or other heterogeneity in the soil (Cuevas and Medina 1988; Hairiah
et al. 1991).
The destructive sampling methods described in this chapter are the basis
for quantifying root length, diameter, surface area, volume, biomass, disease,
mycorrhizal association and ion content. Depending on the goal of the further
analysis, however, one may need correction factors to account for the changes
in root properties during sampling, handling and storage.
Table 6.1 indicates the information that can be obtained from different
methods of root measurement. Auger sampling, ingrowth cores and pinboard
methods are described in this chapter. Core break and root mapping methods
are described in Chapter 7. Minirhizotron and related methods are described
in Chapter 8.
6.2 Methods of Root Sampling
6.2.1 Auger Sampling
The best sampling technique for obtaining volumetric soil-samples is the auger
method. With this method soil samples are taken from the field using hand-
operated or mechanical samplers and washed to separate roots from soil.
.....
00
"
Table 6.1. Aim of different methods for root measurement'
Parameter Auger Ingrowth Pinboard Core break Root Minirhizotron

sampling cores method mapping
Root length cmcm-3 +++ +++ ++(+) + + +

Root weight gcm-3 +++ +++ ++(+)
Distribution pattern + + ++
Root growth and decay ++ +++
Root diameter ++(+) ++(+) ++(+) +
Branching pattern + + +
Time consuming 7-20 5-15 15-30 2 4 4-10
d (8 hours)
Number of samples 120 60 2 Boards 0.5 m' 120 = 12 Boreholes 6 Maps O.sm' 1 Minirhiz./yearb
, Symbols are as follows: - no information on parameter, + qualitative interpretation may be drawn, ++ semi-quantitative interpretation through a
ranking, +++ fully quantitative information obtained.
b length of minirhizotron is 1-2.1 m. :s::
::o
o
Q
~:
...,
Ol
~
~
6.2.1.1 Hand Sampling
The simplest method for taking soil samples is to use a hand-driven corer. One
of the most commonly used corers is that described by Schuurman and Goede-
waagen (1971; Fig. 6.1).1t comprises a cylindrical tube 15cm long with an inside
diameter of 7 cm. A T-handle at the top of the auger shaft facilitates rotation of
the auger to aid penetration into and removal from the soil. Many authors have
modified this simple hand auger, mainly by varying the tube diameter, or by
using mass impact to drive the auger into the soil.
Albrecht (1951) and Albrecht et al. (1953) describe an auger consisting of
two halves held together by a metal ring. These halves can be separated to allow
for recovery of intact soil cores. A new manual cor ing system has been proposed
by Prior and Rogers (1992, 1994) which includes a manual driver of adjustable
weight, a manual core extractor and steel core tubes with clear plastic liners
which encase the soil core for retrieval and transport.
Various materials have been used for the augers but the most frequently
utilised are metal (stainless steel) or Plexiglas, bevelled at one end to minimise
soil dis rupt ion during sampling.
6.2.1.2 Mechanised Techniques
When core samples must be recovered from depth or from difficult soils, mech-
anised equipment is used. One of the simplest ways is to use an auger driven into
the soil by a hand-held motorised drop-hammer and removed by a puller or a
screw-jack (B6hm 1979). Soil core samplers can also be mounted on tractors,
typically on the drawbar, which has hydraulic power for the movement (upward
and downward). It is also possible to utilise special machines constructed for
agricultural engineering purposes, such as for drainage studies and soil survey
mapping. Hydraulic devices transported by large tractors have reduced the time
and the labour required for taking soil cores but their use is limited to situations
where damage of a large portion of the plot is not a problem.
Plunger
Auger and plunger
Fig.6.1. Design of a simple auger based on Schuurman and Goedewaagen (1971)

The method proposed by Baarstad et al. (1993), where a hydraulic probe is

attached to a knuckle boom mounted on a truck, would reduce soil compaction,
although its cost could be a problem.
6.2. 1.3 Core Diameter
One of the most important fac tors in field sampling is the core diameter (see
Bohm 1979). The core must be large enough to obtain a reasonable sample
volume, yet small enough to enable the cores to be obtained by the sampling
method to be used. With decreasing core diameter an increasing number of
replicates must be taken to maintain sampling accuracy (see Chap. 5). Small
diameter cores can be a particular problem where there are low rooting densi-
ties. Furthermore, Schuurmann and Goedewaagen (1971) reported average root
length densities were lower using a 4 cm diameter core than with a 7 cm diam-
eter core. The difference could not be accounted for by the actual soil volume
sampled and was attributed to the increased frictional resistance between the
soil core and the bore of the tube in the smaller diameter tube. They recom-
mended a core diameter of 7 cm, and the most commonly used core diameters
range from 5 to 8 cm (Van Noordwijk 1993).
In dry sandy soils, a smaller auger diameter may be required to avoid soil
loss during core extraction. In very wet soils special augers may be needed
which allow air entry into the mud (Schuurman and Goedewaagen 1971).
6.2.1.4 Sampling Strategy in the Field
Position of Sampling. Sample position depends on the purpose of the mea-

surements. If the aim is to obtain an average value for the field, a form of strat-
ified sampling may be desirable, as rooting density will vary with the spatial
arrangement of plants. In grassland, or other systems where plants are not
grown in rows, a completely randomised design may be followed, unless there
is reason to expect specific spatial patterns in the field (e.g. due to distance to
the nearest drain, or irrigation water patterns), when such factors must be
considered in the sampling design. For row crops, sampling must be stratified
within and between rows. Crozier and King (1993) suggest sampling in a tran-
sect perpendicular to the row, within the rooting zone, while Van Noordwijk
et al. (1985) propose special sampling schemes, depending on row spacing. In
more complex plant arrangements, for example in agroforestry systems, addi-
tional levels of stratification are required and in the case of very complex
systems or in natural vegetation, samples are no longer stratified but rather
collected along a larger number of randomly assigned transects. Within each
stratum, core position must be assigned at random. With any stratified sam-
6 Auger 5ampling, Ingrowth Cores and Pinboard Methods 181
pling scheme care must be taken when estimating a field average value from the
data. Serious biases may be introduced if the average value for alI samples is
assumed to be the average for the field.
For treatment or variety comparisons, one sampling position per treat-
ment may be sufficient, though this would need to be replicated throughout the
experiment. For more detailed process or model calibration studies it may be
necessary to quantify the total root expansion or the root depth or the soil
layers in which the root length density is above a certain threshold (Passioura
1980).
Depth of Sampling. Ideally the core should be driven to the Iim it of rooting
depth but the deepest layers are difficult to reach, and within those horizons
the variability is often high. The maximum sampling depth can be estimated
in some cases by extrapolating from the root length densities in the upper
soil using a negative exponential equation (Van Noordwijk 1993). In any case,
all soils must be sampled to a minimum depth of 30 cm or the bottom of the
plough layer because most of the roots will probably be concentrated in this
layer.
lf clearly differentiated soil horizons are present in the sampled soil, the
first approximation is to separate the different soillayers and then to sub divide
the cores in standard length intervals. The most widely used core length is
10 cm. However, in grasslands and no till systems, it may be worth separating
the 0-10 cm layer into 0-5 and 5-10 cm. Shorter core lengths will increase
sample variability and it is difficult to cut portions of the core shorter than the
diameter of the core itself.
Number of Replications. Particular attentÎon must be given to the number of

replicates (see Chap. 5). It is clear that by increasing the number of replicates
the variability of data decreases, but the standard error of 25 samples is only
five times smaller than that of a single sample (van Noordwijk 1993).
The number of replications adopted by different authors varies from three
to ten per experimental unit, similarly to what is done in aboveground mea-
surements. Van Noordwijk et al. (1985) estimated a coefficient of variation in
grassland root weight for individual auger samples of 385 cm3 (10 cm height and
7 cm diameter) of at least 40%. Values up to 100% are, however, not uncommon.
The number of soil cores required for each plot can be determined at the begin-
ning of the sampling program by extracting the roots from a number of soil
cores corn ing from the same plot (Vogt and Persson 1991). In this way it is pos-
sible to have an idea of the variability of the samples and optimise the sampling
process.
In order to reduce the number of samples to be processed, Schroth and
Kolbe (1994) propose a method consisting of the combination and homogeni-
sation of several cores from a plot, with consequent subsampling for root
extraction. This method, similar to the composite sampling schemes usually fol-
lowed for soil chemical analyses, allows an increase in the number of soil
samples and/or a reduction of processing time. However, the method introduces
a new source of error dependent on the homogeneity of the total sample before
subsampling and it is not appropriate when the root turnover is studied.
6.2.7.5 Special Techniques for Tree Roofs
Sampling tree roots requires special techniques because of the large range of
root diameters within a tree root system and the large volume of soil occupied
by the roots. Knowledge of the species' root system structure is necessary if it
is desired to sample a particular portion of the root system (see Chap.l). Veller
(1971) proposed a sampling scheme consisting of concentric circles, centred on
the stern of the tree, the diameter of each circle being chosen according to the
nature and age of the tree under study.
Generally, for taking soil cores in tree root systems it is necessary to
utilise mechanised sampling techniques because of the hardness of the large
roots.
For some purposes it is relevant to investigate the finer parts of the woody
roots (e.g. spatial distribution, biomass turnover). Methods defined by Kalela
(1950) or Nielsen (1990, 1994) are suitable. The"Nielsen sieve" is 45 x 45 cm with
25-cm-high walls, and ribs, 0.5 cm thick, placed in two perpendicular layers,
defining holes of 3 X 3 cm squares. Depending on the soil and the hypothesis
to be tested, the soil material is extracted separately from a number of soil
horizons to the sieve. Stones and soil material are sieved in the field, and the
collected root material is washed, sorted, dried and weighed in the laboratory.
Roots are sorted into biomass and necromass, the living roots being sorted to
the following diameter classes (root material below 0.1 cm diameter is removed
after drying): fine roots «0.2 cm), thin roots I (0.2-0.5 cm), thin roots II (0.5-
1 cm), thin roots III (1-2 cm) and coarse roots (2-5 cm). Coarse roots can not
be adequately sampled with any auger sampling scheme and one may have to
resort to methods predicting ratios of roots in different diameter classes based
on fractal branching analysis (Spek and van Noordwijk 1994; Van Noordwijk
et al. 1994; Van Noordwijk and Purnomosidhi 1995).
6.2.7.6 Drawbacks
Auger sampling requires a large number of samples since the volume of auger
samples is small compared with the total soil volume. In very small plots prob-
lems may arise if coring damages a relatively large portion of the plots.
Sampling depth using hand-operated apparatus is seldom greater than

100 cm. an stony soils or soils with large amounts of woody roots, inserting the
auger may be difficult. AIso, on dry clay soils the penetration resistance presents
a problem. Sometimes it is helpful to rewet the surface soil before coring, if this
does not interfere with the purposes of sampling.
The use of heavy mechanised equipment for core sampling may be a risk
since it can compact the soil cores, shortening the core itself. This must be taken
into account when cutting the core and interpreting the results.
The total time needed for a single core depends on soil conditions and
depth of sampling, but generally the time needed for taking samples in the field
is much less than the subsequent processing in the laboratory.
6.2.2 Ingrowth Cores
6.2.2.1 Concept
The basic concept of the root ingrowth core technique is to present root-free
soil to the growing root system for a fixed length of time, so that root growth
during that period can be quantified (Steen 1984, 1991; Hansson et al. 1991).
The method reduces uncertainty about the time interval during which the roots
in a particular soil core developed, but may introduce biases, as it is not possi-
bIe to create root-free soil which is identical to the undisturbed soil conditions.
It is therefore best to use the method mainly as a relative indication of the main
periods of root growth, by comparing mesh bags placed at various times or left
in the soil for various time intervals.
Practica! instructions for using ingrowth cores are given in Box 6.1.
6.2.2.2 Drawbacks
The main difficulty in interpreting data from soil ingrowth cores arises from
the unknown degree of disturbance upon introducing the mesh bags. Some
of the roots growing into the core may be branch roots of roots severed by the
coring, and this may cause an overestimate of the root growth which would
otherwise have occurred in this volume of soil. More important1y, however, the
process of sieving and re-packing the soil modifies soil conditions. The alter-
ation of sandy soil properties is small but on other soil types, the disturbance
to the aggregate structure accelerates mineralization of soil organic matter
and affects soil aeration. When results are compared for mesh bags placed
at different time periods, one has to be cautious of the fact that the degree
of disturbance may vary with soil moisture content, even if exact1y the same
BOX 6.1. Procedure for Installing the Mesh Bags for the Root Ingrowth
Cores Technique
The procedure is summarised in Fig. 6.2. Remove long cores of soi! and
sieve the soi! ta remove any roots. Then, place a tubular nylon ar
polypropylene mesh bag (0.5-0.7 cm mesh) in the hale, loosely attached
ta a plastic pipe which is slightly smaller in diameter than the soil core.
Fill the hale with the sieved soi! via this tube (using a funnel), re-packing
the saH ta the original bulk density. The filling procedure is critical and
results depend an saH texture, structure and moisture content. The best
results are obtained by measuring the amount of soi! ta be filled per 1 cm
layer (ar less) and pressing the soillayer by layer with a pestle which fits
into the tube. After pressing a layer, scrape the surface ta create
roughness which allows good contact with the next Iayer of soi!.
a b ( d e f
[1 ..;. ~
Fig.6.2. Procedure for installing the mesh bags for the root ingrowth cores technique
It is convenient ta attach a rope ta the bottom of the mesh bag which

can be used to pull out the mesh bag at harvest time. The other end of
the rope is labelled and remains an the soil surface.
When the bags are recovered, it is often necessary ta cut woody roots
entering the bag with a knife (ar auger of diameter slightly larger than
the one originally used). Non-woody roots can be trimmed after mesh
bag recovery. Wash the mesh bag contents as per a normal auger sample.
6 Auger 5ampling, Ingrowth Cores and Pinboard Methods 185
procedure is used for placing the mesh bags. If large variations in soH moisture
contents are to be expected, it may be better to compare different intervals for
root ingrowth into bags placed at the same time, rather than constant time inter-
vals for bags placed at different times.
6.2.2.3 Results and Perspectives
With sufficient precautions, the method produces a reasonable estimate of

the relative seasonal pattern of fine root growth. The results obtained from this
method can usefully complement data from destructive sampling, which should
be conducted at least once during the period of study. It may be worth collect-
ing the roots sieved from the soH before placing it back into the mesh bag to
correlate new root growth and previous root mass in the core, but sieving indi-
vidual soH cores in the field is time consuming and often soH obtained from
the same soHlayer elsewhere in the experimental plot is used. Using the same,
homogenised soH for alI replicate ingrowth cores may reduce the variability of
the results, but it depends on the purpose of the measurement whether or not
this is desirable.
An alternative use of the ingrowth core technique is to compare different
soH types, soH with different organic or inorganic amendments and/or packed
to different bulk densities. For these applications one accepts that the soH
inside the ingrowth core will be different from the surrounding bulk soil.
Results may indicate the "local response" of root development to soH hetero-
geneity. Hairiah et al. (1991) used this method to investigate whether shallow
root development of Mucuna pruriens on an acid soH was due to an inherent
characteristic of the subsoH, or to the position of the subsoil. In other situations
addition of specific soil organisms (symbionts, rhizosphere organisms, and rhi-
zovores) may be compared. In this way "split-root" experiments can be con-
ducted under field conditions with the advantage of keeping other conditions
equal and focusing on local effects on root development. Most previous split-
root studies have been restricted to pot experiments (e.g. De Jager 1982; Hairiah
et al. 1993). The ingrowth core technique may be especially relevant for trees
and other perennials where other experimental techniques are difficult to
implement.
If dead roots are included in the ingrowth core, root decomposition can
be studied under field conditions, but results may depend on whether the root
material was naturally senesced or killed for the purposes of measurement.
Root growth and senescence may be measured during root ingrowth studies
when alive and dead roots are easHy distinguished and a series of measurement
are made over time.
6.2.3 Pinboard Method
6.2.3.1 Concept
Monolith samples can be obtained with pinboards ("fakir beds"), where the pins
hold the root system in an approximately correct position during washing. The
equipment needed to take a pinboard sample is shown in Box 6.2 and Fig. 6.3.
BOX 6.2. Pinboard Constr uction and Other Requirements

A pinboard ('fakir bed') can be made by inserting U-shaped stainless
steel pins through plywood, ta form a regular grid of steel pins (Fig. 6.3;
Schuurman and Goedewaagen 1971; Bohm 1979). These pins can be
made from bicycle/motorbike spokes bent into a U-shape (A) with a 5-
cm-base and upright length equal to the sum of the required length plus
the thickness of the board (a convenient total upright length is 12cm).
Before inserting the pinboard in to the soil, slide a plastic coarse-mesh
screen over the pins. Other requirements (Fig. 6.3) are trowels (B), metal
blades to smooth the profile wall (D), a rubber hammer, a car jack (C)
.adjusted to support the board, metal string or a steel cable with a
diameter of 2 mm (e.g. motorcycle brake cable) with two handles (E) for
cutting the soil.
Fig. 6.3. Construction of a pinboard and some of the other equipment needed for taking
samples. A pins, B trowel, C car jack, D blade, E string
The method has recently been improved by putting a coarse mesh screen (e.g.
with openings of 0.5 cm) on the pins before the board is pushed into the soil, and
gradually lift ing this screen while washing away the soil. This screen prevents the
movement of roots by "surface runoff" during the washing process.
Pinboard samples can be taken of individual plants, of row crops (usually
perpendicular to the crop row), or of grassland vegetation. The size of the pin-
board is determined by the vegetation rooting habit and practical considera-
tions (samples of 100 x 60 x lOcm of soil weigh about 100kg).
6.2.3.2 Procedure
Practical guidelines for taking a sample are detailed in Box 6.3 and Fig. 6.4.
A set of pictures showing different steps of the pinboard method for sam-
pling crop root systems is presented in Fig. 6.5.
An alternative sampling device for monolith samples, which avoids the
need for a soil pit, is described by Floris and Van Noordwijk (1984). A metal
BOX 6.3. Procedure for Taking Pinboard Samples

Select and mark an appropriate sample location either at random or
according to specified criteria (e.g. restricting samples to plants of a
certain size) .
Dig a soi! pit just outside this sample location; the size of the soil pit
depends on the dimensions of the pinboard. For a 60-cm-wide pinboard,
a pit 100 cm wide allows for sufficient manoeuvrability to set and recover
the pinboard. When plants are grown in rows, the pit is normally dug
perpendicular to the row.
Smooth off the profile wall at about 5 cm from the plant (for 12 cm
pins) so that the plant is in the middle of the pins. Harvest the above-
ground parts of the plant and describe the soi! profile.
Place the pinboard vertically with the pins against the profile face,
and adjust so that the top row of pins is at ground level, and then push
the pinboard into the soil by hammering it in. Alternatively, a jack or
similar hydraulic device can be used to push it in smoothly.
Support the pinboard with a car jack (Fig. 6.4) and remove about 15
cm (a few centimeters beyond the tips of the pins) of soil underneath the
pinboard with a knife. Cut away soi! on both sides of the board, also a
few centimeters further than the tips of the pins and put the steel cable
in place (Fig. 6.3). Tie the plastic mesh screen around the sample to
prevent soilloss during the next step. In a sawing movement, cut
the last surface connecting the monolith to the surrounding soil mass,
leaving the pinboard supported by the jack. Normally one person
Top view Side
.:;<;;~~_ _ _ Sta inless

steel wire
Car jac k
5 mm mesh screen
W
Fig. 6.4. Procedure for taking pinboard samples: after pushing the pinboard into a
smoothed profile wall, the board is supported on a car jack; soil below and on both sides of
the board is cut away by sawing movements with a stainless steel wire
stands in the soil pit to control the process while two people pull the
cable. When loose, pull the board backwards and carefully lift it from
the pit. Cut away the edge of the soil monolith to the level of the tips
of pins.
LabeI the sample and cover it with a plastic bag for transport to the
laboratory.
box is driven into the soi! by a hydraulic cylinder and supported by a strong
metal frame, anchored into the soi!. When the desired depth is reached, the
bottom is cut by pulling a cable (as in the pinboard sample) and a pair of claws
is inserted into the soil by tightening screws on top of the box. Asmall metal
tube provides air and thus prevents suction on the soi! while the box is pulled
to the soil surface. In constructing the box a delicate balan ce must be main-
tained between keeping the walls thin and smooth enough to allow easy inser-
tion into the soil, and providing enough grip on the soil that the sample will
not be lost from the box while it is pulled up. In practice, the equipment
described by Floris and Van Noordwijk (1984) is no more time efficient than
the pinboard method, but it involves less destruction of the surrounding soil
(except for the soil anchors) and may thus be desirable in long-term experi-
Fig.6.5. Pinboard method for sampling crop root systems. a M.A.]. Goedewaagen (author of a
classical description of root research methods) in the 1930s with a washed out sugarbeet (Beta
vulgaris) root system. b Pinboard just pushed into a soi! profile. c Cutting away redundant soi!
after the monolith has been cut free from the profile wall. d Washing a pinboard sample directly
in the field (a small stream in the forest). e Washing a pinboard sample in the laboratory with
a rotating sprinkler. f Sugarbeet root system washed free. (Photographs a and f courtesy of
archive AB-DLO, Haren, the Netherlands, b-e Meine van Noordwijk)
ments with a restricted plot size. If a similar box is mounted on a tractor's

hydraulic power system, the weight of the tractor is generally sufficient to push
the box into the soil. This reduces the destruction of the experimental field to
that caused by a tractor.
6.2.3.3 Drawbacks
Taking the pinboard samples requires some skill and takes more labour than
sampling with augers. The major drawbacks for many situations are probably
the destruction caused by the soil pit and the limits to sample size by the weight
of the soil. Compared with auger samples, larger losses of fine roots during the
washing process can occur, especially for roots entering the sample volume from
the outside, instead of from the plant in the centre of the board.
6.2.3.4 Results and Perspectives
Overall, the pinboard method provides useful information per unit effort. The
distinction between live and dead roots is easier with pinboard sampling
than with methods where the root system is not sampled in its entirety. In
comparison with other sampling methods, based on blocks of soil or auger
samples, pinboard samples contain more coherent parts of a root system
and can thus be washed on a coarse mesh screen rather than on a fine mesh
sieve. This facilitates the washing process and leads to root samples without
much organic debris. The main gain in time is dur ing the sample cleaning stage.
Apart from being faster, the pinboards also allow a more coherent view of the
branching pattern and allow the tracing of individual roots to their origin. For
an analysis of intercropping systems or in situations with a prominence of
"weed" roots, this can be a clear advantage. The washed system can be pho-
tographed to obtain a visual impression or can be conserved as such. If com-
bined with a resin cast of a small slice of the soil profile (Schuurman and
Goedewaagen 1971) a visuallink can be made between soil conditions and root
response. Examples of root systems washed free on pinboards are presented in
Fig. 6.6.
To save time when quantifying roots after washing, a visual estimate of the
amount of roots per block of soil (5 x 5 x12 cm3 ) can be made, with a repre-
sentative selection of samples recovered to calibrate the vis ual ranking. Results
then depend on the consistency of the person doing the visual assesment and
'. "\ - ~
-'
'j (;:..
. , ~
- 1./-- \.;., ~..;...
[~J '~
~ ")
: ' '
Fig.6.6. Examples of root systems washed free on

pinboards. a Calopogonium mucunoides, a legume
cover crop on acid soils in Nigeria. b Potato (Solanum
tuberosum) in a pinboard sample taken along the
ridge. c Helianthus tuberosus in a pinboard sample
perpendicular to the ridge. (Photographs courtesy of
Meine van Noordwijk and archive AB-DLO, Haren, the
Netherlands)
6 Auger Sampling, Ingrowth (ores and Pinboard Methods 191
on the characteristics of the roots. Where large differences in root diameter

occur it is more difficult to obtain accurate estimates. If root data are collected
for a well-specified purpose, the analysis may concentrate on root densities
in a certain range, and a visual estimation procedure can prove helpful. For
example, models of nitrate uptake suggest that root length densities above
0.5 cm cm-3 are redundant for uptake of available N from a soillayer. If nitrate
uptake is the main interest, analysis of the pinboard samples can focus on
the layers or zones around and below this critical point. Generally, cleaning and
measuring the upper soillayers with large amounts of roots takes a large share
of the total time, and for certain questions it may be enough to rank root den-
sities as few, intermediate or many.
6.3 Procedures for Root Washing
Once sampling in the field is finished the roots must be separated from
the soil. This is done in different ways but usually soil is washed away from
the root samples. Ideally, the best approach is to wash roots from the
samples immediately upon arrival from the field but core samples may be
stored in sealed polyethylene bags under refrigeration, or frozen until
processing.
Roots may be separated from soil by dry-sieving but this method is
only appropriate for very sandy soils, and its use is limited to tree root studies,
where roots with a diameter greater than 2 mm are to be investigated
(Bohm 1979).
6.3.1 Hand Washing
Soil Cores. Roots must be washed gently to minimise loss or damage. The water
flow of the sprinkler must be very low because the roots are easily damaged by
high water flows. Large roots may be removed by hand.
If roots and soil have not separated with sprinkling, it may be necessary
to soak and gently agitate the sample. The soil particles should then settle
while the roots and other organic matter float to the top and can be poured
into another bucket or sieved for further cleaning. The procedure must be
repeated as many times as necessary until a clean root sample is obtained. The
sieve size used is critical to minimising root loss during washing (see Sect.
6.3.4.1), though the sieve size used will depend on the type of root material and
the objective of the research. If there is a lot of organic matter, a coarse sieve
(e.g. 0.2 cm mesh) can be used to split the sample into a first fraction, with roots
and large organic particles, and a second fraction, with fine short roots and
debris.
When the main purpose of the research is chemical analysis of roots, it is
important to reduce the washing time and a coarse sieve only can be used to
save time.
Washing Pinboard Samples. Washing the soil from pinboard monoliths requires
a different procedure than that described for auger or ingrowth samples.
Washingpinboard samples is facilitated by soaking overnight in water, deep
freezing (for clay soils), soaking in oxalic acid (for soils with free calcium car-
bonate) or soaking in hexametaphosphate, preferably under vacuum. Whatever
the pretreatment used, gent1e washing must follow; a rotating or oscillating
sprinkler is ideal, but pinboard samples can also be washed under more primi-
tive conditions with a hose attached to pressurised water or a locally constructed
"water tower" or by pouring small buckets of water over the sample. The pin-
board is put at a slight angle and washing starts at the lower part. During the pro-
ce dure some caution is need to avoid the formation of sharp ridges in the profile,
which can cause the breaking of roots. Even when the roots do not break, there
is a change in their positions, moving downwards which affects root distribu-
tion. To avoid this, the coarse mesh screen is lifted, and supported with wooden
strips when the soillayer becomes very thin. The roots then become fixed on
the mesh screen and are not pushed downwards by the water. Root losses from
a pinboard can be collected by placing a large sieve under the waste-pipe. This
indicates how carefully washing is being conducted and how well the root system
is secured. Some losses are acceptable if the main purpose is the rooting pattern,
but for quantitative studies care must be taken to minimise root displacement.
During the washing process large-sized organic debris can be carefully
removed from the board using forceps. Soil layers or other feautures which
become visible during the washing may be marked with string using the pins
as points of reference.
After washing away the soil, the root system is lifted by the mesh screen, pho-
tographed (on a black cloth as background) and/or cut according to soillayers,
depth zones and/or distance to the plant to obtain root biomass and/or root
length. Total plant root biomass can be estimated by integrating root weight
densities measured per zone and deptl). over the relevant soil volume.
6.3.2 Automatic Washing
Various root wash devices have been developed (B6hm 1979). The first to be
commerciallyavailable (the Gillison root washer), was a hydropneumatic elu-
triation system (Fig. 6.7) developed by Smucker et al. (1982).
J
Fig.6.7. Schematic representation of (
Smucker's system. A High kinetic-
energy washing chamber, B elutriation
chamber, C transfer tube, D low
I--~--I
kinetic-energy primary sieve,
o
E secondary sieve
80
This system consists of a high-kinetic energy water vortex, generated in

a chamber in which roots and organic plant residues are separated from soil.
Floating roots and associated materials pass onto low-kinetic energy separa-
tion screens (multiple screens may be stacked) submerged in water. The
mechanical separation of roots is achieved by a closed system using water pres-
sure and compressed air to isolate and deposit roots on a submerged sieve (Byrd
et al. 1976; Smucker et al. 1982; Mackie-Dawson and Atkinson 1991).
According to Smucker et al. (1982) and Smucker (1992) this method retains
almost an roots >0.05 mm diameter, with attached root hairs and nodules Cif
present), as well as some fruiting bodies of fungai pathogens. Nevertheless,
roots can sometimes remain somewhere in the system and a periodic check of
washing water and residues should be made to see if fine roots are being lost.
Such losses should be kept less than 5% of root weight (Van Noordwijk 1993).
More recently another device, based on a design from the CSIRO Cun-
ningham Laboratory, in Australia, has become commercially available (Delta-T
root washer). The method proposed by Pallant et al. (1993) is also suitable for
collecting very fine roots, with a minimum diameter of 0.025 mm, and is an
inexpensive method.
The main advantage of automatic washing is the standardisation of the
process, even when different operators do this work. Labour time is also saved
by this washing system, but after elutriation roots must stiH be sorted by hand,
and this may take several hours (Van Noordwijk 1993). Another advantage is
that automatic washing can be used in research with radioisotopes, since the
system is totally closed.
6.3.3 Chemical Dispersing Agents
For soils with a high clay content, chemical dispersing agents can be used to
help disperse the soil particles and facilitate root washing. A list of some of these
chemicals is presented in Table 6.2. Some form of sodium metaphosphate is
the most commonly used chemical. Weak acids can be used on soils containing
calcium carbonates as binding agents.
Chemical dispersing agents can damage the root system, creating problems
for further observations. For example, Van Noordwijk (1993) mentions that
soaking the samples in a 5% sodium hexametaphosphate solution speeds up
the process of washing roots from clay soil, but the roots become discoloured,
particularly in soil with high organic matter content, making the subsequent
identification of live roots more difficult. So, the decision of when to use these
chemicals must depend on the objectives of a given study.
6.3.4 Errors
The errors that occur during the washing and cleaning of samples are loss of
fine roots, loss of dry weight; changes in nutrients content of the roots; incom-
plete separation of root, soil and debris; poor distinction between live and dead
roots; and personal screening differences. Precautions to solve the problems can
lead to conflicting situations. To quantify errors, samples can be split or a few
samples analysed as a control, checking the Iosses.
6.3.4.1 Loss of Fine Roots
It is almost inevitable that not alI the roots in the soil-root sample are recov-
ered during the washing and cleaning stages. Most of the Iosses during this stage
Table 6.2. Listing of some dispersing chemicals used for root studies by different authors
Chemicals Reference Concentration
Na.P20 7 Schuurman and Goedewaagen (1971) 0.27%

Na3PO. Cassei et al. (1995) 0.2M
(NaP03)6 Van Noordwijk (1993) 5%
NaCI Tardieu and Manichon (1986) Sat. solution
HCI Băhm (1979) 3-5%
(COOH)2 Heringa et al. (1980) IOglI
CH 3COOH Mitchell et al. (1993) 0.25-2.5M
NaHC0 3 + CH 3COOH Pallant et al. (1993) Not given
Detergents Kiicke et al. (1995) Not given
are fine roots, which can comprise a major loss in terms of root length but are
not so critical in terms of root weight.
The mesh size of screens used to collect roots plays an important role
in determining the amount and size of the recovered roots. Amato and Pardo
(1994) compared 2 and 0.2mm mesh sieves (4 and 0.04mm2), and found
that recovered root length for wheat and field beans tended to be an order of
magnitude higher when using the smaller mesh sieve. Roots collected on the
2 mm sieve represented on average 55% of the weight but only 10% of the
total length collected using a 0.2 mm sieve. With a 1 mm sieve the recovered
root weight was 75%, but the recovered root length was stiH only 34% of the 0.2
mm sieve. Caldwell and Fernandez (1975) recovered more than twice as much
root biomass using a mesh of 0.03mm2 instead of 0.2mm2• The use of coarse
screens is therefore more acceptable for root weight than for root length deter-
minations and reduces the time needed for cleaning the samples from organic
debris.
6.3.4.2 Loss of Dry Weight
Loss of root dry weight during washing and storage of root samples has been
observed for several crops grown in nutrient solution (Van Noordwijk and
Floris 1979; Floris and De Jager 1981; Grzebisz et al. 1989). Losses were typically
20-40% of the original root dry weight. Some combinations of washing and
storage treatments resulted in up to 50% loss of the original root dry weight.
Five to 10% losses in dry weight was reported over the 24 h immediately after
harvest (stored at 20 ac in a moist environment), prior to any washing or longer
term storage treatments being imposed. This initialloss was attributed to root
respiration (Floris and De Jager 1981). Using sodium pyrophosphate as a dis-
persing agent tended to increase losses at the washing stage by 10-15% com-
pared with washing without a dispersing agent.
Grzebisz et al. (1989) found that, for most of their washing and storage
treatments, loss of cell walI material was limited. They therefore concluded that
most of the dry matter loss comprised cell contents, not root tissues such as the
epidermis or cortex. This agrees with the lack of change in root diameter during
washing and storage reported by Floris and De Jager (1981).
It has been suggested that a correction factor in the range 1.4 to 2.0 could
be used to account for dry matter losses of 30-50% due to handling of root
samples (Grzebisz et al. 1989). As the authors point out, these correction values
were obtained from roots grown in nutrient solution, which may result in a
different root structure and chemical composition from field-grown roots.
However, Grzebisz et al. (1989) suggest relative root dryweight losses from field-
grown crops could be even higher, particularly when grown in conditions of
limited water or nutrient availability.
Variation in dry weight losses can be reduced by treating alI samples uni-
formly. When working with a large number of samples it is recommend that
the samples be frozen after washing. This treatment gives high dry weight losses
but extra handling of the roots after thawing does not involve new dry
weight losses.
6.3.4.3 Change in Chemical Composition
There are conflict ing reports concerning the importance of nutrient loss from
roots during washing. B6hm (1979) cites work showing nutrient loss during
washing of agricultural root systems was not significant. In contrast, G. Brouwer
(unpubl. data) has observed that the nutrient content (N, P, Ca, K), as a per-
centage of dry matter remaining, can change considerably in a short time. For
calcium, manganese and iron contents, Evdokimova and Grishina (1968) report
a de crease of lO-15% in samples that were washed for 2h compared with
samples which were washed quickly. Grzebisz et al. (1989), comparing different
washing and storage treatments, found that N losses were smaller than dry
matter losses. Consequently, the root N content (as a percentage of dry matter
remaining) in some cases increased by up to lO% of the original value. This can
lead to overestimates of N content.
B6hm (1979) suggested that nutrient loss during root washing depends
mainly on how old the roots are. Brown, decaying roots Iose more nutrients than
white, young roots.
Lignin content typicalIy increases as a percentage of dry matter during
decomposition of organic residues (Parr and Papendick 1978; Jawson and Elliot
1986).
When the root biomass is used for chemical analyses, quick washing on
a coarse sieve, with some control samples is preferred. Samples should not be
frozen before the effect of freezing on the analytical results is known. Nutrient
contents are generalIy only reliable if sample handling is completed within 1
day (Van Noordwijk 1993).
6.3.4.4 Presence of Debris and Soil Partic/es
The amount of debris and soil particles present in washed roots is variable,
depending on the conditions under which roots are grown. In the last stage of
cleaning the distinction between small roots and debris can be difficult and for
this reason it is better to try to standardise the cleaning procedure as much as
possible.
When washing by floatation (as opposed to sprinkling) it is possible to sep-
arate the majority of debris from roots by repeating the decanting process.
Debris which are lighter than living roots fioat more readily than roots and can
be picked up. Some care during this process is needed because some fine roots
can come together with the debris.
The type of soil can modify the amount of roots separated from the soil
itself. In general, the coarser the soil the easier is the recovery of roots. Small
particles can adhere more closely to the roots, especially when the roots have
an abundance of root hairs. This can break the root hairs during washing.
The adherence of fine soil particles to the fine root fraction in particular,
despite careful washing, can be a problem and results in an overestimate of
root dry weight. B6hm (1979) reported that soil particles could account for up
to 50% of the sample weight. This error can be avoided, or at least checked, by
determining the weight of the ash-free organic root matter (see Sect. 6.5).
In organic soils, where the microbialload is usually high, the roots present
in the soil cores can be degraded rapidly.
Separation of soil particles and organic debris from roots stiH remains
a big problem which has led some investigators to concentrate on developing
techniques to determine accurately the totallength of roots in samples without
removing the debris. Dowdy et al. (1995) developed a technique for partition-
ing live root images from debris using imaging techniques, based on the
assumption that the length : width ratio for root segments differs greatly from
that of soil and organic debris.
6.3.4.5 Presence of Dead Roots
It can be difficult to distinguish between dead and live roots. Probably the best
method is to check root elasticity dur ing the washing procedure. Annual roots,
which break when stretched a Httle, are dead and those with some elasticity
are alive. With the auger method it is useful to have a fallow treatment to test if
roots are stiH present from the previous crop, especially if sampling early in the
growth of the current crop.
Many staining techniques are available (see Chap. 10), not only for increas-
ing the discrimination between live and dead roots but also for increasing the
contrast for automatic measurement of roots. The most appropriate technique
for any particular experiment should be determined by testing a few samples
taken before the main sampling date.
6.3.4.6 Operator Differences
Differences between opera tors in sample handling, mainly during the stages
of decantation and cleaning up the root samples, can cause errors. The resolu-
tion of the human eye, time pressure, operator fatigue, different criteria for dis-
tinction between dead and live roots, the decision to stop cleaning, etc. are
reasons for these errors.
It is preferable if the same person does the work, since this gives more
comparable data. When several people are involved in the same work care should
be taken to avoid bias in the measurement results. It is advisable to record the
operator for each sample, to allow calculation of a correction if necessary.
6.3.5 Time Investment
The time invested in cleaning roots, often recognised as a problem in root

systems studies, is highly variable. It depends on: the procedure used; the crop
itself; the stage of development of plants; and on the content of organic debris
and other materials in the sample. Whereas for the upper layers of a grass stand
it can take 1 day to clean up one or two samples, in the same period of time it is
possible to wash and clean 40 samples ofleek (G. Brouwer, unpubl. data).
B6hm et al. (1977) found that 9.5 person h were required to collect, wash
out and measure soybean roots from twelve 15 cm increments of a 180 cm core
from a silt loam soil.
T.C. Kaspar (pers. comm.) and Kaspar et al. (1995), working with core
samples 10 cm in diameter and 15 cm long, for maize roots grown in a silty clay
loam soil, has estimated that it required 0.16 person h/sample to take a core with
a hydraulic sampler and an eight person crew, 0.30 person h/sample to wash out
the soil using a hydropneumatic elutriator based on Smucker's design and a four
person crew, and 0.30 person h/sample to remove organic debris.
6.4 Methods of Storage
Ideally, root measurements should be taken immediately after sampling.

However, the logistics of processing the large number of samples that are
required from an experiment mean that it is often necessary to store roots
before measurement, sometimes for long periods.
6.4.1 Storage Before Washing
As a rule, samples should be washed immediately after being taken from

the field and then stored in a suitable manner. This minimises the period for
root respiration and microbial degradation of the roots after sampling. If it is
not possible to wash samples immediately, they can either be cold-stored or
dried.
Cold-storage should only be considered for temporary storage «2 to 3

days) because the treatment does not stop root decay but only slows it
down. Drying at 60-75 °C (Schuurman and Goedewaagen 1971) offers a longer-
term storage option. An alternative long-term storage option is to freeze the
samples (e.g. Kiicke et al. 1995), butthe sheer bulk of material that would usually
need to be stored often precludes this in practice. Whichever option is used,
because of their bulk it will probably take at least several hours before the
samples attain the desired temperature, during which time some root degrada-
tion will occur.
6.4.2 Storage After Washing
Cold Storage. Washed fresh roots can be stored for up to 48 h in a refrigerator

(Gregory et al. 1992). During storage, the roots should be kept moist to avoid
desiccation. This can be achieved either by wrapping samples in a damp paper
towel (Gregory et al. 1992), or by storing the samples in suspension with water.
If the storage period is longer than 2 days, a chemical preservative should be
added to the suspension (Table 6.3).
Freezing. Freezing at -20°C (Schuurman and Goedewaagen 1971) provides a

long-term storage option. Samples can be stored in small plastic bags or boxes,
with a Httle water. The addition of water helps to avoid root damage during
freezing.
Drying. Drying the root samples is another long-term storage option. Samples
should be dried at 60-75 °C to avoid the pulverisation that can occur when roots
are dried at 100°C (Schuurman and Goedewaagen 1971).
Table 6.3. Listing of some chemicals used for root storage by different authors
Chemicals Reference Concentration
Formalin and thymol Schuurman and Goedewagen (197l) 3-5% and 3%

Ethanol, formalin, acetic Tennant (1976) 50%; 6.5%; and 2.5%
acid and water (FAA)
Ethanol Biihm (1979) 15-20%
Methanol Henry and Deacon (1981) 70%
Formalin Tanaka et al. (1993) 10%
Thymol Marcum et al. (1995) 3%
Isopropanol Kiicke et al. (1995) 15%
Chemicals. A range of chemical agents have been reported in the literature as

satisfactory for long-term root storage (Table 6.3). Formalin and alcohol are the
most commonly used preserving agents. A 5% formalin solution or a 15-20%
alcohol solution can preserve roots satisfactorily for several months at 10 °C
(B6hm 1979). The alcohol concentration required is proportional to the storage
temperature. The higher the storage temperature, the greater the alcohol con-
centration required. Combinations of low temperature with chemical preserv-
ing agents seem to prove most satisfactory.
6.4.3 Effect of Storage Methods on Root Properties
Storage methods can affect the condition of the roots. IdeaHy, a storage method
should be chosen that does not influence the root property that is to be mea-
sured. AH samples must be stored in the same manner.
For chemical analyses, roots need to be dried as quickly as possible before
storage. When this is impossible, it is recommended that a zero control be taken
to quantify the changes prior to drying. Samples must not be stored in a freezer
because soluble nutrients can be lost from the roots when subsequently thawing
the sample for analysis. Nutrient content, as a percentage of dry matter, begins
to change after a few days in cold storage at 4°C.
For physical analyses, a wide range of storage methods is acceptable, though
the best method is to store the fresh samples in a freezer. Alternatively, samples
can be dried at 60-75 °C (Schuurman and Goedewaagen 1971) and then stored.
On re-wetting, the roots more or less completely recover their original root
length and diameter. Any effect of drying and re-wetting on the physical para-
meters should be similar for aH the samples. Williams and Baker (1957) report
that stor ing roots in formalin solution had no significant effect on herbaceous
root dry weight.
6.5 Root Quantification
A number of parameters may be estimated from roots obtained with the

methods described in this chapter, the most frequent being root weight, length,
diameter, surface are a and branching pattern. The parameters measured
depend on the purpose of the experiment and how the researcher intends to
use the root data. For example, when studying the absorptive capacity of the
root system, root length and surface are a are important parameters. Net ion
influx into roots is influenced by root diameter. In other studies, it might be per-
tinent to know the proportion ofbelow-ground dry matter relative to the above-
ground dry matter of the plant, and hence the weight of the root samples would
be recorded. Whatever parameter is measured, it is usualIy related to the volume

of the soil sample, which can be the volume of the entire monolith obtained
with the pinboard, or a smaller volume, such as that obtained with auger
sampling.
6.5.1 Root Weight
Root weight is estimated either from fresh roots or, more frequent1y, from oven-
dried roots. Fresh weight is often used in plant pathology studies, for example
investigating nematodes and fungi in roots (Bohm 1979). For determination of
root fresh weight, it is recommended that a standardised procedure be applied
to remove alI the water adhering to the roots, before weighing. This can be
achieved by submitting the roots to low-speed centrifugation, typicalIy for 30 s
(van Noordwijk and Floris 1979; Grzebisz et al. 1989).
Root dry-matter is a useful measure of plant investment in the root system
and also of the contribution of roots to the soil humus, particularly when related
to shoot biomass. However, it is not valid to assume that root weight is corre-
lated with root activity (Bohm 1979). Root length, particularly of fine roots, is
generalIy considered a better indication of root activity.
Root dry weights are generally recorded for root samples which have been
oven-dried at 65-75 DC until there is no further change in weight (typicalIy
24h). Adherence of fine soil particles to the roots is sometimes a problem. In
these situations the weight of the ash-free organic root matter is determined.
The procedure involves placing the weighed oven-dry roots in a muffle furnace
at 650 DC for 5 h to burn off the organic matter. The weight of the residue is then
recorded. The difference between root dry weight and the weight of ash residues
represents the ash-free organic dry weight.
6.5.2 Root Length
Various techniques, both manual and automated, are available for determining
root length. This section considers primarily manual methods. Automated
methods, in particular the use of image analysis software, are considered
separately in Chapter 10.
6.5.2.1 Direct Measurement
Direct measurements of root length can only practicalIy be made on large diam-
eter roots, e.g. tree roots. Direct measurements of herbaceous plant root lengths
are rarely made, and then only primary roots (e.g. Sivakumar and Salaam 1994).
It is simply not feasible to undertake the direct measurement of the total root
length (including fine roots) of a plant, or even of a root sample.
6.5.2.2 Line Intersect Method
The root length of a sample can be estimated using intercept counting tech-
niques (Newman 1966; Tennant 1975). Newman's technique is based on the rela-
tionship between root length and the number of intercepts between roots,
spread over a surface of defined area, and randomly arranged lines of known
length in that area. Marsh (1971) simplified Newman's method, and Tennant
(1975) tested and popularised this modified method.
The modified line intersect method has become the standard manual tech-
nique for estimation of root length (See Box 6.4). The line intersect method
gives an estimation of the root length of the sample, and is not a direct mea-
surement. Sources of error arise from the random arrangement of the roots on
the grid, root visibility, the definition of an intersection, and operator bias and
fatigue. With care, it is possible for a single individual to obtain coefficients of
BOX 6.4. TheModin.ed Newman Line-intersect Method

Spread the roots, with the minimum overlap possible, over a regular grid
of indeterminate dimensions, but of known regular grid units (e.g. 1 cm
grid squares). Where necessary, tease the root material apart or cut it
into smaller pieces, to avoid overlap. Placing the samples on the grid in a
shallow layer of water in a dish (or on a glass plate) aids separation of
the roots.
Count the total number of intersections between the roots and the
horizontal (H) and vertical (V) grid lines. Assign a count of 1 to a root
crossing a line, a root ending touching a line and a curved root portion
touching a line. Curved root portions which lie on or along a grid line
are assigned a count of 2 (Tennant 1975).
The relationship between the root length and the number of root
intersections with the grid lines is given by:
root length (cm) = ni 4 x00. of intersects x grid unit (cm)
To improve the contrast of roots it can be helpful to stain them.
Alternatively, place a coloured background under the roots. To optimise
working efficiency at an accepta bie random error le vei, choose the grid
size to obtain about 400 intercepts (200 H + 200 V) per sample (Van
Noordwijk 1993).
variation of the root length estimate of 5% or less (Tennant 1975). However, this
is the minimum likely error. Coefficients of variation for the line intersect
method are typically 10 to 15% (Bohm 1979; Bland and Mesarch 1990; Farrell
et al. 1993). Bland and Mesarch (1990) reported a bias in root length estimates
by observers, with some observers yielding consistent1y low or high estimates
relative to the mean.
Steps can be taken to minimise the error associated with the technique.
These include: training new observers and comparing their results against
experienced observers; ensuring adequate lighting; using a magnifying glass to
aid root visibility; improving the colour contrast between the roots and the back-
ground; and ensuring the observers have regular rest breaks to avoid fatigue.
The line intersect technique can be mechanised, using photoelectric coun-
ters such as those of Rowse and Phillips (1974), Richards et al. (1979) and
Wilhelm et al. (1982). The automation eliminates operator bias from the esti-
mation and should therefore reduce error. However, root samples do need to be
cleaned more carefully since it is difficult for automated systems to distinguish
between roots and debris. Details of automated methods for root analysis are
given in Chapter 10.
6.5.2.3 Visual Estimation Method
The visual estimation of the root length is a rapid low-cost method for estimat-
ing root length. At its simplest the sample root length can be described by a sub-
jective rating where, for example, 1 indicates low root length and 9 indicates high
root length. This method has been used to screen for cucumber root length as
part of a cultivar evaluation programme (Walters and Wehner 1994). Visual esti-
mation may also be used to obtain a quantitative root length estimate of samples
recovered using the pinboard method or from auger samples (Naab 1994; Gaze
1996). This essentially involves visual comparison of root samples of unknown
length with standard root samples of known length, enabling a reasonable esti-
mate of the unknown root length to be made (see Box 6.5).
B OX 6.5. Visual Estimation of Root Length

After washing, store the roots in a chemical preserving solution (Table
6.2) until required. Comparison of samples for root length estimation is
easier when the roots are suspended in solution.
Select a sub-set of the washed samples which covers the range of
root lengths in aH the samples, based on a vis ual subjective rating.
Determine the root length of the samples in this sub-set as accurately as
possible using the line intersect method (see Box 6.4). These samples
comprise the standard samples of known root length.
200 (a) .... (b) ;" (e) /

I
I I
I
.!! 150 I
I
I
~ I
I
m I I
I
I
'O 100 I I
I
I
I I
I
~
c: 50
o
o 5 10 15 20 25 30 35 40
. time (man hours)
Fig.6.8. Time required to determine root length using the line intersect method (solid line)
and visual estimation (broken lines). The comparisons assume that 5 samples per h can be
counted using the line intersect method, and that 30 samples per h can be estimated using
the visual estimation method. For the visual estimation method: line a assumes standard
samples are already available; line b assumes 30 standard samples are first counted once using
the line intersect method; line c assumes that 30 standard samples are counted twice before
visual estimation of the remaining samples
Estimate the root length of each of the remaining samples by

comparing them against the standard samples. If necessary, estimate the
root length by interpolation between the two standard samples which
appear most similar to the root length of the sample being estimated. If
the roots are stored correct1y to prevent degradation, it is possible to use
the same standard samples from one sampling date to the next.
Standards should be for the same species (and possibly cultivar), have
roots of similar condition and size, and should be stored in an identical
solution volume and sample container to those samples against which
they will be compared.
The major advantage of the visual estimation of root length is the
time saved when compared with the line intersect method. Gaze (1996)
reported that a single person could score 30-40 samples per h, compared
with counting about five samples per h using the line intersect method.
Figure 6.8 illustrates the major time-saving that can be achieved using
the visual estimation method.
There is inevitably some loss in the accuracy of the estimation of root length
when using a visual estimation method as opposed to the line intersect method.
However, other measurements of root length, such as the line intersect method
and image analysis techniques, are themselves in fact only root length estimates.
For visually estimated root lengths of millet [Pennisetum glaucum (L.) R.Br.], the
coefficient of variat ion was reported as just over twice that of the line intersect
method (Gaze 1996). No systematic bias was found between observers using the
visual estimation method. There can, however, be a tendency to underestimate
root length using the visual method at high root lengths (Walters and Wehner
1994; Gaze 1996). This upper limit is probably a function of root length density
in the sample solution volume, rather than absolute root length.
6.5.3 Root Diameter and Surface Area
Root diameter can be measured on samples spread out on a grid, using a binoc-
ular microscope with an ocular micrometer.According to Van Noordwijk (1993)
at least 20 readings per sample are required.
Total root surface area has been estimated from other root parameters, like
root length and diameter, by solution adsorption measurements, or by using
photoelectric devices (B6hm 1979).
Recently, the availability of image analysing computers has made the deter-
mination of root length, diameter or surface area possible with more accuracy
and with less time investment (see Chap. 10).
6.5.4 Presentation of Root Data
Root length and root weight are often related to soil volume and are referred
to as root length density - Lv (cmrootcm-3soil) or root weight density
(groot cm-3soil). These are particularly useful for describing the spatial distribution
of the root system, and are also useful data for soil-vegetation-atmosphere
transfer models.
Root length and weight can also be related to soil surface area - La (cm root
cm-2soil) and (grootcm-2sOil). This facilitates comparison of root data with mea-
surements of above ground plant production, which are typically made per unit
soil surface area (e.g. yield, leaf area index).
The specific root length (cmroot g-l root dry weight) is used to give an indication
of root thickness.
Comprehensive summaries of published rooting data have been made by
Van Noordwijk and Brouwer (1991), Canadell et al. (1996) and Jackson et al.
(1996).
6.6 Conclusions and Perspectives
Rapid developments in computing hardware and software hold much promise

for assisting in the measurement of root length, root diameter and root branch-
ing patterns and architecture. However, to use these systems, the researcher must
already have taken the samples from the field and washed the soi! away from the
roots. For the foreseeable future, this aspect of root sampling is likely to remain
relatively time consuming and tedious when compared with taking measure-
ments of the aerial parts of plants. Nevertheless, quantifying the location,
amount and role of plant roots is essential to understanding plant growth on an
individual plant basis, as well as understanding the competition and comple-
mentarity between plants and between flora and fauna in the wider ecosystem.
The methods described in this chapter (together with those in Chapter 7) are
likely to remain the basis of quantitative root research for many years to come.
References
Albrecht D (1951) Verbesserung der Spatendiagnose. Dtsch Landwirtsch 2: 41-43

Albrecht D, Fritzsche KH, Winkler S (1953) Weitere Entwicklung des Strukturbohrers. Dtsch
Landwirtsch 4: 206-208
Amato M, Pardo A (1994) Root length and biomass losses dur ing sample preparation with dif-
ferent screen mesh sizes. Plant Soil 161: 299-303
Baarstad LL, Rickman RW, Wilkins D, Morita S (1993) A hydraulic soi! samp!er providing
minimum field plot disruption. Agron J 85: 178-181
Bland WL, Mesarch MA (1990) Counting error in the line-intercept method of measuring root
length. Plant Soil 125: 155-157
B6hm W (1979) Methods of studying root systems. Ecological Studies: Analysis and Synthesis,
voi 33. Springer, Berlin Heidelberg New York
B6hm W, Maduakor H, Taylor HM (1977) Comparison of five methods for characterizing
soybean rooting density and development. Agron J 69: 415-419
Byrd DW Jr, Barker KR, Ferris H, Nusbaum CJ, Griffin WE, Small RH, Stone CA (1976) Two semi-
automatic elutriators for extracting nematodes and certain fungi from soil. J Nematol 8:
206-212
Caldwell MM, Fernandez OA (1975) Dynamics of Great Basin shrub root systems. In: HadleyNF
(eds) Environmental physiology of desert organisms. Dowden, Hutchinson and Ross,
Stroudsburg, Pennsylvania, pp 38-51
Caldwell MM, Virginia RA (1991) Root systems. In: Pearcy RW, Ehleringer J, Mooney HA, Rundel
RW (eds) Plant physiological ecology: field methods and instrumentation. Chapman and
Hali, London, pp 367-398
Canadell J, Jackson RB, Ehleringer JR, Mooney HA, Sala OE, Schulze ED (1996) Maximum
rooting depth of vegetation types at the global scale. Oecologia 108: 583-595
Cassel DK, Raczkowski CW, Denton HP (1995) Tillage effects on corn production and soil
physical conditions. Soil Sci Soc Am J 59: 1436-1443
Crozier CR, King LD (1993) Corn root dry matter and nitrogen distribution as determined by
sampling multiple soi! cores around individual plants. Commun Soi! Sci Plant Anal 24
(11&12): 1127-1138
Cuevas E, Medina E (1988) Nutrient dynamics within Amazonian forests: Part II. Fine root
growth, nutrient availability and !itter decomposition. Oecologia 76: 222-235
De Jager A (1982) Effects oflocalized supply of H,PO., NO" SO., Ca and K on the production
and distribution of dry matter in young maize plants. Neth J Agric Sci 30: 193-203
Dowdy RH, Smucker AJM, Dolan MS, Ferguson JC (1995) Root length measurements of
segments washed from soil cores without removal of extraneous debris. Proc 14th Long
Ashton Int Symp: Plant roots - from cell to systems. Sept 1995. Long Ashton, Bristol,
England
Evdokimova TI, Grishina LA (1968) Productivity of root systems of herbaceous vegetation on
flood plain meadows and methods for its study. In: Methods of productivity studies in root
systems and rhizosphere organisms. Int Symp USSR 1968. USSR Academy of Sciences,
Nauka, Leningrad, pp 24-27
Fabiăo A, Persson HA, Steen E (1985) Growth dynamics of superficial roots in Portuguese plan-
tations of Eucalyptus globulus Labili. studied with a mesh bag technique. Plant Soi! 83:
233-242
Farrell RE, Walley FL, Lukey AP, Germida JJ (1993) Manual and digitalline-intercept methods
of measuring root length: a comparison. Agron J 85: 1233-1237
Floris J, De Jager A (1981) Een schatting van het verlies aan drogestof en van de verandering in
diameter van wortels van engels raaigras (Lolium perenne) dor bemonsteren, bewaren en
spoelen. Inst Bodemvruchtbaarheid Haren-Gr Rapp: 1-81
Floris J, Van Noordwijk M (1984) Improved methods for the extraction of soi! samples for root
research. Plant Soil 77: 369-372
Gaze SR (1996) Water balance of farmer-managed millet and fallow-savannah on sandy soi!s in
south west Niger. PhD Thesis, Department of Soi! Science, University of Reading, Reading
Gregory PJ, Tennant D, Belford RK (1992) Root and shoot growth, and water and !ight use effi-
ciency ofbarley and wheat crops grown on a shallow duplex soil in a Mediterranean-type
environment. Aust J Agric Res 43: 555-573
Grzebisz W, Floris J, Van Noordwijk M (1989) Loss of dry matter and celI contents from fibrous
roots of sugar beet due to samp!ing, storage and washing. Plant Soil 113: 53-57
Hairiah K, Van Noordwijk M, Setijono S (1991) Tolerance to acid soil conditions of the velvet
beans Mucuna pruriens var. utilis and M. deeringiana. 1. Root development. Plant Soi! 134:
95-105
Hairiah K, Van Noordwijk M, Stulen 1, Meijboom FW, Kuiper PJC (1993) P nutrition effects on
aluminium avoidance of Mucuna pruriens var. utilis. Environ Exp Bot 33: 75-83
Hansson A, Andren O, Steen E (1991) Root production of four arable crops in Sweden and its
effect on abundance of soil organisms. In: Atkinson D (ed) Plant root growth: an ecologi-
cal perspective. Blackwell, Oxford, pp 247-266
Henry CM, Deacon JW (1981) Natural (non pathogenic) death of the cortex of wheat and
barley seminal roots, as evidenced by nuclear staining with acridine orange. Plant Soi! 60:
255-274
Heringa JW, Groenwold J, Schoonderbeek D (1980) An improved method for the isolation and
the quantitative measurement of the crop roots. Neth J Agric Sci 28: 127-134
Jackson RB, Canadell J, Ehleringer JR, Mooney HA, Sala OE, Schulze ED (1996) A global analy-
sis of root distributions for terrestrial biomes. Oecologia 108: 389-411
Jawson MD, Elliot LE (1986) Carbon and nitrogen transformations during wheat straw and root
decomposition. Soil Biology and Biochemistry, 18 (1): 15-22
Kalela EK (1950) On the horizontal roots in Pine and Spruce stands.Acta For Fenn 57 (2): 62-68
Kaspar TC, Logsdon SD, Prieksat MA (1995) Traffic pattern and tillage system effects on corn
root and shoot growth. Agron J 87: 1046-lO51
Kolesnikov VA (1971) The root system of fruit plants. Mir Publishers, Moscow
Kiicke M, Schmid H, Spiess A (1995) A comparison of four methods for measuring roots of field
crops in three contrasting soils. Plant Soil 172: 63-71
and the interpretation of results. In: Atkinson D (ed) Plant root growth, an ecological per-
spective. Blackwell, London, 25-47
Marcum KB, Engelke MC, Morton SJ (1995) Rooting characteristics of buffalograsses grown in
fiexible plastic tubes. HortScience 30 (7): 1390-1392
Marsh B.a'B (1971) Measurement of length in random arrangements of lines. J Appl Ecol 8:
265-267
Mitchell AR, Shouse PJ, Rechel EA (1993) Using acetic acid to wash roots from the soil. Soil Sci
Plant Anal 24: 15-16
Naab JB (1994) Interaction of canopy and root system in water use of potato. PhD Thesis, Uni-
versity of Reading, Reading
Newman EI (1966) A method for estimating the totallength of root in a sample. J Appl Ecol3:
139-145
Nielsen CCN (1990) Methodische und okologische Untersuchungen zur Sturmfestigkeit der
Fichte. In: Einfiiisse von Pfianzenabstand und Stammzahlhaltung aufWurzelform, Wurzel-
biomasse, Verankerung sowie auf die Biomassenverteilung im Hinblick auf die Sturmfes-
tigkeit der Fichte, Schriften aus der Forstl Fak Uni Gottingen ud Niedersiichs Forst
Versuchsanst. JD Sauerliinders, Frankfurt
Nielsen CCN (1994) Aspects of sustainability by afforestation of agricultural set-aside areas:
Development of roots and root/shoot-ratios (AIR3-ct93-1269), lst Progress Report, Royal
Vet Agric University, Kirkegaardsvej 3A, DK-2970 Hoersholm
Pallant E, Holmgren RA, Schuler GE, Mccracken KL, Drbal B (1993) Using a fine root extraction
device to quantify small diameter corn roots (::;0.025mm) in field soils. Plant Soil 153:
273-279
Parr JF, Papendick JR (1978) Factors affecting the decomposition of crop residues by mir-
cooorganisms. In: Iscwald WR (ed) Crop residue management systems. American Society
of Aoronomy, Madison, pp lOl-129
Passioura JB (1980) The transport ofwater from soil to shoot in wheat seedlings. J Exp Bot 31:
335-345
Prior SA, Rogers HH (1992) Portable soil coring system that minimizes plot disturbance. Agron
J 84: 1073-1077
Prior SA, Rogers HH (1994) A manual soil coring system for soil-root studies. Commun Soil Sci
Plant Anal 25: 517-522
Richards D, Goubran FH, Garwoli WN, Daly MW (1979) A machine for determining root length.
Rowse HR, Phillips DA (1974) An instrument for estimating the totallength of root in a sample.
J Appl Ecolll: 309-314
Schroth G, Kolbe D (1994) A method of processing soil core samples for root studies by sub-
sampling. Biol Fertil Soils 18: 60-62
Schuurman JJ, Goedewaagen MAJ (1971) Methods for the examination of root systems and
roots, 2nd edn. Pudoc, Wageningen
Sivakumar MVK, Salaam SA (1994) A wet excavation method for root shoot studies of pearl-
millet on the sandy soils of the Sahel. Exp Agric 30: 329-336
Smucker AJM (1992) Contemporary analytical methods for quantifying plant root dynamics.
In: Hiibl E, Lichtenegger E, Persson H, Sobotik M (eds) ISSR Symp. Root ecology and its
practical application. Klagenfurt, Austria, pp 721-726
Smucker AJM, McBurney SL, Srivastava AK (1982) Quantitative separation of roots from
compacted soi! profiles by the hydropneumatic elutriation system. Agron J 74: 500-
503
Spek LY, Van Noordwijk M (1994) Proximal root diameters as predictors of total root system
size for fractal branching models. II. Numerical model. Plant Soi! 164: 119-128
Steen E (1984) Variation of root growth in a grass ley studied with a mesh bag technique. Swed
J Agric Res 14: 93-97
Steen E (1991) Usefulness of the mesh bag method in quantitative root studies. In: Atkinson D
(ed) Plant root growth: an ecological perspective. Blackwell, Oxford, pp 75-86
Tanaka S, Yamauchi A, Kono Y (1993) Cultivar difference in response of root system to nitrogen
application in rice plant. Jpn J Crop Sci 62: 447-455
Tardieu F, Manichon H (1986) Caracterisation en tant que capteur d'eau de l'enracinement du
mals en parcelles cultivees.1. Discussion des criteres d'etude. Agronomie 6: 345-354
Taylor HM, Upchurch DR, Brown JM, Rogers HH (1991) Some methods of root investigations.
In: Mc Michael BL, Persson H (eds) Plant roots and their environment. Elsevier, London,
pp 553-562
Tennant D (1975) A test of a modified line intersect method of estimating root length. J Eco163:
995-1001
Tennant D (1976) Root growth of wheat. 1. Early patterns of multiplication and extension of
wheat roots including effects of levels of nitrogen, phosphorus and potassium. Aust J Agric
Res 27: 183-196
Van Noordwijk M (1993) Roots: length, biomass, production and mortality. Methods for root
research. In: Anderson JM, Ingram JSI (eds) Tropical soi! biology and ferti!ity, a Handbook
of Methods. CAB International, Wallingford, pp 132-144
Van Noordwijk M, Brouwer G (1991) Quantitative root length data in agriculture. In: McMichael
BL, Persson H (eds) Plant roots and their environment. Proc ISRR-Symp,Aug 21-26,1988,
Uppsala, Sweden. Elsevier, Amsterdam, pp 515-525
Van Noordwijk M, Floris J (1979) Loss of dry weight during washing and storage of root samples.
Van Noordwijk M, Purnomosidhi P (1995) Root architecture in reIat ion to tree-soil-crop inter-
actions and shoot pruning in agroforestry.Agrofor Syst 30: 161-173
Van Noordwijk M, Floris J, De Jager A (1985) Sampling schemes for estimat ing root density in
cropped fields. Neth J Agric Sci 33: 241-262
Van Noordwijk M, Spek LY, De Willigen P (1994) Proximal root diameters as predictors of total
root system size for fractal branching models. I. Theory. Plant Soi1164: 107-118
Veller F (1971) A method for studying the distribution of absorbing roots of fruit trees. Exp
Agric 7: 351-361
Vogt KA, Persson H (1991) Measuring growth and development of roots. In: Lassoie JP,
Hinckley (eds) Techniques and approaches in forest tree ecophysiology. CRC Press, Boca
Raton, pp 477-501
210 M.R.G. Oliveira et al.: Auger Sampling, Ingrowth Cores and Pinboard Methods
Walters SA, Wehner TC (1994) Evaluation of the US cucumber germplasm collection for root
size using a subjective rating technique. Euphytica 79: 39-43
Wilhelm WW, Norman JM, Newell RL (1982) Semiautomated x-y-plotter-based method of mea-
suring root length. Agron J 74: 149-152
Williams TE, Baker HK (1957) Studies of root development of herbage plants.l. Techniques of
herbage root investigations. J Br Grassl Soc 12: 49-55
CHAPTER 7
Trench Profile Techniques and Core Break Methods

M. van Noordwijk\ G. Brouwer2 , F. Meijboom 2 , M. do Rosario G. Oliveira\
and A.G. Bengough4
1 ICRAF . li CIFOR, PO Box 161, Situ Gede, Sindang Barang Bogor 16680, 16001Bogor,
Indonesia
2 Plant Research International, Postbus 14, 6700 AA Wageningen, The Netherlands
3 University of Evora, Apartado 94,7002-554 Evora, Portugal
4 Scottish Crop Research Institute (SCRI), Invergrowie, Dundee, DD2 5DA, UK
CONTENTS
7.2 Practical Aspects 212
7.2.1 Preparing Protile Walls for Root Observations 212
Choosing an Observation Plane 212
Preparing Access 213
Smoothing the Observation Plane 213
Removing Last Layers of Soil 213
Recording Roots and Protile Features 214
7.2.2 Root Counts on a Grid 214
7.2.3 Root Maps on a Plastic Overlay 215
7.2.4 Calibrat ion Samples 216
7.2.5 Core-Break Method 217
7.2.6 Time and Labour Requirements of the Methods 218
7.3 Data Analysis 219
7.3.1 Digitizing Root Maps and Overlays 219
7.3.2 Relation Between Intersection Point Density
and Root Length Density 220
7.3.3 Empirical Correlation for Core-Break and Profile Wall
Observations 222
7.3.4 Root Counts as Function of Depth and Horizontal Distance 223
7.3.5 Root Distribution Pattern Within a Zone: Nearest Neighbour
Distances 225
7.3.6 Spatial Correlation of Mapped Features 227
7.3.7 Root Position Effectivity Ratio, R pe r 227
7.3.8 Transition from 2-D to 3-D Distance 230
7.4 Conclusions and Perspectives 231
References 231

212 M. van Noordwijk et al.
7.1 Introduction
This chapter describes methods for root observations based on mapping or

counting root intersections with planes of observation in the soil. Normally
these planes of observation are either vertical or horizontal. Compared with
the methods based on washed root samples discussed in Chapter 6, these "profile
wall" methods have advantages as well as disadvantages. A major disadvantage
of the profile wall methods is that only a small part of a root is visible on such an
intersection and it is not easy to distinguish between roots of different species,
or between live or dead roots. Even the question of whether a whitish thread-like
object sticking out of a plane is a root and not an enchytraeid (pot worm) or
other soil organism may take some experience to answer (potworms move when
touched). Creating access to planes of observation via trenches can be a rather
destructive activity which is not welcome on small experimental plots, especially
those intended for long-term experiments. On the positive side, however, profile
wall methods can give a quick estimate of overall root distribution and can give
detailed information on spatial patterns of roots in their interaction with phys-
ical, chemical and biological characteristics of the soil profile. If maps are made
of root occurrence as well as any other readily observable feature, the toolbox of
geographical information systems and quantitative map analysis can be used to
analyze patterns, be it in only two dimensions.
In this chapter we will describe practical aspects of preparing profile walls
for observations and various methods for recording data (Sect. 7.2). In Section
7.3 we will focus on the analysis of data obtained, especially for root maps, and
discuss the inferences which can be made about three-dimensional reality on
the basis of two-dimensional observations.
7.2 Practical Aspects
7.2.1 Preparing Profile Walls for Root Observations
Field methods for root mapping include the following steps: a plane of obser-
vation must be chosen, and an access trench dug (Fig. 7.1A, B). The number of
profile walls observed per treatment is usually one but further observation
planes can be obtained from the same trench, a few centimeters (depending on
plant space in the row) beyond the previous profile wall. The surface of the plane
is then prepared and root locations recorded on plastic overlays or as tabulated
data (Fig. 7.1C, D). We will consider these steps in turn, as there is a range of
options for each of these.
Choosing an Observation Plane. Usually maps are made on vertical planes of

observation, perpendicular to crop rows and near the base of a plant, where
7 Trench Profile Techniques and Core 8reak Methods 213
Fig. 7.1. Direct root observations on trenches and profile walls. a Direct observations in a deep
trench perpendicular to hedgerows, b direct observations in a plough layer of an arable system,
with minimum disturbance to the soi! profile, c brushing away soil from a sugarbeet root system
to observe its interactions with different layers in the soi! profile, d mapping root intersections
on a polythene sheet, with a grid system as a guide. (Photographs: courtesy of Meine van
Noordwijk)
applicable. Horizontal planes give additional information on root orientation,

but this makes removing the last layers of soil difficult.
Preparing Access. For "plough layer" observations a trench 0.3 m wide and 0.3
m deep may be enough, as the observations can be made without the observer
having to get into the trench. Otherwise, a trench about 1 m across is needed
for access by the observer, with a depth and width depending on the size of the
maps to be made. To reduce damage to long-term experimental plots, sampling
locations are normally chosen close to the border of plots; this may compro-
mise on the representativeness of the sample. The soil from the soil pit or trench
should be put on the outer side, so as not to inBuence the observations. If topsoil
and subsoil are separated by canvas sheets dur ing excavation, disturbance to
the soil profile can be reduced, but the sampling sites should be marked and
avoided in future.
Smoothing the Observation Plane. A long knife blade is normally used to obtain
a Bat and smooth plane of observation. No further preparations are necessary
if the soil and root have very good contrast in colour, provided that the observer
examines the profile closely.
Removing Last Layers of Soil. On most soils it is advisable to wash a few

millimetres of soil from the prepared surface to facilitate observations. This can
be done with a knapsack sprayer, working from the top of the profile down-
wards; a disadvantage of spraying is that in a wet soil 'smearing' of surfaces can
easily cover fine roots. On heavy clay soils washing can be facilitated by cover-
ing the profile with a cloth soaked in sodium pyrophosphate (see Chap. 6) to
disperse the clay. As an alternative one can carefulIy blow away fine soil with a
"vacuum cleaner" (used in reverse) -live roots are remarkably resilient to this
and will generalIy remain visible.
Recording Roots and Profile Features. The number of root intersections can be
counted using a sampling grid, or the positions of the roots can be mapped on
polythene (plastic) overlays. This latter method gives most flexibility in subse-
quent analysis of the data. Appropriate light conditions should be created for
observation. Avoid direct sun-light, as it dries out roots quickly and gives too
sharp a contrast and willlead to water condensation if plastic overlays are used
for mapping. An umbrella or screen can be used to shade the plane of obser-
vation and give diffuse light.
7.2.2 Root Counts on a Grid
The simplest way of recording root occurrences is by overlaying the plane of

observation with a grid and counting the number of roots crossing the plane.
The size of the grid then depends on the purpose of the study and the level of
detail needed. Fine grids require more work, but allow data to be converted to
those of coarser grids (multiples of the fine grid), while the reverse process is
not possible. French researchers in particular (Tardieu and Manichon 1986;
Tardieu 1988) score the presence/absence of roots on a fine grid, rather than
counts of root sections seen on the plane. A frame of strings can be used as a
grid, either prepared in advance, or created on-site with pins, string, and pieces
of lead as weight. It is usualIy worth distinguishing between root diameter
classes in the counts.
Two types of criterion have been used for including roots in the counts.
In the "Reymerink method" alI roots exposed after removing a specified layer
of soil are counted (Schuurman and Goedewaagen 1971; B6hm 1976).
More recently, roots are only counted at the point where they emerge from the
soil in the plane of observation. The latter can be more easily related to root
length density per unit volume of soil (see below), but may not be as visualIy
rewarding as drawings of roots hanging on the profile walI recorded with the
first method.
7.2.3 Root Maps on a Plastic Overlay
Maps on plastic overlays can be made with permanent marker pen; roots are rep-
resented as 'dots', with different colo urs used for diameter classes or roots of dif-
ferent species where these are recognizable. Other features can be included such
as soil horizons, cracks or finer soil structural features, or local accumulations of
organic residues (Fig. 7.2). By spraying suitable dyes onto the excavated soil
surface, spatial patterns in other features of the profile wall such as soil pH or
redox status may be recorded. Water infiltration patterns via macropores can be
visualized with dyes such as methylene blue or fluorescein (the latter requires
UV light for observations and may be poorly visisble when adsorbed to the soil).
Macropores can also be made visible by infiltrating an iodine solution and cov-
ering the soil surface with a fine layer of starch powder; a blue colour will appear
where the iodine has reacted. Root maps should be wrapped in paper on a roll,
rather than being folded to avoid a "printing" process during storage.
As a special case of profile wall observations, resin-impregnated blocks can
be prepared and used for more detailed observations. This is similar to the "thin
sections" used in soil micromorphology (Kooistra and Van Noordwijk 1996;
Ringrose-Voase 1996).
t x
t
r:
o
o\~~~~ '~
x x
o x
o • o o x
x / " x o
x
_' x o
'. - _B.
o x o
x
x x
o
x x
o
X
x
o x
o
~~ o
x o
Fig. 7.2. Example of a root map with a number of features: x and o indicate two types of roots
(different species or diameter), the continuous lines indicate major structural features of the soi!,
the broken lines protile layers, the hatched areas recognizable plant residues ploughed into the
topsoil, and the shaded areas re Rect anaerobic areas (bluish-grey)
7.2.4 Calibration Samples
The number of root intersections per unit area of observation plane can be cal-
ibrated against the root length density per unit volume in the soil directly
behind the plane of observation by taking shallow sample boxes from well-
marked places in the profile (Fig. 7.3). The box size should correspond with (a
multiple of) the grids used for direct counts; where root maps are made, the
size of the sample box is arbitrary, but it should be marked on the root map.
Shallow sample boxes (about 2 cm deep) are better where large gradients of root
length density are expected. In very shallow samples, however, the root seg-
ments obtained may be so small that washing losses increase; variability in
actual sample size may also increase when the boxes are made shallow. The cal-
ibration samples can be washed and treated as described for other washed
samples in Chapter 6. Oliveira-Carvalheiro and Nepstad (1996) made profile
wall observations and empirically calibrated these against small block samples
up to a depth of 9 m.
A. B.
[!J\
10x2x10cm
D. o c.
-------
o
o
Lrv o
o
T o
-
o
sharp edge
Fig. 7.3. Root maps ean be eombined with small bloeks from which roots are washed (see Chap.
6) to derive an empirical relationship between root length density (em per em') and intersee-
tion point density on the map (em em-2 )
7.2.5 Core-Break Method
The core-break method (Schuurman and Knot 1957; Schuurman and

Goedewaagen 1971; B6hm 1979; Drew and Saker 1980) is a rapid and simple
method of observing and recording the presence of roots as a function of depth;
the method is treated here rather than in Chapter 6, as the basic observation of
number of root intersections per specified area of observation is the same as in
the profile wall methods. Cylindrical soil cores (see Chapter 6 for auger design)
are broken in half and the number of roots sticking out of both soil surfaces is
recorded (Fig. 7.4). The two results should be added together, as the same root
cannot be exposed on both sides after breakage; usually roots will break some
distance from the plane of observation and only show on one side. If root
branches are observed these should be ignored, as the count refers to a 'plane
of observation' and not to a volume.
Advantages of the core-break method are that samples are small, allowing
many replicates to be taken. Direct counts are faster than for washing the roots
from a whole sample, and a higher sample variability (the coefficient of varia-
tion for root counts may be 50-100% higher than for root length in washed
auger samples) can be compensated for by larger sample numbers; the time
needed per sample is considerably less. A selection of the samples can be
c.
Fig. 7.4. Core-break method for estimat ing density of roots intersecting with a horizontal plane
of observation
washed to obtain a direct calibration of root length versus counts. Schuurman

and Goedewaagen (l971) gave a series of standardized dot maps with loga-
rithmic increase in density which can be used for rapid assessment ("Knot
method"), especially in grasslands where point densities will be high in the
upper layers.
Disadvantages of the core-break method are that, similar to washed
auger samples, no correlation can be made between local root density and
other features of the soil, as recorded in the profile wall method. Bland (l989,
1991) discussed a number of sources of error: effects of preferential root ori-
entation (see below), systematic gradients of root length density within a core
making the break in the middle non-representative (this can be ignored unless
there are very steep gradients in root length density), random variation of
number of roots intersecting the plane of observation, and counting errors.
Between different people scoring core-break results, however, systematic
differences can be reduced or avoided by mutual cross-checking during a
learning period.
7.2.6 Time and Labour Requirements of the Methods
A rough estimate of the time involved is given in Table 7.1. The number of
replicates to be used, especially for the calibration of samples depends on
Table 7.1. Estimated time involved in profile wall and core-break method; estimates are based
on an experienced crew and soils which are easy to work; they serve as a first approximation
only
Person-days per Suggested number Total time per

sample of replicates per treatment in person-
treatmentl days
observation period
Profile wall
Fieldwork to make 0.25 4 7
maps of 1 x lm2
Digitizing and analysis 1.5
Washed calibration 0.15 16 2.5
samples: processing
Core break
Fieldwork 0.08 25 2
Calibration of samples: 0.25 10 2.5
processing
7 Trench Profile Techniques and Core Break Methods 219
the number of treatments or situations to be compared, as data may be

pooled for analysis. The estimates given refer to three to five treatment
comparisons. Time for processing data in the lab exceeds (as usual) the
amount of time spent in the field. With efficient organization of digitizing
and standard software for data processing, the total time required can be
reduced.
7.3 Data Analysis
Data from trenchwall observations can be analyzed in a number of ways

to derive information on spatial distribution of roots in the plane of obser-
vation, as well as to make inferences on the three-dimensional reality from
which samples were obtained. The first step in any data analysis is digitiz-
ing the root maps at an appropriate level of precision for further analysis.
The theoretical relation between point density on a map and the three-
dimensional density and dis tribut ion of cylinders has been well studied, but
empirical correlations make clear that often considerable "errors" are involved
in the observation process. After describing these, we will discuss a range of
techniques for describing root distribution with depth and horizontal distance
from the plant, and spatial correlations between roots and other mapped
features.
7.3.1 Digitizing Root Maps and Overlays
The purpose of digitizing is to derive a list of x,y coordinates of roots inter-

secting the plane of observation; this corresponds with grid counts with an
infinitely small grid size. A range of technical options exist, from digitizing
tablets where every point has to be touched with a pen to record its position,
to "scanners" which read a section of a map and can be linked with an image
analysis routine which returns the "centres of gravity" to retain a single-pixel
representation of root occurrence. Overlays can be scanned separately to digi-
tize other mapped features as lines or areas. The list of x,y coordinates can be
analyzed in various ways.
The points can be grouped into any grid size and shape for the types
of analysis discussed in section 7.3.2. Dividing the x,y coordinates by a grid
size and taking integer values classifies the roots into a grid cell; a sub-
sequent tabulation can yield point densities in a grid. Similarly, classification
in circular sections can be obtained by transforming the rect angular coor-
dinates to circle coordinates around plant rows or single trees as the point
of origin.
7.3.2 Relation Between Intersection Point Density

and Root Length Density
For a set of randomly oriented line pieces, the basic relationship between length
density per unit volume Lv and the point density of intersections with any plane
of observation N, was derived as Lv = 2N by Lang and Melhuish (1970) and
Melhuish and Lang (1968). The derivation has its origin in the 18th Centuryand
is known as Buffon's problem (Marriot 1972). The derivation considers the
probability that a randomly positioned and randomly oriented line section (of
unit length) with its centre within a specified volume of space will intersect with
a given plane of observation (any plane or a plane bordering the volume), and
then integrates over all possible positions and orientations, to obtain an ave rage
probability of being observed per unit root length. For a population of line
pieces, the expected number of intersections per unit area of observation plane
can then be related to the totallength of line. Taking the inverse of this equa-
tion, the length can be derived from the number of intersections. The fact that
root systems in a unit volume of soil are not independent little sections of line,
but tend to be interconnected may cause some doubt on the applicability of the
Lv = 2 N result, but the connections will increase the variance rather than affect-
ing the expected value and mean.
Random orientation can, however, not be taken for granted in root systems
and a modified form of the equation is Lv = 2XN, where the factor X depends on
root orientation (see Box 7.1). Ifthe N in the equation represents the average for
three mutually perpendicular planes of observation (Nm = (NX1 + N x2 + N y )/3), the
deviations of X from 1.0 are relatively small (Lang and Melhuish 1970; Baldwin
et al. 1971, 1972). In practice, however, one wants to estimate Lv from observa-
tions in one (or maximum two) plane(s) of observation.
Information on root orientation, or at least on the NxlNz ratio (e) is obvi-
ously needed to convert intersection point density measurements on profile
walls or core-break planes into estimates of Lv. Counting root intersections in
both vertical and horizontal planes may give the most direct approach. Avail-
able estimates show that this ratio will at least vary in the 0.5-4 range, and may
thus cause substantial deviations from the Lv = 2 N equation.
Theoretically, one can reconstruct root angles from observations of the
elliptical shapes of root intersections, e.g. on resin-imbedded soil thin sections.
In practice, however, slight deviations of root shapes from a cylinder shape and
errors in measurement make this approach unreliable (Van Noordwijk et al.
1992).
A number of authors (Bengough et al. 1992; Pages and Pellerin 1996;
Pellerin and Pages 1996; Jourdan and Rey 1997; Pages and Bengough 1997;
Lynch et al. 1997; Grabarnik et al. 1998) have used three-dimensional root
distribution models (ef. Chap. 4) to predict root intersections with any plane of
BOX 7.1. Effects of Preferential Root Orientation (Anisotropy)

In the following we denote the point density of intersections of roots
with three mutually perpendicular planes of observations as N x , N y and
Nz> respectively, with N", as their mean. If we can assume the two vertical
planes of observation to yield equal results, we may write N x = Ny = Â.. Nz,
and hence N m = N z (2Â.. + 1)/3. Van Noordwijk (1987) normalized the
anisotropy factor detined by Melhuish and Lang (1968), to obtain:
(l-Nx/ N m/ +(I-Ny/ N m)2+(1 - Nz / Nm )2

a =
n 6
11-Â..1
= 2Â.. + l' (7.1.1)
(this means that an = O for Nx= Ny = Nz = Nm(Â.. = 1), an = 1 if two of the

three observation planes have no roots at all, Â..= O) and aII approaches 0.5
if one of the planes has virtually no roots (Â.. -7 ro).
Marriot (1972) derived estimates of the factor X (in Lv = 2 X N m ) as
function of an. The functions for X can be adequately approximated
(Van Noordwijk 1987) by
X = O.5a,; + 1, and X = 0.8a ~ + 1, (7.1.2)
where the tirst equation applies to a 'linear' and the second to a 'planar'
si tuation with (0,0,1) and (1,1,0) roots in the extreme cases, and Â.. < 1 and
Â.. Q 1, respectively. Root anisotropy will have a relatively mild effect on
the calculated Lvvalues based on this X, provided that the average point
density N m is used. For the core-break method we are, however, mainly
interested in the LvlN: relationship and for the protile wall observations in
LvlNx, or LvlNy.In these ratios we tind much stronger effects of anisotropy,
as Â.. also influences Nm/Nz or Nn,/Nx. Van Noordwijk (1987) derived from
the above equations that for preferentially vertical root orientations with
0$Â..$1:
~ = 3Â..2 +2Â..+ 1 and ~ = 3Â..2 +2Â..+1, (7.1.3)

Nz 2Â.. + 1 Nx 2Â..2 + Â..
and for roots with a preferentially horizontal orientation and Â.. > 1:
16Â..2 + 8Â..+ 6 d 16Â..2 + 8Â..+6

an = (7.1.4)
10Â..+s Nx 1OÂ..2 +sÂ..·
One can see that for Â.. = 1 the equations return to the Lv = 2 N form;
for Â..= O we obtain Lv= N: (and LJNx is infinite), which may represent
the parallel, vertically oriented cylinders of many root uptake models.
The equations for LviNx (profile wall) show a sharp de crease in the
trajectory up to A = 0.5, stabilize around 2.0 for 0.5 < A < 2 and show a
mild decrease for A> 2 (Fig. 7.5). The equations for LviNz (core-break)
are monotonously increasing functions of A start ing at 1.0 for A = o.
14 14
____ Lv/Nx
12 _ _ Lv/Nz 12
10 10
-
Z
N
>
8
...1 6
8
-
x
Z
>
6 ...1
4 4
2 2
O O
O 0.5 1.5 2 2.5 3
Â. = Nx/Nz
Fig. 7.5. Theoretical relationship between root length density per unit volume (L,)
(cm cm-3 ) and intersection point density Non the map (cm cm-2 ) ofhorizontal and vertical
planes of observation for cases where the densities in the two possible mutually perpendi-
cular vertical planes of observation are equal
observation. This has provided a valuable tool for exploring the effect of the
type of deviations from random root orientation to be expected for real-world
root systems.
7.3.3 Empirical Correlation for Core-Break

and Profile Wall Observations
Real world calibrations of the method contain observation errors (in both
profile wall and washed samples) and may deviate from the theoretical values
which primarily depend on root orientation. Table 7.2 presents a selection of
published data on empirical calibration factors. Schuurman and Goedewaagen
(1971) gave calibration results for "root coverage" (related to N) versus root dry
weight for grassland roots.
A point of warning may be that not all these studies have corrected for
trends of root length density and comparisons may be based on intersection
density on a plane external to the volume for which Lv data are obtained rather
than for a plane halfway through this volume (Bengough et al. 1992).
Table 7.2. Examples of published values of calibration factor X in Lv= 2 X N relation between
root intersection section N and root length density Lv. (Extended from Bengough et al. 1992)
Method Crop X for X for Comments Reference

Nv Nh
Profile wall Maize/ 8 Logsdon and

soybean Allmaras (1991)
Maize 1.9 Vepraskas and Hoyt
(1988)
Tobacco 4.1 Vepraskas and Hoyt
(1988)
Mixed 0.41 Profile walls up to Oliveira -Carvalheiro
vegetation depth of9m and Nepstad (1996)
Potato 1.0-4.5 Parker et al. (1991)
Maize 3.0 Van Noordwijk et al.
(1995)
Core-break Oats 2.25 Bragg et al. (1983)
Wheat 3.3-4.4 Drew and Saker (1980)
Cotton/ 0.7-1.2 Bennie et al. (1987)
sorghum
Maize 0.8-1.1 X increases Van Noordwijk et al.
with depth (1995)
Impregnated Cotton 1.0 Melhuish and Lang
soi! blocks (1968)
Maize 1.0 Van Noordwijk et al.
(1992)
7.3.4 Root Counts as Function of Depth

and Horizontal Distance
Grid counts can be made from root maps in the laboratory by using a plastic
sheet overlay. Grid size can be adjusted, and more fiexibility is retained than
with field grid counts. Non-rectangular grids, e.g. following soil horizons, can
be used, as long as the surface area of each section is measured so that results
can be expressed as number of points per area mapped. Point densities of root
intersections can be converted to estimates of root length density Lv on the basis
of empiric al correlations or estimates of root anisotropy (Table 7.2).
Recording the x and y coordinates of each root is equivalent to a very fine
grid, where the data get a presence-absence character rather than a number of
points per grid. When x,y coordinates are recorded one can easily derive a grid
classification (by taking the integer part of the coordinates divided by grid size).
It usually is best to use grid sizes in which inter-row distances are multiples.
A.
1
. \ ....;. (.t .:\~:.... \
1
... . .. / "" :
. J.t·<·
o 1;1 • - ,
-
" , " II
"
- ' 1
" I
~;"I
r , ,1
-'
Fig. 7.6. Methods for analyzing root maps obtained in a vertical plane (A). BBy overlaying the
map with a grid, we can obtain histograms of intersection point density (number of intersections
per unit areal as a function of horizontal distance (Bl), vertical depth (B2) and distance to the
nearest plant (B3). C By classifying the map by distance to cracks (or any other mapped feature)
one can test the homogeneity of the relative point density for the strata so obtained (CI). D If two
types of roots were mapped (x and o in this example) the point density of x can be sampled by
strata differing in distance to the nearest o to test their spatial correlation (Dl)
When a specified layer of soil is washed away from the profile wall, the
number of points per grid or soil depth can also be converted into root length
per unit of soil surface are a (B6hm 1976). This can be achieved for a known
length of profile wall by assuming that each root has the same length in the
removed layer. Calibration of this technique with washed samples (blocks or
cores) is very desirable.
Various root distribution functions can be fitted to the data (see Chap. 4),
once grid counts are made. The grids are classified by their depth z below the
soil surface and radial distance r to the nearest plant and root intersection. Den-
sities are converted to root length densities on the basis of the conversion factor
X(z,r), which may depend on depth z and distance r. In soils without major
restrictions to root growth an exponential decrease of root density with increas-
ing distance from the plant may be expected. Van Noordwijk et al. (1996)
expanded on the one-dimensional form of Gerwitz and Page (1974) and
obtained:
L,(z,r ) = La Ir=oae
- a ~z2+f3r2
. (7.1)
This function has three parameters: Lalr=O is the total root length per unit area
(cmcm-2) at a distance r = O from the tree stern, a is the parameter (m-1)
governing the de crease with depth of root length density (for r = O, at a depth
of 0.699/a the root length density has half of its value at the soil surface), and
f3 is the dimensionless parameter governing the shape of the root system; values
less than 1 indicate shallow-but-wide root systems, values of 1 give a circular
symmetry, and values >1 indicate deep-but-narrow root systems.
If f3 does not differ significantly from 1 and we ignore possible variation
in X(z, r) with distance r, the equation may be simplified to a one-dimensional
distribution of root length density with depth:
Lv(z) = Laae- az •
7.3.5 Root Distribution Pattern Within a Zone:

Nearest Neighbour Distances
Root maps in the horizontal and vertical plane (trench method) not only yield
information on average root length density per soil layer, but also allow
quantification of the distribution pattern within each layer (Barley 1970; Diggle
1983). The 'null model' of root distribution, against which one would usually
want to test, is of independent roots with a homogeneous probability of occur-
rence within (a section of) the plane of observation. A Poisson distribution in
the observation grid thus gives the appropriate comparison. A number of tests
have been proposed, especially in plant ecological literature (Pielou 1969).
Several of these use the fact that in Poisson distributions the mean of the
number of points per cell equals the variance; the ration of mean and variance
in a sample can thus be compared with the confidence intervals for finite
samples. A variance/mean ratio above 1 indicates clustering, a ratio below 1
regularity (Fig. 7.7).
A different category of test is based on a comparison of "point-root" and
"root -root" distances, where "points" are chosen randomly in the plane of obser-
vation. If roots behave as mutually independent points, the two distributions
should be essentially identical. Where roots tend to cluster, root-root distances
tend to decrease while point-root distances increase. Where root patterns tend
towards regularity, the point-root and root-root curves will change in an oppo-
site direction. The difference between these two distributions thus gives a sen-
sitive test (Pielou 1969).
To derive the frequency distribution of nearest neighbour distances, two
basic approaches are:
1. Start with a "source" point of measurement, calculate the distance to alI
neighbours and select the shortest distance (nearest neighbour);
226 Mo van Noordwijk et al.
regular random clustered

o o o o 08 o
o o o o
o
o o o
o o o
o 00
o o 000 o
o o o
o o o o
00
00
o o o o
o o o o o
o o o o o
o o o
o o o
LI/----3
1/ 000: : : : ; - - - - - - - - - - - - - - - - -
Cumulative ., .........., ...............
frequency
distance to nearest reot
Fig. 7.7. Basic terminology for spatial point patterns and a nearest neighbour distance
frequency diagram which can be used to test randomness of the pattern; if the lines for random
points to the nearest root are not essentially different from those for roots-to-roots, the root
pattern may be called random itself
2. Start with alI target points and classify the are a surrounding them by dis-
tance - this is the equivalent of drawing circles of increasing diameter
around alI "target points" o
In the past, method 1 was most used, as it can be defined as a straightfor-

ward algorithm combining logical steps and a distance equation. Within this
approach there is the option of calculating the distance from each source to alI
targets and select the shortest one, or use some prior sorting of target points
which direct1y helps to select a subset of target points to be considered for a
given source point. The choice between these approaches depends on the size
of the data set (Diggle 1983). With the progress made in image analysis tech-
niques, the approach based on drawing circles around alI target points, lines or
polygons can now be easily implemented. Image analysis or GIS software may
have a built-in "distance transform" which can be used, or a distance-like oper-
ation can be based on a sequence of 'filters' using four or eight-neighbour
expansion steps (Van Noordwijk et al. 1993a,c). With the latter one can evalu-
ate the effects of 'barriers' on distance measurements. With some GIS software
one can, similarly, incorporate information on 'resistance' in distance measure-
ments (similar to the use of roads of different qualities).
If statistical tests indicate that observed patterns differ from "random", the
question is what alternative distribution should be used. Diggle (1983) discusses
the use of 'stratified Poisson' processes, which may be useful in root research.
Essentially, these distributions as sume a random set of "mother" points, with

an increased probability of finding 'daughters' in a local neighbourhood. Two
parameters of such a process are then the average number of daughters per
mother, and the distance around the mother point in which daughters are to be
found. By modifying these two parameters the spectrum from highly clustered
to truly random patterns can be generated.
Two interpretations of these parameters are: (1) that the area was a priori
heterogeneous in its attractiveness to root development, and (2) that branch
roots develop necessarily close to the axes from which they originate. The
pattern as such does not allow us to distinguish between these two reasons for
clustering.
7.3.6 Spatial Correlation of Mapped Features
Spatial correlation can be tested by implementing a stratified sampling of roots

with strata chosen with increasing distance to mapped features such as cracks
(Fig. 7.6C; Van Noordwijk et al. 1993b,c).
The following steps can be taken in the analysis:
1. Digitize root (X,Y) coordinates and map other features of interest (here
indicated as focal phenomena),
2. Determine the point density of roots coinciding with the focal phenomena,
by counting the number of roots and measuring the area,
3. "Expand" the image of the focal phenomena by including alI the area within
one unit distance of the previous map of focal phenomena,
4. Determine the point density of roots for the previous distance increment,
5. Repeat steps 3 and 4 until the whole map is covered,
6. Transfer the data to a statistics package and test for spatial correlation; the
null hypothesis of independent random events, is equivalent to the absence
of a significant trend in local point density with distance to other features.
A model of point density as a function of distance can be fitted and used to
test the null hypothesis. When fitting a distance function, one should
acknowledge that the data will have a Poisson rather than normal distribu-
tion (an example, using the Genstat statistical package is provided by Van
Noordwijk et al. 1993c)
7.3.7 Root Position Effectivity Ratio, Rper
The frequency distribution of nearest-neighbour distances of random points

to the nearest root on such a map indicates the possibilities for transport by
diffusion of alI soil resources (Van Noordwijk 1987; Rappoldt, 1990).
BOX 7 .2. Steps in the Derivation of the Root Position Effectivity Ratio,
R per (Van Noordwijk et al. 1993c):
1. Digitize a map of root distribution in a horizontal or vertical plane to
derive a list of X, Y coordinates and estimate the corresponding root
length density Lv>
2. If desired, introduce 'barriers' in the root map, e.g. representing
incomplete root-soil contact or cracks in the soil,
3. Derive a cumulative frequency distribution (Fig. 7.8A, B) of distances
from points in the soil to the nearest neighbour root, taking into
account the existence of barriers (this can be done by 'expanding' the
area taking roots as starting points and measuring surface area after
each distance increment),
4. Derive an 'annulus fraction' representation (Fig. 7.8C) of these nea rest
neighbour distances by dividing the fraction of soil in each distance
increment by that for an equivalent annulus,
5. Transfer the annulus fractions into fractions fi of complete circle
models with radius Ri ('cutting the pie'; Fig. 7.8C).
6. Calculate the total transport capacity to a sum of cylinders with
radius Ri for a given uptake model; since the zero-sink uptake of water
as well as nutrients is proportional to a G -function (De Willigen and
Van Noordwijk 1987a; ef. Chap. 15):
(7.2.1)
with:
1- 3 2 41n 2] 2R 2
G(Ri) = 0.5 [ --p-+ P 2 P and p - i - --;=== (7.2.2)
4 P - 1 - Dr - Dr.JlrL v '
where Dr is the assumed root diameter,

7. Find the R"for which G(R' ) = Gsum, and the corresponding L~ ,
8. Rper = L~/ Lv.
The Rper factor (root position effectivity ratio) was introduced (Van
Noordwijk et al. 1993a; Van Noordwijk and Brouwer 1997) to derive a correc-
tion factor, such that when Lv is multiplied with this factor an equivalent root
length density L~ is obtained for a simple model geometry (regularly spaced,
parallel cylinders), with the same opportunities for transport towards the root
surface. The method (Box 7.2; Fig. 7.8) can potentially be applied to any root
distribution pattern, but it assumes that the soil resources themselves are homo-
geneously distributed (initially). For homogeneous resource distributions, a
regular root spacing maximizes uptake, so Rper will be less than 1.0 for non-
regular root distribution. lf resources are non-homogeneous, non-regular root
distributions may be superior if roots and recources tend to coincide.
(7.3)
Van Noordwijk and Brouwer (1997) reported values of Rper in the range of
0.3 for winter wheat and sugar beet in the plough layer of arable land. Haberle
et al. (1994) explored how Rper varies with the plane of observation based on
three-dimensional root branching models (See Chap. 4).
A. B. c . . 14
0---'
t:
o ,
13 ,
,
~
«1
~
,"
r .... ........
<t: .. ""1"
:~~,'I,,'"''
Distance to
nea rest root
D.
O @
11 + 12 + 13 +
Fig.7.8. Analysis of root distribution pattern. By applying a 'distance transform' to the map on
the basis of ali roots mapped; measurement of the nearest neighbour distances (B) can be used
to derive the area in each distance c\ass for an 'average' root (e); by 'cutting the pie' we can derive
the relative frequency of cylindrical models with 1 . .. x layers which together may represent
the possibilities for uptake by the root pattern (D); for a given root uptake function, we can
finally derive the equivalent homogeneous root density which would allow the same uptake
to occur and express the ratio of this hypothetical and the measured point density as the root
position effectivity ratio, Rp" (Van Noordwijk et al. 1993a,c)
A major assumption in this approach is that a classification of soil accord-

ing to the nearest root (Le. a Dirichlet tessellation) is acceptable. If alI roots start
to function at the same time, this assumption leads to only small errors (De
Willigen and Van Noordwijk 1987b). In the model representation, the cumula-
tive frequency distribution of transport distances should be conserved. For any
frequency distribution of distances to a spatial point pattern, a frequency
distribution of cylinders can be derived which fulfills this criterion (Van
Noordwijk 1987; Rappoldt 1990; Van Noordwijk et al. 1993a).
7.3.8 Transition from 2-0 to 3-0 Oistance
A partly unresolved problem in interpreting root maps is the fact that the
real three-dimensional distances are smaller (except for roots growing perpen-
dicular to the mapped plane) than the two-dimensional distances measured on
the map. For randomly orientated roots, the frequency distribution of three-
dimensional distances coincides with that for 0.71 times the two-dimensional
distances (Van Noordwijk 1987). For non-random root orientation the problem
is unresolved; a serious complication is the fact that a plane represents a biased
sample of roots, as roots growing almost perpendicular to the plane are
over-represented and roots growing almost parallel to the plane are under-
represented. For a 2-dimensional map of the Z-plane, the frequency distribu-
tion of point-root distances in case of a random distribution of roots can be
derived from a Poisson distribution (Pielou 1969; Marriott 1972):
P[d < D2 ] = l_e- trN,Ii';, (7.4)
where D 2 is the two-dimensional distance and d is the distance of a point on the

map to the nearest root.
For three-dimensional distances of points to randomly oriented and spaced
cylinders (lines with density Lv per unit volume and radius Ro) Ogston (1958)
and Barley (1970) derived that:
(7.5)
where D3 is three-dimensional distance, and A is the number of root tips per

unit root length.
The term Lv R~ refers to the volume of the cylinders themselves and is
normally negligible. The last term in the equation (413 A D~) refers to a half
sphere around the root tip for 'end-contacts' and has a considerable modifying
effect on the curve, which is mainly determined by the tangential contacts
of the middle term. For A = O the 3-D equation becomes comparable to the
2-D one if:
(7.6)
This result strict1y depends on random orientation of the roots with

regard to the plane in which two-dimensional distances were measured. If D2
is measured in a plane perpendicular to a parallel set of cylinders, D3 will
equal D2•
7.4 Conclusions and Perspectives
The profile wall methods discussed in this chapter for observing roots span a
wide range of applications: from "quick and not-too-dirty" impressions of
overall spatial patterns, which can be much faster than methods based on
washed samples, to methods aimed at extracting specific spatial information
which cannot be obtained from washed samples at alI.
The geometrical relations between three-dimensional root systems and
observations on any two-dimensional plane of observation are far from trivial
and further progress is to be expected from links between root architectural
models, and predicted and observed point patterns on a range of planes of
observation. For several root ecological questions analysis of "spatial correla-
tion" of roots and other phenomena (roots and cracks and/or anaerobic spots,
roots of species A and those of species B, roots and concentrations of organic
inputs, roots and decaying root channels formed by a previous vegetation) is
crucial, but refined observation techniques may be needed and the standard
"root map" may be too coarse.
References
Baldwin JB, Tinker PB, Marriott FHC (1971) The measurement of length and distribution of
onion roots in the field and the laboratory. J Appl Ecol 8: 543-554
Baldwin JP, Tinker PB, Nye PH (1972) Uptake of solutes by multiple root systems from soil. II.
The theoretical effects of rooting density and pattern on uptake of nutrients form soil.
Barley KP (1970) The configuration of the root system in relation to nutrient uptake. Adv Agron
22: 159-201
Bengough AG, Mackenzie C], Diggle AJ (1992) Relations between root length densities and
root intersections with horizontal and vertical planes using root growth modelling in
3-dimensions. Plant Soil145: 245-252
Bennie ATP, Taylor HM, Georgen PG (1987) An assessment of the core-break method for
estimating rooting density of different crops in the field. Soil Tillage Res 9: 347-353
Bland WL (1989) Estimating root length density by the core-break method. Soil Sci Soc Am J
53: 1595-1597
232 M. van Noordwijk et a!.
Bland WL (1991) Root length density from core-break observations: sources of error. In:
McMichaei BL, Persson H (eds) Plant roots and their environment. Eisevier, Amsterdam,
pp 565-569
Bohm W (1976) In situ estimation of root length at natural soi! profiles. J Agric Sci 87: 365-368
Bohm W (1979) Methods of studying root systems. Berlin Heidelberg New York
Bragg PL, Govi G, Cannell RQ (1983) A comparison of methods, including angled and vertical
minirhizotrons, for studying root growth and distribution in a spring oat crop. Plant Soil
73:435-440
De Willigen P, Van Noordwijk M (1987a) Roots for plant production and nutrient use efficiency.
Doct Thesis, Agricultural University, Wageningen, 282 pp
De Willigen P, Van Noordwijk M (1987b) Uptake potential of non-regularly distributed roots.
J Plant Nutr 10: 1273-1280
Diggle PJ (1983) Statistical analysis of spatial point patterns. Academic Press, London
Drew MC, Saker LR (1980) Assessment of a rapid method, using soi! cores, for estimat ing the
amount and distribution of crop roots in the field. Plant Soi! 55: 297-305
J App!. Ecolll: 75-96
Grabarnik P, Pages L, Bengough AG (1998) Geometric properties of simulated maize root
systems: consequences for length density and intersection density. Plant Soi! 200: 157-167
Haberle J, Van Noordwijk M, Brouwer G (1994) Comparison of root position effectivity ratio of
field-grown and modelled maize root systems. Poster abstract. In: Proc 3rd Congr Eur Soc
for Agronomy, 1994, Abano-Padova
Jourdan C, Rey H (1997) Modelling and simulation of the architecture and development of the
oii palm (Elaeis guineensis Jacq.) root system. II. Estimation of root parameters using the
RACINES postprocessor. Plant Soil 190: 235-246
Kooistra MI, Van Noordwijk M (1996) Soi! architecture and distribution of organic carbon. In:
Carter MR, Stewart BA (eds) Structure and organic carbon storage in agricultural soils.
Advances in Soil Science. CRC Lewis, Boca Raton pp 15-57
Lang ARG, Melhuish FM (1970) Lengths and diameters of plant roots in non-random popu-
lations by analysis of plane surfaces. Biometrics 26: 421-431 (corrections in Biometrics 28:
626-627)
Logsdon SD, Allmaras RR (1991) Maize and soybean root clustering as indicated by root
mapping. Plant Soil131: 169-176
Lynch JP, Nielsen KN, Davis RD, Jablokow AG (1997) SIMROOT: modelling and visualization of
root systems. Plant Soil 188: 139-151
Marriot FHC (1972) Buffon's problem for non-random distribution. Biometrics 28: 621-624
Melhuish FM. Lang ARG (1968) Quantitative studies of roots in soi!. I. Length and diameters of
cotton roots in a clay-Ioam soi! by analysis of surface-ground blocks of resin-impregnated
soi!. Soil Sci 106: 16-22
Ogston AG (1958) The spaces in a uniform random suspension of fibres. Trans Faraday Soc 54:
1754-1757
Oliveira Carvalheiro K de, Nepstad DC (1996) Deep soil heterogeneity and fine root distri-
bution in forests and pastures of eastern Amazonia. Plant Soi1182: 279-285
Pages L, Bengough AG (1997) Modelling mini-rhizotron observations to test experimental
procedures. Plant Soi1189: 81-89
Pages L, Pellerin S (1996) Study of differences between vertical root maps observed in a maize
crop and simulated maps obtained a model for the three-dimensional architecture of the
root system. Plant Soil 182: 329-337
Parker CJ, Carr MKV, Jarvis NJ, Puplampu BO, Lee VH (1991) An evaluation of the minirhizotron
technique for estimat ing root distribution in potatoes. J Agric Sci 116: 341-350
Pellerin S, Pages L (1996) Evaluation in field conditions of a three-dimensional architectural
model of the maize root system: comparison of simulated and observed horizontal root
maps. Plant Soi1178: lOl-112
Pielou EC (1969) An introduction to mathematical ecology. Wiley-Interscience, New York
Rappoldt C (1990) The application of diffusion models to aggregated soi!. Soi! Sci 150: 645-661
Ringrose-Voase AJ (1996) Measurement of soil macropore geometry by image analysis of
sections through impregnated soi!. Plant Soi1183: 27-47
Schuurman JJ, Goedewaagen MAJ (1971) Methods for the examination of root systems and
roots, 2nd edn. Pudoc, Wageningen
Schuurman JJ, Knot L (1957) Het schatten van hoeveelheden wortels in voor wortelonderzoek
genomen monsters. [Estimating root quantities in auger samples; in Dutch] Versl
Landbouwkd Onderz 63: 31
Tardieu F (1988) Analysis of the spatial variability of maize root density. 1. Effect of wheel
compaction on the spatial arrangement of roots. II. Distances between roots. Plant Soi! 107:
259-266 and 267-272
Tardieu F, Manichon H (1986) Caracterisation en tant que capteur d'eau de l'enracinemnent du
mais en parcelle cultivee. 1. Discussion des criteres d'etude. II. Une methode d'etude de la
repartition verticale et horizontale des racines. Agronomie 6: 345-354 and 415-425
Van Noordwijk M (1987) Methods for quantification of root distribution pattern and root
dynamics in the field. 20th Colloq Int Potash Institute, Bern, pp 263-281
Van Noordwijk M, Brouwer G (1997) Roots as sinks and sources of carbon and nutrients in
agricultural systems. In: Brussaard L, Ferrera-Cerrato R (eds) Soil ecology in sustainable
agricultural systems. CRC Lewis, Boca Raton, pp 71-89
Van Noordwijk M, Kooistra MI, Boone FR, Veen BW, Schoonderbeek D (1992) Root-soil contact
of maize, as measured by thin-section technique. 1. Validity of the method. Plant Soi! 139:
109-118
Van Noordwijk M, Brouwer G, Harmanny K (1993a) Concepts and methods for studying
interactions of roots and soi! structure. Geoderma 56: 351-375
Van Noordwijk M, De Ruiter PC, Zwart KB, Bloem J, Moore JC, Van Faassen HG, Burgers S
(1993b) Synlocation of biological activity, roots, cracks and recent organic inputs in a sugar
beet field. Geoderma 56: 265-276
Van Noordwijk M, Brouwer G, Zandt P, Meijboom FW, Burgers S (1993c) Root patterns in space
and time: procedures and programs for quantification. IB-Nota 268, IB-DLO Haren (the
Netherlands)
Van Noordwijk M, Driel W, Brouwer G, Schuurman W (1995) Heavy metal uptake by crops from
harbour sludge covered by non-contaminated topsoi!. II. Cd uptake by maize in reiat ion to
root development and distribution of metals. Plant Soi1175: 105-1l3
Van Noordwijk M, Lawson G, Groot JJR, Hairiah K (1996) Root distribution in relation to nutri-
ents and competition. In: Ong CK, Huxley PA, eds. Tree-crop interactions - a physiological
approach. CAB International, Wallingford. pp 319-364
Vepraskas MJ, Hoyt GD (1988) Comparison of the trench profile and core methods for evaluat-
ing root distributions in tillage studies. Agron J 80: 166-172
CHAPTER8
Root Observations and Measurements

at (Transparent) Interfaces with Soil
A.L. Smie, E. George 2, and J. Groenwold 1
1 Plant Research International (Wageningen UR), Postbus 16, 6700AA Wageningen,
The Netherlands
2 Hohenheim University, Institute for Plant Nutrition, Fruwirthstr. 20, 70593 Stuttgart,
Germany
CONTENTS
8.2 Minirhizotrons 236
8.2.1 Introduction 236
8.2.2 Types of Minirhizotrons 238
8.2.2.1 Solid Tubes 238
Tube Material 238
Dimensions 238
Marking System/Depth Registration 239
8.2.2.2 Flexible or Pressurised Tubes 239
8.2.3 Installation Technique 241
8.2.3.1 Installation Equipment 241
8.2.3.2 Installation 243
Angle of Installation 243
Spatial Distribution in Relation to Rows 244
8.2.3.3 Light Leaks and Temperature Effects 244
8.2.4 Observation Equipment 245
8.2.4.1 Scopes 245
8.2.4.2 Video Cameras 245
The Use of Ultra Violet Light in Root Observation 245
8.2.4.3 Telescopes, Mirrors and Still Cameras 246
An Inexpensive System 246
8.2.5 Observations with Minirhizotrons 247
8.2.5.1 Introduction 247
8.2.5.2 Estimation of Bulk Soil Parameters 248
Root Distribution in the Profile 248
Rooting Depth 248
Root Length Density and Numbers of Roots 249
Conclusions 251
8.2.5.3 Estimation of Root Morphology and Root Dynamics 252
A.L. Smit el al. (Eds.), Rool Melhods

236 A.L. Smit et al.
Root Properties 252

Assessment of Root Directionality 252
Rooting Dynamics 254
8.3 Root Windows 256
8.3.1 Introduction 257
8.3.2 Materials 257
8.3.3 Installation 258
8.3.4 Placement 261
8.3.5 Observation Equipment 262
8.3.6 Applications 263
8.3.6.1 Root Morphology, Growth, and Root Turnover 263
8.3.6.2 Soil and Rhizosphere Measurements 264
8.4 Root Screens 265
Further Reading 266
References 266
8.1 Introduction
Glass windows are often installed in rhizotron facilities to observe root growth.
The same principle has been applied under field conditions, using "root
windows" and "minirhizotrons" (transparent glass or plastic tubes pushed
into the soil: Bates 1937), through which roots are observed non-destructively
in situ at the interface between the glass and the soil. The possible observations
include recording root growth and decay - processes which are difficult
to measure otherwise - by stiH photography or videos, thereby providing
time series that reveal the dynamics of root growth. Root researchers also
use transparent interfaces when estimating the rooting characteristics of
the surrounding bulk soil, such as root length intensity (Lv) and root mass
per unit volume of soil (Wv ), although this often necessitates a (cumbersome)
conversion from 2-D to 3-D.All these methods and tools used to view roots are
reviewed in this chapter.
8.2 Minirhizotrons
8.2.1 Introduction
Inserting clear tubes into the soil through which roots can be observed in situ
with various items of equipment (see Fig. 8.1), has several advantages over other
methods of root quantification.
8 Root Observations and Measurements at (Transparent) Interfaces with Soil 237
Fig.8.1. A common mini-

rhizotron set-up with
observation equipment.
(Photograph courtesy
J. Bartz)
One such advantage is that studying root dynamics at the interface between
a transparent viewing medium and the soil is usually less labour-intensive than
core sampling, in-core growth or profile wall methods. Another is that individ-
ual roots (and their properties) can be followed in time. Root characteristics
such as the growth rate, branching, longevity and the presence of root hairs can
best be assessed through an in situ viewing medium. Moreover, this technique
makes it possible to quantify infection with mycorrhizae, fungi or nematode
species and to assess the diameter distribution of the root system in situ (Majdi
and Nylund 1996).
Minirhizotrons allow the same are a/volume of soil and the same population
of roots to be monitored continuously, but questions have been rai sed about how
representative these observations are compared with bulk soil methods.
8.2.2 Types of Minirhizotrons
8.2.2.1 Solid Tubes
Tube Material. Tubes made of glass (borosilicate, trade name e.g. Pyrex),
acrylic (Perspex, Plexiglas), polybutyrate and polycarbonate (Lexan) have
been used as minirhizotrons. Glass tubes are cheaper than commercially avail-
able polybutyrate tubes, but may break if large roots and rocks are encountered
in the profile, or if the soil is subject to severe freeze-thaw and shrink-swell
movements.
McMichael et al. (1992) mention that polybutyrate is sometimes preferable
to acrylic tubes because of its flexibility and resistance to excessive scratching
during installation. Kloeppel and Gower (1995) described the construction of
tubes from inexpensive acrylic tubing (according to them this is 42% cheaper
than glass), marking them with an easily constructed etcher and installing the
tubes with a tree seedling auger.
Although Taylor and Bohm (1976) suggested that soil adhered more readily
to glass windows than to acrylic plastic windows, they did not specifically
compare the two materials. They also suggested that if the soil touching the
plastic dries, it will shrink at higher soil water contents than if it is touching
glass. They concluded that the resulting cracks and voids encouraged higher
root densities at the plastic face compared with the bulk soil.
It has also been suggested, although not quantified (Voorhees 1976), that
using acrylic material might allow electric charges to develop, which would
interfere with root growth. Although many researchers have voiced concern
about the material used to create an interface to observe root growth, the mate-
rials have never been explicitly compared. This suggests that the material itself
is not a crucial factor for the minirhizotron method. In many cases it is the soil
and environment that determine the choice of material.
Dimensions. Upchurch (1987) argued that larger tube diameters increase the
sample size, which improves precision and reduces variability. Mackie-Dawson
and Atkinson (1991) mentioned that a 5.6cm o.d. (outer diameter) tube would
"sample" a volume of 352cm3, assuming a view 2-mm deep. If a strip of 2cm
only is observed at the upper side of the minirhizotron, only about 40 cm3 would
be "sampled", a small volume compared with the volume of soil sampled by soil
coring. Coring to a depth of 1 m yields samples over 2000cm3 of soil, depend-
ing on the diameter of the corer. But note that when Itoh (1985), cited by Brown
and Upchurch (l987), evaluated tube sizes from 1-3 cm, he found that accuracy
increased with decreasing diameter.
The preferred diameter for round tubes is about 5 cm, mainly because this
will accommodate commercially available video-camera systems. Square tubes
have also been used (Vos and Groenwold 1987; van Noordwijk et al. 1994) but
are not a standard stock item and do not allow the viewing equipment to be
rotated. A further disadvantage is that they require a custom-made soil probe
for creating the minirhizotron access hole.
It has been mentioned that it is more difficult to uniformly illuminate in
round tubes than square tubes (Brown and Upchurch 1987) but modern video
equipment seems to have overcome this problem.
Marking System/Depth Registration. The observation system must be posi-

tioned accurately along the tube; this is particularly imporant if individual roots
have to be monitored over time. Ferguson and Smucker (1989) described an
indexing handle for controlling and identifying the camera position
when making repeated video recordings from minirhizotrons in field experi-
ments. One problem with an indexing handle is that it is probably not accu-
rate enough to enable the images to be semi-automatically proces sed by
image analysis, as horizontal or vertical displacement can still occur (pers.
observ.). This can be overcome by putting a registration mark on the tube, so
that the frames can be properly aligned with image analysis software. The
spring-Ioaded etching trough, described by Kloeppel and Gower (1995) which
etches accurate, equidistant cirdes along acrylic tubes in minutes, would be
useful for this.
8.2.2.2 Flexible or Pressurised Tubes
One factor crucial to the success of the minirhizotron method is the contact
with the surrounding soil. Imperfect contact prevents an accurate estimate of
the bulk soil root density. If there are voids, the roots will grow into them or
will track along the surface of the tube. One of the signs of poor contact between
tube and soil is condensation on the tube surface and a tendency for roots to
produce more root hairs when growing in the space between the tube and soil.
In soils with layers which hinder root penetration, a tunnelling effect can be
created, with these gaps promoting deeper rooting along the tube than in undis-
turbed soils (Vos and Groenwold 1987).
An alternative minirhizotron design has been developed, with improved
contact with the surrounding soil. The pressurised wall minirhizotron devel-
oped by Merrill et al. (1987) and Merrill (1992) consists of (see Fig. 8.2):
1. A central tube of dear, extruded acrylic plastic,
2. A O.05-cm-thick cylindrical piece of dear polyvinyl sheeting fitted over this
central tube and damped to both its ends, and
3. A PVC collar at the upper end of the tube, through which air pressure tubing
is introduced.
Fig. 8.2. A pressurised wall

minirhizotron system (after
Merrill 1992 (left)) and a
solid minirhizotron (right).
Courtesy S. Asseng, CSRIO
These minirhizotrons of 7.5 cm o.d. were installed in holes of 8.9 cm excavated

with a rotary auger and finished with a bucket auger to achieve a final diame-
ter of 9.6 cm. A system of low-pressure air pumps ensures that there is constant
air pressure (10-20kPa) between the sheeting (outer wall) and the tub ing (inner
wall). An advantage of the pressurised wall system is the relative ease with which
the uninftated minirhizotron tubes can be inserted into their access holes.
Gijsman et al. (1991) proposed another method for establishing close
contact between the observation interface and the surrounding soil. Based on
the same idea, they used an inftatable ftexible wall, made from a modified
motorcycle tyre. The method involved inserting a square aluminum or stain-
less steel frame consisting of ribs 1 cm wide into a hole in the ground. An
inftatable rubber tube, made from a cut motorcycle tyre and stretched by a
metal rod, was introduced into the frame and pressurised. The tyre bulges
between the ribs and presses against the soil. Before the interface is observed
the tyre is deftated and carefully removed from the frame, in order not to
change the positions of the roots and to ensure that roots continue to adhere
to the soil.
This design has two additional advantages over Merrill's system:
1. Improved visibility of the roots because it is not limited by the opaqueness
of the minirhizotron wall, nor do the tubes generate disturbing reftections,
and
2. Roots can be sampled/collected in situ with appropriate devices, e.g. to deter-
mine N content, presence of VA mycorrhizae, and physiological functioning
(dead/living).
These aspects underline the potential usefulness of the system in research.
Gijsman et al. (1991) checked for the presence of voids by examining
stereoscopic pairs of photographs. They concluded that no condensation
or smears from alga growth obscured the visibility of the roots and that all
roots were growing in one plane of the soH/tube interface. They advised taking
special care at the points where the frame ribs meet the soH, because voids will
occur here if the frame is too thick or the pressure too low. The transition can
be made more gradual by bevelling the ribs of the frame to prevent gaps from
forming.
Volkmar (1993) compared an adjusted pressurised tyre system with rigid
minirhizotrons (transparent polybutyrate plastic tubes (5.4cm o.d.). From his
observations of root growth at horizontal minirhizotrons at bulk densities
increasing from 1.5 to 1.7 g cm-3 he concluded that significant differences in root
counts between the two systems became apparent with increasing density.
When he examined the relationship between root counts (minirhizotron) and
root density of the bulk soH, he found that with rigid tubes, root counts were
higher at the higher bulk densities. With the flexible tubes, the relationship
between root counts and root density remained more or less stable with dif-
ferent bulk densities. This implies that rigid tubes probably overestimated
rooting, especially at higher bulk densities. Volkmar (1993), however, also noted
that inflatable tubes have a drawback compared with rigid tubes: the pressure
is difficult to maintain and even under controlled environment conditions the
tubes gradually Iose pressure or deflate suddenly if the rubber wall develops a
weakness. For this reason, minirhizotrons have to be checked daHy after instal-
lation. In the pressurised system described by Merrill (1992) the pressure is
maintained by a solar-driven air pressure system. Both these sophisticated mod-
ifications increase the cost of the apparatus and its maintenance, often beyond
research budgets.
8.2.3 Installation Technique
8.2.3.1 Instal/ation Equipment
The technique for installing tubes (glass and plastic) in an undisturbed soH
profile in the field is critical. The main concerns are to establish good contact
between the tube and the surrounding soH, to avoid compaction of the soil
during the installation of the tube in the soH and to minimise the creation of
voids at the tube/soil interface.
Hand augers can be used in conjunction with an angled support (see Fig.
8.3A), to make a hole slightly smaller than the outside diameter of the minirhi-
zotrons. Box et al. (1989) used hydraulic soH coring machines for this purpose
(see Fig. 8.3B).
They took soH cores of 5.5 cm o.d. prior to installing a tube of 5.7 cm
o.d. The cutting points of the soH sampling tubes were machined, so that
soH displaced by the sharpened tip was compressed toward the interior of the
sampling tube, thus preventing compaction at the interface. The angle of the soH
Fig. 8.3. a Installation of

minirhizotrons with hand
auger and an angled
support. (Courtesy
S. Asseng, CSRIO). b
Mechanical hydraulic coring
machine in use at Michigan
State University. (Courtesy
A. Smucker)
probe was maintained by using a large wooden guide constructed at a specific

angle. Afterwards the hole was brushed with a 5.5 cm fine circular steel brush.
The tubes were then forced into the soil manually. Similar procedures have been
used by others (Gregory 1979; Vos and Groenwold 1987; Majdi et al. 1992).
Hummel et al. (1989) studied the effect ofthe installation procedure on the
nature of the contact zone and investigated the importance to the success of the
procedure of stabilising the cor ing unit, guiding the augering and cor ing devices
and of coring bit size. Following field installation in a silty loam they found no
voids, but coring and tube installation did cause soil compression which was
uniform around the tube at a given depth and varied between depths. The com-
pression was minimised when diameter of the cor ing bit was O.16cm less than
the minirhizotron diameter.
In general, data from horizontal minirhizotrons in containers with uni-
formly repacked soil (which allows minirhizotrons to be installed without
voids) best agreed with data from auger sampling (Vos and Groenwold 1983,
1987; Smit et al. 1994; Meyer and Barrs 1985). This underlines the importance
of avoiding voids next to the tubes.
8.2.3.2 Instal/ation
Angle of Installation. A wide range of angles to the vertical has been used
when inserting the tubes. OriginalIy, researchers inserted the tubes verticalIy,
but later various angles were used. Minirhizotrons have even been positioned
horizontalIy (Smit et al. 1994; Dubach and Russelle 1995). Most studies,
however, have used a range of angles between 30° and 45° to the vertical. Bragg
et al. (1983) claimed that installing the tubes vertically would induce preferen-
tial tracking of the roots alongside the tube and that this could be eliminated
with an angled insertion. They reached this conclusion after comparing 45° and
vertical tubes with soil coring. When the data from the top 30 cm of soH were
omitted, angled tubes showed a better relation with coring than vertical tubes.
However, it has been pointed out (De Ruijter et al. 1996), that this conclusion is
not supported by the data shown in the original publication. OveralI, consider-
ing alI depths, vertical tubes showed even better correlation with coring than
angled tubes.
In a pot experiment with Sitka spruce and sycamore, root tracking was
found to occur to the same extent on vertical tubes and on tubes angled at 45°
(Mackie-Dawson et al. 1989). The root counts were higher at vertical tubes, espe-
cialIy in the first 7 cm, but this was attributed not to tracking but to a more hor-
izontal root growth in this layer. Evidence was also found with vertical tubes
that there was greater variance associated with the measurement of root length
(Mackie-Dawson and Atkinson 1991).
Many investigators have installed their tubes at an angle to the vertical,
usualIy 45°, although vertical positioning is still used (Campbell et al. 1994;
De Ruijter et al. 1996), particularly when conditions are unfavourable for
angled tubes. Horgan et al. (1993) acknowledge the importance of the ins talla-
tion angle, especially if root growth orientation deviates from isotropic
(isotropic = alI orientations are equalIy likely) and if the observations of the 2-
D-surface are to be converted to a 3-D environment. If isotropy cannot be
assumed, the "point process" will depend on the orientation of the surface on
which the observations are made. Horgan et al. (1993) derived an optimal angle
of 54° theoreticalIy. At this angle the influence of a non-random orientation of
root growth would be minimal (bias should not exceed 3% whatever the
distribution of root angles). They also suggested two further strategies to
minimise the influence of root anisotropy: one was to give the horizontal
count per unit root length twice the weight of the equivalent vertical count if
horizontal and vertical tubes can be used, as then bias will be close to zero.
Another strategy would be to have groups of three tubes of equallength; two
positioned perpendicular to each other in horizontal planes and the third posi-
tioned vertically. Both strategies alIow examination of variation in root density
bydepth.
Spatial Distribution in Relation to Rows. It is suggested that in row crops, tubes

should be inserted parallel to the row (Upchurch and Ritchie 1983, 1984;
McMichael and Taylor 1987) to avoid a bias with depth. To ascertain the lateral
distribution of roots, more tubes should be installed at various distances
between the rows (Brown 1984 cited by McMichael and Taylor 1987). It is very
difficult to assess the average rooting characteristics of individual plants
(including forest trees): in such cases, minirhizotrons may be inserted at a con-
stant distance from the plant.
8.2.3.3 Light Leaks and Temperature Effects
Light entering a minirhizotron through a gap in the soilltube interface can

travel down the minirhizotron tube to substantial depths and influence roots
approaching the tube wall. The standard treatment to exclude light in minirhi-
zotron studies is to wrap black tape around the above-ground section of the
tube and cap the end with a painted aluminium can which slides snugly over
the tube. Levan et al. (1987) investigated the efficacy of various additional mea-
sures to prevent light leakage: gaskets made of rubber matting (O.16cm thick
rubber, 26 x 16cm) pulled over the minirhizotron flush with the soil surface
and extending about 4 cm laterally around the minirhizotron. The outer 2 cm of
the rubber shield was punctured with pinholes to improve water infiltration.
Finally, a piece of O.15cm black plastic was wrapped over the top of the tube,
covering the junction between the tube and the cap and just reaching the soil
line. The standard treatment (described above) resulted in serious light leaks in
25% of the minirhizotrons. The minirhizotrons with the extra light eXclusion
treatment were found to have fewer roots in the upper 10 cm compared with the
standard treatment. The difference decreased with depth in the profile.
Levan et al. studied the effect of light on root growth in controlled
environment rooms and reported that at unshielded windows the root numbers
were 54% of those at shielded windows. The inhibitory effect of light on root
growth is obvious and is therefore also given as one of the reasons why minirhi-
zotrons usually exhibit fewer roots in the upper soillayers compared with soil
coring. B6hm (1979, Sect. 7.3.4) gives more references on the negative effects of
light on root growth but concludes that the short time that roots are exposed
to the light during recording has no strong influence on the results.
To exclude light and solar warming, Majdi et al. (1992) painted the 15cm
of tube which remained above the soil black, and then white. Others (Smit et
al. 1994) have put insulation foam inside the tube, to equalise the temperature
at the minirhizotron interface.
8.2.4 Observation Equipment
8.2.4.1 Scopes
The scopes developed to inspect interior parts of instruments and human

bodies include endoscopes, borescopes and duodenoscopes. They consist of an
objective lens, a light source, a pathway (flexible or rigid) from the lens to the
eyepiece (e.g. optic fibre) and an eyepiece. The scopes are available in various
lengths, and the objective lens can have various magnifications and fields of
view (Brown and Upchurch 1987).
8.2.4.2 Video Cameras
In contrast to scopes, modern video camera systems are specifically designed

(e.g. by Bartz Technology Corporation, Santa Barbara CA, USA) to be used in
minirhizotrons in order to view root development and the activity of organisms
and soil conditions non-destructively. Many improvements have been made in
resolution, lighting, the indexing handle and the depth registration since the
early developments described by e.g. Upchurch and Ritchie (1984).
The latest commercially available developments include:
1. A camera which can change the field of view from 18mm horizontal to
3.5 mm horizontal, while observing a position on the minirhizotron,
2. A camera system in which a computer moves the camera up and down the
tube. The computer can be programmed to repeat runs at any interval. This
is useful for the "zoom camera" because the length of the interval between
index locations varies, depending on whether the top of the picture must
match the bottom edge of the previous picture, and
3. A digital camera enabling the video pictures to be stored in digital format.
This makes it easier to arrange individual observations (frames) in a
time series (consecutive images of one position on a minirhizotron) and
dispenses with the time spent editing and taping (ef. Dubach and Russelle
1995).
The Use of Ultra Violet Light in Root Observation. It has been reported that the
roots of many plant species fluoresce (emit light in the blue wavelengths) under
UV light, and that there is a relationship between root functionality or age and
the intensity of UV fluorescence (Dyer and Brown 1983). Although some
modern minirhizotron camera systems are equipped optionally with a UV
lighting system, few results obtained with this observation method have been
reported. UV-transmission through the minirhizotron material used should be
checked.
Wang et al. (1995) investigated whether UV lighting was better than visible
light in estimating the proportion of living roots. Their results implied that
visible light should be chosen in grass-dominated communities and UV light
in communities dominated by broad-Ieaved herbs or woody plants. Under UV
light some dead roots apparently stiH fluoresced and were easily misidentified
as living.
Smit and Zuin (1996) studied two plant species: Brussels sprouts (fast
growing, many roots) and leeks (slow growing, low rooting intensity). Leek
roots fluoresced more than Brussels sprouts roots. With increasing age, the UV
fluorescence decreased in Brussels sprouts roots but increased over time in leek
roots. The authors therefore concluded that UV fluorescence could not be used
as a universal indicator of root age or root functionality. However, they sug-
gested that the technique might be used to identify the roots from a species in
mixed cropping experiments.
8.2.4.3 Telescopes, Mirrors and 5till Cameras
An Inexpensive System. Recently, a system for photographing within 8 X 8 cm

square minirhizotrons was described by Poelman et al. (1996). The system
is based on a telescopic lens instead of an endoscope. The main advantage of
this system is that its components are cheap, are usually in stock and can
be repaired easily. Additional advantages are that it is portable, shockproof and
can be operated by a single person. The equipment consists of a long tubular
frame with a mirror angled at 45° at one end and a camera with a telescopic
lens at the other end (see Fig. 8.4). Lighting within the system is achieved with
an electronic flash for taking photographs and a halogen light for focusing. The
strength of this system is its simplicity, as no high-tech parts are involved. The
length of observation tubes is limited to approximately 140cm when using a
600 mm telelens.
Using a camera as described above, Van Noordwijk et al. (1985) obtained a
3-D view of the roots at the minirhizotron surface, using stereoscopic pairs of
photographs of a 6-cm-diameter circle on a square 8 x 8cm Lexan minirhi-
zotron wall, taken every 2.5 cm. They concluded that in most of the photographs
the gap between the tube and the soil was considerable (several times the diam-
eter of a root). Most of the roots grow directly along the Lexan tube, forming
abundant root hairs. A stereoscopic view of the roots can add important infor-
mation about the stage of decay of roots, especially when information about
root growth dynamics is wanted, and makes it easier to assess the functional-
ity of a root.
8 Root Observations and Measurements at (Transparent) Interfaces with Soi! 247
----A
("~
10cm
Fig.8.4. Basic set-up of inexpensive observat ion equipment for minirhizotrons. A camera with
telelens, B minirhizotron wall, C flash, D halogen light, E mirror, Fviewing aperture. See Poelman
et al. (I996) for a detailed description. (Fig. courtesy Dr. Johan van de Koppel)
8.2.5 Observations with Minirhizotrons
8.2.5.1 Introduction
Researchers are interested in responses of roots in the bulk soil. However, only
the roots in contact with the minirhizotron surface can be observed with
minirhizotrons. Therefore, the root parameters measured at this interface
should be directly related to the desired bulk soil root parameter, and the inftu-
ence of the interface on the measured parameter should be minimised
(Upchurch 1987). So, the minirhizotron surface must represent a transect
through the soil rather than a root collection or inhibition zone (Mackie-
Dawson and Atkinson 1991). For this reason it is important to be clear about
what you want to use the minirhizotron method for:
1. To estimate a bulk soil parameter like volumetric root length density, root
distribution etc., or
2. To observe a root characteristic or property such as longevity, diameter,
branching, which is visible at the minirhizotron surface.
In both cases, but particularly in the first case, where a translation from 2-D to
3-D is necessary, the effect of the interface is important.
8.2.5.2 Estimation of Sulk Soi/ Parameters
Root Distribution in the Profile. Minirhizotrons have been widely used to

estimate the rooting distribution in the surrounding soil. However, there have
been numerous reports of discrepancies when comparing minirhizotron data
with soi! coring (Bragg et al. 1983; Hansson and Andren 1987; McMichael and
Taylor 1987; Hansson et al. 1992; Andren et al. 1993; Heeraman and Juma 1993;
Wiesler and Horst 1994). In general, fewer roots are found in the upper soi!
layers when the root distribution in the soi! profile is assessed with minirhi-
zotrons, but the rooting intensity in deeper layers is overestimated. Researchers
have resorted to excluding data from the uppermost soi! layers in order to
achieve a positive correlation between minirhizotron data and soi! coring
(Heeraman and Juma 1993). Various explanations have been given for this dis-
crepancy in root data:
1. Light leakage effects (Levan et al. 1987),
2. Voids, which allow roots to track along minirhizotrons and encourage deeper
rooting (Upchurch and Ritchie 1983; Bragg et al. 1984; Heeraman and Juma
1993),
3. Disrupted hydrology in the soi! above an angled minirhizotron (loss of
contact with the subsoi!; Vos and Groenwold 1987),
4. Loss of structure of the soi! caused by the installation of minirhizotrons
(Hansson et al. 1992), and
5. Temperature differences (McMichael and Taylor 1987).
The interference of voids with rooting observed with minirhizotrons seems to
be the most likely explanation for the discrepancy with coring at deeper layers.
A possible explanation for the discrepancy in shallow layers might be the more
horizontal direction of root growth at those depths, where roots would be less
likely to encounter a minirhizotron. If roots do grow predominant1y horizon-
tally, it seems Iikely that fewer roots wiIl be seen when observing the upper side
of a minirhizotron, especially if the minirhizotron has been installed horizon-
tally (Smit et al. 1994). Beyrouty et al. (1987) suggest root growth orientation as
a possible explanation for observing less root growth in the upper 9 cm with
rice, but they used vertical tubes, which would in fact be preferable for observ-
ing horizontal root growth.
Rooting Depth. Several definitions of "rooting depth" exist in the Iiterature:

1. The first root to appear at a certain depth (Bland 1993). However, in the field
this criterion is subject to large variation,
2. When a critical level of root number/length per cm3 or per cm2 (minirhi-
zotrons) has been exceeded, or
3. The depth at which 95% of the roots (length or mass) are present.
Smit et al. (1996) used data from horizontal minirhizotrons and applied an iter-
ative statistical regres sion procedure to calculate rooting depth, assuming that
rooting increases linearly with (thermal) time over a certain period.
Andren et al. (1993) used the following formula to calculate the mean
rooting depth, although a better description of this calculation might be the
average depth of a root:
(8.1)
in which Zm is depth in cm (average depth of a root), Zi is depth in cm, and Ni

is the root number per cm 2 at depth Zi.
Root Length Density and Numbers of Roots. The following observations can be
made at the tube interface. However, to estimate values of the surrounding soil
conversion procedures are required:
1. Length of roots (cmcm-2 minirhizotron surface, La). The length of the roots
on the minirhizotron surface can be calculated by counting the
number of intersections on a grid for each observation. The grid may be
inscribed on the tube or projected on a recorded image. Using the modified
Newman-line-intersect method (Tennant 1976), the number of roots inter-
secting the grid is then converted to length and can be expressed per cm 2
surface.
Some researchers (e.g. Heeraman and Juma 1993) have subsequent1y
assumed a certain "depth of view" of 0.2 or 0.3 cm and calculated the volu-
metric root length density (Lv) direct1y. Although simple, the assumption
about the depth of view seems rather arbitrary, as good contact between the
soil and tube would imply a depth of view of Ocm!
When assessing this parameter (length) it should be considered whether
the length of roots on the minirhizotron surface is influenced by the minirhi-
zotron itself, as roots may grow slower or faster than in the bulk soil. If
this is so, root length is not the most appropriate parameter to assess. It has
been suggested that the number of roots arriving at the surface (Upchurch
1987) or first intersections as described by Mackie-Dawson and Atkinson
(1991) would show a better relationship with the rooting intensity in the bulk
soil.
2. Number of roots on the minirhizotron surface (N ra ). This parameter is prob-
ably less influenced by the conditions at the interface than length and is inde-
pendent of any property expressed by the root after that root has touched
the minirhizotron. The effect of the interface on results obtained is therefore
probably minimal. Upchurch and Ritchie (1983) proposed the following pro-
cedure. Given a certain minirhizotron area, all roots that intercept the tube
are counted. Each individual root is assigned a count ofl; branched roots
receive a score of 1 for the primary root and an additional score for each
branch. The hypothesis is that the density of the roots in the surrounding
soil determines the number of arrivals of roots at the interface. This is
in contrast to the length of the root at the interface, which is a function of
both the arrival of the root and growth of the root after it has touched the
minirhizotron.
The minirhizotron root counts can be converted to Lv using either an
empirical calibration obtained from the regression of root numbers on
length of roots washed out of soil samples, or a theoretical model. Several
theoretical models have been developed to predict the relationship between
the number of roots on the minirhizotron surface and the root length density
in the surrounding soil,
(8.2)
The theories, with different assumptions, have resulted in a wide range of
values for c (Table 8.1).
Bland and Dugas (1988) compared these c-values in a study with cotton
and sorghum and concluded that c = 2 provided the best estimate of Lv for
cotton and the data were sufficient to reject the c = 1 model (empirically, c
was found to be 2.8). However, no simple relationship between minirhizotron
data and measured root length was found for sorghum.
Using horizontal minirhizotrons in the Wageningen Rhizolab, Smit et al.
(1994) found that most crops largely followed the c = 2 relationship, when
minirhizotron counts had been calibrated with core sampling.
Studies with wheat have yielded c-values of 2.8 (Bragg et al. 1983),3.5
(Meyer and Barrs 1985) and as large as 20 (Belford and Henderson 1985).
Values of c larger than 15 were also found by Murphy et al. (1994) for
creeping bentgrass and annual bluegrass turfs; they concluded that a
single value of c probably does not apply throughout the growing season.
Parker et al. (1991) found likewise for potatoes, as did Smit et al. (1994)
when working in the Wageningen Rhizolab with horizontal minirhizotrons
where no voids were present. However, the Wageningen researchers found a
stable relationship in time for other crops (leeks, wheat, Brussels sprouts).
There are reports of the reverse effect, however: Samson and Sinclair
Table 8.1. Range of values for c with relevant authors
Value of c Reference
Upchurch and Ritchie (1983)

2 Melhuish and Lang (1968)
3.3 Upchurch (1985)
3.0-3.8 Merrill and Upchurch (1994)
B EJ
Fig. 8.5. Points of contact
on minirhizotron frames, O 3
according to Buckland et al.
(1993)
E;J 1 1
[SJ 3 O
(1994) and Wiesier and Horst (1994) found a shift over time for maize and
potatoes.
3. Points of contact. Instead of counting alI root segments (with or without
branches), Buckland et al. (1993) proposed counting onIy the first and last
points of contact with the minirhizotron (see Fig. 8.5). A root that intercepts
a tube, grows away from it, then intercepts it again is considered to have two
points of contact, so that it is not necessary to determine whether one root
intercepts the tube more than once in this method. A tracking root "passing
by" is not counted at alI.
Conversion to root Iength density is done under the assumptions that:
(1) root growth orientation is random, and (2) tubes do not affect the growth
of the root except by diverting.
Volumetric root Iength density is estimated by:
Lv = N/(p Te Tr Ti) = n/(p Te Tr ), (8.3)
in which TI is the total Iength of tubes in the sampIe, Tr is the radius of each
tube, N is the total count of points of contact of roots with tubes in the sample
(comprising first and Iast points of contact and root tips), and p is the pro-
portion of the tube surface that is counted. n = NITl gives the average number
of counts per unit Iength of tube.
According to Buckland et al. the point of contact method agreed well
with measured root Iength densities in cores for wild cherry roots, but not
for roots of pasture species.
Conclusions. Although minirhizotrons are perhaps not the most appropriate

tools for estimating root Iength density in the buIk soil, severai researchers have
tried to evaluate the success of minirhizotrons by comparing the results with

soil coring methods. In general, soil coring is considered as more representa-
tive for measuring the distribution and intensity of rooting (Hansson et al.
1992). Heeraman and Juma (1993) concluded that destructive sampling is still
the best method to quantify root growth in the top 10cm of soil. Underestima-
tion of rooting in shallow layers and overestimation at deeper layers is a par-
ticular problem when studying the effects of bulk density using minirhizotron
methods. De Ruijter et al. (1996) concluded that soil compaction had a major
effect on root length in the bulk soil, which was not reflected in the root number
observed using minirhizotrons. The most appropriate parameter seems to be
to assess the number of root segments or points of contact.
At the moment there is no universal theory to predict the relationship
between Lv and N ra . Different models point to different numbers for the con-
version coefficient c when converting root number to the root length density of
the bulk soil. However, questions remain about the stability of the relationship
with time. This means that the method has to be recalibrated for each crop, and
for different times of the year, which negates the benefits of using minirhi-
zotrons instead of soil coring.
8.2.5.3 Estimation of Root Morphology and Root Dynamics
Root Properties. The resolution of a typical modern video camera system

(VCR) is probably too low to obtain a precise figure for root diameter and its
variation. After transfer from VCR to PC, a root of 0.01 cm diameter would com-
prise only 4 pixels, which considerably reduces the accuracy of measuring the
diameter with image analysis programs. Better results can be expected using
the newest camera systems with zooming facilities (field of view reduced from
18 to 3.Smm). According to the manufacturer (Bartz Technology Company,
USA) the resolution of this camera would allow detailed examination of objects
of 3 Jlm, e.g. infection with mycorrhizal fungi, nematodes and root hairs. Note
that when observing these root properties the same prerequisites exist as for
root length: conditions at the interface should not differ from the conditions in
the bulk soil.
Although branching characteristics can easily be assessed with minirhi-
zotrons, we know of no published results. However, because roots cannot be
traced into the soil, their order of branching cannot be ascertained.
Meisner (1991) observed and quantified root hair characteristics for
groundnut under water stress.
Assessment of Root Directionality. Upchurch et al. (1988) developed a mathe-

matical model, based on probabilistic assumptions about root growth direc-
tions, to characterise the dominant root system orientation. By determining the

ratio of the number of roots that intercept the top of an angled minirhizotron
to the number that intercept the bottom, the model predicts whether the root
system is oriented more horizontally or more vertically. Later, Merrill et al.
(1994) extended the theory to a situation where upward root growth occurs.
After summing the root observations on the upper and lower sides of the
minirhizotron per treatment, the ratio between the two sums then gives the root
directionality ratios (RD). Merrill et al. gave the theoretical RD ratio for four
hypothetical probability distributions of root orientation at minirhizotron
installation angles of 5° to 60° from the vertical (See Table 8.2).
Using vertical minirhizotrons and an endoscope De Ruijter et al. (1996)
counted the number of roots intersecting a horizontalline on the tubes every
2 cm. The roots intersecting a vertical line on the minirhizotrons were counted
at eight compass points. The average orientation of the roots, in degrees from
the vertical was calculated from the ratio between horizontal and vertical
counts. If horizontal and verticallines are not making a square grid, the hori-
zontal or vertical counts have to be corrected for unequal increments (the
authors used a 6 cm o.d. minirhizotron; in this case the horizontal counts were
divided by 0.3751l').
After correction, the ave rage root angle (a) is then:
vertical counts ) (8.4)

a= arctan( .
horizontal counts
This method gives the exact average orientation of roots when growing direc-
tions are distributed symmetrically around one angle, otherwise a small error
is made.
Table 8.2. Root directionality ratios (upper side/lower side) for four hypothetical probability
distributions of root orientation at minirhizotron installation angles from 50 to 600
Minirhizotron Patterns of root directionality distribution

angle
Random ~ Increasing downward ~ More vertical
50 1.04 1.07 1.12 1.22

15 1.13 1.23 1.40 1.79
25 1.22 1.40 1.77 2.66
35 1.31 1.65 2.60 4.07
45 1.39 1.73 3.00 6.51
55 1.46 1.88 4.14 11.34
60 1.49 1.94 5.00 16.76
Rooting Dynamics. Root observations at transparent interfaces enable root

dynamics to be studied, because individual roots can be followed through time.
However, the main problem is that there are no objective criteria for deciding
whether a root is dead or alive. This decis ion has to be made visually, using mor-
phological characteristics of the root; very often this will be the colour of the
root (a change from white to grey or black roots). Unfortunately, the change in
colour varies with crop and soil conditions, so comparisons are difficult to
make. Smit and Zuin (1996) observed that the physical decay of a root is often
preceded by a permanent shrinking of the root and so used this as an extra indi-
cation that the root had ceased functioning.
Keeping track of individual roots is laborious, but so far automatic pro-
cessing of images with image analysis software has not been entirely
satisfactory because roots cannot yet be fully distinguished from the back-
ground automatically or the software cannot recognise the physiological con-
dition of the root. Not surprising, even experienced staff have difficulty in
processing observations. However, image analysis software can be used to make
it easier to assess root dynamics such as mortality and longevity by eye (See
also Chap. 10). One such aid is the MSU ROOTS software developed by Enslin
et al. (1994). This is a video-digitising program with advanced capabilities for
cap turing, displaying, and analysing minirhizotron images of plant root systems
and creating a database for the study of plant root dynamics. After cap turing
an image, an operator can digitise the oudine of individual roots or entities. The
program can be used to identify and measure the length of individual roots
within cohorts at particular sites, while allowing for comparisons between same
site identifications and measurements over time.
Visual comparisons can be made between data digitised earlier and exist-
ing root images. Furthermore, a root identification number, physiological
condition code and branch order number can be assigned. The MSU ROOTS
program allows the user to perform additional analytical operations using data-
base management systems like FoxPro and DBaseIV.
Smit and Zuin (1996) used the NIH-Image program on a Macintosh com-
puter to assess root dynamics in Brussels sprouts and leeks. This is a public
domain image analysis program written by Wayne Rasband at the US National
Institute of Health and available from the Internet. It has also been used to
cap ture the images from a VCR, to make time series, to align images and to
measure length, grey value etc. from root segments in time. Also available for
Macintosh computers and specifically developed for processing minirhizotron
images is RooTracker (Duke University, USA).
Two approaches that can be used to quantify root dynamics in terms of root
longevity are:
BOX 8.1. Calculations on Root Dynamics

If the fate of individual roots is followed it is possible to graph total roots
produced in time (expressed in length, numbers or mass) and roots
which have decayed versus time. Smit and Zuin (1996) regressed the root
data on time with an ordinary logistic model of the form:
c
y"~ and Yd~ =a + 1+e(-b( ' -m» ,
where y denotes the total amount of new roots (y",.) and dead roots (yd,,)
at time t. The sum a + c is the total amount to be reached ultimately, b is
a shape factor, and the value of m is the date at which 50% of the roots
are produced.
The standing root intensity, the amount of roots available at time t
can be ca1culated as:
y", = y .." - Yd,,·

This approach is applicable for annual crops, but for perennial crops and
for forestry where some roots are already present at the start of the
observation period, a variation has been proposed (Van Noordwijk
1987), i.e.:
The total amount of roots (YI,I) present at time t, is:
y", =y .." - YI,I'

in which YI" is the amount of roots present at the beginning of the
observation period.
In this case the standing living root intensity is:
y", = y", - Yd,,·

At the end of the season at time = e, data are obtained for y ..,e> y" OYd,e>
and y",. From these data two ratios can be defined which may summarise
the root dynamics throughout the growing season:
T =Ydl!/ y ,p
where T is the turnover of root length/number during a growing season
or a set time period (- )
and
where R is the root replacement ratio during a growing season or a set

time period (- ).
1. To follow cohorts of roots in time, as explained in detail in Chapter 11, see

also Hendrick and Pregitzer (l992a,b, 1993) and Cheng et al. (l990, 1991),
and
2. To use the cumulative "production" curves of new and dead roots (length,
number), e.g. Rygiewicz et al. (l997), see Box 8.1. See Box 8.2 for some con-
siderations of subsampling minirhizotrons.
BOX 8.2. How orwhen to Subsample with Minirhizotrons

Minirhizotrons are particularly promising tools for assessing root
turnover, but the large number of individual observations suggest that
much labour is required. This raises the question of how or when to sub-
sample. Dubach and Russelle (1995) described that with their system (a
VCR and horizontal minirhizotrons) each replicate of depth (a total of 80
images in one horizontal tube) required 29h oflabour. But a large part of
the processing time (greater than two thirds) is needed for editing tapes
(making a time-series from the individual tapes from different observa-
tion dates) . In the near future the time needed for this editing will proba-
bly be considerably reduced, thanks to the availability of robust,
read-write optica! disks that can record images digitally, together with an
identification code as the file name. Dubach and Russelle (1995) also con-
cluded further that accuracy and precis ion of results were not decreased
by using only every second image along both sides of the tube (40
images), but that precision declined when either only one side of the tube
or only 20 consecutive images on both sides were used. Recording less fre-
quently reduced labour requirements even more: recording intervals could
be lengthened to 2 weeks for shallow depths and even longer for greater
depths without sacrificing accuracy or precision. These changes led to a
70% saving in the labour required for estimating root turnover.
8.3 Root Windows
Root windows are transparent viewing planes installed in soils, allowing the
non-destructive monitoring of root growth and detailed studies of individual
roots and of rhizosphere soil. In contrast to minirhizotrons (see Sect. 8.2), the
viewing area is usually much larger and not curved. In addition to viewing
roots, roots and soil can be assessed, so that small-scale manipulations and sam-
plings are possible in addition to continuous measurements. Root windows are
much cheaper to install than rhizotron facilities, and allow roots and soils to be
observed in situ. They have to be installed with great care, to avoid water or
light penetration into the window-soil interface.
8.3.1 Introduction
The idea of viewing roots growing in field or forest soil behind a transparent
panel is quite old. In most early experiments, glass plates were installed against
vertical soil profiles. Horizontal glass plates were also used to observe root
growth in forest topsoil (McDougallI916). Root windows allowed the study of
in situ root growth periodicity in perennial plant species, for example in fruit
trees (Head 1966), forest trees (McDougallI916), shrubs and bushes (Fordham
1972; Fernandez and Caldwell 1975), in grassland plant communities (Speidel
and Weiss 1974; Hayes and Seastedt 1987) or in crop plants (Pearson and Lund
1968). Root windows allow a continuous and long-term observation of roots
and soil with minimum disturbance. Non-destructive techniques can be com-
bined with coring techniques to increase their validity by helping to assess
the best time to core. The seasonal pattern of root growth and the periods of
minimal and maximal root growth can be determined in root windows (Vogt
et al. 1986).
Non-destructive techniques are also important when dynamic changes
of roots in response to the environment are to be studied (McMichael et al. 1992).
Because their field of observation is larger than in minirhizotrons, root windows
are especially suitable for studying roots in particular soil microsites, interac-
tions with soil micro-organisms (Egli and Kiilin 1991), and soil fauna (Wilson et
al. 1995). They enable results obtained under controlled conditions by using soil-
filled pots or boxes with glass or perspex front plates for root growth studies (for
example, LeNoble et al. 1996) to be verified under field conditions.
At present, root windows are not widely used. However, the need to under-
stand in situ root-soil interactions and to study effects of soil micro-organisms
on root activity rather than "average" root biomass or root length at a site will
foster interest in this inexpensive technique. Root growth along the transpar-
ent panel is, of course, not directly comparable to root growth in bulk soil,
because three-dimensional root exploration is reduced to two-dimensional
growth. The situation at the panel may be similar to, for example, the surface
of a smooth flintstone (Kolesnikov 1971; Bohm 1979).
8.3.2 Materials
The same materials used for minirhizotrons (see Sect. 8.2.2.1) or permanent
rhizotrons may be used for root windows, but Plexiglas (Hiiussling et al. 1985,
1991) or glass (Rutherford and Curran 1981; Egli and Kălin 1991) are most
common. The disadvantages of the individual materials have not been tested
thoroughly. Assumptions that plastic windows induce irregular conditions by
electric charges (Voorhees 1976; Rutherford and Curran 1981) have not been
substantiated (Reid et al. 1992), but close contact between soil and the plastic
plate may be more difficult to obtain with Plexiglas than with glass plates (Taylor
and Bohm 1976; Huck and Taylor 1980). The contact between soil and plate must
be close for meaningful observations at the window.
Root windows can vary in area according to the specific needs of the inves-
tigations. Individual windows for observational purposes are usually as large
as 1-3m2 (Pearson and Lund 1968; Fordham 1972; Teskey and Hinckley 1981;
Tscherning et al. 1995). "Observation pits" (Kolesnikov 1971) or root windows
in a size range of 0.7-0.8m2 are more suitable for replicated observations and
for the study of treatment effects (Egli and Kălin 1991; Hăussling et al. 1991).
Smaller windows are often in better contact with the soil than larger planes
(Rutherford and Curran 1981) and are easier to replicate.
8.3.3 Installation
This description follows the method developed by Hăussling et al. (1985,1991).

A similar method is described by Egli and Kălin (1991). For earlier experiences
with larger root windows see, for example, Kolesnikov (1971), Bohm (1979), and
Rutherford and Curran (1981).
A 50 cm x 40 cm x 0.6 cm Plexiglas plate is put into a laboratory oven. Two
sides of the plate are placed on rods and a heavy weight is laid longitudinally
in the centre of the plate. The oven is heated to approximately 60°C and until
the plate has become slightly concave. It retains its shape after being removed
from the oven and allowed to cool. The root window is then installed by first
hammering a straight sharp-edged steel plate (approximately 54cm wide) into
the soil. The steel plate is rammed into the soil at a specific angle (for example,
45 or 60° from the vertical). To obtain a smooth soil profile, the steel plate must
be heavy and hammered into soil without deflections. Root windows can also
be placed nearly vertical (Bohm 1979) or even with a negative slope (Huck and
Taylor 1980). The preferable angle of installation depends on the plant species
under investigation, the soil type, and the techniques used for observing root
growth at the window.
After the steel plate has been installed, a pit of approximately 55 cm x 55
cm is then dug in front of the steel plate to the depth of installation (see
Fig. 8.6). The excavated soil must be removed from the site, to minimise local
disturbances of soil chemistry. The soil directly behind the window may not
be walked on, as otherwise the physical conditions for root growth may be
Fig. 8.6. A steel plate has been rammed into

the soi! and a pit has been dug in front of it
altered. The steel plate is then carefully removed from the soil profile. No
top soH, not even a small amount, may be transferred down the profile during
the installation. If necessary, the soil profile may be smoothed by gently rubbing
it horizontally from side to side with dry paper. Any small voids in the profile
(for example, from small stones or cutting of woody roots during installation)
can be filled with soil from a corresponding soH depth. Large voids, however,
cannot usually be filled adequately, and if there are any, the window must be
relocated.
The concave-shaped plate is tightly placed with its convex side snugly
against the soil profile. If required, a sheet of clear hard plastic can be inserted
between the soil profile and the plate, to minimise root damage when the plate
is later removed for destructive measurements. To the left and right of the
plates, wooden sticks are hammered deep into soil, perpendicular to the plate,
so that wedges can be pressed between them and the Plexiglas plate (Fig. 8.7).
With the help of the wedges, the sides of the plate are pressed tightly against
the soH profile (Fig. 8.8). Because of the concave shape, the Plexiglas is also in
close contact with the soil in the centre of the plate. In some soils, it is neces-
sary to press down the wedges further several months or years after installa-
tion, in order to maintain the close contact between the Plexiglas plate and the
soil profile. For example, soil movement can be appreciable in soils with mont-
morillonitic clay (Huck and Taylor 1980).
So il surface
Cove r
P lexig las plate
Wedge
Wooden stick (horizo nta l

position)
ou..._..-- Wooden stick (vertica l position)
Fig. 8.7. Schematic drawing of a root window installed into soi!
Fig. 8.8. The root window

(Plexiglas plate) is pressed
tightly against the soil
profile
Alternatively, jacks can be used to press the plate against the soil profile
(Schuurman and Goedewaagen 1971). However, this is more expensive and the
jacks reduce the free viewing area of the window. Elaborate systems with spring-
loaded windows (Rutherford and Curran 1981) or support frames (Kolesnikov
1971) also bring about a tight soil/plate contact, but require a relatively high
technical investment. Large nails or pins are often used to fix the plates directly
onto the soil profile, but this may lead to variable soil contact (Bohm 1979) and
be appropriate for only a relatively short duration, some soil types, and in
certain moisture conditions.
After installation, the soil profile and the pit must be insulated against
light and against temperature fluctuations different from the bulk soil (Fig. 8.8).
This can be done by covering the pit with a block of commercially available
insulation material (e.g. Styrodur). The block must completely cover the pit
to prevent light entering but should not be excessively large. Other materials
used for insulation include thick spun glass stapled to plywood covers (Keyes
and Grier 1981), expanded polyurethane (Rutherford and Curran 1981),
expanded polystyrene (Reid et al. 1992), grass mulch (Fordham 1972), Styro-
foam boards (Tscherning et al. 1995), Styrofoam covered with burlap (Hayes
and Seastedt 1987) or Styrofoam chips and black plastic foiI (Egli and Kălin
1991).
New roots will appear on the window surface soon after installation,
depending on vegetation type and time of year. A high number of new roots
may originate from old roots cut during installation (for instance when
windows are installed into an established forest are a or in grassland; Speidel
and Weiss 1974). In any case, some time should be allowed after installation
before measurements begin. When roots start to appear at the window, it can
be seen whether they have grown in soil and are developing their own rhizos-
phere, or have grown in an air space between soil and the window plate. The
window should not be used if the latter situation applies to many roots, because
air spaces not only provide room for uninhibited root growth but also allow soil
solution to flow freely from the topsoil to deeper soillayers.
In an alternative approach, the plate is fixed a few millimetres from the
original profile, and sieved material from the appropriate soil depths is used
to reconstruct a soil profile directly behind the window surface (Fordham
1972; Hayes and Seastedt 1987; Reid et al. 1992). Though perhaps providing
better control over experimental conditions (Huck and Taylor 1980) this
may, of course, lead to unrepresentative root growth in the reconstructed
profile (Durrant et al. 1973; Rutherford and Curran 1981). Furthermore, the
spatially distinct arrangement of, for example, soil nutrients is lost in this
case, as is the original physical condition of the soil. For these reasons, the
use of reconstructed soil should be avoided (Zillmann 1956), except when
it is impossible to maintain good plate/soil contact with the original soil
profile.
8.3.4 Placement
In many cases, root windows have been used to determine root growth dynam-
ics of plant species in relation to their shoot growth cycles. Observations from
a few or even a single root window were considered adequate for such basic
informat ion. A large number of replicated windows are necessary for more
detailed studies or for comparisons of sites or plots. Four root windows per
plot were used by Keyes and Grier (1981), whereas Hăussling et al. (1991) used
ten windows per plot in a Norway spruce forest. Hahn and Marschner (1998)
installed 16 windows per plot. Window-to-window variation is greater for some

parameters (e.g. total root length) than for others (e.g. weekly root length incre-
ment). The number of root windows required per plot therefore depends on the
parameter measured, but is probably not much less than in comparable studies
with minirhizotrons.
The location of the windows is an important consideration in row crops or
in studies with large perennial plants. In forests, the windows are usualIy placed
at a certain distance from and orientation to the stern of a tree (Egli and Kălin
1991; Hăussling et al. 1991). Such non-random placement of windows reduces
the data variation, but results cannot be extrapolated to the total plot area.
Windows placed in different positions with respect to plant rows or individual
plants enable the spatial variation of, for example, root growth dynamics or
mycorrhizal colonisation of roots to be studied.
8.3.5 Observation Equipment
Almost alI measurements that can be taken in permanent rhizotrons (Huck and
Taylor 1980) can also be carried out on site at root windows. For example, soil
moisture and soil temperature can be measured at different depths and at
different distances behind the root window (Fordham 1972; Fernandez and
CaldwelIl975; Rutherford and Curran 1981; Teskey and Hinckley 1981). Sensors
can easily be installed into the soil through small holes drilled into the plastic
windows. These measurements have usualIy shown that temperature and
moisture conditions behind the plate do not differ distinct1y from those in
the bulk soil (Kolesnikov 1971; Fernandez and Caldwell 1975; Bohm 1979) if
the window is well insulated. Additional measurements could include soil
O2 and CO 2 concentrations and soil pH (Taylor et al. 1990; Reid et al. 1992).
Novel methodology to sample and analyse very small volumes of soil solution
(Gottlein et al. 1996) can be applied to root windows to study ionic concentra-
tions in the rhizosphere and bulk soil solution under natural conditions.
Destructive samplings or measurements can be carried out in smaller addi-
tional windows (Rutherford and Curran 1981) or after briefly removing the
plate from the soil profile (Hăussling et al. 1991). It is possible to remove the
plate if the windows are properly constructed, but this interferes with mea-
surements of root growth dynamics because some roots may be affected or even
displaced during the removal.
Root length along the plate is usually traced with pens onto acetate sheets
placed on the window plane. Different coloured pens can be used to mark new
root growth in different time periods (Fordham 1972; Belgrand et al. 1987; Pages
and Serra 1994; Wilson et al. 1995), different root size classes (Neufeld et al.
1989), brown and white roots (Reid et al. 1992) or roots of different plant species
(Tscherning et al. 1995). During the measurements, exposure of roots to light

should be minimised. The line lengths on the acetate sheets can later be deter-
mined with a grid method, a scanner (Tscherning et al. 1995), or with a digital
image analysis system. These techniques can also be used to determine root
branching pattern and topology (Belgrand et al. 1987; Colin-Belgrand et al.
1989; Pages and Serra 1994). A direct scanning system of roots at the transpar-
ent interface has been described by Pan et al. (1998).
Depending on plant species and soil type, not alI fine roots may be dis-
cernible by eye. Microscopic assessment of these roots may be necessary
(Rutherford and Curran 1981). Microscopic observations are also helpful or
even essential when studying the development of ectomycorrhizal root tips
(Egli and Kălin 1991). A hand-Iens may be sufficient to study meso- and
macroinvertebrate fauna (Wilson et al. 1995). Repeated photography (Wilson et
al. 1995) or video imaging of window sections at different size scales can
document temporal changes in root and soil properties, similar to in minirhi-
zotron tubes.
8.3.6 Applications
8.3.6.1 Root Morphology, Growth, and Root Turnover
Root length and the number of roots can be measured at the window plane.
This results in comparative data sets of, for example, cultivars or species with
different drought resistance (Fernandez and Caldwell 1975) or plants in plots
with different fertiliser levels (Speidel and Weiss 1974), or plants in competition
experiments. This may often be sufficient to test the original hypothesis of an
experiment (Wilson et al. 1995). As with minirhizotron data, measurements
can be converted into values for bulk soil root density and root numbers or
even root biomass (see Sect. 8.2.5.2). These conversions, however, are based on
assumptions or empirical calibrations and are usually only of some predictive
value (Glinski et al. 1993). No interface method can substitute for destructive
soil sampling (see Chap. 6) in the accurate assessment of mean root length den-
sities in the bulk soil. Root windows are better suited than other field methods
to determine root branching patterns (Fernandez and Caldwell 1975; Sword
1998). The measurements are similar to those in rhizoboxes in laboratory
studies (Glinski et al. 1993; Pages and Serra 1994; LeNoble et al. 1996), but have
the additional value of being carried out on site.
The main advantage of the non-destructive techniques (minirhizotrons and
root windows) is, of course, that dynamic changes of root appearance and dis-
appearance, root length and root numbers can be observed. The time between
observations is variable, but 2-week intervals are often used (Speidel and Weiss
1974; Fernandez and Caldwelll975; Keyes and Grier 1981; Haussling et al. 1991;
Wilson et al. 1995). The number of roots at the plane may be a more appropri-
ate parameter than root length to measure rooting density (see Sect. 8.2.5.2).
However, both parameters may be important for dynamic measurements. For
example, in a white oak forest Teskey and Hinckley (1981) observed different
peaks during the season for the number of roots and for root elongation. With
the help of root windows it can be shown that root elongation in temperate
forests is highly positively correlated with soil temperature, except in dry soil
(McDougallI916), but the production of new root tips is correlated with shoot
growth events and peaks in spring and autumn. For measurement of root
distribution in the soil profile, the number of points in contact with the plate
should be assessed (Fernandez and Caldwelll975), because roots are deflected
after their first contact with the plate.
Observations on the appearance and disappearance of new roots can also
be used to estimate root turnover (Keyes and Grier 1981; Wilson et al. 1995). It
has to be kept in mind, however, that roots at the interface for a relatively long
period after installation are comparatively young and not representative of the
total root system including older roots (Hayes and Seastedt 1987). Rates of root
longevity measured at the interface cannot therefore be used uncritically to
calculate turnover of the total root biomass. However, root longevity and root
turnover can be assessed with root windows individually for different root types
and in relation to measurements of the soil environment.
8.3.6.2 Soil and Rhizosphere Measurements
Soil conditions can be tested non-destructively by inserting the appropriate

sensors through the Plexiglas plate. The large spatial heterogeneity over a few
centimetres in, for example, soil water potential and N0 3- concentrat ion or pH
in the soil solution can be shown convincingly (Gottlein et al. 1996). Studies
carried out conventionally in rhizoboxes or rhizotrons could also be set up with
root windows: for example, the study of effects of soil moisture gradients on
root growth (Zillmann 1956; Durrant et al. 1973) and mortality (Kosola and
Eissenstat 1994). The consequences of high concentrations of mineral elements,
pesticides or organochemicals in certain soil zones can also be observed
directly (Haussling et al. 1985, 1991; Blaker and MacDonald 1986; Wilson et al.
1995). Processes such as nodulation or pest attacks (Klepper and Kaspar 1994)
can be observed on site.
A further advantage of root windows in forests is that the seasonal pattern
of mycorrhiza formation and the life span of individual mycorrhizal types can
be determined (Egli and Kalin 1991). Small-scale natural disturbances (soil
mixing by animals, root severance, local changes of pH) may create additional
habitats for mycorrhizal fungi and thus contribute to a high fungal diversity
(Bruns 1995). Such hypotheses can be tested with root windows.
It is also possible to manipulate soil moisture and nutrient supply. For
example, Larigauderie and Richards (1994) injected small amounts of nutrients
into small soil areas. Subsequent1y, they observed the effect of the local nutri-
ent enrichment on root growth of several arid-Iand grass species (Larigauderie
and Richards 1994).
When the Plexiglas plate is taken off, filter papers or agar sheets can be
placed direct1y on the roots. This allows for the quantification of, for example,
soil pH changes in the rhizosphere (Hăussling et al. 1985, 1991). Other possible
measurements include the determination of phosphatase activity, manganese
oxidation or release of phenolics in the rhizosphere (Dinkelaker and Marschner
1992; Dinkelaker et al. 1993).
8.4 Root Screens
Root screens are nylon mesh planes inserted into soil. Some time after
installation, the number of roots crossing the mesh is evaluated. Root
screens are a new and inexpensive technique for studying rooting intensity in
field sites.
To install a root screen, a sharp blade is used to cut a slit into soil, usually
at an angle of 45°. Fibreglass screens are then fitted into the slit at different
soil depths, the blade removed from soil and the soil faces pres sed firmly against
the screen. The screens are 5 to 10 cm wide and have a mesh size of 1 mm
or more (Fahey and Hughes 1994; Parsons et al. 1994). The sites of screen
installation must be marked for later retrieval. In forest soHs, screens can
also be placed horizontally to investigate roots in the soil surface layer. These
screens allow root growth to be observed non-destructively (Fahey and Hughes
1994).
To remove the screens, soH must be lifted from the upward face of the
screen. The number of roots penetrating the screen is counted at the site or
later in the laboratory. Live and dead roots may be distinguished by visual cri-
teria. After repeated samplings, newly produced root length and even biomass
can be calculated from these counts (Fahey and Hughes 1994). As with other
interface methods, these calculations require the use of several conversion
factors.
Similar to the ingrowth core method (see Chap. 6), the root system is dis-
turbed dur ing installation in plant stands. This may affect root productivity. In
contrast to ingrowth cores, no foreign or sieved material needs to be used as a
root growth substrate in the root screen method. However, it may be difficult
to use root screens in dry soH or at high soil bulk density (Parsons et al. 1994).
The main advantage of root screens is that they are relatively quickly installed
and evaluated. This allows observations with a high number of replications at
different times of the growing season or in plots with different treatments.
The effect of small-scale fertilisation near the screen on root growth can also
be studied (Fahey and Hughes 1994). So far, root screens have been used in
temperate forests only, but may well be applicable to root studies in agro-
ecosystems.
Further Reading
Minirhizotron observation tubes: methods and applications for measuring rhizosphere

dynamics. ASA Spec Publ 50: 99-108
Plant Soi! 185: 225-258. Selected papers of the minirhizotron workshop, Sweden 17-20 Sept,
1995
References
Andren O, Hansson AC, Vegh K (1993) Barley nutrient uptake, root growth and depth distribu-
tion in two soi! types in a rhizotron with vertical and horizontal minirhizotrons. Swed J
Agric Res 23 (3): 115-126
Bates GH (1937) A device for the observation of root growth in the soil. Nature (Lond) 139:
966-967
Belford RK, Henderson FKG (1985) Measurement of the growth of wheat roots using a TV
camera system in the field. NATO Adv Study Inst Ser Ser A Life Sci 86: 99-105
Belgrand M, Dreyer E, Joannes H, Velter C, Scuiller I (1987) A semi-automated data processing
system for root growth analysis: application to a growing oak seedling. Tree Physiol 3:
393-404
Beyrouty CA, Wells BR, Norman RJ, Marvel JN, Pillow JA Jr (1987) Characterization of rice roots
using a minirhizotron technique. Minirhizotron observation tubes: Methods and applica-
tions for measuring rhizosphere dynamics. ASA Spec Publ 50: 99-108
Blaker NS, MacDonald JD (1986) The role of salinity in the development of Phytophthora root
rot of citrus. Phytopathology 76: 970-975
Bland WL (1993) Cotton and soybean root system growth in 3 soil temperature regimes. Agron
J 85: 906-911
Bland WL, Dugas WA (1988) Root length density from minirhizotron observations. Agron J 80:
2,271-275
Bohm, W (1979) Methods of studying root systems. Ecological Studies 33. Springer, Berlin
Heidelberg New York, 188 pp
Box JE, Smucker AJM, Ritchie JT (1989) Minirhizotron installation techniques for investigating
root responses to drought and oxygen stresses. Soil Sci Soc Am J 53: 115-118
Bragg PL, Govi G, Cannell RQ (1983) A comparison of methods, included angled and vertical
minirhizotrons, for studying root growth and distribution in a spring oat crop. Plant Soi!
73:435-440
Brown DA (1984) Characterization of soybean root distribution. Arkansas Farm Res 33: 3
Brown DA, Upchurch DR (1987) Minirhizotrons: a summary of methods and instruments in
current use. Minirhizotron observation tubes: Methods and applications for measuring
rhizosphere dynamics. ASA Spec Pub150: 15-30
Bruns TD (1995) Thoughts on the processes that maintain local species diversity of ectomycor-
rhizal fungi. Plant Soil 170: 63-73
Buckland ST, Campbell CD, Mackie-Dawson LA, Horgan GW, Duff EI (1993) A method for
counting roots observed in minirhizotrons and their theoretical conversion to root length
density. Plant Soil 153 (1): 1-9
Campbell CD, Mackie-Dawson LA, Reid EJ, Pratt SM, Duff EI, Buckland ST (1994) Manual
recording of minirhizotron data and its application to study the effect of herbicide and
nitrogen fertiliser on tree and pasture root growth in a silvopastoral system. Agrofor Syst
26 (2): 75-87
Cheng W, Coleman DC, Box JE Jr (1990) Root dynamics, production and distribution in agroe-
cosystems on the Georgia Piedmont using minirhizotrons. J Appl Ecol 27: 2,592-604
Cheng W, Coleman DC, Box JE Jr (1991) Measuring root turnover using the minirhizotron tech-
nique. Agricult Ecosyst Environ 34: 261-267
Christakos G (1992) Random field models in earth sciences. Academic Press, New York
Colin-Belgrand M, Joannes H, Dreyer E, Pages L (1989) A new data processing system for root
growth and ramification analysis: description of methods. Ann Sci For 46 Supp1305s-309s
De Ruijter F, Veen B, Van Oijen M (1996) A comparison of soil core sampling and minirhizotrons
to quantify root development of field-grown potatoes. Plant Soil182 (2): 301-312
Dinkelaker B, Marschner H (1992) In vivo demonstration of acid phosphatase activity in the
rhizosphere of soil-grown plants. Plant Soil144: 199-205
Dinkelaker B, Hahn G, Marschner H (1993) Non-destructive methods for demonstrating chem-
ical changes in the rhizosphere. II. Application of methods. Plant Soil155/156: 71-74
Dubach M, Russelle MP (1995) Reducing the cost of estimating root turnover with horizontally
installed minirhizotrons. Agron J 87 (2): 258-263
Durrant MJ, Love BJG, Messem AB, Draycott AP (1973) Growth of crop roots in relation to soil
moisture extraction. Ann Appl Biol 74: 387-394
Dyer D, Brown DA (1983) Relationship of fluorescent intensity on ion uptake and elongation
rate of soybean roots. Plant Soil 72: 127-l34
Egli S, KaIin I (1991) Root window technique for in vivo observation of ectomycorrhiza on forest
trees. In: Norris JR, Read DJ, Varma AK (eds) Techniques for the study of mycorrhiza.
Methods in microbiology, voI 23. Academic Press, London, pp 423-433
Enslin WR, Pregitzer KS, Hendrick RL (1994) MSU ROOTs: a PC-based program to quantify
plant roots. Center for Remote Sensing, Michigan State Univerity, East Lansing
Fahey TJ, Hughes JW (1994) Fine root dynamics in a northern hardwood forest ecosystem,
Hubbard Brook Experimental Forest, NH. J Ecol 82: 533-548
Ferguson JC, Smucker AJM (1989) Modifications of the minirhizotron video camera system for
measuring spatial and temporal root dynamics. Soil Sci Soc Am J 53: 1601-1605
Fernandez OA, Caldwell MM (1975) Phenology and dynamics of root growth of three cool
semi-desert shrubs under field conditions. J Eco163: 703-714
Fordham R (1972) Observations on the growth ofroots and shoots oftea (Camellia sinensis, 1.)
in southern Malawi. J Hortic Sci 47: 221-229
Gijsman AJ, Floris J, Van Noordwijk M, Brouwer M (1991) An inflatable minirhizotron system
for root observations with improved soil/tube contact. Plant Soil134 (2): 261-270
Glinski DS, Karnok KJ, Carrow RN (1993) Comparison of reporting methods for root growth
data from transparent-interface measurements. Crop Sci 33: 310-314
Gottlein A, Hell U, Blasek R (1996) A system for microscale tensiometry and Iysimetry. Geo-
derma 69: 147-156
Goulard M (1989) Inference in a coregionalization model. In: Armstrong M (ed) Geostatistics,
voi 1. Kluwer, Dordrecht, pp 397-408
Goulard M, Voltz M (1992) Linear coregionalization model: tools for estimation and choice of
cross-variogram matrix. Math Geol 24 (3): 269-286
Gregory PJ (1979) A periscope method for observing root growth and distribution in field soil.
J Exp Bot 30: 205-214
Hahn G, Marschner H (1998) Effect of acid irrigation and liming on root growth of Norway
spruce. Plant Soi! 199: 11-22
Hansson AC, Andren O (1987) Root dynamics in barley, lucerne and meadow fescue investigated
with a minirhizotron technique. Plant Soi! 103: 33-38
Hansson AC, Steen E, Andren O (1992) Root growth of daily irrigated and fertilized barley.
Investigation with ingrowth cores, soil cores and minirhizotrons. Swed J Agric Res 22 (4):
141-152
Hăussling M, Leisen E, Marschner H, Romheld V (1985) An improved method for non-
destructive measurements of the pH at the root-soil interface (rhizosphere). J Plant
Physiol117: 371-375
Hăussling M, George E, Lorenz K, Kreutzer K, Marschner H (1991) Einfluss von saurer Bereg-
nung aufWachstum von Langwurzeln und pH-Werte in der Rhizosphăre von Fichten im
Versuch Hoglwald. In: Kreutzer K, Gottlein A (eds) Okosystem-forschung Hoglwald:
Beitrăge zur Wirkung von saurer Beregnung und Kalkung in einem Fichtenbestand.
Forstwissenschaftliche Forschungen, Heft 39. Parey, Hamburg, pp 44-48
Hayes DC, Seastedt TR (1987) Root dynamics of tallgrass prairie in wet and dry years. Can J Bot
65:787-791
Head GC (1966) Estimating seasonal changes in the quantity of white unsuberized root on fruit
trees. J Hortic Sci 41: 197-206
Heeraman DA, Juma NG (1993) A comparison of minirhizotron, core and monolith methods for
quantifying barley (Hordeum vulgare L.) and faba bean (Vicia faba L.) root distribution.
Plant Soi1148: 29-41
Hendrick RL, Pregitzer KS (1992a) Spatial variation in tree root distribution and growth asso-
ciated with minirhizotrons. Plant Soi! 143 (2): 283-288
Hendrick RL, Pregitzer KS (1992b) The demography of fine roots in a northern hardwood forest.
Ecology 73 (3): 1094-1104
Hendrick RL, Pregitzer KS (1993) Patterns of fine root mortality in two sugar maple forests.
Nature 361: 59-61
Horgan GW, Buckland ST, Mackie-Dawson LA (1993) Estimating three-dimensionalline process
densities from tube counts. Biometrics 49 (3): 899-906
Huck MG, Taylor HM (1980) The rhizotron as a tool for root research. Adv Agron 35: 1-35
Hummel JW, Levan MA, Sudduth KA (1989) Minirhizotron installation in heavy soils. Trans
ASAE 32: 3,770-776; Presented as ASAE Paper No 87-1524
Itoh S (1985) In situ measurement of rooting density by micro-rhizotron. Soil Sci Plant Nutr 31
(4): 653-656
Keyes MR, Grier CC (1981) Above- and below-ground net production in 40-year-old Douglas fir
stands on low and high productivity sites. Can J For Res 11: 599-605
Klepper B, Kaspar TC (1994) Rhizotrons: their development and use in agricultural research.
Agron J 86: 745-753
Kloeppe1 BD, Gower ST (1995) Construction and installation of acrylic minirhizotron tubes in
forest ecosystems. Soil Sci Soc Am J 59 (1): 241-243
Kolesnikov VA (1971) The root system offruit plants. Mir Publishers, Moscow, 269 pp
Kosola KR, Eissenstat DM (1994) The fate of surface roots of cit rus seedlings in dry soi!. J Exp
Bot 45: 1639-1645
Larigauderie A, Richards JH (1994) Root proliferation characteristics of seven perennial arid-
land grasses in nutrient-enriched microsites. Oecologia 99: 102-111
LeNoble ME, Blevins DG, Miles RJ (1996) Prevention of aluminium toxicity with supplemental
boron. II. Stimulation of root growth in an acidic, high-aluminium subsoi!. Plant CeH
Environ 19: 1143-1148
Levan MA, Ycas JW, Hummel JW, Taylor HM (1987) Light leak effects on near-surface soyabean
rooting observed with minirhizotrons. Minirhizotron observation tubes: Methods and
applications for measuring rhizosphere dynamics. ASA Spec Publ 50: 89-98
and the interpretation of results. In: Atkinson D (ed) Plant root growth: an ecologica! per-
spective. Blackwell, Oxford, pp 25-47
Mackie-Dawson LA, Buckland ST, Duff EI, Pratt SM, Reid EJ, Millard P (1989) The use of in situ
techniques for the investigation of root growth. Asp Appl Biol 22: 349-356
Majdi H, Nylund JE (1996) Does liquid fertilization affect fine root dynamics and life span of
mycorrhizal short roots? Plant Soil185: 305-309
Majdi H, Smucker AJM, Persson H (1992) A comparison between minirhizotron and monolith
sampling methods for measuring root growth of maize (Zea mays L.). Plant Soil 147:
127-134
McDougall WB (1916) The growth of forest tree roots. Am J Bot 3: 384-392
McMichael BL, Taylor HM (1987) Applications and limitations of rhizotrons and minirhizotrons.
In: Minirhizotron Observation Tubes: Methods and applications for measuring rhizosphere
McMichael BL, Upchurch DR, Taylor HM (1992) Transparent wall techniques for studying root
growth and function in soi!. J Plant Nutr 15: 753-762
Meisner CA (1991) Peanut roots, shoot, and yield under water stress. Diss Abstr lnt B Sci Eng
52: 1, 3B-4B; Abstr Thesis, University of Georgia, USA, 1990, 125 pp
Melhuish FM, Lang ARG (1968) Quantitative studies of roots in soi!. I. Length and diameters of
cotton roots in a clay-loam soil by analysis of surface-ground blocks of resin-impregnated
soi!. Soil Sci 106: 16-22
Merrill SD (1992) Pressurized-wall minirhizotron for field observation of root growth dynam-
ics.Agron J 84 (4): 755-758
Merrill SD, Upchurch DR (1994) Converting root numbers observed at minirhizotrons to equiv-
alent root length density. Soil Sci Soc Am J 58: 1061-1067
Merrill SD, Doering EJ, Reichman GA (1987) Application of a minirhizotron with flexible,
pressurized walls to a study of corn root growth. Minirhizotron observation tubes:Methods
and applications for measuring rhizosphere dynamics. ASA Spec Publ 50: 131-143
Merrill SD, Upchurch DR, Black AL, Bauer A (1994) Theory of minirhizotron root directional-
ity observation and application to wheat and corn. Soil Sci Soc Am J 58 (3): 664-671
Meyer WS, Barrs HD (1985) Non-destructive measurement of wheat roots in large undisturbed
and repacked clay soil cores. Plant Soi186: 237-247
Murphy JA, Hendricks MG, Rieke PE, Smucker AJM, Branham BE (1994) Turfgrass root systems
evaluated using the minirhizotron and video recording methods.Agron J 86 (2): 247-250
Neufeld HS, Durall DM, Rich PM, Tingey DT (1989) A rootbox for quantitative observations on
intact entire root systems. Plant Soill17: 295-298
Pages L, Serra V (1994) Growth and branching of the taproot of young oak trees - a dynamic
study. J Exp Bot 45: 1327-1334
Pan WL, Bolton RP, Lundquist EJ, Hiller LK (1998) Portable rhizotron and color scanner system
for monitoring root development. Plant Soil200: 107-112
Parker q, Carr MKV, Jarvis NJ, Puplampu BO, Lee VH (1991) An evaluation of the minirhizotron
technique for estimating root distribution in potatoes. J Agric Sci 116 (3): 341-350
Parsons WFJ, Miller SL, Knight DH (1994) Root-gap dynamics in a lodgepole pine forest: ecto-
mycorrhizal and nonmycorrhizal fine root activity after experimental gap formation. Can
J For Res 24: 1531-1538
Pearson RW, Lund ZF (1968) Direct observation of cotlon root growth under field conditions.
Agron J 60: 442-443
Poelman G, Van De Koppel J, Brouwer G (1996) A telescopic method for photographing within
8 x 8cm minirhizotrons. Plant Soil185: 163-167
Reid JB, Winfield D, Sorenson 1, Petrie RA (1992) Test of the reliability of acrylic plastic windows
for studies of kiwifruit roots. N Z J Crop Hortic Sci 20: 463-466
Rutherford MC, Curran B (1981) A root observation chamber for replicated use in a natural
plant community. Plant Soil63: 123-129
Rygiewicz PT, Johnson MG, Ganio LM, Tingey DT, Storm M (1997) Lifetime and temporal occur-
rence of ectomycorrhizae on ponderosa pine (Pinus ponderosa Laws.) seedlings grown
under varied atmospheric CO 2 and nitrogen levels. Plant Soil 189: 275-287
Samson BK, Sinclair TR (1994) Soil core and minirhizotron comparison for the determination
of root length density. Plant Soil161 (2): 225-232
Schuurman II, Goedewaagen MAI (1971) Methods for the examination of root systems and
roots, 2nd edn. Centre for Agricultural Publishing and Documentation (PUDOC),
Wageningen, the Netherlands
Smit AL, Zuin A (1996) Root growth dynamics of Brussels sprouts (Brassica olearacea var
gemmifera) and leeks (Allium porrum L.) as reflected by root length, root colour and UV-
fluorescence. Plant Soil185: 271-280
Smit AL, Groenwold J, Vos J (1994) The Wageningen Rhizolab - a facility to study soil-root-
shoot-atmosphere interactions in crops 2. Methods of root observations. Plant Soil161 (2):
289-298
Smit AL, Booij R, Van der Werf A (1996) The spatial and temporal rooting pattern of Brussels
sprouts and leeks. Neth J Agric Sci 44: 57-72
Speidel B, Weiss A (1974) Untersuchungen zur Wurzelaktivitiit unter einer Goldhaferwiese.
Angew Bot 48: 137-154
Sword MA (1998) Seasonal development of loblolly pine lateral roots in response to stand
density and fertilization. Plant Soil200: 21-25
Taylor HM, Bohm W (1976) Use of acrylic plastic as rhizotron windows. Agron J 68: 693-694
Taylor HM, Upchurch DR, McMichael BL (1990) Applications and limitations of rhizotrons and
minirhizotrons for root studies. Plant Soil129: 29-35
995-1001
Teskey RO, Hinckley TM (1981) lnfluence of temperature and water potential on root growth of
white oak. Physiol Plant 52: 363-369
Tscherning K, Leihner DE, Hilger TH, Miiller-Samann KM, El Sharkawy MA (1995) Grass
barriers in cassava hillside cultivation: rooting patterns and root growth dynamics. Field
Crops Res 43: l31-140
Upchurch DR (1985) Relationship between observations in minirhizotrons and true root length
density. PhD Diss Texas Tech Univ, Lubbock (Diss Abstr 85-28594)
Upchurch DR (1987) Conversion of minirhizotron-root intersections to root length density.
Minirhizotron observation tubes: Methods and applications for measuring rhizosphere
Upchurch DR, Ritchie J (1983) Root observations using a video recording system in mini-
rhizotrons. Agron J 75: 1009-1015
Upchurch DR, JT Ritchie (1984) Battery-operated color video camera for root observations in
minirhizotrons. Agron J 76: 1015-1017
Upchurch DR, McMichael BL, Taylor HM (1988) Use of minirhizotrons to characterize root
system orientation. Soil Sci Soc Am J 52: 2,319-323
Van Noordwijk M (1987) Methods for quantification of root distribution pattern and root
dynamics in the field. In: Proc 20th Colloq Int Potash Institute held in Baden bei Wien,
pp 263-281
Van Noordwijk M, De Jager A, Floris J (1985) A new dimension to observations in minirhi-
zotrons: a stereoscopic view on root photographs. Plant Soil 86: 447-453
Van Noordwijk M, Brouwer G, Koning H, Meijboom FW, Grzebisz W (1994) Production and
decay of structural root material of winter wheat and sugar beet in conventional and inte-
grated cropping systems. Agric Ecosyst Environ 51 (1-2): 99-1l3
Vogt KA, Grier CC, Gower ST, Sprugel DG, Vogt DJ (1986) Overestimation of net root produc-
tion: a real or imaginary problem? Ecology 67: 577-579
Volkmar KM (1993) A comparison of minirhizotron techniques for estimating root length
density in soils of different bulk densities. Plant Soil 157 (2): 239-245
Voorhees WB (1976) Root elongation along a soil-plastic container interface. Agron J 68: 143
Vos J, Groenwold J (1983) Estimation of root densities by observation tubes and endoscope.
Vos J, Groenwold J (1987) The relation between root growth along observation tubes and in bulk
soi!. Minirhizotron observation tubes: Methods and Applications for Measuring Rhizos-
phere Dynamics. ASA Spec Publ 50: 4, pp 39-49
Wang ZQ, Burch WH, Mou P, Jones RH, Mitchell RJ (1995) Accuracy of visible and ultra violet
light for estimating live root proportions with minirhizotrons. Ecology 76 (7): 2330-2334
Wiesler F, Horst WJ (1994) Root growth of maize cultivars under field conditions as studied by
the core and minirhizotron method and relationships to shoot growth. Z Pf!anzenernahr
Bodenkd 157 (5): 351-358
Wilson K, Gunn A, Cherrett JM (1995) The application of a rhizotron to study the subterranean
effects of pesticides. Pedobiologia 39: l32-143
Zillmann K-H (1956) Beobachtung des Wurzelwachstums in Feldbestanden. Dtsch Landwirtsch
7:394-400
CHAPTER9
The Measurement and Analysis

of Fine Root Longevity
J.E. Hooker, R. Hendrick I , and D. Atkinson 2
School of Applied Sciences, University of Glamorgan, Pontypridd, Mid-Glamorgan CF37 lDL
UK
1 School of Forest Resources, University of Georgia, Athens, Georgia 30602-2152, USA
2 SAC, West Mains Rd., Edinburgh, EH9 3JG, UK
CONTENTS
9.1.1 The Importance of Fine Root Longevity in Ecosystem
Processes 274
Carbon Cycling 274
Nutrient Cycling 275
Resources Acquisition 275
9.1.2 Factors Affecting Fine Root Longevity 276
Symbiosis 276
Soil Temperature 276
Soil Moisture 276
Nutrient Availability 277
Carbon Dioxide 277
9.2 Indirect Methods 277
9.3 Direct, Observation-Based Methods 278
9.3.1 Cohort Selection 279
9.3.2 Cohort Size 279
9.3.3 Imaging and Birth Intervals 280
9.3.4 Censoring 284
9.3.5 Survivorship versus Longevity 284
9.3.6 Survivorship Comparisons 285
9.3.7 Longevity 286
9.3.8 Survivorship and Survival Distrihutions 288
9.3.9 Among-Cohort Life Span Comparisons 289
9.3.10 Survival Effects Models 297
9.4 Nutrient Fluxes and Root Longevity 299
9.5 The Future 301
References 302

274 J.E. Hooker et al.
9.1 Introduction
9.1.1 The Importance of Fine Root Longevity

in Ecosystem Processes
A better understanding of ecosystem processes is an important research

goal for scientists in the next millennium. It offers the potential to not only
make better use of our finite natural resources but also to properly understand
the impacts of anthropogenic events such as global climate change. It will
also allow us to make more informed assessments of how any deleterious effects
may be ameliorated. One of the important research targets for terrestrial
ecosystems is to understand biogeochemical cycles, as it is these fluxes (and
the processes which drive them), which are fundamental to ecosystem
stability.
For many years it has been known that roots play an important part in
these processes, firstly by the uptake of nutrients from the soil, and secondly
by returning them back into the soil at mortality. However, practic al difficulties
in quantifying root production, especially in natural ecosystems, have resulted
in many estimates of root production and their role in nutrient cycling
being derived from above-ground data (Atkinson and Fogel 1997). Recently,
more direct methods have become available, and their application has already
provided some evidence to suggest that the level of mortality experienced by
roots means that their impact on terrestrial biogeochemical cycles may be
much greater than previously considered. Essentially, the life of some roots
has been shown to be relatively short (Hendrick and Pregitzer 1992, 1993;
Hooker et al. 1995; Forbes et al. 1997; Black et al. 1998). Previous estimates
based on allometric relationships are thus likely to have been misleading and
have underestimated the contribution of roots to system productivity and the
cycling of nutrients. However, although this recent data suggests that short life-
times may be common for roots, more data needs to be collected. Thus, it is
hoped that this comprehensive account on appropriate methods will facilitate
research.
Carbon Cycling. The need to quantify both root biomass and turnover
has become more important as a consequence of interest in global
climate change. Cannel et al. 1992, summarizing data from a range of sources,
suggested that soil, which contains around 1500Gt relative to 500Gt in the
terrestrial biosphere, is a major component of the global carbon budget.
They acknowledged that, as the part of the biosphere embedded within the
soil, plant root systems represented one of the principal means for the
movement of carbon from both the atmosphere and the terrestrial biosphere
into the soil. However, estimates of the quantity of photosynthate passing
9 The Measurement and Analysis of Fine Root Longevity 275
to root systems as a whole, and to new root production, and of root turn-
over in terrestrial ecosystems, are more uncommon than their importance
would warrant. This deficiency relates principally to methodological problems
in quantifying temporal changes in root production and mortality. Current
estimates include those of Atkinson (1985) who, in a study of an estab-
lished apple plantation, quantified root system production as 2.5 Mgha- I
year- I and estimated that 75% of this new production turned over within a
single year, i.e. an annual root turnover figure of around 1.9Mgha- 1 year- l •
Cannel et al. (1992), obtained figures of a similar magnitude when they esti-
mated fine root production in stick spruce, eucalyptus and beech woodland as
2.8, 1.8 and 2.2Mgha- 1 year- 1 respectively. More recent1y, R. Fogel (unpubl.
data) quantified the annual production for a Populus woodland in northern
Michigan, USA as 16.5Mg and calculated that 5.1Mg i.e. about 30% was in the
form of fine roots. However, neither of the latter studies included estimates of
root survival.
Nutrient Cycling. Estimates of the impact of root longevity on nutrient cycling

are even less common than those for dry matter production. In one, a study
of an established apple orchard, Atkinson (1985) estimated that annual recy-
cling of nitrogen within the root system would amount to approximately
16kgNha-1 year- 1• Moreover, variation in new root production during the
course of a season suggested that the rate of release of nitrogen as a result of
root mortality would also be non-uniform during an annual cycle and that a
maximum of 1.5 kgNha- 1 week- 1 was likely to be released as a consequence of
root mortality and decomposition. Similar estimates have been made for alfalfa
by Andren et al. (1991) who estimated annual nitrogen release due to root
decomposition as 32kgNha-1• In a more recent study, Goins and Russelle (1996)
estimated the annual nitrogen release, as a consequence of fine root turnover
in alfalfa, as 60kgNha- 1• It is thus clear that root decomposition has the poten-
tial to influence the size of fertilizer inputs.
Resources Acquisition. Problems in acquiring information on root length in

soils has led to many estimates of nutrient inflow to plants that are based
upon rather few and infrequent measurements of total root system length.
Where the life of roots in soil is long, infrequent measurements of root
length will be satisfactory for the calculation of variation in inflow during dif-
ferent parts of the season. However, where root longevity is short, and there is
now increasing evidence to show that this is often the case (see above), esti-
mates of nutrient inflow to the root system may be lower than those which will
occur in practice. In addition, the length of the root system actually available
for nutrient uptake in the soil at any time will be critically influenced by root
longevity.
9.1.2 Factors Affecting Fine Root Longevity
Symbiosis. That symbiotic interactions occur between specific groups of

microorganisms and plant roots has been known for many years. However,
although much research effort has been devoted to the study of these symbioses,
studies of their effects on root system development are very limited. It is known
that effects on root architecture can occur (e.g. Hooker et al. 1992; Berta et al.
1995; Norman et al. 1996; for review see Hooker and Atkinson 1996). However,
the effects of symbiotic interactions on the mortality of roots are almost
unknown, with only one report of an appropriate study by Hooker et al. (1995)
who used mini-rhizotrons to study mortality of roots in root systems of poplar
and identified substantial reductions in the longevity of roots when they were
colonized by an arbuscular mycorrhizal fungus (AMF). They also highlighted
the importance of this finding for understanding the role of AMF in ecosystem
functioning. Further studies are clearly needed.
Soil Temperature. There are only a few reports on the effects of soil temperature
on the longevity of plant roots. In the first, Head (1966) studied variation in the
longevity of apple roots over a complete season. He found that root longevity was
maximal during the winter period (approximately 12 weeks), fell to a minimum
value in mid May (approximately 3 weeks) and remained at this rate until
autumn. Later, Hendrick and Pregitzer (1993) compared the pattern of fine root
mortality in two Acer saccharum forests located approximately 80 km apart on a
north-south transect. They measured an enhanced rate of root mortality at the
southern site and hypothesized that this could be linked to the warmer soil tem-
peratures found at this site. These studies thus both suggested that soil temper-
ature has an effect upon longevity and this was recently confirmed by Forbes et
al. (1997) in experiments with Lolium perenne. Here they found that increases in
temperature resulted in a de crease in the longevity of roots but, because they
used soil, could not exclude indirect effects such as changes in soil nutrient levels
due to altered microbial activities. However, Forbes et al. (1996) had previously
reported the direct effect of temperature on root system architecture of Plantago
lanceolata in carefully controlled microcosms. The scale of the effects they
observed i.e. increased branching with increases in temperature would concur
with the data from the later study and (assuming that higher order roots do not
survive for as long as lower order roots) strongly suggest that temperature is an
important determinant of longevity.
Soil Moisture. Although it is well-known that low soil water potential will
adversely effect the rate of root growth (Gregory 1987; Klepper 1987), there have
been relatively few studies investigating the effect of soil water potential on root
longevity. The potential effect of soil water potential on root longevity will be
more complex in perennial crops because of soil water potential effects upon
root branching (Atkinson 1993) which in time interacts with root longevity.
However, the absence of informat ion on root longevity makes it difficult to
assess whether reported effects of low water potentials on root length are a con-
sequence of accelerated root death or impeded new root production, or a com-
bination of both. This question needs to be resolved.
Nutrient Availability. Despite the large number of studies of the effect of nutri-
ent supply on root production and morphology (Drew 1987), there have been
relatively few studies of the effect of nutrient supply on root longevity. However,
in 1993 Pregitzer et al. assessed the effects of enhanced nitrogen supply on root
turnover in Prunus. They found that the supply of nutrients greatly enhanced
root initiation and that the flush of roots initiated in response to an enhanced
nutrient supply had a shorter longevity than those roots which developed in
non-enriched patches. It is known that the appHcation of nutrients to a root
system can cause a major alterat ion to root branching (Drew 1987). It is, though,
unclear as to whether the response identified by Pregitzer et al. was a direct
response in terms of the roots produced, or one consequent upon the produc-
tion of a number of higher order lateral roots which would themselves survive
a shorter period of time than lower order root branches. This is an are a which
is likely to benefit from future investigation.
Carbon Dioxide. Although elevated levels of CO 2 have been shown to increase

root production, e.g. Curtis et al. (1995), Thomas et al. (1996), Httle information
exists about effects on root longevity. In the study by Thomas et al. (1996),
increased root production, which varied in magnitude at different times during
the season, was accompanied by a change in the timing of the onset of new
growth which occurred around 4 weeks earlier in trees growing under elevated
CO 2 conditions. Enhanced CO 2 growing conditions were also associated with an
increase in CO 2 flux from the soil. However, likely errors associated with mea-
surements of both root production and CO 2 flux were such that it was not pos-
sible to use the data of Thomas et al. (1996) to assess whether the roots produced
in the high CO 2 atmosphere showed a higher rate of turnover than those pro-
duced under ambient conditions. This is another area which would benefit from
further investigation.
9.2 Indirect Methods
The use of demographic methods to estimate root production and turnover has
been reviewed by Fogel (1991). A number of studies have attempted to estimate
root production indirect1y by assuming that the ratio of production to mass is
similar for root and shoot systems. However, the ratio of root production to
shoots is known to vary considerably and has been estimated to be between 20
and 100% of the above ground ratio (Harris et al. 1980). Moreover, assumptions
are likely, in most cases, to be invalid and provide misleading data because of
spatial and temporal changes in root systems which may not be reflected above-
ground. The use of this approach has recently been discussed by Atkinson and
Fogel (1997).
An alternative approach is to base estimates upon sequential harvests (see
Chap. 8). Production is estimated from measurements of differences in biomass
on successive sampling dates with these estimates being corrected for losses
(Fogel and Hunt 1979). This process is described by the equation:
npp = Bt+! - Bt+ L,

where npp is net primary production, Bt is the standing crop at time t and L is
the amount lost by rhizodeposition. Estimates obtained using this method are
improved if roots are separated into live and dead categories rather than simply
working upon the difference between maximum and minimum standing crops
for aH roots (Fairley and Alexander 1985). This method is usually described as
the method ofbalancing transfers. One problem is that in practice it can be dif-
ficult (in material obtained from soil cores) to easily distinguish between live
and dead roots. Approaches to resolving these differences have been discussed
by Fogel (1985). However, although the harvest method (adjusted by the bal-
ancing transfers approach) is helpful in producing relative data for different
ecosystems, it does not give sensitive estimates of either production or
decomposition; essentially because of production and undetected mortality
(e.g. roots that rapidly decompose or are consumed) that may have occurred
between sampling dates.
An alternative means of obtaining production estimates is through use of
the mesh bag method (e.g. Steen 1991). In this method the amount of root
grown within a cylindrical mesh bag filled with a given volume of soil, left in
the field for a given period of time, is taken as the rate of root production. The
roots, dead and living, found within the bag after a certain period are assumed
to have grown into the bag after placement and to represent total production.
The method has significant relative value in field trials where experimental
treatments are being compared. Again though the derivation of real rates of pro-
duction are not detectable because of production and undetected mortality that
may have occurred between sampling dates.
9.3 Direct, Observation-Based Methods
Until recently, most observational studies of fine root dynamics have reported
only net temporal changes in root dry weight or at best total root length or
numbers. We now recognize that failing to document changes on an individual
root or root-segment basis can render observational techniques (see Chap. 8 on

minirhizotron methods) no more informative than traditional coring tech-
niques. Fortunately, observation studies lend themselves to demography-based
research, in which the life histories of individual roots are documented from
the time of a root's "birth" until its death. This population-based approach
enables us to quantify root production, longevity and turnover as distinct
processes. Moreover, temporal changes in root dimension and appearance that
may be of functional significance can be measured simultaneously.
The analytical techniques of demographers and population ecologists are
well-developed, and facilitate relatively sophisticated analyses of dynamic root
data. This is particularly true for data derived from studies of root cohorts; that
is, populations of roots of a known age that were produced during the same,
relatively short time period. Cohort studies have a long history in plant, animal
and insect ecology, and form the basis of actuarial science. Their history of use
in root research is much shorter, but the overall approach shows considerable
promise.
9.3.1 Cohort Selection
Cohort-based studies depend upon repeated measures of the same roots, and
hence require the ability to return to the same location on either a minirhi-
zotron tube or an observation window (Atkinson 1985). Repeatedly sampling
the same area can be accomplished by physically marking sampling locations
on tubes or windows, drawing of the root system (Atkinson and Fogel1997) or
via a mechanical guide on the device used to record stiH or video images.
Marking and re-Iocating individual roots can be accomplished either with com-
puter-aided image analysis programs (e.g. ROOTS and RooTracker) or by hand
if using acetate overlays and tracing.
9.3.2 Cohort Size
Cohort sizes should be as large as the time and logistic al constraints involved
in collecting and processing the data wiH allow. The degree of precision with
which survivorship and mortality rates can be estimated necessarily increases
along with cohort size, and accuracy typically increases with sample size as well.
Cohort sizes in our own research have ranged from 26 to 245 individual
roots. The rooting characteristics peculiar to the plant species and soil
environment under study will dictate root densities in observation tubes or
windows, and hence the number of roots or root segments encountered per
unit sampling area. When compiling cohorts in studies in which sampling areas
in the window or rhizotrons are adjacent, it is important that roots growing

out of one sampling area and into another are counted only once, or else that
root will contribute disproportionately to the data set. As stated, it is logical to
conclude that increasing numbers of roots per cohort will result in greater
precision of life span estimates, and presumably greater accuracy as well.
However, the specific number of roots needed to detect treatment or other dif-
ferences in life span during some specific period of time depends upon the mag-
nitude of life span difference one wishes to detect (~), the choice of significance
level with respect to making a Type 1 error (a, the probability of rejecting a true
null hypothesis) and the desired level of the power of the test (1-/3, where /3 =
probability of making a Type II error). The number of samples need for a par-
ticular combination of ~, a and /3 depend upon the distribution of life spans,
but values for an assumed exponential distribution with cohorts of equal size
have been calculated by George and Desu (1974), some of which are summa-
rized in Table 9.1. Makuch and Simon (1982) present a more generalized
approach where the number of treatments is greater than two, and include tab-
ulated data as well as the equation and appropriate constants for various com-
binations of treatments, a and /3. For comparisons of two treatments, their
approach also yields the data in Table 9.1, but in addition to making the gen-
eralized case for comparisons among more treatments, allows flexibility in the
choice of ~, a and /3.
9.3.3 Imaging and Birth Intervals
Given that cohorts are defined as all individual roots produced or "born" at
approximately the same time, it is appropriate to ask as to how wide a time inter-
Table 9.1. The number of individuals required in each cohort

to detect a desired degree of difference in life spans (d) for
particular combinations of a and /3 when treatments are two in
number
Level of d
1-/3 a 10% 20% 50% 100% 200%

0.95 0.01 3749 951 193 66 17
0.05 2384 652 132 46 12
0.90 0.01 2870 785 159 55 14
0.05 1884 515 105 36 9
0.80 0.01 2213 605 123 42 11
0.05 1360 372 76 26 7
val constitutes "approximately the same time"? In human actuarial studies,

periods of one or more years are often considered to constitute a birth inter-
val. However, plant roots may live for as liule as a few days to several years, and
root longevity varies both within and between species (Nadelhoffer et al. 1985;
Hendrick and Pregitzer 1993; Hooker et al. 1995; Black et al. 1998). Rates of root
mortality are age-dependent in some species (Hendrick and Pregitzer 1993;
Black et al. 1998), and may be considerable within a few days ofbirth, dropping
dramatically thereafter (Pregitzer et al. 1993). Moreover, most plant communi-
ties are probably comprised of species with different root longevities, which
may complicate studies of root dynamics in mixed-species environments.
Equally critical is the choice of the proper interval after which the cohort is re-
sampled. Sampling too infrequently will underestimate actual root birth rates
and lead to estimates of survivorship that are too conservative. However, sam-
pling more often than necessary can result in excessive investments of time and
resources that result in liule or no increase in the quality of data collected.
The age of roots within a cohort are usually defined based upon the date
upon which they were first observed. However, the true birth date of a root at
the time it is first observed can be anywhere between the day immediately fol-
lowing the previous sampling date and the day immediately preceding the date
of first observation. Similarly, the first time a root is observed as dead is unlikely
to be the day upon which it died. Without beuer data from prior experience,
one can only as sume that the births of new roots and the deaths of old roots
are distributed evenly throughout the time period between observations. There-
fore, the "ave rage" root is assumed to be born halfway between observations,
and similarly to die halfway between observations. Therefore, aging a cohort as
zero at the date of first observation and then chronologically from that point
forward is numerically equivalent to assuming even distributions of births and
deaths between observations. If the distribution of births and deaths is known
to be other than even, root ages should be adjusted accordingly.
Clearly, any large-scale, intensive study of root longevity should be based
upon some prior knowledge of the life history of roots for that or a similar
species or community. Decisions regarding acceptable cohort birth intervals
and time periods between sampling will rarely be the same for two investiga-
tors or even two successive experiments. However, they should be made with
the following criteria in mind:
1. What percentage of the entire cohort born within a given time interval is
represented at the time of first observation, and is the likelihood of a root's
demise markedly higher in the first few days of existence?
2. How precise must estimates of survivorship and longevity be?
3. In mixed-species experiments, are the observed roots dominated by one or
more species with known or suspected root longevities markedly longer or
shorter than other species in the community? Studies in which species-

specific root longevity is of interest might require sampling intervals struc-
tured around each species.
4. If experimental treatments hypothesized to affect longevity are to be
imposed, in what direction (i.e. increased or decreased longevity) are the
hypothesized effects to be manifested?
5. To what extent will financial and logistical considerations affect the sampling
frequency?
Generally, sampling intensity and frequency will be determined by striving for
a balance between the amount of data desired and the amount that can be col-
lected and analyzed. The interval will be generally influenced by the species
being studied and the normal rate of its root turnover. We have used intervals
of 2 days to 1 month when studying the roots of woody species (e.g. Hendrick
and Pregitzer 1992; Pregitzer et al. 1993; Hooker et al. 1995; Black et al. 1998),
and Dubach and Russelle (1995) concluded that periods of2 weeks were appro-
priate for alfalfa roots. We note that in our experience using video imagery, col-
lecting demographic data adds little to the total cost of the research, and that
most of our resources are devoted to extracting and analyzing the data at a later
date. It may be best to collect data as frequently as possible, even if there is little
chance that alI of it can be analyzed by an experiment's end. Partial analysis of
frequently collected data can give insight into the amount of information lost,
if any, by analyzing data from alternate sampling dates or increasing the inter-
val defining a cohort to increase sample size. For general guidance a sampling
protocol is given in Box 9.2.
Finally, before commencing any measurements, it is vital, given the very sig-
nificant inputs required, that a detailed "cost/benefit analysis" of the protocol
be carried out. As for general guidelines for maximizing the return from one's
efforts to quantify fine root life span or other demographic processes using
observational techniques, we offer the following:
1. Familiarize yourself with the general morphology of the root system
from excavations prior to beginning observational studies. How does the
appearance (e.g. colour, texture, branching habit, etc.) of the root change as
it ages and develops (e.g, from tip to base, or between woody and non-woody
roots).
2. What do dead roots look like? Are there abundant obviously dead roots in
the soil, or are they rare? Given that we have sometimes witnessed what casu-
ally appeared to be dead roots later giving rise to new lateral branches, as
well as the fact that many roots disappear altogether between observation
intervals due to rapid decomposition, herbivory or a combination of both, it
is important that one knows a priori something about what to expect during
the life history of the roots under study.
Table 9.2. Suggested sampling intervals for observation-based techniques
Vegetation type Variable Suggested sampling interval
Annual crops Overall production and mortality 1-2 weeks (best); monthly
or generallife span (coarse)
Early mortality or precise life span 2-5 days (best); 1-2 weeks
(sufficient)
Perennial plants Overall production and mortality 2 weeks - monthly (best);

(growing season, annual) or 2 months (coarse)
generallife span
Overall production and mortality monthly - 2 months

(dormant season) or general
life span
# Early mortality or precise life span 2-7 days (best), 2 weeks or

monthly (sufficient or coarse)
3. Where are most of the roots physically located in the soil? Are the roots of
primary interest mycorrhizal or those most important for nutrient uptake
(likely to be shallow, perhaps within the upper 20-30cm of the soil), or are
deep roots that may be important during periods of high water demand also
of interest? The general vertical distribution of roots, as well as those most
appropriate to the research question being posed, should guide the selection
of minirhizotron length or rhizotron depth.
4. What is the extent of the financial and labour resources, and time, that
can be devoted to the study? For example, a 2 m minirhizotron con-
taining perhaps 150 potential images with roots might take only lOmin
to image in the field, but image analysis of that tube for a single date's
sampling could take from under 30 min to more than 3 h to analyze, dep-
ending upon the image analysis system used, the type of data desired
and the density of roots along the minirhizotron surface. Moreover, it is
our experience that analyzing images in the upper 25-40 cm consumes
as much as 75% of analysis time, as most roots are located there. For a
minirhizotron 1 m in length and inserted at a 25-45° angle to the soil
surface in a perennial plant community (e.g. forest or grassland), perhaps
1.5-2.5 h should be allocated for laboratory image analysis for each sampling
date. Time requirements in agricultural systems could be either greater or
less, depending upon root densities. Although the specific requirements
will vary, completely analyzing images from one sample of 12-15 2m
minirhizotrons, or 24-30 1 m minirhizotrons might take a skilled operator
30-70h.
5. What is the nature of the soil into which the minirhizotron is to be instaHed?
Soils with a high clay content often cause problems when the clay settles
and forms films along the surface of the minirhizotron, thereby obscuring
roots. We have partially overcome this problem by hand-digging narrow
trenches into which the minirhizotrons are laid, after which soil is hand-
packed over the minirhizotrons, rather than sliding the minirhizotrons in a
bore-hole.
9.3.4 Censoring
IdeaHy, cohort studies terminate after aH the subjects (i.e. roots) have died.
However, a root longevity study wiH often end prior to the death of aH roots
under observation. When this occurs, those roots stiH alive at the study's end
are said to be right censored. That is, they are known to live for at least the dura-
tion of the study, but their exact survival times are unknown. When the roots
under study are aH comprised of a single cohort, or if cohorts of the same age
are being compared, the censoring is said to be singular. Another example of
singular censoring is when aH roots within a cohort are foHowed until a certain
percentage die, at which time the study is terminated.
Progressively censored data are encountered in studies in which roots pro-
duced under a single treatment but at different times are stiH alive at the time
of termination. Such roots do not comprise a cohort, but progressively censored
data can be analyzed in ways analogous to singularly censored data. We wiH
concentrate primarily on analyses of singularly censored data in our discus-
sions of data analysis below. TechnicaHy, most of the roots studied via direct
observation wiH also be left censored, in that their exact date of birth is
unknown, regardless of their fate during the course of the experiment. Provided
that an acceptably short period defining the cohort birth interval has been
established, given the investigator's objectives and constraints, this left censor-
ing can be ignored without much consequence.
9.3.5 Survivorship versus Longevity
Before discussing statistical tests appropriate to analyzing root life span data, it
is important to distinguish between survivorship and longevity. Survivorship
refers to the proportion of one or more cohorts alive at a particular point in time,
while longevity refers to the length of time between birth and death. Survivor-
ship analyses are concerned with the number of root deaths within a particular
period of time, while longevity analyses address the length of time until death
occurs. With respect to the former, one might wish to compare the root response
of two different cultivars to a particular pathogen for a specified number of days

after infection, for which statistical contingency tests are appropriate. Alterna-
tively, one might wish to evaluate the effects of two or more levels of soil fertil-
ity on root life spans. Statistical tests of treatment effects on longevity quantify
differences in the distribution of life spans within each cohort.
Below we recommend some statistical tests appropriate for describing and
characterizing the survival and mortality of roots. We also present a number of
statistical tests for comparing the longevity of roots subject to two or more dif-
ferent treatments; those treatments might, for example, be soil fertility, irriga-
tion, plant species or time of birth. In our discussion, our objective is not to give
either an exhaustive or an abstract treatment of statistic al survivorship and life
span quantification or analysis. Rather, our intention is to present a broad
overview of some of the more common and accessible means of describing the
nature of root longevity and mortality, and for testing for differences in root
survivorship and longevity, using both real and hypothetical data in selected
examples. For more detailed discussions, we refer interested readers to a
number of texts and papers addressing the statistical descriptions and com-
parisons of survival and life span data. Among those we suggest are Gehan
(1965), Mantel (1966), Kalbfleisch and Prentice (1980) and Cox and Oakes
(1984), Muenchow (1986), Pyke and Thompson (1986), PoHock et al. (1989), Lee
(1992). AH of these references contain worked examples of one or more statis-
tical applications, and Lee's (1992) is a comprehensive, coherent and highly
readable treatise of statistical methods.
9.3.6 Survivorship Comparisons
Comparing the survivorship of two or more cohorts over a particular interval

of time is among the simplest of root life span analyses. When only two cohorts
are involved, either a simple chi-square test or the G-test can be used to deter-
mine if the proportion of original roots surviving to some point in time are
equivalent. The chi-square test statistic is among the most common of aH non-
parametric statistical tests, and we will not review it here. The G-test is less com-
monly used, but its use in place of the chi-square test has been encouraged in
recent years because of its theoretical appropriateness and greater computa-
tional ease (Sokal and Rohlf 1981).
For comparing the survivorship of two cohorts at the end of a specified
period of time, the G -test takes the foHowing form:
G=2(I,!cln!c - I,j;lnj; +nlnn), (9.1)
where the j;'s are the ceH frequencies in a 2 x 2 contingency table of live and
dead roots within the two cohorts; the j;'s are the row and column totals, n is
the total number of roots in both cohorts, and G is distributed approximately

as chi-square with one degree of freedom. A sample calculat ion for 20 roots in
each of two root cohorts that were observed for a period of 1 year is shown in
Box 9.1.
B O X9.1. G -test for Equivalent 1-Year Survivorship of Two Fine Root

Cohorts Growing in Hardwood Forests in Michigan, USA
Site Alive Dead Totals
North 14 6 20
South 8 12 20
Totals 22 18 40
1. Calculate 'L/cln/c for cell frequencies:
= 14 In 14 + 6 In 6 + 8 In 8 + 121n 12 = 94.1518
2. Calculate 'Lf,lnf, for row and column totals:
= 22ln 22 + 18ln 18 + 20 In 20 + 20 In 20 = 239.6586

3. Compute n In n for total number of roots:
= 40 In 40 = 147.5552
4. G = 94.1518 - 239.6586 + 147.552 = 2.0452
9.3.7 Longevity
Given that it is most appropriate to determine the life span of individual roots,
this presents both practical and analysis questions. Earlier we explained that
the age of the root using any of the currently available direct observational
methods will (in practice) be an estimate, with an accuracy related to frequency
of observation. However, equally important is how to determine, from obser-
vation alone, if a root is alive or not. Discerning live and dead roots is a diffi-
culty long associated with studying root life span and dynamics. A variety of
staining techniques have been devised for distinguishing between live and dead
roots extracted from the soil, none of which have proved either universally
acceptable or consistently accurate. Observational approaches are fraught with
similar difficulties. Dead roots are typified by some combinat ion of a dramatic
darkening, a loss of "intactness" (e.g blurring of root edges, apparent decom-

position, etc.), pronounced shrivelling, colonization by fungal (obviously
non-mycorrhizal) hyphae and either gradual or rapid disappearance. However,
none of these criteria are necessarily reliably or consistently associated
with dead roots. For example, darkening is often associated with root ageing,
shrivelling may be due to the collapse of cortical cells in otherwise live but
ageing roots, and distinguishing the hyphae of decomposer fungi from
mycorrhizal hyphae or hyphae associated with other roots (living or dead) or
the bulk soil is difficult. Wang et al. (1995) used a UV light source to distinguish
between live and dead roots of eight species of plants, including grasses, forbs
and both deciduous and evergreen woody species. The live root assessments of
both UV and visible light were similar in their accuracy (0.88 and 0.85, respec-
tively), but they point out that visible light must be used in alI cases to make
definitive determinations of dead roots. Further testing will be needed to deter-
mine whether UV reflectance is a robust and consistent indicator of vitality. It
is thus clearly an advantage if detailed anatomical studies have been carried out
for the genotype of interest. In apple, for example, MacKenzie (1979) showed
that a colour change from white to brown was associated with phellogen devel-
opment, cortical disintegration, and in roots lacking secondary xylem, death;
useful background information for minirhizotron-based investigations of
longevity.
In many cases, though, in natural systems roots merely disappear i.e. they
will be present on one occasion and absent on the next; presumably due to the
activity of saprophytes or predatory insects or invertebrates. The loss of roots
can often be seen in images collected a week apart. On other occasions the
cortex of a root may be very much reduced and the root will have begun to dis-
integrate and is clearly not functional; this is usually the case in more controlled
laboratory studies where a sterilized soil or sand is often used.
Mortality in roots can usually be determined by slightly different charac-
teristics in species of different types, i.e. annuals, herbaceous perennials and
woody perennials. These are detailed in Table 9.3. Woody roots infected byecto-
mycorrhizas are a special case. Here the long survival of the root in an intact
state, following a colour change which would usually precede disintegration and
disappearance indicates its ectomycorrhizal status and so it needs to be assessed
against a different standard.
In using the minirhizotron method, in conjunction with a database for
analyses, e.g. in the program ROOTS (Michigan State University, East Lansing,
MI, USA), a series of condition codes can be assigned to roots. Examples are
detailed below in Table 9.4.
The data can then be summarized and for simplicity the mean or median
values are often calculated. These can be useful statistics but they should be
used cautiously and the distribution of the data fully appreciated.
Table 9.3. Characteristics used to determine root mortality
Plant type Characteristics for root mortality
Annual species (ali roots) Change colour from white to paie brown.
Decrease in diameter
Roots disappear
Herbaceous perennials ("short roots") Usually change colour from white to dark brown
or black
Roots appear white
Roots disappear
Level of arthropod activity around root increases
Herbaceous perennials ("Iong roots") Become dark brown

Decrease in diameter or start to appear rough in
texture
Roots disappear
Woody perennials ("short roots") Change in colour to mid brown

Decrease in diameter
Become more attractive to collembolans
Roots disappear
Woody perennials ("woody roots")· Change from mid brown to dark brown/black
Loose diameter unevenly
Uneven decrease in diameter
• In ectomycorrhizal species these roots may survive for a period of years.
Table 9.4. Condition codes which can be applied in a database
Code Description
N New root. This is assigned only once in a root's life, on the first occasion it appeared.
W White root. Assigned if the root was white and healthy.
B Brown root. Assigned if the root was brown and intact.
X Disintegrating root. Assigned if the root was clearly starting to break up.
A Absent root. Assigned if the root had disappeared. After a root has disappeared it
continues to get an a coding on every sample date.
9.3.8 Survivorship and Survival Distributions
Survival analysis is a powerful technique for studying censored data. However,

although widely used in medical and industrial studies, it has only recently been
applied to root data. Black et al. (1998) recently first used the technique to
compare the distribution of root longevities between cohorts. Essentially, a sur-
vival distribution is a probability distribution of lifetimes. It includes the sur-

vivor function S(t), which defines the probability of mortality for a root at t.
Probabilities for an individual surviving beyond time tare then determined by
the extent to which they have accumulated hazards and various distributions can
then be applied. These include the exponential, Weibull, gamma and lognormal
distributions, and each of these should be tested to determine which provides
the best fit for the data. For details on how to apply these powerful but reasonably
complex techniques we refer readers to Black et al. 1998 (see also Sect. 9.3.10).
9.3.9 Among-Cohort Life Span Comparisons
If none of the roots in a study are censored (that is, alI roots have died prior to
the complet ion of the study), standard non-parametric tests for equality of dis-
tributions can be used. When only two such distributions are being compared,
the Mann-Whitney U-test is appropriate (Conover 1980; Lee 1992). The Mann-
Whitney U-test is a rank test, in which the life spans of two different samples
of roots are arranged and ranked in ascending order. The ranks of one sample
are then summed (and corrected for rank ties if present), and the resulting test
statistic T is compared with the area under the normal distribution at a desired
level of a. The two samples need not have the same number of roots.An example
is given in Box 9.2.
Censored data are more common than non-censored data in many if not
most longevity studies. Lee (1992) has summarized five non-parametric tests
appropriate for pair-wise comparisons of fine root life spans in which the data
are censored, alI of which are also rank tests. These are: Gehan's generalized
Wilcoxan test, the Cox-Mantel test, the logrank test, Peto and Peto's generalized
Wilcoxan test and Cox's F-test. The first four are appropriate for both singular
and progressively censored data. The particular test to use depends upon the
size and structure of the data set under consideration (Lee 1992).
The two generalized Wilcoxan tests give relatively more weight to roots that
die in the study and are more sensitive to early differences in mortality than
the logrank test, which gives a more equal weight to all roots and is more likely
to detect differences manifested late in the study. None of these three tests are
effective at detecting differences in instances where the mortality or survivor-
ship curves of functions cross; for example, when early mortality is much less
in one cohort than the other, but in which a large number of roots die towards
the end of the study such that final survivorship is lower. If the sample sizes are
less than 50, the data approximate to Weibull or exponential distributions and
the data are singly censored, then Cox's F-test has greater statistical power than
the Wilcoxan tests. The reader is referred to Lee (1992) for more detail on the
strengths and weaknesses of each test.
BOX 9.2. Comparison of Two Non-censored Life Span Distributions for

Roots Growing in Soils Receiving Two Different SoH Treatments Using
a Mann-Whitney U-test
Treatment 1 Treatment 2
Life span (days ) Rank Life span (days) Rank
10 1 18 4
15 2 19 5.S
16 3 24 7
19 5.5 25 9
25 9 25 9
45 13.5 35 11
65 15 37 12
71 18 45 13.5
69 16
70 17
Suppose that an experiment designed to asses the impact of two soil treat-
ments on fine root longevityyields the above hypothetical root life spans,
and that an a level of 0.10 is to be considered statistically significant.
1. Arrange life spans (scores) in ascending order in each treatment.
2. Assign ranks within each treatment based upon score position within
the combined data set. Tied scores receive an average rank that would
have been assigned without ties. For example, the 19-day scores (Iife
spans) are the sth and 6th ordered observations, receiving an average
rank of 5.5. Similarly, the 2s-day scores would be the 8th, 9th and 10th
ordered observations in the absence of ties, receiving an average score
of9.
3. The ranks for one treatment are summed to calculate the test statistic
T (treatment 1 in this case, since the number of observations are
smaller and hence the calculations simpler).
T = 1 + 2 + 3 + 5.5 + 9 + 13.5 + 15 + 18 = 67.0
4. In the absence of ties, the above value of T can be used unadjusted as the
test statistic. With ties, it must be adjusted by subtracting the mean rank
and dividing by the standard deviation using the following formula:
N +l
T - n- -
_ 2
T.dj - 2 '
nm 'f 2 nm(N +1)2)
[
N(N _ l)~Ri - 4(N -1)
where N is the total number of root life spans, n is the number oflife
spans in the sample used to calculate T, m is the number oflife spans in
the other sample, R is the score rank, and L Rf is the sum of squares of alI
N ranks in both samples. In the above example the equation takes the
form:
18 +1
67-8--
_ 2
Tadj - 2 = 0.8022,
(8)(10) (2106) _ (8)(10)(19)2 )
(
(18)(17) 4(17)
which corresponds to a p value of approximately 0.221 and indicates

that the soil treatments did not have a statisticalIy significant affect on
fine root longevity.
We have previously used a variant of Gehan's generalized Wilcoxan

test (Hendrick and Pregitzer 1992) to test for life span differences in fine root
cohorts observed in two hardwood forests in Michigan, USA. Gehan's test can
be used either with individual root life span data, or with observations in which
numbers of surviving roots in each cohort are counted at successive observa-
tion intervals. In Gehan's test, the life span of each root i in sample x (denoted
Xi if dying dur ing the study, or xt if alive and therefore censored at the study's
termination) is compared with the life span of each root in sample Y (denoted
Yj if dying or Y/ if censored). The root in sample x is given a score Uij of 1, O or
-1 according the folIowing rules:
xt
1. If Xi > Yj or 2:: Yj' then Uij
= 1
2. If Xi = Yj or xt < Yj or Y/ < Xi or (xt,y) then Uij =O
3. If Xi < Yj or Xi :::; Y/ then Uij
= -1
The test statistic W is computed as:
w=I.I,uij·
i=l i=l
(9.2)
The variance of W is:
(9.3)
The test statistic W is assumed to be distributed approximately normalIy with

mean and variance of zero. Life span difference between the two samples can
be calculated as
z= W (9.4)
(VarW)1/2
If the number of roots in each sample is large, calculating W can be laborious.
Lee (1992) demonstrates Mantel's (1967) alternative for calculating W that
involves combining both samples and then scoring each root based upon its rel-
ative life span ranking. In effect, each root's life span is compared with aU other
roots in the combined sample. Each root's Ui (where i = 1 to ni + n2) score is
then the sum of the number of roots with a life span less than its own, minus
the number of roots with longer life spans. In this case, the U;'s comprise a pop-
ulation with a mean of zero, and:
(9.5)
;=1
If, instead of root life spans, the data are expressed in terms of the number of
roots in the original cohorts surviving and dying between each observation,
then Pyke and Thompson's (1986) approach combining the methods of both
Gehan (1965) and Mantel (1967) in a tabular format can be used, as demon-
strated in Box 9.3.
BOX 9.3. Gehan's Generalized Wilcoxan Test for Life Span Differences
in Two Root Cohorts
Time 5, 52 S d, d2 D >d , <d , U, d,U,

(t)
O 149 196 345 10 12 22 323 O 323 3230

139 184 323 20 7 27 296 -22 274 5480
2 119 177 296 24 11 35 261 - 49 212 5088
3 95 166 261 10 10 20 241 - 84 157 1570
4 85 156 241 11 22 33 208 - 104 104 1144
5 74 134 208 12 31 43 165 - 137 28 336
6 62 103 165 7 6 13 152 - 180 -28 - 196
7 55 97 152 3 10 13 139 - 193 - 54 - 162
8 46 87 139 6 4 10 129 -206 - 77 - 462
9 46 83 129 5 5 10 119 -216 -97 -485
10 41 78 119 10 5 15 104 - 226 - 122 - 1220
11 31 73 104 3 9 12 92 -241 -149 - 447
12 28' 64' 92' 92' O - 253 -253 -7084
Sum 6972
, Censored
The above table summarizes survival (5) and mortality (d) data for two
root cohorts (ni = 149 roots and n2 = 196 roots) from the time of first
observation (to = month O) untiI the termination of the study 17 months

later. Although only 19% of the roots in cohort 1 were alive at the study's
end, as opposed to 33% in cohort 2, we need to test for differences in
survival distributions to determine if average life spans are the same in
both cohorts (Ho ). The test statistic Wand the determination of a
significant cohort difference in root longevity is calculated as follows:
1. For each observation date t, tabulate the number of roots in each
sample alive at the time of observation (columns 51 and 52 for samples
1 and 2, respectively), and the number of those roots dying between
that observation date and the following observation date (columns d l
and d2, respectively).
2. For each observation date t, sum columns 51 and 52 (= St), and d l and
d 2 (= Dt),
3. For each observation date t, calculate the total number of roots in
both cohorts that survive longer than the time period t to t + 1
(column >dl •t = St - Dt), assigning each root a value of 1 according to
the rules outlined above. For each observation date t, calculate the
total number of roots that died prior to that date during alI time
periods (column <dl - So - St), assigning each root a value of -1
according to the rules outlined above.
5. Calculate Ui at each time t by adding >dl •t to <dl •t ,
6. Calculate W using Eq. 9.2 by summing for all t's the product of d l •t and
Ui•t (column d l •t UI •t):
W = 3230 + 5480 + ... - 7084

= 6972 ..
7. Calculate the variance according to Eq. 9.3, which equates to the
following:
VarW= n l n2 LDt(Ut/
(ni +n2)(nl +n2 -1) t=1
(149)(196) (2 2 2
=( )( )22323) +27(274) + ... 92(-253)
149 + 196 149 + 196-1
=3287278
8. Calculate Z according to Eq. 9.4:

Z -6972/32872781/2 = 3.7461
From a normal distribution table, we find that p > Z = <0.0001, and so we

can reject the hypothesis that the average life spans in the two cohorts
are the same.
When more than two samples or cohorts are to be compared and the data
are not censored, the Kruskal-Wallis H test may be use in a manner analogous
to a one-way analysis ofvariance F-test. The H-test is a rank-based test, in which
the life spans in each treatment sample are ranked relative to all other roots in
the treatments. H is calculated as:
12 R2
I,-' -3(N+l),
K
H= (9.6)
N(N + 1) j=l nj
where N is the total number of roots in the K samples or treatments, nj is the
number of roots in the j th treatment, and Rj is the sum of all ranks in the j th
treatment.
When some of the ranks are tied, H, as computed above, is divided by the
value
1 g
1--3-~)j' (9.7)
N -N j=l
where g is the number of tied groups and Tj = tJ - tj, with tj equal to the number
of tied observations in a tied group (Lee 1992).
The number of roots (nj) in each sample need not be the same. If the number
t
is greater than or equal to five, H is distributed approximately as (Sokal and
Rohlf 1981). When the number of roots is less than 5, exact distributions can be
found in Kruskal and Wallis (1952). An numerical example with hypothetical
root life span data from three different plant species is shown in Box 9.4
When overall significant treatment effects are detected, as in this example,
multiple comparisons ofthe rank sums of each treatment can be made to iden-
tify particular pair-wise differences. The particular form of the test depends
upon whether the number of roots in each sample are equal, and upon the
number of roots in each sample. Lee (1992) has summarized four such tests,
one each for both large and small equal-sized samples; and large and small
unequal-sized samples.
When more than two (e.g. K) samples need to be compared in which some
of the data are censored, an alternative rank test (the K-sample test) can be used
in which the ranks are scored as in the Cox-Mantel test. Instead of one set of
ranks, two sets, one based upon ascending (R 1) and one based on descending
(R 2 ) scores are calculated, from which the difference between Rl and R2 is used
to compute a score Wi for each observation i. These values are used to calculate
a test X2 , which is assumed to be distributed approximately as chi-square, and
is computed as follows:
K S2
I,~
2 j=l nj
X =--2-' (9.8)
S
BOX 9.4. Kruskal-Wallis H-test for S pecies Differences

in Uncensored Root Life Span Data
Fine root life spans (days)
Species 1 Rank 1 Species 2 Rank 2 Species 3 Rank 3
45 4.5 21 1 58 8
55 7 27 2 64 10
67 Il 32 3 75 12
80 13 45 4.5 87 14
90 15 51 6 91 16.5
91 16.5 62 9 102 18
Sums 67 25.5 78.5
1. Rank the life spans in each species in ascending order, with ranks for
n tied scores equal to lIn of the combined ranks, had the scores not
been tied.
2. Compute rank sums (Rj ) for each species,
3. Compute the correction factor using Eq. 9.7. In the example, the
number of tied groups g is 2, and the nu,?ber of tied scores (tj ) in
both groups is 2 and the total number of scores N is 18. The
correction factor is then computed as:
=1- 13 ((2 3 -2)+(23 - 2))

(18) - 18
1
= 1- 5814 (12)
= 0.9979
4. Calculate the test statistic H according to Eq. 9.6:
-12- ((67)2
-- +-(25.5/
-+- (78.5)2) ()
- - 319
18(19) 6 6 6
H = - ---'---- - - - - - - - - < . . . . - -
0.9979
= 9.1138
t
The value at an a of 0.05 with 2 d.f. is 5.991 , and thus we can reject the
hypothesis that the root life spans of aH the species are the same.
where
K
LW;
2
S =i=1- - , (9.9)
N-1
N is the total number of root life spans in the sample, and w/ are the squared
scores for each life span i. The test statistic X2 is compared with the appropri-
ate chi-square value for K -1 degrees of freedom at the desired level of a. The
number of life spans in each sample need not be equal. A numerical example
using this test is given in Box 9.5.
BOX 9.5. K-Sample Test for Equivalent Root Life Spans for Three Plant
Species with Censored Observations
Fine root life spans (days)
Sp. 1 R'.il R2.il Wil Sp.2 R'.i2 R2.i2 W j2 Sp. 3 R'.iJ R2JJ w"
45 4 20 -16 21 1 24 -23 58 10 15 -5
55 8 16 -8 27 2 23 - 21 64 12 13 -1
67 13 12 32 3 22 - 18 75 14 10 4
71+ 14 1 13 45 4 20 - 16 87 15 9 6
90 16 8 8 51 6 19 - 13 91+ 18 1 17
90 16 6 8 54 7 18 -11 93 19 4 15
91 18 6 12 55 8 16 -8 96 20 2 18
94+ 20 19 62 11 14 -3 102+ 21 20
Sums 37 - 113 74
Consider the above hypothetical root life span data for three species of
plants. We wish to test the hypothesis that the life spans in aH three
species are the same, with an acceptable a level of 0.05.
1. Rank aH non-censored observations in ascending order (= R. for
each observation i).
2. AII censored observations receive a score which is one higher than
the adjacent, lower uncensored observation.
3. Reduce the rank scores of tied observations to the lowest value for
the score.
4. Rank ali observations (both censored and non-censored) in
descending order (= R2 for each observation i) .
5. Reduce the rank scores of tied observations to the lowest value for
the score.
6. Reduce the rank of alI censored observations to 1.

7. Calculate wij scores by subtracting R 2,ij from R1,ij and sum the scores
for each sample j.
8. Calculate the sum of w/ = 4352.
9. Compute the value S2:
4532
= (24-1 ) =197.0435.
10. Compute the test statistic X 2 :
(37)2 {_113)2 (74)2
--+ +--
888
197.0435
=12.4427.
The corresponding :t statistic for 2 dJ. and a = 0.05 is 5.9915, and we
can therefore reject the hypothesis that the root life spans of alI three
species are the same.
9.3.10 Survival Effects Models
In root longevity studies, the relationship between root life span and some envi-
ronmental or treatment factor, or an inherent property of the roots themselves,
may be of interest. The covariates presumed to be causalIy related to longevity
and survivorship falI into one of two classes: group-based, or individual-based.
Examples of the former might include irrigation level, soil temperature, or
species, or any other factor to which alI roots in a sample are exposed.
Individual-based covariates might include root diameter, length or colour. The
statistical models used to test these relationships are known as survival effects
models. The details of computing both group and individual-based survival
effects models are too complex to cover here. Moreover, they are too cumber-
some to do by hand. We will thus limit our discussion here to the general prin-
ciples behind the models' analyses, and some current statistical packages
capable of performing such analyses.
Tests of association for individual-based covariates are made with likeli-
hood ratio tests. Likelihood ratio tests are used to test the hypothesis that a
model incorporating the effects of one or more individual-based covariates
better describes root life spans (in terms of the maximum likelihood estimator
It) than a reduced model in which the covariate(s) has been removed. Since It
can be calculated for exponential, WeibulI, lognormal and gamma distributions,
and for both censored and non-censored data, likelihood ratio tests can be used
with a wide variety of data sets. Among survival analysis programs, SURPH
(Center for Quantitative Science, School of Fisheries, University of Washington,
Seattle, WA, USA) uses likelihood ratio tests in the construction of survival
effects models, and further details of the procedure can be found in the SURPH
operating manual and in Hoffmann (1993).
Lee (1992) describes Cox's proportional hazards regres sion model in which
the relationship between group-based covariates and survival times can be
examined. The model is non-parametric, and can be used with both censored
and non-censored data. The hazard function underlying the survival distribu-
tion of the sample serves as the dependent variable, and the covariate effects on
the hazard function are proportional to one another; that is, in a sample in
which the roots are exposed to one of two covariates, the effect of covariate 2
on root longevity is equal to the effect of covariate 1 multiplied by a constant
factor. To estimate the f3 values associated with each covariate in the regression
equation, a maximum likelihood procedure is used where the likelihood func-
tion is based upon a conditional probability of failure associated with each
covariate. The fJs in the model can be added and removed based upon their
effect in the presence of other covariates in a step-wise fashion in a manner
analogous to multiple linear regression using log-likelihood values.
Two other group-based analyses are available in statistical software pro-
grams. The LIFETEST procedure in the SAS (SAS Institute, Cary, NC, USA) uses
a rank test of covariate association with life spans that is based upon the rank
tests for comparing life span distributions discussed above. Rank statistics are
derived from the sum (across all observations) of the product of individual
observation rank scores and corresponding covariate values. Ranks are cor-
rected for censoring in one of two ways, depending upon whether the scores are
log or Wilcoxan values. Chi-square values for each covariate are generated in
the output, as is a stepwise regression model showing the relative improvement
in the fit of the model as covariates are added from most to least significant.
Further computational and output details can be found in the SAS/STAT user
guide. In addition, Allison (1995) has published an excellent manual describing
applications and examples of SAS's survival estimation procedures, including
suggestions based upon particular data types or desired tests.
The effects of group-based covariates for both single and multiple samples
of root life spans are evaluated via analysis of deviance (ANODEV) in the
program SURPH. ANODEV is somewhat analogous to an analysis of variance
(ANOVA). However, instead of being structured around decomposing sums of
squares in order to evaluate treatment effects as with an ANOVA, ANODEV is
based upon decomposing the discrepancy between the log-likelihood value
attained with a "maximum" model incorporating the effects of all covariates and
those models with one or fewer covariates, including the minimum or null
model. The deviance of a particular model is defined as twice the difference

between the model under consideration and the full model. The fit of an inter-
mediate model is assessed by how much of the discrepancy between the full and
null model it explains, accounting for the number of parameters or covariates
in the model in question. The deviance is distributed in an asymptotic X2
manner. Further details on the theory, computation and presentation of
ANODEV can be found in the SURPH users manual.
A newer public-domain program, MARK (Colorado State University,
Boulder, CO, USA, http://www.cnr.colostate.edu/-gwhite/mark/mark.htm) can
be used to construct models incorporating either group- or individual-based
variables. MARK generates survival probabilities for known-fate data (i.e. roots
whose fates are followed over time), and can partition variation in life spans
into time, age (i.e. different cohorts in the same treatment), treatment and other
attributable effects, and their interactions. MARK is based upon Cormack-Jolly-
Seber models (Schwarz and Arnason 1996) of survival probabilities, and gen-
erates AIC (Akaike's information criterion) scores, which are measures of model
parsimony (higher AIC scores = less parsimonious models). Likelihood ratio
tests are used to compare models (e.g. time x age x treatment vs. time x age).
Details, with an application to the effects of browsing on root lifespan, can be
found in Ruess et al. (1998).
9.4 Nutrient Fluxes and Root Longevity
Estimates of fine root longevity, length productivity and length mortality

derived from direct observations of fine root life history can be used to esti-
mate the amount of biomass and nutrients associated with these processes. A
general estimate of annual biomass and nutrient transfers to (and from) roots
on an area basis can be made by multiplying the inverse of fine root longevity
expressed in years (years-1) by mean annual root standing biomass or nutrient
crop (e.g. gm-2 ). For example, if average fine root longevity is 6 months (0.5
year) and biomass is 250gm-2 , annual biomass production (or mortality) can
be calculated as:
1
---x250gm-2 =500gm-2 year-1 •
O.5year
If mean fine root biomass is stable from year to year, biomass and nutrient allo-
cation to production can be assumed to be equivalent to mortality. However,
biomass would not be stable in an immature, developing ecosystem, and the
magnitude of inter-annual fluctuations in root biomass and nutrient content in
even mature systems are unknown. Even if root biomass and nutrients are stable
on an annual basis, several biomass estimates must be made throughout

the year, as they can fluctuate widely on an annual basis. Moreover, fine root
life spans can vary substantially throughout the year, and can dramatically
affect the rate at which the biomass or nutrients associated with any period
of productivity is lost to mortality during a particular time interval. For
example, sugar maple fine roots produced in the autumn can live for several
months longer than roots produced in the spring (Hendrick and Pregitzer
1993). The biomass and nutrients associated with spring root production are
therefore cycled much more rapidly than those associated with autumn root
senescence.
More accurate measures of biomass and nutrients allocated to root pro-
duction and returned through mortality for any period of time, or a series of
time periods, can be made by direct observations of individual fine roots using
some type of rhizotron. The procedure is as follows.
1. With a (mini)rhizotron or other direct observation technique, quantify fine

root length at time t = O, and then periodically thereafter.
2. Sample root biomass and/or nutrient content at time t = O concurrent with
the length sampling, and then once more at some subsequent sampling date,
preferably when root length is either known or suspected to be greater or
less than it was at time t = O.
3. For each successive time period, fine root production and mortality are cal-
culated as a proportion of fine root length at time t = O, as follows:
length productionttot~l proportional root production

lengtht~ for tiIne period t to t + 1
4. Multiply proportional production based upon length by biomass or nutrient

content at time t = O.
5. Check accuracy by adding production estimates and subtracting mortality
estimates for intervals between initial and final biomass samplings, and by
comparing predictions of second biomass observations with actual mass
determinations.
Successful application of this approach is based upon several assumptions: (l)

the dynamics of fine root biomass and length within a narrow diameter class
mimic one another in magnitude, (2) length densities along the (mini)rhizotron
have stabilized (i.e. no further post-installation recolonization), and (3) when
using the approach to calculate nutrient flux, there is no retranslocation of the
nutrient in question from dying roots. Hendrick and Pregitzer (l993) showed
that the technique can be used to accurately calculate root biomass dynamics,
but the extent to which assumption (3) is true has yet to be proved or disproved
conclusively.
9.5 The Future
We believe that the prognosis for continued and improved use of observational
techniques to study fine root longevity is good. With respect to data collection,
continuing improvements in video technology are constantly being made in
consumer products (e.g. video camcorders) and it is these developments that
underpin the advances we are able to make in camera-based research tools.
More specialized instruments, like the Bartz Technology (Santa Barbara, CA,
USA) minirhizotron cameras are also improving. The Bartz carne ras are now
available in two sizes (2.86 and 5.1 cm), and options include a IOOx zoom func-
tion as well as a UV light source. Recent developments by Bartz Technology will
allow direct recording of root imagery onto digital media in the field, bypass-
ing the cumbersome and quality-degrading digital-analog-digital conversions
associated with image acquisition (digital camera), collection (analog video-
tape) and analysis (image redigitization).
Further developments and refinements in image analysis tools are also
being made. ROOTS (Michigan State University, East Lansing, MI, USA), the first
commercially available program developed for analyzing minirhizotron data, is
in the early stages of a much-needed update. RooTracker (Duke University,
Durham, NC, USA), a Macintosh-based program, also continues to be refined.
Scientists working with the Desert Research Institute (Reno, NV, USA) have
developed a more automated system (ARCOS) that is also undergoing refine-
ment based upon user feedback. Scientists at the Rhizolab (Wageningen,
Netherlands) use a public-domain program (NIH-Image, Macintosh or PC)
developed by the US National Institute of Health (http://rsb.info.nih.gov/nih-
image). Finally, the development and application of new software programs (e.g.
MARK, SURPH) and upgrades of existing software (e.g. SAS) facilitate the
design and analysis of more complicated experiments and the extraction of
more information about the dynamics of roots and root systems.
The population of scientists studying root longevity and related demo-
graphic processes is relatively small, and time and monetary investments into
improving our ability to study roots are likely to be also. However, we do stand
to benefit tremendously from continuing improvements in consumer electron-
ies and the importance of demographic statistical and numerical computational
techniques to a wide array of other disciplines, including forestry, wildlife man-
agement, medicine and pharmaceuticals, and the insurance industry. Our task,
then, is perhaps not so much to develop novel techniques, but rather to adopt
or adapt novel means of applying existing, evolving or new techniques and tech-
nology to the study of roots.
References
Allison PD (1995) Survival analysis using the SAS system: a practical guide. SAS Institute, Cary,
North Carolima, 292 pp
Andren O, Rajkai K, Katterer T (1991) A non-destructive technique for studies of root distri-
bution in relation to soH moisture. Agric Eco Environ 34: 269-278
Atkinson D (1983) The growth, activity and distribution of the fruit tree root system. Plant SoH
71: 23-37
Atkinson D (1985) Spatial and temporal aspects of root distribution as indicated by the use of
a root observation laboratory. In: Fitter AH, Atkinson D, Read DJ, Usher MB (eds) Eco-
logical interactions in soH. Blackwell, Oxford, pp 43-65
Atkinson D (1986) The nutrient requirements of fruit trees: some current considerations. Adv
Atkinson D, Fogel R (1997) Roots: measurement, function and dry matter budgets. In:
Vangardingen FR, Foody GM, Curran PJ (eds) Scaling-up from cell to landscape. Cam-
bridge University Press, pp 151-172
Berta G, Trotta A, Fusconi A, Hooker JE, Munro M, Atkinson D, Giovannetti M, Morini S, Fortuna
P, Tisserant B, Gianinazzi-Pearson V, Gianinazzi S (1995) Arbuscular mycorrhizal induced
changes to plant growth and root system morphology in Prunus cerasifera. Tree Physiol
5:281-294
Black KE, Harbron CG, Franklin M, Atkinson D, Hooker JE (1998) Differences in root longevity
of some tree species. Tree Physioll8: 259-264
Cannell MGR, Dewar RC, Thornley JHM (1992) Carbon flux and storage in European forests.
In: Teller A, Mathy P, Jeffers JNR (eds) Responses of forest ecosystem to environmental
changes. Elsevier, London, pp 256-271
Conover WJ (1980) Practical nonparametric statistics, 2nd edn. John Wiley, New York
Cox DR, Oakes D (1984) Analysis of survival data. Chapman and Hall, London
Curtis PS, Zack DR, Pregitzer KS, Teri JA (1995) Above and below ground response of Populus
grandidentata to elevated atmospheric CO 2 and soH N availability. Plant SoH165: 45-54
Drew MC (1987) Function of root tissue in nutrient and water transport. In: Gregory PJ, Lake
JV, Rose DA (eds) Root development and function. Cambridge University Press, Cambridge,
pp 71-102
installed minirhizotrons. Agron J 87: 258-263
Fairley RI, Alexander II (1985) Methods of calculating fine root production in forests. In: Fitter
AH, Atkinson D, Read DJ, Usher MB (eds) Ecological interactions in soH. Blackwell, Oxford,
pp 37-42
Fogel R (1985) Roots as primary producers in below-ground ecosystems. In: Fitter AH,
Atkinson D, Read DJ, Usher MB (eds) Ecological interactions in soH. Blackwell, Oxford, pp
23-36
Fogel R (1991) Root system demography and production in forest ecosystems. In: Atkinson D
(ed) Plant root growth. Blackwell, Oxford, pp 89-102
Fogel R, Hunt G (1979) Fungal and arboreal biomass in a western origin Douglas fir ecosystem:
distribution patterns and turnover. Can J For Res 9: 245-256
Forbes PJ, Ellison CH, Hooker JE (1996) The impact of arbuscular mycorrhizal fungi and tem-
perature on root system development. Agronomie 159: 617-620
Forbes PJ, Black KE, Hooker JE (1997) Temperature induced alterat ion to root longevity of
Lo/ium perenne. Plant Soi1190: 87-90
Gehan EA (1965) A generalized Wilcoxan test for comparing arbitrarily singly-censored data.
Biometrika 52: 650-653
George SL, Desu MM (1974) Planning the size and duration of a clinical trial studying the time
to some critic al event. J Chron Dis 27: 15-24
Goins GD, Russelle MP (1996) Fine root demography in alfalfa (Medicago sativa L). Plant Soil
185:281-291
Gregory PJ (1987) Development and growth of root systems in plant communities. In: Gregory
PJ, Lake JV, Rose DA (eds) Root development and function. Cambridge University Press,
Cambridge, pp 147-166
Harris WF, Santantonio D, McGinty D (1980) The dynamic below ground ecosystem. In: Wearing
RH (ed) Forests: fresh perspectives from ecosystem analysis. pp 119-129
Head GC (1966) Estimating seasonal changes in the quantity of white unsuberized root on fruit
trees. J Hortic Sci 41: 197-206
Hendrick RL, Pregitzer KS (1992) The demography of fine roots in a northern hardwood forest.
Ecology 73: 1094-1104
Hendrick RL, Pregitzer KS (1993) Patterns of fine root mortality in two sugar maple forests.
Nature 361: 59-61
Hoffman A (1993) Quantifying selection in wild populations using known-fate and mark-
recapture designs. PhD Dissertation, University of Washington, Seattle, Washington, 259 pp
Hooker JE, Atkinson D (1996) Arbuscular mycorrhizal fungi-induced alteration to tree-root
architecture and longevity. Pflanzenernaehr Bodenkd 159: 229-234
Hooker JE, Munro M,Atkinson D (1992) Vesicular-arbuscular fungi induced alterat ion in poplar
root system morphology. Plant Soil145: 207-214
Hooker JE, Black KE, Perry RL,Atkinson D (1995) Arbuscular mycorrhizal fungi induced alter-
ation to root longevity of poplar. Plant Soil 172: 327-329
Kalbfleisch JD, Prentice RL (1980) The statistical analysis of failure time data. John Wiley, New
York
Klepper B (1987) Origin, branching and distribution of root systems. In: Gregory PJ, Lake JV,
Rose DA (eds) Root development and function. Cambridge University Press, Cambridge,
pp 103-124
Kruskal WH, Wallis WA (1952) Use of ranks in one-criterion variance analysis. J Am Stat Assoc
47:583-621
Lee ET (1992) Statistical methods for survival data analysis, 2nd edn. Wiley, New York, 482 pp
MacKenzie KAD (1979) The development of the endodermis and phi layer of apple roots. Pro-
toplasma 100: 21-32
Makuch RW, Simon RM (1982) Sample size requirements for comparing time-to-failure among
k treatment groups. J Chron Dis 35: 861-867
Mantel N (1966) Evaluation of survival data and two new rank order statistics arising in its con-
sideration. Cancer Chemother Rep 50: 163-170
Mantel N (1967) Ranking procedures for arbitrarily restricted observations. Biometrics 23:
65-78
Muenchow G (1986) Ecological use of failure time analysis. Ecology 67: 246-250
Nadelhoffer KJ, Aber JD, Melillo JM (1985) Fine roots, net primary production and nitrogen
availability: a new hypothesis. Ecology 66: 1377-1390
304 J.E. Hooker et al.: The Measurement and Analysis of Fine Root Longevity
Norman JR, Atkinson D, Hooker JE (1996) Arbuscular mycorrhizal fungal-induced alteration to

root architecture in strawberry and induced resistance to the root pathogen Phytophthora
Jragariae. Plant Soil185: 191-198
Pollock KH, Winterstein SR, Conroy MJ (1989) Estimation and analysis of survival distributions
for radio-tagged animals. Biometrics 45: 99-109
Pregitzer KS, Hendrick RL, Fogel R (1993) The demography of fine roots in response to patches
of water and nitrogen. New Phytol 125: 575-580
Pyke DA, Thompson JN (1986) Statistical analysis of survival and removal rate experiments.
Ecology 67: 240-245
Ruess RW, Hendrick RL, Bryant JP (1998) Regulation of fine root dynamics by mammalian
browsers in early successional Alaskan taiga forests. Ecology 79: 2706-2720
Schwarz q, Arnason AN (1996) A general methodology for the analysis of capture-recapture
experiments in open populations. Biometrics 52: 860-873
Sokal RR, Rohlf FJ (1981) Biometry, 2nd edn. WH Freeman, New York
Steen E (1991) Usefulness of the mesh bag method in quantitative root studies. In: Atkinson D
(ed) Plant root growth. Blackwell, Oxford, pp 75-88
Thomas SM, Whitehead D, Adams JA, Reid JB, Sherlock RR, Leckie AC (1996) Seasonal root dis-
tribution and soii surface carbon fiuxes for 1-year-old Pinus radiata trees growing at
ambient and elevated carbon dioxide concentrations. Tree Physiol16: 1015-1021
Wang Z, Burch WH, Mou P, Jones RH, Mitchell RJ (1995) Accuracy of visible and ultraviolet light
for estimating live root proportions with minirhizotrons. Ecology 76: 2330-2334
CHAPTER 10
Root Image Analysis and Interpretation

W. Richner, M. Liedgens 1, H. Burge, A. Soldati, and P. Stampl
1 ETH Ziirich, Institute of Plant Sciences, ETH Zentrum, LFW A4, 8092 Ziirich, Switzerland
2 Swiss Federal Office of Information Technology, Systems, and Telecommunication (FOITT),
Monbijoustrasse 74,3003 Bern, Switzerland
CONTENTS
10.2 Image Analysis of Washed Root Samples 306
10.2.1 Preparation of Samples 306
10.2.1.1 Mechanical Separation of Roots from Extraneous Objects 306
10.2.1.2 Staining Roots 307
10.2.1.3 Spreading of Roots in the Recording Tray 309
10.2.2 Image Acquisition 310
10.2.2.1 Recording Devices 310
Cameras 312
Scanners 313
10.2.2.2 Storage of Images 314
10.2.3 Image Analysis Techniques 317
10.2.3.1 Segmentation of Images ofWashed Roots 317
Grey-Level Thresholding 317
Alternative Segmentation Techniques 318
Editing of Segmented Images 319
10.2.3.2 Extraction of Morphological Information from Binary Images
of Washed Roots 320
Line-Intercept vs. Chain Methods 320
Determination of Morphological and Architectural Root
Parameters 320
Discrimination of Roots from Extraneous Objects 327
Differentiation Between Viable and Dead Roots 327
10.2.4 Root Image Analysis Systems 328
10.2.5 Priorities for Image Analysis of Washed Roots 329
10.3 Image Analysis of Minirhizotron Images 330
10.3.1 Segmentation of Minirhizotron Images 330
10.3.1.1 Enhancement of Minirhizotron Images 331
10.3.1.2 Segmentation of Minirhizotron Images 332
10.3.2 Extraction of Information from Binary Minirhizotron Images 334
10.3.2.1 Determination of Morphological and Architectural Root
Parameters 334

306 w. Richner et al.
10.3.2.2 Discrimination of Roots from Extraneous Objects 335

10.3.3 Priorities for Image Analysis of Minirhizotron Images 335
Further Reading 336
References 337
10.1 Introduction
For the description and modelling of root and shoot growth, quantitative infor-
mation on root morphology and architecture has to be gathered in ecological,
agronomic, biological and physiological studies of soil exploration and water
and nutrient uptake by root systems. In the past, root length and surface are a
of washed samples were measured manually, using linear measurement devices
(B6hm 1979), solution adsorption methods (Sattelmacher et al. 1983) or line-
intercept techniques (Newman 1966; Tennant 1975). In more recent years, image
analysis systems have become available for these measurements. Compared
with traditional methods, image analysis is faster, more accurate and less prone
to human error; with these techniques it is also possible to measure morpho-
logical parameters other than the totallength of root samples.
Contrary to classical destructive root investigations, the rhizotron and
minirhizotron techniques allow the nondestructive monitoring of root dynam-
ics in situ. Due to the complexity of root images recorded through transparent
viewing surfaces against the soil, they have been analysed manually so far.
However, automated or at least controlled image analysis techniques are
necessary for measuring other morphological root parameters on images taken
in situ, owing to the large number of pictures that have to be recorded in
rhizotron or minirhizotron studies and the limited amount of informat ion that
can be obtained by manual counting or measuring.
This chapter will present state-of-the-art image analysis of washed root
samples and minirhizotron or rhizotron recordings, describe important steps
in the processing of root samples and discuss research needs in these fields.
10.2 Image Analysis of Washed Root Samples
10.2.1 Preparation of Samples
70.2.7. 7 MechanicalSeparation of Roots from Extraneous Objects
When root samples are washed free from bulk soil, they generally contain soil
aggregates, sand grains, plant residues (remains of shoots and roots of
previous crops or weeds), bodies and eggs of soil animals, as well as weed
10 Root Image Analysis and Interpretation 307
and crop seeds. The kind and amount of such unwanted objects left in the root
samples depends on the soi! type, the crop rotation, the way in which the roots
are washed (see Chap. 6) and especially on the mesh size used for collecting
washed roots. Automated washing systems allow for smaller mesh sizes. Thus,
root samples washed through finer sieves generally contain more fine roots
(Livesley et al. 1998), but also more extraneous objects than samples washed
manually with coarser sieves. Therefore, mechanical separation of roots from
extraneous objects usually precedes measurements of root morphology.
Decanting or flotation techniques (Schuurman and Goedewaagen 1971) can
separate roots from unwanted objects that have a specific weight markedly
different from that of the roots, such as soi! and sand fractions and heavy seeds
of weeds. Murphy and Smucker (1995) separated the majority of extraneous
objects from roots by repeated decanting of the organic debris that floated
more readily than roots. Subsequently, tweezers are used to manually extract
some or all of the remaining extraneous objects. lf an image analysis system
were able to discriminate between roots and extraneous objects according to
differences in their shapes, the manual removal process could be less meticu-
lous and could concentrate on removing large extraneous objects, which
might cover roots partly or completely, and on residue parts that have a shape
similar to that of root objects. lf measurement techniques are employed
which cannot discriminate between roots and unwanted objects, then the use
of tweezers to remove extraneous objects should be done as thoroughly as
possible; remaining objects that have a size above the lower limit of spatial
resolution of the image acquisition device will be detected as roots, and thus
contribute to an overestimation of the total root length. Cleaning roots by hand
is very tedious. Thus, image analysis systems should be able to discriminate
between roots and unwanted objects (see Sect. 10.2.3.2) if these systems are to
be truly advantageous.
10.2.1.2 Staining Roots
The purpose of staining is to enhance the contrast of a root sample. Especially

when using simple thresholding techniques (see Box 10.1), a sharp contrast
between roots and the background is essential. Because roots are often recorded
against a Iight background, they must usually be stained to allow for sufficiently
contrasting images. A variety of dyes of different colours have been used for
staining washed roots (Table 10.1). No general recommendations on the use of
dyes can be given, because staining depends on root morphological and chem-
ical properties. These may be influenced by the plant species, plant age and the
environment in which the plants grew. Because some dyes are toxic to humans,
non-toxic dyes should be used or gloves should be worn when handIing toxic
dyes and stained samples.
308 W. Richner et al.
BOX 10.1. Image Segmentation

Image segmentation or discrimination is the process of dividing grey-Ievel
images into regions or objects of interest for measurements or recognition
(Russ 1995), e.g. into plant roots and background. Segmentation produces
binary (i.e. black-and-white) images which can be used for further analy-
ses. The most important segmentat ion technique is grey-Ievel threshold-
ing, but there are numerous alternative, more sophisticated segmentation
methods (Fu and Mui 1981; Haralick and Shapiro 1988; Glasbey and
Horgan 1995) that are beyond the scope of this chapter.
Grey-Ievel thresholding is very simple and easy to implement.1t
separates regions of an image into a set of c1asses, usually two, according
to one or two threshold grey-Ievels. In other words, thresholding
establishes which range of grey-Ievels belongs to the objects and which
to the background. When using grey-Ievel thresholding for the
segmentation of images, the most critic al step is setting the threshold.
In an ideal situation, object pixels are restricted to a narrow range of
grey-Ievels, the remaining grey-Ievels belonging to the background. In
this case it is very easy to visually set the threshold value. With real-life
images, the brightness histogram of the image can help to reliably
determine the threshold value. In the case of a bimodal brightness
histogram, the grey-Ievel threshold is chosen from the values between
the two peaks (Castleman 1979; Lebowitz 1988). This is, however, not
always consistent from one operator to another (Russ 1995), therefore
a variety of automated thresholding techniques have been developed,
which will not, however, be discussed here.
Roots are usually soaked in the dye solution from a few hours to a few days,
depending on the dye, until staining is complete. After staining, roots of alI
diameter c1asses should be sufficiently and evenly stained; if not, the staining
time may have been too short. Thicker roots sometimes require longer stain-
ing, depending on the plant species or dye. It is not always possible to achieve
complete staining of the whole sample. Thus, several dyes may have to be tested
to determine optimal staining of roots for a given species and plant age. To
prevent a degradat ion of samples, root samples should be placed in a refriger-
ator or cooling room during sta in ing at temperatures slightly above freezing
(usually around 4°C). Prior to the spreading of the roots for recording, they
must be removed from the dye solution with a s ieve whose mesh size is smaller
than the smallest root objects. Then the dye adhering to the roots in the sieve
should be rin sed off with water to avoid staining the scanning tray and the water
Table 10.1. Dyes for staining washed roots to enhance the con-
trast between roots and a light background
Dye Authors
Acid fuchsin Barnett et al. (1987)

Aniline blue Barnett et al. (1987)
Coomassie Brillant Blue Tanaka et al. (1995)
Crystal violet Wilhelm et al. (1983); Kaspar
and Ewing (1997)
Dylon stain Murakami and Yoneyama (1988)
Fluorescein diacetate McGowan et al. (1983)
Malachite green oxalate Murphyand Smucker (1995)
Methyl violet Harris and Campbell (1989)
Methylene blue Sackville Hamilton et al. (1985);
Smit et al. (1994)
Trypan blue Collins et al. (1987)
in which the roots will be placed. If this were to occur, then the contrast between
root boundaries and the background would be strongly reduced.
Contrast can be enhanced by choosing an appropriate background. If white
unstained roots are to be measured, a duH dark background is desirable. We
found that duH matt black velvet is a suitable background under a transparent
scanning tray for recording unstained maize roots, because the rough surface
of the velvet does not reflect light to the same extent as smooth objects painted
black. On the other hand, stained roots should be measured against a light back-
ground. Light boxes or light tables are sometimes placed under a transparent
scanning tray; the emitted light is often diffused by means of a milky glass plate,
such as white opal glass (Barnett et al. 1987).
10.2.1.3 Spreoding of Roots in the Recording Troy
Roots are usually spread out in shaHow transparent trays that aHow for back-lit
camera recording from above or for recording with flatbed scanners from
below. Glass trays are generally preferred to plastic materials, because they are
more resistant to scratches, which may be resolved as false roots. The trays are
usually filled with 2 to 3 mm of water. In this way, roots may be easily separated,
and yet are prevented from floating around in the tray. Because the slightest
movement of the tray will cause smaller root parts to move around, the final
spreading should preferably be done at the place of recording.
Roots should be spread out in the scanning tray so that individual roots do
not overlap, thus causing root length to be underestimated and root diameters
and branching to be overestimated. In addition, roots, especially neighbour

lateral roots, should not stick together after spreading. If the distance between
two parallel roots is in the range of a pixel (short for picture element) of the
recording device, then the two roots may be detected as one thick root, thus
overestimating the mean root diameter and underestimating the total root
length (Pan and Bolton 1991) and branching. It can be advantageous to cut
longer root segments into shorter pieces to ease spreading and to reduce under-
dispersion across the recording are a (Reicosky et al. 1970) if root samples are
large and highly branched. Because many techniques for determining morpho-
logical parameters are quite sensitive to orientation of the objects in the images
(see Sect. 10.2.3.2), roots should be arranged as randomly as possible to achieve
a uniform dis tribut ion of orientations of root segments.
Depending on the size and the type of sample, however, spreading can be
very time-consuming. Therefore, a compromise between accuracy of spreading
and time requirement must be found.
10.2.2 Image Acquisition
10.2.2. 1 Recording Devices
The most frequently used devices to record images of washed root samples for
image analysis are electronic cameras and scanners (Fig. 10.1), both based on
CCD (charge coupled device) sensors. Other devices, such as densitometers
(Voorhees et al. 1980) or photocopiers (Collins et al. 1987), are no longer very
important because of generally limited spatial resolutions or their intricate
Fig. 10.1. Stained roots spread out in a

transparent tray on top of a desktop scanner.
(Photo with permission from U. Zimmermann,
Institute of Plant Sciences, ETH Ziirich)
Table 10.2. Sizes of black-and-white, 8-bit grey-Ievel and 24-

bit colour images of A4 dimension (21.0 x 29.7 cm) scanned at
resolutions of 150,300,600 and 1200 dpi'
Image type Resolution (dpi)
150 300 600 1200
File size (MB b)

Black-and-white 0.3 1.0 4.1 16.6
(single-bit)
256 grey-Ievels (8-bit) 2.1 8.3 33.2 132.8
16.7 million colours 6.2 24.9 99.6 398.3
(24-bit)
, Dots per inch.

b Megabytes.
handling. Imaging devices should produce images of roots which contrast

strongly with the background at a spatial resolution high enough to resolve the
thinnest roots in a sample. Thus, the critical aspects of image acquisition devices
are their radiometric and spatial resolutions (Glasbey and Horgan 1995).
The radiometric resolution is the maximum number of brightness levels
that may be recorded. In an image, a value of brightness is associated with each
pixel. For grey-scale images this value frequently ranges from O to 255 (8-bit
grey scale images); O usualIy stands for black and 255 for white. For most bio-
logical applications, between 16 and 256 grey-Ievels are considered sufficient
(Glasbey and Horgan 1995). Most of the modern CCD cameras and scanners
alIow the recording of coloured images, but almost alI present root image analy-
sis systems stiH rely on monochrome information only, owing to the large file
sizes of digitised colour images (Table 10.2).
The spatial resolution is given by the number of pixels per unit scanning
are a, in the case of scanners usualIy expressed as dpi (dots per inch). It deter-
mines the size of the smallest objects that can be recorded. Recent research has
shown that 56% of the total root length of field-grown maize consisted of roots
thinner than 0.175mm (Pallant et al. 1993). Roots thinner than 0.5 mm may con-
stitute up to 80% of the total root length of a barley root system (Hackett 1968).
Due to the significant proportion of fine roots in most root systems, the
maximum spatial resolution of a recording device should alIow for the record-
ing of the thinnest roots in a sample. Roots with diameters close to one pixel
length may cause problems in thresholding (see Box 10.1) of root images (Smit
et al. 1994), i.e. the division of them into roots and background. As a conserva-
tive rule of thumb, the diameter of the thinnest roots in a sample should be at
Table 10.3. Smallest root diameters that can be theoretically

and practically resolved at scanning resolutions of 150,300,600
and 1200dpi." The calculation of the practically smallest root
widths is based on minimal root widths of three pixels, which is
required by most image analysis algorithms for reliably mea-
suring root diameters
Scanning resolution (dpi)
150 300 600 1200
Resolved root diameter (Ilm)

Theoretical 169 85 42 21
Practical 508 254 127 64
" Dots per inch.
least about three times the pixel size of the recording device to ensure a reli-
able detection of roots and an accurate measurement of root diameter (see Sect.
10.2.3.2). Thus, the practical maximum resolution of an imaging device is some-
what lower than the theoretical maximum resolution (Table 10.3).
Cameras. Electronic cameras turn analogue voltage signals into digital signals
using analogous-to-digital (A/D) converters, such as computer frame grabbers,
whereas digital cameras and scanners do not require such converters. The main
advantage of cameras is their short scanning time and, in the case of CCD
cameras, their high spatial resolution. While the outdated tube cameras often
had optical resolutions too limited to detect higher-order lateral roots
(Voorhees et al. 1980; Lebowitz 1988; Cunningham et al. 1989; Harris and Camp-
bellI989), newer CCD cameras usually have typical spatial resolutions of about
700 x 500 pixels. More expensive CCD cameras have superior resolutions of up
to 5000 x 5000 pixels (Krumenaker 1994).
By keeping scanning areas relatively small, however, the necessary spatial
resolutions for recording fine roots may also be achieved with cheaper
cameras. Especially with low-resolution cameras, however, this may lead to
very small recording areas. Because preparing and spreading roots in a
number of small scanning trays is more time-consuming than using a larger
tray covering the same area, recording samples may be optimised by using
robotic cameras that are mounted on an X- Y moving gantry (Murphy
and Smucker 1995). This allows a large scanning area to be divided into several
sub-images of constant dimension, which are recorded automatically one
after the other. Thus, spatial resolution may be kept sufficient1y large for the
individual sub-images. Additionally, optical distortion at the boundaries of
the recording area is reduced when scanning areas are smaller compared
to recording larger areas with a greater distance between the camera and
the tray. However, some of the recorded objects may be truncated by the
sub-image borders, thus dividing a root into two parts. This is a potential
problem when roots are discriminated from extraneous objects by shape
indices (see Sect. 10.2.3.2), because the truncated root segments may no
longer meet the shape requirements of roots, and thus would be eliminated
as extraneous objects, leading to an underestimation of total root length and
area.
A general problem with camera-recording is achieving uniform lighting of
the scanning area. Samples are lit from the top and/or sides using incandescent
or speciallamps, or from the back using light trays. Lighting from above using
lamps may produce shadows, especially of thick roots, and extraneous light may
infiuence the recording. Therefore, it is usually preferable to back-light the
objects. In this way roots can be recorded as dark objects before a light back-
ground. A milky glass plate is usually used with light-boxes to produce diffused
light. To prevent refiections from the room lighting or sunlight from windows,
shields or boxes are sometimes used to prevent extraneous light from reaching
the scanning area. To reduce a further source of error, Russ (1995) suggests
avoiding automatic gain or brightness circuits that are available in some types
of standard cameras, because they may change the image contrast or linearity
in response to bright or dark regions that do not even lie within the digitised
portion of the image.
Scanners. Like CCD cameras, scanners use a solid-state detector array for
recording images, but in contrast to cameras, this array is linear only and moves
along the scanning area for recording (Russ 1995). Scanners have been used for
recording washed root samples that are spread in transparent trays (Krstansky
and Henderson 1989; Pan and Bolton 1991; Smit et al. 1994). These are generally
standard fiatbed scanners, which are primarily designed to scan fiat objects,
but there are also special 3-D scanners that have a depth of resolution of
several centimetres (Smit et al. 1994). Roots are usually spread in 2 to 3mm
of water; therefore, the depth of resolution of standard flatbed scanners is
usually sufficient for accurate recording of root samples (Kirchhof and
Pendar 1993).
Desktop scanners provide high quality optical characteristics for recording
images, thus enabling accurate and precise estimates of root length and other
morphological root parameters (Pan and Bolton 1991). A major advantage
is their high spatial resolution, which is up to 600dpi even with mid-Ievel
scanners, which is often better than the resolution of cameras (Smit et al. 1994).
Only the optical resolution of scanners should be considered, because the higher
interpolated resolutions that are given by many scanner manufacturers do not
really mean a higher spatial resolution. In addition, lighting problems are

reduced, compared with camera-based systems, although light intensity may be
lower near the boundaries of the scanning area (Russ 1995). To reduce uneven
lighting, the warm-up time of the scanner should be long enough for the light
source to become stable (Russ 1995). Scanners may also cast shadows around
the roots, although usually to a lesser extent than when a light source is placed
above the root tray for the recording of roots with a top-mounted camera. To
reduce this problem, scanned root samples can be simultaneously lit from below
and from above (Bauhus and Messier 1999) using a transparency adapter unit
in place of the scanner lid.
A disadvantage is that scanners are much slower at capturing an image than
are CCD cameras with a fixed scanning area (Berntson 1992), but the time
required for recording with a scanner is stiH much shorter than for spreading
a root sample. On the other hand, camera-based systems mounted on an X- Y
moving gantry may take even longer to record a scanning area that is divided
into a large number of sub-images, although its total size is comparable to that
of a desktop scanner.
Using conventional mid-Ievel desktop scanners with an optical resolution
of 600 dpi, roots as thin as about 130/.1m can be resolved (Table 10.3), provided
that the contrast between roots and the background is sharp and uniform. This
resolution results in image file sizes from 4 to over 30 megabytes (MB) when
scanning A4-sized black-and-white or 256-grey-Ievels images, respectively
(Table 10.2). If lateral roots with diameters of about 50 ţlm are to be measured,
which is in the range of the thinnest reported roots (Pallant et al. 1993),
scanners with a spatial resolution of at least 1200dpi must be used (Table 10.3).
This, however, results in huge image files, e.g. more than l30 MB with 256
grey-Ievels (Table 10.2). This is beyond the processing capability of most
PC-based image analysis programs for measuring roots. To overcome this
limitation, it can be advantageous to split up such large images after recording
into several smaller images or, alternatively, only record parts of the whole
scanning area at once.
10.2.2.2 Storage of Images
While older root image analysis systems (e.g. "Delta-T Mk II", Delta-T Devices,
Burwell, UK; Table 10.4) were designed for real-time processing of root samples,
with neither necessity nor possibility of storing digital images, the majority of
newer systems require the creation of digital images prior to processing. For
most projects involving measurements of morphological traits of washed root
samples, it is worthwhile storing the recorded images at least until the end of
the project so that samples can be remeasured if necessary. Furthermore,
Table 10.4. Overview of widely-used root image analysis systems ...
Q
::1:1
O
System Platform Principle Individual object Major root parameters provided Manufacturerl programmer O
measurements
..3
1»
11:1
Delta-T Mark II Modified area Line-intercept method (Harris and No Length, area, mean diameter Delta-T Devices Ltd., Burwell, ID
meter Campbelll989) (derived from length and area Cambridge, UK

,.
:::II
1»
measurements) -<
VI
iii'
Delta-T SCAN DOS Modified length-intercept method Yes, but not in Length (grouped into user- Delta-T Devices Ltd., Burwell, 1»
:::II
(Kirchhof and Pendar 1993) length- definable diameter dasses), Cambridge, UK Do
measurement area, number of root tips (http://www.delta-t.co.uk) ;-
mode ID
...
'ti
NIH Image 1.62 Macintosh' Macro-based multipurpose system Yes Multipurpose system W. Rasband, National Institutes of ID
1»
(Rasband and Bright 1995); used Health (NIH), Bethesda, MD. USA
......
O·
for root studies by several groups (public domain software; :::II
(e.g. Smit et al. 1994; Dowdy et al. http://rsb.injo.nih.gov/nih-image)
1995; Tanaka et al. 1995)
ROOTEDGE 2.3 DOS Edge chord algorithm (Ewing and Yes Length, width. perim eter R.P. Ewing and T.C. Kaspar.
Kaspar 1995; Kaspar and Ewing USDAIARS National Soi! Tilth
1997) Laboratory, Ames, IA, USA (public
domain software:
http://www.nstl.gov/sojtware/rootedgel)
WinRHIZOI Windowsl Skeletonisation method (Bauhus Yes Length (grouped into user- Regent Instruments Inc., Quebec,
MacRHIZO Macintosh and Messier 1999) definable diameter dasses), Canada
area, diameter, number of (http://www.regent.qc.ca)
root tips and root forks,
topology and fractal analysis,
colour analysis (see Chap.4)
, "Scion Image", a Windows version of"NIH Image", is available from Scion Corporation, Frederick, MD, USA (http://www.scioncorp.com). ,...
""
V1
images can be analysed for additional morphological parameters later when

new algorithms become available.
If images have to be analysed using several computer programs, or if they
are exchanged with co-operating laboratories, a compatible image format
should be chosen. There are a variety of image file formats, but only a few of
them are supported by various computer platforms and programs. The most
widely supported bitmap file format is the TIFF (tagged image file format)
format. TIFF is capable of handling grey-Ievel and colour data and has the
ability to store data in a compressed form. There are, however, a variety of
options for TIFF files, so that a TIFF file created by one program may not be
readable by another (Russ 1995). On the other hand, there are inexpensive pro-
grams for both personal computers (PCs) and Macintosh computers for easily
converting images between different file formats.
As shown in Table 10.2, increasing the spatial and radiometric resolutions
leads to large images, which require more space for storage. This can amount
to several gigabytes of storage capacity, depending on the size of a research
project. Image compression can reduce the amount of storage space, and,
ultimately, the cost, but a discussion of compression techniques is beyond
the scope of this handbook. See Ja~n (1989) for more information on image
compression.
Data are usually stored on tloppy disks, computer hard disks, tapes
and various optical storage devices. Because of their limited capacity, genuine
tloppy disks with a capacity of below 2 MB are not suited for storing images.
Although hard disks with a capacity of 30 gigabytes and more have become
much less expensive, they are still too expensive for the long-term storage
of large amounts of stored information as compared with other media. Hard
disks have the advantage of the fastest access times compared with other storage
media. On the other hand, the storage cost per MB is still too great and the
lifetime of data is short compared with other storage media. Although some-
what slower than hard disks, optical storage media still have relatively short
access times, but like hard disks they are comparatively expensive per amount
of stored information. Compact disc writers are now less expensive and
offer a good way of storing up to 650 MB of data permanently on inexpensive
CD-ROM or temporarily on CD-RW media. Tapes of a variety of standards seem
to be an easy and inexpensive storage medium for large numbers of digital root
images. The disadvantage is that the access time for individual files is much
longer than for hard disks because of sequential access of data. Thus, images
should be grouped logically prior to storage on tape, according to the way they
will be retrieved later for analysis.
Similar to backup procedures for alI kinds of data, it should be kept in mind
that image files must always be saved on a second medium, which should be
stored at a place other than the original data. In this way, the risk of losing
images, and ultimately data, can be kept at a minimum.
10.2.3 Image Analysis Techniques
What is required of an image analysis system for analysing washed root samples
will depend on the type of samples to be analysed and on the parameters to be
measured. If the samples contain roots only and if length is the only parame-
ter to be measured, then automated line-intercept techniques may be sufficient.
If the root sample contains extraneous objects, and more specific measurements
(e.g. diameter, branching) are required, then more sophisticated analysis tech-
niques will be needed.
10.2.3.1 Segmentation of Images of Washed Roots
In image analysis, segmentation (see Box 10.1) of objects from the background
is a necessary task before being able to assess morphological parameters of the
objects. Although the background of washed root samples is relatively uniform,
segmentation is crucial for the automated analysis of root images (Smucker
et al. 1987). This section will discuss grey-Ievel thresholding, which is most often
used for the segmentation of washed root samples, and alternative segmenta-
tion techniques.
Grey-Level Thresholding. For the segmentation of washed root samples, the

grey-Ievel thresholding technique is used most frequent1y. When using thresh-
olding, the most critical step is setting the grey-level threshold (see Box 10.1).
In a few cases, two threshold grey-levels are used instead of just one (Smit et al.
1994; Kaspar and Ewing 1997). The two threshold levels define the upper and
the lower limit of the range of grey-Ievels belonging to the roots.
Contrast is often rather weak in root samples, which usually contain roots
in the range of tens of micrometres (Pallant et al. 1993) to several millimetres,
and in which lighting is not uniform across the whole scanning area. When
decreasing the threshold grey-Ievel to resolve the thinnest roots in a sample,
more background pixels along the root boundaries are classified as object pixels
and, thus, the contours of thicker roots will blur and expand (Fig. 10.2). In addi-
tion, a great deal of noise may occur in the image as a result of a too low grey-
level threshold. On the other hand, when optimising the threshold value for
thicker roots, the thinnest roots will not be resolved. Figure 10.2 shows that
there may be drastic changes in resulting image quality when changing the
threshold by just one grey-Ievel.
Fig. 10.2. Effect of choosing different brightness levels in grey-Ievel thresholding on the quality
of the resulting binary image showing an unstained maize root segment. A threshold grey-
level: 221; B threshold grey-Ievel: 222 (original image: 256 grey-Ievels, spatial resolution:
300dpi)
Staining roots may eliminate some of the problems involved in threshold-

ing, though not completely (see Sect. 10.2.1.2). As a rule of thumb, Kaspar and
Ewing (1997) suggest selecting the lowest possible threshold setting that does
not change background pixels into object pixels for images of the recording tray
filled with water but without roots. Because it is difficult to consistently set the
threshold level in a series of images or when different people are recording
images in turns, a standard threshold level is chosen for alI images.
In summary, simple grey-Ievel thresholding is appropriate only if there
is good contrast between roots and background and a uniform lighting of
the whole scanning area. Because these requirements are most often not met,
it is difficult to visually determine the optimal threshold value (Tollner et al.
1994).
Alternative Segmentation Techniques. When processing root images recorded

by carne ras or scanners, the segmentat ion by thresholding usually leads to
objects with irregular contours producing spurious laterals, and it is difficult to
consistently set the grey-Ievel threshold for a series of samples. Therefore,
attempts have been made to overcome these limitations using alternative seg-
mentation techniques. Apart from grey-Ievel thresholding, however, only a very
limited number of more sophisticated segmentation techniques have been
applied in the analysis of washed root systems. Smit et al. (1994) first estimated
the background, using local maximum and minimum filters. Before the final
histogram-based grey-Ievel thresholding, they applied a Laplacian-like sharp-
ening filter. Smucker et al. (1987) used a combination of edge enhancement and
grey-Ievel thresholding. Koller et al. (1995) proposed a new algorithm for the
segmentat ion and local description of elongated, symmetric line-like structures
using a non-linear combinat ion of linear filters. We adapted and implemented
P ' •
.,
~V.,
~
'- I
J
\
A B .\..
Fig. 10.3. Original 256-grey-scale image (A) of washed white clover (Trifolium pratense 1.)
roots, and resulting image (B) after image segmentat ion and length analysis. Green lines are root
objects, red crosses are ending points of individual root segments and blue circles are extrane-
ous objects discriminated from roots using a 4: 1 length-to-diameter ratio. (Photos with per-
mission from U. Zimmermann, Institute of Plant Sciences, ETH Ziirich)
this algorithm (Fig. 10.3) into a program which provides local shape attributes
(position, width and direction of the centre line) that are necessary for the
morphological description of root segments directly during the detection of
curvilinear objects, i.e. roots.
Editing of Segmented Images. The segmentat ion of images of low contrast

or images that contain very thin roots may result in root objects that are
not completely resolved. If these fragmented roots are measured, length and
area is underestimated. In addition, some of these root fragments may not
satisfy the length-to-width criteria that are used to eliminate extraneous objects
from the image (see Sect. 10.2.3.2). If root fragments are excluded from the
measurements in this way, measurement errors even increase. Therefore,
image-processing algorithms (Russ 1990) can be applied to improve the seg-
mented images prior to morphological measurements. Kaspar and Ewing
(1997) used closing to reduce fragmentation of root objects and to smooth
object edges without deleting small object fragments. Closing involves dilation
(Le. changing any background pixel into an object pixel if at least one of its
neighbour pixels is an object pixel) followed by erosion, which is the reverse of
dilation.
10.2.3.2 Extraction of Morph%gica//nformation

from Binary /mages of Washed Roots
Line-Intercept vs. Chain Methods. When classifying image analysis techniques

that measure morphological parameters of root samples, one can basically dis-
tinguish two approaches (Ewing and Kaspar 1995): line-intercept and chain
methods. Line-intercept techniques, which allow calculation of root length from
the number of intersections of scan lines with root pixels, have been the most
widely used to date. They are sufficiently accurate if the samples contain roots
only and if length and projected area only are measured. If random distribu-
tion of the roots in the image is ensured, additional parameters, such as the
mean root diameter of a sample, can be calculated. However, a major drawback
of such line-intercept approaches is that individual objects within a sample
cannot be described morphologically. Thus, line-intercept techniques produce
only limited information about the sample. Generally, it is of interest to deter-
mine additional parameters:
1. Distribution of root diameters instead of the mean root diameter of the
sample;
2. Branching parameters;
3. Shape parameters of individual root objects.
In contrast to line-intercept techniques, chain methods are able to measure
these parameters by identifying and analysing individual objects in the image
instead of establishing global image features. In principle, chain methods, which
are based on Freeman (1970) chain coding, sum the length of pixels in a chain,
with length correction for pixels connected diagonally (Kaspar and Ewing
1997). Chain algorithms primarily provide the object perimeter by counting the
pixels at the object's border, but they can also be used to measure the length of
skeletonised objects of one pixel width (see below).
Determination of Morphological and Architectural Root Parameters. Root

length is the primary root trait to be measured. The measurement of length
of root samples allows the determination of root length per plant in pot
studies or root length densities in different soil layers in field studies. Root
measurements often include projected root area, from which the root surface
area can be calculated by multiplying it by n, assuming that roots are cylin-
drical (Kokko et al. 1993). From total length and projected area, the mean
diameter of a sample is often calculated. Root length has been rarely classified
according to specified diameter classes (but see Pallant et al. 1993; Dowdy
et al. 1995; Pietola and Smucker 1998), and branching patterns (Zoon and Van
Tienderen 1990) or plant root architecture (Berntson 1992) have seldom been
determined.
Root length has been used more often to describe root systems than other
parameters (Berntson 1992). In the early days of root research, root length was
determined using linear measurement devices such as callipers, rulers or opi-
someters. These techniques are generally accurate and precise (Reicosky et al.
1970), but prone to human error. Additionally, they are too time-consuming for
studies involving large numbers of root samples, but for special purposes, such
as determining the lengths of individual main axes or laterals, these methods
are still applied.
Based on the principle of Newman (1966), later modified by Marsh (197l)
and Tennant (1975), line-intercept methods were widely used for measuring
root length. According to Tennant (1975), root length L is equal to:
L=(tr oN L oDL )j2, (10.1)
where NL is the number of root intersections with parallellines, and DL is the
distance between these lines.
Initially, intersects of roots with parallel, regularly spaced lines were
counted manually, but this was tedious work and prone to human error (Bland
and Mesarch 1990). In the 1970s, photoelectric detection was used increasingly
to determine root intersects with regularly spaced scan lines (Rowse and
Phillips 1974; Goubran and Richards 1979; Richards et al. 1979; Wilhelm et al.
1983; Ottman and Timm 1984; Collins et al. 1987; Morita et al. 1988). Such pho-
toelectric systems, based on line-intercept principles, were reasonably accurate
at relatively high resolutions, but root parameters other than length could not
be measured. Furthermore, they were rather expensive and complex (Harris and
Campbelll989). Later, line-intercept methods were developed to process root
images recorded with video cameras (Voorhees et al. 1980; Barnett et al. 1987;
Burke and LeBlanc 1988; Cunningham et al. 1989; Harris and Campbelll989)
or flatbed scanners (Krstansky and Henderson 1989; Kirchhof 1992). Byestab-
lishing a regression function between the number of pixels and root length
using samples of known size, the total root length in a sample can be estimated.
This "grid-intercept" approach is an improvement over the manual line-
intercept method; the digitised image represents a very fine pixel grid whose
intersections with roots are counted.
Line-intercept based systems are much faster than manual measurement
techniques. Accuracy and precision depend on the root size in a sample, orien-
tation of the root objects and root overlap (Melhuish 1968). Random orienta-
tion of root objects is a requirement of the line-intercept calculation of root
length; non-random placement results in lower accuracy and precision. To
prevent negative effects of non-random root orientation, roots can be reposi-
tioned and recorded for a second time after rotating the scanning trays by
90° (Harris and Campbell 1989). Spreading of roots, however, is very time-
consuming. The line-incercept method is prone to errors due to root overlap at
higher densities of root objects in the scanning area (Barnett et al. 1987). Several
authors (Harrris and Campbell1989; Kirchhof 1992; Kirchhof and Pendar 1993)
provided correction factors for compensat ing for the underestimation of root
length due to root overlap.
Apart from line-intercept methods, chain methods are used for measuring
root length. Using chain algorithms, object length can be measured directly
after thinning the objects to a skeleton (see Box 10.2), i.e. a line of one pixel
width, or by utilising algorithms to calculate length from perim eter and area.
BOX 10.2. Skeletonisation

Apart from perimeter measurements (Ewing and Kaspar 1995), objects in
segmented binary images usually must be reduced to a skeleton or centre
line, i.e. a line of one pixel width, for their morphological description
(Russ 1990). This thinning operat ion is most often accomplished by an
iterative erosion technique (Russ 1990). The skeleton contains both
metric and topological informat ion (e.g. the number of end points or the
number of nodes); it has approximately the same length and topology as
the original object.
Thinning algorithms are needed to create skeletons by eroding pixels
step-by-step from the border to the centre of an object. For accurate
measurements of objects, the thinning process must guarantee
connectivity (avoid removing border pixels and thus causing
discontinuities) and avoid end shortening (avoid removing pixels at the
ends of the line). Alternatively, medial axis transformations (Blum 1967)
might be applied instead of skeletonisation to produce a central line on
which each pixel is at an equal distance from at least two border pixels.
The medial axis transformation has the advantage over the thinned
skeleton in that not only can the length of the object be measured by
counting the pixels on the medial axis, but also the diameter of the object
can be measured at each pixel of the medial axis simuItaneously and
accurately. Although medial axis transformat ion has the potential for
accurate and economic determinations of length and diameter (Russ
1995), it has not been used for root image analysis so far.
The skeletonisation approach has been most often used for the image
analysis of washed root samples (e.g. Lebowitz 1988; Zoon and Van Tienderen
1990; Travis et al. 1993; Smit et al. 1994; Tanaka et al. 1995; Pietola and Smucker
1998). Prior to measurements, a thinning algorithm is applied to create root
skeletons. The success of measurements based on thinning algorithms depends
on the quality of the previous segmentation process. If the borders of the objects
are irregular, skeletonisation may produce spurious lateral branches (Russ
1995), thus leading to an unrealistic description of roots, especially to an over-
estimation of length and branching. After finishing the thinning process, the
length of the object can be measured as a function of the number of vertically,
horizontally and diagonally adjacent pairs of pixels in the skeleton (Lebowitz
1988). It is important to take into account diagonally adjacent pairs of pixels,
because the distance between them is theoretically greater by a factor of..rz than
the distance between vertical or horizontal pairs of pixels. Thus, length L can
be calculated as (Lebowitz 1988):
(10.2)
where D p is the distance between any horizontally or verticalIy adjacent pixels,

and H, V, X are the number of horizontalIy, verticalIy and diagonalIy adjacent
pairs of pixels in the skeleton.
However, Smit et al. (1994) showed that the calculation of Lebowitz (1988)
does not give a cor rect weighting factor for pairs of adjacent pixels when aver-
aged over alI possible directions of root segments. Instead of using correction
factors of 1 and ..rz for horizontalJvertical and diagonal pairs of pixels, respec-
tively, Smit et al. (1994) derived an overall correction factor of 1.12. For a more
accurate calculation of the length of randomly oriented lines, Beckers and
Smeulders (1992) replaced the strictly geometric distances of 1.0 and 1.414 (1.21
on average) between orthogonal and diagonal pixel neighbours by 0.948 and
1.340 (1.14 on average), which is quite close to the correction factor derived by
Smit et al. (1994). Thus, calculation of root length L is simplified as folIowed
(Smit et al. 1994):
L=N·C/R, (10.3)
where N is the number of root skeleton pixels, C the correction factor (1.12),
and R the spatial resolution of the imaging device in pixels per centimetre.
Compared with the direct measurement of root skeleton length, root length
is less often derived from perim eter measurements without a preceding skele-
tonisation of root objects (Pan and Bolton 1991; Kokko et al. 1993; Dowdy et al.
1995; Ewing and Kaspar 1995); this approach is less demanding on computer
memory and time than the above-mentioned skeletonisation process (Ewing
and Kaspar 1995). To overcome the above-mentioned problems of non-random
orientation of objects, Ewing and Kaspar (1995) developed an edge chord algo-
rithm (Fig. IOA) that calculates the length of lines drawn between the edges of
steps along the perimeter of the digitised objects. Its advantages are the small
memory requirement and the fact that the orientation of root objects has only
a minimal effect on measured length. The calculation of object length from area
1.01 Fig. 10.4. Relative length

A (measured/real) of randomly
..r::: oriented wire pieces (A) and
"El
c coefficients of variability of
~
ca 1.00 length measurements (B) by
--
~ the grid-intercept method
u (GRID), a chain method
~
::J according to Freeman (1970)
In
!Il as corrected by Kulpa's
Q)
0.99 --o- GRID
~ -o- CHAIN (1977) estimate of the
---v- ECA theoretical error (CHAIN)
2.0 and the edge chord
B algorithm by Ewing and
~ Kaspar (1995; ECA). Wire
~
~ 1.5 length is shown on a log

:o!Il scale. (Modified from Ewing
.~
and Kaspar (1995), with
> 1.0
'O permis sion from the Journal
1: of Computer-Assisted
Q)
·u 0.5 Microscopy, Volume 7,
:E
Q) Plenum Press)
o
()
10 100 1000
Wire length (cm)
and perim eter, which was measured by a chain-type method and the edge chord
algorithm by Ewing and Kaspar (1995), has been shown to be more accurate
(Fig. IOAA) and more repeatable (Fig. IO.4B) than length measurements by the
grid-intercept method.
There is abundant information on comparisons of true lengths or manual
line-intercept measurements with digitallength measurements of root samples
or calibration objects. Although experimental conditions, imaging devices
and image analysis approaches varied widely, most of the reported coefficients
of determination V) of the association between manual and digital length
measurements were greater than 0.95 (e.g. Barnett et al. 1987; Morita et al.
1988; Farrell et al. 1993; Shuman et al. 1993), and, in some cases, nearly 1
(e.g. Wilhelm et al. 1983; Harris and Campbelll989; Zoon and Van Tienderen
1990; Kirchhof 1992; Smit et al. 1994; Dowdy et al. 1995, 1998; Tanaka et al. 1995).
There were no clear-cut differences in the coefficients of determination
between digitalline-intercept techniques and chain methods. However, results
of a direct comparison of algorithms by Ewing and Kaspar (1995) suggest that
chain-type algorithms may be more accurate and precise than the line-
intercept method.
Farrell et al. (1993) and Murphy and Smucker (1995) attributed weaker rela-
tionships between manually and digitally measured root lengths to the lower
spatial resolution of the imaging devices compared with the human eye, leading
to an underestimation of digital measurements. In the study of Murphy and
Smucker (1995), this was only relevant for perennial ryegrass (Lolium perenne
1.), whose root system is characterised by a large portion of very fine roots, but
not for the coarser-rooted alfalfa (Medicago sativa 1.). This underlines the need
for high-resolution imaging devices that allow reliable recording of the thinnest
roots in a sample (see Sect. 10.2.2.1).
Average coefficients of variation for repeated measurements of root length
and for different operators are usually larger for the manualline-intercept tech-
nique than for digital methods (Farrell et al. 1993). Thus, digital image analysis
techniques may provide more accurate and reproducible measurements of root
length than manual methods.
Projected root area was initially determined with photoelectric devices
designed to measure leaf surface areas (Morrison and Armson 1968; Armson
1972). Because the spatial resolution of these devices was rather poor, this tech-
nique was problematic if too many fine roots were present in the samples
(Kemph 1976). These systems usually required calibration for measuring root
area as was the case when measuring leaf area.
With more recent systems, projected root area is usually measured as the
sum of alI root pixels, which is converted to the appropriate real units of
measure; total root surface area can then be calculated by multiplying the pro-
jected root area of a sample by 1r, assuming that roots are cylindrical (Kokko
et al. 1993). Systems that yield length and mean diameter of roots allow calcu-
lation of root surface area by multiplying the product oflength and mean diam-
eter by 1r. However, this method may be inaccurate when root diameters in
the sample are not evenly distributed above and below the mean diameter
(Kirchhof 1992). Thus, the alternative calculation of total root surface area as
the sum of root surfaces of groups of roots belonging to predefined comple-
menting diameter classes is preferable. The preferred method is to directly
measure the root area of individual root segments or to calculate it from their
lengths and diameters; total root surface area is the sum of the areas of single
root segments. In this way, surface area measurements are expected to be more
accurate, and large, compact extraneous objects, which falsely increase the
area of a sample even more than the length, may be eliminated from the
measurement according to shape parameters.
In line-intercept systems, mean root diameters can be calculated by divid-
ing projected root area by the totallength of a sample, provided that the system
allows for area measurements. In contrast, chain methods using skeletonisation
algorithms enable the direct measurement of the root diameter of individual
objects (Lebowitz 1988). The diameter is, for each point on the object's skele-
ton, directly proportional to the number of iterations necessary in the erosion

process that leads to the skeleton. This approach is more accurate than the cal-
culation of mean root diameter using the line-intercept method.
As a consequence of the thinning process, however, diameter measurements
are unreliable if the diameter of the thinnest roots is close to the lower limit of
spatial resolution (Smucker et al. 1987). If the ratio of root diameter to the pixel
size of the recording device is low, overestimates of root width may occur
because pixels at the root boundary that are only partially occupied by the root
contribute their entire length to the measurement of root diameter (Pan
and Bolton 1991). With increasing root diameters, this overestimation of
root width becomes less important. Therefore, Smucker et al. (1987) suggest that
root segments should be at least four pixels wide for accurate diameter
measurements. This should be considered when selecting scanning resolutions
(see Sect. 10.2.2.1).
Methods that require skeletonisation for length measurements also enable
the determination of branching parameters, which was previously done man-
ually (e.g. Costigan et al. 1982). Image analysis allows the automatic determi-
nation of branching, as shown by Smucker et al. (1987) and Zoon and van
Tienderen (1990) for washed root segments. The thinned skeleton can be used
to determine the degree of branching of a sample. Basically, the number of root
branches can be determined by counting the number of pixels that have three
neighbouring pixels in a 3 x 3 pixel region. In this way, Zoon and van Tienderen
(1990) determined a branching ratio, by dividing the difference between branch
pixels and intersection pixels by root length.
For entire seedling root systems, root architecture (Berntson 1992) can be
determined by image analysis. Large and complex root systems may be divided
into multiple parts, e.g. single main axes, which are recorded and analysed
individually. There are two basic requirements for measurements of root
architecture: first, the root system has to be thinned to a skeleton, represented
by a chain of connected root pixels; second, no loops enclosing background
pixels should occur, and no roots should cross (Berntson 1992). Thus, samples
for such analyses require even more careful spreading than samples for length
measurements. After converting digitised root images into a binary tree data
structure, it has to be checked visually for deviations from the original image,
and, if needed, the image can be edited using a pixel editor program (Berntson
1992). Berntson's (1992) method provides a measurement of the components of
root system architecture (Fitter 1987). The output of this program can be stored
and used in models of the 3-D architecture of plant root systems (Berntson
1994). Theory and measurement of root architecture are described further in
Chapter 4.
Recently, methods have become available for the computer-assisted
description of fractal geometry (Berntson et al. 1998). The reader is referred
to Chapter 4 for more information on theory and measurement of fractal

geometry.
Discrimination of Roots from Extraneous Objects. Most root image analysis

systems do not provide morphological parameters of individual objects, so
there is no way to discriminate extraneous objects from roots. With systems that
provide single-object traits, the classical way of differentiation is to use simple
shape indices.
Murphy and Smucker (1995) used a length-to-width ratio of 4: 1 as a
criterion for discriminating roots from unwanted objects; objects with
length-to-width ratios greater than 4 were assumed to be roots, and objects
with ratios less than 4 were excluded from root measurements. Dowdy et al.
(1998) recently introduced a length-to-width ratio of 15: 1 for discriminating
between extraneous objects and roots, which are assumed to have much
greater length-to-width ratios than unwanted objects. The 15: 1 ratio was
derived from corn root samples through iterations. For hand-picked root
samples with no extraneous objects, an r of 0.97 was found for the correlation
between length measurements by image analysis, using the 15: 1 ratio, and
a manual line-intercept method (Dowdy et al. 1998). The correlation of
root lengths of samples with extraneous objects measured with a 15: 1 and a
4: 1 discrimination ratio gave an r 2 of 0.80 (Dowdy et al. 1998).
More complex shape indices derived from morphological parameters of
objects have not yet been used to discriminate roots from extraneous
objects. Because roots are usually several times longer than they are wide,
they may be discriminated from extraneous objects, which usually have a
more compact shape than roots, by various kinds of morphological parameters,
such as the spherical index (Glasbey and Horgan 1995), describing the
roundness of objects. Colour may be a useful additional criterion for dis-
crimination, because unwanted objects are often darker and more opaque
than roots. Collins et al. (1987) achieved some differential staining of extrane-
ous organic matter with trypan blue, which suggests that there may be some
potential for discriminating roots and unwanted organic objects by selective
staining.
The user has to be aware that the above-mentioned discrimination
approaches are never perfect. Although a given length-to-width ratio may
exclude the majority of extraneous objets from morphological measurements,
an unknown number of unwanted objects will not be eliminated and an
unknown amount of root objects will be falsely removed.
Differentiation Between Viable and Dead Roots. An accurate, automated dis-

crimination between living and dead roots would be highly desirable because
it is essential to know the ratio of living to dead roots in samples for studying
Table 10.5. Dyes used for vital-staining of plant roots
Dye Authors
2,3,5-triphenyltetrazolium chloride Goedewaagen (1954); Sator and Bommer (1971);

Crapo and Coleman (1972); Knievel (1973);
Clemensson-Lindell (1994)
2,3,5-triphenyltetrazolium bromide Jacques and Schwass (1956)

Congo red Ward et al. (1978)
root function and turnover (KnieveI1973). However, differentiation is techni-

cally difficult (Schuurman and Goedewaagen 1971), and basic knowledge of
objective differentiation criteria is scarce (see Chap. 9).
Because certain dyes are used to discriminate between dead and living
roots (Table 10.5), they could be used for a selective measurement of previously
live roots in a sample if the contrast between roots stained in this way
and non-stained roots is great enough to reliably discriminate between both
groups. Ottman and Timm (1984), using an image analyser, measured the
length of viable roots that were effectiveIy differentiated from non-viable
roots and organic residues by preferential staining with trypan bIue. As an
alternative to staining, differences in colour and/or texture information
might be used for discriminating between living and dead roots, but more
information on relationships between these parameters and root viability is
needed first.
10.2.4 Root Image Analysis Systems
Besides a variety of custom-built systems, there are a few recently developed

systems that have been wideIy used (Table IOA). Except for the "NIH Image"
(National Institutes of Health (NIH), Bethesda, MD, USA) and "MacRHIZO"
(Regent Instruments Inc., Quebec, Canada) programs for Macintosh computers
and the "Delta-T Mk II" (Delta-T Devices, Burwell, UK), which is not computer-
based, aH the other programs listed in Table IOA run on PCs. While "Delta-T
SCAN" and "ROOTEDGE" (National Soil Tilth Laboratory, Ames, IA, USA) are
designed for the DOS operating system, "WinRHIZO" (Regent Instruments Inc.,
Quebec, Canada) runs under the Windows environment. There are public
domain programs for Macintosh computers ("NIH Image") and PCs ("ROOT-
EDGE"), but the majority of systems are distributed commercially, often in
combination with imaging devices such as Hat-bed scanners ("Delta-T SCAN",
"MacRHIZO" and "WinRHIZO"). The outdated "Delta-T Mark II" is no longer
commercially available, but was widely used in recent years. While "NIH Image"
is a general purpose program for which task-specific macros have to be written,
the other systems have no option for user-written extensions.
AH these systems are designed for measuring root length and, in most cases,
root surface area. Some of the programs, e.g. "Delta-T SCAN" (Kirchhof 1992;
Kirchhof and Pendar 1993), and "MacRHIZO"I"WinRHIZO" (Arsenault et al.
1995; Bauhus and Messier 1999), have algorithms for correcting errors in the
measurement of root length and area due to root overlaps. In addition to
measurements of root length, area and diameter, some of the systems can be
used for evaluations of other morphological parameters. Among other para-
meters, "Delta-T SCAN" allows for object counting and root tip counting;
"MacRHIZO"/"WinRHIZO" for morphological measurements using user-
definable colour classes, root tip counting and root topology analyses.
If roots washed from soil containing unwanted organic objects are to be
measured, then the systems should aHow for individual object measurements
and, thus, the possibility of discriminating between roots and extraneous
objects. The "MacRHIZO"/"WinRHIZO" and the "ROOTEDGE" (Kaspar and
Ewing 1997) programs and the system described by Murphy and Smucker
(1995) have this feature. However, if only root samples from hydroponics or
solid substrates without soil are measured, then the systems that do not allow
for object-based measurements can also give accurate results because such
samples usually contain no or only few extraneous objects. The "Delta-T SCAN"
programs stands between these two groups, because it allows for single-object
measurements of various parameters but not of root length.
No information can be given yet on the comparative accuracy of the
different systems, because there are no data on comparisons between
image analysis systems or programs. It is not possible to give general recom-
mendations on the use of root image analysis systems, because their options
and drawbacks vary greatly, and the requirements of scientists studying roots
are diverse.
10.2.5 Priorities for Image Analysis of Washed Roots
The major priority in developing more sophisticated image analysis techniques

is to find more accurate image segmentation techniques that do not rely on
simple grey-Ievel thresholding. A second priority is the reliable differentiation
between roots and extraneous objects based on shape parameters and/or
colour. In both cases, new algorithms should be developed together by biolo-
gists, agronomists and ecologists on the one hand and image analysis special-
ists on the other. The adaptation of algorithms from other fields also may be
applicable.
Computer equipment for handling large colour image files is available

today, and, thus, the use of colour information in the analysis of washed samples
may be beneficial in several ways. The inclusion of colour may help to differ-
entiate between living and dead roots, between roots and extraneous objects,
and in assessing the portion of root length or area infected by root diseases, as
shown by Kokko et al. (1993) for subcrown internode discoloration in wheat
caused by common root rot.
None of the image analysis techniques described have provided an
overall solution for the fast and accurate measurement of morphological and
topological root parameters. A special solution for each situation has seemed
to be more suitable to date. Possibly, the ongoing dramatic development of
computer resources will enable scientists to fully exploit the potential of image
analysis techniques in the morphological description of root samples in the
future.
10.3 Image Analysis of Minirhizotron Images
The minirhizotron technique allows for non-destructive measurements and

minimises point-to-point variation because repeated root observations are pos-
sible at the same positions in the soil. The analysis of minirhizotron images,
however, is tedious and prone to human error. Images are usually analysed by
manually counting the number of roots or root elements (Upchurch 1987; see
Chap. 8), which may be quite closely correlated with root length on the same
images (Box and Ramseur 1993). Only in a few cases has root length in minirhi-
zotron images been measured manually by tracing roots on a TV or computer
monitor (Cheng et al. 1990; Hendrick and Pregitzer 1992) or by means of the
modified line-intercept method (Hirasawa et al. 1995). These attempts to
directly measure root lengths yielded more accurate information on root mor-
phology, but required even more time for analysing images compared with
manual counting, especially for tracing of roots. Because minirhizotron inves-
tigations usually require the recording and evaluation of a large number
of images, there is a need for replacing the above-mentioned approaches by
semi- or fully-automated analyses of images.
10.3.1 Segmentation of Minirhizotron Images
Image quality is an important prerequisite for the successful image analysis of

video-recorded roots (Smucker et al. 1987). There are several factors that may
contribute to the poor quality of minirhizotron images. The non-uniform dif-
fus ion of light due to tube curvature may result in inconsistencies in brightness
observed in digitised images (Brown and Upchurch 1987). Non-perpendicular
viewing or camera-tube non-linearity may lead to geometrically distorted
images (Heeraman et al. 1993). Low battery power can also contribute to
photometric distortion due to poor illumination by the incandescent bulbs of
the minirhizotron camera (Heeraman et al. 1993).
The major problem in the automated analysis of minirhizotron images
is the much more complex background compared with washed samples, both
in colour and in texture. A variety of objects, such as reflective soil partieles,
droplets of condensation, scratches on the tubes, electronic noise and soil fauna,
may have similar brightness values as the roots (Smucker et al. 1987; Casarin
et al. 1991; Heeraman et al. 1993; Andren et al. 1996). Furthermore, the
brightness of roots may differ due to variable luminance along the same root,
to loss of root contact with the tube wall, to roots being out of focus (leads
to greater apparent diameters of roots or to a more reflective background)
and to the structure of small roots, which appear to be fragmented (Casarin
et al. 1991).
Additionally, the background of a given minirhizotron image position may
not be constant over time due to varying soil moisture contents, settling of soil
(Bohm 1979), diurnal variations in root diameters (Huck et al. 1970), reorien-
tation of soil partieles during root elongation (Goss 1991), the displacement of
soil partieles into pores adjacent to the tube wall by percolating water, move-
ment of the minirhizotron tube (Heeraman et al. 1993) and activity of soil
animals. As a consequence, the major problem in segmentation (see Box 10.1)
of minirhizotron images is the discrimination of roots from a very complex,
changing background.
10.3.1.1 Enhancement of Minirhizotron Images
Several techniques of image enhancement may have the potential to facilitate

the subsequent segmentation of minirhizotron images. Contrary to the major-
ity of systems reported for the analysis of washed roots, enhancing contrast has
usually been used prior to the analysis of minirhizotron images. Smucker et al.
(1987) applied low-pass filters to eliminate high-frequency noise from the
image, and high-frequency filters averaged the intensity values of each pixel.
Casarin et al. (1991) increased the contrast for roots of small to medium diam-
eter and reduced the noise to some extent on uniform luminosity areas of the
background using linear convolution, a techn[que which is effective for low
dynamic images. Heeraman et al. (1993) used linear contrast stretch enhance-
ment. Andren et al. (1996) reduced background noise, which was created by
Fig. 10.5. Original minirhizotron image of meadow fescue (Festuca pratensis Huds.) roots (A)
and roots identified by median and convolution filtering, multiple grey-level thresholding and
skeletonisation procedures according to Andn!n et al. (1996; B). (Photos with permission from
E Sindh0j, Department of Soil Science, Swedish University of Agricultural Sciences, Uppsala)
reflections from sand particles, by two-step median filtering using a 5 x 5

matrix.
10.3.1.2 Segmentation of Minirhizotron Images
Global techniques, such as grey-scale thresholding, can only be used for the seg-
mentation (see Box 10.1) of images if the objects to be recognised are consis-
tently lighter or darker than the background. Roots in minirhizotron images are
usually lighter than the background, but the resulting light intensity histograms
are not bimodal (Nater et al. 1992). Andn!n et al. (1996) applied grey-level
thresholding using a fixed brightness level for differentiating roots from the soil
background (Fig. 10.5). To avoid background scatter, they chose a conservative
(Le. high) threshold level; as a consequence, only the lightest parts of roots were
detected (Andren et al. 1996). Thus, the discrimination of roots from the soil
background by grey-Ievel thresholding is not reliable, especially in the case of
thin secondary and higher order roots (Casarin et al. 1991). In addition, the
manual selection of a threshold level is dependent on the operator's judgement
and, thus, may lead to error. To overcome this limitation, Smucker et al. (1987)
applied grey-Ievel thresholding, with the mean intensity value of all the pixels
along previously tracked root edges automatically set as the threshold. This
approach gave good results when the roots had a similar and constant inten-
sity. Although roots exhibit large, localised changes in light intensity, other fea-
tures such as pores and cracks may exhibit the same rate of change as do roots
(Nater et al. 1992). Thus, segmentation techniques based on thresholding or
edge detection only are inappropriate.
Because global segmentation techniques are not feasible, several groups
applied local techniques. Casarin et al. (1991) used linear flat-hat transforms
Fig.10.6. Original minirhizotron image of alfalfa (Medicago sativa 1.) roots (A) and the centre
lines of roots resulting from a ridge detection approach (B). (Photos from Smucker (1993), with
permission from the Annual Review of Phytopathology, Volume 31, Annual Reviews)
(Meyer 1979) at different diameters for segmenting images. With this approach,
the extraction of medium and large roots presented few problems (Casarin et
al. 1991). Other approaches for identifying roots in minirhizotron images are
ridge detection by identifying the parallel edges of each root segment
(Ferguson et al. 1990), allowing an identificat ion of 94-96% of the roots,
depending on the complexity of images (Fig. 10.6). Smucker (1993) reported
token-based algorithms, which are an example of a knowledge-based feedback
approach, for improved identification of roots on a complex background. A
further development of Nater et al. (1992) tested several artificial neural system
architectures for segmentat ion. Inputs to their models were horizontal and
vertical derivative images produced from the raw image by smoothing and
differentiation with least square procedures. A training image was created by
hand-editing the original image. After about 40 iterations, most pixels in the
training image were correct1y classified, whereas accuracy was much reduced
when processing other images for which the system had not been trained. The
ability to adjust classification, even in the presence of noise, is a major advan-
tage over classical segmentation approaches (Nater et al. 1992).
In addition to the above-mentioned contour-based approaches, a region
growing approach was recent1y reported by Jankowski et al. (1995). After the
initial detection of seeds, Le. significant root regions, these were expanded
to optimal root contours by a heuristic search for parallel root structures.
Jankowski et al. (1995) detected at least one seed on each root segment in 28
out of 32 tested images, but no seeds could be detected if the inner part of root
segments were strongly structured. Results of the segmentation technique used
by Jankowski et al. (1995) were better than when processing the same test
images with the technique reported by Smucker et al. (1987).
10.3.2 Extraction of Information

from Binary Minirhizotron Images
10.3.2.1 Determination of Morph%gica/

and Architectura/ Root Parameters
After segmentation of images, morphological root traits can be determined in

basically the same way as for washed roots (see Sect. 10.2.3.2). The primary root
parameter to be measured is also root length. Root length was measured in alI
reported attempts to analyse minirhizotron images. In some cases, root length
was categorised by diameter classes (Smucker et al. 1987; Casarin et al. 1991).
Measurements of length from rhizotron and minirhizotron video images
according to the technique by Casarin et al. (1991) showed less than 10% dif-
ference from values determined manuallywith Newman's method (Girardin et
al. 1991). Smucker et al. (1987) reported a difference of 7.6% between manual
line-intercept and digital analysis techniques. On average, Andn!n et al. (1996)
observed that root lengths measured using image analysis were 9% smaller than
manually traced root lengths.
Projected root surface area is easily determined by measuring the image
area occupied by the roots relative to the total image area (Heeraman et al.
1993), but area measurements probably do not very precisely reflect the true
root surface area, because visible root boundaries are usually not smooth and
contours of roots are often partially covered by soil particles. Smucker et al.
(1987) found a difference of 6.2% between manual and digital measurements
of projected area.
Acrylic rhizotron walls do not seem to disturb secondary root ramification
(Jordan 1992), and, thus, branching indices could be determined in minirhi-
zotron images as suggested by Smucker et al. (1987) and Casarin et al. (1991).
However, there are no data available to demonstrate whether root branching
can be determined reliably using the relatively small minirhizotron pictures
only. Before measuring branching indices, thinning of the roots is required
(see Box 10.2).
The development of morphological root parameters over time was evalu-
ated by Casarin et al. (1991) and Heeraman et al. (1993). The latter detected
changes in root appearance and disappearance with time by assigning a unique
primary colour, i.e. red, green or blue, to selected growth stages and then over-
laying the images of a position in soil taken at the different recording times.
Areas of no change appeared white, whereas changed areas appeared as one of
the primary colours or a complementary colour as a result of the mixing of two
primary colo urs. However, the resulting colour composites were interpreted
only visually. Although the automated detection of temporal changes in root
parameters is very desirable, such analyses would be prone to differences in root
detection due to the possibility of a changing background and slight variations

in the position of the camera, despite the use of camera indexing handles as
described by Ferguson and Smucker (1989).
A differentiation between living and dead roots is desirable. Different flu-
orescence of living and dead roots may have potential for discriminating
between them by means of image analysis, provided that viable roots fluoresce
more intensely than dead ones. Brown and Scott (1984) showed that actively
growing parts of soybeans fluoresce strongly under ultraviolet light and that
fluorescence decreases with root age and ceases at the beginning of suberisa-
tion. Wang et al. (1995) found that an ultraviolet light source was more accu-
rate than visible light for estimating live proportions of roots of woody species
in a minirhizotron, whereas the opposite was true for grass species.
10.3.2.2 D;scr;m;nat;on of Roots from Extraneous Objects
There is a variety of extraneous objects potentially visible in minirhizotron

images. Extraneous objects, typical of minirhizotron images, are scratches on
tube walls, water drops or water vapour, and soil animals. Such extraneous
objects may be excluded by shape parameters (see Sect. lO.2.3.2). Smucker et al.
(1987) used a maximum and a minimum root length and a length-to-width
ratio of 4: 1 as criteria for this discrimination. One problem with this approach
is that root segments may quite often have shorter length-to-width ratios than
4: 1, because root segments are either partially covered by soil particles or other
extraneous objects, such as organic residues, or because segmentation results
in fragmented root segments that are separated by gaps. Scratches, water
droplets and certain mineral particles are often isolated and strongly reflect
light, making them easier to eliminate (Casarin et al. 1991).
10.3.3 Priorities for Image Analysis of Minirhizotron Images
Image quality is crucial to the analysis of minirhizotron images and must

be improved. Nater et al. (1992) suggest testing different wavelength filters for
the minirhizotron light source for increased contrast between roots and
background.
Minirhizotron images are usually evaluated by manual counting or tracing
of roots. To enable a more efficient method of analysis, alI the recorded images
must be made available in digital form. This would make the visual evaluation
more flexible, eliminating the necessity of observing the samples in the order
in which they were recorded, thus making a digital analysis possible (Dubach
and Russelle 1995). With digital images, standard image enhancement tech-
niques may be applied to increase the contrast for easier visual or automated
evaluation of images. There are two possible approaches to automatically digi-
tising minirhizotron images. First, images could be directly digitised during
recording, thus eliminating the need to temporarily store them on video tapes.
Second, image code information as provided, for example, by a screen marker
system (Bartz Technology, Santa Barbara, California) could be used to capture
video frames from a video tape and to select an individual frame to be saved
for each tube position according to unique coded information corresponding
to this tube position.
A major conclusion of previous attempts to process minirhizotron
images is that segmentation techniques based only on conventional grey-Ievel
thresholding are inappropriate. Thus, more sophisticated segmentation tech-
niques specifically adapted to discriminate roots from a complex background
have to be developed.An artificial neural systems approach and a contour-based
technique have already shown some potential for automatic analysis, at least
of images with relatively good, uniform contrast between roots and the
soil background. This requirement, however, is usually not met for the
majority of minirhizotron images. Thus, semi-automated systems that allow
for user-input, as proposed, for example, by Daneels et al. (1993) for iden-
tifying and outlining objects in images, should be evaluated as intermediate
steps in the image analysis of minirhizotron images, especially when dealing
with a very complex background or plant species with very thin or translucent
roots.
Results from the digital analysis of images that cover a wide range of soil
conditions, plant species and growth stages should be compared with those
from manual measurements of the same images. To date, such comparisons
have only been made with a very limited number of images. Thus, a conclusive
assessment of exist ing approaches for the digital processing of minirhizotron
images is not yet possible.
Further Reading
Glasbey CA, Horgan GW (1995) Image analysis for the biological sciences. John Wiley, Chichester
Price TV, Osborne CF (1990) Computer imaging and its applications to some problems in
agriculture and plant science. Crit Rev Plant Sci 9: 235-266
Russ JC (1990) Computer-assisted microscopy: the measurement and analysis of images. Plenum
Press, New York
Russ JC (1995) The image processing handbook. 2nd edn. CRC Press, Boca Raton, Florida
Smucker AJM (1993) Soi! environmental modifications of root dynamics and measurement.
Annu Rev Phytopathol31: 191-216
Tillett RD (1991) Image analysis for agricultural processes: a review of potential opportunities.
J Agric Engin Res 50: 247-258
References
Andren O, Elmquist H, Hansson AC (1996) Recording, processing and analysis of grass root
images from a rhizotron. Plant Soi! 185: 259-264
Armson KA (1972) Distribution of conifer seedling roots in a nursery soi!. For Chron 48:
141-143
Arsenault JL, Pouleur S, Messier C, Guay R (1995) WinRHIZO™, a root measuring system with
a unique overlap correction method. HortScience 30: 906 (Abstr)
Barnett CE, White RA, Petrovic AM, Good GL (1987) An automated apparatus for measuring
root length. HortScience 22: 140-144
Bauhus J, Messier C (1999) Evaluation of line root length and diameter measurements obtained
with RHIZO image analysis. Agron J 91: 142-147
Beckers ALD, Smeulders AWM (1992) Optimallength and distance estimation in three dimen-
sions. Comput Vision Graph Image Process 55: 296-306
Berntson GM (1992) A computer program for characterizing root system branching patterns.
Plant Soi! 140: 145-149
Berntson GM (1994) Modelling root architecture: are there tradeoffs between efficiency and
potential of resource acquisition? New Phytol127: 483-493
Berntson GM, Lynch J, Snapp S (1998) Fractal geometry and the description of plant root
systems: current perspectives and future applications. In: Baveye P, Parlange JY, Stewart BA
(eds) Fractals in soi! science. CRC Press, Boca Raton, Florida, pp 113-152
Bland WL, Mesarch MA (1990) Counting error in the line-intercept method of measuring root
length. Plant Soi! 125: 155-157
Blum H (1967) A transformation for extracting new descriptors of shape. In: Whaten-Dunn W
(ed) Models for the perception of speech and visual form. MIT Press, Cambridge, Massa-
chusetts, pp 153-171
Bohm W (1979) Methods of studying root systems. Springer, Berlin Heidelberg New York
Box JE Jr, Ramseur EL (1993) Minirhizotron wheat root data - comparisons to soi! core root
data. Agron J 85: 1058-1060
Brown DA, Scott HD (1984) Dependence of crop growth and yield on root development and
activity. In: Barber SA, Bouldin DR (eds) Roots, nutrient and water influx, and plant growth.
Soi! Science Society of America (SSSA), Madison, Wisconsin, pp 101-136
Brown DA, Upchurch DR (1987) Minirhizotrons: a summary of methods and instruments in
current use. In: Taylor HM (ed) Minirhizotron observation tubes: methods and applica-
tions for measuring rhizosphere dynamics. ASA Spec Publ 50: 15-30
Burke MK, LeBlanc DC (1988) Rapid measurement of fine root length using photoelectronic
image analysis. Ecology 69: 1286-1289
Casarin M, Jacquey S, Fouere A, Girardin P (1991) Digital picture processing applied to the eval-
uation of plant root dynamics. In: McMichael BL, Persson H (eds) Plant roots and their
environment. Proc Symp Int Soc of Root Research (ISRR), 21-26 August, 1988, Uppsala,
Sweden. Elsevier, Amsterdam, pp 570-575
Castleman KR (1979) Digital image processing. Prentice-Hall, Eng1ewood Cliffs, New Jersey
Cheng W, Coleman DC, Box JE Jr (1990) Root dynamics, production and distribution in agroe-
cosystems on the Georgia Piedmont using minirhizotrons. J Appl Ecol 27: 592-604
Clemensson-Lindell A (1994) Triphenyltetrazolium chloride as an indicator of fine-root vitality
and environmental stress in coniferous forest stands - applications and limitations. Plant
Soi! 159: 297-300
Collins RP, Gregory PJ, Rowse HR, Morgan A, Lancashire B (1987) Improved methods of esti-
mating root length using a photocopier, a light box and a bar code reader. Plant Soil 103:
277-280
Costigan PA, Rose JA, McBurney T (1982) A microcomputer-based method for the rapid and
detai!ed measurement of seedling root systems. Plant Soi! 69: 305-309
Crapo NL, Coleman DC (1972) Root distribution and respiration in a Carolina old field. Oikos
23: l37-l39
Cunningham M, Adams MB, Luxmoore RJ, Post WM, DeAngelis DL (1989) Quick estimates of
root length, using a video image analyzer. Can J For Res 19: 335-340
Daneels D, van Campenhout D, Niblack W, Equitz W, Barber R, Bellon E, Fierens F (1993) Inter-
active outlining: an improved approach using active contours. In: Niblack W (ed) Storage
and retrieval for image and video databases. Proc SPIE 1908. SPIE, Bellingham, Washing-
ton, pp 226-233
Dowdy RH, Nater EA, Dolan MS (1995) Quantification of the length and diameter of root seg-
ments with public domain software. Commun Soi! Sci Plant Anal 26: 459-468
Dowdy RH, Smucker AJM, Dolan MS, Ferguson JC (1998) Automated image analyses for sepa-
rating plant roots from soi! debris elutriated from soi! cores. Plant Soi! 200: 91-94
installed minirhizotrons. Agron J 87: 258-263
Ewing RP, Kaspar TC (1995) Accurate perimeter and length measurement using an edge chord
algorithm. J Comput Assist Microsc 7: 99-100
Farrell RE, Walley PL, Lukey AP, Germida JJ (1993) Manual and digitalline-intercept methods
of measuring root length - a comparison. Agron J 85: 1233-1237
Ferguson JC, Smucker AJM (1989) Modifications of the minirhizotron video camera system for
measuring spatial and temporal root dynamics. Soi! Sci Soc Am J 53: 1601-1605
Ferguson JC, Chang SL, Smucker AJM (1990) Automated segmentation of minirhizotron images
based on the coherence of intensity gradients. In: Abstr Conf on analytical methods for
quantifying root and soi! dynamics. St Louis, Missouri, l3-14 August, 7pp
Fitter AH (1987) An architectural approach to the comparative ecology of plant root systems.
New Phytoll06: 61-77
Freeman H (1970) Boundary encoding and processing. In: Lipkin BS, Rosenfeld A (eds) Picture
processing and psychopictorics. Academic Press, New York, pp 241-266
Fu KS, Mui JK (1981) A survey on image segmentation. Pattern Recogn l3: 3-16
Girardin P, Trendel R, Jacquey S, Casarin M, Goss M, Fouere A (1991) Comparison of in situ root
observations with rhizoscope, video recording and endoscope. In: McMichael BL, Persson
H (eds) Plant roots and their environment. Proc Symp Int Soc of Root Research (ISRR),
21-26 August, 1988, Uppsala, Sweden. Elsevier, Amsterdam, pp 603-607
Glasbey CA, Horgan GW (1995) Image analysis for the biological sciences. John Wi!ey, Chichester
Goedewaagen MAJ (1954) Beoordeling van de vitaliteit van het wortelstelsel. Landbouwkd
Tijdschr 66: 553-555
Goss MJ (1991) Consequences of the activity of roots on soi!. In: Atkinson D (ed) Plant root
growth: an ecological perspective. British Ecological Society Special Publication No 10.
Blackwell, Oxford, pp 17l-186
Goubran FH, Richards D (1979) The estimation of root length in samples and sub-samples:
comparison of a visual and an automatic method. Plant Soi! 52: 77-83
Hackett C (1968) A study of the root system of barley. 1. Effects of nutrition on two varieties.
New Phytol67: 287-299
Haralick RM, Shapiro LG (1988) Segmentation and its place in machine vision. Scanning
Microsc (Suppl) 2: 39-54
Harris GA, Campbell GS (1989) Automated quantification of roots using a simple image
analyzer. Agron J 81: 935-938
Heeraman DA, Crown PH, Juma NG (1993) A color composite technique for detecting root
dynamics of barley (Hordeum vulgare 1.) from minirhizotron images. Plant Soil 157:
275-287
Hendrick RL, Pregitzer KS (1992) The demography of fine roots in a northern hardwood forest.
Ecology 73: 1094-1104
Hirasawa T, Masahide T, Ishihara K (1995) A minirhizotron method,for measuring root systems
of soybean plants growing in the field. Jpn J Crop Sci 64: 78-85
Huck MG, Klepper B, Taylor HM (1970) Diurnal variations in root diameter. Plant Physiol45:
529-530
Jacques WA, Schwass RH (1956) Root development in some common New Zealand pasture
plants. 7. Seasonal root replacement in perennial ryegrass (Lolium perenne), Italian rye-
grass (L. multiflorum) and tall fescue (Festuca arundinacea). N Z J Sci Technol Sect A 37:
569-583
Jain AK (1989) Fundamentals of digital image processing. Prentice-Hall, Englewood Cliffs, New
Jersey
Jankowski M, Breckle S-W, Posch S, Sagerer G, Veste M (1995) Automatische Detektion von
Wurzelsystemen in Minirhizotron-Bildern. In: Sagerer G, Posch S, Kummert F (eds)
Mustererkennung 1995 - Verstehen akustischer und visueller Informationen. 17. DAGM-
Symposium, Bielefeld, 13-15 September 1995. Springer, Berlin, Heidelbery New York,
pp 176-185
Jordan MO (1992) Can rhizotrons be used for the study of corn (Zea mays 1.) root ramifica-
tion? Agronomie 12: 3-14
Kaspar TC, Ewing RP (1997) ROOTEDGE: software for measuring root length from desktop
scanner images. Agron J 89: 932-940
Kemph GS (1976) Measuring fibrous roots with a leaf area meter. J Range Manage 29: 85-86
Kirchhof G (1992) Measurement of root length and thickness using a hand-held computer
scanner. Field Crops Res 29: 79-88
Kirchhof G, Pendar K (1993) Delta-T SCAN° user manual. Delta-T Devices, Burwell, Cambridge,
UK
Knievel DP (1973) Procedure for estimating ratio of live to dead root dry matter in root core
samples. Crop Sci 13: 124-126
Kokko EG, Conner RL, Kozub GC, Lee B (1993) Quantification by image analysis of subcrown
internode discoloration in wheat caused by common root rot. Phytopathology 83: 976-981
Koller TM, Gerig G, Szekely G, Dettwiler D (1995) Multiscale detection of curvilinear structures
in 2-D and 3-D image data. Proc 5th Int Conf on Computer Vision, 20-23 June, Cambridge,
Massachusetts. IEEE Computer Society Press, Los Alamitos, California, pp 864-869
Krstansky J], Henderson GS (1989) Computerised measurement of fine root length using a
desktop image scanner. In: Agronomy abstracts. American Society of Agronomy (ASA),
Madison, Wisconsin, 307 pp
Krumenaker L (1994) Hardware, software advances brighten image-analysis systems picture.
Scientist 8: 17-19
Kulpa Z (1977) Area and perim eter measurement of blobs in discrete binary pictures. Comput
Graph Image Process 6: 434-451
340 W. Richner et a!.
Lebowitz RJ (1988) Digital image analysis measurement of root length and diameter. Environ
Exp Bot 28: 267-273
Livesley SJ, Stacey CL, Gregory PJ, Buresh RJ (1998) Sieve size effects on root length and biomass
measurements of maize (Zea mays) and Grevi/lea robusta. Plant Soil207: 183-193
Marsh B (1971) Measurement of length in random arrangements of lines. J Appl Ecol 8: 265-
267
McGowan M, Armstrong MJ, Corrie JA (1983) A rapid fiuorescent-dye technique for measuring
root length. Exp Agric 19: 209-216
Melhuish FM (1968) A precise technique for measurement of roots and root distribution in soils.
Ann Bot 32: 15-22
Meyer F (1979) Cytologie quantitative et morphologie mathematique. Thesis, Ecole des Mines
de Paris, Paris
Morita STS, Haruki Y, Yamazaki K (1988) Morphological characters of rice roots estimated with
a root length scanner. Jpn J Crop Sci 57: 371-376
Morrison JK, Armson KA (1968) The rhizometer - a device for measuring roots of tree seedlings.
For Chron 44: 21-23
Murakami T, Yoneyama T (1988) Comparison of root length of two rice (Oryza sativa 1.) vari-
eties by using an image analyzer. Plant Soil105: 287-289
Murphy SL, Smucker AJM (1995) Evaluation of video image analysis and line-intercept methods
for measuring root systems of alfalfa and ryegrass. Agron J 87: 865-868
Nater EA, Nater KD, Baker JM (1992) Application of artificial neural system algorithms to image
analysis of roots in soi!. 1. Initial results. Geoderma 53: 237-253
Newman EI (1966) A method of estimating the totallength of roots in a sample. J Appl Ecol 3:
139-145
Ottman MJ, Timm H (1984) Measurement of viable plant roots with the image analyzing com-
puter. Agron J 77: 1018-1020
Pallant E, Holmgren RA, Schuler GE, McCracken KL, Drbal B (1993) Using a fine root extrac-
tion device to quantify small diameter corn roots (::; 0.025 mm) in field soils. Plant Soi1153:
273-279
Pan WL, Bolton RP (1991) Root quantification by edge discrimination using a desktop scanner.
Agron J 83: 1047-1052
Pietola L, Smucker AJM (1998) Fibrous carrot root responses to irrigation and compaction of
sandy and organic soils. Plant Soi1200: 95-105
Rasband WS, Bright DS (1995) NIH Image: a public domain image processing program for the
Macintosh. Microbeam Anal Soc J 4: 137-149
Reicosky DC, Millington RJ, Peters DB (1970) A comparison of three methods for estimat ing
root length. Agron J 62: 451-453
Richards D, Goubran FH, Garwoli WN, Daly MW (1979) A machine for determining root length.
Plant Soi152: 69-76
Rowse HR, Phillips DA (1974) An instrument for estimat ing the totallength of root in a sample.
J Appl Ecoll1: 309-314
Russ JC (1990) Computer-assisted microscopy: the measurement and analysis of images. Plenum
Press, New York
Russ JC (1995) The image processing handbook, 2nd edn. CRC Press, Boca Raton, Florida
Sackville-Hamilton NR, Jones M, Harper JL (1985) Automated analysis of roots in bulk soi!. In:
Atkinson D (ed) Plant root growth: an ecological perspective. British Ecological Society
Special Publication No 10. Blackwell, Oxford, pp 291-301
Sator C, Bommer D (1971) Methodological studies to distinguish functional from non-

functional roots of grassland plants. In: Ellenberg H (ed) Integrated experimental ecology:
methods and results of ecosystem research in the German Solling project. Springer, Berlin
Heidelberg New York, pp 72-74
Sattelmacher B, Klotz F, Marschner H (1983) Comparison of two non-destructive methods for
determination of root surface areas. J Plant Nutr Soil Sci 146: 449-459
Schuurman 11, Goedewaagen MAJ (1971) Methods for the examination of root systems and
roots, 2nd edn. Centre for Agricultural Publishing and Documentation, Wageningen, the
Netherlands
Shuman LM, Ramseur EL, Wilson DO (1993) Video image method compared to a hand method
for determining root lengths. J Plant Nutr 16: 563-571
Smit AL, Sprangers JFCM, Sablik PW, Groenwold J (1994) Automated measurement of root
length with a three-dimensional high-resolution scanner and image analysis. Plant Soil
158: 145-149
Smucker AJM (1993) Soil environmental modifications of root dynamics and measurement.
Annu Rev Phytopathol31: 191-216
Smucker AJM, Ferguson JC, DeBruyn WP, Belford RK, Ritchie JT (1987) Image analysis of
video-recorded plant root systems. In: Taylor HM (ed) Minirhizotron observation tubes:
methods and applications for measuring rhizosphere dynamics. ASA Spec Publ 50: 67-80
Tanaka S, Yamauchi A, Kono Y (1995) Easily accessible method for root length measurement
using an image analysis system. Jpn J Crop Sci 64: 144-147
995-1001
Tollner EW, Ramseur EL, Murphy C (1994) Techniques and approaches for documenting plant
root development with X-ray computed tomography. In: Anderson SH, Hopmans JW (eds)
Tomography of soil-water-root processes. Proc Symp of the Soil Science Society of America
(SSSA), Minneapolis, Minnesota, 4 Nov 1992. SSSA Special Publication No 36. Soil Science
Society of America (SSSA), Madison, Wisconsin, pp 115-133
Travis AJ, Murison SD, Chesson A (1993) Estimation of plant-cell wall thickness and cell-size by
image skeletonization. J Agric Sci 120: 279-287
Upchurch DR (1987) Conversion of minirhizotron-root intersections to root length density. In:
Taylor HM (ed) Minirhizotron observation tubes: methods and applications for measur-
ing rhizosphere dynamics. ASA Spec Publ 50: 51-65
Voorhees WB, Carlson VA, Hallauer EA (1980) Root length measurement with a computer-
controlled digital scanning microdensitometer. Agron J 72: 847-851
Wang ZQ, Burch WH, Mou P, Jones RH, Mitchell RJ (1995) Accuracy of visible and ultraviolet
light for estimating live root proportions with minirhizotrons. Ecology 76: 2330-2334
Ward KJ, Klepper B, Rickman RW, Allmaras RR (1978) Quantitative estimation of living
wheat-root lengths in soil cores. Agron J 70: 675-677
Wilhelm WW, Norman JM, Newell RL (1983) Semiautomated X-Y-plotter-based method for
measuring root lengths. Agron J 75: 149-151
Zoon FC, Van Tienderen PH (1990) A rapid quantitative measurement of root length and root
branching by micro computer image analysis. Plant Soil126: 301-308
CHAPTER 11
Computer-Assisted Tomography
and Magnetic Resonance Imaging
S. Asseng 1, L.A.G. Aylmore 2, J.S. MacFalP, J.W. Hopmans 4 and P.J. Gregori
1 CSIRO, Division of Plant Industry, Private Bag, PO, Wembley, Western Australia 6014, Australia
2 The University of Western Australia, Soi! Science and Plant Nutrition, Nedlands, Western
Australia 6907, Australia
3 School of the Environment, Duke University, Durham, North Carolina 27706, USA
4 University of California, Dept. Land, Air and Water Resources, Hydrology Program, 123
Veihmeyer Hali, Davis, California 95616, USA
5 Department of Soil Science, The University of Reading, Whiteknights, PO Box 233, Reading,
RG66DW, UK
CONTENTS
11.2 X-ray Computed Tomography 345
11.2.2 Limitations 348
11.3 Gamma-Ray Computer-Assisted Tomography 351
11.4 Dual-Energy Scanning 353
11.5 Magnetic Resonance Imaging 354
11.6 Future Prospects 358
11.6.1 Conclusions 359
References 360
11.1 Introduction
Until recently, techniques for direct measurements on roots and soil water in
the rooting zone have been largely destructive, or simply lacked the spatial res-
olution necessary for meaningful definition of root and water content distrib-

344 S. Asseng et al.
utions. With advances in X-ray physics, detector technology and mathematical

reconstruction theory, a solution to this problem was essentially achieved by
Hounsfield (1972), who developed the technique known as computer-assisted
tomography (CAT) or computed tomography (CT). In addition to X- and fLrays
(photons), tomographic imaging is also applicable to other types of energy such
as electrons, protons, alpha particles, lasers, radar, seismic and nuclear magnetic
resonance (NMR).
Of particular value is the ability of CT scanners to characterise the inter-
nal structure and the nature of the components present in the soil. For example,
knowledge of the size and distribution of soH pores is important to an under-
standing of soH processes, including water flow and redistribution, aeration,
and root penetration. In particular, the ability to monitor root proliferation
and distribution bya non-invasive technique is essential for detaHed studies of
water uptake by plant roots.
In conventional radiography the transmission of radiation through a three-
dimensional object is used to produce a two-dimensional image of the internal
features of the object on a radiation-sensitive film. Attenuation occurs because
the photons of the incident beam may be absorbed by the material (photoelec-
tric absorption), or may be deflected (Compton scattering), both of which lead
to a decrease in the detected radiation intensity (Fig. 11.1). The final radi-
ographic image represents the spatial variation of radiation attenuation of the
object, perpendicular to the beam. However, it does not provide information on
attenuation variation along the beam direction.
In CT, multiple scans from different angles in a given plane provide a large
number of ray-sums or projections. Using a back projection technique, a two-
dimensional image of a horizontal slice is reconstructed mimerically to yield
the distribution of attenuation coefficients at discrete points within the slice
along the beam path (Aylmore 1993). In the simplest scanning systems these are
obtained using both linear and rotational movements of the source/detector
incident transmitted
,....------..,
-te
absorbed
scattered
Fig. 11.1. Attenuation of a narrow beam of radiation by absorption and scattering. (After
Aylmore 1993)
11 Computer-Assisted Tomography and Magnetic Resonance Imaging 345
system. A complete set of parallel ray-sums represents a single projection for

that view of the cross-sectionallayer of the object. By either moving the object
or the source plus detector, multiple parallel slices yield three-dimensional
images. Alternatively, cone beam imaging in combination with a two-
dimensional detector array allows three-dimensional imaging using a single
object rotation.
Application of CT to X- and y-ray attenuation and nuclear magnetic reso-
nance has provided exciting new methods for the non-destructive imaging of
roots and soils to study root behaviour and soil-plant-water relations in space
and time. In this chapter we discuss the principles of X- and y-ray CT scanning,
the use of dual-energy measurements and nuclear magnetic resonance imaging
and their application and limitations in soil and root studies.
11.2 X-ray Computed Tomography
X-ray beams, generated by electrons, impinge upon a suitable target material.

The beam is polychromatic in nature with a photon energy typically normally
distributed. The subsequent attenuation of an X-ray beam as it passes through
a medium depends on its electron density and packing density, and the energy
of the radiation. Since X-ray CT systems usually do not exceed 140kV excita-
tion, photoelectric absorption is the main mechanism of absorption in soil
materials.
The ability of the scanner to resolve voids and objects of different size and
spacing (spatial resolution) is a function of the picture element size (pixel or
voxel dimensions). The array of attenuation values is graphically displayed as
an image on a high-resolution monitor. Although some scanners allow voxel
sizes from 1.0 x 1.0 x 8.0mm3 to as small as 0.1 x 0.1 x 0.5mm\ even smaller
spatial resolutions of 1O-25,um can be achieved using smaller source beam
diameters and recent developments in detection technology (Steude et al. 1994).
Attenuation coefficients are converted to an internationally standardised scale,
which is expressed in Hounsfield units (HU), where the attenuation of the
scanned medium is measured relative to that of pure water. Optimal contrast is
achieved while maximising the difference in HU-values between the features in
the image, e.g., such as occurs between a water-containing root growing in dry
soil material.
11.2.1 Applications
Numerous workers (Petrovic et al. 1982; Hainsworth and Aylmore 1983,1986;

Crestana et al. 1985; Anderson et al. 1988; Tollner et al. 1987, 1994; Steude et al.
Fig.11.2. An image of 14-day-old bush bean (Phaseolus yulgaris 1.) roots in situ, derived from
a high-energy industrial X-ray CT. (After Heeraman et al. 1997; with kind permission from
Kluwer Academic Publisher)
1989; Vaz et al. 1989; Hopmans et al. 1992) have demonstrated that conventional
X-ray medical scanners can provide information otherwise not obtainable on
the spatial distributions of bulk density and water content in soil columns and
in particular that associated with plant roots (Hainsworth and Aylmore 1983,
1986; Aylmore and Hamza 1990; Hamza and Aylmore 1991, 1992a, b).
Total root length has been estimated from an image of a plant root system
in situ (Fig. 11.2) with a resolution of 0.16mm by Heeraman et al. (1997;
Box 11.1). The image was derived with a high-energy industrial X-ray CT
system. Although alternative CT techniques using, for example, cone beam
geometry in combinat ion with an areal detector array, could potentially
improve the spatial resolution to 5-20 Jlm, it would not necessarily make
possible the imaging of very fine roots of this size. First, it would greatly
reduce the maximum sample size to 1 cm diameter or smaller. Secondly, when
reducing the voxel size further, it would become much more difficult to
separate roots from the soil matrix, as roots and water-filled pores have similar
attenuation values.
BOX 11.1. Root Length Estimate with a High-energy Industrial X-ray

CT System
In an experiment by Heeraman et al. (1997), bush bean plants were trans-
planted to 5-cm-diameter PVC columns containing a fine sandy soiI
maintained at field capacity, after pre-germinating the seeds on paper
towel. Plants were allowed to grow for 14 days in a controlled growth
chamber. Three plants were destructively sampled and analyzed for root
growth parameters, including root length, root dry weight, specific root
length and root volume. One plant was used for the CT measurements. The
stern of the plant was excised at its base, and the root system was imaged
with a high-energy (420keV) third generation industrial X-ray CT scanner
(Steude et al. 1994). The scanner was equipped with a linear detector array
of 125 detectors with each detector being 0.2109 x 0.2109 mm in size. Final
spatial resolution after reconstruction was 0.16mm. Horizontal slices
(tomograms) were obtained by rotating the object; the time to scan a single
slice was 6.2 min. Forty slices were obtained for a total rooting depth of 8
mm (1.2-2cm depth below the soiI surface). Three-dimensional images
were obtained by combining a series of 40 individual tomograms into a
three-dimensional data set of 8 mm height. The three-dimensional data set
was visualized using a modified version ofVIS-5D vers ion 4.0 (Fig. 11.2).
The volumetric rendering image of roots at the 1.2- 2 cm depth uses
this imaging software to construct a three-dimensional isosurface of
attenuation coefficients. Total root length was estimated from root volume,
assuming that the roots are cylindrical. Total root volume as estimated
from CT was 2.97 cm3, whereas destructive root measurements yielded a
total root volume of 2.2 cm3 for the same depth interval. Total root length
as estimated from destructive and CT measurements were 8.65 and 11.7 m
for the 1.2-2.0 cm depth. The overestimation of root volume using CT can
be caused by signal noise and volume averaging assumptions.
An alternative means of improving the spatial resolution while still retain-

ing the ability to scan large samples isto use a microfocal X-ray beam and to
employ"area of interest" scanning techniques. Columns can be scanned rapidly
at several angles to identify areas where roots are likely to be present and then
those areas can be scanned in more detail. This procedure has the advantages
of reducing scanning time and data storage while improving resolution.
Another example of the application of X-ray CT in root research is pre-
sented in Fig. 11.3 from Hamza and Aylmore (1992b) which shows a cross
section through a lupin plant root and surrounding soiI. The total number of
pixels, mean, standard deviat ion and root mean square deviat ion of pixel values
348 s. Asseng et al.
Fig. 11.3. X-ray CT scan showing
a cross-section through a lupin
plant root and surrounding soil.
Hounsfield unit values vary
throughout the section of the root.
The root diameter is about 4 mm.
(After Hamza and Aylmore 1992b)
in the selected region, its area and volume, and the distance between two points
are instantly available.
In root water uptake studies by Hainsworth and Aylmore (1983, 1986, 1989),
changes in volumetric water associated with single radish roots in a non-
swelling 15% kaolinite clay and 85% sand mixture provided the first detailed
observations of this type. Uptake of water along the radish roots was clearly
shown to be non-uniform with depth, with the roots first removing water from
the top, and later drawing water at greater depths, as soil hydraulic resistance
became a major limiting factor in the upper soillayer (Hainsworth and Aylmore
1989). Subsequently, Aylmore and Hamza (1990) and Hamza andAylmore (1991,
1992a, b) used a combinat ion of CT scanning and a Na-ion specific microelec-
trode technique to measure concomitantly the spatial distribution of soil water
content and Na+ ion concentration in close proximity to lupin and radish plant
roots (Box 11.2). Figure 11.4 illustrates the detail and accuracy with which the
CT technique was able to resolve the changes in water content by root water
uptake at different positions along the root resulting from differences in water
content and electrolyte. Recently, Hamza et al. (1996) have used CT to study the
spatial dis tribut ion of water content in the vicinity of interacting multiple radish
roots and the effects of water stress on root shrinkage and recovery.
11.2.2 Limitations
A number of fac tors impose limitations on the applicability of X-ray CT which

are related to spatial resolution, energy absorption, and changing soil bulk
density. For example, the true spatial resolution value in the study by Warner
BOX 11.2. Monitoring SoU WaterContent and Na+ Concentrations Near

Plant Roots
A combination of computer-assisted tomography applied to X- and y-ray
attenuation and Na+ liquid ion-exchanger microelectrodes was used by
Hamza and Aylmore (l992a,b) to monitor soH water contents and Na+ ion
concentrations at root depths of 3, 6 and 9 cm and at zero, 2, 4, 6, and 8 h
intervals from the diurnal commencement of transpiration. The plants
were subjected to two levels of transpirational demand and five Na+ soil
solution concentration levels.
The scan times took upwards of l.4s. The image was displayed
immediately following the scan on a video screen in a 512 x 512 matrix
of 2 x 2mm pixels (Fig. 11.3). The software provided with the scanner
(Somatom DR/H scanner with SCINTILLARC 700) was able to magnify
any area in the scanned slice and pinpoint any pixel (from the root center
to the bulk soH) to determine its water content automatically through
determining its attenuation coefficient. This could be done in very smalt
distance intervals (i.e. 0.1 mm).
The software was also able to perform some statistical functions to
calculate the total water content (through Hounsfield units) of the whole
or part of the scanned slice, thus making it easy to calculate water uptake
by the plant root for any given section and time. Measurable differences
in water content of 0.006 g/cm 3 can be readHy detected by both X- or r-
ray CAT scanning (Hainsworth and Aylmore 1983, 1988).
In combination with the monitoring of soH water content, Na+-LIX
microelectrodes were used, in situ, to determine Na+ concentrations at
the root surface. For this purpose soil pots were specifically designed and
constructed to allow accurate positioning of the microelectrode tips at
the root surface (Hamza and Aylmore 1991). To ensure root/electrode
contact, a 0.1 mm-diameter hole was made from the surface of the soH to
the bottom to encourage the root subsequently planted, to follow. The
microelectrode shaft was manoeuvred inside a slightly larger plastic
tube. Special attention had to be paid to prevent the microelectrode tip
from penetrating the root surface where it would read the cortex or
xylem concentration instead of that at the root surfaee. The
microelectrode pots were amenable to CAT scanning so that the water
drawdown in any layer, and in particular that associated with the
microelectrode, could be monitored eontinuously.
et al. (1989) was about twice the spatial resolution determined by the scanner.
This results from the volume averaging that oecurs if the interface of two or
more materials with different attenuation values are present within a voxel. For
0.30 Fig. 11.4. Water drawdowns
near top, middle, and
o~
0.25 bottom sections of a single
..
Ci radish root after 2 h of tran-
~
c spiration measured by CT
J!!c 0.20
scanning. (After Hamza and
O
...u Aylmore 1992a)
! 0.15
==
0.10
O 2 4 6 8 10 12
Distance from root surface (mm)
example, if two materials each occupy half of the volume of a voxel (e.g., a root
surrounded by a soil matrix), then the attenuation of the voxel will be the arith-
metic average of the two. Consequently, studies have been largely limited to soils
having macropores of a dimension equal to or larger than the voxel size.
Another limitation of X-ray CT in soils arises from the selective absorption
oflow-energy X-rays. The energy spectrum of X-ray scanners is polychromatic,
and preferential absorption of the lower energy X-rays occurs as the beam tra-
verses the soil. This process is referred to as "beam hardening" (Herman 1980).
Physical or mathematical filters can partially correct for beam hardening
(Brooks and Di Chiro 1976), but such artefacts will undoubtedly occur in soils
when using relatively low-energy X-rays.
When only very small differences in attenuation exist between an object
and the surrounding soil, the ability to accurately resolve the object (imaging
contrast) is also reduced. While X-ray CT provides excellent resolution for some
studies, its usefulness in soil systems has invariably been restricted by its
inability to distinguish between changes in water content and bulk density
in swelling soils. Consequently, the applications of X-ray CT to quantitative
soil-water studies in swelling soils have been largely limited to statistical
assessments of macroporosity distributions before and after complete wetting
and drying cyeles (Phogat and Aylmore 1989), and the measurement of water
uptake in proximity to plant roots in non-swelling soils (Hainsworth and
Aylmore 1986; Aylmore and Hamza 1990; Hamza and Aylmore 1992a, b). Even
in non-swelling soils, however, localised changes in bulk density may occur
elose to roots caused by the root displacing partieles when elongating. These
changes are not detected by current X-ray systems' and may lead to errors in
estimates of water uptake.
Their quantitative limitations, great expense and low accessibility have
seriously limited the use of medical X-ray scanners for soil-water-plant studies.
Moreover, these scanners are designed to monitor patients in a horizontal posi-
tion, and their construction is not generally suitable for studies involving plants
growing in vertically-positioned soil columns. Finally, the proprietary nature of
these commercial systems makes software modifications difficult.
11.3 Gamma-Ray Computer-Assisted Tomography
Because of the quantitative limitations and restricted accessibility of X-ray

systems, work in several laboratories has concentrated on potentially more
suitable and less expensive CT systems using y-ray sources (Hainsworth and
Aylmore 1983, 1988; Crestana et al. 1986; Brown et al. 1993). Gamma-ray sources
range in strength from a few millicuries to 500 Ci and have been used to obtain
useful tomographic images in acceptable scanning times, even with a source
brightness which is several orders of magnitude less than that produced by X-
ray sources (MacCuaig et al. 1986). The use of y-sources where the energy spec-
trum is monochromatic, eliminates the beam hardening and absorption edge
problem. Gamma-ray sources also offer additional advantages compared with
X-ray tubes, including lower cost, compactness, portability and easy access to a
very wide range of photon energies.
A block diagram illustrating the operation of a r-CT system (Aylmore 1993)
is shown in Fig. 11.5. The main body consists essentially of two platforms, with
one allowing for vertical motion of the source/detector system, and the other
providing linear translation and rotational motions of the object to be scanned.
detector electtonlcs scenner control COf'I1)UIer
mollon platrorm pnnting data analysis computar storage
Fig.11.5. Block diagram of r-ray CT scanning system constructed at the University ofWestern
Australia, consisting of three subsystems: computer, data acquisition, and mot ion control
systems. (After Aylmore 1993)
Fig. 11.6. Lupin root in

a soil column. Three-
dimensional image
reconstruction obtained by
subtractive imaging using
a r-ray CT. (After Aylmore
1993)
The main platform supports the lead shielding for the source and scintillation
detector. The CT scan platform is placed at a fixed level in an opening in the
middle of the y-platform, so that the y-platform can move independently of the
CT scan platform. This system is capable of scanning columns up to 10 cm in
diameter and 150 cm in length. The systems developed by Crestana et al. (1986)
and Brown et al. (1993) use a similar parallel beam approach.
11.3.1 Applications
Gamma attenuation measurements have been used for many years to monitor
changes in average bulk density and water content in sections through soil
columns (Gurr 1962; Groenvelt et al. 1969; Ryhiner and Pankow 1969).
The application of CT to y-attenuation is illustrated in Fig. 11.6, showing
how the development of a lupin tap root in a soil column can be accurately
defined by subtractive imaging, thereby eliminating the soil's contribution to
the image.
11.3.2 Limitations
The major disadvantage of y-scanners is their low source brightness (Aylmore

1994). The relatively low photon emission from y-sources as compared with X-
ray sources requires much longer scanning times and larger voxel sizes (5 mm 3,
Brown et al. 1993), and measurements are largely restricted to slowly changing
or steady state processes. As with X-ray cr, measurements of the spatial dis-
tribution of water content in soils by a single 'fsource is limited to uniform and
non-swelling soils.
11.4 Dual-Energy Scanning
A major limitation of the use of single energy X- or 'fray cr scanning has been
the necessary assumption of uniform or constant bulk density. As soil aUenua-
tion is a function of both bulk density and volumetric water content, an accu-
rate determination of water content in soils by these methods is impossible
if the bulk density changes during the experiment (Petrovic et al. 1982;
Hainsworth andAylmore 1983; Anderson et al. 1988; Phogat andAylmore 1989).
Even in non-swelling soils, the general non-uniformity of packed soil columns
makes accurate measurement of water content distribution near plant roots
difficult, since the technique requires exact positioning of the soil column in a
sequence of scans.
ro monitor changes in the spatial distribution of both bulk density and
water content in swelling and shrinking soils, independent estimates of at-
tenuation associated with both bulk density and water content with different
energies can be combined. In dual energy 'fCr scanning, a low-energy and a
high-energy 'fsource have been used (e.g., Cs-137 and Am-235 or Yb-169;
Aylmore 1993).
11.4.1 Applications
Effective studies of plant root development and water uptake demand the ability
to accurately and simultaneously monitor changes in water content and bulk
density in close proximity to the roots. Phogat et al. (1991) demonstrated that,
using dual-source 'fCr scanning, it is possible to measure the spatial distribu-
tions of bulk density and water content in swelling soils simultaneously and
non-destructively with a satisfactory level of precision.
11.4.2 Limitations
Unfortunately, the relatively long scanning time required has limited the effi-
cient application of 'fcr to plant root studies. For example, the study of Phogat
et al. (1991) showed that about 169 s was needed for an individual ray-sum. In
their study, 112 h were needed to complete a dual-source scan, which resulted
in an average standard deviation in pixel water content values of 0.025 cm3. Such
large counting times limit its applicability to steady-state systems only.
11.5 Magnetic Resonance Imaging
An alternative approach to studying soil water extraction by plants uses proton

CH) nuclear magnetic resonance imaging (MRI), as presented in Woods et al.
(1989). Nuclei suitable for MRI are characterised by a non-zero spin quantum
number 1 and include IH, 2H, l3C, 31p, 17 0, 19F, 23Na, 27AI, and 31p. Of the NMR
detectable nuclei, IH is intrinsicaUy the most sensitive to detection. It is for this
reason that most imaging experiments focus on IH, with biological and/or soil
water being the most abundant detectable pool of nuclei. In practice, the sample
is surrounded with a radiofrequency (rf) coil, and placed into the bore of a
superconducting magnet. In a uniform magnetic field aU protons resonate at
the same specific frequency (Larmor frequency), which is determined by
magnetic field strength. FoUowing placement in the static magnetic field, the
rf transmitter is used to apply a pulse of radio waves at the Larmor frequency.
This generates a magnetic field, designated Bl, which can be sensed by the
nuclei, causing them to "tip sideways". The degree of tipping is dependent
on the strength and duration of the applied rf pulse (Fig. 11.7). When the rf
radiat ion is turned off, the spins r~align with the static magnetic field. As they
re-align, energy from the transverse magnetisation is detected as an AC current
bya receiver coil placed near the specimen. Detected signals are then amplified,
filtered and sent to a computer for digitisation and recording of the signal
frequency, amplitude and decay. If a magnetic field gradient is applied, protons
at different locations will resonate at different frequencies. If three orthogonal
Fig. 11.7. Vector diagram showing the

orientation of the primary magnetic field (Bo)
which is created by the superconducting
magnet, and the perpendicular field created
by the radio frequency pulse (Bl) . Following
placement in the magnet bore, nuclei align
themselves so that they process in alignment
with Bo
gradients are applied, analysis of the different frequencies can be used to

generate a proton density map or image. By changing the three gradients it
is possible to control such variables as slice position, slice thickness and
resolution, thus producing a magnetic resonance image. These differences in
detected signals provide "contrast" to the image, or the ability to distinguish
different physiological, biochemical or anatomical properties within a speci-
men, including the water distribution in soil surrounding roots as well as
roots themselves.
There are basically five fundamental parameters in nuclear magnetic reso-
nance (NMR) studies which will determine signals detected from a localised
region of tissue, the voxel. The fundamental parameters are spin density (how
many detectable nuclei are present within a voxel, representing volumetric
water content), spin-Iattice relaxation (TI), spin-spin relaxation (T2), experi-
mental spin-spin relaxation (T2*), and movement (e.g. diffusion and flow).
Spin-Iattice relaxation and spin-spin relaxation are functions of the degree of
interaction of the water with its external molecular environment. For example,
soil type affects the binding energy of the water at the liquid-solid interface
which may depend on the chemical composition of the soil and soil solution
as well. The wide range of properties which determine detection ability and
image contrast makes MRI an extremely flexible and powerful technique, but
they also greatly complicate the image interpretation. Choice of image acquisi-
tion protocol will greatly change the appearance of an image and provides
a much wider range of applications than X-ray CT scans, y-detection or dual-
energy scanning.
11.5.1 Applications
It has been demonstrated that large roots (Omasa et al. 1985) and seedlings
(Rogers et al. 1985) can be detected in soil by MRI. In an extensive examination
of the MRI properties of soil, plant roots were imaged when grown in 30 dif-
ferent agricultural and 8 artificial soils. Image quality and the ability to detect
roots suffered significantly when the ferromagnetic particle content was larger
than 4%. Below 4%, there was still significant variation in the suitability of a
soil as a root imaging substrate. Seven native soils, however, presented excellent
delineation of the root systems, suggesting that MRI holds great promise for
the non-destructive study of roots within a soil matrix (Bottomley et al. 1986;
Rogers and Bottomley 1987). Magnetic resonance imaging has also been
demonstrated to give information on the pore structure and wettability of
natural sandstone, providing visualisation of wetting heterogeneities (Guillot et
al. 1994). The TI relaxation time of natural sandstone was shown to increase
significantly with salinization, changing the wettability of the surface from
moderately hydrophilic to hydrophobic. Similarly, the binding properties of

water, as reflected by the relationship between relaxation time and water
content, have been studied for samples of fine sand. When moistened with dis-
tilled water, the TI values increased from 472ms at 5% (w/w) to 1265ms at 25%
(w/w) water. The increase was linear between 5 and 20%, with a slight reduc-
tion in slope between 20 and 25%. The water potential measured by the pres-
sure plate method showed the water potential ranged from 0.01 to O.OMPa (J.S.
MacFall 1990, unpubl. data), indicat ing water was not tightly held by the sand
grains. Further studies with a variety of sand samples of the same approximate
particle size showed a range of TI relaxation times, rang ing from 565 to 1474
ms when the sample contained 25% water (MacFall and Johnson 1994a). In an
early study, MacFall et al. (1991) used a series of reference tubes (sand phan-
toms) filled with acid-washed sand at various water contents to provide a rapid
reference calibration curve. This was used to monitor water uptake by Loblolly
pine seedlings from a fine sand and to quantitatively compare the relative
efficiencies of fine, lateral and taproots in water uptake.
Similar experiments can be conducted for other plants, examining effects
of stress, cultivar comparisons, competition, etc. Image acquisition protocols
have also been developed to enhance signals detected from functional roots,
suppressing signals from water in the surrounding soil/sand. This approach
allows enhanced visualisation of the root system, potentially providing new
tools to study root water content, conductance, growth, distribution, architec-
ture and turnover. In an early work, Bottomley et al. (1986) used a spatial res-
olution of 0.6 mm with an unidentified slice definit ion to observe the movement
of a dilute solution of CuS0 4 (a contrast agent which enhanced the detected
signal) into and through roots of Vicia faba. Brown et al. (1986) were able to
differentiate anatomical regions of the Pelargonium hortorum roots with a
spatial resolution of 0.1 mm x 0.1 mm x 1.2 mm without the use of contrast
agents. In other work, a synthetic soil mix was developed which gave a small
magnetic resonance detectable signal while supporting plant root growth
(Brown et al. 1991).
These image acquisition protocols have been extended from only two-
dimensional image sets to three-dimensional acquisition, reconstruction and
views of intact roots in a soil container (MacFall and Johnson 1994a). Images
acquired with specially tailored pulse sequences provide sufficient contrast
between roots and surrounding soil or sand to allow digital extraction of the
root system or segments from the data set. Roots can be extracted either
through use of a seeding algorithm or simply by adjusting the viewing opacity
and window/level during rendering of the three-dimensional volume (MacFall
and Johnson 1994b; Fig. 11.8). This approach has allowed not only visualisation
but quantitative measurement of changes in root volume, surface area, and total
Fig. 11.8. Magnetic resonance images acquired of a pine seedling root system at two points in
time. Views A (side view) and C (top view) were acquired soon after plants were transplanted
into fine sand. Views B (side view) and D (top view) were acquired 6 weeks after the initial
imaging experiments and ciearly show the extent of root growth over this period. (J.S. Mac Fali,
unpubl. data)
root length of the root systems of potted Loblolly pine seedlings over a period
of several months. Similar studies can be done with other plants, allowing study
of many topics such as pathogen effects, root turnover, comparative growth
rates, etc. More recent work with pulse sequence parameter optimisation has
demonstrated that three-dimensional root images can be acquired, both of
intact, entire root systems within a container and of roots within intact field soil
cores, further enhancing the potential utility of the technique. Roots can also
be studied in a solid, synthetic matrix. Root decay was observed by three-
dimensional imaging techniques of detopped root systems embedded in plaster
of paris. As roots decayed, preferential patterns of water flow could be observed
following the root channels (Liu et al. 1994).
11.5.2 Limitations
The presence of paramagnetic elements such as Fe2+, Fe3+, Mn2+ and Cu2+ at levels
commonly found in soils may cause interference, thus requiring plants to be
grown in porous media other than soil or in synthetic soil mixes (Cassel and
Nielsen 1994).
One major difficulty is in the adsorption of water to soil particles such as
clays. The signal becomes non-detectable when water becomes tightly bound to
a solid substrate and less mobile, creating difficulties in quantitatively measur-
ing water which is physically bound within the soil matrix (MacFall et al. 1991).
In many cases, however, when the water content becomes low enough that
detection becomes limiting, the water potential is also so low that the remaining
water would be unavailable for plant uptake. Preliminary work has shown a cor-
relation between water content (and relaxation time) with soil water potential
(J.S. MacFall, unpubl.), therefore, from the perspective of water flux, the inability
to detect bound water may simply reflect a limiting soil water potential.
Another constraint to the application of MRI is the available scanner
time. For the acquisition of a three-dimensional data set, the imaging time
required is the product of the time for each slice and the number of slices in
the three-dimensional image set. For measurement of water depletion zones,
the acquisition time may be up to 8 h, and if the plant is rapidly transpiring,
there may be substantial changes in the shape of the depletion zone in this
period. Nevertheless, as rapid pulse sequences (e.g. fast spin echo) are devel-
oped for plant imaging applications, much shorter acquisition periods will be
required.
Similar to X- and y-ray CT, most existing measuring systems are non-
portable and plants have to be grown in soil cores or taken as core samples from
the field. The size of MRI systems limits the size of the sample, with sample sizes
ranging from <1 cm to 80 cm diameter depending on the opening size of the
magnet and field strength. Generally, the larger the sample which can be accom-
modated, the larger the voxel sizes comprising the image. A limitation of MRI
is also the high cost of the system.
11.6 Future Prospects
Industrial CT systems designed for object size/weight/density flexibility and

high sensitivity, and not restricted by object motion or radiation dose level, have
recently been developed [e.g., at Advanced Research and Applications Corpo-
ration (ARACOR), Sunnyvale, California, USA, and Surrey Medical Imaging
Systems (SMIS), Surrey, UK]. These include systems capable of inspecting
objects up to 2.4 m in diameter by 5 m long and weighing up to 49500 kg.
Reduction in scanning times to allow more rapid monitor ing of changes in

soil water content would seem a priority for soil and plant studies. Recently, to
reduce scanning time, fourth generation fixed-detector and rotating source
scanners have been developed, in which a large number of detectors are
mounted on a fixed ring and an X-ray tube inside this ring continually rotates
around the object. Furthermore, improved dimensional resolution will enhance
structural definit ion in soil systems. X-ray CT systems have been developed
recently to provide l,um spatial resolution [e.g., synchrotron X-ray computed
macrotomography (CMT), Brookhaven National Synchrotron Light Source X26
beam line, Spanne et al. 1994]. However, current pixel dimensions of the order
of 0.5 to 1 mm are quite adequate to allow meaningful resolution of many of the
controversies associated with water extraction by plant roots. CT scanning has
the potential to provide detailed volume distributions of water content and
potential throughout a complete plant root system and to allow the efficiency
of different parts of the root system in extracting water and interaction between
adjacent roots to be examined. Unfortunately, attempts in this direction (L.A.G.
Aylmore and P.I. Gregory, unpubl.; A. Reid, R.D. Schuller and L.A.G. Aylmore,
unpubl.) have, as yet, been largely confounded by difficulties in obtaining
uniform packing, compensating for variations in soil bulk density and accuracy
of repositioning of the column in the scanner. Future portable CT systems will
allow scans under field conditions. For example, Onoe et al. (1983) described
the use of a portable X-ray CT scanner for measuring annual growth rings of
live trees.
In view of their substantially lower cost and some superior quantitative
characteristics, such as the absence of beam hardening and absorption edge
problems, y-ray tomographic systems are likely to become a useful tool for soil
and plant studies.
Although MRI has its limitations as well, particularly in the study of high
Fe-containing soils, it clearly has enormous potential in the study of plant
physiological processes and water uptake by plant roots. The ftexibility of the
imaging protocols provides the tools to study not only root distributions and
anatomy, but to probe physiological function. Low cost, portable units for vet-
erinary applications are currently in development, and these may open the
doors for greater utility in field applications and to a wider range of users. The
development of broad-line MRI techniques may also allow sub-millimetre
spatial resolution of water distribution in soils and plants.
11.6.1 Conclusions
Possible applications, advantages and present constraints of X-ray, y-ray, dual

energy systems and MRI for root related studies are summarised in Table 11.1.
Table 11.1. Summarised applications, advantages and present constrains of X-ray, J-ray, dual
energy systems and MRI for root related studies. Note that an applications refer to studies with
soil columns
System Applications Advantages Limitations
X-ray CT Dynamics of roots, High resolution Beam hardening and

soH water or bulk Fast scanning edge absorption
density Requires uniform
soH density
Requires non-swelling
soHs
Gamma-ray Steady state roots, No beam hardening Requires uniform

CT soH water or soH and edge absorption soH density
density problems Requires non-swelling
Relative low costs soHs
Low resolution
Slow scanning
Dual energy Soil water content Can simultaneously Very slow scanning
scanning and bulk density quantify changes
in soH water content
and bulk density
(swelling soils)
MRI Dynamics of root High resolution Requires absence

structure and soH No soH density of paramagnetic
water effects elements and high
clay contents
Slow scanning
Despite some of the current limitations there is no doubt that the application
of these innovative techniques will, with further developments, provide major
tools for soi! and plant scientists and have the potential to resolve major con-
troversies with respect to the dynamics of plant root systems and water uptake
by plant roots.
Acknowledgements. We thank Drs D. Rolin and A. Perry for helpful discussions,

Drs K. Shackel and A. McNeill for comments on the manuscript and the Grains
Research and Development Corporation for financial support.
References
Anderson SH, Hopmans JW (eds) (1994) Tomography of soil-water-root processes. Proc Symp
sponsored by Division S-1 and S-6 of the SoH Science Society of America in Minneapolis,
Minnesota,4 November 1992. SSSA Special Publication No 36. American Society of Agron-
omy, Soil Science Society of America, Madison, Wisconsin
Anderson SH, Gantzer CJ, Boone JM, Tully RJ (1988) Rapid non-destructive bulk density
and soil water content determination by computed tomography. Soil Sci Soc Am J 52:
35-40
Aylmore LAG (1993) Use of computer-assisted tomography in studying water movement around
plant roots. Adv Agron 49: 1-54
Aylmore LAG (1994) Application of computer-assisted tomography to soil-plant-water studies
- an overview. In: Anderson SH, Hopmans JW (eds) Tomography of soil-water-root
processes. Proc Symp sponsored by Division S-l and S-6 of the Soil Science Society of
America in Minneapolis, Minnesota, 4 November 1992. SSSA Special Publication No 36.
American Society of Agronomy, SoH Science Society of America, Madison, Wisconsin, pp
7-16
Aylmore LAG, Hamza M (1990) Water and solute movement to plant roots. In: Transactions 14th
Congr of Soil Science. Kyoto, Japan, 12-18 Aug. ISSS, Wageningen, pp 124-129
Bottomley PA, Rogers HH, Foster TH (1986) NMR imaging shows water distribution and
transport in plant root systems in situ. Proc Natl Acad Sci 83: 87-89
Brooks RA, Di Chiro G (1976) Principles of computer-assisted tomography (CAT) in
radiographic and radioisotopic imaging. Phys Med Bio121: 689-732
Brown DP, Pratum TK, Bledsoe C, Ford ED, Cothern JS, Perry D (199l) Noninvasive studies
of conifer roots: nuclear magnetic resonance (NMR) imaging of Douglas-fir seedlings. Can
J For Res 21: 1559-1566
Brown GO, Stone ML, Gazin JE (1993) Accuracy of gamma ray computerised tomography in
porous media. Water Resour Res 29: 479-486
Brown JM, Johnson GA, Kramer PJ (1986) In vitro magnetic resonance microscopy of changing
water content in Pelargonium hortorum roots. Plant Physiol82: 1158-1160
Cassel DK, Nielsen DR (1994) The realization of a dream. In: Anderson SH, Hopmans JW (eds)
Tomography of soH-water-root processes. Proc Symp sponsored by Division S-l and S-6 of
the Soil Science Society of America in Minneapolis, Minnesota, 4 November 1992. SSSA
Special Publication No 36. American Society of Agronomy, Soil Science Society of America,
Madison, Wisconsin, pp 1-6
Crestana S, Mascarenhas S, Pozzi-Mucelli RS (1985) Static and dynamic three-dimensional
studies of water in soil using computed tomographic scanning. SoH Sci 140: 326-332
Crestana S, Cesareo R, Mascarenhas S (1986) Using a computer-assisted tomography miniscan-
ner in soil science. SoH Sci 142: 56-61
Groenvelt PH, De Swart JG, Cisler J (1969) Water content measurement with 60keV gamma ray
attenuation. Bull Int Assoc Sci Hydrol14: 67-78
Guillot G, Chardaire-Riviere C, Bobroff S, Le Roux A, Roussel JC, Cuiec L (1994) Characterisa-
tion of wetting heterogeneities in sandstone rocks by MRI. Magn Reson Imaging 12:
365-368
Gurr CG (1962) Use of gamma rays in measuring water content and permeability in unsaturated
columns of soHs. Soil Sci 94: 224-229
Hainsworth JM, Aylmore LAG (1983) The use of computer-assisted tomography to determine
spatial distribution of soil water content. Aust J Soil Res 21: 435-443
Hainsworth JM, Aylmore LAG (1986) Water extraction by single plant roots. SoH Sci Soc Am J
50:841-848
Hainsworth JM, Aylmore LAG (1988) Application of computer-assisted tomography (CAT) to
gamma attenuation measurement of soH water content. Aust J Soil Res 26: 105-110
Hainsworth JM, Aylmore LAG (1989) Non uniform soil water extraction by plant roots. Plant
Soi11l3: 121-124
Hamza M, Aylmore LAG (1991) Liquid ion exchanger microelectrodes used to study soil solute
concentrations near roots. Soi! Sci Soc Am J 55: 954-958
Hamza M, Aylmore LAG (1992a) Soi! solute concentration and water uptake by single lupin and
radish plant roots.1. Water extraction and solute accumulation. Plant Soi! 145: 187-196
Hamza M,Aylmore LAG (1992b) Soi! solute concentrat ion and water uptake by single lupin and
radish plant roots. II. Driving forces and resistances. Plant Soi1145: 197-206
Hamza MA, Anderson SH,Aylmore LAG (1996) Studies of root function using computer-assisted
tomography. Proc ASSSI and NZSSS National Soi!s Conf, Melbourne, pp 97-98
Heeraman DA, Hopmans JW, Clausnitzer V (1997) Three-dimensional imaging of plant roots
in situ with X-ray computed tomography. Plant Soi1189: 167-179
Herman GT (1980) Image reconstruction from projections: the fundamentals of computed
tomography. Academic Press, New York
Hopmans JW, Vogel T, Koblik PD (1992) X-ray tomography of soi! water distribution in one-step
outfiow experiments. Soil Sci Soc Am J 56: 355-362
Hounsfield GN (1972) A method of an apparatus for examination of a body by radiation such
as X- or gamma-radiation. British Patent No. 12839l5. British Patent Office, London
Liu IY, Waldron LI, Wong STS (1994) The application of nuclear magnetic resonance imaging to
study preferential water fiow through root channels. In: Anderson SH, Hopmans JW (eds)
Tomography of soi!-water-root processes. Proc Symp sponsored by Division S-l and S-6 of
MacCuaig N, Tajuddin AA, Gi!boy WB (1986) Industrial tomography using a position-sensitive
carbon-fibre ano de proportional counter. Nucl Instrum Methods 242: 620-625
MacFall JS, Johnson GA (1994a) Use of magnetic resonance imaging in the study of plants and
soils. In: Anderson SH, Hopmans JW (eds) Tomography of soil-water-root processes. Proc
Symp sponsored by Division S-l and S-6 of the Soil Science Society of America in Min-
neapolis, Minnesota, 4 November 1992. SSSA Special Publication No 36. American Society
of Agronomy, Soil Science Society of America, Madison, Wisconsin, pp 99-114
MacFall JS, Johnson GA (1994b) The architecture of plant vasculature and transport as seen with
magnetic resonance microscopy. Can J Bot 72: 1561-1573
MacFall JS, Johnson GA, Kramer PJ (1991) Comparative water uptake by roots of different ages
in seedlings of Loblolly pine (Pinus taeda L.) New Phytol1l9: 551-560
Omasa K, Onoe M, Yamada H (1985) NMR imaging for measuring root system and soil water
content. Environ Control Biol 23: 99-102
Onoe M, Tsao JM, Nakamura H, Kogure J, Kawamura H, Yoshimatsu M (1983) Computed tomog-
raphy for measuring annual rings oflive trees. Proc IEEE 71: 907-908
Petrovic AM, Siebert JE, Rieke PE (1982) Soil bulk density analysis in three dimensions by com-
puted tomographic scanning. Soil Sci Soc Am J 46: 445-450
Phogat VK, Aylmore LAG (1989) Evaluation of soil structure by using computer-assisted tomog-
raphy. Aust J Soil Res 27: 313-323
Phogat VK, Alymore LAG, Schuller RD (1991) Simultaneous measurement of the spatial distri-
bution of soil water content and bulk density. Soil Sci Soc Am J 55: 908-915
Rogers HH, Bottomley PA (1987) In situ nuclear magnetic resonance imaging of roots: influ-
ence of soil type, ferromagnetic partide content, and soi! water. Agron J 79: 957-965
Rogers HH, Bottomley PA, Foster TH (1985) Application of nuclear magnetic resonance imaging
to plant root studies. Proc Int Conf Soil Dynam 5: 1152-1157
Ryhiner AH, Pankow J (1969) Soil moisture measurement by the gamma transmission method.
J Hydrol 9: 194-205
Spanne P, Jones KW, Prunty L, Anderson SH (1994) Potential applications of synchrotron com-
puted microtomography to soil science. In: Anderson SH, Hopmans JW (eds) Tomography
of soil-water-root processes. Proc Symp sponsored by Division S-l and S-6 of the Soil
Science Society of America in Minneapolis, Minnesota, 4 November 1992. SSSA Special
Publication No 36. American Society of Agronomy, Soil Science Society of America,
Steude JS, Hopkins F, Anders JE (1994) Industrial X-ray computed tomography applied to soil
research. In: Anderson SH, Hopmans JW (eds) Tomography of soil-water-root processes.
Proc Symp sponsored by Division S-l and S-6 of the Soil Science Society of America in
Minneapolis, Minnesota, 4 November 1992. SSSA Special Publication No 36. American
Society of Agronomy, Soil Science Society of America, Madison, Wisconsin, pp 29-42
Steude JS, Sullivan CP, Chaney RC, McKee M, O'Shea LE (1989) Tracer experiments in saturated
sand by computed tomography and inductively coupled plasma. In: Hoddinott K, Lamb
RO (eds) proceeding of the Symposium on Physio-Chemical Properties of Soil, Rock, and
Related Materials. St Louis, Missouri 29 June 1989. American Society of Testing and Mate-
rials, Philadelphia, pp 171-184
Tollner EW, Verma BD, Cheshire JM (1987) Observing soil-root-interactions and soil organisms
using X-ray computer tomography. Trans Am Soc Agric Engin 30: 1605-1610
Tollner EW, Murphy C, Ramseur EL (1994) Techniques and approaches for documenting plant
root development with X-ray computed tomography. In: Anderson SH, Hopmans JW (eds)
Tomography of soil-water-root processes. Proc Symp sponsored by Division S-l and S-6 of
Vaz CMP, Crestana S, Mascarenhas S, Cruvinel PE, Reichardt K, Stolf R (1989) Using a computed
tomography miniscanner for studying tillage-induced soil compaction. Soil Technol 2:
313-321
Warner GS, Nieber JL, Moore ID, Geise RA (1989) Characterizing macropores in soil by
computed tomography. Soil Sci Soc Am J 53: 653-660
Woods RT, Hennessy n, Kwok E, Hammer BE (1989) NMR microscopy - A new biological tool.
Bio Techniques 7: 616-622
CHAPTER 12
Isotope Techniques
1.J. Bingham\ A.D.M. Glass 2, H.J. Kronzucker2 , D. Robinson 3*,
and C.M. Scrimgeour4
1 Scottish Agricultural College, Craibstone Estate, Aberdeen AB21 9YA, UK
2 Department of Botany, University of British Columbia, Vancouver, V6T lZ4, Canada
3 Department of Plant and Soi! Science, University of Aberdeen, Aberdeen AB24 3UU, UK
4 Scottish Crop Research Institute, Invergowrie, Dundee DD2 SDA, UK
CONT ENTS
12.1 Introduction: Why Use Isotopes in Root Studies? 366
12.2 Isotope Notation and Terminology 370

12.2.1 Stable Isotopes 370
12.2.2 Radioisotopes 372
12.3 Instrumentation and Laboratory Facilities
for Isotope Work 373
12.3.1 Instrumentation 373
12.3.1.1 Mass Spectrometry 373
12.3.1.2 High-Sensitivity Mass Spectrometry 373
12.3.1.3 Scintillation Counting 374
12.3.1.4 Y Counting 377
12.3.1.5 Autoradiography 377
12.3.1.6 Magnetic Resonance Techniques 379
12.3.2 Laboratory Facilities 380
12.3.2.1 General 380
12.3.2.2 Radioisotopes 380
12.3.2.3 Stable Isotopes 381
12.4 Sta bie Isotopes 382
12.4.1 2H (D) and 180 382
12.4.2 l3C 383
12.4.3 ISN 383
12.4.4 34S 384
12.5 Radioisotopes 385

12.5.1 3H 385
12.5.2 14C and ll C 386
12.5.3 13N 388
* Principal author

366 LJ. Bingham et al.
12.5.4 32p 389

12.5.5 86Rb and 42 K 390
12.5.6 3SS 390
12.5.7 Other Radioisotopes 391
12.6 Concluding Remarks 391
References 392
Appendix. Commonly Used Isotopes in Root Studies 400
12.1 Introduction: Why Use Isotopes in Root Studies?
The short answer to this question is detection. Isotopes allow materials in certain
pools (e.g. roots) to be distinguished from those in the rest of an experimental
system (e.g. soil). This is useful when studying the detailed locations (Baldwin
and Tinker 1972; Brownlee et al. 1983) or C dynamics of roots in soil (Milchunas
et al. 1985; Swinnen et al. 1994), both ofwhich are almost impossible to accom-
plish using non-isotopic techniques. The effectiveness with which roots access
certain nutrient pools or water in soil can be estimated using isotopically labelled
nutrient (e.g. Powlson et al. 1986; Hodge et al. 1998) and water sources (Caldwell
and Richards 1989). The isotope dynamics of certain pools can provide useful
information about the structure and function of the system. For example, the
efftux kinetics of radioactivity from labelled roots into an unlabelled bathing
solution reflects the sizes and permeabilities of the various compartments in the
root - apoplast, symplast, vacuole, etc. (Elliot et al. 1984; Kronzucker et al. 1997).
These are just some of many of the possible applications of isotopes in root
studies; others are listed in Box 12.1 (see also Appendix).
Most such applications involve introducing an isotopically distinct
(but chemically indistinguishable) tracer into an experimental system. This
approach has been used widely in laboratory and pot-scale studies and field
experiments (Box 12.1). Experiments from <1 h up to 1 year are possible with
this approach. Longer experiments are impracticable as tracers are eventually
diluted beyond detection (but see, e.g. Clark 1977). It is sometimes possible, and
more convenient (especially at field scales), to use distinct natural abundances
of isotopes as natural tracers, provided that isotopic abundances in different
parts of the system of interest are distinguishable and do not change apprecia-
bly during transformation and transport processes. The appearance of an iso-
topic abundance characteristic of a certain source pool in a sink pool may
indicate transfer of material between pools. For example, Watkins et al. (1996)
used differences in the natural abundance of l3C in C3 and C4 plants to measure
12 Isotope Techniques 367
BOX 12.1. Selected Examples of Root Research Utilising Isotope

Techniques
Research area References
Analysis of transport pathways Dick and ap Rees (1975); Farrar (1985)

Biosynthetic rate determination Robbins and Rost (1987); Lee and Lewis
(\994)
Carbon partition ing Caldwell et al. (1977); Atkinson and
Farrar (1983); Fanar (1985); Gordon et al.
(1985); Svejcar et al. (1990)
Carbon transport between plants Watkins et al. (1996)
via mycorrhizal networks
Cell cyele kinetics Clowes (1971); Berta et al. (1991)
Compartmental analysis of ions Cram (1968); Behl and Jeschke (1982); Jeschke
(1982); Fox et al. (1992); Kronzucker et al.
(1995a,b)
FertiJiser recovery by crops Powlson et al. (1986)
Fluxes through competing metabolic Bryce and ap Rees (1985); Dieua ide-Noubhani
pathways et al. (1995)
Hydraulic lift Caldwell and Richards (1989)
Identifying locations of ion uptake along Clarkson et al. (1968); Kashirad et al. (1973);
and with in roots Marschner and Richter (1973); Kochian and
Lucas (1983); Kraus et al. (1987); Lazof et al.
(1992)
Influx and efflux of ions and organic Elliott et al. (1984); Aslam et al. (1994); Jones
molecules and Darrah (1992, 1994, 1996)
Inter-root competition for nutrients Caldwell et al. (1985, 1987)
Kinetics of ion uptake Deane-Drummond and Glass (1983); Glass
et al. (1985); Siddiqi et al. (1989, 1990);
Clarkson et al. (1996)
Mobilisation of ions in the rhizosphere Marschner et al. (1989); Treeby et al. (1989)
Mycorrhizal cont ribution to nutrient Brownlee et al. (1983); Pfeffer et al. (1997)
and water capture
N2 fixation in the field Androssoff et al. (1994)
Nutrient uptake from non-uniformly Drew et al. (1969); Orew and Nye (1970);
distributed sources Robinson et al. (1994)
Rhizodeposition of carbon Keith et al. (1986); Gregory and At\vell (1991);
Mary et al. (1992); Cheng et al. (1994);
Swinnen et al. (1994); Lichtfouse et al.
(1995)
Root growth of competing plants Ludlow et al. (1976); Wong and Osmond
(1991); Polley et al. (1992)
Root-microbe competition for nutrients Jackson et al. (1989)
Root turnover Milchunas el al. (1985)
Shoot to root O2 transport Yoshida and Eguchi (1994)
368 I.J. Bingham et al.
Shoot to root recycling of N Clark (1977); Cooper and Clarkson (1989)

Spatiallocation of roots in soil Baldwin and Tinker (1972)
Turnover of metabolite pools Farrar (1991)
Visualisation of ion depletion zones Bhat and Nye (1973); Krauss et al. (1987)
around roots
Water uptake by and loss from roots Shone and Flood (1980); Baeic and Ratcovic
(1987); Dalton (1989); Dawson (1993);
Thorburn and Ehleringer (1995)
Xenobiotic metabolism and transport Ezra et al. (1985); d'Arcy-Lameta and Bompiex
in roots (1991)
transfer of carbon between such plants linked by a common mycorrhizal

network. Natural tracers may be used over time-scales from <1 day to >10 years
(as with "bomb"- 14C: Harrison et al. 1990). It is impractical to do very short
«1 h) experiments with natural tracers.
It is important to note two things about natural isotope abundances. First,
using natural abundances to trace elements assumes that isotope fractionations
are negligible (Box 12.2). Second, isotope fractionations - when they occur -
can provide valuable insights into the operation of complex biological systems
(e.g. Robinson et al. 1998), but that approach does not use the same ideas and
interpretive tools as tracer studies; it is vital that the two are not confused.
"Fractionation" -based approaches are in their infancy in root studies, and so
will not be considered further, but see Handley and Scrimgeour (1997) and
Scrimgeour and Robinson (2001) for further information.
Many considerations dictate which isotope to use in a root study. These
include the availability of isotope and instrumentation, complexity of sample
preparation, detection limits and precision, half-life (for radioisotopes), costs
and safety. Often there may be little or no choice because only one type of
isotope or instrumentation is available, but several elements have more than
one suitable isotope. For example, C partitioning to roots, and between roots
and the rhizosphere, can be measured with either 13C or 14C (Warembourg and
Kummerow 1991; Boutton 1991b), or N uptake by roots can be measured using
15N or 13N (Clarkson et al. 1996).
Stable isotopes have the advantage that they are non-hazardous and can be
used without special precautions. These lend themselves well to field and large
pot experiments, unlike radioisotopes where containment is usually required
and large quantities of contaminated soil must be disposed of. On the other
hand, radioisotopes can generally be detected in smaller quantities than stable
isotopes. For example, the theoreticallimits of detection for 13N are as low as
5 x 1O-19 mol compared with 1O-9mol for 15N (Meeks 1993).
Another major advantage of radioisotopes is that visualisation of the
isotope is relatively straightforward using autoradiography or electronic
imaging systems (see Sect. 12.3.1.5), enabling the distribution of the tracer to
BOX 12.2. Isotopic Fractionation

The natural abundance of isotopes varies in different materials. These
variations reflect isotopic jractionations between the heavier and lighter
isotope in a pair such as IS/14 N, 18/ 160, etc. Fractionations are caused by
processes such as evaporation, condensation, and diffusion, and by
chemical reactions.
Equilibrium jractionations may occur in certain chemical equilibria;
e.g., NH) + H+ H NH/ , ISN is ca. 20%0 more concentrated in NH/ than in
NH). Isotopes also fractionate in systems which are not at equilibrium,
because chemical reactions may occur at different (usually slower) rates
for molecules containing the heavier isotope of a pair compared with
those containing the lighter isotope. These are known as kinetic
fractionations. For example, when water evaporates, residualliquid is
enriched in D (by up to ca. 19%0) and in 180 (by up to ca. 3%0) and the
vapour is correspondingly depleted in these isotopes. When tracing the
uptake of isotopically different water sources by deep-rooted vegetation
(White et al. 1985) it is essential to measure 8 80 and 8D in xylem water,
not leaf water, as the latter is subject to evaporative enrichments which
fractionate the respective isotopes.
Isotopically enriched tracers should labei pools such that the
influence of fractionations becomes trivialised. Thus, for example, to
distinguish root-derived e from soil-derived e, plants should be labelled
with I3e to be at least 20%0 different from soil e, ideally more. Otherwise,
the 13/12e fractionations which may occur during respiration (up to ca.
5%0; Boutton 1991a) may mask genuine differences between the two e
sources. It is often impossible to use 8 sN measurements to trace N
sources in the environment partly because of ISII4N fractionations which
can occur as an N source (e.g., soil organic N) is converted to plant-
available forms and assimilated by the plant(s) of interest (Handley and
Scrimgeour 1997; see Sect. 12.4.3).
be mapped in root cells, tissues or systems. Imaging of stable isotopes is pos-

sible, but it requires sophisticated instrumentation, Le. a secondary ion mass
spectrometer (Grignon et al. 1992; see Sect.12.3.1.2). For some applications with
radioisotopes, the element used may be less important than its other properties
such as energy, penetrating power and half-life (Appendix).
Our aim is to provide the basic information needed to use stable isotopes
and radioisotopes in root studies. Specific methodologies for studying root
structure and function, e.g. nutrient uptake, water transport and rhizodeposi-
tion, are dealt with in other chapters (e.g. ehaps. 13, 14) and technical details
are given in the original publications cited in the text (including Appendix).
Useful further reading includes: Vose (1980), Slater (1990), Smith (1991a) and
Smith and Cresser (2001) for radioisotopes (although the first is now rather
dated); Lajtha and Michener (1994), Boutton and Yamasaki (1996) and Scrim-
geour and Robinson (2001) for stable isotopes; Coleman and Fry (1991) and
Knowles and Blackburn (1993) for C and N isotopes, respectively; and Rorison
and Robinson (1986) and Caldwell and Virginia (1989) specificalIy for some
isotope applications in root studies.
AlI work with radioisotopes is potentialIy hazardous. It must be done
according to the health and safety regulations of your institution, and only after
you have consulted the appropriate safety officer. The information given here is
for guidance only.
12.2 Isotope Notation and Terminology
12.2.1 Sta bie Isotopes
Stable isotope abundances are quantified as ratios of heavy to light isotopes in

a sample. Suppose we have a sample of N in which 5 out of every 100 atoms are
15N, and the rest 14N; the isotope ratia (usualIy abbreviated to IR and given the
symbol R) of the sample is 5/95 = 0.0526.
Another way to express the abundance of the heavier isotope in an isotope
pair is as the fraction of the total amount of that element. This is the atom frac-
tion, or, because atom fractions are often small, the atom percent abundance (A),
which is 100 times the atom fraction. In the example for 15N given above, A is
[5/(5 + 95)] x 100 = 5 atom %. The mass per cent of 15N in the sample is not,
however, 5%. It is calculated by multiplying A of each isotope by its atomic mass:
(5 x 15)/[(5 x 15) + (95 x 14)] x 100 = 5.338%. In a tracer experiment, even con-
trols to which no tracer has been added contain background amounts of the
isotope. To find the amount of isotope in the sample that originated from the
tracer, de duct the control's A from the sample's. The difference is the sample's
atom percent enrichment or excess.
The A notation is convenient for tracer work. But if we wish to use natural
variations in isotopic abundance, those data sometimes vary in only the 4th
or 5th decimal place when expressed as percentages. We express natural
abundances as the relative difference between the isotopic ratio in the sample
and that in a standard, in parts per thousand or "per miI" (%0); this is the "8'
notation. Dsample is defined as:
Dsample = [(Rsample - RstandarJ / RstandarJ x 1000. (12.1)
and Rstandard are the IRs of sample and standard, respectively (see above).
Rsample
Values of Rstandard are listed in Table 12.l. Laboratory working standards are
Table 12.1. Isotope ratios (IR) in the international standard materials for the analysis of D, J3C,
15N, 18 0 and 34S
Standard material Isotope pair IR, Rstandard
Vienna Standard Mean Ocean Water (V-SMOW) D/H 0.00015576

Vienna Standard Mean Ocean Water (V-SMOW) 18 0/ 16 0 0.00200520
Vienna PeeDee Belemnite (V-PDB) "0/ 160 0.0020671
Vienna PeeDee Belemnite (V-PDB) J3C/l2C 0.0112372
Atmospheric N 2 15N/14N 0.0036765
Canyon Diablo Troilite 34S/35S 0.0450045
calibrated against these primary standards using materials supplied by the

International Atomic Energy Agency (Austria), the Los Alamos National
Laboratory (USA), and other agencies. Suppose we had a sample whose 13I12C
ratio was 0.0111372, i.e. it differed by only -0.0001 from the standard. In o
notation, this difference is -8.90%0. By definition, alI standards have a of o
zero. IRs in samples may be either positive or negative on a o scale. Positive
values mean that the heavier isotope is more abundant compared with the
standard; negative values, that the heavier isotope is less abundant than in the
standard.
If two source materials have isotopic compositions that are (a) constant,
and (b) different, we can deduce how much of each is present in a common
sink using the latter's isotopic composition (if there is little isotopic frac-
tionation during transfer from source to sink: Box 12.2), relative to the dif-
ference in IR between the sources, using the notion of mixing ratios. If an
isotope has two sources, "X' and "B", the percentage derived from "g' in a
common sink is:
(12.2)
o o
Os,mple is the of the sink. OA and Os, the of each source, are the end members
of the line along which alI isotopic mixtures of "g' and "B" must lie (e.g.
Lajtha and Michener 1994, p. 18). The precis ion of fA depends on the difference
between OA and Os relative to the precision with which can be measured.o
If this precision is 1%0, a difference of 5%0 between OA and Os limits the
precis ion of fA to 20%, at best; i.e. we can say only that O, 20, 40, 60, 80, or 100%
of the isotope carne from source "B". In practice, the precis ion will be even
o
less than this, as the analytical error will also apply to the of each end member.
If there is a difference of 20%0 between OA and Os, and the analytical precision
is 0.1 %0, then we can estimate fA to within 1%, good enough for most practical
purposes.
12.2.2 Radioisotopes
Some of the most commonly used radioisotopes (e.g. 3H, 14C, 35S, 32p) decay
by emitting negatively charged beta particles (/3-). Others emit positrons (W)
produced when a proton decays to a neutron. Positrons are unstable and are
annihilated by interacting with electrons, liberating gamma (11 radiation. The
measurement of yradiation can be used to detect /3+ emitters (e.g. 13N: see Sect.
12.5.3). yrays may also be released from some isotopes after the emission of
/3- or alpha (a) particles. a particles consist of two protons and two neutrons,
i.e. they are He nuclei (4He 2+); a emitters are used rarely in biological work. More
detailed accounts of the chemistry of radioisotopes are given in Slater (1990)
and Vose (1980).
Radioactive decay is random. The number of decaying nuclei at a given time
is proportional to the number of atoms present. So, the decay of an isotope over
time is exponential. Decay rate is characteristic of the isotope and is described
by its half life (t1/2). This is the time taken for the radioactivity present to falI by
half. t1/2 varies widely among isotopes: see Appendix. When conducting experi-
ments with short t1/2 isotopes such as 42K and 13N (Behl and Jeschke 1982; Glass
et al. 1985), a large proportion of the activity initially present in a root or soil
sample may be lost through decay before the sample is counted. Thus experi-
ments must be kept short (~6 x t1/2 ) and the final measurements adjusted to
account for the decay occurring during the experiment. High initial activities
must be used to ensure that there is sufficient remaining at the end of the exper-
iment to detect accurately. One of the major reasons for using the analogue 86Rb
instead of 42K in early studies of K+ uptake kinetics was its more convenient t1/2
(Kochian and Lucas 1982,1983; Appendix).
The energy associated with radioactive decay is also characteristic of the
isotope. This is measuredin units ofkeV (1 eV = 1.6 x 1O-16 J). Thetraditional unit
of activity was the Curie (Ci). 1 Ci == 3.7 x 10 10 disintegrations per second (dps)
or 2.22 x 10 12 disintegrations per minute (dpm). Its definition was based on the
amount of radioactivity in 1 g of 226Ra. AIthough stiH widely used, the Ci has been
replaced by the SI unit, the Bequerel (Bq): 1 Bq = 1 dps; i.e. 1 Bq == 2.7 X 10-11 Ci.
Specific activity is the amount of radioactivity (Bq) of a substance per
unit amount of material. A particular radio-labelled compound can sometimes
be purchased at a range of different specific activities. It is then usually diluted
to the required specific activity with a known quantity of a carrier, compris-
ing the stable isotope(s) of the element. An accurate knowledge of the specific
activity is not necessary in applications where only the presence or absence
of activity in a sample is of interest, such as mapping the location of roots
in the soil (Baldwin and Tinker 1972). But the specific activity of the source,
and in some cases individual metabolite pools, must be known when quan-
tifying nutrient fluxes, translocation or compartmentation of ions and metabo-
lites in roots (Behl and Jeschke 1982; Atkinson and Farrar 1983; Chap. 13).
12.3 Instrumentation and Laboratory Facilities

for Isotope Work
12.3.1 Instrumentation
12.3.1.1 Mass Spectrometry
Most root studies which involve stable isotopes will require the measurement
of IRs using a continuous-jLow isotope ratio mass spectrometer (CF-IRMS); see
Scrimgeour and Robinson (2001). This requires relatively large samples (Jlg to
mg dry wt.) and pre-conversion to a suitable form for measurement. An IRMS
measures the IR of only a few simple gases (H l , N l , COl' SOl)' A CF-IRMS is
usually a bench-top system requiring relatively few services and costs ca. $US
150000.
Solid samples (e.g. plant and soi! material) are oven-dried and ground
to powder in a ball-mill to ensure homogeneity. Sub-samples of 1 to 10mg
are weighed into tin-foiI cups and loaded onto an auto-sampler, a standard
feature in modern CF-IRMS. It is good practice to re-dry stored, previ-
ously dried samples before weighing to remove any adsorbed moisture.
Carry-over of samples in successive grindings must be avoided. Scrupu-
lous cleaning of the ball-mill capsule and avoidance of excessively dusty
working conditions are essential, especially if measuring natural isotopic
abundances.
Liquid samples (e.g. soi! and hydroponic solutions) often have to be pre-
concentrated (e.g. by micro-diffusion or ion exchange) so that a small aliquot
can be pipetted into a tin cup, before being freeze dried in situ. If extracted frac-
tions or individual metabolites are to be analysed, they must be highly purified
and free of any solvents or reagents containing the element of isotopic interest.
Gas samples (e.g. COl from root respiration studies) can be introduced
directly into the MS via an auto-sampler if many samples are to be run. Trace
gases require collection over long periods to ensure that enough sample is avai!-
able for analysis.
Standards are included between every 10-20 samples to calibrate the analy-
sis. Calibration is checked by including a standard sample (e.g. plant material
of a known isotopic composition) within each batch of samples.
12.3. 1.2 High-Sensitivity Mass Spectrometry
Gas-chromatograph-MS (GC-MS) or high performance liquid chromatograph-

MS (HPLC-MS) combine separation of minute amounts (pg to ng) of complex
mixtures with IR measurement, but sample enrichments of ca. 1 atom % or
more are required for reliable quantitation. GC-MS or HPLC-MS can measure
only percentage differences in IR. These instruments are useful in metabolic

studies, e.g. to determine the fate of N assimilated by mycorrhizal roots (Martin
and Botton 1993).
IRMS is not sensitive enough to quantify isotopes with small natural abun-
dances, e.g. 14C, 41K, and with very dilute tracers. Greater sensitivity is achieved
by adding a second mass acceleration/separation step, as in tandem MS-MS.
MS-MS requires little or no sample preparation of even complex mixtures and
tissues, and analysis is faster due to the elimination of a lengthy chromato-
graphic separation step required in GC-MS or HPLC-MS.
Other high-sensitivity MS instruments include secondary ion mass spec-
trometry (SIMS). This uses high-energy (MeV) "primary" ion beams to sputter
"secondary" (negatively charged) ions from specimen surfaces. The sensitivity
of SIMS is usualIy limited to 10-7 to 10-8 minor atoms per major atom. One of
its strengths is the ability to image the spatial distributions of isotopes within
tissue sections (Grignon et al. 1992; Lazof et al. 1994). Modified versions of SIMS
- often referred to as accelerator mass spectrometry (AMS; Trumbore 1996) -
can reach detection limits of 10-15 atoms atom- 1. Nepstad et al. (1994) related
root biomass to changes in 14C natural abundance with soil depth, 14C being
measured by AMS.
12.3.1.3 5cintillation Counting
Scintillation counting is the most widely used technique for measuring radioac-
tivity.lts advantages over other techniques include its sensitivity, versatility and
ease of use. The basic principles are the same whether determin ing a, f3, or r
radiat ion.
The kinetic energy associated with a, f3 or r radiation is transferred to a
fluorescent molecule (scintillator or fluor). This transfer is either direct (solid
scintillation or rcounting) or indirect via a solvent (liquid scintillation count-
ing). The resulting excitation of the fluor causes it to emit photons ("scintilla-
tion"). These are then detected by photomultiplier (PM) tubes and counted
electronicalIy. Radioactivity is reported as counts per second or minute (cps or
cpm). Not alI disintegrations are measured and so it is essential to know the
counting efficiency to convert cps to dps. AlI instruments will register back-
ground counts in the absence of a radio active sample. This must be subtracted
from the sample's cps. In most instruments this is do ne automaticalIy. See
Simonnet (1990) and Goulding and SI ater (1994) for more detailed accounts of
the principles of scintillation counting.
Liquid scintillation counting (LSC) is used mostly in biological work for
measuring weak or medium energy f3- emitters. A sample is placed into a vial
with a scintillation cocktail. This cocktail consists of an organic solvent and one
or more fluors. Careful sample preparation is critical for maximising counting

efficiency with LSC (Coupland 1986).
Quenching is a major cause of poor counting efficiency. Colour quenching
occurs when pigmented samples are counted. Problems are most frequently
encountered with leaf extracts containing chlorophyll. Although less of a
problem with roots, colo urat ion may occur with some extraction procedures.
Emitted fluorescence is absorbed or scattered by the pigments before it can be
detected by the PM tubes. Colour quenching can be reduced by bleaching the
sample with, e.g. H20 2 prior to counting (Coupland 1986). However, in some
scintillation cocktails, this may lead to a different problem of chemiluminescence
(see below). An alternative strategy is to reduce the sample volume relative to
that of the cocktail, effectively "diluting out" the colour. But take care that there
is sufficient energy in the vial to register cps significant1y above background.
Chemical quenching refers to interference with the energy transfer from solvent
to fluor by components of the sample. If chemical or colour quenching occurs,
a correction must be applied to calculate the true dps in the sample. Several
techniques are available for quench correction including internal standardisa-
tion and external standardisation (for details see Simonnet 1990; Goulding and
Slater 1994). Modern LSC instruments have automatic external standardisation.
For ease of use and accuracy, this is usually the method of choice. Some instru-
ments also have software for determining dps directly from an analysis of the
full energy spectrum of the isotope without the need for quench correction (e.g.
Direct DPM, Canberra Packard).
Two other important sources of interference in LSC are chemiluminescence
and photoluminescence. Chemiluminescence can result from reactions among
components of the sample and the scintillation cocktail that generate weak
light, but are independent of radio active decay events. Storing samples before
counting is advisable to allow any chemiluminescence to decay. Photolumines-
cence occurs when components of the sample and the vial absorb light and re-
emit it at wavelengths detectable by the PM tubes. It is mostly a problem with
pigmented samples. Sample storage in the dark prior to counting can help over-
come the problem. Some instruments have a facility for identifying and cor-
recting for luminescence, although this should not be regarded as a substitute
for careful sample preparation.
Since [3- particles travel only short distances, it is essential that the sample
is distributed homogeneously in the scintillation cocktail. LSC is best suited
to counting liquid samples. This would include samples of soil solution,
hydroponics or root extracts (Gordon et al. 1985; Jones and Darrah 1994).
Radioactive gases can be counted as solutions after dissolving them in an
appropriate solvent; e.g. 14C02 respired from roots or soil micro-organisms, or
that liberated from oxidation of root tissue (see below), can be trapped in NaOH
or an organic amine (e.g. phenylethylamine or ethanolamine) and counted
(Swinnen et al. 1994). Organic amines are preferable to inorganic bases as

the former have a gre ater trapping capacity, and produce less quenching and
chemiluminescence (Coupland 1986).
Often, the radioactive sample to be counted is a solid. This may be excised
roots, or an insoluble residue after extraction, e.g. ceH waH or DNA (Robbins
and Rost 1987). Although these may be counted directly (Farrar 1985, 1993),
there are two potential problems. First, much radiation may be absorbed by the
sample itself or its solid support, e.g. glass fibre filter (self-absorption). This
leads to a loss in counting efficiency. This is most pronounced with weak f3 emit-
ters, e.g. 3H. Second, counting efficiency will be influenced greatly by the orien-
tation of the sample in the scintillation vial relative to the PMs. Efficiency losses
through this are unpredictable and cannot be corrected by standardisation.
Homogeneity can be increased by first grinding the sample then adding water
and a gel-forming scintillation solution. The sample is dispersed by shaking.
A gel forms after the shaking stops, holding the sample in suspension for
counting.
The best way to treat more bulky material is to solubilize or oxidise it before
counting. Plant samples can be solubilized using strong alkalis such as quater-
nary amines (Zocchi and Hanson 1982; Coupland 1986; Bingham et al. 1998).
This procedure is straightforward. Its main drawback is that only smaH samples
can be treated. Solubilization can lead to chemiluminescence with some cock-
tails, overcome by selecting a compatible cocktail. Mineral elements bound to
or incorporated in structural material can be counted after acid digestion
(Jensen and Kylin 1980). Dalal (1979) and Amato (1983) have described wet oxi-
dation techniques for determining 14C in root and soil samples; organic 14C is
oxidised by strong acid and trapped as 14C02.
For 3H and 14C, an alternative to solubilization and wet oxidation is dry
sample oxidation. Dedicated equipment is available for this. The sample is
combusted in an atmosphere of O2, The H20 and CO2 which are produced are
trapped, and the remaining oxides (e.g. S03 or NOx ) released as waste. Radioac-
tivity in the trapped samples is then counted by LSC. Dry oxidation has several
advantages over other forms of sample preparation. It is quick, large samples
can be handled, and it greatly reduces chemiluminescence and quenching.
Sequential extraction of soluble C fractions can precede sample oxidation (e.g.
Gordon et al. 1985).
A possible alternative to LSC is Cerenkov counting. When high-energy
charged partides pass through a transparent medium (e.g. H20) at a speed
faster than light in that medium, blue-white light is emitted which can be
measured directly in an LS counter (Vose 1980, p. 122). Cerenkov counting is
possible for yemitters and hard f3 emitters with a partide energy greater than
O.5MeV (the Cerenkov lirnit). This indudes many isotopes used in studies of
ion transport in roots; 32p, 36Cl, 24Na, 42K, 86Rb and 13N. Samples, induding root
extracts or excised roots, are placed in water and counted directly (Behl and
Jeschke 1982; Lefebvre and Glass 1982; Glass et al. 1985). It is a simple, cost-
effective method of counting because no organic cocktail is required; it is not
affected by chemical quenching, although colour quenching can occur. Self-
absorption can be a problem with excised roots. To overcome this, establish a
calibration curve by varying the root mass in the vi al (Glass et al. 1985). Count-
ing efficiencies range from 0.1 to 0.8 depending on the isotope. Efficiency
increases with increasing Emax above the Cerenkov limit (Vose 1980; Goulding
and Slater 1994).
72.3. 7.4 r Counting

yrays are more penetrating than most {3- particles, and so tend to pass through
a scintillation cocktail without transferring energy to a solvent or fiuor.
Standard y counters use a cylindrical, solid crystal fiuor (e.g. Nal laced with
thallium) into which a well has been bored in the centre. Sample vials are
transferred to the well for counting (Smith 1991b, p. 350).
Because the sample is external to the fiuor, y counting does not suffer in
the same way as LSC from problems with quenching and luminescence.
Both liquid and solid samples can be counted, but counting efficiency depends
on the sample geometry. The top of the sample must be positioned deep in the
crystal's well and ideally a series of samples should have a constant height. In
root studies, the lack of quenching problems is particularly attractive since
virtually no sample preparation is required other than excision of the parts
of interest (ef. Glass et al. 1985; Kronzucker et al. 1995a,b,1997). Most instru-
ments have calibration procedures that determine automatically the resolution,
background and counting efficiency of each detector (see Simonnet 1990 for
principles ).
Many solid materials, other than the Nal crystals usually employed in y
counters, can be excited to fiuorescence by impacting radiation. These have
potential applications in root research. Minchin et al. (1994) used a series of
scintillation detectors positioned against leaves and two halves of a split-root
system to measure !lC transport within intact barley plants.
72.3.7.5 Autoradiography
Autoradiography is an extremely useful technique for determining the distri-

bution of a radioactive tracer within a sample.
Macro-autoradiographs of whole root systems, and similar structures,
can be made by tap ing the plant material to a piece of stiff card within an
X-ray film cassette or similar holder. To prevent redistribution of the isotope,
378 1.1. Bingham et al.
the roots are oven or freeze dried before a sheet of X-ray film is clamped
tightly against the tissue and exposed for a period of hours to days (Vose
1980, p. 212 et seq.; Farrar 1993). An alternative to drying is to secure the
film over fresh roots (protecting the emulsion with a layer of thin cellophane)
and freezing at -20 ac during exposure. The exposure time required will depend
on the isotope used, its activity, the type of film and the temperature. As in
alI types of autoradiography, a certain amount of experience and trial and
error is required to judge the period accurately. If replicate samples are
available, try developing these for different times to check for adequate expo-
sure. Typical applications include identifying major sinks for assimilates
in roots and root-mycorrhizal associations (e.g. Brownlee et al. 1983) and
determining the systemic activity of root -applied pesticides (d' Arcy-Lameta
and Bompeix 1991).
In situ autoradiography has been useful in mapping the distribution of
roots growing in soil. This has been done in two ways. Soil is loaded with a
radioisotope, roots are grown though the soil, the soil and roots are embedded
in resin and then sectioned horizontally. The cut surface is placed in contact
with X-ray film for a period of time and developed (Nye and Tinker 1977 p.
141). This technique also allows ion depletion zones around roots to be visual-
ized (Kraus et al. 1987). Second, the tops of plants are injected with radioiso-
topes of readily translocated elements. Sheets of X-ray film are driven into the
soil through the root system, or placed against the window of a root observa-
tion box against which roots are growing. The distribution of labelled roots
contacting the plane of the film is indicated by the appearance of dark spots on
the autoradiograph. Suitable isotopes are those which are phloem mobile, have
medium to high energy to overcome absorption by adhering soil, and a t1/2 of
several days or more (e.g. 14C, 32p, 35S; not 45Ca or 3H). Baldwin and Tinker (1972)
visualised the distributions of the inter-penetrating root systems of two plants.
Each plant was injected with a [3 emitter differing in Emax (see Appendix): 35S or
33p in one plant, and 32p in the other. The difference in Emax allowed the smaller
energy to be screened out by a sheet of Melinex covering the film. Baldwin and
Tinker (1972) give some useful hints about the practical application of this
technique.
For studies of cellular and sub-cellular localisation (micro-
autoradiography), liquid emulsion is used to provide the close contact needed
between the section and film (Laskey 1990). After development, the slide can be
viewed under bright or dark field optics to reveal, respectively, black or white
grains where the isotope is located. Isotopes which are not incorporated into
structural material during labelling of the roots may become redistributed
during slide preparation. This can be avoided by freeze-substitution of the root
before embedding (e.g., Kochian and Lucas 1983). Micro-autoradiography is
best suited to the weak [3- emitters 3H, 14C and 35S. Their energy is soon dissi-
pated within the emulsion producing a discrete, high resolution image. Higher
energyemitters (e.g. 32p) generally produce insufficient resolution for micro-
scopic examination.
eomparing the autoradiograph with a photograph of the same structure
allows the distribution ofthe tracer to be determined (e.g. Brownlee et al. 1983).
The autoradiograph can be quantified by densitometry or image analysis
(Kochian and Lucas 1983) against standards of known activity. But note that the
density of image is proportional to the logarithm of exposure and that, above
a threshold exposure, further exposure does not lead to an increase in density
(Farrar 1993).
Electronic systems are now available for imaging radioactivity (e.g.
InstantImager, Packard). By combining direct nuclear counting with image
analysis, the distribution of radioactivity in samples can be quantified in real
time without the need for lengthy exposure and development of film. Its advan-
tages over standard autoradiography are its speed, ease of use, sensitivity and
linearity of response. Because results are generated within 10-30 min, autora-
diography of short tl/2 isotopes is possible. Techniques for autoradiography in
biochemical and molecular work, e.g. visualising 32p probes, are covered by
Laskey (1990).
12.3.1.6 Magnetic Resonance Techniques
Magnetic resonance (MR) involves subjecting molecules to an oscillating elec-

tromagnetic field and measuring the frequency at which they come to resonate
with that field. Most MR work involves detecting nuclear resonance, so it is
known as nuclear MR (NMR). The advantage of NMR is that it is that it is highly
compound-specific. It gives intra-molecular, site-specific information, often
with a minimum of sample preparation, providing that the compound(s) of
interest are sufficiently concentrated and isotopically enriched in the sample.
NMR usually requires at least 10 atom % labelling for quantitation and 1-
100 mg samples, and uses ca. 100% labelled substrates. Nuclei of only certain
isotopes can be detected and quantified by NMR: e.g. 'H, D, l3e, 'SN, 170 and 31p.
NMR spectra do not usually give IRs (or specific activities) directly (unlike MS),
but these may be deduced from, e.g. the extent of 'H_ l3 e couplings in the spec-
trum (Dieuaide-Noubhani et al. 1995). Examples of NMR spectroscopy as
applied to nutrient transport in mycorrhizal root systems are described by
Pfeffer et al. (1997).
In root studies, NMR has been used exclusively for metabolic investiga-
tions. The work of Dieuaide-Noubhani et al. (1995) is a superb example of its
use to clarify pathways of e metabolism in root tips, using both l3e and 14e
labelling. Bacic and Ratkovic (1987) used NMR to monitor water exchange
within roots after labelling with D20. For further information on the use of
NMR in plant nutrition studies see Lee and Ratcliffe (1993).
12.3.2 Laboratory Facilities
No laboratory is designed or equipped idealIy for alI isotope applications. What

folIows is a brief outline of some of the facilities that alIow isotope work to be
done safely and reliably.
12.3.2.1 General
It is a good idea to keep experimental and sample preparation areas separate

from those containing the analytical instruments. It is also sensible, if possible,
to separate areas and equipment (e.g. grinders) in which isotopicalIy labelled
and unlabelled materials are used. Accidental contamination by even a small
amount of a tracer can ruin natural abundance measurements. AlI working
areas should be kept clean and uncluttered. This applies as much to the are as
where soil is prepared as to the analyticallaboratory.
12.3.2.2 Radioisotopes
Basic safety points are described fully by Vose (1980) and Prime and Frith
(1990). Most importantly, alI radioisotope work must conform to acceptable
health and safety standards in addition to those that are in force to ensure good
generallaboratory practices. These differ from country to country, although the
basic rules apply everywhere. AlI workers with radioisotopes should be prop-
erly trained in their use, monitoring and disposal.
Radioisotopes pose an internal hazard if material is ingested, inhaled or
absorbed through the skin and an external hazard from the radiat ion field sur-
rounding an exposed source. The external hazard should be minimised by using
appropriate shielding and by maximising the distance between the operator
and the radioactive source. Access to "hot" areas should be restricted. If pos-
sible, glasshouses or growth rooms should be designated specialIy for radioiso-
tope work. At the very least, the presence of radio active material should be
displayed clearly and shielded from other laboratory areas. When using high
energy sources, the external hazard should be monitored during the experiment
and the acute and cumulative exposure kept as low as possible and certainly
within the legal safety limits. Operators may be required to wear personal
dosimeters or film badges. However, these provide an indication of the level of
exposure only after the dose has been received.
Laboratories should have work surfaces and floors that are as continuous
as possible to minimise the extent to which radio active materials can lodge
and which can be cleaned and decontaminated easily. This includes walls with
washable paintwork.
Operations that generate dust, e.g. grinding, sieving and sectioning of
plant and soil material should be carried out in a negative-pressure glove box
or a fume hood. Particular care is necessary when conducting experi-
ments using e.g. 14C and 35S to avoid accidental release of 14COZ' 35SOZ or
H/ 5S into a confined atmosphere. Degeneration of 35S-amino acids during
storage can release radio active volatile compounds. 14COz may be respired
from plants and microbes labelled with 14C. Where accumulations of radio-
actively labelled gases are likely, apparatus should be vented in a fume
cupboard. Alternatively 14COZ may be trapped by e.g. self-indicating soda
lime or KOH, and the radio active material disposed of as solid or liquid waste
respectively.
The minimum containment for solid or liquid materials is a tray large
enough to contain any spillage on an open bench in a designated laboratory.
The minimum personal protection should be laboratory coat, safety spectacles
or goggles and vinyl or latex gloves. Handle alI radioactive materials with
forceps. Before, during and after each experiment or operation involving
radioisotopes, monitor the working area and laboratory clothes for contami-
nation. This is usualIy done with swab counts or calibrated count rate meters
suitable for the isotope in use.
The use of radioisotopes in the field is possible (e.g. 14C to measure rhi-
zodeposition of C; Swinnen et al. 1994), but presents some practical difficulties
in terms of containment, control of public access, long-term contamination and
waste disposal. A complete risk assessment of the fate of the radioisotope in the
environment is required. As the Chernobyl incident demonstrated alI too welI,
the uncontrolled introduction of radioisotopes to large areas of land can have
unforeseen long-term consequences. Where possible, the use of stable isotopes
should be explored for field experiments.
12.3.2.3 Sta bie Isotopes
Fewer specialised facilities are needed to handle and use stable isotopes.
An important one that is essential for natural abundance 813 c experiments is
to grow plants in a welI-ventilated glasshouse. This avoids unacceptable fluctua-
tions in the 813 C of source COz which might confound later analyses of 813 C.
For this reason, 813 C experiments in controlled environment chambers should
be avoided, unless those chambers have accurate and reliable control of the
internal [CO z) and 813 c: few do.
12.4 Sta bie Isotopes

12.4.1 2H (O) and 180
Deuterium eH or D) and 180 are considered together as these isotopes feature

jointly in many applications, particularly on water uptake by roots, but they
differ in how they are prepared for IRMS. Generally, 180, when used as a tracer,
is expensive but easy to measure, while D is cheap and more difficult to measure.
Suitable working MS standards for D and 180 in H20 are a local tap-water and
tap-water with a little D2 0/H/ 80 added. Both are calibrated against V-SMOW
(Table 12.1).
Water itself is not suitable for IRMS analysis because of memory effects
(Le. water tends to remain adsorbed on surfaces inside the MS inlet). It must
first be converted to, or equilibrated with, more suitable gases: CO2 for 18 0; H2
for 2H analysis. 180 is measured by equilibration of H20 with CO2. The O atoms
in water and CO2 exchange, and the 180 content of the CO 2 reflects that of the
water. This is done readily on large sample sets, with the CO2 being sampled
and measured automatically. For D analysis, water is converted to H2, usually
by reduction with hot Zn or U. Prosser and Scrimgeour (1995) combined an
alternative equilibration method (Le. using H2and a Pt catalyst) with automated
CF-IRMS of H2.
Additional steps are required to measure D or 18 0 in H20 within matrices
such as soil or plant tissue. Existing methods rely mainly on azeotropic
distillation to extract H20. This involves distilling the H20 in the material
with a solvent such as toluene. Each distillation takes several hours,
requires solvent and expensive glassware, and needs samples containing at
least 5 mI of H20. Direct equilibration of such samples avoids the extraction
step and is more suited to large sample sets (Scrimgeour 1995). 18 0 in H20
in plant stems, woody twigs and soil equilibrates with added CO2 at a similar
rate to 180 in free H20. This method gives accuracy and precis ion compara-
bIe to azeotropic extraction. Preparation time is shorter and smaller samples
(containing as little as 100,ul) can be used. With CF-IRMS of the equili-
brated CO2, up to 100 samples can be analysed daily, making ecological-
scale studies a practical proposition. Direct equilibration with H2 can also
be used to analyse H20 in intact plants. However, complete equilibration does
not occur unless the sample is first heated to 100 DC under vacuum. Presumably,
this releases tightly bound H20 that exchanges slowly with other H20 in the
system.
Given sufficient concentration and isotopic enrichment, D is quantifiable
by NMR as Bacic and Ratkovic (1987) showed in their study of H20 exchanges
in maize roots.
12.4.2 13e
l3C tracers are readily available as 13C02 or as a wide range of labelled organic
compounds up to 99 atom %. Most commercially available compressed CO 2 is
isotopically distinct from atmospheric CO 2 (Schnyder 1992). Whereas the 8 3c
of the latter is ca. -8%0, CO 2 produced (industrially in, e.g. distilleries, brew-
eries, petrochemical plants) has 8 3c of -25 to -50%0. A working standard for
natural abundance determination of 8l3 C by CF-IRMS is a leucine/citric acid
mixture (Handley et al. 1993). An internal standard "plant" material of com-
mercially available flour is always included in each batch of samples. To measure
8 3c of CO2 produced by roots and soil microbial respiration, Robinson and
Scrimgeour (1995) used as a standard 5% CO 2 in N2, calibrated against V-PDB
by comparison with other gas samples (Table 12.1).
Natural variations in 8l3 c are potentially very useful tracers of C. The dif-
ference in 8 3c between C3 and C4 species may be used to detect C from one
source against a background of C from the other, e.g. root C of maize (C4 ) in
soil containing soil organic matter derived only from C3 vegetation (Balesdent
and Mariotti 1996).
Within plants, different C-containing molecules have different 8 3c values.
These arise because of l3/12C fractionations that occur during their enzyme-
catalysed synthesis. Compared with hexose, lignin is up to 4%0 more depleted
in l3C; cellulose is ca. 2%0 more enriched; and chitin - not a constituent of
plants, but an important one of pathogenic and mycorrhizal fungi - is up to
4%0 more enriched (Gleixner et al. 1993). Protein and, particularly, lipids
are depleted in l3C relative to carbohydrate. Knowing the J3C composition of
individual compounds allows deeper interpretations to be made of, e.g. the fate
of decomposing root residues (Lichtfouse et al. 1995) or of the metabolic
interactions between roots and associated fungi (Gleixner et al. 1993).
J3C is also quantifiable by NMR given sufficient sample size and isotopic
enrichment (Dieuaide-Noubhani et al. 1995; Sect. 12.3.1.6).
12.4.3 15N
15N has been used most often as a field- or pot-scale tracer for N, in "fertiliser
recovery" trials. 15N enrichments of ca. 1 atom % are usually adequate in the fer-
tiliser. 15N-Iabelled inorganic salts and organic compounds are readily available
with 15N abundances of up to 99 atom %. 15N-enriched N2 is also available. Less
widely known, but of potential use in solution culture experiments, is 15N_
depleted material. This is produced as a by-product of the 15N enrichment
process (industrial 15/14N fractionation), and it is much cheaper than the 15N_
enriched fraction. It is useful for growing plants at a very small background of
ISN as the ISN depleted material can have a ISN abundance of ca. only 0.005 atom
% (Vose 1980, p. 165-7).
ISN is quantified by CF-IRMS as for 8 3c (see above). For natural abundance
measurements of both 8 3c and 8 sN on the same sample, % N is measured first
during a run optimised for 8 3c. This information is then used to calculate how
much sample is required to give enough N in a subsequent analysis optimised
for 8 sN. This can greatly improve the accuracy of the 8 sN determination.
Working standards are as for 8 3c, calibrated against atmospheric N2•
When ISN is used as a tracer in soil, data interpretation may be problemat-
ical if the mixing of ISN labelled materials with native N pools in the soil is
incomplete or, in some way, influences N turnover. This can cause the so-called
added N interactions that Jenkinson et al. (1985) and Powlson and Barraclough
(1993) explain fulIy.
Natural variations in 8 sN are not as useful as are those of 8 3c to trace the
element. Variations in whole-plant 8 sN can be greater than that of the 8 sN of
putative external N sources (be they N0 3-, NH/ or N2 : Handley and Scrimgeour
1997). This means that it is rarely possible to attribute changes in plant 8 sN to
the uptake of different sources of N unless corroboratory evidence is available
(preferably complete analyses of the amounts and 8 sN of alI N sources, some-
thing which is not yet technicalIy feasible). Many 8 sN studies in the literature
should be read with a critical eye and not taken at face value. No methods are
yet available to routinely measure the 81S N values of soil N constituents such as
N0 3- reliably (see Scrimgeour and Robinson 2001). The isolation of N0 3- from
soil by diffusion or distillation is not specific to N0 3- and "N03-" thus isolated
may contain contaminants which contribute to the apparent 8 sN value of the
sample. Samples must be screened for the presence of potential contaminants.
This is done most easily by comparing mass amounts of N measured colori-
metrically (which is specific for N0 3-) with those measured by CF-IRMS (which
are not specific). If the two amounts do not agree to within 10%, say, reject the
data (Robinson and Conroy 1999).
NMR may quantifiably distinguish different ISN-IabelIed fractions in root
tissues or extracts (Fox et al. 1992; Sect. 12.3.1.6).
12.4.4 34 5
There are only a few sources of 34S labelled tracers, available mainly in inorganic
forms; e.g., Ca34S04 which costs around $US 4000 per g isotope. NaturalIy, dis-
similatory reduction of SO/- by microbes depletes 34S in the sulfides that are
produced by up to 50%0 (see Lajtha and Michener 1994, p. 59).
A major reason why S-containing compounds in the environment may
differ in 8 4s is, however, not because of 34/32S fractionations, but because of pol-
lution. Atmospheric inputs of S-compounds produced from, e.g., coal burning,

can be detected in soils and waters if their 8 34S values are sufficiently distinct
from non-anthropogenic sources. However, Kirchmarr et al. (1996) concluded
that "shifts in soil 8 34S cannot be used as a tool for quantitative determinations
of S turnover" in soil, nor, presumably to identify S sources absorbed by roots
in situ.
The determination of 8 34S is more difficult than for the other stable iso-
topes described so far. Analysis is usually done on S02, but reliable conversion
of samples to S02 is tricky. Scrimgeour and Robinson (200l) describe the latest
analytical developments which deal with such problems.
12.5 Radioisotopes
12.5.1 3H
3H (tritium) is usually used in root studies to measure uptake, transport and

metabolism of organic compounds. A wide range of 3H-labelled molecules is
available. For example, amino acid uptake by maize roots was studied by mea-
suring the depletion of 3H-aspartic acid from culture solution (Jones and
Darrah 1994). Rates of biosynthetic reactions can be determined by measuring
the incorporation of 3H-labelled precursors into macromolecules. This
approach has been used to investigate the effects of sulfonylurea herbicides on
DNA synthesis in pea root tips (Robbins and Rost 1987).
Greater resolution can be achieved in micro-autoradiography using
3H rather than 14C-labelled material, although it is less suitable for macro-
autoradiography because of self-absorption by root tissue (but see d' Arcy-
Lameta and Bompeix 1991). A standard technique for measuring cell
proliferation in root meristems involves incubating roots in 3H-thymidine to
label newly synthesised DNA and then microautoradiography of root squashes
or sections to follow the progression of the labelled DNA through the cell cycle
(Clowes 1971; Berta et al. 1991). Visualising the location of inhibitors in roots
has provided valuable information on the spatial patterns of ion uptake.
Microautoradiography of 3H -NEM, a potent inhibitor of K+ uptake, showed that
NEM binding occurred almost exclusively in the peripherallayers of the maize
root cortex (Kochian and Lucas 1983).
Another advantage of 3H over 14C is the greater specific activity of
material that can be obtained. Radio-chemicals containing a single 3H atom
per molecule have a maximum specific activity of 1.07 x 106 GBqmol-1
compared with 2.31 x 103GBqmol-1 for 14c. In metabolic studies where small
amounts of isotope incorporation are anticipated, or where small con-
centrations of source compound are being used, 3H may be the preferred
isotope because of the better sensitivity of detection that its gre ater specific
activity allows.
A potentiallimitation of 3H is isotope exchange. 3H attached to -COOH, -
OH and -NH2 groups may exchange with H20, thus compromis ing the inter-
pretation of tracer experiments (Vose 1980, p. 148). Labelling should therefore
be on parts of the molecule that do not exchange (d' Arcy-Lameta and Bompeix
1991).
12.5.2 14e and 11e

Some of the major questions being addressed in studies of root and rhizosphere
function concern fluxes and partitioning of C within the plant-soil-microbe
system. 14C is widely used as a tracer in such work. Since root systems derive
most of their C from photosynthesis, 14C02 is often used to label photosynthate
to measure the flux of C to the root system and its partitioning between root
and rhizosphere. For details of experimental techniques and numeric al analy-
ses see Keith et al. (1986), Gregory and Atwell (1991) and Swinnen et al. (1994).
General features of 14C02 feeding systems are similar in most laboratories,
differing only in their level of sophistication (Warembourg and Kummerow
1991; Farrar 1993). Whole plants or individual leaves may be sealed in a
feeding-chamber into which 14C02 is released by the addition of acid to 14C03-
or H 14 C0 3-. Uniformity of labelling is improved by ensuring adequate mixing
of air within the chamber or by using a flow-through system.
These techniques are suitable for short-term experiments to pulse-Iabel
plant C pools, e.g., when measuring the time courses of C partitioning between
roots and rhizosphere (Meharg 1994). Longer-term feeding (>24h), during
which a uniform supply of 14C02 is maintained, is necessary to label soluble C
pools to constant specific activity (Farrar 1985). Here a syringe pump can be
used to add, at a constant rate, a 14C03- solution to acid, and the 14C02 produced
is swept by an air stream through a leaf chamber or whole-plant cuvette
(Warembourg and Kummerow 1991; Farrar 1993). Releasing 14C02 into the air
will raise the concentration of CO 2above ambient. When control of the CO 2con-
centration is required, it may be necessary to scrub CO 2 from the airstream
before adding it back at a known dose-rate. Diluting the 14C03- with carrier CO 2
allows both the desired specific activity and total CO 2 concentration to be gen-
erated. Continuous feeding over extended periods is required to uniformly label
root structural (i.e., cell wall) material. The relative merits of pulse-Iabelling and
continuous labelling for measuring rhizodeposition of C have been discussed
by Meharg (1994).
Translocation of specific 14C-Iabelled organic compounds (e.g. phloem-
mobile herbicides) can be measured by applying solutions direct1y to abraded
or non -abraded leaves, or by incubating cut ends of otherwise intact leaves in

14C-solutions (Thompson et al. 1986). This technique, when used with 14C_
sucrose, provides an alternative to 14C02 for labelling root carbohydrate pools.
In an investigation of the pathways of phloem unloading in pea root tips, Dick
and ap Rees (1975) applied 14C-sucrose to the base of roots whose cortex had
been removed to expose the vascular tissue. For studies of root physiology the
simplest, and often most effective way, of supplying 14C-Iabelled organic com-
pounds to roots is via solution in hydroponics. Farrar (1991) pulse-Iabelled
barley roots with 14C-sucrose to measure rates of starch and fructan turnover.
Jones and Darrah (1992, 1996) supplied maize roots with 14C-sugars and organic
acids to study mechanisms of C exudation and re-sorption. However, consider-
ation should be given to whether normal endogenous C pools are labelled by
this method.
An extensive range of 14C-Iabelled organic molecules is available in addi-
tion to 14C03- and H 14C0 3- salts and 14C02. Organic products may be labelled at
specific positions in the molecule (e.g. [1_ 14C] glucose in which the C-l atom is
the 14C isotope, the rest being l2e). These are useful if the metabolic fate of a
particular fraction of the molecule is to be followed or relative fluxes through
different metabolic pathways (e.g., glycolysis versus oxidative pentose-
phosphate pathway) are measured (Bryce and ap Rees 1985; Dieuaide-Noub-
hani et al. 1995). In uniformly labelled products (e.g. [U_ 14C] glucose), each C
atom has an equal chance of being 14C. These are less expensive and are appro-
priate for studies of uptake and metabolism where the metabolic fate of the
whole molecule (or of the C in it) is of interest. Suppliers may even be willing
to synthesise specific products that are not normally available (but at a price!).
Although ubiquitous in atmospheric CO2, most atmospheric 14C is not
strict1y "natural" in occurrence. It was produced largely from thermonuclear
weapons testing in the late 1950s and early 1960s (Goh 1991). The abundance
of atmospheric 14C is changing in a predictable (and, for us, a useful) way as it
decays. We can use this decay to distinguish C fixed by plants recently (Le., since
ca. 1960) and introduced to soil via their roots from C fixed earlier (Harrison
et al. 1990; Nepstad et al. 1994).
14C can be detected by LSC or autoradiography. Its extremely long t1l2
(Appendix) means that any contamination is serious, particularly if liquids or
solids find their way into cracks in floors or benches, because the activity will
not quickly decay to safe levels. Detection by AMS is possible (Nepstad et al.
1994), as is 14C-NMR (Dieuaide-Noubhani et al. 1995), given appropriate sample
sizes and specific activities.
In certain types of work, the long t1l2 of l4C can be a limitation. llC, in con-
trast, has a t1l2 of 20.3 min which means that after administering a pulse of llC0 2
most of its activity will have decayed within 1 h. Measured activity then repre-
sents mainly current assimilate, rather than material in storage pools (unlike
14C). By re-supplying llC0 2 at set intervals, repeat or continuous labelling exper-

iments are possible on the same plant, to investigate short-term changes in
translocation from leaves to roots, roots to rhizosphere, or between competing
roots (Spence and Sharpe 1991; Minchin et al. 1994; Minchin and Thorpe 1996).
llC in intact tissues may be detected continuously using scintillation detectors
coupled to a ratemeter. Many of the precautions required for llC experiments
are the same as those which apply to 13N (see Sect. 12.5.3).
12.5.3 13N
The opportunities and practicalities of using high energy, short t 1l2 isotopes in
root studies are illustrated by 13N. The greater sensitivity of detection possible
with 13N compared with lsN has alIowed low capacity constitutive N transport
systems in roots (Lee and Drew 1986; Kronzucker et al. 1995a), N influx (Siddiqi
et al. 1990; Kronzucker et al. 1995a) and efftux (Lee and Clarkson 1986;
Kronzucker et al. 1995a,b) to be analysed in detail. The ability to measure efftux
direct1y and sensitively, makes it possible to estimate N exchange times for sub-
celIular compartments. Then, using labelling times <t1l2 for exchange of a com-
partment, it is possible to measure the unidirectional flux across a particular
membrane rather than the net N flux (balance between influx and efftux). For
example, knowing that t 1l2 for exchange of the cell walI and cytoplasm was ca.
20 to 30s, and ca. 6 to 8min, respectively, Kronzucker et al. (l995a,b) designed
protocols to measure plasma membrane influx in roots of Picea glauca (white
spruce), rather than the "quasi-steady" flux to the vacuole (Cram 1968). It is
impossible to distinguish the component fluxes (influx and efftux) by ion deple-
tion (Chap. 13).
As with llC, the high energy of the 13N yemission (0.5 MeV maximum emis-
sion energy) permits non-invasive translocation studies using whole plants
and repeated observations on the same plant after alIowing for isotopic decay
(Caldwell et al. 1984). Yet, experiments lasting as long as 75 to 90min are feasi-
bIe if coupled to methods such as HPLC, to investigate the metabolic fates of
13N tracer (Meeks 1993). However, because of the short t 1l2 , longer-term studies
(» 1 h, e.g., to resolve vacuolar characteristics by compartmental analysis), are
impossible. In this case, lsN becomes the isotope of choice.
Special problems associated with the use of 13N falI into two categories;
both arise from its short t 1l2 (9.96 min). First, rapid decay dictates that much
more radioactivity (> 100x) be delivered to the laboratory than is required
for longer-lived tracers, to compensate for decay during the experiment.
Second, alI procedures must be rigorously pre-planned to avoid needless
time losses, and experiments performed close to a cyclotron where the 13N is
produced.
The high energies of the positron emission and r radiation (Appendix)

demand extensive shielding of the experimental area with lead bricks as well as
the development of automated methods to minimise exposure to this radiation.
Radiation dosimeters are mandatory as well as Pb aprons and eye shielding.
Precautions adopted for the use of 13N at the University of British Columbia are
as folIows. Generation of 13N03- by proton irradiation of water (Meeks 1993) is
controlled remotely from outside a "hot" celIlined with 50 mm Pb bricks. Trans-
fer from the cyclotron to the u.B.C. campus hospital is in an underground pneu-
matic tube, the 2.5 km distance being covered in 2 min. Transfer of 13N from the
collection point to the laboratory (O.5km) is by car, the tracer being contained
inside a Pb cylinder. The cylinder is carried on a specially constructed trolIey
to distance the experimenter as far as possible from the 13N. Contaminants
(13 N0 2-, 13NH/, 18p) are removed in a fume hood lined on alI sides with 50mm
Pb bricks (Lee and Clarkson 1986; Lee and Drew 1986; Siddiqi et al. 1989), where
conversions to 13N02-, or 13NH/ are done as required. It is important to con-
sider the possibility of exposing individuals in adjacent laboratories or offices.
A sheet of 15mm Pb-impregnated glass (180 x 270mm) mounted on a sliding
track over the open fume hood, provides a "safe" viewing window. Within the
hood itself is a large mirror so that direct viewing of the isotope is avoided.
Isotope spills decay within the hour leaving behind only trace amounts of 13N.
12.5.4 32p
32p is one of the most widely used radioisotopes in root studies. It has been used
to measure P uptake, to visualise root locations in soil, and P depletion zones
around roots. Some of the most interesting applications have been in con-
junction with its sister isotope, 33p, such as interspecific root competition for P
(Caldwell et al. 1987) and distinguishing between P influx and its simultaneous
efflux (Elliot et al. 1984).
The usual aim of adding 32p to soil is to label the plant-available pool(s)
without increasing P availability significantly. But labelling soil with 32p presents
problems due to the exchange of inorganic P between solid and liquid phases.
Drew and Nye (1970) incubated 32P-Iabelled soil for up to 2 months before using
it in a P uptake experiment to allow isotopic equilibration to occur. In such
applications, the 32p content of the soil solution should, idealIy, be measured
sequentially to check the approach of the soil to isotopic equilibrium. Phosphate
does not move greatly in soil (unlike N0 3-, for example) and considerable
spatial variability of [Pl can occur over short distances «10 cm). If a uniform
distribution of 32p with depth is required, inject it at several depths rather
than rely on downward percolation of solution which produces very uneven
distributions of 32p (Jupp et al. 1987).
The high energy [3- particles represent a significant external hazard when
dispensing isotope stocks and conducting experiments in solution culture.
However, field experiments are possible because the activity is greatly reduced
when P becomes adsorbed in the soil. The risk of leaching is low, and the short
t 1l2 means that activity rapidly decays. Vose (1980, p. 268 et seq.) discusses the
practicalities of using 32p in field experiments. If 32p is to be used in a p-fertiliser
experiment, uniformly labelled material cannot be produced by adding carrier-
free 32p to a unlabelled P fertiliser; it must be purchased ready-Iabelled (Vose
1980, p. 278).
JupP et al. (1987) described a neat, non-destructive, semi-quantitative
method to measure 32p uptake by plants in soil. They measured the activity of
32p in a leaf to which a scintillation probe had been attached. This allowed the
time-course, but not the amount, of 32p uptake to be obtained.
12.5.5 86Rb and 42K
Many physiological studies of K uptake by roots have actually used an analogue,

86Rb. This is because 86Rb has a more convenient t 1l2 compared with that of the
obvious K isotope, 42K (Appendix 12.1). 86Rb has been used extensively in studies
of uptake rate-ion concentration relations in the laboratory (e.g. Kochian and
Lucas 1982, 1983). 86Rb has also been used to estimate root activity at specific
locations in soil (Vose 1980, p. 291). However, a number of studies demonstrate
that discrimination between Rb and K may occur dur ing influx and efflux
across root membranes; the effects depend on species and experimental condi-
tions (Jensen and Kylin 1980; Behl and Jeschke 1982; Drew and Saker 1984).
Experiments of this type using 86Rb must, therefore, be interpreted with caution;
at best they provide a qualitative indication of possible treatment effects on K
uptake and transport.
86Rb is usually quantified by Cerenkov counting, and 42K by Cerenkov or r
counting (Behl and Jeschke 1982; Jeschke 1982), although other techniques have
been used, e.g. LSC and end window Geiger counting (Jensen and Kylin 1980;
Drew and Saker 1984).
12.5.6 35 5
This is the isotope of choice in studies of S uptake by roots. But as suitable

IRMS systems become available, we may see greater use of 34S in the future.
Robinson et al. (1994) supplied a fraction of a wheat root system in soil with 35S
labelled K2S0 4 solution at a [S] of 2 moI m- 3 and 35S activity of ca. 460 kBq cm-3;
this was more than adequate to allow the isotope to be quantified in shoots 7
days later. Other examples of the use of 3SS with hydroponically grown plants
are in Lee (1982), Holobradâ and Kubica (1989) and Sunarpi and Anderson
(1996).
3SS is also widely used in studies of root metabolism. The effects of treat-
ments on current protein synthesis can be determined by incubating roots in
3SS-methionine and measuring its incorporation into proteins. Quantitative
autoradiography of 3sS-labelled proteins on two-dimensional gels has shown
that synthesis of specific proteins may be regulated by the carbohydrate supply
to the root (Baysdorfer and VanDerWoude 1988).
12.5.7 Other Radioisotopes
Convenient isotopes are available for a number of other elements of interest to

root scientists induding some important plant micro-nutrients, e.g. Mn C4Mn)
e
Zn (6S Zn ), Mo (99Mo), Fe C9 Fe), Cu (64 Cu), CI (36 C1), plus Ca (4S Ca) and Na 2Na).
In most cases they have been used as tracers for measuring ion uptake, translo-
cation and compartmentation. For example, Lasat and Kochian (1997) used 6SZn
to study the mechanisms of Zn hyper-accumulation in Thlaspi caerulescens, a
candidate species for phyto-remediation of Zn-contaminated soils. Similarly,
s9Fe has been used to investigate the role of phyto-siderophores (non-
proteinogenic amino acids) in Fe mobilisation in the rhizospheres of and uptake
by Fe-deficient grasses (Marschner et al. 1989; Treeby et al. 1989), and the sites
of Fe uptake by maize roots (Kashirad et al. 1973). Details of radiation type, Emax,
and t1/2 can be found in Vose (1980).
12.6 Concluding Remarks
This chapter has illustrated the breadth of applications that stable and radioiso-
topes have in root studies. As instrumentation evolves, this range is sure to
increase, e.g. we expect to see in the near future many uses of electronic auto-
radiography to show how roots work in real time. In some are as, further devel-
opments are needed in basic chemistry, rather than in isotope chemistry per se,
e.g. in the isolation of plant and soil N species for reliable 81S N determinations.
If that particular development occurs, there will be a parallel need for improved
theories to usefully interpret such 81S N data. In physiology, too, certain break-
throughs are required, e.g. micro-sampling of plant tissues and organelles for
the measurement of their isotopic composition to address still-unanswered
questions about metabolite compartmentation. Whatever the developments, the
use of isotopic (or any other) techniques should be driven by biological ques-
tions, and not the other way around.
Acknowledgments. The Scottish Crop Research Institute and the Scottish

Agricultural College receive grant-in-aid from the Scottish Executive Rural
Affairs Department.
References
Amato M (1983) Determination of carbon \2C and 14C in plant and soil. Soil Biol Biochem 15:
611-612
Androssoff GL, van Kessel C, Pennock DJ (1994) Landscape-scale estimates of dinitrogen
fixation by Pisum sativum by nitrogen-15 natural abundance and enriched isotope
dilution. Biol Fertil Soils 20: 33-40
Aslam M, Travis RL, Huffaker RC (1994) Stimulation of nitrate and nitrate effiux by ammonium
in barley (Hordeum vulgare 1.) seedlings. Plant Physiol106: 1293-1301
Atkinson q, Farrar JF (1983) Allocation of photosynthetically-fixed carbon in Festuca ovina 1.
and Nardus stricta 1. New Phytol95: 519-531
Bacic G, Ratkovic S (1987) NMR studies of radial exchange and distribution of water in maize
roots: the relevance of modelling and exchange kinetics. J Exp Bot 38: 1284-1297
Baldwin JP, Tinker PB (1972) A method for estimating the lengths and spatial patterns of two
interpenetrating root systems. Plant Soil37: 209-213
Balesdent J, Mariotti A (1996) Measurement of soil organic matter turnover using \3C natural
abundance. In: Boutton TW, Yamasaki S (eds) Mass spectrometry of soils. Dekker, New
York, pp 83-111
Baysdorfer C, VanDerWoude WJ (1988) Carbohydrate responsive proteins in the roots of
Pennisetum americanum. Plant Physiol 87: 566-570
Behl R, Jeschke WD (1982) Potassium fluxes in excised barley roots. J Exp Bot 33: 584-600
Bengtsson B (1982) Uptake and translocation of calcium in cucumber. Physiol Plant 54: 107-111
Berta G, Tagliasacchi AM, Fusconi A, Gerlero D, Trotta A, Scannerini S (1991) The mitotic cyele
in root apical meristems of Allium porrum 1. is controlled by the endomycorrhizal fungus
Glomus sp. strain E3' Protoplasma 161: 12-16
Bhat KKS, Nye PH (1973) Diffusion of phosphate to plant roots in soil. 1. Quantitative
autoradiography of the depletion zone. Plant Soil38: 161-175
Bingham IT, Blackwood JM, Stevenson EA (1998) Relationship between tissue sugar content,
phloem import and lateral root initiation in wheat. Physiol Plant 103: 107-113
Blackburn TH, Knowles R (1993) Introduction. In: Knowles R, Blackburn TH (eds) Nitrogen
isotope techniques. Academic Press, London, pp 1-10
Boutton TW (1991a) Stable carbon isotope ratios of natural materials: II. Atmospheric,
terrestrial, marine and freshwater environments. In: Coleman DC, Fry B (eds) Carbon
Boutton TW (1991b) Stable carbon isotope ratios of natural materials: 1. Sample preparation
and mass spectrometric analysis. In: Coleman DC, Fry B (eds) Carbon isotope techniques.
Academic Press, London, pp 155-171
Boutton TW, Yamasaki S (eds) (1996) Mass spectrometry of soils. Marcel Dekker, New York
Brownlee C, Duddridge JA, Malibari A, Read DJ (1983) The structure and function of mycelial
systems of ectomycorrhizal roots with special reference to their role in forming inter-plant
connections and providing pathways for assimilate and water transport. Plant Soil 71:
433-443
Bryce JH, ap Rees T (1985) Comparison of the respiratory metabolism of Plantago lanceolata 1.
and Plantago major 1. J Exp Bot 36: 1559-1565
Caldwell CD, Fensom DS, Bordeleau L, Thompson RG, Drouin R, Didsburg R (1984)
Translocation of 13N and llC between nodulated roots and leaves in alfalfa seedlings.
J Exp Bot 35: 431-443
Caldwell MM, Richards JH (1989) Hydraulic lift: water efflux from upper roots improves
effectiveness of water uptake by deep roots. Oecologia 79: 1-5
Caldwell MM, Virginia RA (1989) Root systems. In: Pearcy RW, Ehleringer 1, Mooney HA, Runde!
PW (eds) Plant physiological ecology: field methods and instrumentation. Chapman and
Hali, London, pp 367-398
Caldwell MM, White RS, Moore, Camp LB (1977) Carbon balance, productivity and water use of
cold-winter desert shrub communities dominated by C3 and C4 species. Oecologia 29:
275-300
Caldwell MM, Eissenstat DM, Richards JH, Allen MH (1985) Competition for phosphorus: dif-
ferential uptake from dual-isotope labeled soi! interspaces between shrub and grass.
Science 229: 384-386
Caldwell MM, Richards JH, Manwaring JH, Eissenstat DM (1987) Rapid shifts in phosphate
acquisition show direct competition between neighbouring plants. Nature 327: 615-616
Cheng W, Coleman DC, Carroll CR, Hoffman CA (1994) Investigating short-term carbon fiows
in the rhizospheres of different plant species using isotopic trapping. Agron J 86: 782-788
Clark FE (1977) Internal cycling of ISnitrogen in a shortgrass prairie. Ecology 58: 1322-1333
Clarkson DT, Sanderson J, Russell RS (1968) Ion uptake and root age. Nature 220: 805-806
Clarkson DT, Gojon A, Saker LR, Wiersema PK, Purves JV, Tillard P, Arnold GM, Paans AJM,
Vaalburg W, Stulen I (1996) Nitrate and ammonium infiuxes in soybean (Glycine max)
roots - direct comparison of 13N and IsN tracing. Plant Cell Environ 19: 859-868
Clowes FAL (1971) The proportion of cells that divide in root meristems of Zea mays 1. Ann
Bot 35: 249-261
Coleman DC, Fry B (eds) (1991) Carbon isotope techniques. Academic Press, London
Cooper HD, Clarkson DT (1989) Cycling of amino-nitrogen and other nutrients between shoots
and roots in cereals - a possible mechanism integrating shoot and root in the regulation
of nutrient uptake. J Exp Bot 40: 753-762
Coupland D (1986) Sample preparation for liquid scintillation counting. Aspects Appl Biol 11:
55-66
Cram WJ (1968) Compartmentation and exchange of chloride in carrot root tissue. Biochim
Biophys Acta 163: 339-353
Dalal RC (1979) Simple procedure for the determination of total carbon and its radioactivity in
soils and plant materials. Analyst 104: 151-154
Dalton FN (1989) Plant root water extraction studies using stable isotopes. In: Loughman BC
(ed) Structural and functional aspects of transport in roots. Kluwer, Dordrecht, pp 151-155
d'Arcy-Lameta A, Bompeix G (1991) Systemic transport of tritiated phosphonate in tomato
plantlets (Lycopersicon esculentum Mill). Pestic Sci 32: 7-14
Dawson TE (1993) Water sources of plants as determined from xylem-water isotopic com-
position: perspective on plant competition, distribution and water relations. In: Ehleringer
JR, Hali AE, Farquhar GD (eds) Stable isotopes and plant carbon-water re!ations. Acade-
mic Press, London, pp 465-496
Deane-Drummond CE, Glass ADM (1983) Compensatory adjustments in ion fiuxes into barley
(Hordeum vulgare 1. cv. Betzes) seedlings in response to differential root/shoot growth
temperature. J Exp Bot 34: 1711-1719
Dick PS, ap Rees T (1975) The pathway of sugar transport in roots of Pisum sativum. J Exp Bot
26:305-314
Dieuaide-Noubhani M, Raffard G, Canion P, Pradet A, Raymond P (1995) Quantification of
compartmented metabolic fluxes in maize root tips using isotope distribution from l3C_ or
14C-Iabelled glucose. J Biol Chem 270: 13147-13159
Drew MC, Nye PH (1970) The supply of nutrient ions by diffusion to plant roots in soil. III.
Uptake of phosphate by roots of onion, leek and rye-grass. Plant Soil 33: 545-563
Drew MC, Saker LR (1984) Uptake and long distance transport of phosphate, potassium and
chloride in relation to internal ion concentrations in barley: evidence of non-allosteric
regulation. Planta 160: 500-507
Drew MC, Nye PH, Vaidyanathan LV (1969) The supply of nutrient ions by diffusion to plant
roots in soil. 1. Absorption of potassium by cylindrical roots of onion and leek. Plant Soil
30:252-270
Ehleringer JR (1991) l3C/12C fractionation and its utility in terrestrial plant studies. In: Coleman
DC, Fry B (eds) Carbon isotope techniques. Academic Press, London, pp 187-200
Elliot GC, Lynch J, Lăuchli A (1984) Influx and efflux of Pin roots of intact maize plants. Plant
Physiol 76: 336-341
Ezra G, Rusness DG, Lamoureux GL, Stephenson GR (1985) The effect of CDAA (N,N,-
Diallyl-2-chloroactetamide) pretreatments on subsequent CDD injury to corn (Zea mays
1..). Pestic Biochem Physiol23: 108-115
Farrar JF (1985) Fluxes of carbon in roots of barley plants. New Phytol99: 57-69
Farrar JF (1991) Starch turnover: its role in source-sink relations and a comparison with the
role of sucrose. In: Bonnemain JL, Delrot S, Lucas WJ, Dainty J (eds) Recent advances in
phloem transport and assimilate compartmentation. Quest Editions, Nantes, pp 213-223
Farrar JF (1993) Carbon partitioning. In: HalI DO, Scurlock JMO, Bolhar-Nordenkampf HR,
Leegood RC, Long SP (eds) Photosynthesis and production in a changing environment: a
field and laboratory manual. Chapman and HaU, London, PP 232-246
Fox GG, Ratcliffe RG, Robinson SA, Slade AP, Stewart GR (1992) Detection of 15N-Iabelled
ammonium. In vivo 15N NMR versus mass spectrometry. J Magnet Reson 96: 146-153
Glass ADM, Thompson RG, Bordeleau L (1985) Regulation ofN0 3- influx in barley. Studies using
13NO,-. Plant Physiol 77: 379-381
Gleixner G, Danier H-J, Werner RA, Schmidt H-L (1993) Correlations between the l3C content
of primary and secondary plant products in different ceU compartments and that in
decomposing basdiomycetes. Plant Physiol102: 1287-1290
Goh KM (1991) Bomb carbon. In: Coleman DC, Fry B (eds) Carbon isotope techniques.
Gordon AL, Ryle GJA, Mitchell DF, Powell CE (1985) The flux of 14C-Iabelled photosynthate
through soyabean root nodules during N2 fixation. J Exp Bot 36: 756-769
Goulding KH, Slater RJ (1994) Radioisotope techniques. In: Wilson K, Walker J (eds) Principles
and techniques of practical biochemistry. Cambridge University Press, Cambridge, pp
227-274
Gregory PJ, Atwell BJ (1991) The fate of carbon in pulse-Iabelled crops of barley and wheat. Plant
Soil136: 205-213
Grignon N, Halpern S, Gojon A, Fragu P (1992) 14N and 15N imaging by SIMS microscopy in
soybean leaves. Biol Cell 74: 143-146
Handley LL, Raven JA (1992) The use of natural abundance of nitrogen isotopes in plant
physiology and ecology. Plant Cell Environ 15: 965-985
Handley LL, Scrimgeour CM (1997) Terrestrial plant ecology and 15N natural abundance; the
present limits to interpretation for uncultivated systems with original data from a Scottish
old-field. Adv Ecol Res 27: l33-212
Handley LL, Daft MJ, Wilson J, Scrimgeour CM, Ingleby K, Sattar MA (1993) Effects of the
ecto- and VA-mycorrhizal fungi Hydnagium carneum and Glomus clarum in the 8'N
and 8'c values of Eucalyptus globulus and Ricinus communis. Plant CeH Environ 16: 375-
382
Harris D, Paul EA (1991) Techniques for examining the carbon relationships of plant-microbial
symbioses. In: Coleman DC, Fry B (eds) Carbon isotope techniques. Academic Press,
London, pp 39-52
Harrison AF, Harkness DD, Bacon PJ (1990) The use of bomb- 14C for studying organic
matter and N and P dynamics in a woodland soi!. In: Harrison AF, Ineson P, Heal OW (eds)
Nutrient cycling in terrestrial ecosystems: field methods, application and interpretation.
Eisevier, London, pp 246-258
Hodge A, Stewart J, Robinson D, Griffiths BS, Fitter AH (1998) Root proliferation, soil fauna and
plant nitrogen capture from nutrient-rich patches in soi!. New Phytoll39: 479-494
Holobrada M, Kubica S (1989) The role of maize root tissues in sulphate absorption and
radial transport. In: Loughman BC (ed) Structural and functional aspects of transport in
roots. Kluwer, Dordrecht, pp 101-105
Jackson LE, Schimei JP, Firestone MK (1989) Short-term partitioning of ammonium and nitrate
between plants and microbes in a annual grassland. Soil Biol Biochem 21: 409-415
Jenkinson DS, Fox RH, Rayner JH (1985) Interactions between fertilizer nitrogen and soil
nitrogen - the so caHed 'priming' effect. J Soil Sci 36: 425-444
Jensen P, Kylin A (1980) Effects ofionic strength and relative humidity on the efflux ofK+ (86Rb)
and Ca2+ ("Ca) from roots of intact seedlings of cucumber, oat and wheat. Physiol Plant
50: 199-207
Jeschke WD (1982) Shoot-dependent regulation of sodium and potassium fluxes in roots of
whole barley seedlings. J Exp Bot 33: 601-618
Jones DL, Darrah PR (1992) Re-sorption of organic components by roots of Zea mays 1. and its
consequences in the rhizosphere 1. Re-sorption of l·C labelled glucose, mannose and citric
acid. Plant Soil143: 259-266
Jones DL, Darrah PR (1994) Amino-acid influx at the soil-root interface of Zea mays 1. and its
implications in the rhizosphere. Plant Soil163: 1-12
Jones DL, Darrah PR (1996) Re-sorption of organic components by roots of Zea mays 1. and its
consequences in the rhizosphere III. Characteristics of sugar influx and efflux. Plant Soil
178: 153-160
JupP AP, Newman EI, Ritz K (1987) Phosphorus turnover in soil and its uptake byestablished
Lolium perenne plants. J Appl Ecol 24: 969-978
Kashirad A, Marschner H, Richter C (1973) Absorption and translocation of 59Fe from various
parts of the corn root. Z Pflanzenernahr Bodenk l34: l36-147
Keith H, Oades JM, Martin JK (1986) Input of carbon to soil from wheat plants. Soil Biol Biochem
18:445-449
Kirchmarr H, Pichlmayer F, Gerzabek MH (1996) Sulfur balances and sulfur-34 abundance in a
long-term fertilizer experiment. Soil Biol Biochem 59: 174-178
Knowles R, Blackburn TH (eds) (1993) Nitrogen isotope techniques. Academic Press, London
Kochian LV, Lucas WJ (1982) Potassium transport in corn roots. 1. Resolution of kinetics into a
saturable and linear component. Plant Physiol 70: 1723-1731
Kochian LV, Lucas WJ (1983) Potassium transport in corn roots. II. The significance of the root
periphery. Plant Physiol 73: 208-215
Kraus M, Fusseder A, Beck E (1987) Development and replenishment of the P-depletion zone
around the primary root of maize during the vegetative period. Plant Soil101: 247-255
Kronzucker HJ, Siddiqi MY, Glass ADM (1995a) Compartmentation and flux characteristics of
nitrate in spruce. Planta 196: 674-682
Kronzucker HJ, Siddiqi MY, Glass ADM (1995b) Compartmentation and flux characteristics of
ammonium in spruce. Planta 196: 691-698
Kronzucker HJ, Siddiqi MY, Glass ADM (1997) Conifer root discrimination against soil nitrate
and the ecology of forest succession. Nature 385: 59-61
Kuhn AJ, Bauch J, SchrOder WH (1995) Monitoring uptake and contents of Mg, Ca and K in
Norway spruce as influenced by pH and Al, using microprobe analysis and stable isotope
labelling. Plant SoilI68-169: 135-150
Lasat MM, Kochian LV (1997) Physiological basis for Zn hyperaccumulation in Thlaspi
caerulescens. In: Flores HE, Lynch JP, Eissenstat D (eds) Radical biology: advances and
perspectives on the function of plant roots. American Society of Plant Physiologists,
Rockville, pp 139-149
Laskey RA (1990) Radioisotope detection using X-ray film. In: Slater RJ (ed) Radioisotopes in
biology: a practical approach. Oxford University Press, Oxford, pp 87-107
Lajtha K, Michener RH (eds) (1994) Stable isotopes in ecology and environmental science.
Blackwell, Oxford
Lazof DB, Rufty TW, Redinbaugh MG (1992) Localization of nitrate absorption and trans-
location within morphological regions of the corn root. Plant Physioll00: 1251-1258
Lazof DB, Goldsmith JKG, Rufty TW, Suggs C, Linton RW (1994) The preparation of cryosec-
tions from plant tissue: an alternative method appropriate for secondary ion mass
spectrometry studies of nutrient tracers and trace metals. J Microsc 176: 99-109
Lee RB (1982) Selectivity and kinetics of ion uptake by barley plants following nutrient
deficiency. Ann Bot 50: 429-449
Lee RB, Clarkson DT (1986) Nitrogen-13 studies of nitrate fluxes in barley roots. 1. Com-
partmental analysis from measurements of 13N effiux. J Exp Bot 37: 1753-1767
Lee RB, Drew MC (1986) Nitrogen-13 studies of nitrate fluxes in barley roots. II. Effect of plant
N-status on the kinetic parameters of nitrate influx. J Exp Bot 37: 1753-1767
Lee RB, Lewis MJ (1994) Synthesis and 15N-labelling of glutamine and glutamate in maize roots
during early stages of 15N-ammonium assimilation. J Exp Bot 45: 767-777
Lee RB, Ratcliffe RG (1993) Nuclear magnetic resonance studies of the location and function of
plant nutrients in vivo. Plant SoilI55-156: 45-55
Lefebvre DD, Glass ADM (1982) Regulation of phosphate influx in bariey roots: effects of
phosphate deprivation and reduction of influx with provision of orthophosphate. Physiol
Plant 54: 199-206
Lichtfouse E, Dou S, Girardin C, Grably M, Balesdent J, Behar F, Vandenbroucke M (1995)
Unexpected 13C-enrichment of organic components from wheat crop soils: evidence for the
in situ origin of soil organic matter. Org Geochem 23: 865-868
Ludlow MM, Troughton JH, Jones RJ (1976) A technique for determining the proportion of C3
and C4 species in plant samples using stable natural isotopes of carbon. J Agric Sci 87:
625-632
Macklon AES, Sim A (1981) Cortical ceH fluxes and transport to the stele in excised root
segments of Allium cepa L. 4. Calcium as affected by its external concentration. Planta 152:
381-387
Macko SA, Ostrom NE (1994) Pollution studies using stable isotopes. In: Lajtha K, Michener RH
(eds) Stable isotopes in ecology and environmental science. Blackwell, Oxford, pp 45-62
Marschner H, Richter C (1973) Akkumulation und Translokation von K+, Na+ und Caz+ bei
Angebot zu einzelnen Wurzelzonen von Maiskeimpflanzen. Z Pflanzenerniihr Bodenkd
135: 1-15
Marschner H, Treeby M, Romheld V (1989) Role of root-induced changes in the rhizosphere for
iron acquisition in higher plants. Z Pflanzenerniihr Bodenkd 152: 237-240
Martin F, Botton B (1993) Nitrogen metabolism of ectomycorrhizal fungi and ectomycorrhiza.
Adv Plant Pathol 9: 83-102
Mary B, Mariotti A, Morei JL (1992) Use of l3C variations at natural abundance for studying
the biodegradation of root muci!age, roots and glucose in soi!. Soil Biol Biochem 24:
1065-1072
McLaughiin MJ, Aiston AM, Martin JK (1987) Transformations and movement of P in the
rhizosphere. Plant Soil97: 391-399
Meeks JC (1993) l3N techniques. In: Knowles R, Blackburn TH (eds) Nitrogen isotope techniques.
Meharg AA (1994) A critical review of labeling techniques used to quantify rhizosphere
carbon-flow. Plant Soil166: 55-62
Milchunas DG, Lauenroth WK, Singh JS, Cole CV, Hunt HW (1985) Root turnover and
production by 14C dilution: implications of carbon partitioning in plants. Plant Soil88: 353-
365
Minchin PEH, Thorpe MR (1996) What determines carbon partitioning between competing
sinks? J Exp Bot 47: 1293-1296
Minchin PEH, Thorpe MR, Farrar JF (1994) Short-term control of root:shoot partitioning. J Exp
Bot 45: 615-622
Mordacq L, Mousseau M, Deleens E (1986) A C-13 method of estimation of carbon allocation
to roots in a young chestnut coppice. Plant Cell Environ 9: 735-739
Nepstad DC, de Carvalho CR, Davidson EA, Jipp PH, Lefebvre PA, Negreiros GH, da Si!va ED,
Stone TA, Trumbore SE, Viera S (1994) The role of deep roots in the hydrological and
carbon cycles of Amazonian forests and pastures. Nature 372: 666-669
Nye PH, Tinker PB (1977) Solute movement in the soil-root system. Oxford: Blackwell Scientific
Publications
Pfeffer PE, Shachar-Hill Y, Becard G, Rolin D, Douds DD (1997) Nutrient transport and
metabolism in the life cycle of arbuscular mycorrhizae as examined by NMR spectroscopy.
In: Flores HE, Lynch JP, Eissenstat D (eds) Radical biology: advances and perspectives
on the function of plant roots. American Society of Plant Physiologists, Rockville, pp 187-
209
Polley HW, Johnson HB, Mayeux HS (1992) Determination of root biomasses of three species
grown in a mixture using stable isotopes of carbon and nitrogen. Plant Soil142: 97-106
Powlson DS, Barraclough D (1993) Mineralization and assimilation in soil-plant systems.
In: Knowles R, Blackburn TH (eds) Nitrogen isotope techniques. Academic Press, London,
pp 209-242
Powlson DS, Hart PBS, Pruden G, Jenkinson DS (1986) Recovery of ISN -labelled fertilizer applied
in autumn to winter wheat at four sites in eastern England. J Agric Sci 107: 611-620
Prime D, Frith B (1990) Radioisotope use. In: Slater RJ (ed) Radioisotopes in biology: a
practical approach. Oxford University Press, Oxford, pp 7-30
Prosser SJ, Scrimgeour CM (1995) High-precision determination of zH/IH in Hz and HzO by
continuous-flow isotope ratio mass spectrometry. Anal Chem 34: 1992-1997
Raney F, Vaadia Y (1965) Movement of tritiated water in the root system of Helianthus annuus
in the presence and absence of transpiration. Plant Physiol40: 378-388
Robbins 1, Rost TL (1987) Chlorsulfuron inhibition of cell cyele progres sion and the recovery of
Gl arrested celIs by!le and Val. J Plant Growth Regul 6: 67-74
Robinson D, Conroy JP (1999) A possible plant mediated feedback between elevated CO"
denitrification and the enhanced greenhouse effect. Soil Biol Biochem 31: 43-53.
Robinson D, Millard P (1986) Short-term uptake rates and partitioning of nitrogen in a potato
crop. J Exp Bot 38: 551-557
Robinson D, Scrimgeour CM (1995) The contribution of plant C to soil CO, measured using 8'c.
Soil Biol Biochem 27: 1653-1656
Robinson D, Smith KA (1991) Analysis of nitrogen isotope ratios by mass spectrometry.
In: Smith KA (ed) Soil analysis, 2nd edn. Dekker, New York, pp 465-503
Robinson D, Linehan DJ, Gordon DC (1994) Capture of nitrate from soil by wheat in relation to
root length, nitrogen inflow and availability. New Phytol128: 297-305
Robinson D, Handley LL, Scrimgeour CM (1998) A theory for I'N/ 14N fractionation in
nitrate-grown vascular plants. Planta 205: 397-406
Rorison IH, Robinson D (1986) Mineral nutrition. In: Moore PD, Chapman SB (eds) Methods in
plant ecology, 2nd edn. BlackwelI, Oxford, pp 145-213
Schnyder H (1992) Long-term steady-state labelling of wheat plants by use of natural 13COi l 'CO,
mixtures in an open, rapidly turned-over system. Planta 187: 128-135
Scrimgeour CM (1995) Measurement of plant and soil water isotope composition by direct
equilibration methods. J Hydroll72: 261-274
Scrimgeour CM, Robinson D (2001) Stable isotope analyses and applications in soil and
environmental science. In: Smith KA, Cresser MS (eds) Soil and environmental analysis.
Dekker, New York
Shone MGT, Flood AV (1980) Studies on the uptake and loss of water by barley roots in relation
to changes in root resistance. J Exp Bot 31: 1147-1159
Siddiqi MY, Glass ADM, Ruth TJ, Fernando M (1989) Studies of the regulation of nitrate influx
by barley seedlings using 13NO,-. Plant Physiol 90: 806-813
Siddiqi MY, Glass ADM, Ruth TJ, Rufty TW (1990) Studies of the uptake of nitrate in barley 1.
Kinetics of 13NO,- influx. Plant Physiol 93: 1426-1432
Simonnet G (1990) The scintillation counter. In: Slater RJ (ed) Radioisotopes in biology: a
practical approach. Oxford University Press, Oxford, pp 31-85
Slater RJ (ed) (1990) Radioisotopes in biology: a practical approach. Oxford University Press,
Oxford
Smith KA (ed) (1991a) Soil analysis, 2nd edn. Dekker, New York
Smith KA (1991b) Nuclear and radio-chemical analysis. In: Smith KA (ed) Soil analysis, 2nd
edn. Dekker, New York, pp 325-375
Smith KA, Cresser MS (eds) (2001) Soil and environmental analysis. Dekker, New York
Spence RD, Sharpe PJH (1991) Uses and procedures for llC. In: Coleman DC, Fry B (eds) Carbon
Sunarpi, Anderson JW (1996) Distribution and redistribution of sulfur supplied as [35S) sulfate
to roots during vegetative growth of soybean. Plant Physiol110: 1151-1157
Svejcar T, Boutton TW (1985) The use of stable carbon isotope analysis in rooting studies.
Oecologia 67: 205-208
Svejcar T, Boutton TW, Trent JD (1990) Assessment of carbon allocation with stable isotope
labelling. Agron J 82: 18-21
Swinnen J, van Veen JA, Merckx R (1994) I'C pulse-Iabelling of field-grown spring wheat: an
evaluation of its use in carbon budget estimations. Soil Biol Biochem 26: 161-170
Thompson LML, Sanders GE, Pallett KE (1986) Experimental studies into the uptake and trans-
location of foliage applied herbicides. Aspects Appl Biolll: 45-54
Thorburn PJ, Ehleringer J (1995) Root water uptake of field-growing plants indicated by
measurements of natural-abundance deuterium. Plant Soill77: 225-233
Treeby M, Marschner H, Romheld V (1989) Mobilization of iron and other micronutrients from
a calcareous soil by plant-borne, microbial, and synthetic metal chelators. Plant Soil114:
217-226
Trumbore SE (1996) Applications of accelerator mass spectrometry to soil science. In: Boutton
TW, Yamasaki S (eds) Mass spectrometry of soils. Dekker, New York, pp 311-340
Voroney RP, Winter JP, Gregorich EG (1991) Microbe/plant/soil interactions. In: Coleman DC,
Fry B (eds) Carbon isotope techniques. Academic Press, London, pp 77-99
Vose PB (1980) Introduction to nuclear techniques in agronomy and plant biology. Pergamon
Press, Oxford
Warembourg FR, Kummerow J (1991) Photosynthesis/translocation studies in terrestrial
ecosystems. In: Coleman DC, Fry B (eds) Carbon isotope techniques. Academic Press,
London, pp 11-37
Wassenaar LI (1995) Evaluation of the origin and fate of nitrate in the Abbotsford Aquifer using
the isotopes of 15N and 180 in N0 3-. Appl Geochem lO: 391-405
Watkins NK, Fitter AH, Graves JD, Robinson D (1996) Quantification using stable carbon
isotopes of carbon transfer between C3 and C. plants linked by a common mycorrhizal
network. Soil Biol Biochem 28: 47l-477
White JWC, Cook ER, Lawrence JR, Broecker WS (1985) The D/H ratios of sap in trees:
implications for water sources and tree ring D/H ratios. Geochim Cosmochim Acta 49:
237-246
White PJ, Clarkson DT, Earnshaw MJ (1987) Acclimation of potassium influx in rye (Secale
cereale) to low root temperatures. Planta 17l: 377-385
Wong SC, Osmond CB (1991) Elevated atmospheric partial pressure of CO2 and plant growth.
III. Interactions between Triticum aestivum (C3 ) and Echinochloa frumentacea (C.) during
growth in mixed culture under different CO 2, N nutrition and irradiance, with emphasis
on below-ground responses using (jJ3 C value of root biomass. Aust J Plant Physiol 18:
137-152
Yoshida S, Eguchi H (1994) Environmental analysis of aerial O2 transport through leaves for
root respiration in relation to water uptake in cucumber plants (Cucumis sativus L.) in
02-deficient nutrient solution. J Exp Bot 45: 187-192
Zocchi G, Hanson JB (1982) Calcium influx into corn roots as a result of cold shock. Plant Physiol
70:318-319
o
,o
""
Appendix. Commonly Used Isotopes in Root Studies (Based on Vose, 1980, pp. 143-144)'
Element Isotope Emax (MeV) tl/2 Method Selected examples of applications

in root studies
f3 r
Calcium 40Ca Sb
45Ca 0.254 162 days LSC;AR T: Marschner and Richter (1973); Jensen and
Kylin (1980); Macklon and Sim (1981);
Bengtsson (1982); Zocchi and Hanson (1992)
Carbon "C 0.9600 20.5min (c.p.) rcounting GR: Spence and Sharpe (1991)
T: Minchin et al. (1994); Minchin and Thorpe
(1996)
l2C Sb
l3C MS GR: Boutton (1991b); Ehleringer (1991)

T: Svejcar et al. (1990); Mordacq et al. (1986);
Lichtfouse et al. (1995); Hodge et al. (1998)
NT: Ludlow et al. (1976); Svejcar and
Boutton (1985); Wong and Osmond (1991);
Mary et al. (1992); polley et al. (1992);
Robinson and Scrimgeour (1995); Watkins
et al. (1996)
14C 0.156 5730 years LSC;AR; MS GR: Warembourg and Kummerow (1991);
.....
(for NT); NMR Harris and Paul (1991); Voroney et al. (1991); -;-.
Goh (1991); Meharg (1994) I:P
5·
Ot>
T: Caldwell et al. (1977); Swinnen et al. (1994); ::r
P>
NT: Harrison et al. (1990); Nepstad et al. (1994) S
AR: Brownlee et al. (1983) ~
~
Hydrogen IH Sb ...
N
2H (D) MS;NMR GR: Dawson (1993) iA
T: Caldwell and Richards (1989) ...oo
NT: White et al. (1985); Thorburn and 'ti
ID
Ehleringer (1995); NA: Dalton (1989) ~
n
AR: Kochian and Lucas (1983) ::r
::J
3H (T) 0.0181 12.26 years LSC:AR GR: Vose (1980) li·
c
T: Raney and Vaadia (1965); Shone and Flood ID
(1980) '"
Nitrogen 13N 1.19 0.51 9.7min (c.p.) ycounting GR: Meeks (1993)
T: Lee and Clarkson (1986); Siddiqi et al.
(1989,1990); Kronzucker et al. (1995a, b);
Clarkson et al. (1996)
14N Sb
15N s MS;NMR GR: Jenkinson et al. (1985); Robinson and
Smith (1991); Handley and Raven (1992);
Blackburn and Knowles (1993)
T: Clark (1977); Powlson et al. (1986); Robinson
and Miliard (1986); Jackson et al. (1989);
Grignon et al. (1992); Clarkson et al. (1996)
NT: Androssoff et al. (1994)
Oxygen 16 0 Sb
18 0 S MS T: Yoshida and Eguchi (1994)

NT: Wassenaar (1995)
Phosphorus 31p Sb
32p 1.71 14.3 days Cerenkov; AR GR: Vose (1980)
T: Drew and Nye (1970); Elliott et al. (1984);
Caldwell et al. (1985); JupP et al. (1987);
McLaughlin et al. (1987)
AR: Bhat and Nye (1973) le
>l>-
Appendix. (cont.) a
N
Element Isotope Emax (MeV) tll2 Method Selected examples of applications

in root studies
f3 r
33p 0.25 25 days LSC;AR T: Elliott et al. (1984); Caldwell et al. (1985);
McLaughlin et al. (1987)
AR: Kraus et al. (1987)
Potassium 39K Sb
41K MS T: Kuhn et al. (1995)

42K 3.55, 1.52 12.4h r counting; Cerenkov T: Marschner and Richter (1973); Behl and
2.0 Jeschke (1982)
Rubidium BSRb Sb
86Rb 1.77, 1.08 18.7 days r counting; Cerenkov T: Clarkson et al. (1968); Drew et al. (1969);
0.7 Kochian and Lucas (1982,1983); White et al.
(1987)
Sulphur 32S Sb
34S MS GR: Macko and Ostrom (1994)

NT: Kirchmarr et al. (1996)
35S 0.168 86.7 days LSC;AR T: Lee (1982); Holobrada and Kubica (1989);
Robinson et al. (1994); Sunarpi and
Anderson (1996)
AR: Baldwin and Tinker (1972); Baysdorfer
......
and Vanderwoude (1988)
t:!:I
-
, Abbreviations are as follows: AR autoradiography, c.p. cyclotron produced, GR general reference, LSC liquid scintillation counting, MS mass spec- ~
trometry, NMR nuclear magnetic resonance, NT natural tracer, s stable, T tracer. El
b most abundant isotope. :!
~
CHAPTER 13
Assessing the Ability of Roots

for Nutrient Acquisition
Ch. Engels l , G. Neumann 2 , T.S. Gahoonia3, E. George 2 , and M. Schenk4
1 Agraroko!ogie, Universitat Bayreuth, 95440 Bayreuth, Germany
2 Institut fUr pftanzenernahrung (330), Universităt Hohenheim, 70593 Stuttgart, Germany
3 Department of Agricultural Science, The Royal Veterinary & Agricultura! University,
1871 Frederiksberg C, Copenhagen, Denmark
4 Institut ftir Pftanzenernăhrung, Universităt Hannover, 30419 Hannover, Germany
CONTENTS
13.2 Modification of Nutrient Availability in the Rhizosphere 406

13.2.1 Collection of Root Exudates 406
13.2.1.1 Collection in Trap Solutions 407
Aerated Trap Solutions 407
Percolation Techniques 407
Composition of Trap Solutions and Sample Processing 408
13.2.1.2 Localized Sampling 409
13.2.1.3 Collection from Soil-Grown Plants 410
13.2.1.4 Collection of Mucilage 411
13.2.1.5 Factors Affecting Recovery of Root Exudates 411
Externa! Concentration 411
Effects of Microorganisms 412
Root Injury 414
13.2.2 Separation of Rhizosphere from Bulk Soil 414
Shaking Off Soil Adhering to Roots 414
Collection of Soil at Well-Defined Distances from Roots 415
Thin Slicing of Rhizosphere Soil 415
13.2.3 In Situ Measurements in the Rhizosphere 417
13.2.3.1 Rhizosphere pH 418
13.2.3.2 Redox Processes 419
Reduction of Fe(III) 421
Reduction of Mn(IV) 421
Redox State in General 421
13.2.3.3 Complexation of Aluminium 422
Dedicated to Professor Dr. Dres. h.c. H. Marschner

404 Ch. Engels et al.
13.2.3.4 Detection of Low Molecular Weight Rhizosphere

Compounds 422
13.2.3.5 Detection of Enzyme Activities 423
13.2.3.6 Assessment of Nutrient Distribution in the Rhizosphere
by Autoradiography 423
Plant Culture 423
Soil Labelling, Preparation and Evaluation
of the Radioautogram 424
13.3 Nutrient Uptake 424
13.3.1 Principles of Nutrient Uptake 424
Mechanisms for Nutrient Uptake 424
Transporters 425
Metabolic Control 425
Parameters for Estimating the Ability of Roots
to Take Up Nutrients 426
13.3.2 Nutrient Uptake into Intact Plants 426
13.3.2.1 Nutrient Accumulation in Plants Under In Situ Conditions
(Williams Formula) 426
13.3.2.2 Methods to Assess the Variation of Nutrient Uptake Within
an Intact Root System 429
Injection of Tracers into Various Soil Zones 429
Local Application via Nutrient-Containing Agar 430
Compartmented Containers 431
Separation of Root Regions After Short Periods of Uptake 431
Measurement of Ion Activities at the Root Surface
with Microelectrodes 432
13.3.3 Nutrient Uptake into Excised Roots 434
Principles 434
Methodology 434
Root Characteristics Assessed 435
Problems 435
13.3.4 Collection of Xylem Exudate 436
Principle 436
Collection of Xylem Sap 436
Assessment 437
13.3.5 Contribution of Mycorrhiza to Nutrient Acquisition 438
General 438
Methods for Determining Mycorrhizal Colonization 439
Methods for Determining the Effect of Mycorrhizal Fungi
on Nutrient Uptake 439
13.3.6 Measurement of Influx, Efflux, and Uptake Kinetics 440
13 Assessing the Ability of Roots for Nutrient Acquisition 405
Estimation of Influx 440

Estimation of Efflux 442
Estimation of Uptake Kinetics 443
Measurement of Rate of Depletion of the External Solution 444
Calculations 445
13.3.7 Nutrient Analysis 446
Sample Preparation 446
Chemical Analysis 447
Instruments for Analysis 447
13.4 Concluding Remarks 447
Further Reading 448
References 449
13.1 Introduction
Nutrient acquisition by roots from soil is a complex process which is dependent

on several root features: (1) morphological root characteristics, including myc-
orrhizal associations, which determine the extent of the interface between
plant and soil (2) ability to modify the nutrient availability in the rhizosphere,
and (3) ability for nutrient uptake through the plasma membranes (for reviews
see Barber 1984; Clarkson 1985; Marschner 1995). The relative importance of
these factors for nutrient acquisition is dependent on environmental conditions
and the specific nutrient, particularly its chemical availability and mobility
in the soil.
The most convenient way to estimate nutrient acquisition by roots is to
measure the nutrient accumulation rate in the whole plant between two harvest
dates. Such a measurement, however, gives no information about the contribu-
tion of the various root characteristics mentioned above. Thus, statements e.g.
on the ability for nutrient acquisition of a particular genotype, or on the con-
sequence of a specific agronomic measure for nutrient uptake, can be drawn
only for the specific environmental conditions under which the plants have been
cultured. Furthermore, accumulation rate is difficult to measure in trees which
are several years old. Therefore, it is often necessary to determine separately
the contributions of the various root characteristics involved in nutrient ac-
quisition. The measurement of morphological root characteristics has been
described in detail in the preceding chapters. In this chapter an overview is
given on the principal methods of measuring or estimating root-induced mod-
ifications in the rhizosphere, the ability of roots to take up nutrients, and the
contribution of mycorrhizal fungi to these processes.
13.2 Modification of Nutrient Availability
in the Rhizosphere
Roots may modify the physical, chemical and biochemical characteristics of the
soil in their vicinity (rhizosphere) by uptake and rele ase of different compounds
(Grayston et al. 1996). These alterations lead to modifications of the biomass
and species composition of microorganisms which in turn change rhizosphere
conditions. In view of the large diversity of substances which can modify the
rhizosphere soil, this chapter does not cover analytical methods for measure-
ment of the various compounds. It is confined to techniques for collection of
root exudates directly and for sampling rhizosphere, as opposed to bulk, soil
for subsequent extraction and analysis of rhizosphere compounds.
13.2.1 Collection of Root Exudates
Various techniques have been used for the collection of root exudates in a solu-
tion surrounding the roots (Fig. 13.1). Of course only water soluble, diffusible
('_._....' ; C
A B
! ~
.
Closed tap glass
tube (remeved
i alter preculture)
.. Nutrient solution
Nr Pipelte tip
_ Nroutiet
XAD-4 resin .... -..
Membrane filler
hydrophobic.
Glasswool ..... . . 0-2 IIm
Dialysis membrane - Nrinletl

nutrient
solution
Fig. 13.1. Examples of plant culture systems used for the collection of root exudates. APlant
culture in solid medium (quartz sand); root exudates are eluted from the sand with the circu-
lating nutrient solution and may be selectively retained in exchange resin at the bottom of the
system (Tang and Young 1982, reprinted with permission of ASPP). B Plant culture in nutrient
solution; root exudates are concentrated within a dialysis membrane around the roots (Tadano
and Sakai 1991, reprinted with permission of Japanese Society of Soi! Science and Plant Nutri-
tion). C Hydroponic system for plant culture and collection of root exudates under axenic
culture; for further details see Box 13.1 (von Win!n et al. 1995, reprinted with permission of The
New Phytologist)
exudates can be measured in the solution, whereas exudates adhering to the root
surface (e.g. mucilage) are not or only partially sampled by these techniques.
13.2.1.1 Collection in Trap Solutions
Aerated Trap Solutions. Root exudates may be collected by transferring the

roots from the culture medium into an aerated trap solution. By this technique
root exudation may be assessed during a well-defined period of time. How-
ever, its application should be restricted to plants grown in nutrient solution.
Attempts to apply the technique to culture systems using solid media (soil,
sand) almost certainly cause root injury and thus introduce artefacts. On the
other hand, the physical nature of the medium surrounding the roots affects
the total amount and composition of substances released by the roots (Boeuf-
Tremblay et al. 1995), and exudation is stimulated by the mechanical impedance
during growth in solid media. Simulation of the mechanical forces imposed on
roots growing in soil may be achieved by addition of glass ballotini to the nutri-
ent solution (Barber and Gunn 1974; Sch6nwitz and Ziegler 1982) or by growing
the plants in sand culture using percolation techniques for exudate sampling
(Fig. l3.1A).
For the collection of high molecular weight compounds, roots may be
enclosed in dialysis membranes which are non permeable to compounds
exceeding a certain molecular weight, to increase the concentration of these
compounds in the vicinity of the roots, and thus, to facilitate their analytical
detection (Fig.l3.1B; Sakai and Tadano 1993). In open (flowing culture) systems
(Fig. l3.1A) the exudates are continuously displaced and may be removed
from the circulating nutrient solution e.g. by exchange resins (Petersen and
Bottger 1991).
Percolation Techniques. Collection of exudates from undisturbed root systems

of plants grown in solid media (e.g. sand, vermiculite) may be achieved by per-
colating the culture vessels with the trap solution. Prior to collection, rhizos-
phere products accumulated during the culture period due to chemi cal and
microbial modification of root exudates, should be removed by repeated wash-
ings with the trap solution. Thereafter, the vessels are incubated for a short
period (e.g. 30-60 min), or repeatedly percolated with the trap solution, and
exudates released during this time are subsequently collected with the perco-
late (Miyasaka et al. 1991; Hiilster and Marschner 1994; Shepherd and Davies
1994a; Johnson et al. 1996). For quantitative evaluation of rhizodeposition of
organic compounds, isotopic labelling techniques using 14C02 (Keith et al. 1986;
Hodge et al. 1996; Johnson et al. 1996) or 15NH3 (Janzen 1990; Reining et al. 1995)
may be employed before percolation.
Control samples should be collected from culture vessels without plants.

Since sorption of certain exudate compounds to the solid matrix of the culture
medium cannot be excluded, recovery experiments and comparison with results
obtained from experiments in nutrient solution are essential.
Composition of Trap Solutions and Sample Processing. Generally, trap solutions

used for collection of root exudates are nutrient solutions of the same compo-
sition as the culture medium (Miyasaka et al. 1991; Shepherd and Davies 1994a;
Johnson et al. 1996), solutions of 0.5-2 mM CaS04 or CaCl2 (Ohwaki and Hirata
1990), or distilled water (Lipton et al. 1987; Hiilster and Marschner 1994; von
Win!n et al. 1994). The nature of the trap solution may influence exudation rates
of certain compounds and subsequent sample preparation as well. Since the
osmotic strength of commonly used nutrient solutions is generally low, osmo-
tic effects of the trap solution on the membrane permeability are not to be
expected even if distilled water is used as the trap solution. However, exudation
of low molecular weight compounds may be affected by the ionic composition
of the medium surrounding the roots, providing counter ions for transport
mechanisms. For example, efflux of amino acids and sugars from plant roots
seems to occur by passive diffusion, but effective resorption by active transport
mechanisms has been reported (Jones and Darrah 1993, 1994) which can be
affected by the ionic composition of trap solutions. Release of organic acids
from plant roots, which is probably mediated by active efflux mechanisms
(Dinkelaker et al. 1989; Jones and Darrah 1995), may be affected in a similar
way. However, comparison of either nutrient solution, or water, ar 20 mM KCl
as trap solutions for amina acids released from roots of Brassica napus 1.,
showed no differences in exudation during collection periods between 0.5 and
6h (Shepherd and Davies 1994b). In order to avoid nutrient depletion in the
roots during collection of exudates, water or CaSOJCaCl2 solutions should only
be used if collection is confined to short time periods.
Collection of root exudates in trap solutions generally yields a very diluted
solution of exudates. Before analysis, it is often necessary to increase the exudate
concentration by decreasing the volume of the trap solution e.g. by freeze drying
ar rotavaporation, depending on the stability of the compounds of interest.
Depending on the composition of the trap solution, volume reduction may lead
to high salt concentrations which may interfere with subsequent analysis or even
result in precipitation of exudates (e.g. Ca-citrate, Ca-oxalate, proteins). Thus, if
possible, interfering salts should be removed by use of ion exchange resins prior
to volume reduction. Exudates may be also enriched and prefractionated by tech-
niques of solid phase extraction (ion exchange resins for charged compounds,
reversed phase matrices for lipophilic compounds). High molecular weight con-
stituents (polysaccharides, proteins) can be precipitated by addition of organic
solvents (e.g. 80% v/v) or separated by ultrafiltration.
13.2.1.2 Loca/ized Samp/ing
Exudation is not uniform over the whole root system, and considerable longi-
tudinal gradients may exist in exudation rates along the roots. For example,
enhanced release of organic acids in P-deficient rape (Hoffiand et al. 1989) or
of phytosiderophores in Fe-deficient maize (Romheld 1991) is largely confined
to the root tip (apical cm). As the density of microorganisms which may catab-
olize exudates is higher in basal root zones (Uren and Reisenauer 1988), the
localization of enhanced exudation in the root tip is expected to increase the
efficiency of exudates in mobilizing P and Fe in the rhizosphere. Moreover, a
distinct diurnal rhythm exists for the release of phytosiderophores from Fe-
deficient plants (Marschner et al. 1986), which may also increase the proba-
bility that phytosiderophores mobilize Fe in the rhizosphere before being
degraded by microorganisms. For the assessment of the ecological relevance of
exudation for nutrient acquisition, it may therefore be important to measure
the variability of exudation in space and time.
The variability of exudation in time can be measured well with the above
mentioned methods, e.g. by frequent exchange of the solution surrounding the
roots. There are also several methods to evaluate the variability of exudation
along the root axis. For' example, Hoffiand et al. (1989) spread roots of plants
grown in nutrient solution on a glass plate, and placed small plastic rings (diam-
eter 1.2 cm) over the root zone from which exudation was to be measured. After
covering the roots outside the rings with agar solution, a small volume of nutri-
ent solution (200,ul) was pipetted into the rings. The solution in the rings was
collected after incubation for 2 h at room temperature and high humidity, and
analyzed for exudates. Delhaize et al. (1993) transferred solution-grown plants
to large Petri dishes filled with nutrient solution, and sealed the plastic rings
for separation of distinct root zones with vacuum grease. Alternatively, the
whole root system may be covered with a thin prefixed layer of 1% agarose (low
electroendosmosis type; Marschner et al. 1987). During incubation the agarose
layer should be moistened with distilled water to prevent desiccation. After
incubation the agarose is cut into pieces according to the position of the
root segments below, from which 'exudation is to be measured, and subsequently
analysed for exudates (Marschner et al. 1987). However, exact localization of
the diffusion zones of exudates is difficult when using agarose sheets for
exudate collection. It is also possible to spread the root system on a layer of
fleece which is covered by a sheet of filter paper and moistened with water. Roots
selected for exudate sampling are underlaid by small sheets of polyethylene
foil, and different root zones are placed between two discs of moist chro-
matography paper (5 mm in diameter, preparative quality) which was previ-
ously washed with methanol and distilled water. The rest of the root system is
covered with filter paper, and moistened with nutrient solution to avoid desic-
cation. During incubation the filter discs are occasionally remoistened with
lO.u1 of distilled water. After 3 h the filter discs are removed and exudates are
extracted in a small volume (I00.u1 per three discs) of distilled water by cen-
trifugation (Iamin) in a microliter centrifuge. In the case of organic acids, the
samples can be analyzed directly by HPLC, and quantitative determinations of
exudates originating from single root tips or root segments with a spatial
resolution of 5 mm even from plants with low exudation rates are possible
(Neumann, unpublished results). Similarly, spatial variation of fiavonoids
released from roots of soybean seedlings has been examined by placing the
roots between two layers of cellulose acetate filters as a medium for sorption of
root exudates (Kape et al. 1992).
13.2. 1.3 Collection from Soil-Grown Plants
For collection of root exudates from soil-grown plants, as described above, pieces
of absorbent material (e.g. chromatography paper, glass fibre filters, solid agar
sheets, polyacrylamide gels) may be placed on the roots for several hours. During
collection the absorbent material has to be pres sed onto the roots to ensure
proper contact with them. To avoid root injury during uncovering from the soil,
exudates can be collected from roots growing along the internal surface of root
observation windows (rhizoboxes or windows installed in the field, see Chap. 8).
This also allows localized collection from well-defined root segments. To prevent
roots sticking to the plastic lid when the window is opened, a transparent plastic
sheet is placed between the front lid and the soil. Using small sheets of filter paper
as an absorbent material, temporal changes have been monitored in the concen-
tration of organic acids in the soil solution obtained from the rhizosphere of pro-
teoid roots of Hakea undulata (Neumann et al. 1995; Dinkelaker et al. 1996).
When using agarose gels for exudate collection, gel shrinking sometimes occurs
due to diffusion of water into the soil matrix.
In principle, in soil-grown plants the solution surrounding the roots can
also be collected and compared with the solution from the bulk soil. Recently,
Gottlein et al. (I996) described the construction of miniaturized suction cups
which allow the microscale collection of soil solution from the rhizosphere. The
solution is sampled in micro ceramic cells (outer diameter 1 mm, length 5 mm)
which are connected to a vacuum collecting device. AIso, extraction of root exu-
dates from rhizosphere soil (see Sect. 13.2.2), collected from proteoid roots of
Banksia integrifolia L. (Grierson 1992) and Lupinus albus L. (Dinkelaker et al.
1989; Gerke et al. 1994) has been reported.
For the techniques of exudate sampling from soil-grown plants described
above it has to be considered that exudates may be degraded by micro organ-
isms and adsorbed to the soil matrix. Thus, for collection of root exudates,
sterile absorbent material should be used (e.g. agar sheets, methanol-washed

filter paper), and the collection should be terminated before substantial micro-
bial colonization of the absorbent material occurs. Furthermore, recovery
experiments may be helpful to assess microbial degradation or sorption of exu-
dates to the soil matrix under the specific experimental conditions. In such
experiments, well-defined amounts of exudate compounds are added to exp-
erimental soil and recovery of these compounds is measured after incubation
under sterile (e.g. chloroform atmosphere) and non-sterile conditions.
13.2.1.4 Collection of Mucilage
Mucilage is a gelatinous, root-produced material mainly consisting of polysac-

charides and polyuronic acids, and is deposited primarily in the region around
the root tip. Possible functions of mucilage include protection of the root from
desiccation (Leiser 1968), reduction of friction between the growing root tip
and the soil, improvement of the root-soil contact, and complexation of toxic
minerals such as aluminium (Horst et al. 1982). For collection, it may be
removed from the root surface with a soft brush and transferred to cellulose
acetate filters (Horst et al. 1982), or it may be simply removed by use of forceps
(Dinkelaker et al. 1996). Mucilage formed on nodal roots of maize after immer-
sion into distilled water can be collected by vacuum suction (MoreI et al. 1986).
Also, ultrasonication treatments have been described to remove exudate com-
pounds adhering to the root surface without visible damage or influence on sub-
sequent plant growth (Barber and Gunn 1974).
13.2.1.5 Factors AHecting Recovery of Root Exudates
External Concentration. The total amount of substances released by the roots

is dependent on their concentration in the medium surrounding the roots. Low
external concentrations increase the transmembrane concentration gradient
and enhance the release, and decrease the reabsorption (retrieval) of exudates.
Therefore, it may be expected that exudate collection in systems with circulat-
ing nutrient solution or frequent replacement of the nutrient solution in static
systems may lead to unrealistically high exudation rates (Prikryl and Vancura
1980; Jones and Darrah 1993). On the other hand, in closed (static) systems (Fig.
13.1B,C) the concentration of exudates in the nutrient solution is increasing
with time, leading to higher retrieval into the roots as was demonstrated for
sugars and amino acids in root exudates of maize (Jones and Darrah 1993,
1994). Estimates of exudation for the two techniques may differ by an order of
magnitude and it is not clear which method most closely resembles exudation
in soil (Darrah 1996).
Effects of Microorganisms. Microorganisms can modify and degrade low mol-

ecular weight exudates. This process can be diminished in open systems when
exudates may be protected from microbial degradat ion by sorption to exchange
resins (Fig. 13.1A). Alternatively, plants can be grown in axenic culture (Fig.
13.1C). Various systems have been described to grow sterile plants in hydro-
ponic culture (e.g. Lipton et al. 1987; Jones and Darrah 1992; von Wiren et al.
1995; Box 13.1). It has to be kept in mind, however, that microorganisms affect
both the quality and quantity of exudates released from the roots. In culture
systems where the microorganisms are spatially separated from the roots by
a Millipore membrane ("microcosm", see Meharg and Killham 1991), it was
demonstrated that exudation may not only be affected through root infection
but also by microbial metabolites. Depending on the microbial inoculum, exu-
dation varied in the range of 3-34% of recently assimilated carbon, compared
with 1% in the absence of microorganisms (Meharg and Killham 1995).
BOX 13.1. Example of a Hydroponic System f or Sterile Plant Culture

(Wiren et al. 1995, Fig. 13.1C)
Prior to germination, surface sterilization of seeds is performed by
immersion for 1- 5min into 96% ethanol and 25min into 18% (v/v) H2 0 2 ,
followed by rinsing the seeds five times in sterile 2.5 mM CaS04 solution.
To select sterile seedlings, the seeds are germinated on sterile agar plates
containing half-strength casein-peptone/soymeal-peptone nutrient broth
(Merck No. 5459) for 3 days at 25°C in the dark. Uninfected seedlings are
inserted into pipette tips in autoclaved culture vessels containing 600 mI
autoclaved nutrient solution and the shoot is covered with a closed top-
glass tube for a preculture period of 9- 10 days. Thereafter the top glass
tube is removed and the pipette tips are sealed with 45°C melted
paraffin. Sterilized cotton plugs are inserted between the pipette tips to
maintain axenic conditions in the culture vessel. The nutrient solution is
continuously aerated through 0.2 f.1M membrane filters and the air inlet
filter is also used to replace the nutrient solution. Aliquots of the nutrient
solution are regularly checked for sterility via incubation for 2 days at
28°C on nutrient agar.
Several methods are avaHable for soil sterilization such as sterilization by

heat, gamma-irradiation and fumigation with toxic chemicals (Johnson and
CurlI972; Box 13.2). However, soH sterilization procedures can change chemi-
cal soil properties and may lead to increased plant availability of nutrients such
as nitrogen, phosphorus and toxic elements (Johnson and CurlI972). Further-
more, it is very difficult to prevent re-colonizat ion of sterilized soH by microor-
ganisms, particularly dur ing long-term cultivation of plants.
BOX 13.2. Methods for Soil Sterilization

Sterilization by Heat
Dry heat pasteurization may be used to rid soi! of most fungal
propagules. Sieved soi! samples are heated twice to 70-80 DC for 24 h, each
followed by incubation at room temperature for 24 h. Steaming at 83 DC
for 30 min results in a better heat distribution in the soil and kiHs most
pathogenic fungi, nematodes, insects and weed seeds, but some bacteria
and viruses may survive dry or moist heat treatment. The soi! structure
seems not to be significantly affected by these methods (Johnson and
Curl1972). Autoc1aving of homogenized soi! samples (120 DC, 1.1 atm,
30- 60 min) destroys most microorganisms but the soi! structure is
altered significantly, and may result in the liberation of nutrients and
toxic elements (Alef 1995).
Chemical Sterilization
Soi! samples (250 g or more) can be disinfected by fumigation in a large
desiccator containing a beaker of alcohol-free chloroform. The desiccator
is evacuated and incubated in the dark for 18- 24 h at 25 DC. Thereafter
re-evacuation is repeated three times for 3 min to remove the chloroform.
Fumigation with chloroform destroys a11 microorganisms but may also
significantly alter the soi! structure (Powlson and ]enkinson 1976a).
Fumigation with other chemicals such as ethylene oxide (explosive) or
methyl bromide can be used with simi!ar results (Johnson and Cur!
1972).
Gamma lrradiation
Soil samples sealed in polyethylene bags are exposed to an irradiation
dose of 2.5 Mrad at a rate of about 2 Mrad h- 1 • The procedure eliminates
a11 microorganisms and there seems to be only little effect on soi!
structure. However, enzyme activities such as those of phosphatase and
urease are stiH detectable after the sterilization treatment (Johnson and
Curl 1972).
To prevent microbial degradat ion of exudate compounds, various antibi-

otics such as Rifampicin/Tetracyc1in (50/25ppm; Schwab et al. 1983; Ohwaki
and Hirata 1992), Cefatoxim/Trimethaprim (30120ppm; Azaizeh et al. 1995) or
"Micropur" (Amann and Amberger 1989) have been added to trap solutions,
or agar sheets used as collection media for root exudates, or washing solutions
used for root pretreatment prior to exudate collection. However, antibiotics may
have phytotoxic effects depending on plant species and dosage (Smart et al.
1995). Reducing the time per iod of collection of root exudates from plants
grown under non-sterile conditions (Tagaki et al. 1984; Cakmak and Marschner
1988) is another strategy to overcome problems of microbial degradation of

exudate compounds.
Root Injury. During transfer of root systems into trap solutions, as well as during
application of absorbent materials or preparation of root systems for exudate
collection, roots may be exposed to stress due to mechanical damage (e.g. rupture
of roots hairs, epidermal celIs) or rapid change in the environmental conditions
(e.g. temperature, pH, oxygen availability), which may affect exudation rates. The
general influence ofhandling during exudate collection on plant growth may be
assessed by comparing plants either subjected or not subjected to the collection
procedure (Barber and Gunn 1974). The effects ofthe routine procedure of han-
dling on root exudation may roughly be estimated by comparison of the quali-
tative and quantitative changes in patterns of exudation of plants subjected to
handling procedures of different intensity.
13.2.2 Separation of Rhizosphere from Bulk Soil
Shaking Of! Soil Adhering to Roots. This simple method for separation of rhi-
zosphere and bulk soil involves careful excavation of plant roots. The soil
loosely adhering to the roots (regarded as bulk soil) is collected by gently shaking
the roots until most soil has dropped onto elean paper. Separation of soil
from roots is facilitated by a short period of air drying, which dries the soil
without making the roots brittle. The more tightly adhering soil (regarded as
rhizosphere soil) is then shaken off separately on another elean paper. Alterna-
tively, tightly adhering soil can be washed off from the roots and air dried.
AlI visible root debris should be removed using forceps or by sieving to
avoid contamination of the rhizosphere soil with root residues during analysis.
The method can be applied to plants grown in pots or in the field (Smiley 1974).
The method was first described by Starkey (1931) and was used to study e.g. the
pH (Riley and Barber 1971; Smiley 1974), or the concentrations of various
mineral elements in the rhizosphere (Sihna and Singh 1976; Hendriks and
Jungk 1981).
Using this method, rhizosphere soil can be obtained from plants growing
under natural conditions, both in pot and field experiments, without the need
for technical equipment. The main disadvantage of the method, however, is
the low spatial resolution of sampling not only along the root axis but also in a
radial direction. Depending on the sand and elay content and thus the forma-
tion of aggregates, different amounts of soil are removed from the roots by
shaking (Hendriks and Jungk 1981). Therefore, the method is applicable mainly
for qualitative assessment of modifications in the rhizosphere, particularly if
large differences exist between rhizosphere and bulk soil.
Collection of Soil at Well-Defined Distances from Roots. Papavizas and Davey

(1961) designed a microsampler for obtaining soil at increasing distances (3 mm
increments) from the surface ofindividuallupin roots. The sample collector con-
sisted of seven thin-walled steel tubes, 50mm long with an inside diameter of
3 mm, soldered adjacent to each other in a plane with an overall width of 22 mm.
To collect the soil samples, the collector was inserted vertically into the soil im-
mediately adjacent and parallel to the primary root of lupin seedlings. Another
possibility for obtaining soil at well defined distances from the roots is to confine
the roots to certain soil zones by nets with a mesh width preventing penetration
by roots or root hairs but allowing free exchange of exudates and soil solution.
Boero and Thien (1979) used a system of concentric cylinders made of a nylon
netting and confined the root system to the central core. Helal and Sauerbeck
(1983) divided pots filled with soil byvertical stainless steel screens (pore diam-
eter 31.5 pm, open area 25%) into a central root zone, an adjacent soil zone of
10 mm width which could be penetrated by root hairs and an outer soil zone
which which was free from roots and root hairs. The principle of this method has
also been used to studyphosphorus depletion and pH changes at the soil-hyphae
interface of plants infected with mycorrhiza (Li et al. 1991). Instead of dividing
the pots vertically into different soil zones, horizontal separation can also be
achieved (Kuchenbuch and Jungk 1982, see below).
Thin Slicing of Rhizosphere Soil. Many of the processes involved in nutrient

acquisition occur in the soil zone very close to the roots. Therefore, it is neces-
sary to investigate this zone with high spatial resolution. This can be achieved
by slicing the rhizosphere soil into thin layers with a microtome. Farr et al.
(1969) held onion roots between two moist soil blocks. Onion roots do not have
many root hairs and do not branch. Therefore, they did not grow into the soil
blocks, and a defined root-soil interface was obtained. After 12 days, soil and
plant roots were separated. The soil was frozen in liquid nitrogen, sliced into
thin sections using a freezing microtome and analyzed for K, Ca and pH.
Kuchenbuch and Jungk (1982) separated roots from soil columns horizontally
by using a fine mesh (30 J..lm) nylon screen allowing the penetration of root hairs
but not that of roots. At harvest, the roots had formed a dense root mat on the
screen, so that the whole screen could be regarded as a root surface. The soil
column was separated from the screen (root surface) with a knife, quickly
frozen into liquid nitrogen and sliced at -4°C into 0.06 mm layers with a
freezing microtome. The method has been applied to study the rhizosphere
concentrations of potassium (Kuchenbuch and Jungk 1984), magnesium
(Seggewiss and Jungk 1988), phosphorus (Gahoonia et al. 1992), and alumin-
ium (Gahoonia 1993) and. phosphatase activity (Tarafdar and Jungk 1987).
Gahoonia and Nielsen (1991, 1992) developed an improved method for study-
ing the rhizosphere of plants at more advanced stages of development under
controlled nutritional conditions (Fig. 13.2). Plants were precultured for about
10 days in PVC tubes filled with vermiculite as illustrated in Fig. 13.2A. The
bottoms of the tubes were closed with fine nylon cloth impervious to roots. The
plants were supplied with a nutrient solution via two wicks covered with black
PVC foiI. When a root mat had developed on the nylon cloth after about 9 days,
the nylon was removed and the PVC tube with plants transferred to the test soil
columns (Fig. 13.2B). A nylon screen of inner mesh size 53 ţ1m at the bottom of
the upper soil column prevented penetrat ion of roots to the lower column. The
roots grew quickly into the 1 cm soH layer of the lower column and formed a
new root mat on top of the nylon screen. To maintain defined soH moisture
content, the soil columns were placed on a small cup-shaped sand bath, con-
nected to a reservoir of distilled water with a wick. To obtain rhizosphere soil
of defined distance from the roots, the columns were separated from the root
mats, quickly frozen in liquid nitrogen and sliced into thin layers using a freez-
ing microtome.
Legend
a wick covered with

black PVC foii
b vermiculite filled
in PVC tube
(L 10 cm, 0 4.4 cm)
c test soH filled in PVC tube
(L 3 cm, 0 5.6 cm)
d fine mesh nylon screen
(53 pM)
e root mat in 1 cm soil
layer
fine mesh nylon cloth
9 cup-shaped sand bath
h sand bath
I __~J __; water
~
nutrient solution
k pump
Fig. 13.2. Example of a plant culture system for obtaining rhizosphere soi! at defined distance
from the root surface (root mat). A Pre-experimental system. B Experimental system. For further
explanation see text (Gahoonia and Nielsen 1992; with kind permission from Kluwer Academic
Publishers)
13.2.3 In Situ Measurements in the Rhizosphere
Various non-destructive staining techniques have been developed to demon-

strate chemical changes in the rhizosphere (Fig. 13.3; Box 13.3; Box 13.4). Gen-
erally, sheets of agar, agarose, polyacrylamide or paper are used as carrier
matrices for indicators and substrates for enzyme or colour reactions. After
application to the surface of root systems of plants grown in hydroponic or soil
culture (rhizoboxes, root windows; see Chap. 8), qualitative, semiquantitative
or in some cases even quantitative evaluat ion of changes in the rhizosphere is
possible.
BOX 13.3. Methods for Visualization of Chemical Changes

in the Rhizosphere
Rhizosphere pH
Root systems are embedded in 0.75% (w/v) agar solutions (35 ac, pH 6)
conta in ing nutrient solution or CaS0 4 [1 mM], and 0.006% (w/v)
bromokresol-purple as a pH-indicator. After an incubation period of
between 15min and 2h, pH-induced colour changes along the roots
become visible. For sensitive detection of pH changes, the use of agar
with low buffering capacity (e.g. Merck No. 1614) or even agarose (1%
w/v) is recommended. A stock solution of 1% (w/v) bromokresol-purple
is prepared, dissolving the pH indicator in distilled water by dropwise
addition of 1 N NaOH during 20- 40 min. Complete dissolution is
indicated by a constant pH. Thereafter, the pH of the indicator solution
is adjusted to 6.0 by addition 1 N H2S04•
Reduction of Fe(III)
A 0.75% (w/v) agar solution conta in ing 100,uM Fe(III)EDTA and 300 ţLM
BPDS as chelator for Fe(II) can be used for embedding of the root
systems (Marschner et al. 1982) or for application of agar sheets (3 mm)
onto the root surface (Dinkelaker et al. 1993a). Reduction of Fe(III) is
indicated by formation of a red complex after 30 min to several hours.
Alternatively, the agar solution may be prepared with 100 ţLM
Fe(III)EDTA, 10mM Mes-buffer pH 4-4.6, and 200ţLM Ferrozin,
resulting in the formation of a violet complex of Fe(II)-Ferrozin.
Reduction of Mn (IV)
KMn0 4 [1 mM] is added to the agar solution (0.75% w/v) pH 6.0, and
kept at 50 ac for 2 h to induce formation of Mn02' After application to
plant roots, Mn02 reduction is indicated by decoloration in the brown
agar layer during an incubation period of between one and several days
(Marschner et al. 1982). Alternatively, filter pa pers (e.g. MN-260;
Macchery and Nagel, Diiren, Germany) are impregnated in a solution of

10mM KMn0 4 for 7h, forming Mn02 (brown) by oxidation of cellulose
in the paper. After washing (distilled water) and drying, the papers can
be stored for months without loss in colour intensity. The impregnated
paper sheets are applied with proper contact to the root surface (e.g.
rhizoboxes, root windows) and fixed with pins. Decoloration can be
observed after a period of 30 min to several hours.
Complexation of Aluminium
Gels of polyacrylamide (1 mm, 10% T); agar (3 mm, 0.75% w/v; Dinkelaker
et al. 1993a), or agarose (3 mm, 1% w/v; Dinkelaker et al. 1997) are used
as a carrier matrix for red-coloured Al-Aluminon complexes. For gel
preparation, a solution of Al(N0 3h [250 ,uM], Aluminon [88,uM] and
ascorbic acid [1.4,uM] adjusted to pH 4.2 is used for either incubation (2h,
polyacrylamide gels) or direct addition to the solutions of agar or agarose
after cooling to approximately 50 ac. Thereafter, gel sheets are placed onto
the root surface and covered with plastic foiI. Complexation of Al by
chelators with a higher affinity to Al compared with Aluminon (e.g.
organic acids in root exudates) is indicated by zones of decoloration
around the root, developing within a few hours.
13.2.3. 1 Rhizosphere pH
Root-induced changes of pH in the rhizosphere are mainly due to an imbalance

in cation-anion uptake which depends particularly on the nitrogen source,
or due to a deficiency in nutrients e.g. in phosphate or iron (Marschner and
Romheld 1983). For demonstration of pH changes in the rhizosphere, root
systems are embedded in agar solutions containing nutrient solution or CaS0 4 ,
and bromocresol-purple as a pH-indicator (Box 13.3). After an incubation
period of between 15 min and 2h, pH-induced colour changes along the roots
become visible (Fig. 13.3A), a yellow colour indicat ing acidification, and purple
or violet colours indicating alkalization. The technique is applicable for plants
grown in nutrient solution (Marschner et al. 1982) and in rhizoboxes in soil
(Marschner and Romheld 1983). As an alternative to embedding root systems
in agar solution, the roots may be covered with solid agar sheets, permitting
repeated measurements with the same plants. Quantitative measurements of the
pH changes in the rhizosphere are possible by insertion of microelectrodes into
the various pH zones qualitatively indicated by the colour changes in the agar
layer (Haussling et al. 1985; Gollany and Schumacher 1993). Recently, computer
controlled videodensitometry has been used to quantify pH changes in the rhi-
zosphere after visual demonstrat ion using the agar technique described above
(Jaillard et al. 1996).
8 O X13.4. Methods for Estimation of Enzyme Activity

in the Rhizosphere
Peroxidase
Peroxidase released from roots of sterile grown Festuca rubra has been
photometrically detected in the nutrient solution by addition of an equal
volume of 100mM citrate-phosphate buffer, pH 5.5, using 100mM H20 2
as a substrate and phenolic acids or guaiacol [10 mM] as hydrogen
donors for the enzymatic reaction (Vaughan et al. 1994).
Acid Phosphatase and Protease
For determination of acid phosphatase activity, the production of
p-nitrophenol by hydrolysis of p-nitrophenyl phosphate is measured
photometrically (Bartlett and Lewis 1973). Sin ce in many cases acid
phosphatase was found to be predominant1y associated with rhizodermal
ceH walls (Bieleski and Johnson 1972) or released in the mucilage (Felipe
et al. 1979), whole root systems or root cuttings are incubated in the
reaction buffer at 20°C for 20- 60min. (Bielski and Johnson 1972).
A non-destructive method for visualization of phosphatase activity in
soil-grown plants (rhizoboxes, root windows) has been described by
Dinkelaker and Marschner (1992) based on the enzymatic hydrolysis of
l -naphtylphosphate by root acid phosphatase, yielding l-naphtol as a
reaction product which forms a red complex with Fast Red TR
(diazotized 2-amino-5-chloro-toluene 1,5-naphtalene disulfonate). Filter
papers impregnated with the reagents are placed onto the root surface of
plants grown in rhizoboxes and activity of acid phosphatase is indicated
by development of red colour zones within approximately 2 h (Dinkelaker
and Marschner 1992; Dinkelaker et al. 1996). Similarly, protease activity
was visualized by decoloration of polyacrylamide gels (10% T)
conta in ing gelatine (0.5%), dye-Iabelled by Remazol-Brilliant-Blue, which
were placed onto the root surface (8. Dinkelaker et al. unpubl.).
13.2.3.2 Redox Processes
Plant roots are able to influence the redox state of the rhizosphere soil either
by reduction or oxidation. Reduction of iron or manganese oxides due to the
activity of plasma membrane-bound reductases (Bienfait et al. 1983) or by
release of reducing substances (Marschner et al. 1986) such as phenolic acids
or carboxylic acids (Hether et al. 1984; Dinkelaker et al. 1993b, 1996) can
increase the availability of iron and manganese especially in soils of neutral
or alkaline pH (Cairney and Ashford 1989). In contrast, in marsh plants such as
rice, growing in wet or flooded soils, excessive uptake of reduced iron, man-
Fig. 13.3. Examples for the demonstrat ion of chemical and biochemical modifications of the
rhizosphere; further details in the text, and Boxes 13.3 and 13.4. a Formation of yellow colour
in agar containing bromocresol-purple, induced by dec1ine of pH around roots of P-deficient
buckwheat (V. Romheld, unpubl.). b Formation of a red complex in agar containing BPDS,
induced by the reduction of Fe3+ to Fe'+ in Fe-deficient groundnut (left; V. Romheld, unpubl.). c
Decolorization in agar containing brown MnO" induced by the reduction of Mn4+ to Mn' +
(Dinkelaker et al. 1993b; with kind permission from Kluwer Academic Publishers). d Oxidation
power of roots of Phragmites australis (Armstrong et al. 1992, reprinted with permis sion ofThe
New Phytologist). Above: formation of blue colour in agar containing colourless leuco methyl-
ene blue, induced by oxidation to methylene blue. Below: iron oxidation also becomes visible as
an orange-brown precipitate in the rhizosphere of soil-grown plants. e Decoloration in agar con-
taining red Al-aluminon complexes induced by root exudates of Lupinus luteus forming more
stable complexes with Al (B. Dinkelaker, unpubl.). fFormation of redish-brown colour on filter
papers soaked with 1-naphtyl phosphate and Fast Red TR, induced by the formation of 1-
naphtol through phosphatases (B. Dinkelaker, unpubl.)
ganese and sulfides is avoided by an increase in the redox potential of the rhi-
zosphere relative to the bulk soil. In these plant species an internal channel
system (aerenchyma) allows diffusion of oxygen from the shoot to the roots and
subsequent release into the rhizosphere (Trolldenier 1988). For demonstration
of redox changes in the rhizosphere agar gels or filter papers impregnated with
redox indicators are applied to the root surface (Box 13.3).
Reduction of Pe(III). The method most frequently used for demonstration

of Fe(III) reduction is an agar technique where reduction of Fe(III) is
indicated by the formation of a red complex between Fe(II) and BPDS
(bathophenanthroline-disulfonic acid; Marschner et al. 1982; Fig. 13.3B). Instead
of BPDS, K3Fe( CN)6 has also been used as a redox indicator (Brown and Ambler
1974). For quantitative evaluat ion of the Fe(III) reducing capacity of roots from
plants grown in nutrient solution, the formation of a red complex between
Fe(II) and Ferrozine (3-(2-pyridyl)-S,6-diphenyl-1,2,4-triazine-p,p'-disulfonic
acid) has been measured spectrophotometrically at 562 nm in nutrient solution
(pH 5.4) containing 100,uM Fe(III)EDTA and 300,uM Ferrozine (Horst et al.
1992).
Reduction of Mn(IV). Reduction of Mn(IV) in the rhizosphere has also been

demonstrated by use of the agar technique described above (Box 13.3). After
application of agar sheets with Mn02 to plant roots, Mn0 2 reduction is indi-
cated by decolorat ion in the brown agar layer (Marschner et al. 1982; Fig. 13.3C).
A modification of the method to increase its sensitivity and applicability for
soil-grown plants has been described by Dinkelaker et al. (1993a). Based on the
method of Uren (1981), Mn0 2--impregnated paper sheets are applied to the
surface of roots growing e.g. in rhizoboxes (Box 13.3).
Redox State in General. For quantitative measurements of redox changes in the

rhizosphere, platin microelectrodes have been employed (Schaller and Fischer
1985; Flessa and Fischer 1992) which are inserted into the rhizosphere soil. Leuco
methylene blue (Methylene blue [20-25 mgl-I] reduced to the colourless form by
addition of Na2S204 [200 mgl-Il) has been frequently used in combination with
agar techniques to visualize oxidative changes in the rhizosphere (Trolldenier
1988; Armstrong et al. 1992; Fig. 13.3D), whereas reductive changes were demon-
strated by use of methylene blue (Karpov and Potapov 1975).
An agar medium which was blackened by precipitated ferrous sulfide was
used to demonstrate the oxidizing power of rice roots (Trolldenier 1988). FeS
precipitat ion was obtained by reduction of FeS04 [5mM] with Na 2S [2-4mM]
which was added to the agar solution at 50°e. FeS oxidation was indicated by
transparent zones in the rhizosphere of rice plants cultivated for 11 days in the
semi-solid agar medium containing FeS and nutrient solution.
13.2.3.3 Complexation of Aluminium
Aluminium (Al) toxicity is a major factor limiting plant growth in strongly

acidic soils. Depression of root elongation is the most obvious toxic effect of Al
on plant growth (Delhaize and Ryan 1995). Plant species and cultivars vary
widely in their resistance to Al toxicity. One possible cause for Al tolerance is
detoxification of Al by complexation either within the plant or in the rhizos-
phere (Miyasaka et al. 1991; Delhaize and Ryan 1995).
Complexation of Al in the mucilage covering the root apex has been
demonstrated by hematoxylin staining (Horst et al. 1982). AIso, root-derived
organic acids such as malic, citric and oxalic acids may complex and thus detox-
ify Al in the rhizosphere (Miyasaka et al. 1991; Delhaize et al. 1993; Pellet et al.
1995). A technique to demonstrate Al complexation in the rhizosphere of soil-
grown plants by applying agar containing red-coloured Al-aluminon complexes
to the roots has been described by Dinkelaker et al. (l993a,b, 1997; Box l3.3).
Complexation of Al by chelators with a higher affinity to Al compared with alu-
minon (e.g. organic acids in root exudates) is indicated by zones of decoloration
around the root, developing within a few hours (Fig. 13.3E). Control experi-
ments should be carried out, measuring Al concentrations in the different gel
zones, to make sure that decoloration is really a result of Al complexation and
not of root uptake of the water-soluble Al-aluminon complex.
13.2.3.4 Detection of Low Molecular Weight

Rhizosphere Compounds
Exudation of citric acid from proteoid roots of white lupin (Lupinus albus L.)
grown in a calcareous soil has been detected by eye due to precipitation of crys-
taIs of calcium citrate in the rhizosphere (Dinkelaker et al. 1989). Low mol-
ecular weight compounds may be sampled on absorbent materials such as
chromatography paper, agarose gels (Dinkelaker et al. 1996) or cellulose acetate
filters (Kape et al. 1992). For example, phenolic compounds in the rhizosphere
of several tree species grown in rhizoboxes have been visualized after a 4-h
collection period with chromatography paper (MN 260, Macchery and Nagel,
Diiren, Germany) or agarose gels (1 % w/v, 2 mm), and subsequent spraying with
Folin Ciocalteau reagent (Merck, Darmstadt, Germany) and 20% w/v Na2C03
(Dinkelaker et al. 1996). Similarly, other spray reagents commonly used for
detection of different groups of compounds in paper- or thinlayer chromatog-
raphy may be suitable for visualization of rhizosphere compounds (e.g. ninhy-
drin for amino acids, ani1inphtalate for sugars). Moreover, many phenolics
become visible by examination of the carrier media in UV-light (360 nm,
254nm) due to intense autofluorescence (Kape et al. 1992).
13.2.3.5 Detection of Enzyme Activities
Activities of various enzymes, such as acid phosphatase (Bieleski and Johnson

1972; Bartlett and Lewis 1973) peroxidases (Vaughan et al. 1994), invertase, cel-
lobiase, adenosinetriphosphatase, pyrophosphatase and nudease (Chang and
Bandurski 1964) have been detected in root bathing solutions of different plant
species. For measurements of enzyme activities, the solutions are generally
mixed with an appropriate buffer and a natural or artificial substrate of the
enzymatic reaction (Box 13.4). Reaction products are detected by eye (Fig.
13.3F) or spectrophotometrically.
13.2.3.6 Assessment of Nutrient Distribution in the Rhizosphere

by Autoradiograpy
Autoradiography has been used to study the nutrient concentration profile in

soil in the vicinity of plant roots. More details are given in Chapter 12. In prin-
ciple, plants are grown in soillabelled with radio active tracers. After a few days,
X-ray film is placed on the soil around roots, and tracer concentration is esti-
mated from blackening of the film.
Plant Culture. Most frequently, plants are grown in flat rhizoboxes (depth of the
soillayer 0.5 cm) in which the roots grow along the removable front cover. A
thin (e.g. 8 J.lffi) plastic foil or mylar film can be placed between the front cover
and labelled soil, on which the film is placed at the end of the experimental
period (Bhat and Nye 1973; Claassen et al. 1981). As this technique is non-
destructive, several autoradiograms can be made with the same plants during
their development. Ernst et al. (1989) cultured the plants in rhizoboxes with non-
labelled soil. After different growing periods, the front cover was removed and
the soH-root interface covered with a prefixed sheet (2.1 mm thickness) of 0.75%
(w/v) agar containing 32p. After an uptake period of 24h the soil-root interface
induding the agar sheet was covered by X-ray film for autoradiography. The
above mentioned methods of plant culture have been criticized because nutri-
ent flow towards the root takes place only in the direction of the plane but not
radially from aH around the root as would be the case under "natural" condi-
tions (Kraus et al. 1987). To overcome this problem, plants were grown in cylin-
drical pots· containing an additional small cylinder in the centre, which was
filled with labelled soil. The top of the inner cylinder was covered with a nylon
net. A small maize seedling was transplanted to the experimental device with
its primary root forced into a small hole which was bored vertically through the
nylon and the first 2-3 cm of the central cylinder. The rest of the roots were
allowed to grow in the non-Iabelled soil outside the central cylinder. At the end
of the experimental period the soil core from the inner cylinder was separated
and frozen in liquid nitrogen. The whole soil core was cut into disks of about
0.5 cm thickness using a stone cutting machine and these disks were used for
autoradiography (Kraus et al. 1987).
Soil Labelling, Preparation and Evaluation of the Radioautogram. Various

radio active tracers have been used for autoradiography, inc1uding 32p and 33p
(Bhat and Nye 1973; Claassen et al. 1981; Kraus et al. 1987), 86Rb (Claassen
et al. 1981), 33K (Claassen and Jungk 1982), 45Ca (Barber and Ozanne 1970;
Wilkinson et al. 1968a), 99Mo (Lavy and Barber 1964), 65Zn (Wilkinson et al.
1968b) and 35S (Sanders 1971). The tracer has to be thoroughly mixed with the
soil. The initial specific activity needed is dependent on the halflife of the isotope
and the duration of the experiment, and is normally in the range of I-
lO JlCig- 1 soil. For the radioautogram, X-ray film is brought into c10se contact
with the foil covering the soil and exposed to the radiation e.g. for 4 h or 1 day
either at root temperature (Claassen et al. 1981) or in a freezer (Kraus et al. 1987).
For quantitative evaluation of the radioautograms, blackening of the film
is measured densitometrically (Claassen et al. 1981; Kraus et al. 1987). For stan-
dardization of the densiometer readings, radioautograms have to be prepared
from soil samples containing various amounts of tracer. Further details of quan-
titative autoradiography are given e.g. by Passioura (1972).
13.3 Nutrient Uptake
13.3.1 Principles of Nutrient Uptake
Mechanisms for Nutrient Uptake. In the soil, nutrients are transported to the
root surface by massflow and diffusion. Transport of nutrients by these physi-
cal processes is restricted within the apoplast of roots by hydrophobic suberin
deposits in the endodermis, and in some plant species also in the exodermis.
So, even if membrane transport does not occur at the root periphery (root hairs,
epidermis), nutrients stiH have to cross the plasma membranes at the endoder-
mis before they can reach the symplasm of the roots. The plasma membrane is
an effective barrier to diffusion of nutrients and is the site where selection and
metabolic control of nutrient uptake takes place. Across the plasma membrane
between the external solution and the cytoplasm, an electrical potential differ-
ence of 100-200mV (inside negative) exists, which is generated by the activity
of the membrane-bound H+-ATPase, pumping H+ from the cytoplasm into the
apoplasm. The electrochemical potential gradient of H+, generated by this
enzyme, is used for the membrane transport of nutrients via transport proteins.
The electric potential drives ions through channels. The activity of both trans-
porters and ion channels is metabolicalIy controlIed. The direction of ion flow
through an open channel is determined by the electrochemical gradient across
the plasma membrane. Therefore, ion channels are very likely to provide the
mechanism for ion efflux, i.e. release of ions from the cytoplasm to the exter-
nal solution. In principle, one can envisage that net uptake of ions into the cell
may be controlled by both influx and efflux (Deane-Drummond 1990; White et
al. 1992; Lee 1993).
Transporters. At low external concentrations (below 1 mM) the membrane

transport of nutrients is mediated mainly by specific high affinity transport pro-
teins (Glass et al. 1992; Maathuis and Sanders 1996; Miller and Smith 1996), which
show Michaelis-Menten kinetics, i.e. saturation at high external concentrations.
At high external concentrations a second set of quite distinct low affinity trans-
porters come to dominate the overall absorption of ions. Uptake by low affinity
transporters often linearly increases with increasing external nutrient concen-
trations. Thus, depending on the external nutrient concentration in an experi-
mental set up, the activity of different uptake systems may be assessed. In soil,
the concentration of most nutrients at the root surface is expected to be low, so
that membrane transport is mainly mediated by high affinity uptake systems.
At low nutrient concentrations in the external solution the outermost cell
layers of the roots (rhizodermis) absorb most of the nutrients delivered by dif-
fusion in the soil solution (Rufty et al. 1986; Clarkson 1996). At high external
concentrations a nutrient-rich solution may enter the apoplast of the root
cortex, and alI celIs of the cortex may contribute to nutrient uptake. Thus,
depending on the nutrient concentration in an experimental set up, different
tissues or locations within the roots might be assessed for their ability for nutri-
ent uptake.
Metabolic Control. Nutrient transport across plasma membranes of root celIs

may be modified in the short-term, i.e. within seconds to minutes, by a change
in the transmembrane potential which is established by the membrane-bound
H+ ATPase. Rapid changes of the root environmental conditions, e.g. in tem-
perature (Minorsky and Spanswick 1989), or mechanical disturbance of the
roots occuring even when roots are gently transferred from one nutrient solu-
tion to another (Miller 1981; Bloom and Sukrapanna 1990) may lead to a sudden
decrease in nutrient uptake which is presumably caused by a de crease in the
transmembrane potential. If the roots are not seriously injured nutrient uptake
recovers within minutes.
In the long-term (hours to days), nutrient uptake may also be regulated by
the concentrations of mineral nutrients and their metabolites in the roots, or
by various shoot signals translocated to the roots in the phloem, including
sugars, malate, nutrients and their metabolites (Marschner 1995). Excision of
roots stops transfer of theses signals. Therefore, measurements on nutrient

uptake in excised roots should be completed before interruption of signalling
from the shoot becomes effective, i.e. within minutes to a few hours.
Parameters for Estimating the Ability of Roots ta Take Up Nutrients. The ability
of roots for nutrient uptake and/or translocation to the shoot can be estimated
by various parameters which assess different root characteristics. The actual
activity of the transport systems mediating membrane transport of ions from
the external solution into the cells may be estimated by measurement of ion
influx, i.e. the unidirectional flux of ions from the external solution into the cells.
Influx can only be measured under laboratory conditions (see Sect. 13.3.6),
but the roots used for influx studies can be sampled from the field (Chapin
et al. 1986).
The ability of roots to absorb nutrients from the external solution (soil solu-
tion or nutrient solution) may be estimated by measurement of net uptake (net
influx) of ions into roots. Net uptake of an ion is the result of mechanisms medi-
ating influx and efflux.1t can be determined for roots from field-grown plants in
situ (see Sect. 13.3.2) or under laboratory conditions (see Sect. 13.3.3).
lf ion influx is mediated by saturable transport mechanisms, the relation-
ship between influx and extern al concentration obeys Michaelis-Menten
kinetics and may be characterized by lmax (maximum influx at external ion
concentrations saturating the activity of the transport system) and Krn
(Michaelis-Menten constant, external concentration where influx = Imax12). If the
relationship between net uptake and externa! concentration obeys Michaelis-
Menten kinetics, the potential ability of roots for net uptake can also be char-
acterized by lmm Krn, and Cmim where Cmiu is the minimum concentration to which
roots may deplete the external solution, i.e. the concentrat ion where influx =
efflux. The kinetic parameters for influx and net uptake can only be determined
under laboratory conditions (see Sect. 13.3.6).
The ability of roots to translocate nutrients to the shoot may be assessed
from measurement of net accumulation rates of ions in the shoot (net translo-
cation rate), or from ion fluxes in the exudate of decapitated plants (gross
translocation rate). Both net translocation rates and xylem fluxes can be
assessed in plants grown in nutrient solution or soil (see Sect. 13.3.4).
13.3.2 Nutrient Uptake into Intact Plants
13.3.2.1 Nutrient Accumulation in Plants Under In Situ Conditions

(Williams Formula)
Nutrient accumulation in plants may be determined from plant biomass incre-

ment between sequential harvests, and nutrient concentrations in the biomass
(see Sect.13.3.7 for methods to measure nutrient concentrations in plant tissue).

Depending on the root parameters which are measured, nutrient accumulation
between two harvest dates can be based e.g. on root biomass, root length, or
root surface area. According to Williams (1948) net uptake rates (U) can be cal-
culated with the following formula:
U = (W2 - ~)/(t2 -tl)x ln(R2/RI )/(R2 -RI ),
whereby W I and W2 are the nutrient contents (biomass x concentration) of the

plants at times (t) 1 and 2, and RI and R2 are the root parameters (e.g. root
biomass, length) measured at ti and t2 • For the above formula it is assumed that
root growth is exponential between the two harvest dates. For the assumption
of linear root growth the formula is modified to
U =(W2 - ~)/(t2 -tl )/(R2 +RI )/2.
U is a parameter to estimate mean uptake rates of the roots between the har-
vests. The time interval between sequential harvests (t2 - ti) may vary from a
few days (Engels and Marschner 1992) to several months (Headleyet al. 1985).
The lower limits of the time interval are given by the need for a significant
increment in either biomass or nutrient concentratiop. between two harvest
dates. With long intervals between harvests, U, as calculated by the Williams
formula, is only a rough parameter for estimating root activity, and may be
falsified by (1) biomass losses e.g. through herbivores or leaf falI, (2) nutrient
losses e.g. by leaching or retranslocation to the roots and subsequent release
into the rhizosphere, and (3) nutrient uptake or release via aerial parts of
the plant, which may be considerable for nitrogen and sulfur. Furthermore,
it is obvious that increasing the time interval between harvests complicates
the assessment of the mean root biomass that has contributed to the
observed nutrient uptake, because root growth under field conditions is neither
linear nor exponential but is strongly dependent on plant age and environ-
mental conditions in the shoot and root zone. If uptake rates are based not on
root biomass but on root length or surface area, additional sources of error may
arise because specific root length and root diameter may also vary depending
on plant age and soil conditions (e.g. temperature, mechanical impedance,
moisture).
In the field it is difficult to obtain data for root biomass. Furthermore, con-
siderable nutrient losses from roots may occur during washing out from soil
(Grzebisz et al. 1989). Therefore, often only nutrient accumulation rates in the
shoots are considered in the Williams formula, i.e. net translocation rates from
the roots to the shoot. In large perennial plant species like trees, it is not easy
to gain data for the nutrient accumulation in the above-ground organs of the
plant. From data on nutrient accumulation rates in individual plant parts, e.g.
young growing leaves, conclusions on root activity can be drawn only if tracers
are applied to the soil, because increment of nutrient content in the organs may
be the result of both uptake by roots or remobilization from other plant organs,
e.g. roots or stems.
The importance of adequate experimental design for the statistical analy-
sis of data was emphasized byChapin and van Cleve (1989). For example,a com-
parison of the effect of two treatments on nutrient uptake requires that a block
containing sufficient individuals of both treatments for aH harvest dates is
established. This block constitutes an individual replicate and should be repli-
cated sufficiently to provide the necessary sample size. At each harvest date a
plot is randomly selected from within each block, and individuals of the two
treatments are harvested from that plot. At the next harvest date another plot
is randomly selected from each replicate block etc. An example of an experi-
mental design for four individuals per treatment, tree harvest dates and five
individual replicates is given in Fig. 13.4.
In conclusion, growth analysis and calculation of the net uptake rate by
the Williams formula estimate the in situ root activity of intact plants growing
under specific environmental conditions, as an average of the whole root
system and the time interval between sequential harvest dates. The main
advantage of this method is that there is no need for any manipulations of
the plants or soil surrounding the roots during the uptake study, thus exclud-
ing the risk of measuring artefacts. Disadvantages of the method include
the poor resolution of nutrient uptake in time and space and, at least for field-
grown plants, the difficulty of estimat ing the root size on which net uptake
is to be based. Furthermore, from nutrient accumulation rates in field-
grown plants it is not possible to differentiate whether observed variations in
nutrient uptake between treatments are due to differences in the ability of the
roots to take up nutrients or to modifications of nutrient availability in the
rhizosphere.
Replicate 1 Replicate2 Replicate 3 Replicate4 Replicate 5
Harvest 1
Fig. 13.4. Experimental design for comparing net uptake rates of treatments 1 and 2 during
two time intervals, using the Williams formula
13.3.2.2 Methods to Assess the Variation of Nutrient Uptake

Within an Intact Root System
Nutrient uptake within a root system can vary depending on e.g. root age
(Marschner and Richter 1973; Ernst et al. 1989) or local nutrient availability
(Drew and Saker 1975; van Vuuren et al. 1996). In principle, differences between
various root regions in their ability to take up nutrients can be assessed using
excised roots (see Sect. 13.3.3) or intact root systems. With intact root systems,
several methods have been used which may be differentiated according to
whether nutrients are supplied to the whole root system or only to specific root
regions, whether mineral nutrients or tracers are supplied, whether nutrient or
tracer depletion is measured in the root medium or whether tracer accumula-
tion is in whole plants or root segments (Fig. 13.5). With intact root systems,
nutrient uptake into various root regions is most commonly assessed by local-
izing nutrient supply to specific root zones, e.g. by injection of tracers into
various soil zones, by applications of nutrient -containing agar on specific root
zones, or by forcing the roots into compartmented containers.
Injection of Tracers into Various Soil Zones. The variation in time and space of
the relative activity of roots for nutrient uptake can be assessed by injecting
tracers into the soil at various dates and depths or distances from the plant, and
subsequent measurement of tracer depletion from soil or accumulation within
the plant. For tracer injection, holes are usually drilled into the soil with an
auger. The application of the tracer (solid or solution) is made through pipes
which are inserted into the holes. After tracer applications, the holes are refilled
carefully with soil. Further details of tracer application are described by
B6hm (1979).
Tracer injection Local application Compartmented Local measurement Separation of root

into the soil via agar root containers of ion activities at regions alter short-
the root surface with term uptake of
microelectrodes tracers
Measurement Measurement Measurement Measurement

of tracer con- of tracer con- of tracer con- of tracer con-
tent in the tent in the tent in the tent in the
the plant the plant or the plant or root segment
nutrientJtracer nutrientJtracer
content in the content in the
agar compartment
Fig. 13.5. Methods to assess the variation of nutrient uptake within an intact root system
Various tracers have been used, including rare chemical tracers such as Li,
Sr, Rb and Br (Fitter 1986; Benjamin et al. 1996), stable isotopes e.g. lsN (McKane
and Grigal 1990), or radioisotopes e.g. 32p (Kurien et al. 1992). As mineral
nutrients are absorbed by uptake systems which are specific to each nutrient,
studies with rare chemical tracers may give an indication of the distribution of
living roots rather than of the ability of the roots for uptake of essential min-
erals. Even with isotopes, the quantitative assessment of the ability of specific
root regions for nutrient uptake is tied to several prerequisites: (1) supplyof
the tracer to the uptake sites is not limited, (2) supply of the tracer is confined
to the specific root region, and (3) the specific activity of the tracer at the
uptake sites is uniform in the various root regions which are tested (Pearson
1974). To meet these requirements, the amount of applied tracers should be suf-
ficiently high (possibly a slow-release tracer source), the uptake period should
be sufficiently short, the tracer should be biologically, chemically and physically
inert under the specific soil conditions (e.g. potassium in a non-fixing soil),
and the nutrient/tracer application should at least transiently equilibrate dif-
ferences between various root regions in the rhizospheric concentrations of
native nutrients (resulting e.g. from differences between root regions in root
activity). From this listing, which is certainly not complete, it becomes obvious
that tracer injection techniques under field conditions may at best lead to semi-
quantitative answers concerning the ability of different root regions to take up
nutrients.
Local Application via Nutrient-Containing Agar. If the roots are readily acces-
sible, e.g. roots growing along a removable plexiglas cover either in rhizoboxes
or in the field (see Chap. 8), nutrients or tracers can be applied carefully directly
to specific root segments via agar. The uptake ability of the root segment may
be assessed by measuring tracer uptake into the plant (Ernst et al. 1989; Brady
et al. 1993) or nutrient depletion from the agar (Reidenbach and Horst 1995;
Box 13.5). Generally, measuring nutrient depletion from the medium sur-
rounding specific regions of the root system allows the assessment of the uptake
ability of different root regions within one plant, and measurements can be
repeated in the same plant or even root segment e.g. in course of its develop-
ment. Furthermore, nutrient concentration in the rooting medium can be a
parameter which responds much more sensitively to nutrient uptake than the
nutrient concentration in the plant, particularly if nutrients are supplied to
small root segments in large plants, but the method is indirect. The change in
nutrient concentration in the rooting medium may result not only from uptake
by roots but also from physical (gaseous losses, leaching, desorption, adsorp-
tion), chemical (precipitation) and biological processes (mineralization, immo-
bilization) in the medium. To assess the importance of the latter processes,
which should be of minor importance in nutrient-containing agar, the change
in nutrient concentration in the rooting medium should be tested under the

experimental conditions in the absence of roots.
Compartmented Containers. The ability of different root zones to take up nutri-

ents may be assessed by placing roots in compartmented containers containing
soil or nutrient solution. Various substances have been used to seal the root seg-
ments in the compartments, including vaseline (Moritsugu et al. 1993), silicone
grease (Siebrecht et al. 1995) and silicon putty (Hăussling et al. 1988). Uptake
of different root segments can be estimated by measuring the nutrient deple-
tion from the individual compartments (Hăussling et al. 1988), or by supply of
tracers to one compartment and measuring tracer accumulation in the plant
(Drew and Saker 1986).
Various experimental sets have been used (Fig. 13.6) which differ in tech-
nical expenditure and thus suitability for field measurements. Marschner et al.
(1991) investigated the gradients in nutrient uptake ability along individual
roots of spruce in a 60-year-old stand. The roots were gently uncovered from
the upper soil horizon and placed into a simple compartmented container with
nutrient solution (Fig. 13.6A), in which ion uptake rates from the individual
compartments were determined over a 3-day-period from the decrease in solu-
tion volume and ion concentration. As a control, solution volume and ion con-
centration were also measured in a compartment containing no roots. This
non-destructive field technique is suitable even for large trees and measure-
ments can be repeated within the same plant or root segment. Under labora-
tory conditions, root segments can be placed in plastic tubes or compartments
in which the root segment is supplied with a continuous flow of nutrient solu-
tion during the uptake period, allowing the measurement of tracer accumula-
tion in the plant at constant external concentrations. In the experimental set up
shown in Fig. 13.6B, the two compartments adjacent to the root segment under
study were supplied with tracer-free solution to flush away tracer ions diffus-
ing through the cortex and thus to prevent contamination of the outer solution
with tracers. The detail with which variation of uptake along the root axis or
with increasing root age can be studied depends on the length of the segment
used for labeling. Harrison-Murray and Clarkson (1973) used smalilabeling
chambers (3 mm diameter) which could be fitted into the gaps between the
lateral roots. Thus, nutrient uptake of the main root could be studied
with increasing root age without the confounding effect of young laterals
(Clarkson 1996).
Separation of Root Regions After Short Periods of Uptake. The variation of

nutrient uptake within an intact root system may also be assessed after short
exposure of the entire root system to labelled nutrients followed by rapid sepa-
ration of the various root regions (Canning and Cramer 1958; Yoneyama et al.
Fig. 13.6. Examples for

compartmented containers;
for further details see text.
A Simple compartmented
container used for field
studies (Hăussling et al.
1988, reprinted with
permission from Gustav
A
Fischer Verlag). BApparatus
for measurement of tracer
uptake under controlled
environmental conditions.
(Drew and Saker 1986; with
permission of Oxford
University Press)
De ta il of t reatment zone
1975; Lazof et al. 1992; Box 13.5). Exposure to labelled solutions must be suffi-
ciently short to minimize translocation among root regions. Results from short-
term time courses of uptake by intact roots indicate that radial ion transport
across the root to the xylem can occur within a few minutes (Siddiqi et al. 1991;
Lazof et al. 1992), and considerable interregional transport is likely to occur
with exposure to labelled nutrients for 30 min or more. In view of the brief expo-
sure of the roots to labelled nutrients (tracers), the ratio of apoplastic to sym-
plastic tracers is high. To remove tracers from the apoplast before analysis,
careful rinsing of the roots is necessary, e.g. with a solution containing non-
labelled nutrients (see Sect. 13.3.3).
Measurement of Ion Activities at the Root Surface with Microelectrodes. The

variation of nutrient uptake along the root axes can also be assessed by mea-
BOX 13.5. Experimental Protocols for Measurement

of Nutrient Uptake Along the Root
Measurement of Nutrient Depletion from the External Medium
Using the agar technique, Reidenbach and Horst (1995) measured the
ability of different root regions for N0 3- uptake in maize growing in
large rhizoboxes up to maturity. To minimize the disturbance of roots
directly before the uptake measurement, non-branched apical regions of
roots growing along the removable plexiglass front cover were underlaid
with gauze (4 x 4cm, 40 Jlm pore width) which allowed nutrient and
water transfer from the soi! to the root. One day before measurement, the
gauze was exchanged with waterproof plastic. The next day a small block
of agarose (0.8% weight/volume, 2 x 1 x 06cm, . 1 mM KN0 3, 1 mM
CaS0 4 , pH 5) was put on the underlaid root segments and covered with
plastic to prevent desiccation.After 8h, the N03- concentration in the
agar blocks was determined by dissolving the blocks for 10 min in water
at 90°C, followed by centrifugation and measurement of N0 3-
concentrat ion in the filtered supernatant with an autoanalyser. The root
segments were photographed before agar application and after removal,
and the mean root length during the uptake period was estimated from
the photographs.
Measurement of Short-term Accumulation Within Root Segments
of Intact Plants
Lazof et al. (1992) examined the transport characteristics of six different
root regions by measuring the change in accumulation rates during short
time courses «15min) of 15N03- supply to the whole root system. The
roots were harvested and separated into regions after 1,3,6 or 15min of
15N supply. The rate of 15N accumulation dur ing the first minute was
taken as an estimate of unidirectional influx. The 15-min accumulation
rate was taken as the steady state accumulation rate which is the result of
both influx and efflux from the external solution, and translocation to or
import from other root regions. Within 15 min, accumulation rates of 15N
increased in the apical root regions indicating import from other root
regions, and decreased in basal root regions indicating translocation to
root tip and shoot. The accumulation rates of 15N in the whole plant were
constant during the 15 min of 15N supply indicating that efflux was
negligible. The results on translocation to and import from other root
regions were confirmed with pulse chase experiments where the change
in 15N concentrations in different root regions were measured after 15N
supply for 6 min followed by a 12 min chase period with a supply of non-
labelled N0 3-.
suring ion activities in the unstirred layer immediately external to the root
surface using ion-selective electrodes (Newman et al. 1987; Henriksen et al.
1992). From the measurements of ion activity at several positions in the
diffusion boundary layer net fluxes of ions into the root can be estimated
(Henriksen et al. 1990, 1992). With this method ion fluxes can be estimated with
high resolution in time and space, and with minimal disturbance of the root
system. But, the method is only suitable for plants growing in nutrient solution.
13.3.3 Nutrient Uptake into Excised Roots
Principles. The ability of roots to take up nutrients may be estimated from iso-
lated root segments by measuring either nutrient depletion from solutions
(Cruz et al. 1995), or nutrient accumulation in the roots (Hogberg et al. 1995).
As uptake studies with isolated roots have to be confined to a short time dura-
tion (see below), the roots are usually supplied with tracers to increase the sen-
sitivity of measurements. For uptake studies roots may be sampled from plants
growing in the field (Jackson et al. 1990), or in pots with soil (Matzner and
Richards 1996) or nutrient solution (Cruz et al. 1995).
Methodology. In principle, the method involves the following steps: (1) adap-
tation of the roots to the temperature and pH at which uptake will be measured,
(2) transfer to the uptake solution containing the tracer, (3) termination of
uptake and removal of tracer from the root surface and free space of the cortex
by rinsing the roots in cold (e.g. 4°C) washing solution, and (4) measurement
of tracer content in the roots, or in the uptake and washing solutions. To facil-
itate the transfer of roots from one solution to another, the roots may be
enclosed in "tea bags" (Epstein et al. 1963) or "swimfeeders" (Harrison et al.
1978). Instead of transferring the roots from one solution to another, the solu-
tions may be exchanged e.g. via vacuum withdrawal (Fox et al. 1996). For steps
(1) to (3), the solutions must be aerated, and it is recommended to add Ca (final
concentration about 0.2 to 0.5 mM) to maintain membrane integrity. The pH
may be buffered during the uptake period, e.g. with 10 mM MES, pH 5.5 (Leon
et al. 1995) or 5mM Mes-Tris, pH 5.0 (Fox et al. 1996). The washing solutions
(step 3) may contain the non-labelled form ofthe nutrient being studied at con-
centrations much higher (10-100 fold) than that in the uptake solutions. Alter-
natively, the washing solutions may contain CaC12 for the exchange of ions.
Cations like Cu which are tightly bound to exchange sites in the free space may
be exchanged with Pb(N03)2 (Harrison et al. 1979). The adequacy of the des-
orption regime must be checked by measuring desorption (tracer content of the
washing solution) in relation to the duration or frequency of rinsing. In chill-
ing sensitive plants like cucumber or maize, a rapid decrease in solution tem-
perature during transfer of the roots from the uptake to the washing solution
may not only stop nutrient uptake into the root cells but cause substantial efflux
from the cells into the washing solution (Minorsky and Spanswick 1989). There-
fore, rapid changes in root temperature during washing should be avoided. To
test the nutrient loss from the symplast to the solution, nutrient efflux during
the washing procedure can be measured in roots from plants which were fed
with labelled nutrients during preculture (Lazof et al. 1992).
Root Characteristics Assessed. With excised roots, their ability to take up nutri-
ents may be assessed under the specific experimental conditions provided in
the uptake solution (e.g. pH, temperature, nutrient concentration). Because the
experimental conditions in the uptake solution and the length of the uptake
period can be exacdy regulated, the kinetics of net uptake or influx (see Sect.
13.3.6) may also be measured (Chapin et al. 1986). If root samples are taken
from specific regions along the root axes, uptake characteristics can be related
to root age or distance from the root tip (Cruz et al. 1995). The method has also
been used to assess the uptake ability of roots subjected to drought (Matzner
and Richards 1996) or a non-homogenous supply of nutrients in the field
(Jackson et al. 1990).
Problems. Measuring nutrient uptake in excised roots suffers from two severe
problems: (1) root injuries occuring during handling of the roots and (2) phys-
iological changes associated with isolation of roots or root segments from the
rest of the plant. Both factors impair nutrient uptake. Therefore, this method is
appropriate for comparing relative differences between treatments in the ability
of roots to take up nutrients rather than absolute values or actual in situ uptake
rates. The extent of root injuries, e.g. loss of fine lateral roots and root hairs or
damage of root epidermal cells which are of particular importance for nutrient
uptake from soil (Marschner 1995), varies with the experimental system. It may
be expected that injuries are more severe in root samples washed from soH cores
taken from the field than from plants grown in nutrient solution. But, even
gently cutting the roots or transferring roots from one solution to another may
decrease nutrient uptake (Bloom and Sukrapanna 1990), presumably because
of a transient decrease in the transmembrane electrochemical potential gradi-
ent (see Sect. 13.3.1). Pretreatment of the roots e.g. for 30 min in a solution con-
taining Ca (see above, step 1) may lead to a recovery of the roots from this
transient decrease in uptake ability. The effect of root injuries on nutrient
uptake can be roughly estimated from comparison of the measured data on
uptake with data in the literature obtained from intact plants.
Isolation of roots from the rest of the plant may de crease nutrient uptake
within hours because of the interruption of both import of sugars and other
"shoot signals" via the phloem, and export of nutrients via the xylem (see Sect.
13.3.1). The effect of these factors on nutrient uptake increases with time from
excision, and is dependent on the specific nutrient (Bloom and CaldwellI988).
In order to find out the period after which root isolation leads to a marked
depression of uptake, it is therefore necessary to make a time study of nutrient
uptake with increasing time from excision.
13.3.4 Collection of Xylem Exudate
Prin cip le. Nutrients are translocated from the roots to the shoot in the xylem
by massflow with water. In transpiring plants, the massflow in the xylem vessels
is driven by the water loss from the plants to the atmosphere, which occurs
mainly via the open stomata. Under conditions of low or no transpiration, e.g.
during the night or before leaf emergence in spring, or in decapitated plants,
the massflow is driven by"root pressure". The permeability of plant membranes
to water is much higher than to ions. Therefore, roots behave as osmometers,
and "root pressure" may develop as a consequence of metabolically driven
nutrient uptake into the roots and rele ase into the xylem vessels, creating a dif-
ference in the osmotic potential between the extern al solution and the xylem
sap (Marschner 1995). The amount of nutrients exuded with the xylem sap from
detopped plants or individual roots can be taken as parameter to estimate the
actual delivery of nutrients from the roots to the shoot. For the estimation of
the amount of nutrients exuded, the nutrient concentrat ion in the xylem sap,
and the amount of xylem sap which is exuded per unit of time have to be
measured.
Collection of Xylem Sap. In field-grown plants bleeding sap has been obtained
from the stern cut just above the ground (Engels et al. 1994), or from individ-
ual nodal roots (Canny and McCully 1988) and laterals (Jeschke and Pate 1995)
after care fuI uncovering of their distal end from the soil while the rest of the
roots were still undisturbed in the soil. To collect the xylem sap a sloping cut
can be made and the sap running down to the lower edge of the cut sampled
with a micropipette (Canny and McCully 1988). For quantitative analysis of
xylem sap, a sleeve of silicone rubber tubing is fitted over the cut surface and
sealed at the base e.g. with silicone grease. To reduce bacterial growth during
exudation, the cut surface can be cleaned with a paper towel and moistened
with a drop of Micropur.
The duration of the collection period should be as short as possible to
collect enough xylem sap for nutrient analysis, because translocation of nutri-
ents from the roots to the shoot is increasingly affected by interruption of
phloem supply from the shoot with carbohydrates and other "shoot signals" (see
Sect. 13.3.1). Exudate sampled during the first 5-lOmin after decapitation is
often discarded for technical reasons (time needed to fit and seal the rubber
tubing), and also because it may contain abnormally high concentrations of
nutrients and hormones due to mechanical stress on the stern tissue during
cutting or contaminat ion with the contents of severed cells from the cut surface
(Else et al. 1994).
The instantaneous content of the xylem vessels of field-grown roots or root
segments may be sampled by applying a mild vacuum or pressure to excavated
roots (Canny and McCully 1988; Jeschke and Pate 1995). To protect the root seg-
ments from drying they should be wrapped e.g. in a damp paper towel. In maize
nodal roots to which reduced pressure was applied with a syringe, each 1 cm of
bare root yielded about 1 mg of xylem sap (Canny and McCully 1988).
In plants growing in pots or nutrient solution, root xylem sap may be
obtained by applying (1) a positive hydrostatic pressure to the whole root
system (Else et al. 1994) or (2) a negative hydrostatic pressure (suction) to the
cut surface to simulate transpiration (Gil de Carrasco et al. 1994). To pressur-
ize root systems (method 1) the culture vessels includ ing the roots are placed
in Scholander-bomb-like vessels in which the pressure can be raised with com-
pressed air (see Fig. 13.7). By varying the hydrostatic pressure, the sap flow rates
can be adjusted to those measured in intact plants. Aiso roots (including the
pot) of intact plants can be placed in pressure chambers allowing collection of
xylem sap exuding from the cut leaf tips (Passioura and Munns 1984; Jeschke
et al. 1996).
Assessment. Collection and analysis of xylem exudate allows an estimation of

the actual root activity under field conditions. Compared with the method using
the Williams formula (see Sect. 13.3.2.1) the resolution in time of the estimate
for root activity is substantially improved. The disadvantages of the method are
similar to those for the Williams formula, namely difficulties of estimating the
. O' ring seal Colleelion lubing

Pressure inlet
Fig. 13.7. Diagram of vessel

used to pressurize root
WingnUI ___
systems to obtain xylem sap.
To avoid damage to the . O' ring seal --~~~~~~l~~~~~'- Slainl ess
sleellid
20 mm hypocotyl stump, Stainless sleel - - -rHlr-- Aubber sleeve

sealing disc
pressure sealing was made Hypoeotyl stump
with an "O" ring between the Plant pot and
Clamping rod
lid and a sealing disc rather reot system in
eompost
than against the hypocotyl
Aerylie side wall
itself (Else et al. 1994; with
permis sion of Oxford
Sta inless __ s~~~~~~~~~~
University Press) sleel base
root size in the field, and problems in differentiating between root characteris-
tics associated directly with nutrient uptake and translocation to the shoot and
those modifying the nutrient availability in the rhizosphere. Furthermore,
several fac tors have to be considered in the interpretation of xylem exudate
analysis. Xylem sap contains shoot-derived nutrients cycled from the shoot to
the roots in the phloem and reloaded to the xylem again in the roots (Marschner
1995). Translocation of solutes in the exudate of decapitated plants may sub-
stantially differ from that of intact plants because of not only the interruption of
the supply of "shoot signals" (see Sect. 13.3.1) but also the differences in water
flux rates. However, in contrast to the translocation rates of hormones, those of
most nutrients seem to be not very sensitive to the rate of water flux in the xylem
either in intact (Tanner and Beevers 1990) or decapitated plants (Else et al. 1995).
13.3.5 Contribution of Mycorrhiza to Nutrient Acquisition
General. Mycorrhizas are symbiotic associations between plant roots and soil
fungi. In annual plants and also in fruit trees, the most common type of asso-
ciation is formed by vesicular-arbuscular mycorrhizal (VAM) fungi (also called
arbuscular mycorrhizal [AM] fungi by some authors) of the order Glomales
(Zygomycotina). Roots colonized by VAM fungi are characterized by intern al
fungal structures (hyphae, and often arbuscules and vesicles) and by external
hyphae growing away from the root (Fig. 13.8). Spores are formed in the soil on
Roo!
hair
Fig. 13.8. Schematic

presentation of internal and
extern al structures of root
colonisation with a vesicular-
arbuscular mycorrhizal fungus
the extern al mycelium. The external hyphae are well suited to absorb mineral
elements from root-distant soil and transport them to the plant.
Vesicular arbuscular mycorrhizal fungi occur in most terrestrial ecosys-
tems, in agricultural fields, and can colonize most crop and pasture plants. Myc-
orrhizal fungi have probably been associated with roots since the occurrence
of vascular land plants more than 300 million years ago (Taylor et al. 1995;
Simon 1996) and only a few plant species (for example, lupins or some Brassica
species) have developed a resistance against mycorrhizal colonization. In addi-
tion to a contribution to plant nutrient uptake, mycorrhizal fungi also modify
the carbon economy, the water use, and the pathogen susceptibility of plants.
In particular, non-nutritional effects of mycorrhizal fungi in disease resistance
have recent1y received much interest (St-Arnaud et al. 1997). A summary of the
current view on the role of mycorrhizas in plant growth is given by Smith and
Read (1997) and Marschner (1995).
In this chapter, only a small part of the methodology used in mycorrhiza
research is described. For a more detailed description, the reader is referred
to, for example Schenck (1982), Sieverding (1991), Norris et al. (1992), and
Brundrett et al. (1994).
Methods for Determining Mycorrhizal Colonization. It is not possible to see root

colonization with VAM fungi without staining and microscopic observation.
However, the staining process does not require expensive instrumentation. For
assessment of mycorrhizal colonization, roots must be separated from soil and
the tissue cleared with 10% potassium hydroxide. After washing and acidifica-
tion of the root sample, fungal structures in the roots are stained by Trypan
blue, acid fuchsin, chlorazole black or other compounds (Brundrett et al. 1984;
Grace and Stribley 1991). A staining of active fungal parts only is also possible
(Schaffer and Peterson 1993). After staining, in most cases the percentage of the
total root length colonised by mycorrhizal fungi is determined (Giovanetti and
Mosse 1980).
Spores of mycorrhizal fungi can be isolated from soil by a wet sieving and
decanting technique (Gerdemann and Nicolson 1963). AIso, the length of the
extern al mycorrhizal mycelium in the soil can be determined after washing the
hyphae from soil and subsequent staining (Hamel et al. 1990). In a recent devel-
opment, fatty acid profiles are used to determine mycorrhizal biomass in the
soil and within roots (Olsson et al. 1997). In the future, DNA-based methods
will be used to characterise the distribution of different fungal populations in
the soil and in the roots.
Methods for Determining the Effect of Mycorrhizal Pungi on Nutrient Uptake.

Because of the ubiquity of mycorrhizal colonisation in most plant species,
usually non-mycorrhizal individuals of these species cannot be found in fields
or natural environments. The colonisation of roots by VAM fungi can be

reduced by use of pesticides, but side-effects of these pesticides on other micro-
organisms and on nutrient availability in soil have to be accounted for (Weber
et al. 1993).
Most of the current knowledge on the contribution of mycorrhiza to plant
nutrient uptake and growth is derived from pot experiments with partially ster-
ilised (mycorrhiza-free) soils, soil mixtures or other growing substrates. Heating
a soil to 70-80°C for one day in most cases eradicates all mycorrhizal propag-
ules. In appropriate treatments, mycorrhizal colonization of roots is then re-
established by using fungal inoculum. Such mycorrhizal inoculum can be
produced by isolating spores from soil and subsequent multiplication in a plant
pot culture, but is also commercially available. In the experiments, nutrient
uptake, water use, biomass production and other properties of mycorrhizal
plants are compared with those of non-mycorrhizal plants, and the difference
is attributed to mycorrhizal colonization.
Up to now, it has not been possible to grow mycorrhizal fungi on nutrient
media without a host plant. Thus, little is known about the physiological
requirements of different VAM fungal isolates. The use of in vitro cultures of
mycorrhizal fungi with Ri T-DNA-transformed plant roots (Chabot et al. 1992)
will be a means of investigating functions of mycorrhizal hyphae and roots
independent of shoot influences.
To specifically study the extraradical mycelium, nets with a fine mesh size
of 10-30 J.1m (required mesh size depends on minimum diameter of roots of
plant species grown in the experiment) can be used to spatially separate root-
and hyphae-growing zones in soil (George et al. 1995). Such meshes can be
obtained from commercial suppliers of bolting cloth. The method requires the
construction of pots where a net can be used to separate the growth substrate.
With even less investment, nutrient depletion in root-distant soil can be studied
in conventional pot experiments with the help of a soil tube accessible to hyphae
only (Fig. 13.9).
When soil is sieved and stored, the indigenous mycorrhizal flora may
become inactive. AIso, mycorrhizal fungi rarely become established on nutrient
solutions, except when the special requirements of the symbiosis are taken into
account (Hawkins and George 1997). Thus, up to now many studies in root
physiology are inadvertently carried out with non-mycorrhizal root systems,
although field-grown roots are usually colonized by mycorrhizal fungi.
13.3.6 Measurement of Influx, Efflux, and Uptake Kinetics
Estimation of Influx. Net uptake of mineral nutrients into the roots is the result
of simultaneous influx i.e. nutrient flux from the external solution into the cells,
Hyphal tube - Extractable P in soil of

Potwith 5011 accessible to hyphal tube after harvest
plants hyphae only
30
~ 20
Ol
E
10
o '-'-------1----'="--
Freshly -AM +AM
Tube closed with ferti-
lised
20-J.lm nylon net soil
Fig. 13.9. Concentration of extractable phosphorus in the soi! of a hyphal tube at the onset of
the experiment (freshly fertilized soi!) and after 40 days of growth of non-mycorrhizal (-AM)
and mycorrhizal (+AM) ryegrass plants (right side). Uptake of phosphorus from the soi! tube
by mycorrhizal hyphae is indicated by depletion of extractable phosphorus in the soi! enclosed
bya 20,um nylon net (lefi side; E. George et al. unpubl.)
and efflux i.e. nutrient flux from the cells into the external solution. For sepa-
rate estimation of influx, accumulation of labelled nutrients (tracers) in the
plants or depletion from the external solution may be measured in time dura-
tions short enough to exclude substantial efflux of labelled nutrients into the
external solution. Because efflux of labelled nutrients from the cytoplasmic
pool of the roots to the external solution willlead to a time-dependent under-
estimation of influx, the measurements have to be done well within the time it
takes to fill the cytoplasmic pools (Clarkson et al. 1996). Typical periods for the
measurement of tracer accumulation in plant tissue which take into consider-
ation the probable turnover times of cytoplasmic pools are 5 min for 4SCa2+
(White et al. 1992),5-15 min for lSN-nitrate and -ammonium (Muller et al. 1995;
Clarkson et al. 1996), and 10 to 25 min for 32p-phosphate, 33P-phosphate, 86Rb,
and 3sS-sulfate (Elliott et al. 1984; Siddiqi and Glass 1987; Clarkson et al. 1989;
Lee 1993).
The lower time limits of influx studies are given by the sensitivity of mea-
surements of tracer accumulation in the plant tissue or tracer depletion from
the extern al solution. This is no problem with radioactive tracers which can be
measured very sensitively against a very low background in the plant tissue, but
possibly is with lsN, particularly if the kinetic parameters of influx are to be
determined, because this requires measurement of influx from low extern al
concentrations (see below). The discrimination of lsN absorbed from the uptake
solution from that already in the tissue should be improved if the natural "back-
ground" is reduced by supply of lsN-depleted N sources during preculture.
However, it has been suggested that the trade-off for improved discrimination
would be decreased precision, because the lsN "background" of the lsN -depleted
plants may become more variable. This would result from differences in the con-
tribution of N sources with natural lsN abundance to the N supply of the plant,
such as seed N reserves, NH 3or NO x absorbed via the stomata, or symbiotically
fixed N2 (Clarkson et al. 1996). For laboratories sufficiently close to a cyclotron,
the radio active isotope 13N provides a sensitive tracer for investigating fluxes of
nit ro gen in plant material over short periods (Lee and Drew 1986; Kronzucker
et al. 1995; Clarkson et al. 1996; see also Chap. 12).
It is obvious that influx studies over short, well-defined periods of time can
only be done with roots to which the tracers are supplied in a nutrient solution.
The procedure for removal of tracers from the apoplast at the end of the influx
study is described in Section 13.3.3.
Estimation of Efflux. Efflux of nutrients from the roots to the external solution
may be estimated either as the difference between influx and net uptake
(Macduff and Jackson 1992; Lee 1993), or directly by measurement of nutrient
release from the roots into the external solution (Lee and Clarkson 1986; Aslam
et al. 1994). For the latter method the roots are transferred from an uptake solu-
tion which contains the nutrient or tracer to be studied to an efflux solution
which is identic al to the uptake solution except for lack of the specific nutrient
or tracer (Aslam et al. 1994). possibly, efflux is affected by the nutrient concen-
trations in the external solution, and related to the concurrently occuring influx.
To study efflux from roots into which concurrently influx takes place, the plants
are loaded with a tracer for time periods long enough to ensure isotopic equi-
librium, or near equilibrium, even in the slowest -exchanging compartments
under investigation. Preloading periods reported are variable, e.g. 15 h for
loading of lsN into soybean seedlings (Muller et al. 1995), or 2 days to label the
P pools of maize with 32p (Elliott et al. 1984). After preloading and short rinsing
of the roots to remove apoplastic tracers (see Sect. 13.3.3) plants are transferred
to efflux solutions which are identical in composition to the preloading solu-
tions except that the nutrient being studied is supplied in its natural form (non
radio active, natural mass), and tracer accumulation is measured in this efflux
solution. As tracers released from the roots by efflux may be recaptured by the
roots again, leading to an underestimation of efflux, the periods for efflux
should be short (l0-20min) and the volume of the efflux solution large.
However, tracer efflux is not linear with time even within short time periods
after transfer of plants to the efflux solution (Lee and Clarkson 1986; Muller et
al. 1995). This is because the cytoplasmic tracer pool which is the immediate
source for efflux is continuously diluted e.g. by nutrient influx and tracer
fluxes to the external solution and the xylem. This becomes a problem partic-
ularly for nitrate and ammonium where the cytoplasmic pool size in the roots
is small (about 0.25 to 1.25 .umol g-l root fresh weight, calculated on the basis of
the cytoplasm occupying 5% of the total root volume and nitrate concentra-
tions in the cytoplasm ranging between 5 and 25 mM) and influx is high (about
10 to 20,umolesh- 1 g- 1 root fresh weight; Muller et al. 1995). In maize, influx
and efflux of P in roots of intact plants have been simultaneously measured
by preloading of the plants with 32p for 48 h. Then the plants were transferred
to a solution containing 33p for 10 min to measure 32p accumulation in the
external solution (effiux) and 33p accumulation in the plants (influx; Elliott
et al. 1984).
Estimation of Uptake Kinetics. As a rule, ion uptake by plant roots shows

Michaelis-Menten kinetics at least in the concentration range below 1 mM
that is typical for soil solutions (Fig. 13.10). Thus, the potential ability of roots
for ion uptake may be characterized by Irnax, Krn, and Crnin (see Sect. 13.3.1).
Methods to determine the transport kinetics of nutrients by plant roots involve
time-course recording of ion/tracer depletion from the external solution
(Claassen and Barber 1974; Drew et al. 1984) or tracer accumulation in the plant
(Siddiqi et al. 1990). As the rate of tracer accumulation in the plants has to be
measured at different external concentrations, this method needs a separate set
of plants for each external concentrat ion to be tested. For the estimation ofV rnax
and Krn of enzymatic reactions it has been recommended to measure enzyme
activities (corresponding to influx) at least at 10 different concentrations
ranging from a factor 10 smaller to a factor 10 higher than Krn (Bisswanger
1994).
In contrast to tracer accumulation in the plants, nutrient/tracer depletion
from the external solution can be measured with one set of plants over the
whole range from saturating concentrations to Crnin. The use of tracers either for
the measurements of depletion from the external solution or accumulation in
the plants, allows the estimation of Irnax and Krn for the unidirectional influx
from the external solution into the roots, if the measurements at each discrete
external concentration are confined to a time period short enough to minimize
. __ .. _-_._-_._._. __ I max
._--_._---~.
Fig. 13.10. Schematic presentation of the

relationships between net uptake rates (net
influx = influx - effiux) of ions and their
externa! concentrations; Cmin = net uptake
zero (influx = effiux); Km = Michaelis-Menten
constant (externa! ion concentration \ External concentration (Cs )
allowing ha!f maxima! net uptake) C min
tracer and nutrient efflux (see above). It is obvious that for the estimation of
kinetic parameters for influx several sets of plants have to be used regardless
of whether tracer depletion from the external solution or tracer accumulation
in the plant is measured.
Measurement of Rate of Depletion of the External Solution. Depletion of the

external solution by roots can be measured at several discrete concentrations
ranging from saturating concentrations to minimum concentrations by trans-
ferring the plants from one solution to another. In each solution normally a
20-25% decrease in ion concentration is allowed so that a change in concen-
tration can be reliably detected. This means that only one data point is gener-
ated over a 20-25% change in concentration, while the uptake rate could have
been continually changing throughout that interval (Goyal and Huffaker 1986).
Therefore, ideally uptake kinetics should be determined by measuring nutrient
depletion continuously over the whole concentration range. In this case, the
volume of the vessels used for the upake study should be coordinated with the
root size of the experimental plants (number of plants and root fresh weight
per plant) to reach Cmin in reasonable time (6-lOh). This helps to avoid adapta-
tional responses of the roots to the diminishing external concentrations such
as decreases in Cmin and Km. Drew et al. (1984) suggest volume ratios of root:
solution between 1 : 10 and 1 : 40.
In the first minutes after transfer of the roots to the "depletion solution", a
large efflux may occur, particularly if the plants were precultured at high nutri-
ent concentrations. Therefore, before the depletion rate is measured, the plants
may be transferred for a few hours to a nutrient solution deficient in the nutri-
ent to be studied (Drew et al. 1984). After transferring the plants to the deple-
tion solution, samples are taken at regular intervals for tracer or nutrient
analysis. Goyal and Huffaker (1986) designed a fully automated micro computer-
based system for monitoring nitrate, nitrite and ammonium depletion from an
uptake solution. In this system samples of the nutrient solution taken over time
intervals of 2.5 min were directly injected into an HPLC, allowing measurement
of these nutrients in real-time. Alternatively, samples can also be taken using a
peristaltic pump (Swiader and Freiji 1996) or by hand. Samples are collected
until 30-60 min after the nutrient concentration in the depletion solution has
reached a minimum. Vigorous aeration of the nutrient solution is recom-
mended to promote mixing. Water losses from the nutrient solution caused by
evapotranspiration and sampling can be replaced by adding water according to
volume changes of the solution in the pots or weight changes of the whole
experimental unit (pots, solution, plants). Alternatively, water losses are deter-
mined from the difference in weight between start and end of the uptake study.
Then the water losses as well as the nutrient losses from sampling can be con-
sidered for the calculations of uptake rates.
At the end of the experiment the plants are harvested for determination of
the root parameter on which uptake rates are to be based (e.g. weight, length,
surface area). When depletion is measured over time periods long enough to
have an appreciable root growth, root size (R) at time (t) can be calculated using
the expression
where Ro is the root size (weight, length, surface area) at the start of the exper-
iment and k is a constant describing the rate of root size increase with time
(Claassen and Barber 1974).
Calculations. The net influx (In) between two sampling dates is calculated from
the volume of the nutrient solution (V), the decrease in its concentration per
unit time (6C/6t) and the root size (R e.g. root fresh weight or root length or
root surface area) according to the following formula:
In = -VIR 6C/6t.
To calculate the kinetic parameters for net uptake, the experimental data are
fitted to the Michaelis-Menten equation as described by Claassen and Barber
(1974):
In =[Imax x Cj(K m +C)]-C min ,
where In is the net influx, expressed as .umolh-l (unit roott\ Imax is the
maximum net influx at saturating concentrations, C is the nutrient concentra-
tion in the nutrient solution (.umoU-1) and Km the Michaelis constant. Cmin may
be directly obtained from the final concentration reached in the "depletion solu-
tion" (see above). Imax and Km may be estimated by the least-squares method, or
after linear transformation of the Michaelis-Menten equation, allowing the
results to be plotted as points on a straight line (Cornish-Bowden 1995; see
Table 13.1).
Due to experimental errors, frequent sampling is associated with large vari-
ability in uptake rates if calculated for an interval of two subsequent samplings.
To overcome this problem, the data obtained from the depletion experiment
can be fit to a series of cubic equations using a cubic spline computer program
(Swiader and Freiji 1996). The equation of the cubic spline is differentiated
to obtain In as a function of time. Imax is calculated from the negative slopes of
the linear portion of each curve. Km is calculated from the derivative equations
of the cubic spline by computing the solution concentration at which Imax/2
occurs (Swiader and Freiji 1996). Further methods for the calculation of the
kinetic parameters are described by Claassen and Barber (1974) and Drew
et al. (1984).
Table 13.1. Linear transformations of the Michaelis-Menten equation
Name of the plot Y X Intercepts with Slope of the line
X-axis Y-axis
Eadie-Hofstee In In/C Imax/Km Imax -Km

Lineweaver-Burk lIIn lIC -lIKm lIImax Km/lmax
Hanes C/ln C -Km Km/lmax lIImax
13.3.7 Nutrient Analysis
The methods of sampling, handling and analyzing plant tissue for mineral
nutrients are well described in handbooks of plant analysis (Jones and Case
1990).
Sample Preparation. If the plant samples are contaminated with dust or pesti-
cide residues, they should be washed e.g. with a weak (0.1-0.3%) detergent
solution, followed by rinsing in deionized water. Only fresh plant samples
should be washed and the washing procedure should be done quickly and
gently, to avoid leaching of mineral nutrients from the plant tissue. Washing is
always recommended if the tissue is to be analyzed for Fe, Al or Si (Jones and
Case 1990).
After washing, the samples are dried at 65 °C to constant weight to stop bio-
logical activity of the tissue which can reduce the plant mass by respiration.
Drying at temperatures much lower than 65°C, e.g. 50°C, may leave too much
moisture in the tissue and lead to enzymatic degradation of the tissue during
storage. Drying at temperatures much higher than 65°C, e.g. 105°C, may lead
to thermal decomposition of the tissue, particularly if ground tissue samples
are re-dried before nutrient analysis (Steyn 1959).
If only a subsample of the collected plant tissue is used for nutrient analy-
sis, the dried material has to be ground. The purpose of grinding is to obtain
subsamples of uniform composition. The partide size which is necessary to
obtain uniformity is mainly dependent on the size of the subsample needed for
analysis. Grinding with a commercial coffee miU (partide size reduction to pass
a 20-mesh screen) may be sufficient if 500 mg aliquots of plant tissue are ana-
lyzed, whereas fine pulverization of the tissue is necessary if only 3-10mg
aliquots are analyzed e.g. for 15N in a mass-spectrometer. During grinding, the
plant tissue may be contaminated mainly with Al, Cu, Fe or Zn (Jones and Case
1990). So, when these elements have to be analyzed, the tissue should be ground
with an agate miU (Steyn 1959). The ground material should be dried again for
24 h at 65°C to remove any moisture added during grinding, and then stored
e.g. in sealed bottles until analysis.
Chemical Analysis. The dried and ground material may be directly used for
analysis of S in automated sulfur analyzers and N in automated Dumas instru-
ments or by Kjeldahl digestion. Methods for the determination of other ele-
ments involve destruction of the tissue's organic component to convert the
elements to a soluble form for analysis. There are essentially two decomposi-
tion procedures, wet acid digestion and high temperature dry oxidation, fre-
quently referred to as wet and dry ashing, respectively (for a detailed discussion
ofwet and dry ashing see Jones and Case 1990). Wet ashing is not recommended
for B, since this element can be partially lost by volatilization during wet oxi-
dation (Feldman 1961).
For dry ashing, aliquots of the sample are weighed into commercially avail-
able ashing vessels or small glass bottles in which contamination of the sample
e.g. by B or Na should be exduded. The plant tissue is ashed for 6-8 h in a muffle
furnace at 450-500 aC. Complete oxidation of the plant tissue is indicated by the
"white" appearance of the remaining ash. If a dean white ash is not obtained after
8 h, re-ashing for 2 h after addition of ashing aids such as HN0 3 or 10% H2S04 is
necessary. Thereafter, the ash is dissolved in dilute acid (HCl or HN0 3 ). The dear
solution is ready for elemental assay, with or without further dilution.
Wet ashing is the destruction of organic matter by high temperature acid
digestion. The common acids used are H2 S04 , HN0 3, and HCl0 4, usually in some
combination of two or alI three. The wet oxidation is frequently done in com-
mercially available sealed digestion tubes inserted into ports of a temperature-
controlled digestion block. Alternatively, plant tissue can also be digested in a
mixture of HN0 3 and 30% H20 2 heated by ultraviolet radiation (Ogner 1983)
or in a microwave oven (White and Douthit 1985).
Instruments for Analysis. Colorimetric methods involving measurement in

spectrophotometers are used for B (Basson et al. 1969) and P (Gericke and
Kurmies 1952; Murphy and Riley 1962). The most commonly used methods for
determination of the other elements involve flame emis sion spectrometry,
atomic absorption spectrometry, and plasma spectrometry (Table 13.2). Ions
may be measured by HPLC with anion and cation exchange columns (WeiB
1991; Table 13.2).
13.4 Concluding Remarks
In future, the development of more precise analytical techniques in the labora-

tory will allow collection of smaller samples, and thus, sampling of roots or
Table 13.2. Instruments for the measurement of mineral nutrients
Instrument Mineral nutrient
Automated sulfur analyzer S

Automated Dumas instrument N
Spectrophotometer P,Mo
Flame photometer Ca, K, Na, Cu, Fe, Mn
Atomic absorption spectrometer Ca, Mg, Cu, Fe, Mn, Zn
Inductively coupled plasma emission spectrometer P, K, Ca, Mg, Cu, Fe, Mn, Zn, B
(ICP)
HPLC (ion chromatography) sol- , HPol-, NO,-, MoOl-, cations
compounds in the immediate vicinity of roots with higher resolution in

space and time. The critical step in obtaining information about nutrient acqui-
sition capacity of roots will remain sampling itself. There is no generally
accepted and recommended method for measuring root capacity for nutrient
acquisition. Artificial growing conditions e.g. in rhizoboxes or along root
windows may allow sampling and in situ measurements of root activity with
minimal disturbance of the roots during sampling, but possibly do not exact1y
reflect root behaviour under natural conditions. Natural growing conditions
often require strong disturbance of the root/soil system prior to or during mea-
surement of acquisition capacity, or give liule resolution of root activity in time
and space. Thus, the best way to collect data on root activity will strongly
depend on the specific research question to be solved.
Further Reading
Brundrett MC, Melville L, Peterson RL (1994) Practical methods in mycorrhizal research. Myco-
logue Publications, Waterloo, Canada
Clarkson DT (1996) Root structure and sites of ion uptake. In: Waisel Y, Eshel A, Kafkafi U (eds)
Plant roots: the hidden half. 2nd edn. Marcel Dekker, New York, pp 483-510
Grayston SJ, Vaughan D, Jones D (1996) Rhizosphere carbon fiow in trees, in comparison with
annual plants: the importance of root exudation and its impact on microbial activity and
nutrient availability. Appl Soil Ecol 5: 29-56
Jones JB Jr, Case VW (1990) Sampling, handling and analyzing plant tissue samples. In:
Westerman RL (ed) Soil testing and plant analysis. Soil Science Society of America,
Madison, pp 389-427
References
Alef K (1995) Sterilization of soi! and inhibition of microbial activity. In: Alef K, Nannipieri P
(eds) Methods in applied soi! microbiology and biochemistry. Academic Press, London, pp
52-54
Amann C, Amberger A (1989) Phosphorus efficiency of buckwheat (Fagopyrum esculentum). Z
Pflanzenernăhr Bodenkd 152: 181-189
Armstrong J, Armstrong W, Beckett PM (1992) Venturi- and humidity-induced pressure flows
enhance rhizome aeration and rhizosphere oxidation. New Phytol120: 197-207
Aslam M, Travis RL, Huffaker RC (1994) Stimulation of nitrate and nitrite effiux by ammonium
in barley (Hordeum vulgare 1.) seedlings. Plant Physioll06: 1293-1301
Azaizeh HA, Marschner H, Romheld V, Wittenmayer L (1995) Effects of vesicular-arbuscular
fungus and other soi! microorganisms on growth, mineral nutrient acquisition and root
exudation of soi!-grown maize plants. Mycorrhiza 5: 321-327
Barber DA, Gunn KB (1974) The effect of mechanical forces on the exudation of organic
substances by the roots of cereal plants grown under steri!e conditions. New Phytol 73:
39-45
Barber SA (1984) Soi! nutrient bioavai!ability. A mechanistic approach. John Wi!ey, New York
Barber SA, Ozanne PG (1970) Autoradiographic evidence for the differential effect of four
plant species in altering the Ca content of the rhizosphere soi!. Soi! Sci Soc Am Proc 34:
635-637
Bartlett EM, Lewis DH (1973) Surface phosphatase activity of mycorrhizal roots of beech. Soi!
Biol Biochem 5: 249-257
Basson WD, Bohne RG, Stanton DA (1969) An automated procedure for the determination of
boron in plant tissue. Analyst 94: 1135-1141
Benjamin LR, Peach L, van Woerden IC, Oppelaar A (1996) A technique to estimate the radial
extent of active mineral absorption by individual plants in carrot stands. J Exp Bot 47:
687-692
Bhat KKS, Nye PH (1973) Diffusion of phosphate to the plant roots in soi!. Quantitative autora-
diography of the depletion zone. Plant Soi! 38: 161-175
Bieleski RL, Johnson PN (1972) The externallocation of phosphatase activity in phosphorus-
deficient Spirodela oligorrhiza. Aust J Biol Sci 25: 707-720
Bienfait HF, Bino RJ, van der Bliek AM, Duivenvoorden JR, Fontaine JM (1983) Characterization
of ferric reducing activity in roots of Fe-deficient Phaseolus vulgaris. Physiol Plant 59:
196-202
Bisswanger H (1994) Enzymkinetik. Theorie und Methoden. VCH-Verlag Weinheim
Bloom AJ, Caldwell RM (1988) Root excision decreases nutrient absorption and gas fluxes. Plant
Physiol 87: 794-796
Bloom AJ, Sukrapanna SS (1990) Effects of exposure to ammonium and transplant shock upon
the induction of nitrate absorption. Plant Physiol 94: 85-90
Boero G, Thien S (1979) Phosphatase activity and phosphorus avai!ability in the rhizosphere of
corn roots. In: Harley JL, Russell S (eds) Soi!-root interface. Academic Press, London
Boeuf-Tremblay V, Plantureux S, Guckert A (1995) Influence of mechanical impedance on root
exudation of maize seedlings at two developmental stages. Plant Soi! 172: 279-287
Bohm W (1979) Methods of studying root systems. Springer, Berlin Heidelberg New York
Brady DJ, Gregory PJ, Fillery IRP (1993) The contribution of different regions of the seminal
roots of wheat to uptake of nitrate from soi!. Plant Soi! 1551156: 155-158
Brown JC, Ambler JE (1974) Iron stress response in tomato (Lycopersicon esculentum). 1. Sites
of Fe reduction, absorption and transport. Physiol Plant 31: 221-224
Brundrett MC, Piche Y, Peterson RL (1984) A new method for observing the morphology of
vesicular-arbuscular mycorrhizae. Can J Bot 62: 2128-2134
Brundrett MC, Melville L, Peterson RL (1994) Practical methods in mycorrhizal research. Myco-
logue Publications, Waterloo, Canada
Cairney JWG, Ashford AE (1989) Reducing activity at the root surface in Eucalyptus pilularis-
Pisolithus tinctorius ectomycorrhizas. Aust J Plant Physiol 16: 99-105
Cakmak 1, Marschner H (1988) Increase in membrane permeability and exudation in roots of
zinc deficient plants. J Plant Physiol132: 356-361
Canning RE, Kramer PJ (1958) Salt absorption and accumulation in various regions of roots.
Am J Bot 45: 378-382
Canny MJ, McCully ME (1988) The xylem sap of maize roots: its collection, composition and
formation. Aust J Plant Physiol 15: 557-566
Chabot S, Becard G, Piche Y (1992) Life cyele of Glomus intraradix in root organ culture. Mycolo-
gia 84: 315-321
Chang CW, Bandurski RS (1964) Exocellular enzymes of corn roots. Plant Physiol 39:
60-64
Chapin FS III, Van Cleve K (1989) Approaches to studying nutrient uptake, use and loss in plants.
In: Pearcy RW, Ehleringer JR, Mooney HA, Rundel PW (eds) Plant physiological ecology.
Field methods and instrumentation. Chapman and Hall, London, pp 185-207
Chapin FS III, Van Cleeve K, Tyron PR (1986) Relationship of ion absorption to growth rate in
taiga trees. Oecologia 69: 238-242
Claassen N, Barber SA (1974) A method for characterizing the relation between nutrient con-
centration and flux into roots of intact plants. Plant Physiol 54: 564-568
Claassen N, Jungk A (1982) Kaliumdynamik im wurzelnahen Boden in Beziehung zur Kaliu-
maufnahme von Maispflanzen. Z Pflanzenerniihr Bodenkd 145: 513-525
Claassen N, Hendriks L, Jungk A (1981) Erfassung der Mineralstoffverteilung im wurzelnahen
Boden durch Autoradiographie. Z Pflanzenerniihr Bodenkd 144: 306-316
Clarkson DT (1985) Factors affecting mineral nutrient acquisition by plants. Annu Rev Plant
Physiol36: 77-115
Clarkson DT (1996) Root structure and sites of ion uptake. In: Waisel Y, Eshel A, Kafkafi U (eds)
Plant roots: the hidden half, 2nd edn. Marcel Dekker, New York, pp 483-470
Clarkson DT, Saker LR, Purves JV (1989) Depression of nitrate and ammonium transport in
barley plants with diminished sulphate status. Evidence of co-regulation of nitrogen and
sulphate intake. J Exp Bot 40: 953-963
Clarkson DT, Gojon A, Saker LR, Wiersema PK, Purves Jv, Tillard P, Arnold GM, Paans AJM,
Vaalburg W, Stulen I (1996) Nitrate and ammonium influxes in soybean (Glycine max)
roots: direct comparison of 13N and 15N tracing. Plant Cell Environ 19: 859-868
Cornish-Bowden A (1995) Analysis of enzyme kinetic data. Oxford University Press, Oxford
Cruz C, Lips SH, Martins-Louc;:ao MA (1995) Uptake regions of inorganic nitrogen in roots of
carob seedlings. Physiol Plant 95: 167-175
Darrah PR (1996) Rhizodeposition under ambient and elevated CO z levels. Plant Soi! 187:
265-275
Deane-Drummond CE (1990) Biochemical and biophysical aspects of nitrate uptake and its reg-
ulation. In: YP Abrol (ed) Nitrogen in higher plants. Research Studies Press, Taunton, UK,
37 pp
Delhaize E, Ryan PR (1995) Aluminum toxicity and tolerance in plants. Plant Physiol 107:
315-321
Delhaize E, Ryan PR, Randall PJ (1993) Aluminum tolerance in wheat (Triticum aestivum 1.). II.
Aluminum-stimulated excretion of malic acid from root apices. Plant Physiol103: 695-702
Dinkelaker B, Marschner H (1992) In vivo demonstration of acid phosphatase activity in the
rhizosphere of soil-grown plants. Plant Soi1144: 199-205
Dinkelaker B, Romheld V, Marschner H (1989) Citric acid excretion and precipitation of calcium
citrate in the rhizosphere of white lupin (Lupinus albus 1.). Plant CeH Environ 12: 285-292
Dinkelaker B, Hahn G, Romheld V, Wolf GA, Marschner H (1993a) Non-destructive methods for
demonstrating chemical changes in the rhizosphere 1. Description of methods. Plant Soi!
155/156: 67-70
Dinkelaker B, Hahn G, Marschner H (1993b) Non -destructive methods for demonstrating chem-
ical changes in the rhizosphere II. Application of methods. Plant Soil155/156: 71-74
Dinkelaker B, Hengeler C, Neumann G, Eltrop L, Marschner H (1996) Root exudates and mobi-
lization of nutrients. In: Rennenberg H, Eschrich W (eds) Trees - contributions to modern
tree physiology. SPB Academic Publishing, Amsterdam, pp 3-14
Dinkelaker B, Hengeler C, Neumann G, Eltrop L, Marschner H (1997) Root exudates and mobi-
lization of nutrients. In: Rennenberg H, Eschrich W, Ziegler H (eds) Trees - contributions
to modern tree physiology, Backhuys, Leiden, The Netherlands, pp 441-451
Drew MC, Saker LR (1975) Nutrient supply and the growth of the seminal root system in barley.
II. Localized compensatory increases in lateral root growth and rates of nitrate uptake when
nitrate supply is restricted to only part of the root system. J Exp Bot 29: 79-90
Drew MC, Saker LR (1986) Ion transport to the xylem in aerenchymatous roots of Zea mays 1.
J Exp Bot 37: 22-33
Drew MC, Saker LR, Barber SA, Jenkins W (1984) Changes in the kinetics of phosphate and
potassium absorption in nutrient-deficient barley roots measured by a solution-depletion
technique. Planta 160: 490-499
Elliott GC, Lynch J, Lăuchli A (1984) Influx and efflux of Pin roots of intact maize plants. Plant
Physiol 76: 336-341
Else MA, Davies WJ, Whitford PN, Hali KC, Jackson MB (1994) Concentrations of abscisic acid
and other solutes in xylem sap from root systems of tomato and castor-oi! plants are dis-
torted by wounding and variable sap flow rates. J Exp Bot 45: 317-323
Else MA, Hali KC, Arnold GM, Davies WJ, Jackson MB (1995) Export of abscisic acid, 1-
aminocyclopropane-l-carboxylic acid, phosphate, and nitrate from roots to shoots of
flooded tomato plants. Accounting for effects of xylem sap flow rate on concentration and
delivery. Plant Physioll07: 377-384
Engels C, Marschner H (1992) Root to shoot translocation of macronutrients in relation to
shoot demand in maize (Zea mays 1.) grown at different root zone temperatures.
Engels C, Buerkert B, Marschner H (1994) Nitrogen and sugar concentrations in the xylem
exudate of field-grown maize at different growth stages and levels of nitrogen ferti!ization.
Eur J Agron 3: 197-204
Epstein E, Schmid WE, Rains DW (1963) Significance and technique of short-term experiments
on solute absorption by plant tissue. Plant CeH Physiol 4: 79-84
Ernst M, Romheld V, Marschner H (1989) Estimation of phosphorus uptake capacity by differ-
ent zones of the primary root of soi!-grown maize (Zea mays 1.). Z Pflanzenernăhr
Bodenkd 152: 21-25
Farr E, Vaidynathan V, Nye PH (1969) Measurement of ionic concentration gradients in soil near
roots. Soil Sci 107: 385-391
Feldman C (1961) Evaporation of boron from acid solutions and residues. Anal Chem 33:
1916-1920
Felipe MR, Pozuelo JM, Cintas AM (1979) Acid phosphatase localization at the surface of young
corn roots. Agrochimica 23: 143-150
Fitter AH (1986) Spatial and temporal patterns of root activity in a species-rich alluvial grass-
land. Oecologia 69: 594-599
Flessa H, Fischer WR (1992) Redoxprozesse in der Rhizosphăre von Land- und Sumpfpflanzen.
Fox TC, Shaff JE, Grusak MA, Norvell WA, Chen Y, Chaney RL, Kochian LV (1996) Direct mea-
surement of 59Fe-Iabeled Fe2+influx in roots of pea using a chelator buffer system to control
free Fe2+ in solution. Plant Physioll11: 93-100
Gahoonia TS (1993) Influence of root-induced pH on the solubilityof soil aluminium in the rhi-
zosphere. Plant Soil149: 289-291
Gahoonia TS, Nielsen NE (1991) A method to study rhizosphere processes in thin soillayers of
different proximity to roots. Plant Soil135: 143-148
Gahoonia TS, Nielsen NE (1992) Control of pH at soil-root interface. Plant Soil140: 49-54
Gahoonia TS, Claassen N, Jungk A (1992) Mobilization of phosphorus in different soils by rye-
grass supplied with ammonium and nitrate. Plant Soil143: 241-248
George E, Marschner H, Jakobsen I (1995) Role of arbuscular mycorrhizal fungi in uptake of
phosphorus and nitrogen from soil. Crit Rev Biotechnol15: 257-270
George E, Gorgus E, Schmeisser A, Marschner H (1996) A method to measure nutrient uptake
from soil by mycorrhizal hyphae. In: Azcon-Aguilar C, Barea JM. Mycorrhizas in integrated
systems from genes to plant development. Oftice for Official Publications of the European
Community, Luxembourg, pp 535-538
Gerdemann JW, Nicolson TH (1963) Spores of mycorrhizal Endogone species extracted from soil
by wet-sieving and decanting. Trans Br Mycol Soc 46: 235-244
Gericke S, Kurmies B (1952) Die colorimetrische Phosphorsăurebestimmung mit Ammonium-
Vanadat-Molybdat und ihre Anwendung in der Pflanzenanalyse. Z Pflanzenernăhr Diing
Bodenkd 59: 235-247
Gerke J, Romer W, Jungk A (1994) The excretion of citric and malic acid by proteoid roots of
Lupinus albus 1.: effects on soil solution concentrations of phosphate, iron, and aluminium
in the proteoid rhizosphere samples of an oxisol and a luvisol. Z Pflanzenernăhr Bodenkd
157:289-294
Gil de Carrasco C, Guzman M, Lorente FA, Urrestarazu M (1994) Xylem sap extraction: a
method. Commun Soil Sci Plant Anal 25: 1829-1839
Giovanetti M, Mosse B (1980) An evaluation of techniques for measuring vesicular-arbuscular
mycorrhizal infection in roots. New Phytol 84: 489-500
Glass ADM, Shaff JE, Kochian LV (1992) Studies of the uptake of nitrate in badey IV, Electro-
physiology. Plant Physiol 93: 456-463
Gollany HT, Schumacher TE (1993) Combined use of colorimetric and microelectrode methods
for evaluating rhizosphere pH. Plant Soil154: 151-159
Gottlein A, Hell U, Blasek R (1996) A system for microscale tensiometry and lysimetry. Geo-
derma 69: 147-156
Goyal SS, Huffaker RC (1986) A novel approach and a fully automated microcomputer-based
system to study kinetics of N0 3-, N0 2-, and NH.+ transport simultaneously by intact wheat
seedlings. Plant Cell Environ 9: 209-215
Grace C, Stribley DP (1991) A safer procedure for routine staining ofvesicular-arbuscular myc-
orrhizal fungi. Mycol Res 95: 1160-1162
Grayston SJ, Vaughan D, Jones D (1996) Rhizosphere carbon flow in trees, in comparison with
annual plants: the importance of root exudation and its impact on microbial activity and
nutrient availability. Appl Soil Ecol 5: 29-56
Grierson PF (1992) Organic acids in the rhizosphere of Banksia integrifolia 1. Plant Soil 144:
259-265
Grzebisz W, Floris J, van Noordwijk M (1989) Loss of dry matter and cell contents from fibrous
roots of sugar beet due to sampling, storage and washing. Plant Soil 113: 53-57
Hamei C, Fyles H, Smith DL (1990) Measurement of development of endomycorrhizal mycelium
using three different vital stains. New Phytol115: 297-302
Harrison SI, Lepp NW, Phipps DA (1978) Uptake of copper by excised roots. 1. A modified
experimental technique for measuring ion uptake by excised roots, and its application in
determining uptake characteristics of "free" copper ions in excised Hordeum roots.
Z Pflanzenphysiol 90: 443-450
Harrison SJ, Lepp NW, Phipps DA (1979) Uptake of copper by excised roots. II. Copper desorp-
tion from the free space. Z pflanzenphysiol 94: 27-34
Harrison-Murray RS, Clarkson DT (1973) Relationships between structural development and
the absorption of ions by the root system of Cucurbita pepo. Planta 114: 1-16
Hăussling M, Leisen E, Marschner H, Riimheld V (1985) An improved method for non-
destructive measurements of the pH at the root-soil interface (rhizosphere). J Plant Physiol
117: 371-375
Hăussling M, Jorns CA, Lehmbecker G, Hecht-Buchholz Ch, Marschner H (1988) Ion and water
uptake in relation to root development in Norway spruce [Picea abies (1.) Karst.]. J Plant
Physiol133: 486-491
Hawkins H-J, George E (1997) Hydroponic culture of the mycorrhizal fungus Glomus messeae
with Linum usitatissimum 1., Sorghum bicolor 1. and rriticum aestivum 1. Plant Soil196:
143-149
Headley AD, Callaghan TV, Lee JA (1985) The phosphorus economy of the evergreen tundra
plant, Lycopodium annotinum. Oikos 45: 235-245
Helal HM, Sauerbeck D (l983) A method to study turnover processes in soillayers of different
proximity to roots. Soil Biol Biochem 15: 223-225
Hendriks L, Jungk A (1981) Erfassung der Mineralstoffverteilung in Wurzelnăhe durch getren-
nte Analyse von Rhizo- und Restboden. Z Pflanzenernăhr Bodenkd 144: 276-282
Henriksen GH, Bloom AJ, Spanswick RM (1990) Measurement of net fluxes of ammonium and
nitrate at the surface of barley roots using ion-selective microelectrodes. Plant Physiol 93:
271-280
Henriksen GH, Raman DR, Walker LP, Spanswick RM (1992) Measurement of net fluxes of
ammonium and nitrate at the surface of barley roots using ion-selective microelectrodes.
II. Patterns of uptake along the root axis and evaluation of the microelectrode flux esti-
mation technique. Plant Physiol 99: 734-747
Hether NH, Olsen RA, Jackson LL (1984) Chemical identification of iron reductants exuded by
plant roots. J Plant Nutr 7: 667-676
Hodge A, Grayston SI, Ord BG (1996) A novel method for characterisation and quantification of
plant root exudates. Plant Soi1184: 97-104
Hoffland E, Findenegg GR, Nelemans JA (1989) Solubilization of rock phosphate by rape.
II. Local root exudation of organic acids as a response to P-starvation. Plant Soil 113:
161-165
Hogberg P, Jensen P, Nasholm T, Ohlsson H (1995) Uptake of 24Mg by excised pine roots: a pre-
liminary study. Plant Soill72: 323-326
Horst WJ, Wagner A, Marschner H (1982) Mucilage protects root meristems from aluminium
injury. Z Pflanzenphysioll05: 435-444
Horst WJ, Asher CJ, Cakmak 1, Szulkiewicz P, Wissemeier AH (1992) Short-term responses of
soybean roots to aluminium. J Plant Physiol140: 174-178
Hillster A, Marschner H (1994) PCDD/PCDF-Transfer in Zuchini und Tomaten. Veroff PAl> 8:
579-589
Jackson RB, Manwaring JH, Caldwell MM (1990) Rapid physiological adjustment of roots to
localized soil enrichment. Nature 344: 58-60
Jaillard B, Ruiz L, Arvieu IC (1996) pH mapping in transparent gel using color videodensitom-
etry. Plant Soil183: 85-95
Janzen HH (1990) Deposition of nitrogen into rhizosphere by wheat root. Soil Biol Biochem 22:
1155-1160
Jeschke WD, Pate JS (1995) Mineral nutrition and transport in xylem and phloem of Banksia
prionotes (Proteaceae), a tree with dimorphic root morphology. J Exp Bot 46: 895-905
Jeschke WD, Klagges S, Bhatti AS (1996) Collection and composition of xylem sap and root struc-
ture in two halophytic species. Plant Soill72: 97-106
Johnson LF, Cur! EA (1972) Control of soil environment. In: Johnson LF, Cur! EA (eds) Methods
for research on the ecology of soil-borne plant pathogens. Burgess Publishing Company,
Minneapolis, pp 82-91
Johnson JF, AlIan DL, Vance CP, Weiblen G (1996) Root carbon dioxide fixation by phosphorus-
deficient Lupinus albus. Plant Physiol112: 19-30
Jones DL, Darrah PR (1992) Re-sorption of organic compounds by roots of Zea mays 1. and its
consequences in the rhizosphere. I. Resorption of 14C labelled glucose, mannose and citric
acid. Plant Soil143: 259-266
Jones DL, Darrah PR (1993) Re-sorption of organic compounds by roots of Zea mays 1. and its
consequences in the rhizosphere II. Plant Soil153: 47-59
Jones DL, Darrah PR (1994) Amino-acid influx at the soil-root interface of Zea mays 1. and its
implications in the rhizosphere. Plant Soil163: 1-12
Jones DL, Darrah PR (1995) Influx and efflux of organic acids across the soil-root interface of
Zea mays 1. and its implications in rhizosphere C flow. Plant Soil173: 103-109
Jones JB Jr, Case VW (1990) Sampling, handling, and analyzing plant tissue samples. In:
Westerman RL (ed) Soil testing and plant analysis. Soil Science Society of America,
Madison, pp 389-427
Kape R, Wex K, Parniske M, Gorge E, Wetzel A, Werner D (1992) Legume root metabolites and
VA-mycorrhiza development. J Plant Physiol141: 54-60
Karpov EA, Potapov NG (1975) Reducing activity of the root surfaces in corn plants in connec-
tion with its differentiation and absorptive capacity. Fiziol Rast 22: 298-304
Keith H, Oades IM, Martin IK (1986) Input of carbon to soil from wheat plants. Soil Biol Biochem
18:445-449
Kraus M, Fusseder A, Beck E (1987) In situ determination of the phosphate gradient around a
root by radioautography of frozen soil sections. Plant Soil 97: 407-418
Kronzucker HJ, Siddiqi MY, Glass ADM (1995) Compartmentation and flux characteristics of
ammonium in spruce. Planta 196: 691-698
Kuchenbuch R, Jungk A (1982) A method for determining concentration profiles at soil-root
interface by thin slicing rhizosphere soil. Plant Soil 68: 391-394
Kuchenbuch R, Jungk A (1984) Wirkung der Kaliumdiingung auf die Kaliumverfiigbarkeit in

der Rhizosphare von Raps. Z Pflanzenernahr Bodenkd 147: 435-448
Kurien S, Goswami AM, Deb DL (1992) Root activity of two citrus rootstocks assessed using
radiotracer techniques. J Hortic Sci 67: 87-94
Lavy TL, Barber SA (1964) Movement ofmolybdenum in soil and its effect on availability to the
plants. Soil Sci Soc Am Proc 28: 93-97
Lazof DB, Rufty TW Jr, Redinbaugh MG (1992) Localization of nitrate absorption and translo-
cation within morphological regions of the corn root. Plant Physioll00: 1251-1258
Lee RB (1993) Control of net uptake of nutrients by regulation of influx in barley plants recov-
ering from nutrient deficiency. Ann Bot 72: 225-230
Lee RB, Clarkson DT (1986) Nitrogen-13 studies of nitrate fluxes in barley roots. 1. Compart-
mental analysis from measurements of 13N efflux. J Exp Bot 37: 1753-1767
Lee RB, Drew MC (1986) Nitrogen-13 studies of nitrate fluxes in barley roots. II. Effect of plant
N-status on the kinetic parameters of nitrate influx. J Exp Bot 37: 1768-1779
Leiser AT (1968) A mucilaginous root sheat in Ericaceae. Am J Bot 55: 391-398
Leon M, Laine P, Ourry A, Boucaud J (1995) Increased uptake of native soil nitrogen by roots of
Lolium multiflorum Lam. after nitrogen fertilization is explained by a stimulation of the
uptake process itself. Plant Soil173: 197-203
Li XL, George E, Marschner H (1991) Phosphorus depletion and pH decrease at the root-soil
and hyphae-soil interfaces of VA mycorrhizal white clover fertilized with ammonium. New
PhytoII19:397-404
Lipton DS, Blanchar RW, Blevins DG (1987) Citrate, malate, and succinate concentration in
exudates from P-sufficient and P-stressed Medicago sativa 1. seedlings. Plant Physiol
85:315-317
Maathuis FJM, Sanders D (1996) Mechanisms of potassium absorption by higher plant roots.
Physiol Plant 96: 158-168
Macduff JH, Jackson SB (1992) Influx and efflux of nitrate and ammonium in Italian ryegrass
and white clover roots: comparisons between effects of darkness and defoliation. J Exp Bot
43:525-535
Marschner H, Richter C (1973) Akkumulation und Translokation von K+, Na+ und Ca2+ bei
Angebot zu einzelnen Wurzelzonen von Maiskeimpflanzen. Z Pflanzenernahr Bodenkd
135: 1-15
Marschner H, Romheld V (1983) In vivo measurement of root-induced pH changes at the
soil-root interface: effect of plant species and nitrogen source. Z pflanzenphysiol 111:
241-251
Marschner H, Romheld V, Ossenberg-Neuhaus H (1982) Rapid method for measuring changes
in pH and reducing processes along roots of intact plants. Z Pflanzenphysioll05: 407-416
Marschner H, Romheld V, Kissel M (1986) Different strategies in higher plants in mobilization
and uptake of iron. J Plant Nutr 9: 695-713
Marschner H, Romheld V, Kissel M (1987) Localization of phytosiderophore release and iron
uptake along intact barley roots. Physiol Plant 7l: 157-162
Marschner H, Haussling M, George E (1991) Ammonium and nitrate uptake rates and rhizos-
phere pH in non-mycorrhizal roots of Norway spruce [Picea abies (1.) Karst.]. Trees 5:
14-21
Matzner SL, Richards JH (1996) Sagebrush (Artemisia tridentata Nutt.) roots maintain nutrient
uptake capacity under water stress. J Exp Bot 47: 1045-1056
McKane RB, Grigal DF (1990) Spatiotemporal differences in I'N uptake and the organization of
an old-field plant community. Ecology 71: 1126-1132
Meharg AA, Kilham K (1991) A novel method of quantifying root exudation in the presence of
soil microflora. Plant Soil133: 111-116
Meharg AA, Kilham K (1995) Loss of exudates from the roots of perennial ryegrass inoculated
with a range of micro-organisms. Plant Soil 170: 345-349
Miller AJ, Smith SJ (1996) Nitrate transport and compartmentation in cereai root cells. J Exp
Bot 47: 843-854
Miller DM (1981) Pressure-flow characteristics of the roots of Zea mays. Plant Soil63: 15-18
Minorsky PV, Spanswick RM (1989) Electrophysiological evidence for a role for calcium in tem-
perature sens ing by roots of cucumber seedlings. Plant Cell Environ 12: 137-143
Miyasaka SC, Buta JG, Howell RK, Foy CD (1991) Mechanism of aluminum tolerance in snap-
beans. Root exudation of citric acid. Plant Physiol 96: 737-743
Morei JL, Mench M, Guckert A (1986) Measurement ofPb+, Cu 2+ and Cd 2+binding with mucilage
exudates from maize (Zea mays 1.) roots. Biol Fertil Soils 2: 29-34
Moritsugu M, Shibasaka M, Kawasaki T (1993) Where is the most important and efficient site
for absorption and translocation of cations in excised barley roots? Soil Sci Plant Nutr 39:
299-307
Muller B, Tillard P, Touraine B (1995) Nitrate fluxes in soybean seedling roots and their response
to amin o acids: an approach using I'N. Plant Cell Environ 18: 1267-1279
Murphy J, Riley JP (1962) A modified single solution method for the determination of phos-
phate in natural water. Anal Chim Acta 27: 31-36
Neumann G, Dinkelaker B, Marschner H (1995) Kurzzeitige Abgabe organischer Sauren aus Pro-
teoidwurzeln von Hakea undu/ata (Proteaceae). In: Merbach W (ed) Pflanzliche Stoffauf-
nahme und mikrobielle Wechselwirkungen in der Rhizosphare. BG Teubner, Stuttgart, pp
129-136
Newman IA, Kochian LV, Grusak MA, Lucas WJ (1987) Fluxes of H+ and K+ in corn roots:
characterization and stoichiometries using microelectrodes. Plant Physiol 84: 1177-
1184
Norris JR, Read DJ, Varma AK (1992) Methods in microbiology, voi 24. Academic Press, London
Ogner G (1983) Digestion of plants and organic soils using nitric acid, hydrogen peroxide and
UV radiation. Commun Soil Sci Plant Anal 14: 936-943
Ohwaki Y, Hirata H (1990) Phosphorus absorption by chickpea (Cicer arietinum) as affected by
VA mycorrhiza and carboxylic acids in root exudates. In: van Beusichem ML (ed) Plant
nutrition - physiology and applications. Kluwer, Amsterdam, pp 171-177
Ohwaki Y, Hirata H (1992) Differences in carboxylic acid exudation among P-starved legumi-
nous crops in reiat ion to carboxylic acid contents in plant tissues and phospholipid levels
in roots. Soil Sci Plant Nutr 38: 235-243
Olsson PA, Băăth E, Jakobson 1 (1997) Phosphorus effects on the mycelium and storage struc-
tures of an arbuscular mycorrhizal fungus as studied in the soil and roots by analysis of
fatty acid signatures. Appl Environ Microbiol 63: 3531-3538
Papavizas GC, Davey CB (1961) Extent and nature of the rhizosphere of lupinus. Plant Soil14:
215-236
Passioura JB (1972) Quantitative autoradiography in the presence of crossfire. In: Liittge U (ed)
Microradioautography and Electron Probe Analysis, Springer, Berlin Heidelberg New York,
pp 49-59
Passioura JB, Munns R (1984) Hydraulic resistance of plants. II. Effects of rooting medium, and
time of day, in barley and lupin. Aust J Plant Physiol11: 341-350
Pearson RW (1974) Significance of rooting pattern to crop production and some problems of
root research. In: Carson EW (ed) The Plant Root and Its Environment. The University
Press of Virginia, Charlottesville, pp 247-270
Pellet OM, Grunes OL, Kochian LV (1995) Organic acid exudation as an aluminum-tolerance
mechanism in maize (Zea mays L.). Planta 196: 788-795
Petersen W, Bottger M (1991) Contribution of organic acids to the acidification of the rhizos-
phere of maize seedlings. Plant Soi! 132: 159-163
Powlson OS, Jenkinson OS (1976a) The effects of biocidal treatments on metabolism in soi!. 1.
Fumigation with chloroform. Soi! Biol Biochem 8: 167-177
Powlson OS, Jenkinson OS (1976b) The effects of biocidal treatments on metabolism in soi!. II.
Gamma irradiation, autoclaving. Soi! Biol Biochem 8: 179-188
Prikryl Z, Vancura V (1980) Root exudates of plants. VI. Wheat exudation as dependent on
growth, concentration gradient of exudates and the presence of bacteria. Plant Soi! 57:
69-83
Reidenbach G, Horst WJ (1995) Bedeutung verschiedener Wurzelzonen fUr die Nitrataufnah-
merate bei Mais (Zea mays L.). VOLUFA-Schriftenr 40: 121-124
Reining E, Merbach W, Knof G (1995) 15N distribution in wheat and chemical fractionation of
root-borne 15N in the soi!. Isotopes Environ Health Stud 31: 345-349
Ri!ey O, Barber SA (1971) Effect of ammonium and nitrate fertilization on phosphorus uptake
as related to root induced pH changes at the root-soi! interface. Soi! Sci Soc Am Proc 35:
301-306
Romheld V (1991) The role of phytosiderophores in acquisition of iron and other micronutri-
ents in graminaceous species: an ecological approach. Plant Soi! 130: 127-134
Rufty TW Jr, Thomas RF, Remmler JL, Campbell WH, Volk RJ (1986) Intercellular localization
of nitrate reductase in roots. Plant Physiol 82: 675-680
Sakai H, Tadano T (1993) Characteristics of response of acid phosphatase secreted by the
roots of several crops to various conditions in the growth media. Soi! Sci Plant Nutr 39:
437-444
Sanders FE (1971) Effect of root and soi! properties on the uptake of nutrients by competing
roots. O Phi! Thesis, Oxford, England
Schaffer GF, Peterson RL (1993) Modifications to clearing methods used in combination with
vital staining of roots colonized with vesicular-arbuscular mycorrhizal fungi. Mycorrhiza
4:29-35
Schaller G, Fischer WR (1985) pH-Ănderungen in der Rhizosphare von Mais und
ErdnuBwurzeln. Z pftanzenernăhr Bodenkd 148: 306-320
Schenk NC (1982) Methods and principles of mycorrhizal research. The American Phytopatho-
logical Society, St Paul, Minnesota
Schonwitz R, Ziegler H (1982) Exudation of water soluble vitamins and some carbohydrates by
intact roots of maize seedlings (Zea mays L.) into a mineral nutrient solution. Z Pftanzen-
physioll07: 7-14
Schwab SM, Menge JA, Leonard RT (1983) Quantitative and qualitative effects of phosphorus on
extracts and exudates of sudangrass roots in relation to vesicular-arbuscular mycorrhiza
formation. Plant Physiol 73: 761-765
Seggewiss B, Jungk A (1988) Einftuss der Kaliumdynamik im wurzelnahen Boden auf die
Magnesiumaufnahme von Pftanzen. Z pftanzenernăhr Bodenkd 151: 91-96
Shepherd T, Oavies HV (1994a) Effect of exogenous amina acids, glucose and citric acid on
the patterns of short-term accumulation and loss of amina acids in the root-zone of sand-
cultured forage rape (Brassica napus L.). Plant Soi! 158: 111-118
Shepherd T, Davies HV (1994b) Patterns of short-term amin o acid accumulation and loss
in the root-zone of liquid cultured forage rape (Brassica napus L.). Plant Soi! 158: 99-
lO9
Siddiqi MY, Glass ADM (1987) Regulation of K+ influx in bariey: evidence for a direct control of
influx by K+ concentrat ion of root cells. J Exp Bot 38: 935-947
Siddiqi MY, Glass ADM, Ruth TI, Rufty TW Jr (1990) Studies of the uptake of nitrate in bariey.
1. Kinetics of 13NO,- influx. Plant Physiol93: 1426-1432
Siddiqi MY, Glass ADM, Ruth TJ (1991) Studies of the uptake of nitrate in bariey. III. Compart-
mentation of N0 3-. J Exp Bot 42: 1455-1463
Siebrecht S, Măck G, Tischner R (1995) Function and contribution of the root tip in the induc-
tion of NO,- uptake along the bariey root axis. J Exp Bot 46: 1669-1676
Sieverding E (1991) Vesicular-arbuscular mycorrhiza management in tropical agriculture.
Deutsche Gesellschaft fUr Technische Zusammenarbeit (GTZ), Eschborn, Germany
Sihna BK, Singh NT (1976) Salt distribution around roots of wheat under different transpira-
tion rates. Plant Soi144: 141-147
Simon L (1996) Phylogeny of the Glomales - deciphering the past to understand the present.
Smart DR, Ferro A, Ritchie K, Bugbee BG (1995) On the use of antibiotics to reduce rhizoplane
microbial populations in root physiology and ecology investigations. Physiol Plant 95:
533-540
Smi!ey RW (1974) Rhizosphere pH as influenced by plants, soi!s and nitrogen fertilizers. Soil Sci
Soc Am Proc 38: 795-799
Smith SE, Read DJ (1997) Mycorrhizal symbiosis, 2nd edn. Academic Press, San Diego
Starkey RL (1931) Some influences of the development of higher plants upon the micro-
organisms in the soi!: IV. Influences of proximity to roots on abundance and activity of
micro-organisms. Soi! Sci 32: 367-393
St-Arnaud M, Hamei C, Vimard B, Caron M, Fortin JA (1997) Inhibition of Fusarium
oxysporum f. sp. dianthi in the non-VAM species Dianthus caryophyllus by co-culture
with Tagetes patula companion plants colonized by Glomus intraradices. Can J Bot 75:
998-lO05
Steyn WJA (1959) Leaf analysis. Errors involved in the preparative phase. J Agric Food Chem 7:
344-348
Swiader JM, Freiji FG (1996) Characterizing nitrate uptake in lettuce using very-sensitive ion
chromatography. J Plant Nutr 19: 15-27
Tadano T, Sakai H (1991) Secretion of acid phosphatase by the roots of several crop species
under phosphorus-deficient conditions. Soil Sci Plant Nutr 37: 129-140
Tagaki S, Nomoto K, Takemoto T (1984) Physiological aspect of muginieic acid, a possible phy-
tosiderophore of graminaceous plants. J Plant Nutr 7: 469-477
Tang C-S, Young C-C (1982) Collection and identification of allelopathic compounds from the
undisturbed root system of Bigalte Limpograss (Hemarthria altissima). Plant Physiol 69:
155-160
Tanner W, Beevers H (1990) Does transpiration have an essential function in long-distance
transport in plants? Plant Cell Environ 13: 745-750
Tarafdar JC, Jungk A (1987) Phosphatase activity in the rhizosphere and its relation to the deple-
tion of soi! organic phosphorus. Biol Fertil Soi!s 3: 199-204
Taylor TN, RemyW, Hass H, Kerp H (1995) Fossil arbuscular mycorrhizae from the eariy Devon-
ian. Mycologia 87: 560-573
Trolldenier G (1988) Visualisation of oxidizing power of rice roots and of possible participation
of bacteria in iron deposition. Z Pflanzenernăhr Bodenkd 151: 117-121
Uren NC (1981) Chemical reduction of an insoluble higher oxide of manganese by plant roots.
J Plant Nutr 4: 64-71
Uren NC, Reisenauer HM (1988) The role of root exudates in nutrient aquisition. Adv Plant Nutr
3: 79-114
Van Vuuren MMI, Robinson D, Griffiths BS (1996) Nutrient inflow and root proliferation during
the exploitation of a temporally and spatially discrete source of nitrogen in soi!. Plant Soil
178: 185-192
Vaughan V, Cheshire MV, Ord BG (1994) Exudation of peroxidase from roots of Festuca rubra
and its effects on exuded phenolic acids. Plant Soil 160: 153-155
von Wiren N, Mori S, Marschner H, Romheld V (1994) Iron inefficiency in maize mutant Ys1
(Zea mays 1. cv Yellow-Stripe) is caused by a defect in uptake of iron phytosiderophores.
Plant Physiol106: 71-77
von Wiren N, Romheld V, Shioiri T, Marschner H (1995) Competition between micro-organisms
and roots of barley and sorghum for iron accumulated in the root apoplasm. New Phytol
130: 511-521
Weber E, Saxena MC, George E, Marschner H (1993) Effect of vesicular-arbuscular mycorrhiza
on vegetative growth and harvest index of chickpea grown in northern Syria. Field Crops
Res 32: 115-128
WeiB J (1991) Ionenchromatographie, 2nd edn. VCH Verlagsgesellschaft, Weinheim
White PJ, Banfield J, Diaz M (1992) Unidirectional Ca2+ fluxes in roots of rye (Secale cereale 1.).
A comparison of excised roots with roots of intact plants. J Exp Bot 43: 1061-1074
White RT Jr, Douthit GE (1985) Use of microwave oven and nitric acid-hydrogen peroxide diges-
tion to prepare botanical materials for elemental analysis. JI Assoc Off Anal Chem 68:
766-769
Wi!kinson HF, Loneragen JF, Quirk JP (1968a) Calcium supply to plant roots. Science 161:
1245-1246
Wilkinson HF, Loneragen JF, Quirk JP (1968b) The movement of zinc to plant roots. Soi! Sci Soc
Am Proc 32: 831-833
Williams RF (1948) The effects of phosphorus supply on the rates of intake of phosphorus and
nitrogen and upon certain aspects of phosphorus metabolism in gramineous plants. Aust
J Sci Res B1: 333-361
Yoneyama T, Komamura K, Kumazawa K (1975) Nitrogen transport in intact corn roots. Soil Sci
CHAPTER 14
Water Uptake
J.E. Fernandez', B.E. Clothier2, and M. van Noordwijk 3
1 Instituto de Recursos Naturales y Agrobiologia de Sevilla, Avenida de Reina Mercedes,
No. 10, Apartado 1052,41080 Sevilla, Spain
2 Environment and Risk Management Group, HortResearch, PBII -030, 5301 Palmerston
North, New Zealand
3 ICRAF,1I CI FOR, PO Box 161, Situ Gede, Sindang Barang Bogor 16680,16001 Bogor,lndonesia
CONTENTS
14.2 Evaluation ofWater Uptake from Changes
in Soil Water Status 465
14.2.1 Methods for Determining Soil Water Content 465
The Gravimetric Method 465
The Neutron Scattering Method 467
Gamma Densiometry 467
Dielectric Methods 468
14.2.1.1 Time Domain Reflectometry (TDR) 468
14.2.1.2 Frequency Domain Method (FD) 469
14.2.2 Methods for Measuring Soi! Water Potential 471
Electrical Resistance Blocks 471
Tensiometers and Microtensiometers 472
Soi! Psychrometers 472
14.2.3 The Nuclear Magnetic Resonance (NMR) Technique 474
14.2.4 Applications to the Study of Water Movement Between Soi!
Layers via the Root System 474
Measurement of Water Efflux and Hydraulic Lift 475
Split-Root Boxes 475
14.3 Evaluation ofWater Uptake from Water Movement in Plants 476
Heat-Pulse 478
Heat Probe 478
Heat Balance 479
Sap-flow Measurements in Individual Roots 479
The Water-Budget-Meter 483
14.4 Measuring Stable Isotopes in Plants to Assess Water Uptake
from Different SoH Depths 483
Deuterium 483
The 18 0/ 16 0 Ratio 485

462 J.E. Fernândez et al.
14.5 Measuring the Ability of Roots to Take Up Water 485

14.5.1 Soil and Root Hydraulic Conductivity 485
The Pressure Probe Techniques 488
14.15.2 Resistance to Flow 489
Principle 489
Methods 490
14.5.3 Root-Shoot Communication 491
14.6 Measurement of Plant Water Status 493
14.6.1 Root Water Potential 493
The Psychrometer Technique 493
The Pressure-Chamber Technique (Sucker Shoots) 495
14.6.2 Stern Diameter Changes 497
14.7 Summary and Conclusions 498
References 499
14.1 Introduction
Uptake of water by the root system is critic al for plant functioning, as it

balances aboveground water losses by transpiration and facilitates movement
towards the roots of nutrients and other chemicaIs. The coupled processes in
the soil-plant-atmosphere continuum are controlled at the plant, the cellular,
or even molecular level (Kramer and Boyer 1995; Steudle and Peterson 1998).
A basic understanding is needed of the process of root water uptake, combin-
ing soil physical and plant physiological perspectives in models of plant behav-
iour. This can be related to the water status in their environment to determine
accurate plot-Ievel soil water balances, evaluate plant adaptation to drought, and
analyse below-ground competition in mixed vegetation systems.
At the most basic level, the soil-plant-atmosphere continuum concept
(Huber 1924; Gradmann 1928; Van den Honert 1948) assumes steady-state flow
and constant resistance. Obviously this is an oversimplification (Kramer and
Boyer 1995), but it serves to remind us of the interrelationships between the
soil, plant, and atmospheric fac tors that have to be taken into account when
determining plant water status. Any analytical description, or model, of the
soil water balance needs to consider many processes that act simultaneously
across a number of different time and spatial scales. These processes include
(Fig. 14.1):
l. Water entry into the soil by either rainfall or irrigation and its separation
into either infiltration or surface run -off. This process of water entry into the
14 Water Uptake 463
Fig. 14.1. Soi! water balance evapo~ /_----- --- -- --" rainfall and
!~:~I~~ ""~.~~. .
for a unit of soi! with
vertical as well as lateral ••• ation
g
inputs and outputs / plant surface (j)
:' water evapo- 1 surface
run-on !f@ uptake ration 0 inftltration ~ run-off
- _-o\:SO~I~:t:;-;;-1---®' -
"~~---(j)-~---~--0~
o
e:: ('D ""1
.>.
Yo
' - - retentlOn 0 top- r::r
1--1.1--- -~---~::~::---l:r~!
-
subsurfa- subsurfa-
-
ce lateral ce lateral
El- ·1 b o'"
jlaw e. = . G) jlaw
::; 1» CD
I:t. 1» SOI water SU -
00
::;o g: retenuon
3
soli ~
(storage)
deep
infiltration ~
' - - - leaching
root zone operates on a seconds-to-minutes time scale throughout the depth

of the root zone, say centimetres to many metres.
2. Infiltration through and drainage from the cascade of soillayers, and/or via
bypass flow through macropores. This internal movement operates on a
seconds-to-hours time scale across distances from millimetres to metres.
3. Evaporation from surface soil which operates on an hours-to-days time scale
across the scale of centimeters.
4. Water uptake across the spatially distributed root system, which operates
on a hour-to-days time scale up to a spatial scale of many (at least up to
50 in some trees) metres. Uptake normally occurs during daytime when the
stomata are open, although capacitance recharge of the plant itself can result
in nocturnal uptake.
5. Root uptake and discharge from and into soil ("hydraulic lift") can occur
through hydrostatic equilibration across the root systems on an hour-to-days
time scale, but mostly at night when plant transpiration is negligible. The
spatial scale is that of the entire root system.
6. Root uptake strategies in response to hormonal signals can occur locally on
an hours-to-weeks time scale.
Better understanding of root uptake can thus be seen to demand knowledge of
the roots themselves, as well as information about the coupled biophysical
processes operating in the soil and the atmosphere.
When the water balance of a plant, or tree, is calculated over a whole
growing season, total water uptake will often be constrained by the amount of
water freely available in the soil. The traditional definition of "available" water
on the basis of a water retention curve as the difference between "field capac-
ity" and "wilting point" may be adequate for the topsoil, but in the deeper layers
of the root zone, not alI of this "available" water might actually be acces sed by
464 J.E. Fernandez et al.
A soil water content B soil water content
rooting--. ----t
depth !; water
.-
i extrac-
1 !ion
, profile
f
wilting field C Transport model:
point capacity
+-----+
available
water
Fig. 14.2. A Definition of "effective rooting depth" for water uptake and "available water" (McK-
eague et al. 1984) In this pragmatic definit ion error 1 cancels error 2, so that the total amount
of water available for the plant per unit soil area can be estimated from the difference in volu-
metric soil water content between field capacity and wilting point, times the "effective rooting
depth"; B and C provide alternative concepts to estimate total water uptake on the basis of the
complete root system and relative water depletion efficiencyas a function on root length density
(de Willigen and van Noordwijk 1987)
plants either in quantity, or at the rate at which it is needed. As illustrated in

Fig. 14.2, a simple way around this conundrum is via the definition of an
"effective rooting depth". This then makes up for the difficulties in defining
"available" water because the "effective rooting depth" is defined in such a way
that the overestimate of possible water uptake above the "effective" rooting
depth equals the underestimate due to lack of complete water uptake below
this depth. This definition (McKeague et al. 1984) couples and may appear to
render useful both the concepts of "available water" and "effective rooting
depth", even though it is not very satisfactory from a rigorous scientific view-
point. Thus the challenge is to quantify better the form of the entire root system,
and to predict better the biophysical functioning of roots in their quest to take
up water and chemicals. To achieve this link between form and function, we
need better descriptions of the root system, and improved comprehension of
the chain of resistances to water flow in the soil-root-shoot-Ieaf-atmosphere
pathway.
Researchers of root system form and function, being aware of the impor-
tance of understanding the processes of water uptake, have developed a con-
siderable number of techniques to improve the measurement and modelling
of water uptake (Fig. 14.3). The objective of this chapter is to give a compre-
hensive and detailed description of the most effective of these techniques. We
14 Water Uptake 465
Table 14.1. Spatial resolution of the methods used for assessing water uptake by roots
Measured variable Resolution l-lOOJlm O.l-lOmm O.Ol-lm
Water content, e Gamma densiometry Gravimetry

Neutron scattering
NMR
TDR
FD
Soil water Microtensiometer Tensiometer
potential,'IIs Psychrometer Resistance
blocks
Rootwater Pressure probe Psychrometer Pressure chamber
potential,'Pr
and resistance, lILp
Flowwithin Sap flow techiques
roots, Jv SpIit-rootbox
first discuss methods for measuring water uptake by roots that can be detected
via changes in the soH's water content or water potential. We then describe
methods for determination of water fluxes in plants, as well as methods to assess
the sources of water absorbed by roots, and determination of the resistance to
flow and uptake. We conclude with methods to quantify root-shoot communi-
cation via plant hormones. These methods alI differ in their temporal and
spatial resolution, and thus they are intended for use for different purposes
(Table 14.1).
14.2 Evaluation of Water Uptake from Changes

in Soil Water Status
A variety of methods can be used to determine changes in both soH water

content and potential for identifying areas of the soH where roots are actively
taking up water. Especially interesting are those suitable for continuously
monitoring changes in soH water content with the required resolution in both
space and time, and with minimum disturbance to soH conditions.
14.2.1 Methods for Determining Soil Water Content
The Gravimetric Method. The water content of soils can be determined from
the moist and dry weights of soH samples. To obtain the moist weight, samples
Fig. 14.3. An apple tree whose root zone has been isolated by a 3 x 3m trench. This isolated
tree can also be covered by a ground level rainout shelter. The tree is instrumented with heat-
pulse sensors of sap fiow measurement in both the trunk and diametrically opposite structural
roots. The soil is instrumented with TDR probes, minirhizotron tubes, and tensiometers. (Cour-
tesy of Peter Spring, HortResearch, New Zealand)
are weighed at the time of sampling, or later if they are kept in sealed contain-
ers in a cold environment. To determine the dry weight, the sample is dried to
a constant weight in an oven at 105 De. The water content per unit soil dry weight
or gravimetric wetness is the ratio of the weight loss by drying, to the weight
of the dried sample. lf bulk density of the soil is known or measured, volumet-
ric soil water content can be calculated by multiplying the gravimetric reading
by the bulk density.
The method does not require expensive instrumentation. However, it is
destructive and time consuming. Errors can be made at every step - sampling,
transporting, drying and weighing. Recommendations about the use of the
technique can be found elsewhere (Hillel1980; American Society of Agronomy
14 Water Uptake 467
1986). Errors can be reduced by increasing the size and number of samples, but
this requires more time and effort, and eventualIy disturbs the experimental
plot.
The Neutron Scattering Method. This is one of the methods most widely used
for determining soil water content. It requires the use of a neutron probe, an
instrument with a source of fast neutrons and a detector of slow neutrons
(Gardner and Kirkham 1952). When the probe is lowered into an access tube
inserted into the soil, the fast neutrons that are emitted collide with various
atomic nuclei in the surrounding medium. Through repeated collisions the
neutrons are deflected and "scattered", and Iose kinetic energy. Their speed
decreases to the characteristic speed for particles at the ambient temperature.
Those slow neutrons are said to be thermalised. The nuclei of hydrogen, being
of similar size, are the most effective in thermalising the fast neutrons, and so
the flux of slow neutrons scattered by the soil is proportional to the soil water
content. The source of fast neutrons is obtained by mixing a radioactive emitter
of alpha particles (radium or americium) with beryllium. The flux of slow neu-
trons is converted to a count rate by an electronic scaler.
The neutron scattering method is less time consuming and not as labori-
ous as the gravimetric method. It is less destructive, and repeated measure-
ments can be made at the same locations, directly surrounding the access tubes.
The main disadvantages are the high cost of the neutron probe and the emis-
sion of neutron and gamma radiation. In addition, the method is not suitable
close to the soil surface zone due to the escape of fast neutrons through the
surface. The effective volume of measurement is a sphere which varies in size
with soil water content. It ranges from a radius of less than 10 cm in a wet soil,
to more than 25 cm in a dry soil (Stone et al. 1955; Holmes 1956). The neutron
probe has to be calibrated, usualIy by making many measurements of soil mois-
ture with the neutron and the gravimetric methods simultaneously. Once cali-
brated, it should work on different soil types, except in unusual soils, like very
heavy clays or in soils with a high boron content. When changing soils, one or
two readings can be obtained gravimetricalIy to see if points falI on the cali-
bration line, to tind out whether a new calibration is required. Different soil
layers may require different calibration lines. A good example of the use of
the neutron probe to in fer root uptake is the work by Rambal (1984). He used
the technique for 7 consecutive years to examine seasonal water extraction to
4.5 m in a Quercus coccifera evergreen scrub system. Different calibration curves
were required for soils with different rock distributions.
Gamma Densiometry. The theory and practice of gamma densiometry were

documented by van Bavel (1959) and Gardner (1986). Briefly, when a source
of gamma radiation is placed in the soil close to a detector, the amount of
468 J.E. Fermindez et al.
radiation passing through the soH decreases in proportion to soH water content,
if bulk density is constant. Simultaneous measurements of bulk density and
soH moisture changes can be made with dual-source gamma-ray scanners, in
which both 137Cs and 24IAm are used (Gardner 1965). Small changes in soH
water content can be detected with spatial resolution and precision that is
difficult to attain using other methods (Baker and van Bavel 1986, 1988). For
example, the effective depth resolution of measurement is about 1 cm, much
better than the neutron probe (Table 14.1). The access tubes for the source and
detector probes must be strictly parallel, which make installation difficult in
certain soHs. The technique is likely to be used less in future, as non-nuclear
methods improve.
Dielectric Methods. Over the last decade a revolution has occurred in the tech-
nology of soH water measurement. The water content and electrolyte concen-
tration of a soH can be accurately determined from the measurement of its
dielectric properties. Two methods can be used, one with measurement in the
frequency domain between 30-3000 MHz (Schmugge et al. 1980), and the other
in the time domain at frequencies above a GHz (Topp and Davis 1985a). These
methods, based on the measurement of the electrical capacitance of the soH,
depend on the fact that water has a much higher dielectric constant than either
air or the dry constituents of soil. The dielectric constant for free water is about
80, and for soH-bound water varies between about 4 and 80, whHe for dry soH
it is about 5 and for air, 1.
14.2.1.1 Time Domain Reflectometry (TDR)
For the time domain method, the most popular instrument uses time domain
reflectometry (TDR). The principle of the technique is well explained by Topp
and Davis (1985a). TDR determinations are based on the propagation of high
frequency (=GHz) electromagnetic waves through the soil. The propagation
constants for electromagnetic waves in the soH depend on the soH properties,
especially its water content and electrical conductivity. In the TDR technique,
a voltage pulse is propagated along a transmis sion line which terminates with
a wave guide inserted in the soil. Wave guides are composed of two or three
parallel metal rods. The signal is reflected from the end of the wave guide
and returns to the TDR receiver. The propagation velocity and the amplitude
of the reflected signal are then recorded. Because the propagation velocity
depends on the dielectric constant of the medium, it is proportional to volu-
metric water content, becoming smaller as water content increases (Fig. 14.4).
The determination of water content is essentially independent of soH texture,
temperature and salt content. However, Herkelrath et al. (1991) found that
calibrations run on soH cores did not compare well with published curves
14 Water Uptake 469
Fig. 14.4. Essential BALUN

TRANSMISSION
~(
components for measurement TOR LINE
of soi! water content by TDR UNIT
(tap) and idealised output ~
Length l.
trace. Electrical conductivity
is calculated from the signa!
travel time (t) between the A
beginning (A) and the end
(B) of the soil transmis sion
TOR
line, and from the refiected PULSE Ţ
V2·VI
signal amplitude (v2-vl). (J)
~ ~
(From: Topp and Davis o I
> I B
1985b)
TIME
developed for mineral soils, probably because of high soil organic matter
content. The authors did not explain why organic soils shift the calibration.
Measurements in dry cracking soils may also be difficult, due to the poor
contact between probes and soil.
The TDR technique has the same order of accuracy and reproducibility as
nuclear methods, and usually does not require site-specific calibration (Wraith
and Baker 1991). Several wave guides can be multiplexed and continuously
monitored by an unattended TDR unit (Baker and Allmaras 1990; Heimovaara
and Bouten 1990). This makes TDR a powerful technique for studying root
water uptake. Wraith and Baker (1991) used an automated TDR system to study
water uptake and reported high spatial and temporal resolution with high
precis ion; standard deviations in water-content measurements were between
0.0008 and 0.001 m 3 m-3• Clothier and Green (1994) used 22 TDR probes around
a kiwifruit vine to monitor the pattern of root water uptake (Fig. 14.5). The ne ar
surface preference of roots for wetted soil can be seen in the lower figure. They
also used a bank of tensiometers and sap ftow measurements in the stern of the
tree, in an example of the combined use of several techniques to understand
and quantify the spatial and temporal pattern of water uptake by roots.
Further developments of the TDR technique will continue to improve the
precis ion and accuracy of measurement and reduce cost.
14.2.1.2 Frequency Domain Method (FD)
The first dielectric soil water content sensors that worked with frequency
domain (FD) were developed in the 1930s (Ferguson 1933; Smith-Rose 1933;
Liick 1964). Modern instruments (Turski and Malicki 1974; Wobschall 1978;
South Norlh
0.0
.....
E
...., -0.5
..c
c..
II
O
-1.0
-2.0 -1.0 0.0 1.0 2.0

0.0
.....
....,
E
-0.5
..c AfJ
D.
II
O
-1.0
-2.0 -1.0 0.0 1.0 2.0

Radial distance (In).
Fig. 14.5. An example of the spatial resolution of measurements by TDR where the ordinate is
soi! depth and the abscissa is the radial distance from the kiwifruit vine. The upper part shows
the average dai!y change in soi! water content around a kiwifruit vine over a 4-week period; the
lower part gives results for a 2-week period following irrigation of the south side (left side); the
rate of water extraction !+.O/!+.t is given in m 3 m-3 day-l. (From Clothier and Green 1994)
Hilhorst et al. 1992) use improvements that make the technique an accurate
method of measuring soil water content, suitable for automatic measurements
in situ (Hilhorst and Dirksen 1994; Dirksen and Hilhorst 1994). The FD method
involves the measurement of capacitance and conductivity at a fixed frequency.
Most commercially available FD sensors have an operating frequency between
10 and 150 MHz. A sinusoidal current is fed through an impedance formed by
two electrodes with the soil as the dielectric. From the amplitude of the voltage
developed across the electrodes and the phase angle between current and
voltage, the dielectric properties of the soil can be calculated. Electrodes can be
plates, rods or rings. The electronic equipment associated with the FD technique
is less expensive than that of TDR, and they are easy to operate. Calibration,
however, is soil specific and non linear (Ben et al. 1987). A description of dif-
ferent instruments using capacitance probes and a comparative study with the
TDR method can be found in Paltineanu and Starr (1997) and Robinson et al.
(1998).
14 Water Uptake 471
Measurements of the dielectric properties of soil in the time dom ain and
the frequency domain were compared by Hilhorst and Dirksen (1994). Despite
the advantages attributed by the authors to the FD technique, TDR instruments
have become more widely used.
14.2.2 Methods for Measuring Soil Water Potential
The measurement of water potential both in the soil and in roots is used for
determining the state of water in both systems. It is the thermodynamic poten-
tial of water that provides the continuous and causallink in the transport chain
from water in the soil, to the plant and to the atmosphere.
The main components of the total soil water potential are the matric poten-
tial, the osmotic potential and the gravitational potential (Kramer and Boyer
1995). The matric potential results from the capillary and adsorptive forces
imposed on water by the soil matrix. The osmotic potential is due to the pres-
ence of solutes in the soil water. The gravitational potential of the soil water is
deterrriined by the elevation of a given point relative to a reference level, usually
the soil surface or the water table. For measuring matric potential - the main
component of the total water potential in the field - resistance blocks and ten-
siometers are used. Total soil water potential can be measured with thermo-
couple psychrometers. Hydraulic gradients can also be determined, with several
tensiometers placed at different depths, so that Darcy's Law can be used to
compute fluxes of water in the soil.
Electrical Resistance Blocks. Blocks of porous materials, such as gypsum,

nylon or fiberglass, containing a pair of electrodes, can be used to measure
matric potential (Bouyoucos and Mick 1940; Colman and Hendrix 1949). The
underlying principle relies on the fact that the electrical resistance of many
materials depends on water content. The resistance of a porous block in soil
decreases as the quantity of soil water absorbed by the block increases. At
equilibrium, the water content of the block is dependent on the matric
potential of surrounding soil, which is therefore proportional to the resistance
of the block. Porous blocks have to be calibrated per individual soils. Measure-
ments are made by an ohmmeter connected to the electrodes of the resistance
block.
Blocks made of gypsum behave better than blocks made of inert materi-
als, since they are less sensitive to variations in the salinity of the soil solution.
However, the solubility of gypsum makes the blocks eventually deteriorate
in the soil. Gypsum blocks are more responsive in the dry range, whereas
porous nylon blocks, because of their larger pore sizes, are more sensitive in
the wet range of soil moisture variation (Hillel 1980). Porous blocks can be
472 }.E. Fermindez et al.
considered a complementary tool to tensiometers, since they work better in

soils drier than -0.05 MPa. Reasonably accurate measurements can be obtained
at matric potentials as low as -2.0 to -3.0MPa. Goltz et al. (1981) described
electronic improvements to make resistance blocks useful at water potentials
as high as -0.03 MPa. Although the porous blocks are not very accurate,
they are frequently used to schedule irrigation, partly because they can be
used for automatic recording of in situ changes in soil suction (Schlub and
Maine 1979).
Tensiometers and Microtensiometers. A tensiometer consists of a water-filled

container with porous cup, usually ceramic, which is inserted into the soil. The
water container is connected to a manometer. The de-aired water within the
cup equilibrates with the soil water through the pores of the ceramic walls. Soil
water, usually under suction, draws water out of the tensiometer to achieve
equilibrium, causing a drop in pressure inside the container which is indicated
by the manometer. The manometer can be a water, or mercury-filled U tube, a
vacuum gauge or an electrical transducer.
Tensiometers may be used at water potentials between O and -0.08 MPa.
Higher suction causes air entry through the cup. For dry areas, as near the soil
surface, however, potentials can often go below -0.08 MPa. The size of the
ceramic cup will determine the sphere of influence within which the soil matric
potential is measured. Several tensiometers are needed for characterising the
distribution of moisture in the soil. Maintenance and measurement operations
are not greatly time-consuming, and Httle training is required.
Vetterlein et al. (1993) described microtensiometers for measuring
variations in soil matric potential in the vicinity of a living root system. They
used ceramic capillaries with a length of 5 mm, a diameter of 1 mm and a wall
thickness of 0.3 mm. The ceramic cells were melted into a glass tube, which
was connected to a pressure transducer via a plexiglass body. Measurements
were taken every 36min throughout a 14-day period. The microtensiometers
were operational at matric potentials between Oand -0.08 MPa, with a standard
error <5% below -0.008 MPa. Temporal resolution was in the range of seconds.
Soil Psychrometers. Soil psychrometers are thermocouple psychrometers

housed in a ceramic cup (Fig. 14.6), which is embedded in the soil. The ther-
mocouple is used to measure the relative humidity inside the cup, which, at
equilibrium, is equal to that of the soil atmosphere. This is accomplished by
using the Peltier effect to condense water on the thermocouple junction within
the ceramic cup. Relative humidity is then determined, in the psychrometric
mode of operation, from the wet-bulb depression of the junction, or, in dew
point mode, from its dew point temperature (see Sect. 14.6.1). Soil psychrome-
14 Water Uptake 473
l - - - - Heol - shrink lubing

epoxy resin
Thermocouple
(Iemperolure meosuremenl)
u++-+-- Pin
-ft+H-- - Teflon spocer
Thermocouple
(psychromeler junclion) - --f'I--=--- Porus ceromic
Teflon plug
Fig. 14.6. Structure of a typica! soi! psychrometer. (From Pearcy et al. 1989)
ters can be used when the potential is below the tensiometer range, but they are
not accurate above -0.2 MPa.
Thermocouple psychrometers have been used to measure soil water poten-
tial by many authors. Savage et al. (1987) described the use, calibration, instal-
lation and maintenance of commercial soil psychrometers, both in disturbed
and undisturbed soil. They listed precautions necessary for accurate measure-
ment of soil water potential in the laboratory, glasshouse and lysimeters. Thus,
direct sun on the aluminium covered lead wires of the soil psychrometers has
to be avoided to achieve an acceptable value of zero offset. In addition, the
screen cage cover cannot be in contact with soil, in order to reduce thermal gra-
dients within the psychrometer to acceptable levels. Differences in temperature
between soil and instrument, as well as within the psychrometer circuitry itself,
can greatly affect the accuracy and precision of measured water potentials.
Brown and Chambers (1987) discussed the magnitude of these errors and
analysed the influence on results of temperature gradients between psych-
rometer and soil. Fonteyn et al. (1987) and Schlesinger et al. (1987) used soil
psychrometers in the field. They found a good cor reIat ion between soil water
potential and predawn plant xylem sap potential.
Soil psychrometers are easily installed, but should be placed parallel to the
soil surface to minimise thermal gradients. They retain calibration characteris-
tics for several years, but salt deposition in saline soils can seriously degrade
their performance.
14.2.3 The Nuclear Magnetic Resonance (NMR) Technique
The principles and other features of proton nuclear magnetic resonance (NMR)
imaging are described in Chapter Il. Here we show main applications to the
study of water distribution and transport in plant root systems, a field in which
the technique has been widely used since the 1980s (Bottomley et al. 1986, 1993;
Brown et al. 1986). The technique is ideal for determining water distribution
around and within roots. The signal intensity in NMR images is most sensitive
to protons with a high degree of molecular-Ievel mobility such as those of
hydrogen in water (Rogers and Bottomley 1987). Bottomley et al. (1986) were
able to analyse uptake and movement of water within Vicia faba plants growing
in pots with different potting media. They were able to monitor the process of
plant wilt and recovery by using a CuS04·5H20 solution with Cu2+ as a tracer
(Cu2+ is paramagnetic) and water application alone to a plant growing in a plot
with dry soil. NMR imaging has been used by Omasa et al. (1985) for monitor-
ing changes in root and soil water content. MacFall et al. (1990, 1991) used it for
observing the development of a water depletion zone around suberized pine
roots.
The major advantage of the technique is that it is non-destructive, as it
employs non-ionising radiation that penetrates the soil media. This has no
known effect on plants (Bottomley et al. 1986). The technique is sensitive to the
presence of ferromagnetic particles occurring naturally in the soil, which can
disturb the magnetic field and thereby dis tort the NMR information. Rogers
and Bottomley (1987) investigated the range of potential applications of NMR
imaging in root systems and assessed the impact of soil magnetic properties
and water content on the results from the use of the technique. They found that
the NMR technique was not useful in soils with ferromagnetic particle contents
greater than about 4% by weight. The equipment is expensive and requires
highly trained technicians. The technique cannot provide images of large root
systems.
14.2.4 Applications to the Study of Water Movement

Between Soil Layers via the Root System
Water efflux from roots into surrounding soil can occur under certain condi-
tions. Water will move out of roots that are permeable to water when the water
potential of the soil falls below the water potential within the root. Hydraulic
continuity via root systems can lead to transfers of water between soil layers,
on the basis of water potential and resistance. If the subsoil is wet and the
surface layers are dry, this process is called "hydraulic lift" (Baker and van Bavel
1986; Caldwell and Richards 1989; Xu and Bland 1993). The reverse process,
14 Water Uptake 475
transfers from wet surface layers to dry subsoil is possible as well and has
recently been observed in Machakos (Kenya; Smith et al. 2000). Although the
total quantities involved in these water transfers may be relatively small, it
can be important in the competition between shallow and deep rooted plants.
Hydraulic lift can re-wet nutrient-rich dry topsoil layers and thus facilitate
nutrient uptake. The reverse process, deep water storage by deep rooted plants
after moderate rainfall which only infiltrates into the topsoil, can increase their
overall resource capture vis-a-vis shallow rooted plants.
Water efflux from roots is a phenomenon that needs to be taken into
account when considering survival and competition within a plant community.
Thus, hydraulic lift can help the plant to withstand a period of drought. It can
even benefit neighbouring plants with shallow root systems. On the other hand,
rapid storage of water in subsoil around the deep roots of trees, beyond the
wetting front, may give trees a competitive edge over rather shallow rooted
plants.
Measurement of Water Efflux and Hydraulic Lift. Water efflux and hydraulic lift
can be determined by detailed measurement of local changes in soil water con-
ditions. Vetterlein and Marschner (1993) used microtensiometers to measure
the transfer of water to upper soillayers from lower layers by hydraulic lift. The
microtensiometer technique was sensitive enough to detect water release by
roots in the top soil. Vetterlein and Marschner (1993) mentioned other attempts
to measure hydraulic lift by using gamma densiometry or psychrometers
(Mooneyet al. 1980; Dirksen and Raats 1985; Baker and van Bavel1988; Saab
and Sharp 1989). Dawson (1993) used standard water relations methods
and stable hydrogen isotope analysis to determine the magnitude and
radial extent of hydraulic lift by sugar maple trees. He observed that ground
water taken up by the trees was redeposited at night into the upper 35 cm of
soil, benefiting neighbouring plants. Direct measurements of the direction and
rate ofwater movement in roots based on heat-pulse methods will be discussed
below.
Split-Root Boxes. Water movements within the root system can be also studied
using split-root boxes. In these experiments, the root zones of plants are divided
by barriers so that water can move from one side to another only through root
systems which bridge the partitions. Different conditions can be imposed on
each side. Split-root experiments were used by Zhang and Kirkham (1995a) to
determine the effect on crop behaviour of having half of the root system under
water and nutrient stress. The authors explain a way of splitting the root systems
of grain sorghum and sunflower. Baker and van Bavel (1986) worked with
Cynodon growing in a split-root box. After creating a large difference in soil
water potential between the two sides of the root system, they observed noc-
Fig. 14.7. A split-root box

for quantifying hydraulic
equilibration of soi! via plant
roots from wet to dry
compartments. (From Baker
and van Bave11986)
14 0.3 m ~ 1.2 m ~ 0.3m ~
turnal transfer through the plant of significant amounts of water from one side
to the other. This was detected by frequent measurements of volumetric water
content in the box with a gamma probe. Overnight movements of water from
wet to dry soil in a split-box (Fig. 14.7) were also observed by Baker and van
Bavel (1988) when the potential gradient between dry and wetted compart-
ments reached 1.0 MPa.
14.3 Evaluation of Water Uptake from Water Movement

in Plants
Some 60 years ago Huber (1932) used heat as a qualitative tracer of sap fiow in
plants, by monitoring the downstream temperature rise following application
of a short pulse of heat. He observed the rate of ascent of sap. Three thermal
tracer techniques have been developed to quantify the volumetric rise of sap;
the heat pulse method, the heat probe, and that of heat balance technique. These
methods have been applied to a wide range of woody and herbaceous plants,
both in the stem and in roots (van Bave11992; Swanson and Perttu 1994; Zhang
14 Water Uptake 477
~~,.
. __
...,
__~~ W EATHER SH~D
!tt-...
STEM AIRGAP
HEATEOZONE , - .... FOAM INSULA1l0N
I
HEATER Ht-+-- .... CORI(
mERMOPll.E
......~+-If--~ THERMOCOUPLE
1UNcnON
'-'~' T'
a
•
-p '
CA
,'1 - - --, B,
I I
I I
I I
I I
1- I
----
--- - J
xl vH
b
q,
~
o
~
q, +-- ---+- q, ~
p ----.
...
o'"
1
.-- p
~
8
!
qv
~
o
Fig_ 14_8_ Principles of heat balance and heat pulse technique for recording water fiow through
roots and sterns. (Frorn Srnith and Allen 1996). a Vertical cross section through a stern heat
balance gauge. The dashed box in a is expanded in b to show wiring details of therrnocouples
in the gauge; copper wires are shown as solid lines and constantan wires as dotted lines. For the
determinat ion of sap fiow, the ternperature differentials tlT., tlTb and tlT, are obtained frorn
rneasurernents of the voltages across AH, BH and CH, respectively. c The heat balance of the
length of stern heated by the gauge, where P is heat applied to the stern, qv is the rate of verti-
cal heat loss by conduction, q, is radial heat loss by conduction and qf is heat uptake by the sap
strearn. d Configurat ion of a single set of heat-pulse probes irnplanted radially into a stern of
radius R at the carnbiurn and h at the heartwood boundary. The upstrearn ternperature sensor
is installed at a distance X u below the heater and the downstrearn sensor at a distance Xci above
the heater. For calculat ing the heat-pulse velocity, the systern rneasures the tirne delay for the
ternperatures at points Xu and Xd to becorne equal
R
1" DOWNSTREAM
h SENSOR PROBE
THERMlSTOR OR
THERMOCOUPLE
HEATER PROBE
THERMlSTOR OR
THERMOCOUPLE
UPSTREAM
SENSOR PROBE
Î Î Î Î
CENTRE HEAR1WOOD SAPWOOD BARK + CAMBIUM
OFSTEM d
Fig. 14.8. (cont.)
and Kirkham 1995b). Smith and Allen (1996) reviewed principles and applica-
tions of both the heat balance and heat pulse methods (Fig. 14.8). Scaling up
from measurements of individual plants to that of a whole stand over a growing
season requires special attention.
Heat-Pulse. In the heat pulse technique, which is best adapted to woody plants,
a line-source heater is inserted radially into the stern, or root (Fig. 14.9). Every
so often, say every 15-30 minutes, a short (1-2 s) pulse of heat is applied by the
heater. Temperature sensors can be located both upstream and downstream of
the heater, as in the case of the compensation method (Swanson and Whitfield
1981); or simply on the downstream side (Cohen et al. 1981). Solution of the
heat ftow equation (MarshaIl1958), with appropriate wound factors (Green and
Clothier 1988), allows the local velocity of sap ftow to be inferred from the mea-
sured time-of-travel of the heat pulse. Sap ftow rate is then obtained by inte-
gration of sap velocity over the cross-sectional area of the stern.
Heat Probe. In the probe technique of Granier (1985) the dissipation of heat
from a heater probe is empirically related to sap ftow rates. Two probes are
inserted radially into the stern or root of a woody plant; one probe is both a
heater and temperature sensor, whereas the other measures the reference
temperature upstream of the heater. From the regularly measured rise in
temperature whilst the heater is supplied with a constant current, an empirical
reIat ion provides the time-varying trace in the local sap velocity.
14 Water Uptake 479
Fig.14.9. Heat-pulse
sensors for both
measurement in the trunk
using thermistors, and a
miniaturised version for use
in roots which uses
thermocouples. Also shown
is the stainless steel heater
probe. (Courtesy of Dr. Brent
Clothier, HortResearch, New
Zealand)
Heat Balance. Unlike the two previous techniques, wherein a tempera ture
change is related to a sap velocity, the heat-balance technique relies on
measuring all-but-one of the components of the thermal balance of a stern or
root section. The basic equation is:
qi = qv + qr + qf> (14.1)
where qi is the heat input to the stern, qv is the vertical heat loss by conduction
in the stern, qr is the radial heat loss and lJiis the heat uptake by the moving sap
stream, the only term of the equation which cannot be measured directly; it
must be found by difference.
In closing the thermal balan ce, the amount of heat carried away by the
ascending sap can be used to determine the volumetric efflux of sap from the
stern, or root section. This technique is well adapted to herbaceous plants or
fruit rachilla (Sakuratani 1981), as well as for larger trunks of woody species
using a different configurat ion (Cermak et al. 1973).
Over the last 25 years, these three techniques have been extensively applied
in the trunks and stems of a wide variety of plants. These data have provided
many insights into the aerial mechanisms controlling the amount of water
transpired by plants, and also into the mechanisms of plant hydraulic
functioning. More recently these techniques have been applied in roots so that
the subterranean functioning of plants has become observable in more detail.
Sap-Flow Measurements in Individual Roots. Green and Clothier (1991) minia-

turised their heat-pulse equipment so that they could observe sap fiow in
25-mm-diameter roots ofkiwifruit. Their results (Green and Clothier 1995) pro-
vided a clearer picture of how vine roots that have dried down the soil can
respond preferentially to localised wetting (Fig. 14.10). In Fig. 14.10, The rise of
sap fiow in Rl 3-4 days after wetting on DOY 84 was thought to have resulted
0.8
0.2
0.0+-~--.-~-,--~-r--r-4-~--.-~~--~-rL
42 49 56 113 70 77 84 91
(b)
iJ
o 1.0
;:
'O
e
-II
.:!!:
c
"ii
~
0.5
0.0+-~--.-~-,--~-r--r-4-~--.--r~--~-rL
42 49 5& &3 70 77 84 91
Dcy of Yecr
Fig. 14.10. Sap fiow in kiwifruit roots measured with the heat-pulse technique: a relative sap
fiow in two roots, Rl and R2, relative to their total, Rr. RlIRr is the top line, and R2/Rr the lawer.
In b the root fiow is expressed as a ratio to that measured in a control vine R3. Here RlIR3 is
the tap line, and R2/R3 the lower. On day 70 the soil surrounding Rl only was wetted and on
day 84 the entire root zone was wetted. (From Green and Clothier 1995)
from growth of new roots, as sap flow in Rl was higher, after rewetting, than
prior to the dry down.
For apples, Green, Clothier, McLeod (1996) reported quite a different
response to localised wetting (Fig. 14.11). The sap flow in Rl shows an imme-
diate and sustained response to the localised wetting of DOY 47. The rapidity
of the response, and the subsequent1y unchanging ratio to stern flow would be
the response expected from Gardner's (1960) hydraulic model of water uptake.
14 Water Uptake 481
Apple-1995
I
......
..r:: ".0
I~
,!
ţ
Rool R1
\\ 0.10 I - ..r::
......
...J i i ...,.
...J
~ ,I
I '\
~
1
O
;: r/ O
;:
2.0 O.OS
E
CII
'OO
Vi It:
a 0.0 0.00
...... /Iem ~ .,
I ".0
[1
i
i~
i A ." li 0.10
.......
"i
.l ) \
..r:: I \ROOI R2
• ..r::
......
...J \ ~. I
\
...J
.......
I
- -
~ ~
~ o
;:
o.os
-
2.0
E o
CII o
1/) It:
0.0 0.00
b ..6 <4a 50
Doy of Yeor
Fig. 14.11. Response of sap fiow in two apple roots to localised wetting around R1 at the time
indicated by the arrow in the top diagram (from Green et al. 1996). a The dotted line is the sap
fiow measured in the trunk and the ordinate scale is on the left. The sap flow in root R1 is given
as the solid line and the fiow can be seen to rise immediately following the irrigation on Day 47.
The ordinate for root fiow is on the right. b Again stern sap flow is given as the dotted line, and
here the sap flow in root R2 is given as the solid line
In the case of apples, the root form is one of high occupancy in the explored
zone, and so it would appear that with such a root system form, apple uptake
functioning does not rely on root growth. Kiwifruit root system form, in com-
parison, has Iow occupancy in the explored zone, so that uptake functioning
relies in part on the growth of new roots, much as Nobel et al. (1990) found with
desert succulents.
Using the heat-probe technique, Cabibel and Do (1991) observed the
diurnal trace of sap flow in apple roots that were either drawing water from a
zone of localised wetting, or from drier soil. They found sap flow in the dry root
tended to be out of phase with that in the trunk such that flow actualIy peaked
at night, and they attributed this to be a result of nocturnal recharge of the tree's
capacitance.
Using the heat-balance technique on the roots of the African tree Croton
melagocarpus, Ong and Khan (1993) were able to examine, by branch pruning
and root amputation, the ability of the root system to function when the soil's
water is depleted, or when the root system is damaged. A commercialIy
available brand of heat balance equipment (Dynamax Inc., Houston, Texas,
USA) was used to monitor sap flow in 20-30 mm diameter roots of Eucalyptus
grandis (Lightbody et al. 1994). By sequential root severance, they noted that
other remaining roots were able to increase their flow to accommodate the flow
required by the tree. Thus flow in individual roots and shoots of perennials can
be quite decoupled, whereas this might not be the case in annuals.
For the sclerophyllous olive, Moreno et al. (1996) used the heat pulse tech-
nique to observe the inability of near-surface roots to recover immediately from
emboli caused by water stress, despite a thorough but single wetting of the soil
around the once very-dry root.
By extension into roots of heat tracer techniques of sap flow monitor ing,
new observations are being made of the link between root form and function-
ing (Fig. 14.12). This includes details of root response to localised wetting, the
nocturnal recharge of a plant's capacitance, the respective roles of near-surface
roots and roots deeper in the profile, the impact of emboli, and the fraction of
the plant's uptake functioning that can be carried out by just some fraction of
the root system. Further use of heat-tracer techniques will be useful in provid-
ing modellers with data that have hitherto been unavailable. It is important to
realise that water flow in any part of a stern or root can be in alI directions, and
deep roots may pump water out of subsoil during droughts, but may also leak
water taken up after rainfall has rewetted the surface layers.
Fig. 14.12. A set of three

instrumented apple tree
roots. Each of the roots dips
away from the trunk at a
different angle, so that
comparison between the
measured fiows can be used
to indicate the depthwise
uptake strategy by the root
system. (Courtesy of Dr.
Brent Clothier,
HortResearch, New Zealand)
14 Water Uptake 483
Fig. 14.13. Schematic design of water- Main flow

budget meter: 1 control unit with attached -+
-
inner and outer containers mounted on a
balan ce; 2 gas exchange cuvette for
determining co, exchange and
transpirat ion; 3 steel cage connecting control
unit with platform of the balance; 4 outer
container on the pan of the balance with oii
seal to prevent evaporat ion to ambient air; 5
air-tight seals fixing the plant to the control
unit and gas exchange cuvette; wind and
radiation shields not shown. (From Flach et
al. 1995)
8alance
The Water-Budget-Meter. For detailed laboratory-scale experiments, Flach et al.

(1995) used a improved potometric water-budget-meter (WBM; Fig. 14.13),
based on a system described by Eller et al. (1993). Plants are mounted in a gas
exchange cuvette where CO 2 exchange and transpirat ion can be measured. The
roots are bathed in water in a container on a balance which is used to deter-
mine water uptake. Water uptake is the weight change of the container. They
reported an error of less than ±30/0 for water uptake, if a time period of more
than 1 day was considered. For shorter periods (6-24h) the error was larger,
but stilliess than ±60/0. This improved WBM is able to quantify water uptake of
a whole plant during time periods of several weeks. It will be a useful instru-
ment for detailed and controlled studies of the processes controlling root water
uptake.
14.4 Measuring Sta bie Isotopes in Plants

to Assess Water Uptake from Different Depths
The principles underlying the use of stable isotopes for understanding of water
uptake processes are explained in Chapter 12. In this section we only mention
illustrative applications of these techniques that can be used for assessing the
sources of water absorbed by roots.
DeuterÎum. The isotopic composition of a sample is usually expressed in terms

of delta units relative to some standard (Friedman and O'Neil 1978):
(14.2)
where OXstd is the isotopic ratio in delta units relative to the standard, and
R sam and R std are the isotope abundance ratios of the sample and standard
respectively. Water is the only hydrogen source for a plant under the natural
conditions of rainfall. Water absorbed by roots can either come from pre-
cipitation during the growing season, or from stored soil water, or from the
groundwater. The values of OD, the deuterium/hydrogen ratio of the water
from those sources can be quite different. This difference arises from the fact
that in temperate areas, with a continental elimate, OD values of precipitation
commonly show a seasonal cyele. High OD values occur in summer, with
lower OD values in the winter (Dansgaard 1964; Gat 1980). Thus, the OD values
of rain falling during the summer growing season are generally different
from those of stored groundwater. Flanagan et al. (1992) used measurements
of the D/H ratio in stern water to determine the relative uptake of summer
precipitation by four co-occurring plant species in southern Utah. They found
that it was possible to determine the uptake of summer precipitation by mea-
suring the stable isotopic composition of stern water (Flanagan and Ehleringer
1991). Since no isotopic fractionation occurs during water uptake by plant
roots (Dawson and Ehleringer 1991), the isotopic composition of water in
roots and stems reflects that of water taken up by roots. White et al. (1985)
carried out measurements of D/H ratios of tree sap to determine the source of
water. They took sap samples from the trunk around noon, when sap flow
rate was high. Wood samples were taken from the trunk using an increment
borer and the sap drawn into capillary tubes. They demonstrated that the OD
values of sap could be used to quantify the relative contributions of sum-
mertime rainfall and groundwater. Dawson and Pate (1996) were able to
identify the sources of the water taken up by the lateral and tap roots belong-
ing to several woody shrubs and trees. They did this by studying the seasonal
changes in the OD of xylem water within different parts of the root system
and lower shoots.
Dalton (1988) studied water extracted by the roots of tomato plants. He
explained that water will be either H20, characterised by water when both
hydrogen atoms are the isotope lHI> or HDO (D for deuterium) when one hydro-
gen atom is the isotope 2Hl' The vapour pressure of the lighter H20 molecule is
slightly greater than that of the DHO molecule. Therefore, vapour phase water
in equilibrium with liquid phase water will have a slightly lower ratio HDOIH 20
than that of the liquid phase. lf there is no vapour component to root water
extraction, then the proportion of HDO in the residual soil water will remain
constant. This experimental technique and isotopic ratio measurements were
shown to be useful for determining the vapour component of water uptake by
the plant roots.
14 Water Uptake 485
The 180i6o Ratia. Measurements of 18 0/ 160 ratios in samples of water drawn

from the xylem can also be used for determining whether the water absorbed
by roots is coming from either surface or deep waters. Allison and Hughes
(1983) analysed isotope ratios of 18 0 and D in soil profiles down to 15 m depth.
This was carried out in are as of Eucalyptus, and Eucalyptus are as that had been
converted to wheat. During the rainy season, the soil surface of the cropped land
showed an increase of the isotopic ratio, as compared with the measurements
taken dur ing summer. This indicates a degree of infiltration of the precipitat ion
that possesses a higher 8D and 8 18 0. Such an increase was not observed in the
soils with Eucalyptus, indicat ing negligible infiltration since most of the pre-
cipitation has been intercepted by the tree canopy.
The 18 0/ 16 0 composition of water at equilibrium is known. The 180/ 160 com-
position of water can be determined by equilibration with a small volume of
CO 2, followed by repeated cryogenic distillation to free the CO 2 which has
reached isotopic equilibrium (Epstein and Mayeda 1953; Compston and Epstein
1958). For very small samples, the 18 0/ 160 composition can be found by a
reaction with guanidine hydrochloride to produce CO 2 (Dugan et al. 1985; Wong
et al. 1987).
Smith et al. (1996, 1997) found a clear difference in the 18 0/ 16 0 isotopic ratio
between late and early rainfall. In the early season rain, the rain results from
the northward advance of the inter tropical convergence zone (ITCZ), and
during the later part of the growing season in the Sahelian region of Africa, the
rain is due to the southward return of the ITCZ. This isotopic discrimination
can be used to distinguish between surface and deep water uptake by the roots
of Sahelian trees.
14.5 Measuring the Ability of Roots to Take Up Water
Water uptake by roots can be assessed with the methods described so far. In
this section, we refer to methods for studying the ability of roots to take up
water, and the role of root-shoot signalling mechanisms which lead to stomatal
closure and, ultimately, to water uptake.
14.5.1 Soil and Root Hydraulic Conductivity
The water status of a plant is largely determined by the balance of transpira-

tion and water uptake from the soil. Water uptake depends on the soil water
status, root distribution and activity in the soil, and the hydraulic properties
along the pathway of water movement. The four main components of the soil-
root-shoot pathway are: (1) the soil, (2) the root-soil interface, (3) entrance to
the xylem in the root, and (4) the axial pathway from root to shoot in the xylem
(Huang and Nobel 1994). The conductance of this pathway is not independent
of rate of flow or soil water content, and the relative importance of the four main
components varies. Changes in the conductivity of the soil-root-shoot pathway
can be due to:
l. Reduction in soil water conductivity as the soil dries out,
2. Changes in continuity of the soil-root pathway by air gaps and reduced
root-soil contact,
3. Changes in osmotic potential outside of the root,
4. Changes in true permeability of the cell walls along the pathway,
5. Changes in osmotic potential inside the root,
6. Changes in the continuity of the axial pathway by embolism.
Different perspectives exist on the relative importance of these processes (Sands
et al. 1982; Weatherley 1982; Passioura 1988; Riidinger et al. 1994; Stirzaker and
Passioura 1996), but there is now general agreement that past conclusions on
the importance of (4) ignored a number of alternative explanations. When the
soil's water content is high, water movement from the soil to the shoots depends
mainly on the hydraulic conductivity of the root itself (3-5 and possibly 6;
Blizzard and Boyer 1980; Passioura 1988). As the soil dries, the conductance of
the soil-root interface (2) can become limiting for water movement, due to root
shrinkage (Taylor and Willatt 1983; Fernândez et al. 1992), but possibly also
because of increases in the osmotic potential of the soil solution (3; Stirzaker
and Passioura 1996). At lower soil water contents, the soil conductivity (1)
becomes the primary limitat ion on water movement in the root-soil pathway.
See Box 14.1 for the components related to the permeability of roots.
A range of techniques is available for determining Lp, Lr and Kh' usually
based on measuring the volume of water flowing through a root that is exposed
to a known difference in water potential, by varying hydrostatic or osmotic
pressure (Huang and Nobel 1994). L p has been determined for individual roots
by measuring Iv induced by applying a partial vacuum to the distal end of an
excised root (Nobel et al. 1990), but also by applying external pressure to the
whole root system. Typical results for Lp' Lr and Kh' as measured by different
authors in different species are reviewed by Huang and Nobel (1994). They show
how hydraulic conductivity changes with plant stage and environmental
conditions. From all the techniques, the root pressure probe technique seems
to be the most widely accepted method to determine Lp' Lr and Kh (Sect. 14.5.1).
The relative importance of the axial and radial component of Lp varies with
plant species and the developmental stage of the root (Huang and Nobel 1994).
Rowse and Goodman (1981) studied water movement in broad bean roots. They
analysed the relative effects ofaxial resistance, calculated from the xylem
dimensions, and radial resistance calculated using a potometer developed to
14 Water Uptake 487
BOX 14.1. The Permeability of Roots

Water from the surrounding soil moves radially into the root and axially
along the xylem. Therefore, the permeability of roots to water depends
on the root hydraulic conductivity, Lp (ms- I Pa- I ) which can be defined as
(Moreshet and Huck 1991; Nobel 1991):
Lp = J (~su'face - ~xylem ),
y / (14.3)
where Jv (m 3 m-2 S- I) is the volumetric flux density of water into the root,
'Vsurfoce (Pa) is the water potential at the root surface, and 'Vxylem is the
water potential of the root xylem. Also,
Lp = Qv/(I1P A), ( 14.4)
where A (m 2) is the root surface area, Q v is the volumetric flow rate of
water (m 3 S- l), and I1P (Pa) is the pressure drop along the root. Lp can be
partitioned into two components: radial conductivity (L r ) for water
movement into the roots and up to the xylem, and axial conductance (K,.)
for water movement along the root xylem (Dryden and van Alfen 1983;
Nobel et al. 1990). For a root segment of length L (m), K" (m 4 S- I Pa- 1) can
be expressed as:
K" = (L Qy)/ I1P. (1 4.5)

The radial conductivity Lr (ms- 1 Pa- 1) averaged over the root segment
can be calculated by incorporating Lp' K,,, L, and the root radius Ro (m)
into the model of Landsberg and Fowkes (1978):
Lr = (L p a L)/ tanh(a L), (14.6)
where a (m- I ) equals (2 n Ro L,/ Kh)ln, which represents the inverse of the
length along the root xylem between which the pressure drops to a half
(Landsberg and Fowkes 1978; Frensch and Steudle 1989; Alm et al. 1992).
measure the rates of water uptake by small portions of an intact root system.
They found that axial resistance was not likely to be important. This conclusion
may apply to most dicotyledonous plants with secondary development of xylem
transport capacity; in monocots without the capacity to adjust xylem dimen-
sions, however, axial resistance may dominate L p in well-branched root systems.
The continuity of the axial pathway may be dramatically reduced in alI plants,
however, by embolism.
Sperry and Tyree (1990) observed that embolism caused by water stress
reduced hydraulic conductivity in roots of Picea rubens. The appearance of air
embolism in the xylem, as the soil dries, caused a marked reduction of Kh. The
stereo
seol
Intoct
root system ---l::-+--rfifţ~ circuloting
nutrient
quortz sond solution
--'"""'...........-pump
L.!::.=~ strip-chort
recorder
Fig. 14.14. Device for measuring root pressure and root pressure relaxation on spruce roots;
root pressure was monitored by a pressure transducer connected to an amplifier and display;
movements of the oii water meniscus are recorded in a measuring capillary (stereo microscope);
hydrostatic pressure relaxations are measured by moving the micrometer screw rapidly into and
out of the pressure probe. (From Riidinger et al. 1994)
extent of air embolism can be determined by the ratio ofaxial conductance

measured on segments before, and after, pressurisation in solution to remove
the emboli (Sperry et al. 1994; Magnani and Borghetti 1995). Xylem at the junc-
tions between lateral roots and main roots, and between main roots and stems,
differs from that elsewhere in the root system. Embolism occurs more easily at
the junctions.
The Pressure Probe Techniques. The cell pressure probe technique has been
adapted for use on excised roots (Steudle and Jeschke 1983). The root pressure
probe is described by Hallgren et al. (1994), Melchior and Steudle (1994), and
Steudle (1994). Root pressure is measured using a small manometer tightly fixed
to the cut end of excised roots (Fig. 14.14). Excised roots exude xylem sap
because of the active accumulation of solutes in the xylem. When exudation is
stopped by attaching a "root manometer", the system reaches a root pressure
which can be recorded continuously. Half of the interior of the root pressure
probe is filled with silicone oiI and the other half with water, so a meniscus
is formed. The position of the meniscus is monitored with the aid of a stereo-
microscope. Volume changes can be induced in the measuring system, being
responses of root pressure monitored. Measurements begin when a steady root
14 Water Uptake 489
pressure is attained. Changes in pressure with time can be converted into water
flows (Melchior and Steudle 1993). Hydraulic conductivity can be calculated
from analysis of the pressurelfiow relations (Peterson and Steudle 1993;
Peterson et al. 1993). Riidinger et al. (1994) used the root pressure probe to
determine the hydraulic and osmotic properties of a spruce root.
Root pressures can be measured on root tips, segments, or whole root
systems. In a single root tip, measurements are usually possible for 1 to 2 days.
In root systems, measurements have been performed for up to 10 days (Steudle
1994). Measurements on excised roots may not, however, be representative of
those of intact plants.
Pressure in individual xylem vessels of intact plants can be directly mea-
sured by the xylem pressure probe, described in detail by Balling and
Zimmermann (1990) and Zimmerman et al. (1994). Zhu et al. (1995) used this
technique to measure hydraulic properties in roots of maize. This technique
allowed, for the first time in intact plants, the direct measurement of diurnal
and seasonal changes in xylem pressure, plus xylem flow and, in principle, the
solute composition of the xylem sap at the single vessel level. The xylem
pressure probe is filled with "denucleated" water and incorporates a pressure
transducer. A microcapillary probe is advanced slowly through xylem tissue
until the transducer registers a pressure below atmospheric. In order to verify
that the probe is present in a vessel, the microcapillary is loaded within a dye
solution prior to insertion, so that the penetrated vessel can be seen in cross-
sections after the removal of the probe (Benkert et al. 1991). Model experiments
have shown that the probe can accurately read pressure values down to at least
about -1 MPa (Balling et al. 1988; Zimmermann et al. 1994).
Zimmermann et al. (1995) stated that single measurements with the xylem
pressure probe indicate that the xylem tension in the leaves of intact, transpir-
ing plants is often much smaller than that predicted for transpiration-driven
water ascent through continuous water columns. Studies with the xylem pres-
sure probe, together with other evidence, indicates that different mechanisms
need to be considered to explain water movement within plants.
14.15.2 Resistance to Flow
Principle. The relation between applied hydrostatic pressure and flow out of
decapitated root systems is non-linear (Fig. 14.15). This non-linearity was
initially interpreted as a change in the inherent conductance of roots with
increasing flow rates. Fiscus (1975) and Dalton et al. (1975) showed that the
non-linearity can be explained by the interaction of osmotic and hydrostatic
pressures, at a constant hydraulic conductance of the roots. In fact, the osmotic
A measurement B
pipette _ _ I
I
I
coUection of -.> I
air bubbles .:i I
ti!
""
pot with ""
"
decapitated
sample Applied pressure
plant
coUection
regulating vaI ve
and manometer
Fig. 14.1 S. A Pressure bomb design to measure the rate of water flow through excised
root systems at different external hydrostatic pressures. B Typical result, showing non-linear
relationship of flow rate and applied pressure. (From de Willigen and van Noordwijk 1987)
pressure component changes with changing flow rates, depending on the solute
reflection coefficient eTr of the membrane. The models were initially based on a
single membrane in a well-stirred solution, but can be extended to a membrane
in an unstirred solution with salt accumulation in front of the membrane, coun-
teracted by diffusion away from the root surface (Fiscus 1977; Passioura 1984;
de Willigen and van Noordwijk 1987). Recently, Steudle and Peterson (1998)
described the composite transport model, to explain how external fac tors may
influence water passage across roots, and how plants optimize water uptake
according to demands from the shoot.
Methods. Rieger (1995) measured the hydraulic conductivity of the roots of

fruit trees using the technique of Rieger and Motisi (1990). The canopy of each
plant was enclosed in a gas exchange chamber, with the root container and one
basal side shoot left outside the chamber. The container was then placed in
1 cm distilled water to ensure that the soil water potential was virtually zero.
Leaves on the bas al shoot outside the chamber were wrapped in parafilm and
foiI, and allowed to equilibrate with the water potential of the stern base. At
steady-state, the water potential gradient across the root system was assumed
equal in magnitude, but opposite in sign, to the water potential of the wrapped
leaf. Total water flow through the plant was measured by pumping chamber air
through a desiccant column at a rate that maintained a constant dewpoint. A
regression equation was developed for water potential gradient (y) versus water
14 Water Uptake 491
flow on a root length basis (x) from the two points obtained for each plant,
assuming linearity. The y-intercept of the regres sion gives the minimum gradi-
ent required for flow through the root system (Passioura and Munns 1984;
Rieger and Motisi 1990), and the reciprocal of the slope gives L p• With this
technique, Lp measurements are time consuming.
Measurements of the flow rate through decapitated root systems can be
made for single roots or whole root systems in permeable pots, by enclosing the
roots in a pressure chamber with the stern attached to a pipette (Fig. 14.15). At
low applied pressures (around 0.1 MPa) the osmotic component is not negligible
and samples of the exudate can be collected for analysis. At high applied pres-
sures (around O.5MPa) the osmotic component is negligible, but physical
damage to the roots may occur. If root systems are used which were grown under
incomplete aeration and have created cortical air spaces in response to this, a sub-
stantial volume of air may bubble out of the root, as the cut end of the stern
includes the cortex. The air bubbles can disturb the measurement of flow rates,
but their escape from the root can also cause physical damage to the root. A
sequence oflow and high pressures should be used, including zero-pressures to
check for root leakiness under the pressure of the water column in the pipette.
The method can be applied with large potted plants, provided the dimen-
sions of the pressure chamber are suitable. For roots in their natural habitat,
however, a revers al of the direction of flow is needed. Tyree et al. (1994, 1995)
developed a novel method of measuring hydraulic conductance of tree root
systems using a high-pressure flow meter in the laboratory and field. They mea-
sured the amount of water flowing into root systems through a cut end on the
proximal side (either close to the tree stern or at some distance away) at a range
of pressures. By rapidly testing a range of pressures, total water flow during the
measurements is small and therefore osmotic adjustments or changes in soH
water content around the roots are probably negligible. By pressuring the roots
internally, root-soH contact may be improved. At pressures above 0.1 MPa a
linear response of flow Iv to applied pressure ~P can be obtained, from which
L p for the whole root system can be derived if total root surface area is known.
The latter will need destructive measurement on the root after completing mea-
surements, and/or the use of allometric equations relating tree root diameter
to total root surface area. When the method is applied to dead roots very high
flow rates can be obtained, so a check on the integrity of the root system is
needed, especially for "oudiers".
14.5.3 Root-Shoot Communication
Tardieu and Davies (1993) consider that the root-shoot signalling mechanism
acting over the stomatal conductance of plants in drying soH is better explained
by considering the integrated effect of hydraulic and chemical signals, and

not only chemical signals or water relations alone. They studied the ABA con-
centration in the xylem sap. ABA was synthesized by dehydrating the roots
(Cornish and Zeevaart 1985). This synthesis can take place in different root
tissues (Hartung and Davies 1991). Synthesis is roughly proportional to the root
water potential. Therefore, the "root message" in drying soil should be linked to
the water status of the root system. The authors mention different hypothesis
for root messaging.
Ternesi et al. (1994) analysed root-shoot signalling in sunflower. They
collected the sap of container-grown sunflower plants by cutting the shoot of
the plant at 2.0 cm above the beginning of the root, and placing a latex tube
on the top. Sap was collected in test tubes during 6 hand immediately frozen
and stored at -20 ac until required. The ABA content in collected sap was
measured by using an ABA-specific monoclonal antibody (Ephyscience EP120,
Phytoscience, France) using the protocol suggested by the manufacturer.
Results suggested the synthesis of a chemical signal in roots of plants under
mechanical stress - a seven times higher ABA concentrat ion in confined
roots than in roots of control plants -, which can induce the inhibition of plant
growth.
Jackson (1994) mentioned the difficulties of correct measurement of the
amounts of hormones transported from root to shoot by the transpiration
stream. The problem is that sap flow rates have a marked diluting influence on
hormone concentration. If the sap taken for analysis flows at a slower rate than
in the intact plants, which is usual, estimates of concentration present in the
xylem sap of intact plants will be in error. Thus, variable sap flows mean that
concentration alone is not a reliable guide to the amounts moving into the
shoot. What is needed is a delivery rate per unit of time. Such a value is obtained
easily by multiplying concentration with the sap flow rate. The most appropri-
ate concentration measurements will be those made in sap flowing from the
roots at rates as close as possible to the transpiration rate. Jackson (1994) men-
tioned that in pot-grown plants, the latter can be obtained gravimetrically over
a few minutes immediately before the shoot is removed, or by using similar
whole plants set aside for this purpose. One alternative approach has been to
expel small amounts of sap from cut veins of leaves after pressurizing the roots
of intact plants (Schurr et al. 1992). A second alternative method has been to
sample the first few drops of sap issuing from the cut surface of the detopped
root system (Zhang and Davies 1990), though the samples can be contaminated
with ABA, sucrose, and other substances as a consequence of compression
forces exerted on the stern when tubing is applied to collect the sap (Else et al.
1994).
Else et al. (1995) mentioned the difficulties in determining whether root
signals are hydraulic or chemical. They observed that ABA, ethylene, and the
14 Water Uptake 493
ethylene precursor ACC increased in shoots of tlooded plants. Trejo et al. (1995)
determined ABA concentration in leaves of Phaseolus acutifolius. Sampled
leaves were frozen in liquid nitrogen and freeze-dried. Bulk leaf ABA concen-
tration was determined using a radioimmunoassay. The monoclonal antibody
AFR MAC 62 was used, which is specific for (+)-ABA. Samples were extracted
using different ratios of extraction (leaf dry weight: solvent) depending on the
concentration of the ABA fed (1 :65-1: 150), and further dilutions were made
when found necessary.
14.6 Measurement of Plant Water Status
Measurements of the plant water status at a certain time could give us valuable
information for estimating water uptake rates by the roots. Techniques for
measuring stern diameter changes are also discussed in this section, as a more
direct way to estimate water uptake.
14.6.1 Root Water Potential
Root water potential can be indirectly calculated (Jones 1983) from measure-
ments of leaf water potential made using the pressure-chamber technique
(Scholander et al. 1965). For more direct determination, thermocouple psy-
chrometers, the sucker method or the root pressure probe can be used. However,
these are alI techniques that, by virtue of their delicate nature, are limited to use
in the laboratory.
The Psychrometer Technique. Thermocouple psychrometers have been widely

used for measuring root water potential. See Box 14.2 for details on the
techniques. In situ psychrometers were developed in the 1970s for leaves
(Boyer 1972; Campbell and Campbell 1974), stems (Michel 1977) and roots
(Fiscus 1972). The last is easy to install on roots of various diameters, just by
changing sleeves and bottom caps. By placing soil psychrometers close to root
psychrometers, the potential gradient from the bulk soil to the root surface can
be estimated.
Errors associated with the use of thermocouple psychrometers are caused
by vapour pressure disequilibria; the resistance to vapour diffusion from the
sample to the chamber atmosphere; thermal gradients and instability caused by
tluctuations in environmental temperature; respiratory heat production from
the sample; and by solar heating (Koide et al. 1989). Changes in sample water
potential can also result from excision of the sample and growth during the
equilibration period. This causes a reduction in water potential since the turgor
BOX 14.2. Thermocouple Psychrometers

After placing a tissue sample in an enclosed psychrometer chamber held
at a constant temperature, the water in the sample equilibrates with the
chamber atmosphere. The value of the sample water potential can then
be determined by measuring the equilibrium relative humidity (h) of this
atmosphere (Turner 1981):
RT ln(h)
1JI = , (14.7)
Vw
where R (J K- 1 mot 1) is the universal gas constant, T is the temperature (K)
and Vw (m l mol- 1) is the partial molar volume of water. Since a water poten-
tial of - 2.0 MPa represents a wet bulb depression of about 0.25 °C, facilities
for accurate temperature control and measurement have to be available.
Slow drifts in temperature can be tolerated, so long as the reference tem-
perature is known at the time when the sample is measured. In practice,
the sealed chambers are usually placed in an accurate water bath in a con-
stant temperature room controlled 1 or 2 °C below that of the water bath.
There are two main techniques to measure h in the psychrometer chamber.
In the first, the psychrometric mode, pure water is allowed to evaporate
into the chamber atmosphere from the surface of a thermocouple junction.
The resulting wet-bulb depression at the thermocouple junction is related
tohby
h =ew- ay(T - Tw), (14.8)

e
where a (Pa) is the atmospheric pressure, y(K- 1) is the psychrometric
constant, T the dry-bulb temperature of the psychrometer, T w is the
wet-bulb temperature, e (Pa) is the saturation vapour pressure at T,
and ew (Pa) is the saturation vapour pressure at T w(List 1968). A
procedure based on the Peltier effect is commonly used for placement
of the pure water on the thermocouple junction. A cooling current
reduces the temperature of the junction below the dew point of the
chamber atmosphere, so water is condensed on it (Spanner 1951).
With the second technique, called dewpoint hygrometry, h is
determined by measuring inside the psychrometer chamber the dew-
point temperature (Neumann and ThurtellI972; Campbell et al. 1973).
Dew-point hygrometers hold the thermocouple junction at the dewpoint
temperature using a pulsed cooling current. An advantage of this
technique is that the dew-point depression is larger than the wet-bulb
depression; the thermocouple signal is therefore larger. In addition, the
signal is stable, as there is no net movement of water at the dew point, so
that vapour equilibrium in the chamber is not disturbed by measurement.
14 Water Uptake 495
pressure will decline in the absence of a water supply. To minimise errors,

psychrometers must be kept clean and the sample surface area per unit chamber
volume should me maximised. Errors due to thermal gradients and instability
can be reduced by assuring good contact between the sample and the chamber
and by using chamber materials with a high thermal conductivity. Errors due
to changes in water potential resulting from the excision and growth of the
sample in the chamber can be reduced by working at low temperatures or with
mature tissue.
Brown and Collins (1980) noted errors caused by temperature gradients
after using different types of psychrometers to measure water potential, and
they developed a new screen-caged psychrometer. Oosterhuis (1987) described
a method to measure components of the root water potential in cotton plants
grown in pots. He used screen-caged thermocouple psychrometers mounted in
sample chambers. Water potential was determined after a 4-h equilibration time
in a 25°C isothermal water bath. Osmotic potential was measured by freezing
the sample-chamber in liquid nitrogen for 3 min, and then allowing it to thaw
at room temperature for 30 min, before reequilibration in the water bath for
4 h. Pressure potential was obtained by subtraction.
Simonneau and Habib (1991) measured root water potential on small root
samples from peach trees with psychrometers. For calibration, they used strips
of filter paper soaked in standard sodium chloride solutions (Brown and Collins
1980); calibrat ion curves were then obtained for temperatures between 20 and
35°C. Schildwacht (1988) avoided errors resulting from the nutrient solution
adhering to roots by centrifuging the root sample in wire mesh on a plastic tube
before transferring it to the psychrometer chamber. Bennet and Cortes (1985),
among others, stated that the volume of sample has to be large enough to limit
the effect of water adsorption by the chamber.
Yamauchi et al. (1995) used a modification of the method proposed by
Fiscus (1972; Fig. 14.16) to measure xylem water potential in cotton taproots.
They constructed teflon chambers so that screen-caged thermocouple
psychrometers could be attached to the taproot at points 0.4 and 0.6 m apart.
A screen-caged thermocouple psychrometer was buried near each root
psychrometer to measure the soil water potential. Subsequent changes in
root and soil water potential were measured. Temperature fluctuations
were minimized by isolation of the system. Psychrometer readings were
corrected for any observed temperature difference from the psychrometer
calibration temperature at the time of the reading. Axial resistance was
also measured by root psychrometers attached to roots growing in a device
that ensured all the water flowed only through the taproot to which the
psychrometer was attached.
The Pressure-Chamber Technique (Sucker Shoots). When leaves are not tran-
spiring, the pressure chamber (Scholander et al. 1965) can be used to determine
Polystyrene Insulol ion

(17mm in thicknessl
Fiber 610 ss
(10 mm in Ihicknessl
Soil
Screen- coged l.;!;.1:.-Hl~kA?;E3--- Tef Ion C homber
Thermocouple
Psychromeler
4~ffi~~~r--- Thermocoup le
Psychrometer
Collon
Toproot ~I--- Foom Insulotion
( 14 mm in thickness)
Sond (Oryl ~B-- V inyl Tubing
( 13mmOO.IOmm 10)
PVC Pipe
1000mm (113 mm 00, 101 mm ID)
Wox Loyer Foom Insu lotion

(14 mm in thicknessl
(3mm in
th icknessl
Rubber
Omm Sto pp er
Sond (Wet) -E~~NfI Groduoted

Cylinder
(IOOml)
PVC Cop - -E3'<'1'"'Hi
Fig. 14.16. Apparatus for measuring root axial resistance to water flow. (From Yamauchi et al.
1995)
root water potential. Simonneau and Habib (1991) developed the sucker shoot
method. They used sucker shoots of young peach trees (Fig. 14.17) to estimate
the water potential at the root-sucker branching point. They enclosed the sucker
shoots in plastic bags and aluminium foil to prevent transpiration and mea-
sured the water potential of sucker shoots with a pressure chamber. As there is
no ftow in the suckers, the measured water potential should be equal to the water
potential in the root xylem at the branching point.
Turner (1981, 1987) detailed the main precautions for measuring water
potential with a pressure chamber. Thus, in rapidly transpiring leaves, mea-
surements with the pressure chamber have to be taken as soon as possible after
14 Water Uptake 497
Fig. 14.17. Water potential in trees at

different positions; LWP leaf water
potential, RWP root water potential,
ESWP enclosed sucker water potential.
(From Simonneau and Habib 1991)
'--_ _ ESWP
RW P
excision (not later than 10-30 s); the leaf petiole should not be recut; the length
of the petiole left outside of the pressure vessel should be as short as possible;
gas must not escape from the pressure vessel during the measurement process;
pressurization in the vessel has to be slower than 0.005MPas- 1• The pressure
chamber is robust and inexpensive. Time required for each measurement varies
among species. The technique is, however, destructive and, in some species, the
determination of the end point is complicated by resin exudation from ducts in
the xylem or fluid exudation from the pith and cortex (Ritchie and Hinckley
1975).
14.6.2 Stem Diameter Changes
Water stress in a plant can lead to shrinking and swelling of the stern. Mea-
surement of stern diameter variations and the analysis of the relation with the
water status of the plant have been made in both herbaceous and woody plants
(Molz and Klepper 1973, in cotton; So et al. 1979, in soybean; Brough et al. 1986,
in apple trees; Castel 1994, in citrus trees). Molz and Keppler (1972) found that
the relationship between stern diameter and leaf water potential had significant
hysteresis, mainly due to the time lag in transmittance of water potential
TOP - VI EW Fig. 14.18. Design (top, side and front view)

of linear variable difference transforrner
(LVDT) attached to a plant stern to record
c changes in stern diarneter; a LVDT, b perspex
holder, c perspex backing plate, d plant stern,
e rnachine screw for adjusting rnechanical
zero,! rubber bands, g LVDT-core, h light
metal spring. (Frorn So et al. 1979)
2. 8 3 cm
I
d
e 9
SI DE- V IEW FRONT-VIE W
changes from the xylem into the phloem and its surrounding cor tic al tissues.
Using an analogy between a tensiometer-soil system, and a linear variable dif-
ferential transformer (LVDT - other authors use the word 'transducer' instead
of'transformer') plant system, So et al. (1979) developed a dynamic method to
correct for the time lag observed when using stern diameter fluctuations to
predict changes in plant water potential. They used a LVDT system continuously
to monitor changes in stern diameter. The holders were similar to that used by
Klepper et al. (1971; Fig. 14.18). They reported small errors due to wind fluctu-
ations. LVDT holders with a low expansion coefficient have also been described
by Huguet (1985) and Li et al. (1989). The holder and sensor system can reach
a precis ion of 5 Jlm. The system can be mounted on a stern or branch. Outputs
can be electronically recorded and stored in a data logger.
Simmoneau et al. (1993) used a LV DT system in roots of a 4-year-old peach
tree growing in pots supplied with nutrient solution. The sensor and holder was
of the type described by Huguet (1985). They observed that the diurn al evolu-
tion in diameter of the terminal shoot, main branch, trunk and main root
followed a similar pattern, close to changes in climatic conditions.
14.7 Summary and Conclusions
The biophysical processes involved in the movement of water through soil to

roots, hydraulic lift (and the reverse process), and root-shoot communication,
provide many challenges for the researcher. A considerable variety of tech-
niques have been developed for such studies, with different spatial and tempo-
14 Water Uptake 499
raI resolution. Sophisticated technological methods for evaluation of water

movement, and water distribution around and within roots, e.g. the NMR tech-
nique and the use of isotopes, are now at the disposal of researchers. Also, new
electronic devices and innovative sensors are now capable of working in the
soil, in the vicinity of roots, and are miniaturised and automated to provide
the researcher with powerful tools for studying the biophysical functioning of
the root system. Using such high-resolution tools, however, will not necessarily
contribute to the understanding of water balances on ecologically relevant time
scales, unless such study is accompanied by a better understanding of the range
of internal regulation and feedback processes operating in the plant as a whole.
The functionality of plant responses to temporary or local droughts depends
on the chances it provides for long-term survival and reproduction, rather than
in maximizing short-term productivity. The tools that the researcher chooses
will thus continue to depend on the goals of the study, and the way detailed
knowledge contributes to the overall objectives.
References
Allison GB, Hughes MW (1983) The use of natural tracers as indicators of soil-water movement
in a temperate semi-arid region. J Hydro160: 157-173
Alm DM, Cavelier J, Nobel PS (1992) A finite-element model of radial and axial conductivities
for individual roots: development and validation for two dessert succulents. Ann Bot 69:
87-92
American Society of Agronomy - Soil Science Division (1986) Methods of soil analysis, Part 1,
Physical and mineralogical methods, Agronomy Monograph No 9, 2nd edn., American
Society of Agronomy Dnc., Madison, Wisconsin
Baker JM, Allmaras RR (1990) System for automating and multiplexing soil moisture measure-
ment by time-domain reflectometry. Soil Sci Soc Am J 54: 1-6
Baker JM, van Bavel CHM (1986) Resistance of plant roots to water loss. Agron J 78: 641-644
Baker JM, van Bavel CHM (1988) Water transfer through cotton plants connecting soil regions
of differing water potential. Agron J 80: 993-997
Balling A, Zimmermann U (1990) Comparative measurements of the xylem pressure of Nico-
tiana plants by means of the pressure bomb and pressure probe. Planta 182: 325-338
Balling A, Zimmermann U, Biichner KH, Lange LO (1988) Direct measurement of negative pres-
sure in artificial-biological systems. Naturwissenschaften 75: 409-411
Bell JP, Dean TJ, Hodnett MG (1987) Soil moisture measurement by an improved capacitance
technique, part II. Field techniques, evaluation and calibration. J Hydrol 93: 79-90
Benkert R, Balling A, Zimmermann U (1991) Direct measurements of pressure and flow in the
xylem vessels of Nicotiana tobacum and their dependence on flow resistance and the tran-
spiration rate. Bot Acta 104: 423-432
Bennet JM, Cortes PM (1985) Errors in measuring water potentials of small samples resulting
from water adsorption by thermocouple psychrometer chambers. Plant Physiol 79: 184-188
Blizzard WE, Boyer JS (1980) Comparative resistance of the soil and the plant to water trans-
port. Plant Physio166: 809-814
Bottomley PA, Rogers HH, Foster TH (1986) NMR imaging shows water distribution and
transport in plant root systems in situ. Proc Natl Acad Sci USA 83: 87-89
Bottomley PA, Rogers HH, Prior SA (1993) NMR imaging of root water distribution in intact
Vicia faba L. plants in elevated atmospheric CO 2 • Plant Cell Environ 16: 335-338
Bouyoucus GJ, Mick AH (1940) An electrical resistance method for the continuous
measurement of soil moisture under field conditions. Mich Agric Exp Stn Techn Bull 172,
pp 1-38
Boyer JS (1972) Use of isopiestic technique in thermocouple psychrometry. II. Construction. In:
Brown RW, van Haveren BP (eds) Psychrometry in water relations research. Utah Agricul-
tural Experimental Station, Utah State University, pp 98-lO2
Brough D, Jones H, Grace J (1986) Diurnal changes in water content of the stems of apple trees,
as influenced by irrigation. Plant Cell Environ 9: 1-7
Brown JM, Johnson GA, Kramer PJ (1986) In vivo magnetic resonance microscopy of changing
water content in Pelargonium hortorum roots. Plant Physiol 82: 1158-1160
Brown RW, Chambers JC (1987) Measurements of in situ water potential with thermocouple
psychrometers: a critical evaluation. Proceedings of the International Conference on
Measurement of Soi! and Plant Water Status. July 6-10, Utah. Logan, USA 125-136
Brown WR, Collins JM (1980) A screen-caged thermocouple psychrometer and calibration
chamber for measurements of plant and soi! water potential. Agron J 72: 851-854
Cabibel B, Do F (1991) Mesures thermiques des flux de seve et comportement hydrique des
arbres. II. Evolution dans le temps des flux de seve et comportement hydrique des arbres
en presence ou non d'une irrigation localisee. Agronomie 11: 757-766
Caldwell MM, Richards JH (1989) Hydraulic lift: water effux from upper roots improves
effectiveness of water uptake by deep roots. Oecologia 79: 1-5
Campbell GS, Campbell MD (1974) Evaluation of a thermocouple hygrometer for measuring leaf
water potential in situ. Agron J 66: 24-27
Campbell EC, Campbell GS, Barlow WK (1973) A dewpoint hygrometer for water potential
measurement. Agric Meteorol12: 113-121
Castel JR (1994) Response of young clementine citrus trees to drip irrigation. I. Irrigation
amount and number of drippers. J Hortic Sci 69: 481-489
Cermak J, Dernl J, Penka M (1973) A new method of sap flow rate determination in trees. Biol
Plant 15: 171-178
Clothier BE, Green SR (1994) Root zone processes and the efficient use of irrigation water. Agric
Water Manage 25: 1-12
Cohen Y, Fuchs M, Green GC (1981) Improvement of the heat pulse method for determining sap
flow in trees. Plant Cell Environ 4: 391-397
Colman EA, Hendrix TM (1949) Fiberglass electrical soi! moisture instrument. Soi! Sci 67:
425-438
Compston W, Epstein S (1958) A method for the preparation of carbon dioxide from water vapor
for oxygen isotope analysis. Trans Am Geophys Union 39: 511-512
Cornish K, Zeevaart JAD (1985) Abscisic acid accumulation by roots of Xanthium strumarium
L. and Lycopersicon esculentum Mill. in reiat ion to water stress. Plant Physiol 79: 653-658
Dalton FN (1988) Plant root water extraction studies using stable isotopes. Plant Soi! 111:
217-221
Dalton FN, Raats PAC, Gardner WR (1975) Simultaneous uptake of water and solutes by plant
roots. Agron J 67: 334-339
Dansgaard W (1964) Stable isotopes in precipitation. Tellus 16: 399-401
14 Water Uptake 501
Dawson TE (1993) Hydraulic lift and water use by plants: implications for water balance,
performance and plant-plant interactions. Oecologia 95: 565-574
Dawson TE, Ehleringer JR (1991) Streamside trees that do not use stream water. Nature 350:
335-337
Dawson TE, Pate JS (1996) Seasonal water uptake and movement in root systems of Australian
phraeatophytic plants of dimorphic root morphology: a stable isotope investigation.
Oecologia 107: 13-20
de Willigen P, van Noordwijk M (1987) Roots, plant production and nutrient use efficiency. Doct
Thesis Wageningen Agricultural University, Wageningen, the Netherlands
Dirksen C, HiIhorst MA (1994) Calibration of a new frequency domain sensor for soil water
content and bulk electrical conductivity. Proc Symp on TDR in Environmental,
Infrastructure and Mining Applications. September 7-9, 1994 Evanston, Illinois
Dirksen C, Raats PAC (1985) Water uptake and release byalfalfa roots. Agron J 77: 621-626
Dryden P, van Alfen NK (1983) Use of the pressure bomb for hydraulic conductance study. J Exp
Bot 34: 523-528
Dugan JP, Borthwick J, Harmon RS, Gagnier MA, Glahn JE, Kinsel EP, MacLeod S, Viglino JA,
Hess JW (1985) Guanidine hydrochloride method for determination of water oxygen
isotope ratios and the oxygen-18 fractionation between carbon dioxide and water at 25°C.
Anal Chem 57: 1734-1736
Eller BM, Egli A, Flach BM-T (1993) Influence of transpiration and photosynthetic pathway on
water uptake of Cotyledon orbiculata (CAM) and Othonna opima (C3). Bot Helv 103: 207-221
Else MA, Davies WJ, Whitford PN, Hali KC, Jackson MB (1994) Concentrations of abscisic
acid and other solutes in xylem sap from root systems of tomato and castor-oil plants are
distorted by wounding and variable sap flow rates. J Exp Bot 45: 317-323
Else MA, Davies WJ, Malone M, Jackson MJ (1995) A negative hydraulic message from oxygen-
deficient roots of tomato plants? Plant Physiol 109: 10 17-1024
Epstein S, Mayeda T (1953) Variation of 180 content of waters from natural sources. Geochim
Cosmochim Acta 4: 213-224
Ferguson JG (1933) Classification ofbridge methods of measuring impedances. Bell Syst Technol
J 12: 452-459
Fermindez JE, Moreno F, Martin-Aranda J, Fereres E (1992) Olive-tree root dynamics under
different soil water regimes. Agric Mediterr 122: 225-235
Fiscus EL (1972) In situ measurement of root-water potential. Plant Physiol 50: 191-193
Fiscus EL (1975) The interactions between osmotic- and pressure-induced water flow in plant
roots. Plant Physiol55: 917-922
Fiscus EL (1977) Effects of coupled solute and water flow in plant roots with special reference
to Brouwer's experiment. Plant Physiol55: 917-922
Flach BM-T, Eller BM, Egli A (1995) Transpiration and water uptake of Senecio medley-woodii
and Aloe jucunda under changing environmental conditions: measurements with a poto-
metric water-budget-meter. J Exp Bot 291: 1615-1624
Flanagan LB, Ehleringer JR (1991) Stable isotope composition of stern and leaf water: applica-
tions to the study of plat water use. Funct Ecol 5: 270-277
Flanagan LB, Ehleringer JR, Marshall JD (1992) Differential uptake of summer precipitation
among co-occurring trees and shrubs in a pinyon woodland. Plant Cell juniper Environ
15:831-836
Fonteyn PJ, Schlesinger WH, Marion GM (1987) Accuracy of soil thermocouple hygrometer mea-
surements in desert ecosystems. Ecology 68: 1121-1124
Frensch J, Steudle E (1989) Axial and radial hydraulic resistance in roots of maize (Zea mays L.).
Plant Physiol91: 719-726
Friedman 1, O'Nei! JR (1978) Isotopes in nature. In: Wedepohl KH (ed) Handbook of geochem-
istry. Springer, Berlin Heidelbery New York 1B1-1 B8
Gardner WR (1960) Dynamic aspects of water and avai!ability to plants. Soil Sci 89: 63-73
Gardner WH (1965) Water content. In: American Society of Agronomy, Methods of soil analy-
sis, Monograph 9. Klute, Madison, Wisconsin pp 92-127
Gardner WH (1986) Water content. In: American Society of Agronomy, Methods of soi! analy-
sis. Parti, 2nd edn. Klute, Madison, Wisconsin pp 43-544
Gardner WR, Kirkham D (1952) Determination of soil moisture by neutron scattering. Soil Sci
73:391-401
Gat JR (1980) The isotopes ofhydrogen and oxygen in precipitation. In: Fritz P, Fontes Ch (eds)
Handbook of environmental isotope geochemistry, voi 1. Elsevier, Amsterdam, pp 24-48
Goltz SM, Benoit G, Schimmelpfennig H (1981) New circuitry for measuring soi! water matric
potential with moisture blocks. Agric Meteorol 24: 25-82
Gradmann H (1928) Untersuchungen iiber die Wasserverhăltnisse des Bodens als Grundlage des
pflanzenwachstums. Jahrb Wiss Bot 69: 1-100
Granier A (1985) Une nouvelle methode pour la mesure du flux de seve brute dans le tronc des
arbres. Ann Sci For 42: 193-200
Green SR, Clothier BE (1988) Water use ofkiwifruit vines and apple trees bythe heat-pulse tech-
nique. J Exp Bot 39: 115-123
Green SR, Clothier BE (1991) Heat pulse measurement of sap flow in trees and vines. In: van
Bavel CHM (ed) 1992 Symposium on sap flow measurements. Collected Summaries of
papers at the 83rd ASA meeting, Denver CO, Oct27-Nov2 1992 37pp. Dynamax Inc,
Houston, Texas. 37 pp
Green SR, Clothier BE (1995) Root water uptake by kiwifruit vines following partial wetting of
the root zone. Plant Soil173: 317-328
Green SR, Clothier BE, McLeod DJ (1996) The preferential uptake response of sap flow in apple
roots due to localised irrigation. Agric Water Manage 33: 63-78
Hallgren SW, Riidinger M, Steudle E (1994) Root hydraulic properties of spruce measured with
the pressure probe. Plant Soi! 167: 91-98
Hartung W, Davies WJ (1991) Drought-induced changes in physiology and ABA. In: Davies WJ,
Lones HG (eds) Abcisic acid, physiology and biochemistry. Bios Scientific Publishers,
Oxford, pp 63-79
Heimovaara TJ, Bouten W (1990) A computer-controlled 36-channel time domain reflectome-
try system for monitor ing soi! water contents. Water Resour Res 26: 2311-2316
Herkelrath WN, Hamburg SO, Murphy F (1991) Automatic, real-time monitoring of soi! mois-
ture in a remote field area with time domain reflectometry. Water Resour Res 27: 857-864
Hi!horst MA, Dirksen (1994) Dielectric water content sensors: time dom ain versus frequency
domain. Proc Symp on TDR in Environmental, Infrastructure and Mining Applications.
September 7-9, 1994 Evanston, Illinois
Hilhorst MA, Groenwold J, Groot JF (1992) Water content measurements in soi! and rockwool
substrates: dielectric sensors for automatic in situ measurements. Sensors in Horticulture.
Acta Hortic 304: 209-218
Hillel D (1980) Fundamentals of soi! physics. Academic Press, New York
Holmes JW (1956) Calibration and field use of the neutron scattering method of measuring soi!
water content. Aust J Appl Sci 7: 45-58
14 Water Uptake 503
Huang B, Nobel PS (1994) Root hydraulic conductivity and its components, with emphasis on
desert succulents. Agron J 86: 767-774
Huber B (1924) Die Beurteilung des Wasserhaushaltes der Pflanze. Jahrb Wiss Bot 64: 1-120
Huber B (1932) Beobachtung und Messung pflanzlicher Saftstr6me. Ber Dtsch Bot Ges 50:
89-109
Huguet J (1985) Appreciation de l'etat hydrique d'une plante il partir des variations
micrometriques de la dimension des fruits ou des tiges au cours de la journee. Agronomie
5:733-741
Jackson MB (1994) Root-to-shoot communication in flooded plants: involvement of abscisic
acid, ethylene, and 1-aminocyclopropane-l-carboxylic acid. Agron J 86: 775-782
Jones HG (1983) Estimation of an effective soil water potential at the root surface of transpir-
ing plants. Plant Cell Environ 6: 671-674
Klepper B, Browning VD, Taylor HM (1971) Stern diameter as a measure of plant water status.
Plant Physiol 46: 683-685
Koide RT, Robichaux RH, Morse SR, Smith CM (1989) Plant water status, hydraulic resistance and
capacitance. In: Pearcy RW, Ehleringer J, Mooney HA, Rundel PW (eds) Plant physiological
ecology. Field methods and instrumentation. Chapman and Hall, London, pp 161-183
Kramer PJ, Boyer JS (1995) Water relations of plants and soils. Academic Press, London
Landsberg JJ, Fowkes ND (1978) Water movement through plant roots. Ann Bot (Lond) 42:
493-508
Li S, Hughet J, Bussi C (1989) Irrigation scheduling in a mature peach orchard using tensiome-
ters and dendrometers. Irrig Drainage Syst 3: 1-12
Lightbody KE, Savage MJ, Graham ADN (1994) In situ measurement of the sap flow rate in lateral
roots and stems of Euealyptus grandis under conditions of marginality, using a steady state
heat balance technique. J S Afr Soc Hortic Sci 4: 1-7
List RJ (1968) Smithsonian meteorological tab les, 6th revised edn. Smithsonian Institution Press,
Washington, DC
Liick W (1964) Physikalische Grundlagen und Begriff der Feuchtemessung: Gebergerăte fUr Gas-
feuchte und Materialfeuchte. VDI - Bildungswerkhandbuch. BW 1421, S 1-2, and BW 1429,
S 1-7
MacFall JS, Johnson GA, Kramer PJ (1990) Observation of a water-depletion region surround-
ing loblolly pine roots by magnetic resonance imaging. Proc Natl Acad Sci USA 87:
1203-1207
MacFall JS, Johson GA, Kramer PJ (1991) Comparative water uptake by roots of different ages
in seedlings of loblolly pine (Pinus taeda 1.). New Phytol119: 551-560
Magnani F, Borghetti M (1995) Interpretation of seasonal changes of xylem embolism and plant
hydraulic resistance in Fagus sylvatiea. Plant Cell Environ 18: 689-696
Marshall DC (1958) Measurement of sap flow in conifers by heat transport. Plant Physiol 33:
385-396
McKeague JA, Eilers RG, Thomasson A, Reeve MJ, Bouma J, Grossman RB, Favrot JC, Renger M,
StrebelO (1984) Tentative assessment of soil survey approaches to the characterisation and
interpretation of air-water properties of soils. Geoderma 34: 69-100
Melchior W, Steudle E (1993) Water transport in onion (Allium eepa 1.) roots. Changes ofaxial
and radial hydraulic conductivities during root development. Plant Physiol 10 1: 1305-1315
Melchior W, Steudle E (1994) Hydrostatic and osmotic hydraulic conductivities of onion (AlIium
eepa 1.) roots. In: Baluska F, Gasparikova O, Ciamporova M (eds) Structure and function
of roots. Kluwer, Dordrecht
Michel BE (1977) A miniature stern thermocouple hygrometer. Plant Physiol60: 645-647

Molz FJ, Klepper B (1972) Radial propagation of water potential in stems. Agron J 64: 469-473
Molz FJ, Klepper B (1973) On the mechanism of water-stress-induced stern deformation. Agron
J 65: 304-306
Mooney HA, Gulman SL, Rundel PW, Ehleringer J (1980) Further observations on the water rela-
tions of Prosopis tamarugo of the northern Atamama desert. Oecologia 44: 177-180
Moreno F, Fernandez JE, Clothier BE, Green SR (1996) Transpiration and root uptake by olive
trees. Plant SoiI184: 85-96
Moreshet S, Huck MG (1991) Dynamics of water permeability, In: Waisel V, Eshel A, Kafkafi U
(eds) Plant roots: the hidden half. Marcel Dekker, New York pp 605-623
Neumann HH, Thurtell GW (1972) A Peltier-cooled thermocouple dewpoint hygrometer for in
situ measurement of water potentials. In: Brown RW, van Haveren BP (eds) Psychrometry
in water relations research. Utah Agricultural Experiment Station, Utah State University,
Utah, pp 103-112
Nobel PS (1991) Physicochemical and environmental plant physiology.Academic Press, San Diego
Nobel PS, Schulte PJ, North GB (1990) Water influx characteristics and hydraulic conductivity
for roots of Agave deserti Engelm. J Exp Bot 41: 409-415
Omasa K, Onoe M, Yamada H (1985) NMR imaging for measuring root systems and soiI water
content. Environ Control Bio123: 99-102
Ong CK, Khan AAH (1993) The direct measurement of water uptake by individual tree roots.
Agrofor Today Oct-Dec 2-5
Oosterhuis DM (1987) A technique to measure the components of root water potential using
screen-caged thermocouple psychrometers. Plant SoiII03: 285-228
Paltineanu IC, Starr JL (1997) Real-time soiI water dynamics using multisensor capacitance
probes: laboratory calibration. SoiI Sci Soc Am J 61: 1576-1585
Passioura JB (1984) Hydraulic resistance of plants. 1. Constant or variable? Aust J Plant Physiol
11: 333-339
Passioura JB (1988) Water transport in and to roots. Annu Rev of Plant Physiol 39: 245-265
Passioura JB, Munns R (1984) Hydraulic resistance of plants. II. Effects of rooting medium, and
time of day, in barley and lupin. Aust J Plant Physiol11: 341-350
Pearcy RW, Ehleringer J, Mooney HA, Rundel PW (eds) (1989) Plant physiological ecology.
Chapman and HalI, New York
Peterson CA, Steudle E (1993) Lateral hydraulic conductivity of early metaxylem vessels in Zea
mays 1. roots. Planta 189: 288-297
Peterson CA, Murrmann M, Steudle E (1993) Location of the major barriers to water and ion
movement in young roots of Zea mays 1. Planta 190: 127-136
Rambal S (1984) Water balance and pattern of root water uptake by a Quercus coccifera ever-
green scrub. Oecologia 62: 18-25
Rieger M (1995) Offsetting effects of reduced root hydraulic conductivity and osmotic adjust-
ment following drought. Tree Physiol 15: 379-385
Rieger M, Motisi A (1990) Estimation of root hydrulic conductivity on intact peach and citrus
rootstocks. HortScience 25: 1631-1634
Ritchie GA, Hinckley TM (1975) The pressure chamber as an instrument for ecological research.
Adv Ecol Res 9: 165-254
Robinson DA, Gardner CMK, Evans J, Cooper JD, Hodnett MG, Bell JP (1998) The dielectric cal-
ibration of capacitance probes for soiI hydrology using an oscillation frequency response
model. Hydrol Earth Syst Sci 2: 83-92
14 Water Uptake 505
Rogers HH, Bottomley PA (1987) In situ nuclear magnetic resonance imaging of roots: infiu-
ence of soi! type, ferromagnetic particle content, and soil water. Agron J 79: 957-965
Rowse HR, Goodman D (1981) Axial resistance to water movement in broad bean (Vicia faba)
roots. J Exp Bot 128: 591-598
Riidinger M, Hallgren SW, Steudle E, Schulze E-D (1994) Hydraulic and osmotic properties of
spruce roots. J Exp Bot 279: 1413-1425
Saab IN, Sharp RE (1989) Nonhydraulic signals from maize roots in drying soi!s: inhibition of
leaf elongation but not stomatal conductance. Planta 179: 466-474
Sakuratani T (1981) A heat balance method for measuring water fiow rate in the stern of intact
plants. J Agric Meteorol 40: 273-277
Sands R, Fiscus EL, Reid CPP (1982) Hydraulic properties of pine and bean roots with varying
degrees of suberation. Vascular differentiation and mycorrhizal infection. Aust J Plant
Physiol 9: 959-69
Savage MI, Ritchie JT, Khuvutlu IN (1987) Soi! hygrometers for obtaining water potential. Proc
Int Conf on Measurement of Soil and Plant Water Status. July 6-10, Utah. Logan, USA, pp
119-124
Schildwacht PM (1988) Changes in the osmotic potential of the root as a factor in the decrease
in the root-shoot ratio of Zea mays plants under water stress. Plant Soill11: 271-275
Schlesinger WH, Fonteyn PJ, Marion GM (1987) Soil moisture content and plant transpirat ion
in the Chihuahuan Desert of New Mexico. J Arid Environ 12: 119-126
Schlub RL, Maine JW (1979) Portable recorder for the continuous monitoring of soi! moisture
resistance blocks. J Agric Engin Res 24: 319-323
Schmugge TJ, Jackson TJ, McKim HL (1980) Survey of methods for soi! moisture determination.
Water Resour Res 16: 961-979
Scholander PF, Hammel HT, Bradstreet ED, Hemmingsen EA (1965) Sap pressure in vascular
plants. Science 148: 339-113
Schurr U, Gollan T, Sculze E-D (1992) Stomatal responses to soi! drying in relation to changes
in the xylem sap composition of Helianthus annuus: II. Stomatal sensitivity to abscisic acid
imported from xylem sap. Plant Cell Environ 15: 551-560
Simonneau T, Habib R (1991) The use of tree root suckers to estimate root water potential. Plant
Cell Environ 14: 585-591
Simonneau T, Habib R, Goutouly JP, Huguet (1993) Diurnal changes is stern diameter depend
upon variations in water content: direct evidence in peach trees. J Exp Bot 260: 615-621
Smith DM, Allen SJ (1996) Measurement of sap fiow in plant stems. J Exp Bot 47: 1833-1844
Smith DM, Jarvis PG, Odongo JCW (1996) Water use by Sahelian windbreak trees. In: Buerkert
B, Allison BE, von Oppen M (eds) Wind erosion in West Africa: the problem and its control.
MargafVerlag, Weikersheim, pp 115-128
Smith DM, Jarvis PG, Odongo JCW (1997) Sources ofwater used by trees and millet in Sahelian
windbreak systems. J Hydrol198: 140-153
Smith DM, Jackson NA, Roberts JM (2000) A new direction in hydraulic lift: can tree roots siphon
water downwards? Agrofor Forum 8 (1): 23-26
Smith-Rose RL (1933) The electrical properties of soi!s for alternating currents at radio fre-
quencies. Proc R Soc Lond 140: 359-377
So HB, Reicosky DC, Taylor HM (1979) Utility of stern diameter changes as predictors of plant
canopy water potential. Agron J 71: 707-713
Spanner DC (1951) The Peltier effect and its use in the measurement of suction pressure. J Exp
Bot 2: 145-168
Sperry JS, Tyree MT (1990) Water-stress-induced xylem embolism in three species of conifers.
Plant CeH Environ 13: 427-436
Sperry JS, Nichols KL, Sullivan JEM, Eastlack SE (1994) Xylem embolism in ring-porous, diffuse-
porous, and coniferous trees of northern Utah and interior Alaska. Ecology 75: 1736-1752
Steudle E (1994) Water transport across roots. Plant Soi! 167: 79-90
Steudle E, Jeschke WO (1983) Water transport in barley roots. Planta 158: 237-248
Steudle E, Peterson CA (1998) How does water get through roots? J Exp Bot 49: 775-788
Stirzaker RJ, Passioura JB (1996) The water relations of the root -soi! interface. Plant CeH Environ
19:201-208
Stone JF, Kirkham O, Read AA (1955) Soi! moisture determination by a portable neutron scat-
tering moisture meter. Soi! Sci Soc Am Proc 19: 419-423
Swanson RH, Perttu AAH (1994) Significant historical developments in thermal methods for
measuring sap flow in trees. Agric For Meteorol 72: 113-132
Swanson RH, Whitfield OWA (1981) A numerical analysis of heat pulse velocity and practice. J
Exp Bot 32: 221-239
Tardieu F, Oavies WJ (1993) Integration of hydraulic and chemical signalling in the control of
stomatal conductance and water status of droughted plants. Plant CeH Environ 16: 341-349
Taylor HM, Willatt T (1983) Shrinkage of soybean roots. Agron J 75: 818-820
Ternesi M, Andrade AP, Jorrin J, BenHoch M (1994) Root-shoot signalling in sunflower plant
with confined root systems. Plant Soi! 166: 31-36
Topp GC, Oavis JL (1985a) Time-domain reflectometry (TOR) and its application to irrigation
scheduling. Adv Irrig 3: 107-127
Topp GC, Oavis JL (1985b) Measurement of soi! water content using time domain reflectome-
try (TOR): a field evaluation. Soi! Sci Soc Am J 49: 19-24
Trejo CL, Clephan AL, Oavies WJ (1995) How do stomata read abscisic acid signals? Plant Physiol
lO9: 803-811
Turner NC (1981) Techniques and experimental approaches for the measurement of plant water
status. Plant Soi! 58: 339-66
Turner NC (1987) The use of the pressure chamber in studies of plant water status. Proc. Int.
Conf. Measurement Soi! Plant Water Status 2: 13-24
Turski R, Malicki M (1974) A precise laboratory meter of a dielectric constant of soi! of a dif-
ferent moisture. Pol J Soi! Sci 7: 71-79
Tyree MT, Yang S, Cruizat P, Sinclair B (1994) Novel method of measuring conductivity of tree
root systems and interpretation using AMAIZEO. Plant Physiol104: 189-199
Tyree ME, Patino S, Bennink J, Alexander J (1995) Oynamic measurements of hydraulic con-
ductance using a high-pressure flow meter in the laboratory and field. J Exp Bot 46: 83-94
van Bavel CHM (1959) Oensitometry by gamma transmission. Soi! Sci 87: 50-58
van Bavel CHM (ed) (1992) Symposium on sap flow measurement. CoHected Summaries of
papers at the 83rd ASA meeting, Oenver CO, Oct27-Nov2 1992. Oynamax, Houston, Texas
37pp
van den Honert TH (1948) Water transport in plants as a catenary process. Oiscuss Faraday Soc
3: 146-153
Vetterlein O, Marschner H (1993) Use of microtensiometer technique to study hydraulic lift
in a sandy soi! planted pearl millet (Pennisetum americanum 1. Leeke). Plant Soi! 149:
275-282
Vetterlein O, Marschner H, Horn R (1993) Microtensiometer technique for in situ measurement
of soi! matric potential and root water extraction from a sandy soi!. Plant Soi! 149: 263-273
14 Water Uptake 507
Weatherley PE (1982) Water and flow in roots. In: Lange OL, Nobel PS, Osmond CB, Ziegler H
(eds) Encyclopedia of plant physiology. Springer, Berlin Heidelbery New York, Voi 12B, pp
79-109
White JWC, Cook ER, Lawrence JR, Broecker WS (1985) The D/H ratios of sap in trees: Impli-
cations for water sources and tree ring D/H ratios. Geochim Cosmochim Acta 49: 237-246
Wobschall D (1978) A frequency shift dielectric soil moisture sensor. IEEE Trans Geosci Elec-
tron 15: 112-118
Wong WW, Lee LS, Klein PD (1987) Oxygen isotope ratio measurements on carbon dioxide gen-
erated by reaction of microliter quantities of biological fluids with guanidine hydrochlo-
ride. Anal Chem 59: 690-693
Wraith JM, Baker JM (1991) High -resolution measurement of root water uptake using automated
time-domain reflectometry. Soil Sci Am J 55: 928-932
Xu X, Bland WL (1993) The short life and replant problems of deciduous fruit trees.Agron J 85:
384-388
Yamauchi A, Taylor HM, Upchurch DR, McMichael BL (1995) Axial resistance to water flow of
intact coUon taproots. Agron J 87: 439-445
Zhang J, Davies WJ (1990) Does ABA in the xylem control the rate of leaf growth in soil-dried
maize and sunflower plants? J Exp Bot 41: 1125-1132
Zhang J, Kirkham MB (1995a). Water relations ofwater-stressed, split-root C4 (Sorghum bicolor,
Poaceae) and C3 (Helianthus annuus, Asteraceae) plants. Am J Bot 82: 1220-1229
Zhang J, Kirkham MB (199Sb) Sap flow in a dicotyledon (sunflower) and a monocotyledon
(sorghum) by the heat-balance method. Agron J 87: 1106-1114
Zhu J], Zimmermann U, Thiirmer F, Haase A (1995) Xylem pressure response in maize roots
subjected to osmotic stress: determination of radial reflection coefficients by use of the
xylem pressure probe. Plant CeH Environ 18: 906-912
Zimmermann U, Meinzer FC, Benker R, Zhu J], Schneider H, Goldstein G, Kuchenbrod E, Haase
A (1994) Xylem water transport: is the available evidence consistent with the cohesion
theory? Plant CeH Environ 17: 1169-1181
Zimmermann U, Meinzer F, Bentrup F-W (1995) How does water ascend in tall trees and other
vascular plants? Ann Bot 76: 545-551
CHAPTER lS
Modelling Water and Nutrient Uptake

P. de Willigen 1, N.E. Nielsen 2, N. Claassen\ and A.M. Castrignano 4
1 Alterra, Green World Research, Postbus 47,6700 AA Wageningen, The Netherlands
2 The Royal Veterinary and Agricultural University, 40 Thorvaldsensvej, 1871 Frederiksberg C,
Copenhagen, Denmark
3 Georg-August University, von Siebold-Str. 6, 37075 Gottingen, Germany
4 Istituto Sperimentale Agronomico, Via C. Ulsiani 5,70125 Bari, Italy
CONTENTS
15.2 Single Root Models 511

15.2.1 Basic Transport Equation 511
15.2.2 Geometry and Boundary Conditions 511
15.2.3 Initial Condition 512
15.2.4 Nutrients 512
15.2.4.1 Transport Equation 512
15.2.4.2 Boundary Conditions 514
15.2.4.3 Analytical Solutions and Approximations for Simple
Geometry 515
15.2.4.4 Examples and Applications 518
Availability 518
Application to Pot Experiments 519
15.2.5 Water 520
15.2.5.1 Transport Equation 520
15.2.5.2 Boundary Conditions 523
15.2.5.3 Analytical Solutions and Approximations 525
Availability 526
Root and Soil Resistance 527
15.3 Modelling Uptake by a Crop Root System 529
15.3.1 Nutrients 529
Stationary Root Systems 532
Growing Root Systems 532
A.L. Smil el al. (Eds.J, Rool Melhods

510 P. de Willigen et al.
15.3.2 Water 536

15.4 Conclusions and Suggestions for Further Research 537
References 539
15.1 Introduction
Supplying plants with in organic nutrients and water is a major function of

roots (Nye and Tinker 1977; Nielsen 1983; Barber 1984; Jungk and Claassen
1989; Gahoonia and Nielsen 1991, 1992a,b). It is stiH a major challenge to
agricultural research to quantitatively predict production of a crop as
influenced by uptake of nutrients and water. Models can be of great assis-
tance to understand how different factors and their interactions influence
crop functioning. A model is per definition a simplification of part of reality.
To quote Van Wijk (1963): "Generally it is necessary to introduce drastic
simplifications in an agricultural problem to make it amenable to a mathe-
matical analysis. Objections are sometimes raised that such simplifications
restrict or even preclude the applicability.... However this need not be the
case. The very purpose of the mathematical physical treatment is to isolate the
essential characteristics of the problem and to abstract them from the less
essential ones".
Nye (1992a,b,c) mentioned the major advantages of modelling: "a) The act
of constructing a model ... clarifies thinking, b) the overall model serves to
identify gaps in knowledge or understanding, c) the model provides a frame-
work for discussion and argument, and serves to identify precisely the points
at issue, d) it helps to decide research priorities by highlighting the most worth-
while experiments:' This chapter gives an overview of different models in use
and discusses the advantages and disadvantages of the different approaches, at
the scale of a single root and that of a crop root system. Further detail can be
found in several books and review papers dealing with uptake of nutrients (Nye
and Tinker 1977; Barber 1984; Rengell999) and ofwater (Molz 1981; Jones and
Ritchie 1991). In this chapter we emphasise the similarity of the transport prob-
lems involved in water and nutrient uptake, and discuss these in an analogous
fashion. We first begin with the fundamentals of transport to and uptake by a
single root.
15 Modelling Water and Nutrient Uptake 511
1 S.2 Single Root Models
15.2.1 Basic Transport Equation
The equation of continuity for a substance in a porous region reads:

as
ar =-V.F+U, (15.1)
where S is the bulk density of the substance [L.T-3 ], F is the flux [M.L-2.T-1], v
is the gradient operator [L- 1], T is time [T], and U is the production or
consumption [M.L-3.T-1]. The flux of the substances under consideration can,
at least within the scope of this book, be assumed to consist of two components:
a convective component and a diffusive component:
F=Fc+FD. (15.2)
The convective flux is due to movement of the fluid which contains the sub-
stance, brought about by e.g. gradients of total pressure. The moving fluid
carries with it the substance it contains:
Fc= VC, (15.3)
where Vis the volume flux of the fluid [L. T-1], and C is the concentration of the
substance in the fluid [M.L-3]. The diffusive flux is assumed here to be given by
Fick's first law, Le. proportional to the gradient of the concentration:
FD=-DVC, (15.4)
where D is the diffusion coefficient W.T-1]. Substitution of Eqs. (15.2), (15.3),
and (15.4) into (15.1) results in:
as
ar=-V.VC+V.DVC+U, (15.5)
and this then is the general equation that is used to describe transport in the soH
to the root. It assumes a particular form where water or solutes are concerned.
The same symbols C, S, U and D (with the exception of diffusivity Dw) will be
used. From the context it will be clear whether they refer to water or solutes.
15.2.2 Geometry and Boundary Conditions
Though the earlier descriptions of nutrient transport and uptake bya root con-
sidered a single root in an infinite medium (Bouldin 1961; Nye 1966), most single
root models consider a single cylindrical root confined in an isolated soH
cylinder. The single root is assumed to be representative of a set of roots alI having
the same dimensions and operating under the same conditions. In order to solve
the partial differential equation (15.5) boundary conditions have to be formu-
lated. Two boundary conditions have to be formulated, one at the outer bound-
ary of the medium and one at the inner boundary, the surface of the root.
For a root in an infinite medium the boundary condition at infinity is that
of finite bulk density of the substance. For a root confined within a soil cylin-
der, with radius R J the condition at the outer boundary of the latter is that of
vanishing flux:
R=R J, Fc+FD=O (15.6)
At the inner boundary, Le. at the root surface where R = Ro, the boundary
condition is usually that of prescribed flux:
R = Ro, Fc + FD= Fo, and (15.7)
sometimes that of given bulk density:
R=Ro,S=So (15.8)
where both Fo and So can be functions of the spatial co-ordinates, time, and in
the case of flux of the concentration at Ro.
15.2.3 Initial Condition
The initial distribution of the bulk density in the soi! cylinder is generally
given by:
T=O, S=Sj (15.9)
where the initial bulk density can be a function of the spatial co-ordinates.
15.2.4 Nutrients
15.2.4.1 Transport Equation
The total bulk density of nutrient in inorganic form is composed of the amount
in solution, which is mobile, and that in the solid phase, which is considered
immobile. The latter consists of a pool Q which exchanges at a fast rate with the
pool in solution and a pool from which transfer to the solution is so slow that
in that can be ignored for our purposes:
S =Q+ ec, (15.10)
It will be assumed here that at any time there exists a unique functional
relation - the adsorption isotherm - between Q and C, symbolised by:
Q =fiC). (15.1I)
Substitution of (15.1I) and (15.lO) into (15.5) yields an equation with only
C as the dependent variable:
Jc
[f'(C)+O] ar =V ·DVC- V· VC+U. (15.12)
In the following we will generally neglect the production/consumption term

U. In Eq. (15.12) the role of the soil chemical properties is given by the deriva-
tive of the adsorption isotherm with respect to concentration as shown in the left
hand side of Eq. (15.12). The soil physical properties and conditions manifest
themselves via their influence on the transport parameters V and D. Water is both
the carrier of solutes - reflected in the role of the flux V - and the medium
through which diffusion takes place. The diffusion coefficient in soil is very much
dependent on water content and can be calculated as (Nye and Tinker 1977):
D = DrJiO, (15.13)
where Do is the diffusion coefficient in water, and fi the impedance factor, a
function of water content.
Equation (15.12) assumes special forms, some examples of which follow,
depending on subsequent assumptions. An obvious and widely used assump-
tion is that of the cylindrical form of the root, which leads to the choice of
expressing the gradient and divergence in cylindrical co-ordinates. Moreover,
when tangential and vertical gradients are assumed negligible, Eq. (15.12)
assumes the form:
(15.14)
When transport is dominated by diffusion and the mass flow component
can be neglected, Eq. (15.14) simplifies to:
[f'(C) + O] ac
ar =~~DRac.
RR aR
(15.15)
When the root has contact with the soil solution only over a part of its radial
surface, tangential gradients will develop, and when transport is by diffusion
alone, Eq. (15.12) becomes:
[f'(C)+O]Jc =~~DRJc +_1 ~ aDC (15.16)

ar R aR aR R2 dIfI dIfI '
where lfI is the tangential co-ordinate.
With some special experimental techniques it is possible to obtain condi-
tions where transport is one-dimensional in a rectangular geometry (Dunham
and Nye 1973; Dunham and Nye 1974; Gahoonia et al. 1994). In such conditions
the appropriate transport equation reads:
[f'(C)+8]de =~Dde _ avxC. (15.17)

(JT ax ax ax
When the adsorption isotherm is linear then J(C) = K•. C, and j'(C) is constant.
Provided D is independent of C, Eqs. (15.15), (15.16) and (15.17) are linear
and can be solved analytically by classical mathematical techniques. If the
adsorption isotherm is non-linear, the equations generally have to be solved by
numerical methods.
75.2.4.2 Boundary Conditions
In the case of an infinite medium, the outer boundary condition states that the
concentration converges to a finite value:
R~oo, C~Ci' (15.18)
When the root is confined within a soil cylinder, the boundary condition
at the outer boundary of the soil cylinder is that of vanishing flux Eq. (15.6), in
cylindrical co-ordinates:
de
R=RJ, -D aR +VC=o. (15.19)
The boundary condition at the root surface usually involves some assump-
tions about physiological properties of the root. Bouldin (1961) and later the
school of Nye assumed a relationship between uptake rate and concentration at
the root surface. Nye and Tinker (1977, p. 116) state: "It is most usual to con-
sider root uptake in relation to solution concentration at the surface as the
boundary condition". The earlier models (Bouldin 1961; Nye 1966) assumed
uptake to be proportional to the concentration at the root's surface:
(15.20)
where a is the root absorbing power [L.r J ]. Later models used Michaelis-
Menten kinetics, sometimes with some modification, to describe the relation
between concentration and uptake (Nye and Marriott 1969):
de C
R=Ro, -D-;-+VC=Umax - - - , (15.21)
dR Km+C
where Umax is the maximum uptake rate, occurring at very high values of C, and
Km is the Michaelis-Menten constant.
De Willigen and van Noordwijk (1978, 1987) took as a starting point the
demand of the plant, which generally is a function of time, but in the case of
crops with closed canopies is constant during a considerable fraction of the

growing season. They assumed that, as long as the concentrat ion at the root
surface exceeded a certain minimum value, uptake equalled demand. When the
minimum value is reached, uptake is such that the minimum value C/jm of the
concentration is maintained:
Jc A
R=Ro, ifC~C/jm, D-= , elseC=C/jm. (15.22)
aR 2nHLrvRo
75.2.4.3 Analytical Solutions and Approximations

for Simple Geometry
Generally, analytical solutions can only be found for linear conditions, i.e. a
linear adsorption isotherm and the diffusion coefficient and flow of water being
independent of the concentration.
The exact solutions are generally quite complicated, consisting of infinite
series of special functions, therefore simpler approximations are used. These
can be employed in macroscopic uptake models. Baldwin et al. (1973) proposed
to describe the concentration-distance profile around a root, taking up with a
rate proportional to the concentrat ion at its surface, by a sequence of steady-
state solutions. In this approach, at each time the average concentration in the
soil cylinder - when transport is by diffusion alone - is given by (for conve-
nience of notation dimensionless variables are used, Table 15.1 gives the
meaning of the symbols used.):
(15.23)
The uptake rate can then be calculated as:
p
ro = 2 c. (15.24)
1- 1 + p_P- Inp
2 p 2 _1
The approximate cumulative uptake calculated with Eqs. (15.23) and
(15.24) was shown to give good agreement with the uptake calculated by a
numerical model (Nye and Mariott 1969) especially for p < 5. Taking the limit
of Eq. (15.24) as p ~ yields the solution for zero concentration at the root
00
surface. Moreover, if one can assume p » 1, as usually is the case, Eq. (15.24)
simplifies to:
Table 15.1. List of symbols
Symbol Name Dimension Dimensionless symbol
T Time T t= DT/Ro2
R Radial co-ordinate L r=RlRo
Ro Root radius L
RI Soil cylinder radius L p
&. Root length (Iayer thickness) L 11
A Plant demand for nutrients M.L-2.11 ro = p' B/(2</111)
Si Initial amount of nutrient M.L-3
</1 Supply/demand parameter </1= D MARo)
C Concentration M.L-3 c
Ci Initial concentration M.L-3
B (=K.+0) Buffer capacity f3 = (K. + 0)/0
w Water uptake rate L3.T-I
V Flux of water 1.11 v = WI(27rHD)
Il = EI{2nHD(p'l)}
L" Root density L-2
K. Adsorption constant
0 Water content
D Diffusion coefficient L2.11
a Root absorbing power 1.11 P= aRolD
Km Michaelis-Menten constant M.L-3 1(= Km/Ci
1
W= C. (15.25)
1
lnp--
2
Claassen (1989) used the same approach as Baldwin et al. (1973), but used
Michaelis-Menten kinetics. The uptake rate can now be given as:
C-Co
W= , (15.26)
1
lnp--
2
where Co is the concentration at the root surface, which itself is a non-linear

function of the average concentration:
C-IC-1Cpf(p) ~[C -IC-1Cpf(p)]2 _

Co = + +ICC (15.27)
2 4
where f(p) is defined as: f(p) = lnp - 1/2. In the case of zero-sink uptake, Le.
when Co = 0, Eq. (15.26) is identical to Eq. (15.25).
De Willigen and van Noordwijk (1987) described the development of the
concentration distribution as a sequence of steady rate situations, in accordance
with the boundary condition of zero concentration at the root surface. Now the
average concentration is given by:
c- =exp[ -2t J (15.28)

Br-!p + p~_llnp}
2 4 '
and the uptake rate by:

p2_ 1 _
0)= c. (15.29)
1-3 p 2 p4
--+--lnp
4 p2_ 1
When p» 1 Eq. (15.29) simplifies to:
1 _
0)= c. (15.30)
3
lnp--
4
The cumulative uptake can be calculated in both cases with:

Q= 1- c. (15.31)
Both Eqs. (15.24) and (15.29) give the uptake rate as a function of the
average concentration, the buffer power does not play a role in this relation. De
Willigen (1990) showed that the zero-sink uptake rate can be approximated by
Eq. (15.29) even if the adsorption isotherm is non-linear.
Similar approximations as Eqs. (15.24) and (15.29) apply in cases of trans-
port by mass flow playing a role where the replenishment of water takes place
at the outer boundary of the soil cylinder (Cushman 1980; De Willigen and van
Noordwijk 1994a,b).
Up to now it has been assumed that the root over its total (active) surface
had direct contact with either the soil solution or the soil solid phase. However,
inspection of root-soil contact in the field, both macroscopically, and micro-
scopically by observing thin sections (Altemiiller and Haag 1983), reveals that
complete contact between the root and the soilliquid and solid phases may be
the exception rather than the rule, especially for heavy soils. De Willigen and
van Noordwijk (1987) derived a steady rate solution of the distribution of con-
centration around a root of which only a fraction of its radial surface partakes
in uptake. From their solution an approximate zero-sink uptake rate can be
derived as:
(15.32)
75.2.4.4 Examples and Applications
Availability. The concepts of soil fertility and nutrient availability have shifted
from essentially static approaches to ones in which the transport of solute in
the soil-root system plays an important role, a tale well told by Nye and Tinker
(1977).
Bray (1954) was among the first to point out the significance of the
mobility of nutrients in the soil with respect to plant nutrition. Barber (1962)
formulated this more precisely when he distinguished two mechanisms of
transport towards the plant root: transport by mass flow and transport by dif-
fus ion, concepts which are now generally accepted (Nye and Tinker 1977). How
much of the available amount will actually be taken up by the plant at the
required rate depends on the rate itself, the path length of transport (which is
a function of root length density), the transport possibilities in the soil, and the
rate of depletion by sinks other than the plant root. The available amount (in
the definition of de Willigen and van Noordwijk 1987) is that part of the total
amount present in the root zone which can be taken up by the crop when trans-
port through soil is not limiting. Within the root zone during the uptake period
many processes can occur which render part of the originally available pool, at
least temporarily, unavailable (chemical fixation, microbial immobilisation), or
transform former unavailable fractions into a readily available forms (mineral-
ization, rele ase from minerals). When relevant, considerations about crop
uptake have to take into account these amounts released or fixed.
Because of its finite root density and because transport rates in soil are
finite, the plant can take up only a fraction of the available pool. The amount
that can be taken up at the required rate, will be indicated as the unconstrained
uptake capacity. Then the total uptake capacity is that amount which will be
taken up in a certain period, e.g. a growing season, with a rate less than or equal
to the required rate. The difference between total uptake capacity and available
pool is the amount remaining in the soil due to transport limitations. De Willi-
gen and van Noordwijk (1987) applied the steady rate solution to estimate
the amount of a nutrient which is unconstrained and available, for p» 1, it is
given by:
Q = S _ A(Ka +0) Rj2 {In _ ~}. (15.33)

1 D~z 2 P 4
The unconstrained uptake capacity thus equals the available amount Si

diminished by a term representing the amount which cannot be taken up at the
required rate. The unconstrained uptake capacity equals the available amount
if root length density and/or the diffusion coefficient becomes infinite, i.e. when
transport is not limited. This can be expressed as a fraction of the available
amount, the fractional depletion:
9f3p2 AR~ {1-3 P2 p4lnp}

Fd =1- 2 ---+-2-- . (15.34)
2(p -1) DSjD.z 4 P -1
Figure 15.1 shows the fractional depletion, Le. the fraction of the available
nutrient which can be taken up with the required rate. For the parameters used
it shows that in absence of adsorption, e.g. in case of nitrate, a rather sparse root
system can take up almost alI of the available nutrient.
Application to Pot Experiments. The single root models are especialIy suitable
for analysing laboratory and pot experiments. A good example of the use of a
model with simple geometry is given by the study of Gahoonia et al. (1994) con-
cerning the transport of phosphorus to the rhizosphere as influenced by soil
moisture. By employing a special technique in pot experiments, the effects of
water content on transport of P and water could be studied separately, Le. iso-
lated from effects of water availability on root growth and physiology. The
experimental set up was such that transport in one direction in a plan ar
geometry and zero-sink uptake could be assumed. The adsorption isotherm of
P was taken to consist of two linear branches, so that for each branch separately
an analytical solution could be used. There were two treatments comprising a
(constant) moisture content of 0.14 and 0.20 respectively leading to diffusion
K,.= OmI cm-3

K,.= 20 mI cm-3
K,.= 50 mI cm-3
Fig. 15.1. Fractional depletion as a

function of root length density, mass
flow and adsorption strength.
Parameters: uptake rate 2kgha-1.day-l,
available amount 400kgha-1,
diffusion coefficient 0.1 cm2 day-l.
Interrupted lines: transpiration
O.scm day-l, uninterrupted lines no
transpiration, transport by diffusion
only. The numbers at the curves give 4.00 6.00
the values of the adsorption constant Root length density em em· 3
coefficients differing by a factor of 4. Figure 15.2 a and b show the good agree-
ment between theory and experiment, i.e. the differences in distribution of
(available) P around the root could, to agreat extent, be explained by the
differences in diffusion coefficients, which in their turn are caused by difference
in water content, so confirming the validity of the transport theory.
Claassen (1989) added to the usual form of the transport equation (15.14) a
sink term to account for the uptake of root hairs. The strength of the sink
was calculated by using an equation like (15.26), where now Ro and RI de note the
radius of the root hair and the half-distance between root hairs respectively. The
model was put to the test by calculating P-uptake by a number of species in pot
experiments. Taking uptake by root hairs into account considerably improved
the agreement between measured and calculated uptake (Fig. 15.3a and b).
15.2.5 Water
75.2.5.7 Transport Equation
In temporate regions the soil solution under conditions of interest is very

dilute [in the order of 0.01 M (Bolt and Bruggenwert 1979)], accordingly,
gradients of the concentration of water in the soil solution will be very small,
and can be neglected for our purposes, i.e. in the case of water one can assume
[ef. Eq. (15.4)]:
as = -V.Pc.
(ff (15.35)
The convective flux is given by Darcy's law which states that the flux of the
soil solution is proportional to the gradient of total potential. The total potential
is the sum of various component potentials, but as we deal with horizontal
transport only, we will only take into account matric potential, reflecting the
forces by which the water is held by the soil matrix. The treatment below is taken,
with some modifications, from Stroosnijder (1976). Darcy's law reads:
(15.36)
where Kw is the hydraulic conductivity [M- I .L3 .T], Pr the total potential
[M.L- I .T-2 )], and Pm the matric potential. It is convenient and commonly done,
to express the potential in terms of length of water column. The flux equation
then becomes:
v = -KVh, (15.37)
where h is the pressure head [L], and K the hydraulic conductivity [L.r- I ].
Substitution of Eq. (15.3) and Eq. (15.37) into Eq. (15.35) results in:
Fig. 15.2a,b. Concentration of

inorganic P in soH solution (points
observed, lines calculated) at
varying distance from a root mat of
maize, at two soH moisture levels.
(After Gahoonia et al. 1994)
0.05
-0.05 0.05 0.15 0.25 0.35

a Distanee from root mat cm
0.5
•
0.45
0.4
0.35
":'13
u
0.3
":i
]
c::
o
0.25
•
o
'"
.5 0.2
~
0.15 • ••
0.1
0.05
-0.05 0.05 0.15 0.25 035

b Distanee from root mat em
10 Fig. 15.3. Comparison of

measured and ealeulated
influx of P for seven plant
speeies. a Calculations
without root hairs. b
Calculations taking the
7
presenee of root hairs into
'" aecount. (After Claassen
"'8u 1989)
6
"O
8
;';
b....,
"el
~ 4
"3
..sl
~
u
.'
o~--~----,-·-------,-------,-------,-------,
o 2 4 8 la
a measured 10- 14 moi em-1 S-1
10
""'"
""8
u
"O
8
;';
o
....,
"el
!l
~
..
"3
u
"@
u
.
'
10
b measured 10-14 moi em- 1 S-1

dS
dT =V·KCVh. (15.38)
As S = ee, where Ois the volumetric water content of the soil and since C
can be taken to be constant because of the low concentration of solutes, one
finally obtains:
(15.39)
To solve Eq. (15.39) the relation between Pm (or Hp ) and Oshould be known.
This relation is usually called the water retention curve. Now the water
diffusivity Dw( O) can be defined as:
dh
D =-K- (15.40)
W dO·
Substitution of Eq. (15.40) into Eq. (15.39) leads to:

dO
dr =VDwVO. (15.41)
In cylindrical co-ordinates with neglect of tangential and vertical gradients

term one obtains:
dO =.!~RD dO. (15.42)

dr R dR W dR
Because of the non-linear relations between Dw and O, and K and O, a solution
of Eq. (15.41) can usually be found only by numerical methods, though in
special cases analytical solutions can be found (e.g. Knight and Philip 1974).
75.2.5.2 Boundary Conditions
The same type of boundary conditions as with nutrient uptake apply, viz. given
flux or water content. The former is mostly used, where the flux at the root
surface is related to root length density and transpiration.
But contrary to uptake of nutrients, uptake of water is usually assumed to
be a passive process, a consequence of gradients in water potential in the path:
bulk soil-rhizosphere-root-Ieaf-atmosphere. The models dealt with here pertain
to a part of the path viz. to the transport of water from bulk soil to the rhizos-
phere and from there into the root. This leads to a boundary condition of the
form (CampbellI985; De Willigen 1990):
R=Ro, D w dO =~ , (15.43)
dR 2nRo
where Ro is the root radius (L), and FI is the uptake rate per unit length of
root (L2.rl ). The uptake rate is assumed to be proportional to the difference in
pressure between water in the rhizosphere and in the roots (the root water
potential):
(15.44)
where hRO is the pressure head of the rhizosphere (L), hp is the pressure head in
the root (L), and K p is the conductance of the root (L.T- I ). Equation (15.44) thus
describes the flow across the root surface. On the other hand, FI is also equal to
the transpiration rate per unit root length:
E
1\=--, (15.45)
ilzL rv
where E is the actual transpiration rate (L.T- I ), and Llz (L) is the layer thickness
(root length). Equation (15.45) describes the flow of water from bulk soil to the
root surface. From Eqs. (15.44) and (15.45) it follows that
(15.46)
The actual transpiration rate in its turn may depend on the root water
potential:
(15.47)
where Epal is the potential transpiration rate, as determined by atmospheric evap-
orative demand (L. T- I ) and f(h p) denotes the reduction of potential transpiration
due to root water potential. Campbell (1991) proposed the function:
1
f(h p ) = - - - (15.48)
1+ ( hp )a
hP,l/2
where hp.1/2 is the leaf water potential where E = Tpa l2, and the exponent a is a
species dependent constant.
When the level of dissolved salts in the soil solution becomes appreciable,
the difference in total soil moisture potential (including the osmotic as well as
the matric potential) between the soil and the root-water potential must be
taken into account. A detailed description of water and solute transport from
soil to root must consider both the differential hydraulic conductance and the
differential osmotic permeability of the various tissues and flow paths, with the
osmotic factor being particularly important at the endodermis.
Because the membranes of the roots exclude most solutes from the stream
of solution passing through them, there will be a build-up of solutes at these
membranes and hence the osmotic pressure at the membranes will be greater
than that of the extern al solution. Localised draw-downs of both matric and
osmotic components of soil moisture potential around the roots can be impor-
tant, particularly in the case of young plants and/or if the irrigation water or
the soil is saline. Water and solute transport through root systems or segments
of roots is more precisely expressed in the plant physiologicalliterature by the
relationship:
(15.49)
which derives from irreversible thermodynamics and is routinely used for
describing the flow of water across an individual semipermeable membrane
(Dainty 1976). llP is the drop in hydrostatic pressure across the membrane; (J'
is the reflection coefficient, which takes into account the permeability of the
membrane to a given solute; and lln: is the rise in osmotic pressure, across the
membrane. Equation (15.49) applies, strictly, only to a well-defined membrane
with uniform properties.
75.2.5.3 Analytical Solutions and Approximations
Contrary to the flux equation for solutes, which for our purposes can be
assumed to be linear, i.e. the diffusion coefficient and the flux of water are inde-
pendent of the solute concentration, the flux equations for soil water are highly
non-linear, the conductivity decreasing strongly with de cre as ing pressure head
and water content. Next to the non-linearity of the flux equation there is a non-
linear relation between the pressure head and the water content. Due to these
non-linearities, the transport equation generally has to be solved numerically.
However, some analytical approximations are mentioned in the literature and
will be discussed here.
Van den Honert (1948) introduced the attractive analogue of a electric al
current flowing through a series of resistances. For the analogy to apply, a
steady-state situation should exist. Gardner (1960) described the drying of soil,
due to root water uptake, as a sequence of steady-states.
For a homogeneous soil and horizontal transport, the non-linearity of the
flux can be removed by introducing the matric flux potential (Raats 1970),
defined by:
e
f f
h
eI> = Kdh = Dwde, (15.50)

href Bre!
where eI> is the matric flux potential (L2.r-1), hrej is a reference value of the
pressure head and the corresponding water content. This transforms Eq.
(15.37) into:
v = -VeI>. (15.51)
Next to the non-linearity of the flux equation there is a non-linear reiat ion
between the pressure head and the water content. By assuming a steady-rate
situation, i.e. supposing d()/dt = constant, Passioura (1980) obtained an analyt-

ical expres sion relating uptake of water to root length density and difference in
matric flux potential in the bulk soil and at the root surface:
E (p2 p2 -1 )
<1> -<1> = - - --lnp--'--- (15.52)
Rl Il" 27rL",H p2 -1 2(p2 -1) .
De Willigen and van Noordwijk (1987) have shown the distribution of water
content around the root calculated using Eq. (15.52) is in excellent agreement
with that calculated using a model, which solves Eq. (15.39) numerically. From
Eq. (15.52), the water flow from the bulk soil towards the root surface can be
expressed as a function of the average value of the matric flux potential in the
bulk soil and in the rhizosphere:
p2_ 1
}\= 2 4 [<1>-<1>Il"J. (l5.53)
- - + -P- n1p 2
1-3p
-
4 p2_ 1
75.2.5.4 Examples and Applications
Availability. As in the case of nutrients, the following definition of availability

is used: the available amount of water in the root zone is that amount of water
which can be taken up at the rate required to fulfil the demand of the plant for
water, if transport in the soil/root system were not limiting uptake, i.e. if the dif-
fusivity and root conductance were infinitely large. An infinitely small gradient
of pressure head would then suffice to transport water to the root surface and
from there into the root at the required rate. Suppose that the transpiration of
the plant is reduced when the pressure head of water in the plant equals a
certain value hp,c' In the absence of any gradient, the pressure head and water
content are the same everywhere in the soil, and the uptake of water by such
a root system can proceed unhampered until the value of the pressure head
in the rhizosphere equals hp,c with a corresponding water content of ()p,c' The
amount of available water W (L) in a soillayer with given water content ()i> ()p,c
and thickness ,1z (L) can thus be given as:
f R«()i - ()p", )dR

Rj
27r,1z
W= Il" =~Lrv«()i-()pJ(R~-R5). (15.54)

7rR~
Availability in this definition thus depends on the crop only through the
plant-specific value of hp,c' In reality, of course, the diffusivity and the root con-
ductance have finite values. This means that when the root water potential equals
hp,c> the pressure head in the rhizosphere equals (ef. Eqs. (15.44) and (15.45)):
(15.55)
Moreover, there will be a gradient in the water content around the root. The
unconstrained available water Wu which can be taken up at the required
rate, taking into account the transport possibilities of the soil/root system, is
defined by:
f f
~ ~
Wu = 27rL"A.z R(O; - O}1R = W - 2nL"A.z R(O - Op,;; }1R. (15.56)

Ro Ro
The integral on the right hand side of Eq. (15.56) can be evaluated using
Eq. (15.52), where <I>Ro is evaluated with (15.50) at h = hRO' with hRO given by Eq.
(15.55). The fractional depletion of the available water realised in the period of
unconstrained uptake is:
Wu
FDw =-=1
, W (15.57)
Figure 15.4 shows the fractional depletion as a function of root length

density for a Dutch clay soil (clayblO, Wosten 1987) with hp,c = -2 MPa, corre-
sponding to an amount of available water of 3.3 7. 10-2 m, or a maximum uptake
period of 6.74 days, Epot being 5.1O-3m day-l. The unconstrained availability and
uptake periods are zero, when the pressure head in the rhizosphere equals the
value corresponding to field capacity, with the parameter values used here at a
root length density of 2.51 103 mm-3. If root length density is only slightly
higher, the unconstrained fractional depletion increases fast, e.g. to 0.4 and 0.7
at root length densities of 3.103mm-3 and 6.103mm-3 respectively. Above a root
length density of 0.5 103m m-3 further increase in root length density causes a
limited increase of the fractional depletion, i.e. from 0.9 at L TV = 0.5cmcm-3 to
0.98 at L TV = 5 103mm-3. This means that doubling the total root length within
a layer, e.g. increasing the root length density by a factor two, is much less
favourable than maintaining the same average root length density and doubling
the root length, i.e. in the case discussed here distributing the total root length
over 0.4 rather than over 0.2 m.
Root and Soil Resistance. The single root uptake model can be used to asses the
relative importance of soil resistance with respect to root resistance. The root
resistance is calculated as:
(15.58)
Fractional depletion
0.8
0.6
0.4
0.2
o~------~--------~--------~--------~------~--------~
o 2 3 4 5 6
Lrv cm cm- 3
Fig. 15.4. Fractional depletion as a function of root length density for a clay soil. Critical root
water potential -2 MPa, potential transpiration rate 5.10-3 m day-l
where 0Jp is the resistance of the root (d), sa:

1
(J)p =--- (15.59)
KpLrvLlz
The soil resistance is, in analogy ta Eq. (15.58), equal ta the difference
between pressure head in the bulk soil, fi, and that found in the rhizosphere,
hRO, divided by the transpiration:
h -hRo
(J)s =--E-. (15.60)
Figure 15.5 shows the plant and soil resistance as a function of the average
pressure head (given as pF), for a root length densityof 103 mm-3, and trans-
piration of 5 10-3 m day-l, for a clay, sand, and peat soil in the Netherlands
(Wosten 1987), and a root conductance of 23.IO--4 m d- 1, the highest value found
in the literature (De Willigen and van Noordwijk 1987). Clearly, even with this
high value of root conductance, plant resistance exceeds the soil resistance for
pF-values up ta 3 (sand) ar even 4.5 (clay). This is an illustration of the con-
clusion of Molz (1981): "... it [the root resistance] seems certain ta dominate
in the upper 75% of the water content under normal rooting conditions".
Herckelrath et al. (1977) were the first ta quantitatively evaluate the effects
of limited root-soil contact for water uptake. The uptake rate in their approach
is multiplied with the water content relative ta the water content at saturation.
Fig. 15.5. Soi! and plant

resistance as a function of
ave rage pF-value [=log(hl]
SAND PEAT
of the soi!. Parameters: root
radius 0.01 cm, root length
20 cm, root length density
1 cmcm-3 , root conductance
23. 10-6 cm day-l
pF
The same procedure has been used by Hansen et al. (1991) and Bouten (1992).
The approximation in Eq. (15.32) can be used to estimate the extra resistance
in the soil due to incomplete contact, again the matric flux potential plays an
analogous role to the concentration. For the clay soil and the parameters of Fig.
15.5, the point where the soil resistance exceeds the root resistance shifts from
2.77 MPa in case of complete contact to 1.57 MPa when contact is 10%.
15.3 Modelling Uptake by a Crop Root System
15.3.1 Nutrients
For a root system consisting of regularly distributed roots confined to a single

layer of uniform initial nutrient concentrat ion, upscaling from a single root
model is rather trivial, the uptake rate per unit soil surface simply is the product
of the root length density and the uptake rate of a single root. Upscaling is
considerably more difficult, even for parallel identical roots confined to a single
soillayer, when roots or nutrient are not regularly distributed. Barley (1970)
proposed constructing a polygon around each root, consisting of the locus of
points in the soil nearer to that root than to any other. Such a construction
is called the Dirichlet tessellation (Green and Sibson 1977). Barley then
considered the uptake from a cylinder with the same area as the polygon.
Uptake rate was assumed to be proportional to the concentration at the
root surface. He found small effects of non-regular root distribution. Baldwin
et al. (1973), using an electrical analogue, found considerably larger effects. De
Willigen and van Noordwijk (1987) constructed a Dirichlet tessellation for root
maps made in the field, and calculated the potential uptake from equivalent
cylinders, as Barley had done, but also took into account the eccentric position
of the root in such a cylinder. They found no reduction in uptake for mobile
solutes like nitrate, but a reduction of 75% for a nutrient with an adsorption
constant of 20.
Rappoldt (1992) presented a highly interesting theory on the use of an
equivalent model system to which the relevant diffusion equation can be
applied. The model system consists of spherical, cylindrical or plane aggregates,
which do not represent the individual aggregates of the soil, however. The
length scales in the real and model system are the same. A model of diffusion in
soil aggregates is evaluated by solving a differential equation in the model system.
The overall result for the soil is then found as a weighted sum of the results for
the various length scales (ef. Sect. 7.3.7 on the root effectivity ratio Rper ).
Claassen and Barber (1976) presented a model of a growing root system,
where it was assumed that the roots do not interfere [boundary condition,
Eq. (15.18)], Le. it deals with a single root, the length of which is increasing in
time. Barber and Cushman (1981) extended this model to one where the bound-
ary condition Eq. (15.19) was used, but where root length density apparent1y
stayed constant.
Hoffland et al. (1990) simulated phosphate uptake for a regularly distrib-
uted root system where roots behave as zero-sinks, increasing in root length
density in time. With increasing density the radius of the soil cylinder allocated
to each root decreases and the concentration gradients in the soil volumes sur-
rounding the root are truncated. Newly grown roots sample the soil with the
largest concentration. The model was successfully applied to a pot experiment
with rape. Later (Hoffland 1991) this was extended to include exudation of
organic acids by the root system in order to explain the unexpected1y high
uptake when rock phosphate was applied.
Geelhoed et al. (1997) extended this model with a sink term for zero sink
uptake by root hairs. The zero sink influx term was taken from de Willigen and
van Noordwijk [1994b, ef. Eq. (15.29)]. The model appeared to give good results
for P-uptake of maize plants grown in a medium of quartz sand coated with
goethite, at least when the level of adsorbed phosphate was high (Fig. 15.6). Low
level uptake was severely underestimated, due to the large decrease in pH in the
course of the experiment, which was not incorporated into the model.
Most crop growth models contain a module which deals with uptake by the
root system. Van Keulen et al. (1975) used a numerical model to calculate the
uptake possibilities of a root system. They carne to the conclusion that in case
of non-adsorbed nutrients, like nitrate, under normal conditions availability is
almost 100%, even at low root length densities (see also Fig. 15.1). Based on such
calculations, in many crop growth models (Greenwood et al. 1985; van Keulen
and Seligman 1987; Rijtema and Kroes 1991) dealing with uptake of nitrate the
(distribution of) root length density does not play a role, the only relevant
parameter being the vertical extent of the root system.
140
120 •
'";'-100
=
<Il
•
'i5..
'O
~ 80
-
Q)
~
•
<Il
c..
;:s
- ••
p., 60
"O
Q)
<Il
~
u
<il
u 40
20 •
Fig.15.6. Observed and
calculated P uptake by maize O
plants. Uptake calculated
O 20 40 60 80 100 120 140
with zero sink Eq. (15.29).
Observed P uptake Ilffiol plant- 1
(After Geelhoed 1998)
In the models or modules where distribution of root length density is

taken into account, the root zone is divided into layers or compartments
within which the roots are assumed to be distributed regularly (e.g. Jones
and Kiniry 1986; Hansen et al. 1991; De Willigen and van Noordwijk 1995).
For calculation of the uptake rate the steady-state, steady-rate and zero-sink
approximations are used.
15.3. 1. 1 Examples and Applications
Stationary Root Systems. The assumption of a stationary root system in

one layer of soil is acceptable for phosphate uptake by a full-grown arable
crop. Due to the low mobility of phosphate ions, transport of P from the
plough layer to the deeper layers can be ignored, and uptake will mainly take
place from the plough layer (the upper 20 to 25 cm of the soil profile, Prummel
1957). The phosphate recommendation for arable crops in the Netherlands
is based on the so-called Pw-value of the plough layer. This value gives the
amount of phosphate, expressed in mg P20sl- 1 soil, soluble in water at an
extraction ratio of 60 volumes of water to 1 volume of soil (van der Paauw
1971). It was shown by de Willigen and van Noordwijk (1978) and van
Noordwijk et al. (1990) that the Pw-value can be calculated with reasonable
accuracy on the basis of 24 h P-adsorption isotherms and the total amount of
labile P as measured by anion exchange resins or iron-hydroxide-impregnated
filter paper.
Van Noordwijk et al. (1990) used Eq. (15.29) to calculate the minimum
required phosphate supply of the soil for maximum growth of a number of
crops and two groups of soils, and compared these with experimentally estab-
lished values for required phosphate supply. For each crop, representative values
for daily P demand and root length density were used. Although the agreement
between theoretical and experimental results was not 1: 1, the correlation
between the two appeared considerable. This at least indicates that the P level
in the soil necessary to meet the phosphate requirement of crops can be esti-
mated using Eq. (15.29).
Growing Root Systems. Three examples of models of uptake by a growing root

system will be treated here. These models are in fact modules and as such a part
of a model on crop growth, in which above ground growth and soil processes
like mineralization, leaching, etc. are also dealt with.
The first example is the nutrient uptake module of the Danish model DAISY
(Hansen et al. 1990, 1991). In this model the root depth and root density profile
are calculated on a daily basis and are used as input variables in the models for
water and nutrient uptake.
, 5 Modelling Water and Nutrient Uptake 533
The nitrogen uptake model is based on the concept of a potential nitrogen

demand simulated by the crop model, and the plant availability of soil nitro-
gen, i.e. the rate by which nitrogen can be made available at the root surfaces.
The transport of nitrogen from the bulk soil to the root surfaces is based on the
assumption that each root exploits an average effective volume of soil which is
a cylinder around each root. The radius of this cylinder corresponds to the
average half distance between the assumed parallel roots. The nit ro gen trans-
fer to the root surface takes place by mass flow and diffusion. It is assumed that
the concentration distance profile develops in time in a stepwise manner and,
at each time step, approximates to a steady state profile. In the present model it
is assumed that nitrogen uptake equals the nit ro gen flux towards the root
surface. Based on these assumptions, the flux 1 of nitrogen towards the root
surface is calculated as:
_ [ ln p21] p
2nD(C-C r) ---- v=O
p2 -1 2
(p2 -1)(' - G,l np 2
1= v=2,
qr (p2 -1)-l np 2
(p2 -1)(1- v)(' _ (p2-2V -l)C r
qr (2 )( ) (2-2V -1 ) else
p-11-v-p
where qr is the radial water flow rate (L 3 .L- 1.T- 1), and Cr is the concentration in
the soil solution at the root surface (M.L- 3).
If the uptake is limited by the availability of nitrogen, the concentrat ion
at the root surface is assumed to be equal to zero and hence the root will
act as a zero sink. In this case total uptake of nitrogen is calculated by
integrat ing the flux over the entire root system. In the case of ample nitrogen
supply, the total nitrogen uptake is determined by the potential nitrogen
demand. Then total uptake is distributed over the entire root zone by
assuming a common concentration to exist along the root surfaces of the
entire root system. Soil layers in which the concentration is less than the
common concentration are assumed not to contribute to nitrogen uptake.
The calculations are performed for both ammonium and nit rate. It is assumed
that ammonium is taken up by the plant roots in preference to nitrate. The
mobility of the ammonium in soil is considered less than that of nitrate
due to adsorption of ammonium to soi! colloids which is described by an
adsorption-desorption isotherm. The vertical movement of nitrogen is model-
led by means of a numerical solution of the convection-dispersion equation
for ammonium as well as for nitrate. The source sink term in the convection
dispersion equation integrates the transformation processes for ammonium as
well as for nitrate.
A second example pertains to the uptake module developed by De Willigen

and van Noordwijk (1987, 1995) applied in the crop growth model of CP-BKF
(Cultures Pluviales - Burkina Faso, AB DLO Haren). The potential uptake rate
where the root is supposed to act as a zero-sink is first calculated based on Eq.
(15.29), if the required uptake is less than the potential, the actual uptake rate
equals the required uptake rate, otherwise it equals the potential uptake rate.
Furthermore, in the module it is supposed, as indeed is often found, that roots
in favourable positions can compensate for roots in less favourable positions by
increasing their uptake rates (De Jager 1985). It is therefore assumed that the
uptake rate of roots in a certain layer depends on the uptake potential of roots
in other layers. The calculation procedure implies that roots growing under
favourable conditions will compensate as much as possible for roots growing
under less favourable conditions. It is thus assumed that information about the
necessary behaviour, as far as uptake is concerned, is instantaneously available
throughout the complete root system. The procedure described above can also
be applied when roots within a horizontallayer are not distributed regularly.
The layer then is divided over a number of compartments within which root
distribution can be assumed to be regular.
The third example stems from the much used CERES-model, in this case
the CERES-Maize (Jones and Kinity 1986). Here the potential uptake of nitrate
from a certain layer is calculated as:
(15.61)
where U is uptake in kgha- I day-I, Lv,; is the root length density in cmcm-3,
lXw is a reduction factor depending on relative water content, Azi is the
thickness of the layer in cm, the maximum uptake rate per unit root length
(mgcm-1d- I) and Fav an availability factor depending on the nitrate content
(NN03,i in mgkg- I):
Fav,i = l-exp(-0.03* NN03,;). (15.62)
The factor 100 is for conversion to kgha- I day-I. Total uptake is calculated
by summing the uptake rates from the individuallayers. Then a nitrate uptake
factor is computed as the ratio of demand and total potential uptake. If this
factor exceeds 1, the actual uptake equals the potential uptake, otherwise the
actual uptake is calculated as the potential uptake multiplied by the nitrate
uptake factor.
For a situation as given in Table 15.2, and a plant demand of 4kgha- I
day-I, the uptake distribution was calculated using the three models. Results are
given in Fig. 15.7, where it is shown that the Ritchie formulation leads to a total
uptake rate of less than half the demand, whereas the two other uptake models
do satisfy demand completely or almost.
Table 15.2. Distribution of roots, nitrogen and water with depth as used for the calculations
of distribution of nitrogen uptake by three different uptake models
Layer Root length density Nitrogen content Water content

em em cm-3 kg ha- 1
0-20 1. 20 0.2
20-40 0.5 10 0.2
40-60 0.25 8 0.2
60-80 0.12 5 0.2
80-100 0.06 2 0.2
4.0
w H
3.5
-:.o
";'~
3.0
..c
0Jl
..I<i 2.5
~
~
...
-
~
2.0
- - - - - -R
..I<i
~
o..
-
;::1
.:::
~
1.5
ro
'3
8;::1
U 1.0
Fig. 15.7. Cumulative uptake
rate with depth for the situation
given in Table 15.2, as 0.5
caleulated with the uptake
module of Hansen et al. (1991;
indicated with H) the Ceres 0.0
model (R) and that of de O 50 100
Willigen and van Noordwijk
(1995) (W) Depth em
15.3.2 Water
Obviously, as with nutrients, upscaling from a single root to a set of regularly

distributed parallel roots is simply a matter of multiplying the root length
density by the uptake rate. Lafolie et al. (1991) used a numeric al model to
calculate uptake of water by a (stationary) root system consisting of parallel
non-regularly distributed roots. A Dirichlet tessellation was constructed from
root maps, and based on this a triangle finite element grid was generated.
As with nutrients, one can distinguish two types of uptake models. In the
first group where root length density and root water potential do not play a
role, uptake is determined by the depth of the root zone and the pressure head
or water content of the soil layers (e.g. Feddes et al. 1978; van Keulen and
Seligman 1987), the other is based upon steady-state or steady-rate approxi-
mations of the single root model (e.g. Nimah and Hanks 1973; Hillel et al. 1976;
De Willigen and van Noordwijk 1995).
75.3.2. 7 Examples and Applications
A first example is the water uptake module as used in the SWATRE model
(Feddes et al. 1978). First, a potential water uptake rate is calculated, which is
corrected for the pressure head in the soil. In its simplest form the potential
water extraction rate (5 maJ is assumed to be equally distributed over the root
zone (Zr):
(15.63)
Later (Bouten 1992), the potential uptake was distributed according to the rel-
ative distribution of the root length density:
(15.64)
The actual uptake rate (5) in a certain layer depends on the pressure head of
that layer:
5 = a(h)5 max• (15.65)
De Willigen and van Noordwijk (1995) used a water uptake model (WATUP)
based on an extension of the single root model treated in Section 15.2.5. Again,
for a root system distributed over several soillayers (or compartments), they
assumed that within each layer (compartment) the roots are regularly distrib-
uted and that again in each layer flow of water from the bulk soil to the rhi-
zosphere equals that from the rhizosphere into the root:
(15.66)
where the subscript i denotes the number of the layer. AIso, the total uptake
should equal transpiration:
LL
N
rvJf1z;Kp(h Ro ,; -hp)= Eaet =!(hp)E pot ' (15.67)
This yields N + 1 equations with N + 1 unknowns: the N values for hRo,; and the
root water potential hp • Again, the equations have to be solved by iteration
because of the non-linear relation between <l> and h, and the non-linear nature
of f(h p ). For a relatively wet soil the major resistance is found in the root (Fig.
15.5), and the gradients in the soil are accordingly very small, i.e. <l>Ro,; "" <Îl;, and
hp is much larger than h;. The uptake from a certain layer is then proportional
to the relative root length density in that layer as in Eq. (15.64). So for relatively
wet conditions, both models yield the same distribution of uptake.
The models are applied to results of an experiment of Sharp and Davies
(1985). They measured root length and water uptake distribution of maize
plants. After a preliminary growth period, half of the plants were watered daily,
the other half were not watered at alI. Figure 15.8a shows the results of the welI-
watered treatment. The results show that measured water uptake is more or less
proportional to root length density as predicted by WATUP and SWATRE with
a potential sink term as given in Eq. (15.64). The sink term in Eq. (15.63) leads
to a uniform distribution of water uptake.
Figure 15.8b shows the uptake distribution for the treatment where no
water was given during 18 days. Now the distribution of uptake is not deter-
mined by distribution of root length density but by distribution of pressure
head, as presumably the resistance in the soil now exceeds that in the roots. The
three models yielded more or less the same distribution as found in the exper-
iment although the measured transpiration amounted to 0.93 cm day-l, WATUP
calculated 0.85cm day-l, and SWATRE, using Eq. (15.63), yielded 0.62cm day-l,
and using Eq. (15.64) arrived at 0,49cm day-l.
15.4 Conclusions and Suggestions for Further Research
Though it can be said that we have come a long way in modelling uptake by
root systems since the beginning in the eariy 1960s, a large number of uncer-
tainties stiH exist. One of the difficulties is the validation of the uptake models
as such in field situations. EspecialIy in the case of uptake of nutrients, it is
compulsory to consider other processes as well e.g. mineralization, dissolution,
';"
>o-
0.6
1\1
"C 0.5
E
u 0.4
-...
~
1\1 0.3
c..
SI
-
:1
0.2
al
~ 0.1 M
O
O 20 40 60 80
a
... 0.6
Depth cm
~
"C
0.5 W
E
u 0.4 M
-...
~
1\1 0.3 SI
D..
:1 0.2 S2
-
al 0.1
~ O
-0.1
O 20 40 60 80
b Depth cm
Fig. 1 S.8. a Distribution of water uptake with depth as observed (Sharp and Davies 1985, M),
and calculated by SWATRE (Feddes et al. 1978, S1), modified SWATRE (Bouten 1992, S2) and
WATUP (de Willigen and van Noordwijk 1995, W). Well-watered treatment. b As for a but during
18 days no water was given
aboveground growth, distribution of dry matter etc. Even when uptake is

predicted satisfactorily this could be due to compensating errors. Moreover,
differences between models when acting at a certain instant as shown in
Section 15.3.1, may disappear when considered during longer periods due
to feedback mechanisms, which makes it difficult to decide which is the
"correct" one.
Many models perform reasonably when simulating uptake in a relative rich
environment, and do much worse under poor conditions (De Willigen et al.
1998), this seems especiallythe case for uptake of phosphate.Amijee et al. (1991)
state: "... in soils of low P concentration ... the predicted value always under-
estimates the observed value". This has to do with the many possible adapta-
tions of the roots and the root system, apart from simply increasing its size and
exploring new areas, which are not implemented in most models. Many P-
starved plants exude organic acids, thereby influencing the adsorption proper-
ties of the soil with respect to phosphate in such a way that availability is
increased.A fewmodels (e.g.Hoffland 1991; Geelhoed 1998) have included these
processes. Other processes triggered by root activity are temporal immobilisa-
tion of N and P by micro-organisms, and production and exudation of iron
mobilising substances. Darrah (1993) gives a nice review of the relative impor-
tance of such processes. Future models should, where relevant, include more
details of the processes in the rhizosphere and their interaction with root activ-
ity and functioning.
Another interesting area, but now at the scale of a root system rather than
of a single root, is that of the influence of root architecture on efficiency of
uptake. Fitter (1987) estimated this by combining his topological model with
the depletion zone as defined by Baldwin and Nye (1974). Doussan et al.
(1998a,b) coupled a model ofroot system architecture (Pages et al. 1989) with
laws describing flow of water into and within roots. They used the model to
investigate the influence of conductance varying with time on the distribution
of water potential in a maize root system. They found that considerable gradi-
ents in water potential could develop. It is to be expected that coupling of archi-
tectural models with appropriate uptake models wiH be one of the main
activities in the field of modelling of uptake and functioning of root systems,
but the influence of growth and decay of roots should also be dealt with.
On a stiHlarger scale, there are models describing the competition between
root systems of different species, which is of relevance for e.g. agroforestry
systems or where interaction between crops and weeds is studied. Recently, two
models investigating agroforestry systems have been published, Wimisa (Mayus
1998) and Wanulcas (Van Noordwijk and Lusiana 1998), the first one dealing
with competition between trees and a crop for water, the second also dealing
with competition for nutrients. Wimisa has been put to the test with reasonable
results, Wanulcas is stiH at the development stage. Developing and validating
such complicated models offer a major challenge to future research.
References
Abramowitz M, Stegun IA (1970) Handbook of mathematical functions. Dover Publications,

New York
Altemiiller HJ, Haag Th (1983) Mikroskopische Untersuchungen an Maiswurzeln im ungestiirten
Bodenverband. Kali-Briefe (Biintehof) 16 (6): 349-363
Amijee F, Barradough PB, Tinker PB (1991) Modelling phosphorus uptake and utilisation by
plants. In: Johansen C, Lee KK, Sahrawat KL (eds) Phosphorus nutrition of grain legumes
in the semi-arid tropics, pp 63-77 ICRISAT, Pabanchever, Judia
Baldwin JP, Nye PH (1974) A model to calculate the uptake by a developing root system or root
hair system of solutes with concentration variable diffusion coefficients. Plant Soil 40:
703-706
540 P. de Willigen et a!.
Baldwin JP, Nye PH, Tinker PB (1973) Uptake of solutes by multiple root systems from soi!.
III. A model for calculating the solute uptake by a randomly dispersed root system
developing in a finite volume of soi!. Plant Soil38: 612-625
Barber SA (1962) A diffusion and mass flow concept of soi! nutrient avai!ability. Soil Sci 93:
39-48
Barber SA (1984) Soil nutrient bioavai!ability. A mechanistic approach. Wi!ey, New York
Barber SA, Cushman JH (1981) Nitrogen uptake model for agronomic crops. In: Iskandar IK
(ed) Modelling wastewater renovation. John Wi!ey, New York, pp 382-410
Barley KP (1970) The configuration of the root system in reiat ion to nutrient uptake. Adv Agron
22: 159-201
Bear J (1972) Dynamics of fluids in porous media. American Elsevier, New York, 764 pp
Bolt GH, Bruggenwert MGM (1979) Composition of the soi!. In: Bolt GH, Bruggenwert MGM
(eds) Soi! chemistry A. Basic elements. Elsevier, Amsterdam, 1979, pp 1-11
Bouldin DR (1961) Mathematical description of diffusion processes in the soi!-plant system. Soi!
Sci Soc Amer Proc 25: 476-480
Bouten W (1992) Monitoring and modelling forest hydrological processes in support of
acidification research. Doct Thesis, University of Amsterdam, Amsterdam
Bray RH (1954) A nutrient mobility concept of soi! plant relationships. Soi! Sci 78: 9-22
Campbell GS (1985) Soi! physics with BASIC. Elsevier, Amsterdam
Campbell GS (1991) Simulation of water uptake by plant roots. In: Hanks 1, Ritchie JT (eds)
Modeling plant and soil systems. Agronomy Monograph 31, American Society of
Agronomy, Madison, Wisconsin, pp 273-285
Carlslaw HS, Jaeger JC (1959) Conduction of heat in solids. Oxford University Press, Oxford, 510
pp
Claassen N (1989) Die Aufnahme vom Năhrstoffen aus dem Boden durch die hiihere Pflanze als
Ergebnis von Verfiigbarkeit und Aneignungsvermiigen. Habilitationsschrift, Georg-August
Universităt, Giittingen, 327 pp
Claassen N, Barber SA (1976) Simulation model for nutrient uptake from soi! by a growing plant
root system. Agron J 68: 961-964
Cushman JH (1979) An analytical solution to solute transport near root surfaces for low initial
concentration: 1. Equations development. Soil Sci Soc Am J 43: 1087-1090
Cushman JH (1980) Complet ion ofthe list of analytical solutions for nutrient transport to roots.
1. Exact linear models. Water Resour Res 16: 891-896
Dainty J (1976) Water relations of plant cells. In: Liittge U, Pitman MG (eds) Transport in plants.
II.A Cells. Encyclopaedia of Plant Physiology. New Series, voi 2. Springer, Berlin Heidelberg
New York, pp 12-35
Darrah PR (1993) The rhizosphere and plant nutrit ion: a quantitative approach. In: Barrow NJ
(ed) Plant nutrition from genetic engineering to field practice. Kluwer, Dordrecht, the
Netherlands, pp 3-23
De Jager A (1985) Response of plants to localised nutrient supply. PhD Thesis, University of
Utrecht, Utrecht, 137 pp
De Willigen P (1981) Mathematical analysis of diffusion and mass flow of solutes to a root
assuming constant uptake. Inst Bodemvruchtbaarheid, Rapport 6-81, 56 pp
De Willigen P (1990) Calculat ion of uptake of nutrients and water by a root system. Nota 210,
Institute for Soi! Fertility, Haren, the Netherlands
De Willigen P, van Noordwijk M (1978) Calculations on phosphate uptake by a crop in reiat ion
to rooting density and P-status of the soi! (In Dutch). Inst Bodemvruchtbaarheid Rapport
19-78,63 pp
De Willigen P, van Noordwijk M (1987) Roots, plant production and nutrient use efficiency,
PhD Thesis, Wageningen Agricultural University, Wageningen
De Willigen P, van Noordwijk M (1994a) Mass flow and diffusion of nutrients to a root with
constant or zero-sink uptake.1. Constant uptake. Soil Sci 157: 162-170
De Willigen P, van Noordwijk M (1994b) Mass flow and diffusion of nutrients to a root with
constant or zero-sink uptake. II. Zero-sink uptake. Soi! Sci 157: 171-175
De Willigen P, van Noordwijk M (1995) Model for interactions between water and nutrient
uptake. In: Kabat P, Marshall B, van den Broek BJ, Vos J, van Keulen H (eds) Modelling and
parametrization of the soil-plant-atmosphere system, a comparison of potato growth
models. Wageningen Press, Wageningen, pp l35-153
De Willigen P, Dijksterhuis G, Chardon W (1998) Dynamique et disponibilite du Phosphore dans
les sols. In: Breman H, Sissiko K (eds) L'intensification agricole au Sahel. Karthala, Paris.
pp 266-284
Doussan C, Vercambre G, Pages L (1998a) Modelling of the hydraulic architecture of root
systems: an integrated approach of water absorption - model description. Ann Bot 81:
2l3-223
Doussan C, Vercambre G, Pages L (1998b) Modelling of the hydraulic architecture of root
systems: an integrated approach of water absorption - distribution ofaxial and radial
conductances in maize. Ann Bot 81: 225-232
Dunham RJ, Nye PH (1973) The influence of soil water content on the uptake of ions by roots.
1. Soil water content gradients near a plane of onion roots. J Appl Ecol lO: 585-598
Dunham RJ, Nye PH (1974) The influence of soil water content on the uptake of ions by roots.
II. Chloride uptake and concentration gradients in soil. J Appl Ecolll: 581-596
Feddes RA, Kowalik PJ, Zaradny H (1978) Simulation of field water use and crop yield.
Simulation monographs, PUDOC, Wageningen, 189 pp
Fiiter AH (1987) An architectural approach to the comparative ecology of plant root systems.
New Phytol. (Suppl.) 106: 61-77
Gahoonia TS, Nielsen NE (1991) A method to study rhizosphere processes in thin soillayers of
different proximity to roots. Plant Soil l35: 143-146
Gahoonia TS, Nielsen NE (1992a) Control of pH at soil root interface. Plant Soil140: 49-54
Gahoonia TS, Nielsen NE (1992b) The effect of root-induced pH change on the depletion of
inorganic and organic phosphate. Plant Soi! 143: 185-191
Gahoonia TS, Raza S, Nielsen NE (1994) Phosphorus depletion in the rhizosphere as influenced
by soil moisture. Plant Soil159: 2l3-218
Gardner WR (1960) Dynamic aspects of water availability to plants. Soil Sci 89: 63-73
Geelhoed J (1998) Phosphateavailability in the soil-root ysytem: integration of oxide
surface chemistry, transport and uptake. PhD Thesis, Wageningen Agricultural University,
177 pp
Geelhoed JS, Findenegg GR, van Riemsdijk WH (1997) Availability to plants of phosphate
adsorbed on goethite: experiment and simulation. Eur J Soil Sci 48: 473-481
Green PJ, Sibson R (1977) Computing Dirichlet tessellations in the plane. Comput J 21:
168173
Greenwood DJ, Neeteson JJ, Draycott A (1985) Response of potatoes to N fertiliser. Plant Soil85:
163-183
Hansen S, Jensen HE, Nielsen NE, Svendsen H (1990) DAISY: A Soil Plant System Model. Danish
simulation model for transformation and transport of energy and matter in the soil plant
athmosphere system. The national Agency for Environmental Protection, Copenhagen,
369 pp
Hansen S, Jensen HE, Nielsen NE, Svendsen H (1991) Simulation of nitrogen dynamics and
biomass production in winter wheat using the Danish simulation model DAISY. Fert Res
27:245-261
Herckelrath WM, Miller EE, Gardner WR (1977) Water uptake by plants. II The root contact
model. Soi! Sci Soc Am Proc 41: 1039-1043
Hillel D, Talpaz H, Van Keulen H (1976) A macroscopic-scale model ofwater uptake by a nonuni-
form root system and of water and salt movement in the soi! profile. Soi! Sci 121: 242-255
Hoffland E (1991) Mobilisation of rock phosphate by rape. PhD Thesis, Wageningen
Agricultural University, Wageningen
Hoffland E, Bloemhof HS, Leffelaar PA, Findenegg GR, Nelemans JA (1990) Simulation of
nutrient uptake by a growing root system considering increasing root density and
interroot competition. Plant Soi! 124: 149-155
Jones CA, Kinity JR (eds) (1986) CERES-maize: a Simulation model of maize growth and
development. Texas A&M University Press, College Station, 194 pp
Jones JW, Ritchie JT (1991) Crop growth models. In: Hoffman GJ, Howell TA, Solomon KH (ed)
Management of farm irrigation systems. ASAE, St. Joseph, MI. pp 63-89
Jungk A, Claassen N (1989) Availability in soi! and acquisition by plants as the basis for
phosphorus and potassium supply to plants. Z Pflanzenernaehr Bodenkd 152: 151-
157
Knight J, Philip JR (1974) Exact Solutions in non linear diffusion. J Eng math 8: 219-227
Lafolie F, Bruckler L, Tardieu F (1991) Modelling root water potential and soi!-root water
transport: 1. Model presentation. Soil Sci Soc Am J 55: 1203-1211
Mayus M (1998) Millet growth in windbreak -shielded fields in the Sahel, experiment and model.
PhD Thesis, Wageningen Agricultural University, Wageningen
Molz FJ (1981) Models of water transport in the soi!-plant system: a review. Water Resour Res
17: 1245-1260
Nielsen NE (1983) Plant parameters controlling the efficiency of nutrient uptake from soil. In:
United Nations Economic Commission for Europe (ed) efficient use of fertilisers in agri-
culture. Developments in Plant and Soil Sciences, voi 10. Martinus Nijhoff, Dordrecht, the
Netherlands, pp 199-217
Nimah A, Hanks RJ (1973) Model for estimat ing soil water, plant, and atmospheric interrela-
tions.I Description and sensitivity. Soi! Sci Soc Amer Proc 37: 522-527
Nye PH (1966) The effect of the nutrient intensity and buffering power of a soi!, and the absorb-
ing power, size and root hairs of a root, on nutrient absorption by diffusion. Plant Soil 25:
81-105
Nye PH (1992a) Towards the quantitative control of crop production and quality. 1. The role of
computer models in soi! and plant research. J Plant Nutr 15 (6&7): 1131-1150
Nye PH (1992b) Towards the quantitative control of crop production and quality. II. The
scientific basis for guiding fertiliser and management practice, particularly in poorer
countries. J Plant Nutr 15 (6&7): 1151-1173
Nye PH (1992c) Towards the quantitative control of crop production and quality. III. Some
recent developments in research into the root-soi! interface. J Plant Nutr 15 (6&7):
1175-1192
Nye PH, Marriott FH C (1969) A theoretical study of the distribution of substances around roots
resulting from simultaneous diffusion and mass flow. Plant Soi! 30: 459-472
Nye PH, Tinker PB (1977) Solute movement in the soi!-root system. Sudies in Ecology voi 4.
Blackwell Scientific Publications, Oxford, 342 pp
Olsen SR, Kemper WD (1968) Movement of nutrients to plant roots. Adv Agron 20: 91-151
Pages L, Jordan MO, Picard D (1989) Simulation of the three-dimensional architecture of the
maize root system. Plant SoH1l9: 147-154
Passioura JB (1980) The transport of water from soil to shoot in wheat seedlings. J Exp Bot 31:
333-345
Prummel J (1957) The significance of the thickness of the arable layer for the evaluation of soil
analysis data. (In Dutch). Landbouwkd Tijdschr 69: 703-7l2
Raats PAC (1970) Steady infiltration from line sources and furrows. SoH Sci Am Proc 34: 709-7l4
Rappoldt C (1992) Diffusion in aggregated soil. PhD Thesis, Wageningen Agricultural Univer-
sity, 162 pp
Rengel Z (1999) Mineral nutrition of crops. Food Products Press, New York, 399 pp
Rijtema PE, Kroes JG (1991) Some results of nitrogen simulations with the model ANIMO. In:
Groot JJR, de Willigen P, Verberne ELJ (eds) Nitrogen turnover in the soH-crop system.
Kluwer, Dordrecht, pp 189-199
Sharp RE, Davies WJ (1985) Root growth and water uptake by maize plants in drying soil. J Exp
Bot 36: 1441-1456
Stroosnijder L (1976) Infiltratie en herverdeling van water in grond. PUDOC, Wageningen, 213
pp
Turner NC (1974) Stomatal behaviour and water status of maize, sorghum and tobacco under
field conditions. II At low water potential. Plant Physiol 53: 360-365
Van den Broek BJ (1992) Predicting times of irrigation with a simulation model. M Sc Thesis,
University of Alberta, Alberta, 122 pp
Van den Honert TH (1948) Water transport in plants as a catenary process. Discuss Faraday Soc
3: 146-153
Van der Paauw F (197l) An effective water extraction method for the determination of
plant-available soH phosphorus. Plant Soil34: 467-481
Van Keulen H, Seligman NG (1987) Simulation of water use, nitrogen nutrition and growth of
a spring wheat crop. PUDOC, Wageningen, Simulation Monographs, 310 pp
Van Keulen H, Seligman NH, Goudriaan J (1975) Availability of anions in the growth medium
to roots of an actively growing plant. Neth J Agric Sci 23: 131-138
Van Noordwijk M, Lusiana B (1990) WaNulCAS 1.1. Background of a model of water nutrient
and light capture in agroforestry systems. Working Document, International Centre for
Research in Agroforestry (ICRAF), Bogor, Indonesia
Van Noordwijk M, de Willigen P, Ehlert PAI, Chardon WJ (1990) A simple model of P uptake as
a possible basis for P fertiliser recommendations. Neth J Agric Sci 38: 317-322
Van Wijk WR (1963) Physiscs of plant environment. North-Holland publ. Co., Amsterdam,
382 pp.
Wosten JHM (1987) Beschrijving van de waterretentie- en doorlatendheidskarakteristieken uit
de Staringreeks met analytische functies. Stiboka rapport 2019, Stiboka Wageningen, 53 pp
Youngs EG, Gardner WR (1963) A problem of diffusion in the infinite hollow cylinder. Soil Sci
Soc Am Proc 27: 475-476
CHAPTER 16
Plant Anchorage
A.R. Ennos l and S. Pellerin 2*
1 The University of Manchester, School of Biological Sciences, 3.614 Stopford Building, Oxford
Road, Manchester, M13 9PT, UK
2 INRA, Domaine de la Grande Ferrade, B.P. 81,71 Av. Ed. Bourleaux, 33883 Villenave d'Ornon
Cedex,France
CONTENTS
16.2 Mechanics of Anchorage 547
16.2.1 Resistance to Uprooting 547
16.2.1 Resistance to Lodging and Windthrow 548
16.2.2.1 Systems Transmitting Vertical Forces to the Soi! 548
16.2.2.2 Systems Transmitting Horizontal Forces to the Soil 549
16.3 Methods for Measuring Anchorage Components 551
16.3.1 Measuring the Relative Contributions of Anchorage
Components 551
16.3.2 Measuring Root Properties 553
16.3.2.1 Tensile Tests 553
16.3.2.2 Bending Tests 555
16.3.3 Measuring Soi! Properties 556
16.4 Methods for Evaluating Plant Anchorage Taken as a Whole 558
16.4.1 Direct Methods 558
16.4.1.1 Vertical Tests 558
16.4.1.2 Horizontal Pushing or Pulling Tests 559
16.4.2 Indirect Methods 559
16.5 Conclusion 560
References 562
Further Reading 565
16.1 Introduction
Together with water and nutrient absorption, a major function of roots is to

anchor plants. This chapter is dedicated to methods used for studying root
systems when consider ing this particular function.
* Principal author

546 A. R. Ennos and S. Pellerin
Under natural conditions, plants may be subjected to two types of external

forces which could cause failure to the anchorage: lateral or vertical pulling by
a grazing herbivore, or lateral pushing by wind. The resistance of plants to
uprooting inftuences the durability of grazed pastures, whereas resistance to
lodging of arable crops such as wheat, maize or rape has consequences on yield
(Carter and Hudelson 1988; Minami and Ujihara 1991), harvest quality (in the
case of silage maize for instance) and time required for harvesting operations.
Similarly, damage by windthrow may cause huge economic losses for forestry.
Methods for evaluating root anchorage are therefore needed in these different
contexts: resistance to uprooting of grassland species; resistance to lodging of
arable crops; and resistance to windthrow of trees.
Resistance of grassland species to uprooting has been rarely studied.
Devices for measuring forces exerted by grazing herbivores were, however,
proposed by Hughes et al. (1991) and Laca et al. (1992). In contrast, lodging of
arable crops and windthrow of trees have given rise to an abundant literature.
Both lodging and windthrow are the consequence of very complex phenomena
involving airftow above and within the canopy, the mechanics of plants under
wind loading, and their anchorage characteristics (Baker 1995). Moreover, it has
been shown that plants exhibit adaptive growth when subjected to mechanical
stimulation, which modifies both shoot and root morphology (Stokes et al. 1995;
Crook and Ennos 1996a; Goodman and Ennos 1996,1997). Lodging or wind-
throw may happen to healthy plants, or to plants previously weakened either
by remobilisation of carbohydrates from the stalk or roots, or by attacks of
pathogens or pests. In some cases, stems or trunks may break or buckle, so that
the problem is dependant on the mechanical properties of the stern or trunk.
In most cases, however, plants are displaced from the vertical or uprooted
without any stern damage occurring, so that the anchorage of the plant is clearly
involved. In either case, resistance to lodging or windthrow involves the
dynamic equilibrium between the aerial part, which is subject to wind, and the
soil-root system, which anchors the plant.
The mechanics of trees, and to a les ser extent arable crops, under wind
loading has given rise to numerous studies (Coutts and Grace 1995; Flesch
and Grant 1992a,b). Models exist for calculating the transfer of wind momen-
turn into canopies. The aerial part of individual plants is subjected to wind
pressure and also creates a torque when displaced from the vertical because of
its own weight. The characteristics of plants which are involved therefore
include the area of the leaf and stern, their vertical distribution, and drag
coefficient; the vertical distribution of their weight; and the mechanical prop-
erties of the stems. Plant root development, and the way in which roots and soil
anchor plants has received less attention than what occurs above-ground.
Pioneer studies have only relatively recent1y been carried out on trees and
arable crops.
16 Plant Anchorage 547
This chapter focuses on methods used for measuring plant anchorage. The
reader must bear in mind, however, that most anchorage studies are only valu-
able if combined with investigations on aerial parts of the plant. Basic descrip-
tions of the mechanics of plant anchorage are given in Section 16.2. Available
methods for measuring anchorage components, or anchorage as a whole, are
given in Sections 16.3 and 16.4.
16.2 The Mechanics of Anchorage
If we are to prevent catastrophic failure of crop plants it is c1early important to

understand exact1y how roots anchor plants in the soil. Unfortunately, the
mechanics of anchorage are extremely complex, even when a plant is simply
being pu1led out of the ground. As we shall see, anchorage strength depends not
only on the strength of individual roots, but also on their length, distribution
and branching pattern, and also on the properties of the soil. Changes in each
of these factors can affect not only the anchorage strength but also the pattern
of failure itself.
16.2.1 Resistance to Uprooting
Roots resist uprooting in just the same way as tent pegs (Ennos 1989, 1990);
friction transfers tension from the roots into the surrounding soil. Such a
system can fail in several ways depending on the size and strength of the root
and the strength of the soil. If a root is short and strong and the soil weak it
will pull out of the ground. In contrast, a long, weak root in strong soil will break
before the soil fails. In practise, unstrengthened roots more than a few mil-
limetres in length would break long before being pulled out of the soil, and at
very low forces.
Most c1imbing and procumbent plants which have to resist being uprooted
have their anchorage systems improved over a single unstrengthened root in
three ways (Ennos 1991a, 1993). First, the friction between root and soil is
increased by possession of root hairs which are glued to soil partic1es. However,
even then the rate at which tension can be transferred is ultimately limited by
the shear strength of the soil. Hence in weak, wet soil friction is stiH transferred
slowly. Second, their roots are strengthened towards their base where they are
most highly stressed. Third, most of these plants produce fibrous root systems
using large numbers of adventitious roots or a highly branched primary root
to produce a root system with a large surface area. In these ways, therefore, the
anchorage strength is maximised for a minimum investment in strengthening
material.
In practise, the failure of real root systems occurs by a combination of three

modes offailure (Ennos 1991a, 1993). When a plant is pulled up the soil fails in
tension near the base of the stern and a baU of soil attached to the basal roots
is broken off. Distal to this further failure occurs. Some of the longer or weaker
roots fail in tension. The break is sometimes clean but often the inner stele puUs
out from the cortex. However, the shorter or stronger ones are puUed out of the
soil intact.
Exactly where failure occurs depends on the soil conditions. The shear
strength of hard, dry soil is high so tension will be transferred rapidly from
roots to soil. The roots will mostly snap ne ar their base and only a smaU root
baU will break free with the stern base. The uprooting resistance will be high
and in extreme conditions will exceed the strength of the stern. In contrast, soft,
wet soil has a low shear strength, so tension will be transferred only slowly into
the soil and failure will occur much further away from the stern base. More roots
will be puUed out intact from the soil together with a larger root baU, and the
force required to uproot the plant will be much lower. This is why soil around
plants should be dampened before transplant ing them.
16.2.2 Resistance to Lodging and Windthrow
The mechanics of anchorage of self-supporting plants against lodging or wind-

throw is even more complex since the root system must transmit a torque into
the soil rather than a simple upward force. There are many possible designs for
such an anchorage system and the manner in which they fail is even more
dependent on the properties of the soil in which the plant is growing. However,
there is one fundamental aspect of the design of aU these root systems; each
contains at least one rigid element without which a torque could not be trans-
mitted (Ennos and Fitter 1992; Ennos 1993). A plant with a fibrous root system
would simply coUapse if its stern was not propped up. The rigid element aUows
the soil to be loaded in one of two ways. In some systems an upward force is
transmitted to the soil on the windward side of the stern and an equal down-
ward force to the soil on the leeward side. Alternatively, upper soillayers can be
loaded to the leeward side while lower soillayers are loaded by an equal wind-
ward force. Both patterns of forces produce a net torque which props up the
plant and, if failure does occur, the mode will depend on the relative ability of
the roots and soil to withstand these forces.
16.2.2. 1 Systems Transmitting Vertical Forces to the Soil
At one extreme of the range of anchorage systems which transmit vertical forces
to the soil are plants which have a broad, spreading root system with many hor-
izontallateral roots and some vertical sinkers. Such plants include many trees
including spruce (Coutts 1983,1986), larch (Crook and Ennos 1996b), poplar
and many buttressed rainforest trees. In these spreading systems the resistance
of the soil to downward movement of the laterals will be very high because of
the large are a of the root system and the high compressive strength of soil. For
these reasons failure occurs mostly on the windward side, the plant being
levered up about a hinge on the leeward side (Fig. 16.lA). Exactly what happens
depends on the soil conditions, just as it does during uprooting of climbers. In
wet soil a large root plate may be levered out of the ground together with many
sinkers which are pulled out intact. In contrast, in drier, stronger soil a smaller
root plate may be formed, windward roots may be broken and sinkers sheared
off. In such systems three components of anchorage may be identified (Fig.
16.lA): the resistance of the leeward hinge to bending; the resistance of the
windward side to uprooting and the weight of the root -soil plate.
At the other extreme of the range of anchorage types are plants which have
a narrower root system with large numbers of nearly vertical roots. Such
systems are found in the coronal root systems of winter wheat (Crook and
Ennos 1993) and the prop root systems of maize (Ennos et al. 1993b). In such
systems the resistance of the windward roots to being pulled out of the soil is
higher than the combined resistance of the soil beneath the narrow root plate
to being compressed and the leeward roots to buckling. As a result, the system
will rotate about a windward hinge (Fig. 16.lB) and a cone of roots and attached
soil will be levered into the ground. The size of the cone will depend on the soil
strength, being large in weak wet soil and smaller in strong dry soil. There are
only two possible components of anchorage: the resistance of the soil to com-
pres sion and the buckling resistance of the leeward roots.
16.2.2.2 Systems Transmitting Horizontal Forces to the Soil
A final extreme form of anchorage system is the simple tap root system seen in
many members of the Brassicaceae (Ennos and Fitter 1992) and some pioneer
rainforest trees, such as members of the Euphorbiaceae (Crook et al. 1997). In
such systems anchorage is almost entirely provided by the vertical root which
acts just like a foundation pile. When such a plant is pushed over the tap root
rotates about a point some way below the soil surface, the top moving to leeward
and the bottom to windward (Fig. 16.lC). These movements are resisted by two
components: the compressive resistance of the soil to lateral motion and the
bending resistance of the tap root itself. Resistance of the lower regions of the
tap root to uprooting may also play some slight role. As in the other systems
the exact mode of failure will depend on the soil strength: soft, wet soil will fail
deep into the ground and the plant willlean over permanently; in stronger, drier
soil the tap root or stern are more likely to snap.
Hinge
Fig. 16.1A-C
Understanding the behaviour of these three extreme types of anchorage

systems is all very well but plants do not always behave in the textbook manner.
Some species may have morphologies which are intermediate between two
types. Sunflowers, for instance, have a large tap root as well as spreading later-
als. In other species the pattern of failure may vary between varieties. In some
varieties of maize, for example, the windward roots tend to shear off while in
others the leeward roots can buckle above-ground. Similarly, spring wheat may
fail in a way which is intermediate between winter wheat and trees, the plant
rotating about a point direct1y beneath its stern (Ennos 1991b). The mode of
failure may even depend on the conditions in which plants have been grown or
on the soil moisture. Himalayan balsam, a dicot with a root system which looks
like that of maize, only rarely fails in the same way (Ennos et al. 1993a); more
often there is windward failure similar to that characteristic of many tree
species.
16.3 Methods for Measuring Anchorage Components
16.3.1 Measuring the Relative Contributions

of Anchorage Components
The first stages in determining the components of anchorage in a particular

plant are to investigate the morphology of the root system and to determine
how the anchorage fails. This is best achieved by cutting vertical trenches in the
soil alongside the stern to expose a cross section of the root system (Coutts
1983). This has two advantages.1t allows direct observation of the roots in situ
and allows the movements of the roots and surrounding soil to be observed as
the plant is pulled over. These movements may be better visualised by coating
the sides of the trench with silver spray paint, which shows up cracks, and
filming the process of uprooting using a video camera. Sounds of breaking also
give clues about how failure is occurring.
Fig. 16.1. Failure modes due to horizontal forces in three types of root systems. In widely
spreading root systems with sinker roots such as those possessed by trees and some herbaceous
dicots (A) the system rotates up around a leeward hinge. Uprooting is resisted by three compo-
nents: the stiffness of the hinge; the resistance to pullout of the windward sinker roots; and the
weight of the root/soil plate. In the narrower systems seen in cereals (8) and other monocots,
rotation occurs about a windward hinge. Lodging is resisted by two components: the resistance
of the soil beneath the root/soil cone to compression; and the buckling resistance of the coronal
roots. Simple tap root systems (C) are supported by the resistance of the soil on either side to
compression. Rotation occurs directly beneath the stern
Once the pattern of failure has been observed, and the possible components
of anchorage identified, their relative importance can be determined by carry-
ing out a further series of tests.
In plate systems in particular it has proved possible to determine the rela-
tive size of the different components. This is do ne by comparing the torques
required to pull over single plants both before and after particular components
are destroyed. Typical procedures involve cutting of the leeward hinge and
trenching around the root system (Coutts 1986; Ennos et al. 1993a). It is
also possible to estimate the strength of the leeward hinge by summing the
strengths of individual lateral roots on one side of the trunk against being bent
upwards (Crook and Ennos 1996). In all the species that have so far been
investigated the largest anchorage component is the uprooting resistance of
windward roots.
In other systems it has proved much harder to work out the relative size of
the different anchorage components. For some systems this is due to practical
difficulties. In cone systems, for instance, it is hard to determine the importance
of the buckling resistance of the basal roots because each buckles at a different
depth and at a different distance from the centre of rotation (Crook and
Ennos 1993). The importance of the resistance of soil to compression in such
root systems can be determined, however, by rotating a model cone into the soil.
For other anchorage systems the difficulty in determining the relative size of
the anchorage components is intrinsic to the mechanics. The soil will only
help prevent tap-rooted systems from lodging, for example, if the tap root
has bending rigidity. It is therefore theoretically impossible to separate the
components.
The most obvious technique to determine anchorage components, and
one which has been extensively used by many early workers who investigated
the lodging resistance of wheat and maize (Neenan and Spencer-Smith
1975), is to measure the force required to extract individual roots. Unfortunately
this technique has very little value because roots are not pulled out of the
ground when these species are pulled over. A further problem is that the
failure which occurs when an individual root is pulled out is quite different from
that which happens when the whole system fails and many roots pull out in
a block.
In cases in which it is difficult to calculate the size of the components of
anchorage it is perhaps best to simply measure the mechanical properties of the
roots and the soil in which they are embedded. Measurement of root proper-
ties can provide much more information about the anchorage system and will
allow much easier direct comparison between different varieties or species
of plants.
16.3.2 Measuring Root Properties
Two sets of mechanical tests can be carried out on roots depending on their
morphology and function. These are best performed on a universal loading
frame that has been purpose built for materials testing (eg. Instron Corpora-
tion, Lloyd's).
16.3.2. 1 Tensile Tests
Fibrous roots resist only tensile forces so their properties are best investigated
by carrying out tensile tests, in which the upper regions of the roots are held
in the clamps of a universal testing machine and stretched (Fig. 16.2A). The
machine will produce a graph of force against extension (Fig. 16.2C) which can
be used (see Box 16.1) to determine the root's mechanical properties: breaking
strength, and the breaking stress and stiffness of the material from which it is
made. The "overalI" strength of the root system can finalIy be characterised as
the sum of the basal strengths of alI the roots.
BOX 16.1. Equations to Calculate Root Properties

from Tensile Tests
The breaking strength of a root, Fmax> is simply the greatest force it can
withstand (see Fig. 16.2b). From the graph of force against extension the
material properties of the material from which the root is made can also
be calculated.
The breaking stress, amax> is the force per unit area the root can
withstand. If the root is circular, breaking stress is given by the
expression:
(16.1)
where D is the diameter of the root.
Finally, the material stiffness or Young's modulus, E, of the root, is
the stress which would be required to double its length. This has units of
newtons per square metre and is given by the expression:
E= dF 4L , (16.2)
dy nD 2
where dF/dy is the initial slope of the force/extension curve, and L is the
initiallength of the root between the clamps (Ennos 1989).
A B
t
Crosshead
-+---++--Load Cell
Clamp
Probe~~~~----I
Root---tî~-ln~~~~~
Support -
Force
_ ~reaking~trength
I
/
dF/dy/
/
I
I
"
Deflection
Fig. 16.2. Methods for determining the mechanical properties of roots using a universal testing
machine. In tensile tests (A), fibrous roots are gripped by two clamps and are stretched by
upward movement of the crosshead. The load cell measures the force required. In three point
bending tests (B), a length of rigid root is balanced between two supports and is bent by a probe
which is moved down and hits it midway between them. Both tests produce a force/deflection
graph (C) which can be used to determine both the maximum bending moment a root can with-
stand, its rigidity, and the stiffness of the material of which it is composed
There are two main problems with tensile tests. The first is a theoretical
problem; the strength of roots decreases with increasing distance from their base,
so the results obtained may not correspond well with uprooting resistance. The
second is practical; when stretched by the machine, roots tend to break at the
clamps, at forces well below their breaking strength. Gluing the roots to the clamp
using cyanoacrylate "superglues" may reduce this problem to an extent but when
this is done the stele may simply slip out of the cortex at the clamps.
76.3.2.2 Bending Tests
The more rigid laterals and tap roots are broken more frequently in nature by
being bent rather than by being stretched, so their properties are better inves-
tigated by carrying out bending tests. To study uniform bending forces on their
own without additionally imparting shear to the roots it would be theoretically
best to carry out four point bending tests, which are equivalent to holding a root
section at its two ends and bending it. This provides a uniform bending load to
the whole section. However, these four point bending tests are tricky to perform
and roots are never, in fact, loaded this way. It is easier and just as valid to carry
out three point bending tests. In such tests, the upper region of the root is rested
between two supports and the root is bent by a probe which impacts on it half
way between the supports (Fig. 16.2B). In these tests the centre of the section
is subjected to the greatest bending loads, while the ends get no loading in
bending. However, from the force/deflection curve {Fig. 16.2C) both the bending
strength and rigidity of the root can be calculated, along with the stiffness, or
Young's modulus, of the root material (Ennos et al. 1993b), but the calculations
(see Box 16.2) are rather more complex than for tensile tests.
Bending tests have the advantage that they are very quick and easy to
perform. Their disadvantages are that roots are seldom cylindrical, as the test
assumes, but taper gradually. Hence the calculations of the strength, rigidity
and Young's modulus are subject to some error. Despite this, three point bend-
ing tests have been extremely useful in comparing anchorage systems of dif-
ferent species (Ennos et al. 1993a,b), varieties ofplants (Crook and Ennos 1994)
and plants subjected to different treatments (Crook and Ennos 1994, 1995;
Goodman and Ennos 1996). The rigidity and strength of the stern can also be
measured for comparison using this technique. A simplified pocket device exists
for field measurement of the material properties of woody tree roots (Mattheck
et al. 1994). This measures both the lateral breaking stress and longitudinal
compressive breaking stress of the wood (Stokes and Mattheck 1996). Though
these properties are not the same as those calculated by the three point bending
test, they are still useful because the breaking stress of wood tends to be pro-
portional to its stiffness.
BOX 16.2. Equations to Calculate Root Properties

from Bending Tests
The maximum bending moment a root can withstand, M max , is a
structural property of the whole organ. It has units newton metres and
is given by the expres sion:
(16.3)
where F rnax is the force required ta break ar bend the root past its failure
point, and L is the length between the supports.
The rigidity of the root, R, which is also a structural property of the
whole organ, is its resistance ta being curved. It has units newton metres
squared and is given by the expression:
dF L3
R =- - (16.4)
dy 48
where dF/dy is the initial slope of the force/deftection curve.
The rigidity of the root is, in turn, the product of the stiffness
(ar Young's modulus) of the material, E, and 1, the second moment of
area of its cross section, sa:
R = EI. (l6.5)
For a cylindrical root of diameter, D, 1 is given by the expression:

(16.6)
The stiffness of the material, E, which has units of newtons per metre
squared, can therefore also be readily calculated for each root section by
combining the last two equations ta give:
(16.7)
Just as with the tensile tests, the sum of the bending strengths ar
rigidities of aU the roots can then give a measure of the strength of the
root system.
16.3.3 Measuring Soil Properties
As we have seen, the strength of a plant's anchorage system and even its uproot-
ing behaviour depends greatly an the strength of the soil in which its roots are
embedded. In turn this crucially depends an the water potential of the soil. It
is therefore extremely important when performing lodging and uprooting tests
to control soil properties, or at least be able to measure them. In this section we

mention briefly some techniques used in measuring and controlling soil prop-
erties. The re ader should refer to Chapter 3 of this book, and standard soil
physics texts for further information (eg. Marshall and Holmes 1998; Smith and
Mullins 1999).
Unfortunately, controlling soil water potential is extremely difficult, partic-
ulady in the fie1d next to a transpiring plant. Perhaps the best that can be
achieved in fie1d conditions is to carry out tests when the soil is wet and anchor-
age failure is most likely to occur. This is best done by saturating the soil by
heavily overwatering it before leaving it for over an hour to drain. In the labo-
ratory, control of water potential is rather easier. Soil at field capacity can be
achieved by saturating cores of soil, and allowing them to drain. A tension table
can also be used to obtain a known matric potential between O and about
-20 kPa. This range can be extended to about -80 kPa using ceramic tables with
a vacuum pump attached. To do this the saturated cores are laid on filter paper
which covers the tension table. The matric potential is adjusted by raising or
lowering a water reservoir, and leaving the samples to equilibrate. The time
taken for equilibration is gre ater for more negative matric potentials and taller
soil samples. Equilibrium is achieved when the mass of the soil core remains
constant with time, and may require a week or longer. Soil strength increases
considerably in the range -50 to -1500 kPa, but it is unfortunately virtually
impossible to evenly dry soil that contains a transpiring plant to such negative
potentials.
No easy tests exist to reliably measure the tensile or compressive strengths
of undisturbed soil in the fie1d, as re1ated to anchorage. Penetrometers measure
the pressure required to push a metal cone through the soil (Chap. 3), involv-
ing compression of the soil around the probe tip. Penetrometer resistance
involves a large component of frictional resistance, and a complex mode of
failure around the probe tip that cannot be simply re1ated to the failure of soil
which occurs dur ing anchorage tests.
The shear strength, r, of soils is most easily measured using a simple vane
shearmeter which can be modified to measure a wide range of soil strengths by
changing the size of the vane. Soil is a variable material so as many tests as pos-
sible should be carried out close to the plants which are being examined; ten
to twenty tests should give a good idea of the mean and variance of the soil
strength. These tests should always be carried out on undisturbed soil as
remoulding dramatically alters soi! mechanical properties.
16.4 Methods for Evaluating Plant Anchorage Taken

as aWhole
16.4.1 Direct Methods
16.4.1.1 Vertical Tests
Because the mechanics of plant anchorage is complex, and involves several root
and soH characteristics, a convenient way to evaluate anchorage resistance is to
measure it directly. The most commonly used method is to measure the verti-
cal root-pulling resistance of the plant. Its basic principle is to measure the peak
force required to pull the plant's root system from the soil. Practical problems
may happen if stern breakage occurs instead of uprooting. This is most likely
when the soH is very dry, so these conditions should be avoided. Simple exper-
imental devices have been proposed which may be used for field studies (Nass
and Zuber 1971; Rogers et al. 1976; O'Toole and Soemartono 1981; Peters et al.
1982; Jenison et al. 1981; Kevern and Hallauer 1983; Penny 1981). The equip-
ment used includes a scale to measure force, a lever and a hand-operated clamp
to grasp the plant, and a tripod or some other portable platform. The clamp,
sash cord, or leather strap is placed around the base of the stalk and the plant
is uprooted with a hand-powered winch, the force necessary being measured
by a force transducer. This type of device has some limitations with respect to
the standardisation of the test and the time required. More sophisticated devices
have been proposed with a tractor or a skid-steer loader providing the required
hydraulic power. These devices are less labour intensive and use a standardised
root-pull measurement (Donovan et al. 1982; Beck et al. 1987). A device aimed
at more accurate vertical tests performed on young plants in laboratory
conditions was proposed by Ennos (l991b). In this latter case, the whole
force/vertical displacement curve was recorded.
Significant correlations were found by some authors between this vertical
test and the resistance to lodging of a range of maize genotypes (Fincher et al.
1985). Contradictory results were, however, found by other authors who found
low or no correlations between this test and lodging resistance (Armara and
Crosbie 1982; Kevern and Hallauer 1983; Melchinger et al. 1986). One major
advantage of this method is its simplicity, but the mode of faHure is very dif-
ferent from what happens during natural lodging; the anchorage is tested
vertically, whereas lodging involves resistance to a horizontal strain. Weak cor-
relations between this test and resistance to lodging are therefore not surpris-
ing. Clearly, this method seems more adapted to studies of plant resistance to
uprooting.
16.4.1.2 Horizontal Pushing or Pulling Tests
Another group of methods is based on the measurement of the horizontal

pushing or pulling resistance of the plant. The basic principle of these methods
is to measure the resistant torque of the plant versus its angle of displacement
when it is horizontally pulled or pushed. Such a test was used in field or labo-
ratory conditions on rice (Terashima et al. 1992, 1995), maize (Kushibiki 1979;
Koinuma et al. 1990; Ennos et al. 1993), wheat (Ennos 1991b; Crook and Ennos
1993b; Crook et al. 1994) and sunflower (Ennos et al. 1993). Similar tests, using
powerful winches and high tension force transducers, have been used to test the
anchorage of trees (Fraser and Gardiner 1967; Coutts 1983, 1986; Stokes et al.
1995; Crook and Ennos 1996b; Crook et al. 1997). For laboratory trials, an accu-
rate system was proposed by Ennos (1991b) using a modified tensile testing
machine which could simultaneously push over the plant and measure the
restoring anchorage moment supplied by its root system. In field conditions,
many authors have used quite simple manual devices with only a force gauge
and a ruler to measure angle displacement of the stern. A more sophisticated
device was proposed recent1y by Fouere et al. (1995; Fig. 16.3). The apparatus
consists of a support, a force sensor, an angle sensor and a control head. It
records a curve relating the resisting torque of the anchorage system to the incli-
nation of the stern during an artificial pushing test. Data may be transferred to
a computer using an RS232 serial transfer protocol. A typical response curve is
given in Fig. 16.4.
Significant correlations were found between this horizontal test and resis-
tance to lodging (Kushibiki 1979; Koinuma et al. 1990; Terashima et al. 1992;
Fouere et al. 1995). Although they do not exact1y reproduce what happens
during naturallodging, horizontal tests mimic lodging far better than vertical
root-pull tests. The horizontal test may also be used to evaluate the relative con-
tribution of the different components of the anchorage performing manipula-
tions as proposed by Ennos et al. (1993a,b).
16.4.2 Indirect Methods
Many authors have tried to find correlations between important features of the
root system and plant anchorage, measured with one of the previously men-
tioned methods, or resistance to lodging. Correlations with individual root
characters are often weak because of the complexity of anchorage mechanics.
The closest correlations were often found with characters which affect the size
of the root-soil cone, such as the angle of spread of adventitious roots from
higher internodes (Hebert et al. 1992). In wheat, the link between the diameter
of the root-soil cone and anchorage resistance was theoretically and experi-
..
[lS)
''''''''I.
•••
(front view)
HANDLE
~+--+lI--MAIN SOCKET
J11.i"i--t.f--- ON/OFF swilch

nlt-+--display sockel
ANGLE SENSOR - - - - -
(fn box)
O.20m
AOJUSTABLE NAILS
(ANCHORING IN GROUNO)
a b
Fig. 16.3. General view of the portable electronic device for horizontal pushing test proposed
by Fouere et al. (1995). The free rotating nails are anchored in the ground, thus creat ing a hor-
izontal rotation axis passing through the plant base and preventing back-sliding of the device
during the pushing operation. During the pushing operation, the force sensor measures and
transmits the force to the stalk. Afreely rotating PVC pulley prevents any friction between the
force sensor and the stalk. (Reproduced from Agronomy Journal)
mentally demonstrated by Crook and Ennos (1993). Obviously the search for
root characters which are correlated with plant anchorage resistance should be
based on a preliminary investigation of the anchorage mechanics of the species
under considerat ion. Most promising developments are based on attempts to
model root anchorage, which may help in identifying the most important root
and soil characters for anchorage (Ennos 1991b; Crook and Ennos 1993).
16.5 Conclusion
A classical objective of people interested in plant anchorage is to compare the

ability of different genotypes to withstand conditions which may cause lodging
or windthrow. They have also investigated the relative performance of plants
subjected to different treatments, such as growing them at different densities,
and in different soils. Static methods for evaluating anchorage as a whole, such
as vertical uprooting or lateral pushing or pulling tests, have proved helpful
and pragmatic tools in this context, even though they do not exact1y mimic
the natural challenge to anchorage when aerial parts are subjected to
dynamic wind pressures. Until now, methods for measuring root mechanical
properties have also been used in the context of comparison between species
or genotypes.
Because the scientific investigation of anchorage is stiH in its early stages,
there is, however, stiH a strong need for simply understanding how it works.
Observation-based methods, which describe what happens when a plant is
pushed or pulled, are stiH useful in this context. An important contribution may
be expected from the use of strain gauges which record the distribution of
mechanical strains in the stern and roots (Stokes et al. 1995; Crook and Ennos
1996b; Crook et al. 1997). A further step wiH be to model plant anchorage by
simultaneously considering the root system architecture, the mechanical prop-
erties of the roots, and the mechanical properties of soils. Pioneer models have
been proposed by Ennos (1991b), and Crook and Ennos (1993), but future
models wiH probably have to be more sophisticated and make use of the criti-
cal state theory of soil mechanics, which up until now has mostly investigated
the movement of narrow tines through topsoil. Such models, combined with
experimental studies in which model roots are pulled through the soil, wiH help
in identifying the most important root and soil characteristics for anchorage.
Future work must also investigate the development of root systems over time,
and the effect on the secondary growth of roots of the constriction imposed by
soils of different strengths.
Finally, if the ultimate aim is to account for the ability of plants to with-
stand adverse climatic conditions, anchorage factors have to be combined with
aerial factors. Although plant response to wind pressure and anchorage
mechanics are both very complex, the most promising approaches for under-
standing plant resistance to lodging are those based on the simultaneous con-
sideration of these two phenomena. As a plant develops and becomes taller, the
mechanical demands imposed on the stems and root system increase. The equi-
librium between aerial parts and anchorage has to be considered with a
dynamic approach. Examples of such "safety factors" accounting for the equi-
librium between shoot and root system have been proposed by Crook et al.
(1994) for wheat, Duparque and Pellerin (1994) for maize, and Nielsen (1995)
for trees. An even more sophisticated general approach has been developed by
Baker (1995) and is being applied to the lodging of wheat.
Acknowledgements. The authors wish to thank Alexia Stokes and John Graham
for helpful comments on the manuscript.
References
Anderson q, Coutts MP, Ritchie RM, Campbell DJ (1989) Root extraction force measurements
for Sitkas pruce. Forestry 62: 127-137
Arihara J, Crosbie TM (1982) Relationships among seedling and mature root system traits of
maize. Crop Sci 22: 1197-1202
Baker CJ (1995) The development of a theoretical model for the windthrow of plantsr. J Theor
Bio1175: 355-372
Beck DL, Darrah LL, Zuber MS (1987) An improved technique for measuring resistance to root
pulling in maize. Crop Sci 27: 356-358
Carter PR, Hudelson KD (1988) Influence of simulated wind lodging on corn growth and grain
yield. J Prod Agric 1: 295-299
Coutts MP (1983) Root architecture and tree stability. Plant Soil 71: 171-188
Coutts MP (1986) Components of tree stability in Sitka spruce on peaty gley soil. Forestry 59:
173-197
Coutts MP, Grace J (1995) Wind and trees. Cambridge University Press, Cambridge
Crook MJ, Ennos AR (1993) The mechanics of root lodging in winter wheat Triticum aestivum
1. J Exp Bot 44: 1219-1224
Crook MJ, Ennos AR (1994) Stern and root characteristics associated with lodging resistance
in four winter wheat varieties. J Agric Sci 123: 167-174
Crook MJ, Ennos AR (1995) The effects of nitrogen and growth regulators on stern and
root characteristics associated with lodging resistance in winter wheat. J Exp Bot 46:
931-938
Crook MJ, Ennos AR (1996a) Mechanical differences between free-standing and supported
wheat plants Triticum aestivum 1. Ann Bot 77: 197-202
Crook MJ, Ennos AR (1996b) The anchorage mechanics of mature larch Larix europea x 1.
japonica. J Exp Bot 47: 1509-1517
Crook MJ, Ennos AR, Sellers EK (1994) Structural development of the shoot and root systems
of two winter wheat cultivars, Triticum aestivum 1. J Exp Bot 45: 857-863
Crook MJ, Ennos AR, Banks JR (1997) The function of buttress roots: a comparative study of
the anchorage systems of buttressed (Aglaia and Nephelium ramboutan species) and non-
buttressed (Mallotus wrayi) tropical trees. J Exp Bot 48: 1703-1716
Donovan LS, Jui P, Kloek M, Nicholls CF (1982) An improved method of measuring root strength
in corn (Zea mays 1.). Can J Plant Sci 62: 223-227
Duparque A, Pellerin S (1994) An attempt to model the mechanism of root 10dging resistance
in maize. 3rd Congr European Society for Agronomy. Proceedings, 18-22 September 1994,
Padova, Italy, pp 360-361
Ennos AR (1989) The mechanics of anchorage in seedlings of sunflowers Helianthus annuus.
Ennos AR (1990) The anchorage of leek seedlings: the effect of root length and soil strength.
Ann Bot 65: 409-416
Ennos AR (1991a) The mechanics of anchorage in wheat Triticum aestivum 1.1. The anchorage
ofwheat seedlings. J Exp Bot 42: 1601-1606
Ennos AR (1991b) The mechanics of anchorage in wheat Triticum aestivum 1. II. Anchorage of
mature wheat against lodging. J Exp Bot 42: 1607-1613
Ennos AR (1993) The scaling of root anchorage. J Theor Bio1161: 61-75
Ennos AR, Fitter AH (1992) Comparative functional morphology of the anchorage systems of
annual dicots. Funct Ecol6: 71-78
Ennos AR, Crook MI, Grimshaw C (1993a) A comparative study of the anchorage systems of
Himalayan balsam Impatiens glandulifera and mature sunflower. Helianthus annuus. J Exp
Bot 44: 133-146
Ennos AR, Crook MJ, Grimshaw C (1993b) The anchorage mechanics of maize Zea mays. J Exp
Bot 44: 147-153
Fincher RR, Darrah LL, et Zuber MS (1985) Root development in maize as measured by verti-
cal pulling resistance. Maydica, 30: 383-394
Flesch TK, Grant RH (1992a) Corn motion in wind during senescence: 1. Motion characteristics.
Agron J 84: 742-747
Flesch TK, Grant RH (1992b) Corn motion in wind during senescence: II. Effect of dynamic
plant characteristics. Agron J 84: 748-751
Fouere A, Pellerin S, Duparque A (1995) A portable electronic device for evaluating root lodging
resistance in maize. Agron J 87: 1020-1024
Fraser AI, Gardiner JBH (1967) Rooting and stability in Sitka spruce. Forestry Commission Bul-
letin 40. HMSQ, London
Goodman AM, Ennos AR (1996) A comparative study of the response of roots and shoots of
sunflower and maize to mechanical stimulation. J Exp Bot 47: 1499-1507
Goodman AM, Ennos AR (1997) The responses of field-grown sunflower and maize to mechan-
ical support. Ann Bot 79: 703-711
Hebert Y, Barriere Y, Bertholeau JC (1992) Root lodging resistance in forage maize: genetic vari-
ability of root system and aerial part. Maydica 37: 173-183
Hughes TP, Sykes AR, Poppi DP, Hodgson J (1991) The influence of sward structure on peak bite
force and bite weight in sheep. Proc N Z Soc Anim Prod 51: 153-158
Jenison JR, Shank DB, Penny LH (1981) Root characteristics of 44 maize inbreds evaluated in
four environments. Crop Sci 21: 233-237
Kevern TC, Hallauer AR (1983) Relation of vertical root-pull resistance and flowering in maize.
Crop Sci 23: 357-363
Koinuma K, Inoue Y, Kato A (1990) Evaluation of lodging resistance of maize (Zea mays L.) by
the measurement of the horizontal pull resistance. Bull Nat Grassl Res Inst 43: 23-29
Kushibiki H (1979) A simple method of testing lodging resistance of maize. Bull Hokkaido
Prefect Agric Exp Stn 42: 21-27
Laca EA, Ungar ED, Seligman NG, Ramey MR, Demment MW (1992) Am integrated
methodology for studying short-term grazing behaviour of cattle. Grass forage sci 47:
81-90
Marshall TJ, Holmes JW (1998) Soil physics, 3rd edn. Cambridge University Press, Cambridge
Mattheck C, Bethge K, Zipse A (1994) Das Fractometer. Alig Forstztg 14: 790-791
Melchinger AE, Geiger HH, Schmidt GA (1986) Vertical root-pull resistance and its relationship
to root lodging and forage traits in early maturing European inbred lines and FI hybrids
of maize. Maydica 31: 335-348
Minami M, Ujihara A (1991) Effect of lodging on dry matter production, grain yield and nutri-
tional composition at different growth stages in maize. Jpn J Crop Sci 60: 107-115
Nass HG, Zuber MS (1971) Correlation of corn (Zea mays 1.) roots early in development to
mature root development. Crop Sci 11: 655-658
Neenan M, Spencer-Smith JL (1975) An analysis of the problem of lodging with particular
reference to wheat and barley. J Agric Sci 85: 495-507
Nielsen CCN (1995) Recommendations for stabilisation of Norway spruce stands based on eco-
logical survey. In: Coutts MP, Grace J (eds) Wind and trees. Cambridge University Press,
Cambridge
Q'Toole JC, Soemartono (1981) Evaluation of a simple technique for characterizing rice root
systems in relation to drought resistance. Euphytica 30: 283-290
Pellerin S, Trendel R, Duparque A (1990) Relation entre quelques caracteres morphologiques

et la sensibilite il la verse en vegetation du mals (Zea mays 1.). Agronomie 6: 439-446
Penny LH (1981) Vertical-puII resistance of maize inbreds and their test crosses. Crop Sci 21:
237-240
Peters DW, Shank DB, Nyquist WE (1982) Root-pulling resistance and its relationship to grain
yield in FI hybrids of maize. Crop Sci 22: 1112-1114
Rogers RR, Russell WA, Owens JC (1976) Evaluation of a vertical-pull technique in population
improvement of maize for corn rootworm tolerance. Crop Sci 16: 591-594
Smith KA, Mullins CE (1999) Soi! analysis: physical methods, 2nd edn. Marcel Dekker, New York
Stokes A, Mattheck C (1996) Variation of wood strength in tree roots. J Exp Bot 47: 693-699
Stokes A, Drexhage M, Heinze P, Guitard D (1995) Anchorage strength and distribution of strain
in tree root systems. Colloque I:arbre, Montpellier, France
Terashima K, Akita S, Sakai N (1992) Ecophysiological characteristics related with lodging tol-
erance of rice in direct sowing cultivation. 1: Comparison of the root lodging tolerance
among cultivars by the measurement of pushing resistance. Jpn J Crop Sci 61: 380-387
Terashima K, Akita S, Sakai N (1995) Ecophysiological characteristics related with lodging tol-
erance of rice in direct sowing cultivation. III: Relationship between the characteristics of
root distribution in the soi! and lodging tolerance. Jpn J Crop Sci 64: 243-250
Further Reading
Ennos AR (1994) The biomechanics of root anchorage. Biomimetics 2: 129-137

Appendix: Suppliers
Company Address Chapter Items
Bartz Technology 116-A East Yanonali Street 8,9,10 Minirhizotron camera

Corporation Santa Barbara, systems, digital image
CA 93101, USA capture systems
tel.: + 1 8058845156
fax: +1 8058845168
www. bartztechnology.com
e-mail: bartz@silcom.com
BASF pic Wembley Park Drive 151 8 Styodur insulation
HA9 8HQ Middlesex, UK material
BioRad Laboratories Harbour Way South 1414 13 Ion exchange resins,
Richmond, CA 94804, USA material for solid phase
extraction
Campbell Scientific Field Street 14-20 3 Temperature sensors
Ltd. LE12 9AL Leicestershire, UK
Classen and Co. Grevenweg 89 8 Endoscopes
2000 Hamburg 26, Germany
tel.: +49 40 254031
Decagon 950 EN Nelson Court 14 Plant water and osmotic
Pullman, WA 99163, USA potential, soil water
tel.: +1 509 332-2756 potential
fax: +1 509332-5158
www.decagon.com
e-mail: bryan@decagon.com
Delta-T Devices Ltd. Low road 128 3,6,10,8 Temperature sensors,
CB5 OEI Cambridge, UK root washer, root image
tel.: +44 1638742922 analysis system,
fax: +44 168 743 155 minirhizotrons, desktop
www.delta-t.co.uk scanner systems for
e-mail: sales@delta-t.co.uk scanning root samples
DOW Europe S.A. Bachtobelstrasse 3 8 Styrofoam insulation
8810 Horgen, Switzerland material
Dynamax Inc. Fallstone 10808 14 Sap flow
Houston, TX 77099, USA
fax: +1 713 5645200
Easy Test Ltd. P.O. box 24 3 Redox probes, etc.
Solarza 8b
20-815 Lublin 56, Poland

568 Appendix: Suppliers
Eijkelkamp Postbus 4 6,14 Root augers,

Agriseareh 6987 ZG, Giesbeek, soil water eontent and
Equipment The Netherlands matrix potential, (peat)
tel.:+31 313631941 profile samplers,
fax: +31 313 632167 penetrometers, plant
www.eijkelkamp.eom physiological researeh
e-mail: equipment
eijkelkamp@eijkelkamp.eom
ELE International Eastman Way 3,16 Penetrometers,
Ltd. HP2 7HB Herts, UK temperature sensors,
tel.: +44 1442218355 shear meters
fax: +44 1442252474
www.eleint.eo.uk.
e-mail: ele@eleint.eo.uk
Environmental Turisticka 5 14 Sap ftow
Measuring Systems 62100 Brno,
Czeeh Republie
tel.: +420 (O) 541225344
fax: +42 (O) 541 225 344
e-mail: jkueera-ems@iol.ez
Fluka Chemie Messersehmittstrasse 17 13 Agarose (low
89231 Neu-Ulm, endosmosis type), pH
Germany and redox indicators,
ion exehange resins,
material for solid phase
extraetion
Fluroehem Ltd. Wesley Street 3 Polyethylene glycol
SK13 9RY Derbyshire, UK
Gillisons Variety 3033 Benzie HWY 6 Hydropneumatic root
Fabrieation Ine. Benzona, MI 496 washer
16-9747,USA
tel.: +1231 8825921/
fax: +1231 8825637
e-mail: gillison@gtiLeom
Greenspan 22 Palmerin Street 14 Sap ftow

Teehnology Pty Ltd. Warwick QLD 4370,
Australia
tel.: +61 (O) 7 4660 1888
fax: +61 (O) 74660 1800
www.greenspan.eom.au
e-mail:
admin@greenspan.eom.au
Appendix: Suppliers 569
Hawker De 5108 Salisbury, Australia 6 Comair root scanner

Havilland tel.: +61 82583422
Victoria Ud. fax: +61 8281 2529
Hortresearch PB 11-030 14 Sap flow
Environment & Palmerston North
Risk Management New Zealand
Group tel.: +64 (06) 356-8080
fax: +64 (06) 3546731
www.hort.cri.nz
e-mail: sgreen@hort.cri.nz
Ingold - Mettler Siemensstrasse 9 13 pH electrodes
Toledo 61449 Steinbach, Germany and microelectrodes
Institute of Crowmarsh Gifford 14 Soil water content
Hydrology OXI0 8BB Oxfordshire, UK and matrix potential
Instron Ud. Coronation Road 16 Universal mechanical
HP12 3SY Bucks, UK testing machines
tel.: + 01494 464646
fax: + 01494 456123
Kiyowa Hatfield Road 135 16 Strain gauges,
ALI 4LZ St Albans, UK strain bridges
tel.: + 01727 859 373
fax: + 01727 844272
Machery & Nagel Duren, Germany 13 Filter papers, membrane
GmbH&Co.KG filters, phosphatase test
papers, ion exchange
resins, material for solid
phase extraction
Mecmesin Ud. Lawson Hunt Industrial 16 Force gauges, torque
Park Horsham, gauges, force testing,
West Sus sex, UK torque testing, force
tel.: + 01403256276 measurement, torque
fax: + 01403240205 measurement, quality
www.mecmesin.com control, tension,
e-mail: info@mecmesin.com compression,
dynamometes, materials
testing, software,
universal testing medina,
force transduces, torque
transducers
MerckKG 64271 Darmstadt, Germany 13 Agar, ion exchange
resins, material for solid
phase extraction
570 Appendix: 5uppliers
Millipore Maple Street 34 13 Filter papers,

Corporation Milford, MA 01757, USA membrane filters
Olympus Optical Wndenstrasse 14-16 8 Borescopes, fibroscopes
Co. 2 Hamburg, Germany
Regent rue Blain 4040 8,lO Root image analysis
Instruments Inc. G2B 5C3 Quebec, Canada system, desktop scanner
tel.: +1(418) 840-1347 systems for scanning
fax: +1(418) 840-1350 root samples, root
www.regent.qc.ca morphological
e-mail: sales@regent.qc.ca measurement, root
length measurement
Rhizosphere Dolderstraat 62 13 Equipment for soil,
Research 6706 JG Wageningen, rhizosphere and root
Products The Netherlands research
tel.: +31-317-4lO961
fax: +31-317-422415
www.rhizosphere.com
e-mail:
info@rhizosphere.com
Schott Rickingham Road 302 13 pH electrodes
Microelectrodes Lindindery, and microelectrodes
Inc. NH 03053, USA
Schott -Ruhrglas Theodor Schmidt Str. 25 8 Duran borosilicate glass
GmbH 8589 Bayreuth, Germany minirhizotrons
Sentek PTY Ltd. King William Street 69 14 Soil water content
5067 Kent Town Australia and matrix potential
fax: + (08) 3628400
Sigma Chemie Postbus 260 13 pH and redox indicators
9470 Buchs, Switzerland
Skye Instruments Ddole Industrial Estate 14 Plant water
Ltd. Landrindod Wells LD1 6DF and osmotic potential
Powys UK
fax: +44(0) 1597824812
Soilmoisture 801 S. Kellogg Ave. 14 Plant water

Equipment Corp. Goleta, CA93117, USA and osmotic potential
tel.: + 1 805 9643525
fax: +1 8056832189
www.soilmoisture.com
e-mail:
websales@soilmoisture.com
Appendix: 5uppliers 571
Thermoplastic Valley Road 1268 8 Clear tenite butyrate

Proeesses Ine. Stirling, NJ 07980, USA tubes and eylinders,
TML (Tokyo Sokki tel.: +1 201 6471000 clear polyearbonate
Kenkyojo Co. fax: +12017536749 tubing (minirhizotrons)
Ud. Japan)
Transdueers; agent 59 Aleester Road, Studley 16 Foree transdueers,
UK/Ireland: Teehni B80 7NJ Warwickshire, UK strain gauge-based
Measure tel.: +44 1527854103 transdueers
fax: +44 1527 853267
e-mail:
sales@teehni-measure.eo.uk
UpGmbH Ruffinstr.12 14 Sap flow
80637 Munich, Germany
fax: +49 (O) 89 1665 135
Zurich Bolting Rusehlikon 8803, 13 Bolting cloth
Cloth Mfg. Co. Ud. Switzerland
Subject Index
A (s. atom percent abundance) percent abundance (A) 370

abscisic acid (ABA) 13,77,492 percent enrichment (excess) 370
abundance, natural 366,370,371,374,380, augers 155,159,177,178,217
381 hand 241
Acer saccharum 276 root 568
acetonitrile 44 autoclaving 38
acid soil 185, 190 automatic analysis 336
acidified dimethoxypropane 44 autoradiography 368,377,385,387,391,
acrylic resins 45, 55 423,424
activity, specific 372 axial resistance 487
added N-interactions 384
adsorption isotherm 512-514,519,532,533
linear 515 ballottini 87, 407
non-linear 517 Banksia integrifolia 410
aerenchyma 421 bariey 6,24,25,59,81,88,116,311,377
agroforestry 539 bean 81, 148,346
alfalfa (luccme) 116,275,282,325 beech 275
algorithms bending
chain (type) method 320,324 resistance 549
closing 319 tests 555, 556
dilation 319 bentgrass, creeping 250
edge chord 323, 324 Bequerel (Bq) 372
erosion 319 beta ({3) partic1es 372,375
skeletonisation/thinning 322 Betula pendula 9,23
alpha (a) particles 372 biogeochemical cyc1es 274
aluminium toxicity 121 biopores 11
AMF (s. mycorrhizae) blue light excitation 56
ammonium 533 bolting cloth 440, 571
AMS (s. mass spectrometry) borescopes 245,570
analogue 372,390 boundary conditions 511,512,514,517,
anchorage (components) 3,12,17,547-562 523,530
identifying, measuring 551-563 branching order (s. also root branching)
angle sensors 559 16, 125
anoxic 93,96,105 Brassica napus 408
anti-frost additives 42 Brassicaceae 549
apple 10, 11,84,275 brightness levels 311
architecture guided sampling 157 Brussels sprouts 246, 250
atom buffer power 517
fraction 370 bulk density 86,95,97,99,100,104,241,265
574 Subject Index
spatial distribution of 346, 353 chloral hydrate 38

bush bean 347 chlorine (CI) 391
bushes 257 chloroform 39
chromosome number 40
36Cl 376,391
"C 377,386,388,400 Clarke fixatives 38,40,44
l3C 69,366-369,379,381,383,400 elimate change 274
l·C ("bomb") 368 CMI (s. computed macro tomography)
l·C 368,372,374-376,378,379,380, coagulating fixatives 42, 43
385-388,400 coefficient of variation 162-164
C,-plants/C.-plants 366,383 cohort(s) 16,17,21,27,279,280,281,
.5Ca 378, 391, 400 284-287,289,291,292,294,299
Calapogonium mucunoides 190 compartmental analysis 367,388
calcium (Ca) 391,400 compartmented containers 431,432
calibration 190, 222 compensation method 478
root area 325 competition (experiments) 10,263,367
procedure 59 computed macro tomography (CMI) 359
Canada balsam 40, 53, 54 computed tomography (CI) 21,344,358
capacitance 482 computer assisted tomography (CAI) 344
carbon (C) 400 conductance 486, 539
allocation/partitioning 2, 6, 14, 15, 17, 28, root 524,526,529
367,368,386 hydraulic 20,489,491,524
cost of root systems l37 conductivity 486
cyeling 274,381,386 hydraulic 80, 86, 486, 489, 490, 520
dioxide (CO,), increased 8, l37, 277 constraints, mechanical 128
dynamics 366 contrast (enhancement) 309,317
transport 367 control
Casparian strips 51 negative 51
CAI (s. computer assisted tomography) positive 50
CCD (s. charged coupled device) conversion, 2-D to 3-D 27,220,224,236
cell distance transition 230
cyele kinetics 367 co-ordinates, cylindrical 513
turgor 90 copper (Cu) 391
wall tension 90 core diameter 180
cells, triaxial 87,93 core-break technique 155,178,217,218,
cellulose 52 221,222
centrifugation 201 coUon 83, 90, 97, 116, 223
cereals 152, 161, 163-165 CRAF fixatives 44
Cerenkov critical factors during sectioning 49
counting 376,390 critical point drying 55
limit 376,377 crop rotation 6
CERES-model 534 cryoprotective substances 42
CF-IRMS (s. mass spectrometry) cryoscanning electron microscopy 56, 57
charged coupled device (CCD) 310,311 CI (s. computed tomography)
charges, electric 238, 258 64Cu 391
chemiluminescence 375,376 Curie (Ci) 372
chitin 64 cytokinin 14
Subject Index 575
D CZH) 369,379,382,401 enzyme histochemistry 46

3-D view of roots 246 epoxy resins 45, 55
DAISY-model 532 equation
Darcy's law 520 of continuity 511
debris, organic 190,192,196,198 (basic) transport 511,513,520
decanting technique 63, 307 flux 520
decay, radioactive 372 equilibrium, functional 9
degradat ion of cytoplasm 37 estimation procedure/method, visual 191,
demography 279 203
densitometry 379 ethylene (ethene) 93,492
depletion, fractional 518,519,527 eucalypt 90, 275
depth of field 60 Euphorbiaceae 549
deterministic 126 evaporation 463
deuterium 483 excitation, blue light 56
deuterium/hydrogen ratio 484 experimental design 159
dewaxing agent 39 exudation (s. root exudation, xylem sap)
dielectric constant 468
diffusion coefficient 511,513,515,518,520
diffusion 513,515,518,530,533 FD (s. frequency domain)
diffusivity 511,523,526 59Fe 391
digitising 218,219,228,254,336 fertiliser recovery 367, 383
dimethoxypropane, acidified 44 fescue 332
direct method 61,64 fibroscopes 570
Dirichlet tessellation 230, 530, 536 Fick's first law 511
distance transform 226 field bean 195
distances, three dimensional 230 field experiments 381,383
drought resistance 263 fixatives 43
dual energy scanning 353 coagulating 42, 43
dual-source gamma-ray scanners 468 CRAFT 44
duodenoscopes 245 chemical 42
dye(s) (s. also stain) 307,309 gel forming 42
solution 308 flooding 93, 95
fluorescein 215
fluorescence microscopy 53, 54
Eadie-Hofstee plot 446 flux(es) 367
edge enhancement 318 convective 511,520
efflux kinetics 366 diffusive 511
electrical resistance blocks 471 vanishing 514
electrodes force
platinum 91 sensors 559
ion -selective 434 transducers 89, 57l
electron energy loss spectroscopy 57 fractal(s) (pseudo) 123,125,136,158,
electrostatic charge 50 166
embolism 482, 487 branching analysis 182
EMF (s. mycorrhizae) dimension 126
endodermis 524 geometry 326
endoscopes 245, 567 frequency domain method (FD) 469,470
576 Subject Index
gamma (r) counting 374,377 Histoclear 39,47

gamma densiometry 467,475 histogram 332
gamma probe 476 brightness 308
gamma (r) radiation 372 HPLC-MS (s. mass spectrometry)
gamma (r) ray 344,351-352,359 hydraulic conductance 20,489,491, 524
beam hardening 351 hydraulic conductivity 80,86,486,487,489,
gas sampling 101 490,520
gas-chromatograph-MS (GC-MS) 373 hydraulic lift 85,367,463,474,475
gauges hydrogen (H) 401
strain 562, 569, 571 hypoxia 83
force 559, 569
torque 569
Geiger counting 390 illumination 59
gel forming fixatives 42 image analysis 59,305-336,379
geometry, planar 519 so~are 226,254,283,328,570
geostatistics 166 system 29,262,314,328,567
co-kriging 169 image(s)
kriging 169 acquisition 310
semi-variance 167, 168 binary 320, 334
semi-variograms 167, 168 compression 316
gerbera 94 editing of segmented 319
GIS-software 226 enhancement 335
gradients, tangential 513 file size 314
grass (land) 10,116,127,151,152,161-164, formal 316
180,187,198,218,222,257,325,335 quality 330, 335
gravitropism 153 resolution 58
grid intersect method (s.line-intersect) segmentation 308,317,318,329,330,332,
groundnut 97,116 333
storage 314
imaging
lH 379 electronic 368
2H 369,379,382,401 spatial distributions of isotopes 374
3H (T) 372,376,378,385,386,401 immobilisation 539
Hakea undulata 410 immuno-analysis 47
half-life 368,372,378,387,388,391 immunological
Hanes plot 446 properties 41
Hartig Net 68 techniques 65
heat impedance factor 513
balance technique 476,479,482 impedance, mechanical 77,87,90,93-95,
probe 476,478 97,98,104,128,407
pulse method 20,476,478,479 indexing handle 239, 335
tracer technique 482 infiltration 463, 485
Helianthus tuberosus 190 ingrowth core 21,177,178,183,184,265,
herbs 246 278
heterogeneity 185 inhibitors, visualising the location of 385
high pressure freezing 41 initial condition 512
Himalayan balsam 551 insulation material 567
Subject Index 577
insurance strategy 5, 6, 9 lodging (resistance) 548,552

intercepts 251 Lolium perenne 7,276
interconversions of root parameters 22-24 London resins 45
iodine 215 LR white resin 45, 68
ion exchange resins 568, 569 LSC (5. liquid scintillation counting)
IRMS (5, mass spectrometry) L-systems 123
iron (Fe) 391 lupin 347,352,415
isopropanol 47 Lupinus albus 410
isotope( 5) 483 Lupinus luteus 420
fractionation 368,369, 37l, 383, 384
equilibrium 369
kinetic 369 macropores 215
mixing ratios 37l magnetic resonance (MR) 379
notation and terminology 370 magnetic resonance imaging (MRI) 354,
ratio (IR) 370 355,359
stable 381 magnification 59
isotopic composition 484 maize 10,26,81,84,88,90,93-95,97,102,
isotropic 243 116,127,132,134,135,148,153,157,
159,163,198,223,309,382,383,387,
391,411,521,531,539,549,551,552,
41K 374,402 561
42K 372,376, 390, 402 manganese (Mn) 391
knives oxidation 265
glass 49 mannitol 82
metal 49 mass flow 513,517,518,533
kriging 169 mass spectrometry 373
accelerator (AMS) 374,387
continuous-flow isotope ratio (CF-IRMS)
labeling 373,382,384
continuous 386, 388 high-sensitivity- 373
pulse 386, 387 high performance liquid chromatograph-
laboratory facilities 380 (HPLC-MS) 373
lactophenol 38 isotope ratio- (IRMS) 373,374,390
leek 198,246,250 secondary ion- 374
Lepidium sativum 37 median longitudinal sections 68
light box 309 membrane 5, 524
light microscopy 41,43,45, 55 filters 570
lighting, uniform 313 integrity 434
lignin 51 semi-permeable 84, 525
Limnobium 61 memory effects 382
line-intersect method 64,66,68,202,203, mesh bag (see ingrowth core)
249,263,320,321,324,325,327,334 mesh size 195,307
Lineweaver-Burk plot 446 metabolite pools, (turnover of) 368,372,
liquid scintillation counting (LSC) 387, 390 373,379
livingldead roots (5. roots, type of) method
LM (5. microscopy) dielectric 468
localisation of callose 51 frequency domain 469
578 Subject Index
gravimetric 465 models

point of contact 215 allometric 2, 28
polarographic 91,100 branching 123,158,229
sucker shoot 496 crop (growth) 13,28,117,120,531,534
time-domain 468 numerical 515,531
methylene blue 215 root anchorage 560
Michaelis-Menten (s. nutrient uptake) root architecture 76, 114, 123, 138,539
microcosms 276 evaluation/validation of 133
microelectrodes 61,432,569,570 root distribution 115, 117, 118, 165, 220
microscopy root length density 4,7,26,114-117,137,
cryoscanning electron 56, 57 224,532
fiuorescence 53, 54, 56 root system 510
light (LM) 41,43,45,55 rooting depth 114-116,137,152,224,
scanning electron (SEM) 35,43,55 532
transmission electron (TEM) 41,43,45, single root 510,511,519,527,529,530,
49,55 536
microtensiometer 472, 475 topological 539
microtome 49 uptake 100,191,514,515,529-537
sliding 48 molybdenum (Mo) 391
millet 205 monolith 14,177,187,189
mineral oiI 55 MR (s. magnetic resonance)
minirhizotrons 14,21,23,132,155,166, MRI (s. magnetic resonance imaging)
178,236-256,257,276,279,283,287, mucilage 411
330,334,570,571 Mucuna pruriens 185
dimension of 238 mycorrhizae 11,15,36,62-69,240,252,262,
horizontal 249, 250 264,265,283,287,367,368,383,415
images 279 vesicular-arbuscular (VAM)/
analysing 254, 330-336 arbuscular m. fungi (AMF) 7,8,10,14,
capturing 254 28,62-66,276,438
displaying 254 arbuscule 17, 63, 438
segmentation 330,331-333,336 external mycelium 438, 439
infiatable (fiexible) 241 extraradical mycelium 440
installation 241 fatty acid profiles 439
installation angle 242, 243, 253 hyphae 438
light leaks 244, 248 hyphae-growing zone 440
observation equipment 301 infection (%) 17
pressurised 239 inoculum 440
rigid 241 metabolic interaction with roots 383
square 238 mycorrhiza-free soil 440
subsampling of 256 nutrient uptake 374,439
tube material 238 root colonisation 62,66,377,439,440
tube diameter 238 spores 438
video cameras, for 245,252,301,312, staining technique 17,63,439
331,567 vesic1e 17,63,438
54Mn 391 ecto- (EMF) 63,66-69,263,287
99Mo 391 fruiting bodies 66
mobilisation 367 infected root tips 68
5ubject Index 579
spores (quantification) 67 nutrient depletion 430

NMR-application for 69 rate 6,514,515,516,517
root age 429
root tissue 425
13N 372,376,388,389,401 root injury 425,435
14N 368,401 transport(ers) 376,388,425
15N 69,368-370,379,383,384,388 variation along root axis 367,429,433,
natural abundance 442 435
depleted N source 441 variation in time 429
22Na 391 Williams Formula 426-428
24Na 376 xylem flux 426
nearest neighbour distance 225 nutrient (re)cycling 274,275,368
needleboard (see pinboard) nutrient accumulation 433
negative control 51 nutrient acquisition 405
nematodes 252 nutrient analysis
neural system architectures 333, 336 ashing 447
neutron probe 467 drying 446
neutron scattering method 467 grinding 446
Newman (see line-intersect method) Kjeldahl digestion 447
nitrate (concentration) 264,519,533 sample preparation 446
nitrogen (N), (soiI) 119,275,277,401 wet acid digestion 447
fixation 61 nutrient depletion 443, 444
uptake 368 zone 368,539
NMR (s. nuclear magnetic resonance) nutrient fluxes 372
nocturnal transfer 476 nutrient solution, hydroponics 57,59,387,
Nod factors 61 391,418,444
nodulation 264 nutrient staus 14
nuclear magnetic resonance (NMR) 21,69, nutrient supply (local) 265
344,354-358,379,380,382-384,387,
474
magnetic field 354 17 0 379
nutrient (ion) compartmentation 372,391 18 0 369,382,401
nutrient (ion) translocation 372,388,391 18 0/ 16 0 Ratio 485
nutrient (ion) uptake 391,523 oats 93, 116, 223
Cmin 19,20,426,443-445 observation pits 258
efflux 367,388,426,441,442 oiI, mineral 55
excised roots 426, 434 onion 415
influx 367,388,426,440-442 organic compounds
Km 19,426,443-445,514 uptake, transport, metabolism 385
kinetics 367,372,435,440,443 translocation 386, 388
local nutrient avaiIability 429,518,533 organic farming 6
maximum uptake rate (lmax) 426, 443, organochemicals 264
445,514,534 orientation of root segments 310,321
mechanisms of 424 osmotica 78, 80
metabolic control 425 overlap of roots 321
Michaelis-Menten kinetics 425,426,445, oxygen (O) 401
514,516 concentration 92
580 5ubject Index
concentration profiles 83 Picea glauca 388

flux 91-94,100,101 picture element size 345
partial pressure 95 pixel 60,310,345
solubility 82 voxel 140,345
transport 367 pinboard 14,21,95,156,178,186-191
pine 148,356
plane of observation 213,214,219,220
31p 69,379 plant demand for nutrients/water 516,526,
32p 61,372,376,378,379,389,401 533,534
33p 389,401 Plantago lanceolata 276
paraffin wax 44, 47 plants
parameter estimation 131 perennials 158,257,262
pathways, metabolic 367 woody 10,246
pea 86,90 plasmolysis 42
PEG (s. polyethylene glycol) plastic-embedded material 50
Pelargonium hortorum 356 plasticity 5
Peltier effect 472,494 plastics 48
penetrometers 86,87,91,97,98,100,104, plate, porous 87
557,568 plough pan 104, 153,213,229
perennials 158,257 points of contact 251,252
perimeter measurement 323 polyethylene glycol (PEG) 79,81,83,84,568
permanent slides 40 polyphenols 37,52
permeability (root) 487 pools
osmotic 524 cytoplasmic 441
pest attacks 264 tracer 442
pesticides 264 nutrient 366
pF 529 poplar 28
pH 531 Populus 275
electrodes 569, 570 porosity, air-filled 86,95, 102, 104
soil 215,262,264,265 positive control 50
phenolic compounds 65 positrons 372
phosphatase pot size 88
activity 66,265,419,423 potassium (K) 391,402
test 569,419 uptake 390
phosphate 26, 28, 539 efflux 390
demand 532 influx 390
depletion zones 389 potato 151,161, 190,223,250
efflux 389 potential
fertiliser 390 gravitational 471
influx 389 leaf water 85, 497, 524
uptake 389,390,521,530,531,532,538 matric 78,80,81,84,85,87,90-92, 104,
phosphorus(P) 61,401,519 471,520,524,568,557,570
photoluminescence 375 matric flux 525, 526
Phragmites australis 420 osmotic 84, 495, 524, 567, 570
phytomers 127 plant 498,567,570
phytoremediation 391 profiles 81
phytosiderophore 391,409 root water 493,495,524, 526, 528, 537
Subject Index 581
total 520 resin(s)

(soil) water 24,78,81,84-86,93, 127, acrylic 45, 55
264,276,356,471,495,494,556,557, epoxy 45,55
567 impregnated blocks 215
xylem water 495 ion exchange 568, 569
preservative, chemical 199 London 45
pressure chamber 20,437,491,493,495 LR white 45, 68
pressure Spurr's 48
head 520,524-526,528,536 resistance blocks, electrical 471
hydrostatic 489, 525 resistance
(maximum) axial root growth 88,90 axial 487
osmotic 524, 525 bending 549
probe techniques 488 resolution 311-313
total 511 interpolated 313
transducers 90 optical 313,314
probe, micro-pressure 90 radiometric 311
profile spatial 311,345-347
samplers 568 resource allocation 3, 26
wall method 14,17,18,21,189,212,215, resource/nutrient capture 2,6, 7
218,221,222 rewetting 480
propylene oxide 44 Rhizobium 61
Prunus 277 rhizoboxes 132,263,264,410,418,423
psychrometers 24,472,475,493,495 rhizodeposition 381,386
pulse-Iabel 386, 387 isotopic labeling technique 367,407
Pw-value 532 quantitative evaluation 407
pycnometer 95 rhizosheats 36
rhizosphere 3,7,256,262,264,265,367,
368,386,388,391,406-424,519,524,
quenching 375,376 526,528,536,570
chemical 375,377 rhizosphere soil 20
colour 375, 377 culture system 416
in situ measurements 417-424
Al complexation 418,422
radioisotope phosphorus 61 autofluorescence 422
rate of nutrient uptake 6,514-517 enzyme activity 419,423
86Rb 372,390,402 Fe(III) reduction 417, 421
RD (s. root directionality ratios) low molecular weight compounds 422
recording devices 310 Mn(IV) reduction 417,421
redox nutrient distribution 423
indicators 568, 570 pH 417,418
potential 93, 94, 100 redox state 419,421
probes 567 staining techniques 56,417,420
status 215 micro sampler 415
region growing approach 333 microtome 415,416
regulations, health and safety 370,380 root mat 415,416,521
release of phenolics 265 sampling 414
replication 165,177,181 thin slicing 56,415
582 5ubject Index
rhizotron 14,21,95,236,257,264,280,283 collection 406, 408, 410

rice 419,559 absorbent material 41O,4ll
ridge detection 333 agarose 410
risk strategy 5, 6, 9 axenic culture 406,412
robotic cameras 312 cellulose acetate filter 410
rock phosphate 530 dialysis membrane 406, 407
root abscission l31 exchange resin 406,407,412
root absorbing power 514 filter paper (disk) 410
root age 245 microscale collection 410
root anatomy 36-55 percolation technique 407
root anisotropy 152,221,223,243 recovery 4ll
root biomass 263,265,274,299,300,374, suction cups 410
427 trap solution 407,408
root branching (indices) 8, 16, 101, 123, diurnal rhythm 409
126,178,237,250,252,262,263,276, effect of micro-organisms 412
277,282,320,326,334,547 microbial degradation 41O-4l3
acropetal 129, l30 release 16,367,381,408,411
root cap 36,37 organic acids 530, 538
root carbon 383 iron-mobilising substances 539
root colonisation 62 retrieval 4ll
root colour 254, 262, 282 resorption 387,408
root compartments 366 variation along root axis 409
root competition 388, 389 root fiuorescence 246
root counts (number of) 14,214,223,241, root functionality 245
249,263,264 root growth 99,100,126,178,183,276,367
root decay/decomposition l31, 178, 185, secondary 158,562
246,254,275,278,282 root growth dynamics 246,254,261,262,
root density 15, 135,241,264 263
root diameter (radius) 7,8,15,60,84,158, root growth models 165
178,182,191,195,205,252,310-312, three-dimensional 166
314,320,325,331,334,427,516,524, two-dimensional projections 166
529 root growth orientation (direction) 127,
root directionality ratios (RD) 253 128,213,220,230,248,251,252,253
root discrimination (from objects) 327,335 root growth periodicity 257
root distribution 8, ll7, 158, 160,219,225, root growth rate 17
229,247,248,264,547 root hair deformation 61
anisotropy 152,243 root hair vitality 60
dustering/aggregation l35, 152, 160,227 root hairs 20,36,57-62,101,237,239,246,
function of 224 252,415,520,522,530,547
non-regular/regular 229,529,530 root histology 36-55
random 227 root imaging 355-357
root dynamics 237,255 rootinjury 407,411,414,425
root elongation (rate) 83,84,86-88,91,93, root intersections 159, 166,216,220,221,
ll5, 127,264 224,249
root end distribution 155 root length 6-8,10,13,14,103,195,201,
root exudates, exudation 19,20,121,265, 205,262,265,275,278,297,310,323,
387,406-414,488 334,427,516
Subject Index 583
per unit area 151,205,224,225 root system 114,530,533

specific (SRL) 7, 13, 15,427 architecture 123-130,276,326,334,539
root length density design 5
cumulative 15 diameter 18
volumetric 10,13, 15, 151, 155, 158, 163, growing 532
178,191,205,216,220,223,224,247, morphology 124
249,251,321,518,523,527,529,532, properties 155
536 spatial distribution 24, 148, 150,219,536
root length intensity 236, 263, 265, 347 spatial variability 166
root longevity 16,21,27,205,237,254,264, structure 150
275-277,279 stationary 532, 536
root mapping 60,135,155,159,178,215, systematic trends 150,160
219,224,225,530 types 148
horizontal 135 root tip counting 14,329
root metabolism 391 root topology 14,263,329
root morphology 157,252,263,334 root tracking 239,251
root mortality 121, 122, 131,254,264,265, rootturnover 177,255,263,264,274,276,
274-276,278,279,283,285,287-289, 279,282,367
299,300 root turnover/dynamics 252,255
root nitrogen content 196 root volume 14
root position effectivity ratio 227,228,530 root wash devices 192,193
root pressure 488, 489, 436 hydropneumatic elutriation system 192,
root production 17,265,275,277-279 567,568
root products 3, 12 root water potential (s. potential)
root proliferation 158 root weight density 205
root properties 252 root weight/mass 14,178,195,201,205,236
root replacement ratio 255 root windows 256-264,410
root resistance 527,529 angle 258
axial 487,495 installation 258
root respiration 20,121, 195,373,383 insulation 260,261
root samples (washed) 306 light leaks 260
storage 195, 198 soil-contact 260
storage temperature 200 root: root interactions 10
root sampling 35, 177 root: shoot ratio 2,6-9, 14,26, 120, 121
hand 179 rooting depth 17,102,115-117,151,166,
mechanised (hydraulic) 179,198,241 248
root screens 265-266 average 249
root severance 264 effective 18, 464
root shape indices 327 mean 249
root shrinking 254 rooting intensity 246
root signals 492 root-microbe competition 367
root size classes 262 roots of different plant species,
root soil interface 79 distinction between 246, 262
root spreading 309 roots, types of
root surface (area) 57,205,320,325,334, adventitious 127,547,559
427,432,524 axile 127
effective 57 branch 157
584 5ubject Index
brown 15 experimental technique 154

coarse 182 gas 101
coronal 549-551 location 159
development 353 position 180
fibrous 11,547,548,553 procedure 154
fine 182,183,346,356 root 278,279-281
functional 19,20 scheme 154
intact 356 soil solution 101
lateral (n lh order) 16,90, 127, 130, 148, strategy 147,180
150,157,356,549,555 timing 165
living or dead volume 180
(differentiating between) 185,190,194, sap fiow 437,478,479,481,482,492,567,
197,198,254,286,287,327,328,330, 568,569,571
335 sap velocity 478
nodal 103,148,157 scaling 147
primary 16,90,127,250,547 scanner 262,311,314,567,569
proteoid 410 3-D 313
secondairy 16 desktop 313
tertiary 16 fiatbed 313
quaternary 16 scintillation 376
quinary 16 scintillation cocktail 374,375
radic1e 16,90, 127 scintillation counting, liquid (LSC) 374,
secondary 16 387,390
seminal 16,86,157 scopes 245
storage 13 sectioning 48
tap 16,130,148,150,153,356,549,551, critical factors dur ing 49
552,555 sections, median longitudinal 68
thin (I or II) 182 seed sterilisation 412
white 15 SEM (s, microscopy)
root-shoot interaction 91, 120, 122 semi-automated systems 336
root -shoot signalling 491 semi-variance 167,168
root-soil contact 7,100,513,517,528 semi-variograms 167,168
root-soil interactions 257 sensors 262
row crops 244 angle 559
rubbertree 130,131,134,148 charge coupled device (CCD) 310-312,
Rubidium (Rb) 402 314
rules, developmental 126 force 559
rye 163 temperature 102,567,568
sensors, angle 559
sensors, force 559
34S 384,385,390,402 separation
35S 372,378,380,390,391,402 mechanical, of roots and objects 306
sampling shadows 314
architecture guided 157 shear meters 557,568
depth 181,183 shear strength 548, 557
design 157 shoot signals 435
experimental design 159 shoot: root ratio (s. root: shoot ratio)
Subject Index 585
shrubs 257 soil-root-shoot pathway 485

simulation 165 solid phase extraction 567 -569
single root (see model) solid scintillation 374
skeleton 323, 326 solutions, analytical SIS, 525
skeletonisation/thinning 322, 334 sorghum 116,223
soaking 192 soybean 100,116,159,223,335
sodium (Na) 391 spatial correlation 227,231
sodium pyrophosphate 214 specific root length (SRL) 7, 13, 15,427
software split -root 90
ARCOS 301 boxes 475,476
Delta-T Mark II 301,315,328 systems 85,94,95,185,377
Delta-T SCAN 315,328,329 spruce 85,243,261,275,549
GEN STAT 227 Spurr's resin 48
MSU ROOTS software 254,279,287,301 SRL (s. specific root length)
NIH-Image 254,301,315,328,329 staining 49
RooTracker 254,301 preferential 328
ROOTEDGE 315,328,329 procedures SI
SURPH 298 plant tissue 36
WinRhizo/MacRhizo 315,328,329 roots 197,307,318
soil vital 328
aeration 86, 91, 94, 98, 104, 119 stains 50, 51, 60
CO, concentration 262 fluorescent 39
constraints 122 standard (isotope ratio) 370-371,373,383
coring 14,17,21,36,177,248,252 standing (Iiving) root intensity 255
hydraulic 241 statistics
fauna 257,263 coefficient of variation 162-164
horizons 215, 223 replication 165
microorganisms/microbes 3,4,7,8,257, t-test 160, 162
384,386 statoliths 36
respiration 383 steady-rate 516-518,525,532,536
moisture 119,265, 276 steady-state 515,525,532,536
gradients 264 Steedman's wax 44,47,50
nitrogen 119,391 stern diameter 497
O, concentration 262 stochastic 126
physical properties 121 storage devices 316
profile (rooted) 117, 119 stress 119
resistance 528, 529 mechanical 86, 87
solution 5, 521 oxygen 83,84,87,90,91,94
sampling 101 temperature 90
sterilisation 412,413 water (drought) 78,79,83-85,87,90,98,
structure 3, 10, 99, 100 104,435
temperature 102, 119,276,297 stresses, combinat ion of 103
water balance 462 suberin 51
water content 366, 568, 570 sucker shoot 496
spatial distribution 346, 353 sugar beet 150, 161, 163, 189,229
volumetric 465, 523 sugar maple 300
soil-plant-water continuum 462 sulphur (S) 384,385,402
586 Subject Index
uptake 390 natural 366

sunflower 116,550,559 preloading 442
supporting medium 41 radioisotope 21,193,368,430
survival rare chemical 430
analysis 288, 298 removal of 434
distribution 288,289, 298 stable isotope 368, 383, 384, 430
survivorship 281,284-286,288,289,297 transformer, linear variable differential
SWATRE-model 536-538 (LVDT) 498
sycamore 243 transmis sion electron microscopy 41,43,
symbiosis 276 45,49,55
symbiotic associations 62 transpiration (rate) 523,524,526,528,
537
actual 524
tannins 37 potential 524
TDR (s. time domain reflectrometry) transport model, the composite 490
TEM (s. microscopy) transport pathways 367
tensile tests 553-555 trees 66, 150, 158, 182, 185,244,549,551,
tensiometer 24,81,472 555,559
tests (mechanical) buttressed rain forest 549
bending 555, 556 forest 257,262, 264
(lateral) pushing or pulling 559,561 fruit 257
tensile 553-555 trench (wall) 155,212,225
thermal tracer techniques 476 triaxial cells 87, 93
thermocouple psychrometers 473,494 trichoblasts 57
thin-sections 21,100,190,215 trough, etching 239
Thlaspi caerulescens 391 t-test 160,162
thresholding 307 tubes
grey-level 308,317,318,332 acrylic 239
tillage pan 104, 153,213,229 ceramic 80
tillage systems 103 pressurised 239
time dom ain reflectrometry (TDR) 468 turf 250
time requirements of methods 218 turnips 81
time, thermal 103, 127 type of roots (s. roots, types of)
tissue
clearing 37
dehydration 44 ultramicrotome 55
embedding 44, 49 ultrasound treatment 36, 57
fixation 41 unit soil area 159
infiltration 37, 38, 43 uprooting 547-549
tobacco 223 resistance 548, 552, 558
tomato 84, 88, 94 upscaling 27,529,530,536
tracers 366,369,374,380,382,386,429, uptake capacity
442 unconstrained 518
accumulation 441 total 518
depletion 441,443 uptake, potential 536
desorption 434 1JV-fluorescence 215,245,246,287,335
distribution 377 1JV-light(ing) 245,246,301,335
Subject Index 587
VAM (s. mycorrhizae) wheat 10,88,97,104,116,127,153,163,195,

vermiculite, equilibrated 79, 83 223,250,330,552,559
Vicia faba 356 spring 551
voids 101,242,248,259 winter 103, 120, 163,229,549,551
Wimisa-model 539
windthrow 548, 560
Wageningen Rhizolab 250 woody tissue 39
Wanulcas-model 539
washing solution 434
washing 63, 190, 191, 195 x-ray 344,345-351,359
automatic 192 beam hardening 350
dispersing agents 194,195 X-Y moving gantry 312
hand 191 xylem
pin-board samples 192 exudate 436
root losses (dry weight, nutrient) root pressure 436
194-196 sap 436
soil cores 191 bacterial growth 436
time 198 fiow rate 437
water nutrients cycled 438
available (unconstrained) 10,79,463, positive hydrostatic pressure 437
464,526,527 pressure chamber 437
efflux 474,475 shoot signal 436, 438
infiltration 215 negative hydrostatic pressure 437
isotopically different 369 xylene 49,54
status 493
supply 79
unconstrained uptake of 527 Young's modulus 553, 555
uptake (rate) 80,348,353,356,368,382,
462,480,516,523,526
water-budget-meter 483 zero-sink 517,519,530,532,534
WATUP-model 536-538 zinc (Zn) 391
wax layers 87,88,95 65Zn 391
wetting,localised 479,480,482 zoom camera 245

Root Methods - A Handbook (PDFDrive) PDF

Uploaded by

Copyright:

Available Formats

Root Methods - A Handbook (PDFDrive) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Root Methods - A Handbook (PDFDrive) PDF

Uploaded by

Copyright:

Available Formats

Root Methods: A Handbook

Springer-Verlag Berlin Heidelberg GmbH

With 108 Figures, 4 in Color, and 39 Tables

Dr. A. Glyn Bengough Dr. Sylvain Pellerin

Prof. Dr. Christof Engels Dr. Siebe C. van de Geijn

Why a New Handbook on Root Methods?

Uptake of water and nutrients is a key process in agricultural and natural

"Root research under natural jield conditions is a step-child of science. The

Acknowledgement. The production of this handbook was possible as part of a

On behalf of the Editors, A.L. Smit

We would like to gratefully acknowledge the kind help of the following

Dr. J. Alegre, ICRAF-PERU, Lima, Peru

Dr. M.B. Kirkham, Department of Agronomy, Kansas State University,

Chapter 1. Root Characteristics: Why and What to Measure ............. 1

Chapter 2. Anatomy and Histology of Roots and Root-Soil Boundary 33

Chapter 3. Control and Measurement of the Physical Environment

Chapter 4. Modelling Root System Growth and Architecture ........... 113

Chapter 5. Sampling Strategies, Scaling and Statistics ................. 147

Chapter 8. Root Observations and Measurements at (Transparent)

Chapter 9. The Measurement and Analysis of Fine Root Longevity . . . . .. 273

Chapter 10. Root Image Analysis and Interpretation .................. 305

Chapter 11. Computer-Assisted Tomography

Chapter 12. Isotope Techniques ................................... 365

Chapter 14. Water Uptake ........................................ 461

Chapter 15. Modelling Water and Nutrient Uptake ................... 509

Chapter 16. Plant Anchorage 545

Appendix: Suppliers ............................................. 567

Subject Index .................................................. 573

Chapter 1: Prof. Dr. D. Atkinson, The Scottish Agricultural College (SAC),

Chapter 2: Dr. Tara S. Gahoonia, The Royal Veterinary and Agricultural

Chapter 4: Dr. L. Pages, INRA-Centre d' Avignon, Ecophysiologie

Chapter 5: Dr. A.G. Bengough, Scottish Crop Research Institute (SCRI),

Chapter 6: Prof. Dr. Maria do Rosârio G. Oliveira, Universidade de Evora,

Chapter 8: Dr. A.L. Smit, Plant Research International (Wageningen UR),

Chapter 9: Dr. J.E. Hooker, School of Applied Sciences, University

Root Characteristics: Why and What to Measure

A.L. Smit el al. (Eds.l, Rool Methods

1.2 Why Study Roots?

Ecological Significance. In many situations, but especially in relation to natural

Resource Allocation. Information on the relative allocation of resources be1ow-

Root Products. Roots may be used as an energy source in tropical production

Basic Biological Information. To obtain basic information on a part of the plant

1.2.1 Ecological Significance

The ecological significance of roots is poorly understood (Atkinson 1991).

roots and root systems.

Resource plentiful Optimal Habitats with acute

Low C allocation High C allocation

1.2.2 Resource Capture

In cultivated situations, but rarely in more natural situations, nutrient supply

1.2.3 Soil Microbes

Soil microbial characteristics must relate to the morphological and physiolog-

ical to the understanding of processes as diverse as the stabilisation of soil

1.2.4 Resource Allocation

There is no clear consensus on the functional significance of individual

Shoot Dry Weight (g)

system, while under conditions of good N supply, shoot growth increased in

1.2.5 Plant Interactions

The existence of different communities of plants on different soils, some of

1.2.6 Soil Structure