Freshwater Biology (1999) 41, 333±345

Ecosystem measures of river health and their response to

riparian and catchment degradation
S. E. BUNN*, P. M. DAVIESy AND T. D. MOSISCH*
*Centre for Catchment and In-Stream Research, Faculty of Environmental Sciences, Griffith University, Nathan, QLD 4111,
Australia
yDepartment of Zoology, The University of Western Australia, Nedlands, WA 6907, Australia

SUMMARY

1. Measurements of ecological patterns are often used as primary biological indicators of

river health. However, these patterns provide little information about important stream
ecosystem processes (e.g. the sources and fate of energy and nutrients). The direct
measurement of these processes is considered fundamental to the determination of the
health of stream and river ecosystems.
2. In this paper we used two basic approaches to assess stream ecosystem response to
catchment disturbance and, particularly, to the loss of riparian vegetation in different
forested biomes across Australia. Benthic gross primary production (GPP) and respiration
(R24) provided measures of the amounts of organic carbon produced and consumed within
the system, respectively. Stable isotope analysis was used to trace the fate of terrestrial and
instream sources of organic matter in the aquatic food web. In a focal catchment in SE
Queensland, additional measurements were taken of riparian attributes, catchment
features and water quality.
3. Baseline measurements of GPP and R24 from undisturbed forest streams provided
reference values for healthy streams for comparison with sites where the catchment or
riparian vegetation had been disturbed. These values of metabolism were low by world
standards in all biomes examined. Preliminary data from the Mary River catchment in SE
Queensland indicated that these parameters were sensitive to variations in riparian canopy
cover and, to a lesser extent, catchment clearing, and predictive models were developed.
The ratio P : R (GPP : R24) was used to determine whether sites were net consumers
(P < R) or producers (P > R) of carbon but this was not considered a reliable indicator of
stream health on its own.
4. Although forest streams were typically net consumers of carbon (P < < R), stable
isotope analysis of metazoan food webs indicated a high dependence on inconspicuous
epilithic algae in some biomes.
5. A dramatic decline in the health of forest streams was observed when GPP substantially
exceeded R24, especially when instream primary producers shifted from palatable
unicellular algae to prolific filamentous green algae and macrophytes. These sources of
instream production do not appear to enter aquatic food webs, either directly through
grazing or indirectly through a detrital loop. Accumulation of these plants has led to
changes in channel morphology, loss of aquatic habitat and often a major decline in water
quality in some of the streams studied.

Keywords: ecosystem health, streams, riparian, primary production, benthic metabolism

Correspondence: S. E. Bunn, Centre for Catchment and In-Stream Research, Faculty of Environmental Sciences, Griffith University,
Nathan, QLD 4111, Australia. E-mail: s.bunn@mail.box.guedu.au

ã 1999 Blackwell Science Ltd. 333

334 S. E. Bunn et al.

Introduction understanding of ecosystem processes in Australian

rivers is poor and there has been little attempt to
The ecological health of rivers and streams is a incorporate these measures into biological monitoring
fundamental and increasingly important water man- programs (e.g. Bunn, 1993, 1995).
agement issue in Australia (Schofield & Davies, 1996). Here we use two basic approaches to assess stream
Definitions of river health and selection of methods ecosystem response to catchment disturbance and,
appropriate for its assessment, however, are subject to particularly, to the loss of riparian vegetation in
considerable debate (Bunn, 1995; Norris & Norris, different forested biomes across Australia. This
1995; Wicklum & Davies, 1995). Commonly, biologists includes the direct measurement of amounts of
measure biodiversity (e.g. Magurran, 1988) and the organic carbon produced and consumed within the
structure of benthic communities (e.g. Furse et al., system, and analysis of the fate of terrestrial and
1984; Marchant, Mitchell & Norris, 1984; Faith, instream sources of organic matter in the aquatic food
Dostine & Humphrey, 1995; Wright, 1995). Others web. These processes are clearly linked to catchment
advocate a focus on the distribution and abundance of and riparian condition, and underlying causes of
specific taxonomic groups, including diatoms (Reid variation are relatively well known. The major aim of
et al., 1995) and other aquatic plants (Whitton & Kelly, this paper is to illustrate how such measures of
1995), and fish (Harris, 1995). ecosystem processes can be used as an important
Patterns of species distribution and abundance are diagnostic tool in the assessment of stream health.
undoubtedly important elements of river health but
often contribute little to an understanding of how a Materials and methods
system works (Harris, 1994), and therefore should not
be the sole consideration. Many goals relating to river Study areas
management and protection refer to ecosystem-level
processes, with the notion that streams and rivers Most of the data presented in this paper are derived
should function in the same way they do in un- from large-scale comparative studies currently in
impacted catchments. This is an important element of progress in three different forested biomes within
the principles of ecologically sustainable development Australia. Data on undisturbed forested streams have
that espouse not only the protection of biodiversity been compiled from studies on the Mary River
but also the maintenance of essential ecological catchment (subtropical rainforest) in SE Queensland,
processes and life support systems (Council of the Johnstone River catchment (tropical rainforest) in
Australian Governments, 1992). Often, patterns of far north Queensland and from the northern jarrah
species richness and abundance are used as surrogate forest (Mediterranean climate, dry sclerophyll forest)
measures of these fundamental processes. Although in south-western Western Australia (Table 1). Addi-
altered function may affect patterns, pattern and tional features of the study streams can be found in
process are not necessarily linked (Barmuta, 1987; Bunn, Edward & Loneragan (1986), Davies (1994),
Bunn, 1995). Hughes et al. (1996), Hancock & Bunn (1997) and
Ecosystem-level processes are ideal measures of Mosisch & Bunn (1997).
stream health because they provide an integrated In addition, we include data from 15 disturbed sites
response to a broad range of catchment disturbances. within the Mary River catchment to illustrate the
Disturbances include increased nutrient concentra- relative importance of riparian and catchment influ-
tions from diffuse or point sources, changes to the ences on river health (Table 2). Measurements of
quantity and composition of organic carbon inputs, catchment features, riparian attributes and water
alterations to the light regime (e.g. riparian shading quality (e.g. nutrients) were collected from 20 stream
and turbidity), and sedimentation. Furthermore, sites in total (including the five undisturbed sites in
ecosystem measurements integrate these factors at a Table 1), ranging from 0 to 100% forest riparian
large spatial scale and allow the health of streams and canopy. In most cases, single sites were sampled in
rivers to be viewed in a catchment context (Minshall individual tributary streams, and streams were
et al., 1985; Frissell et al., 1986). Unfortunately, our considered to be replicates within biomes.

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

 Ecosystem measures of river health 335
Table 1 Features of undisturbed forest stream sites from the three forested biomes in Australia. Values represent means (‹ 1 SE)

Riparian Bankfull Slope Catchment

Biome n Latitude, longitude Climate and vegetation canopy (%) width (m) (%) area (km2)

Johnstone River, 3 145°399E, 17°309S tropical wet rainforest, 83.6 ‹ 0.3 6.3 ‹ 0.5 1.67 ‹ 0.34 1.8 ‹ 0.4
far north Queensland summer rainfall
Mary River, 5 152°409E, 26°459S subtropical rainforest, 72.8 ‹ 2.1 11.7 ‹ 0.7 0.24 ‹ 0.07 9.8 ‹ 1.2
SE Queensland summer/autumn rainfall
Northern jarrah forest, 3 116°059E, 32°509S Mediterranean dry 48.0 ‹ 3.6 3.9 ‹ 1.3 0.96 ‹ 0.36 2.2 ‹ 1.3
SW Western Australia sclerophyll forest,
winter rainfall

n, number of sites.

Catchment and stream channel attributes

sites) were obtained from topographic maps (scale
Data from a GIS on catchment land use (total hectares) 1 : 25 000).
above each of the sites in the Mary River were
supplied by the Queensland Department of Natural Riparian attributes
Resources for the following categories: crops/pasture,
isolated trees (< 0.2 canopy cover projection (CCP)), Photographs of the riparian canopy were taken mid-
open woodland (0.2±20 CCP), woodland (20±50 CCP), stream at 50±100 cm above the water surface at each
open forest (50±80 CCP), and closed forest (> 80 CCP). of the study sites, using a 35-mm camera (Nikonus V)
Latitude and longitude of each site were determined and fish-eye lens. The camera was mounted horizon-
by a hand-held global positioning system (GPS). tally on a surveyor's tripod with an adjustable base
Channel slope over the study reaches was surveyed and oriented with the top edge of the film facing
using a dumpy level. Additional field measurements magnetic north, and the lens facing vertically.
included active channel width (m) and stream aspect Negatives of the resultant hemispherical photographs
(upstream±downstream alignment in degrees). Other (ASA25 black and white document copy film) were
site-specific data (altitude, stream order, catchment digitized and enhanced to maximize the distinction
area, and total kilometres of stream channel above between canopy and sky. Images were then analysed

Table 2 Features of disturbed forest stream sites in the Mary River, SE Queensland

Latitude and Riparian Bankfull Slope Catchment

Site longitude Land use upstream canopy (%) width (m) (%) area (km2)

NF 152°459E, 26°499S Rainforest, pasture 88 8.6 5.6 0.6

KM 152°459E, 26°499S Pasture (dairy) 0 4.7 1.7 0.2
NI 152°449E, 26°489S Partially cleared 76 9.0 2.3 5.8
KC 152°429E, 26°479S Rainforest strip for 60 8.1 0.3 25.0
» 20 m, pasture
MR 153°439E, 26°479S Pasture 77 12.3 2.1 21.5
SR2 152°419E, 26°409S Pasture with trees along bank 44 7.3 0.1 24.2
SA2 152°419E, 26°429S Pasture, trees along bank 49 18.9 0.1 34.3
ST 152°449E, 26°479S Pasture, cleared with trees 36 18.8 2.6 18.2
along bank
GC 152°449E, 26°479S Open forest after » 100 m 57 12.5 2.8 8.7
GF 152°449E, 26°479S Open forest 57 13.0 2.2 8.3
TF 152°459E, 26°479S Open forest with rainforest 52 11.0 1.7 4.2
pockets
TC 152°459E, 26°479S Open forest 75 13.3 0.4 4.2
CC1 152°459E, 26°429S Open forest 58 11.7 1.5 12.0
CC2 152°459E, 26°429S Open forest 0 7.2 0.2 13.1
CC3 152°459E, 26°429S Open forest 35 9.0 0.2 12.1

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

336 S. E. Bunn et al.
for percentage canopy cover, and used to estimate 1985), was found to be negligible in forested streams
direct and diffuse photosynthetically active photon (Davies, 1994) and benthic metabolism alone was
flux densities (PPFD), above and below the canopy considered to be indicative of whole-stream processes.
(ter Steege, 1993). Daily estimates were made for the Small Perspex chambers were used to measure
period of measurement, as well as annual estimates, metabolism from individual cobbles and were dome-
taking into consideration the height of the vegetation shaped (diameter = 29.5 cm, height = 17.5 cm) with a
(measured by clinometer), latitude and longitude, removable watertight plastic base. Larger open-
relevant time zone, altitude, and Julian day. bottom chambers (diameter = 29.5 cm, height =
35 cm) were used in gravel, sand and mud habitats,
Water quality and were sealed by pushing them at least 5 cm into
the substrate.
Water samples were collected from each of the 20 All chambers had a central port for the polaro-
stream sites in the Mary River catchment during graphic oxygen sensor (POS; TPS model 501) and side
August 1996, in 500 mL acid-washed polyethylene ports for a 12-V recirculating pump. Flow within the
bottles. All samples were stored on ice in the field and chambers was maintained at a velocity similar to that
subsequently frozen, and later analysed for the of the particular habitat within the stream. Possible
following nutrients (lower reporting limits as indi- effects of leakage of water (between the open-bottom
cated, all in mg L±1): ammonium nitrogen (0.001), chambers and stream water or exchanges with
oxidized nitrogen (0.001), Kjeldahl nitrogen (0.03), hyporheos) on dissolved oxygen concentrations were
orthophosphate phosphorus (0.001), and total phos- tested in situ at most sites. These showed no net
phorus (0.01). Kjeldahl nitrogen and total phosphorus exchanges of oxygen, even in the presence of a
(TKN and TKP) were determined from acid digests substantial oxygen gradient between the chamber
with automated colorimetric finishes (Bran & Luebbe, water and overlying stream water.
1990). The TKN finish for ammonium was based on Calibration of the POS was conducted in the field
the indophenol reaction (Searle, 1984), while the TKP immediately before deployment. Thick (0.1 mm) POS
finish for orthophosphate was based on the ascorbic membranes were used to minimize drift (see Bunn,
acid acid/phosphomolybdic blue reaction (Murphy & Davies & Kellaway, 1997). Dissolved oxygen within
Riley, 1962). Analytical methods for ammonium each chamber was monitored electronically by the
nitrogen, oxidized nitrogen and orthophosphate POS and datalogger (Dataflow 389) over at least 24 h.
phosphorus (AMM_N, OX_N, PHOS_P) broadly Each datalogger was equipped with a preamplifier to
followed the automated, continuous flow methods power the associated POS. Data were recorded at 10
described in Eaton, Clesceri & Greenberg (1995). The minute intervals and stored in RAM. After the data
method for AMM_N was based on the indophenol were downloaded from the datalogger, a series of
reaction (Searle, 1984), the OX_N method on the computer subroutines were used to convert the
Griess±Ilosvay reaction following cadmium reduction information into units of carbon. This was done by
(Eaton et al., 1995), and the method for PHOS_P on the assuming both a photosynthetic quotient of 1 (e.g.
ascorbic acid acid/phosphomolybdic blue reaction Atkinson & Grigg, 1984; Lambert, 1984; Bender et al.,
(Murphy & Riley, 1962). 1987) and 1 mg O2 = 0.375 mg C.
In initial trials where nutrient depletion was
Community metabolism considered to be a limiting factor (e.g. in habitats of
elevated algal biomass), chambers were flushed at
A widely used technique for determining ecosystem least once over the 24 h measurement period with
level flux of carbon is the measurement of metabolism stream water by opening a side port. However, rates
(e.g. Odum, 1956; Mann, 1975; Naiman, 1983; Bott of primary production in the 10 minute interval before
et al., 1985; Minshall et al., 1992). Community meta- flushing were not significantly different (paired t-test,
bolism was measured by monitoring dissolved oxy- t = 0.091, d.f. = 9, P > 0.05) to those in the 10 minute
gen within benthic Perspex chambers over 24 h (e.g. interval immediately after. This indicates that the
Bott et al., 1978). Water column metabolism, assessed volumes of water within chambers were sufficient to
using a light±dark bottle technique (Keithan & Lowe, prevent nutrient depletion.

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

 Ecosystem measures of river health 337
At the end of the measurement period, the volume quently frozen until prepared for stable isotope
of water enclosed by each chamber was measured in analysis.
situ by determining the depth the chamber was Primary sources were cleaned and rinsed in
pushed into the substrate or, in the case of enclosed distilled water and oven-dried at 60 °C for 36±48 h.
cobbles, using the volume of water displaced by the The dried material was then ground to a powder-like
cobble. Cobbles were wrapped in aluminum foil with consistency using a ring grinder. Exoskeletons of
care taken to trim off excess edges. The foil was later shrimps and crayfish were removed, as were gastro-
dried and weighed, and cobble surface area estimated pod shells to avoid possible contamination from
using a weight/area regression (e.g. McCreadie & nondietary carbonates (see Bunn, Loneragan &
Colbo, 1991). The active (photosynthetic) surface area Kempster, 1995). Aquatic insects were prepared
of the cobbles was determined to be half the total whole and smaller individuals typically were pooled
surface area (Davies, 1994). to obtain sufficient-sized samples. Most invertebrates
Most of the undisturbed upland rainforest streams had voided their gut contents before freezing and any
were characterized by cobble and boulder substrates undigested material remaining is unlikely to represent
(Mosisch & Bunn, 1997). In contrast, streams of the a significant source of contamination. Samples of
northern jarrah forest had predominantly gravel and muscle tissue were filleted from each of the fish. All
coarse sand substrates with cobbles formed from animal samples were oven-dried at 60 °C for 24 h and
concreted laterite (Bunn et al., 1986). Accumulations of ground by hand with a mortar and pestle.
leaf litter and fine organic debris were generally Dried, ground samples were oxidized at high
associated with pools in all streams. The relative temperature and the resultant CO2 and N2 were
proportion of each of the major benthic habitats was analysed with a continuous flow-isotope ratio mass
estimated at each site and a single benthic chamber spectrometer (Europa Tracermass and Roboprep,
was randomly assigned to each habitat representing Crewe, UK). Ratios of 13C/12C and 15N/14N were
> 30% of the stream bed. The amount of carbon expressed as the relative per mille (p.p.t.) difference
produced (benthic gross primary production; GPP) between the sample and conventional standards (PDB
and consumed (respiration; R24) at the stream reach carbonate and N2 in air):
scale was estimated by patch-weighting the habitat-
level metabolism values (e.g. Naiman, 1983).
[( ) ]
δX = Rsample / Rs tan dard − 1 × 1000 (p. p. t.)

where X = 13C or 15N and R = 13C/12C or 15N/14N.

Stable isotope analysis Carbon isotope ratios of aquatic plants can show
considerable spatial and temporal variation, depend-
Stable isotope analysis was used to determine the ing on local conditions (France, 1995; 1996a). This can
structure of the stream food web in three undisturbed complicate interpretation of the role of aquatic vs.
forest sites in each biome: Booloumba, Bundaroo and terrestrial carbon sources to aquatic food webs,
Kilcoy creeks (SE Queensland), Opposum and Dirran especially when algal values overlap with terrestrial
creeks, and an unnamed tributary of Opposum (far values. Algal carbon isotope ratios from most sites in
north Queensland), and Foster, Kronin and Wilson all three regions were, however, distinct from those of
brooks (SW Western Australia). Major primary terrestrial (riparian) sources, and the possible con-
sources were collected by hand, including leaves tributions of the epilithic algae and terrestrial vegeta-
from riparian trees, aquatic macrophytes, filamentous tion to the assimilated carbon in consumers were
algae, and epilithic algae (mostly diatoms). Samples calculated using the following simple two end-point
of fine (250 mm±1 mm) and coarse (> 1 mm) particu- mixing model:
late organic matter were also collected from the
benthos. Aquatic macroinvertebrates were collected (
Palgae = δ13Cconsumer − f − δ13Criparian / )
with a long-handled kick-net (250 mm mesh) from
each of the major habitats. Additional samples of
(δ 13
Calgae − δ Criparian
13
)
shrimps (Atyidae and Palaemonidae), crayfish (Para- where Palgae = proportion of algal carbon; and f = iso-
stacidae) and fish were collected using a seine net. All topic fractionation (p.p.t.). In all cases, an isotopic
samples were kept on ice in the field and subse- fractionation (f) of 0.2 was used (see France, 1996a).

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

338 S. E. Bunn et al.
Macrophytes (mainly Vallisneria) and filamentous day±1 in far north Queensland during summer
algae (mainly Spirogyra) were present in some of the (Table 3). Typically, rates of GPP and R24 in SE
streams, particularly in the open canopy sites in the Queensland were intermediate. All P:R ratios in all
Mary River. However, stable carbon and nitrogen biomes and seasons were less than 1 and rates of GPP
isotope ratios of macrophytes and filamentous algae and R24 were higher in summer than in winter
were distinct from those of most consumers and were (Table 3).
not considered to be a major contributor to the food
web (see also Bunn & Boon, 1993; France, 1996b; Bunn Stable isotope analysis of forest stream food webs
et al., 1997).
In SW Western Australia and SE Queensland forest
Data analysis streams, algal carbon isotope ratios were more 13C-
depleted than terrestrial (riparian) sources (Table 4).
Mean (SEM) rates of GPP, R24 and P : R ratios were In contrast, algal carbon isotope ratios from streams in
determined for the forested sites from the three biomes the wet tropics in northern Queensland were more
by patch-weighting habitat-level data. In the Mary enriched (mean d13C ‹ 1 SE = ± 21.8 ‹ 0.1 p.p.t) than
River catchment, the relationship between metabolism terrestrial sources (mean d13C ‹ 1 SE = ± 30.4 ‹ 0.2
and water quality, catchment conditions and riparian p.p.t.). These values were used to estimate the
attributes was also assessed. Stepwise multiple regres- contributions of algal carbon to consumer biomass.
sion, with hierarchical inclusion, was used to deter- Mean d13C values for fine and coarse particle detritus
mine which of the independent variables accounted (FPOM and CPOM) were similar to those for riparian
for most of the variation in stream metabolism (GPP vegetation at all sites.
and R24). Appropriate transformations were used to Stream invertebrates in SW Western Australia were
normalize the distribution of the data. The regression clearly dependent on terrestrial carbon sources in
analysis was conducted using stream reach data gravel and pool habitats (Table 4). Although the
(estimated by patch-weighting habitat measures) and contribution of algae to invertebrates sampled from
data from individual cobbles. Cobbles were a major cobbles was higher (56%), these represented a minor
component of the benthic habitat at all sites sampled in proportion of the total stream habitat. Most inverte-
the Mary River catchment. brates analysed from general collections in SE
Queensland had carbon isotope signatures more
Results similar to terrestrial than algal sources (Table 4).
However, animals collected from cobbles had carbon
Benthic metabolism in forest streams isotope signatures very similar to epilithic algae. Most
consumers (both invertebrates and fish) sampled from
Rates of benthic metabolism were low in forested sites the three forest streams in far north Queensland had
from all three biomes (Table 3). Mean rates of GPP values similar to epilithic algae and quite distinct
ranged from 86 mg C m±2 day±1 in SW Western from terrestrial sources (Table 4).
Australia during winter to 198 mg C m±2 day±1 in
far north Queensland during summer. Mean rates of Riparian vs. catchment influences on ecosystem pro-
benthic respiration (R24) were much higher than GPP cesses
with values ranging from 168 mg C m±2 day±1 in SW
Western Australia during winter to 380 mg C m±2 Of the entire suite of catchment, riparian and water
Table 3 Mean rates of reach-scale metabolism from undisturbed forested sites in the three biomes sampled. Standard errors are
indicated in parentheses

GPP mg C m±2 day±1 R24 mg C m±2 day±1 P:R

Biome Summer Winter Summer Winter Summer Winter

Johnstone River 198 (31) 164 (23) 380 (36) 264 (63) 0.52 (0.04) 0.62 (0.06)
Mary River ± 152 (58) ± 174 (43) ± 0.87 (0.22)
Northern jarrah forest 125 (12) 86 (21) 255 (16) 168 (22) 0.49 (0.02) 0.51 (0.19)

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

 Ecosystem measures of river health 339
Table 4 Stable carbon isotope ratios of epilithic algae, riparian vegetation and aquatic consumers from undisturbed sites in three forest
biomes. The estimated contribution of algal carbon (%) to the biomass of consumers is also given. Numbers of individuals analysed are
given in parentheses

Mean d13C ‹ 1 SE (ppt) (n) Mean d13C ‹ 1 SE (ppt) (n) % algal contribution
of sources of consumers to consumers

Biome Epilithic Riparian Benthic Benthic

habitat algae vegetation invertebrates Fish invertebrates Fish

Johnstone River ±30.4 ‹ 0.2 (9)

Cobbles ±21.8 ‹ 0.1 (9) ±22.5 ‹ 0.6 (10) 86
General collection ±24.0 ‹ 0.2 (56) ±22.0 ‹ 0.3 (15) 72 100
Mary River ±27.4 ‹ 0.1 (4)
Cobbles ±31.0 ‹ 1.1 (4) ±31.8 ‹ 0.4 (7) 100
General collection ±26.3 ‹ 0.2 (59) Not sampled 26 Not sampled
Northern jarrah forest ±26.9 ‹ 0.1 (235)
Gravel ±27.6 ‹ 0.2 (48) Not sampled 19 Not sampled
Cobbles ±31.7 ‹ 0.5 (64) ±29.4 ‹ 0.2 (87) 56
Pools ±26.7 ‹ 0.2 (54) 0

quality parameters, riparian cover accounted for most estimating percentage of habitats and scaling from
of the variation in rates of benthic metabolism in the patch-weighted measurements may contribute to the
Mary River during winter (Table 5). Riparian canopy unexplained variation in the reach-scale data. Ripar-
cover explained 44% of the variation in GPP at the ian canopy cover explained 68% of the variation in
stream reach scale (P < 0.01). Similarly, canopy cover GPP (P < 0.01) and 66% of the variation in R24
explained 32% of the variation in R24 (P < 0.01). There (P < 0.01) on cobbles, the major habitat in these
was a significant correlation (r = 0.92, P < 0.001) streams (Table 5, Fig. 1).
between GPP and R24, suggesting that much of the In both the stream reach and cobble-only analyses,
measured respiration was due to algae. Errors in the proportion of land cleared for crops or pasture
was the next most important parameter, explaining an
additional 14% of the variation in GPP in both cases. It
Table 5 Results of stepwise multiple regression with hierarch- is notable that no other catchment parameters and no
ical inclusion of benthic metabolism against catchment, riparian
water quality parameters contributed significantly to
and instream parameters, for 19 stream sites in the Mary River,
winter 1996. The distribution of percentage crops was normal-
the explained variation in stream metabolism
ized by an arc-sine square root transformation. Variation (Table 5).
explained by each variable (R2), and significance levels (P) are
shown. Positive or negative values of the slope of the relation-
ships are presented in parentheses

Dependent variable Independent variables R2 P

Stream reach
GPP Riparian canopy 0.44 (±) < 0.01
% Crops and pasture 0.14 (+) < 0.01

R24 Riparian canopy 0.32 (±) < 0.01

Velocity 0.19 (±) < 0.01
% Crops and pasture 0.11 (+) < 0.01
Cobbles
GPP Riparian canopy 0.68 (±) < 0.01
% Crops and pasture 0.14 (+) < 0.01 Fig. 1 Relationship between benthic metabolism (O, GPP and X,
R24) and riparian canopy cover for the Mary River, winter 1996.
R24 Riparian canopy 0.66 (±) < 0.01 Measurements are based on cobble samples from 19 of the 20
% Closed forest 0.08 (±) < 0.01 sites (site SA2 had only a gravel/sand substrate).

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

340 S. E. Bunn et al.
Discussion (SE Queensland). Grazers are a conspicuous compo-
Carbon flux in healthy forest streams nent of the fauna in these forest streams (Bunn &
Hughes, 1997).
Measurements of benthic metabolism in undisturbed Based on these observations of carbon flux, we
forest streams provide a baseline or reference condi- would define healthy forest streams as having low
tion for comparison of river ecosystem processes in rates of GPP and R24, and as being net consumers of
disturbed streams, and are indicative of natural carbon (P < R), although the ratio P : R is not
ecosystem function. As predicted for forest streams considered to be an adequate measure of the health
elsewhere in the world (Vannote et al., 1980; Minshall of stream systems on its own. Although P : R ratios
et al., 1983, 1985), undisturbed streams in the three can be estimated relatively simply using open system
biomes were net consumers of carbon and can be measurements without the need for detailed measure-
regarded as heterotrophic (P : R < 1). However, ments of rates per unit area (Owens, 1974), they
although P : R ratios were similar to those reported require a substantial metabolic contribution to over-
elsewhere, values of GPP recorded here are at the low ride the effects of atmospheric re-aeration. This is not
end of the world scale (Woodwell et al., 1978; Webster always possible in forested upland streams where
& Meyer, 1997). The GPP values of these streams are background metabolic rates are very low (e.g. Bunn &
about half that measured in a glacial meltwater stream Davies, 1990). Furthermore, Bunn et al. (1997) found
in the Antarctic (McKnight & Tate, 1997) and similar that P : R ratios in a highly disturbed cane-land
to subarctic streams studied by Naiman & Link (1997). stream in the wet tropics fluctuated considerably on a
Clearly, factors other than shade and temperature daily basis. Variations in cloud cover were thought to
contribute to the low productivity observed in these have influenced the unstable balance between high
Australian streams (see Bunn & Davies, 1990). rates of production and respiration.
Treadwell, Campbell & Edwards (1997) reported Although terrestrial carbon is a major source of
higher rates of GPP in Keppel's Creek in Victoria food for aquatic consumers, microalgae can also be
than those observed here. However, they reported important, at least at patch scale and possibly at
`considerably elevated' instream nitrogen levels as the whole-reach scale in some streams and/or seasons
result of recent logging, which may have stimulated (see also Rosenfeld & Roff, 1992). This is despite the
algal production to levels higher than expected for low levels of primary productivity and despite
undisturbed forest streams. enormous inputs of riparian carbon. Factors that
Even though undisturbed streams in all biomes influence the productivity and composition of stream
were net consumers of carbon, this does not necessa- microalgae could have a significant impact on stream
rily imply that stream food webs were driven by food webs.
terrestrial inputs of organic matter. As in most aquatic
systems, only a small fraction of the total carbon What are unhealthy streams?
present is likely to be assimilated by metazoan
consumers. Not all carbon is of sufficiently high The amounts of organic carbon produced within the
quality for metazoan consumers and not all is truly stream system, and the fate of this and terrestrial
available, because other factors prevent consumers sources of organic matter in the aquatic food web, are
from reaching some sources. Stable isotope data undoubtedly sensitive to changes in catchment land
confirmed that stream invertebrates in the northern use (e.g. increased nutrient runoff) and, particularly,
jarrah forest were almost entirely dependent on to changes in riparian condition. In the absence of
terrestrial carbon. The extremely low rates of GPP in riparian shade, large vascular plants and filamentous
these streams (Davies, 1994; this study) and the low algae often proliferate, restricting flow, trapping
diversity of algal grazers (Bunn & Davies, 1990) sediment and ultimately resulting in marked changes
suggest that algal sources of carbon are not an to available habitat and lowered water quality (Wade,
important food source. However, epilithic algae 1994). A spectacular example of this is the excessive
played a disproportionately important role in the growth of para grass in stream channels in the cane
food web of streams in the tropics (far north Queens- lands of northern Queensland, which clearly has had
land) and, at least in cobble habitats in the subtropics a dramatic effect on channel morphology, flood

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

 Ecosystem measures of river health 341
capacity and aquatic ecosystem function (Bunn et al., Data obtained for the Mary River catchment
1997, 1998). Although macrophytes can be conspic- indicate that powerful predictive models can be
uous components of larger river systems and are often developed that link ecosystem measures of stream
assumed to be important sources of carbon for aquatic health with likely causal factors. Baseline values for
consumers (e.g. Webster & Benfield, 1986; Mann, GPP and R24 obtained for undisturbed forest sites give
1988), recent studies provide little evidence of a the natural state or reference condition (see Fig. 2).
significant contribution to aquatic food webs (Hamil- These are associated with riparian cover of > 73% (the
ton, Lewis & Sippel, 1992; Bunn & Boon, 1993; France, mean for undisturbed forest streams, see Table 1).
1996b). Using the regression equations developed for the
Direct changes to the carbon dynamics of streams Mary River, reductions in canopy cover below 73%
and rivers associated with riparian removal have a would be expected to result in the stimulation of GPP.
tremendous impact on ecosystem function and river This could result in increased secondary production
health, particularly if coupled with increased nutrient initially if the increased GPP originates from palatable
inputs. Although eutrophication is a consequence of microalgae. In the true sense, this might be regarded
high nutrient levels, it is the accumulation of uncon- as an impact (i.e. departure from reference state), but
sumed plant biomass that ultimately leads to water one that may be regarded as acceptable. However, we
quality problems, loss of habitat and major decline in would predict major changes in stream health when
stream ecosystem health and biodiversity. In many GPP exceeds R24 and the stream shifts to autotrophic
cases, changes to the carbon dynamics are also dominance; in this case, when canopy cover is less
accompanied by other impacts triggered by the loss than 40±50%. This relationship is specific to the Mary
of riparian vegetation. Increased water temperatures, River catchment, however, and has yet to be tested in
loss of instream habitat and sedimentation make an the other biomes.
important contribution to the demise of the stream. An important consideration is the ability to predict
The distinction between extreme states of degrada- when increased GPP switches from palatable micro-
tion, as occur with the total loss of riparian cover algae to filamentous algae and/or macrophytes.
described above, and undisturbed healthy streams is Filamentous algae are known to have higher light
so obvious that elaborate measurements of carbon requirements than diatoms or benthic cyanobacteria
flux would tell us little more than a visual inspection. (Richardson, Beardall & Raven, 1983; Langdon, 1988;
However, measurements of these ecosystem processes Hill, 1996). Much of the variation in below-canopy
may provide an early warning of potential problems light (estimated from canopy attributes) in sites in the
(see Bunn, 1995). Slight increases in light and Mary River can be explained by canopy cover. Mean
nutrients (e.g. associated with reduced riparian daily light (mol m±2 d±1) = 48.1±0.48 ´ canopy (%)
cover) may have a positive effect on stream produc- (r = 0.90, P < 0.001). Using this relationship, we
tivity (e.g. Hawkins et al., 1983), but at what threshold
does the system switch to excessive production of
nonpalatable forms of plant production?

Predicting changes in stream health

If measures of stream ecosystem processes (e.g. GPP

and R24) are good indicators of stream health they
should at least be sensitive to variation in catchment
and, specifically, riparian condition (levels of shading
and nutrients in particular). It should be possible to
develop predictive models between these parameters
and potentially causal factors and, if stream health (as
measured by these processes) has declined, strategies Fig. 2 Predictive model of changes in stream metabolism with
for rehabilitation can be recommended that are land use, based on cobble data from the Mary River catchment,
directed at the suspected cause. SE Queensland. Undisturbed forest sites are also shown.

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

342 S. E. Bunn et al.
would estimate that the optimal light threshold for Conclusions
diatoms (< 9 mol m±2 d±1; Steinman et al., 1989)
would be where canopy cover exceeds 80%. In Direct measures of ecosystem processes (e.g. GPP and
contrast, filamentous green algae such as Cladophora R24) and of the fate of organic carbon in aquatic food
(» 26 mol m±2 d±1; Graham et al., 1995), would require webs should be an integral component of the routine
a canopy cover of less than 50%. Although other assessment of stream health. These two approaches
biological and physical factors (e.g. nutrients, grazing, can be used to establish baseline values for undis-
stream discharge) could also affect the productivity turbed systems (as in other biomonitoring
and composition of aquatic plants in these streams, approaches) and incorporated into monitoring and
we would predict that nuisance growths of filamen- compliance guidelines (e.g. ANZECC, 1992). Impor-
tous algae would be prevented if riparian canopy tantly, this approach satisfies many of the essential
cover was maintained (or restored) to above this level. criteria for bioindicators of stream health. The
Similar relationships have been established elsewhere methods are relatively rapid and inexpensive com-
between the extent of riparian cover and plant pared with laborious assessments of benthic commu-
biomass (e.g. Canfield & Hoyer, 1988) or production nity structure (see Bunn, 1995). Perhaps more
(e.g. Gregory et al., 1991). importantly, the measures are sensitive to variation
Given that riparian canopy cover is a powerful in factors known to be causal, particularly riparian
predictor of stream ecosystem function in the Mary and catchment condition, and, potentially, to other
River, it could be argued that measures of canopy aspects of water quality. Powerful but simple pre-
cover alone could be used to predict stream health dictive models can be developed and, if stream health
without the need for detailed measures of ecosystem (as measured by these processes) has declined,
processes. Although catchment clearing also explains strategies for rehabilitation can be identified.
a significant amount of the intersite variation, there
are no other confounding factors that may affect the Acknowledgements
predicted relationship between riparian canopy cover
and stream function. We gratefully acknowledge the assistance of staff
This does not appear to be the case in other from the Queensland Department of Natural
biomes we have studied. Preliminary data from Resources and the Water and Rivers Commission of
degraded streams in the wet tropics of far north Western Australia, especially John Amprimo, Heather
Queensland indicate that high benthic sediment Hunter, Damon Telfer, Steve Kelly and Jeff Kite.
loads can limit GPP in open pasture sites (zero Permission to collect in state forests and national
canopy cover), but lead to very high rates of R24. For parks was given by the Department of Conservation
example, the mean GPP for Clemensen Creek, a and Land Management (Western Australia), and the
degraded tributary of the upper Johnstone River, Queensland Departments of Natural Resources and
was only a quarter of that recorded from nearby Environment. Chris Marshall, Sue Creagh and
forested streams, whereas R24 was approximately Michelle Jenkins assisted with the collection of field
four times higher (see also Bunn et al., 1998). High data and laboratory analysis of samples. This project
respiration rates in sediments have also been found was funded by the Land and Water Resources
in clear-cut streams in the Cascade Mountains, Research and Development Corporation under a
Oregon (Murphy, Hawkins & Anderson, 1981). In National Riparian Lands Program grant (GRU17) to
some salinized catchments in SW Western Australia, SB, and a National River Health Program grant
rates of benthic metabolism are lowest at saline sites (UWA14) to PD.
(unpublished data). Departures from the predicted
relationship between canopy and stream ecosystem
function could therefore be used to infer the nature References
of these other impacts. For example, increased ANZECC (1992) National Water Quality Management
turbidity would be reflected in lower benthic GPP Strategy. Australian Water Quality Guidelines for Fresh
than expected, whereas high nutrient inputs may and Marine Waters. Australian and New Zealand
have the opposite effect. Environment and Conservation Council, Canberra.

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

 Ecosystem measures of river health 343
Atkinson M.J. & Grigg R.W. (1984) Model of a coral reef Bunn S.E. & Hughes J.M. (1997) Dispersal and recruit-
ecosystem. II Gross and net benthic primary production ment in streams: Evidence from genetic studies. Journal
at French Frigate Shoals, Hawaii. Coral Reefs, 3, 13±22. of the North American Benthological Society, 16, 338±
Barmuta L.A. (1987) Polemics, aquatic insects and 346.
biomonitoring: an appraisal. The Role of Invertebrates Bunn S.E., Loneragan N.R. & Kempster M.A. (1995)
in Conservation and Biological Survey (Ed. J.D. Majer), Effects of acid washing on stable isotope ratios of C
pp. 65±72. Department of Conservation and Land and N in seagrass and penaeid shrimp: implications for
Management, Western Australia. food web studies using multiple stable isotopes.
Bender M., Grande K., Johnson K., Marra J., Williams Limnology and Oceanography, 40, 622±625.
P.J.L., Sieburth J., Pilson M., Langdon C., Hitchcock G., Canfield D.E. Jr & Hoyer M.V. (1988) Influence of
Orchardo J., Hunt C., Donaghay P. & Heinemann K. nutrient enrichment and light availability on the
(1987) A comparison of four methods for determining abundance of aquatic macrophytes in Florida streams.
planktonic community production. Limnology and Canadian Journal of Fisheries and Aquatic Sciences, 45,
Oceanography, 32, 1085±1098. 1467±1472.
Bott T.L., Brock J.J., Cushing C.E., Gregory S.V., King D. Council of Australian Governments (1992) Strategy for
& Petersen R.C. (1978) A comparison of methods for Ecologically Sustainable Development. Australian Govern-
measuring primary productivity and community re- ment Printing Service, Canberra.
spiration in streams. Hydrobiologia, 60, 3±12. Davies P.M. (1994) Ecosystem ecology of upland streams of
Bott T.L., Brock J.T., Dunn C.S., Naiman R.J., Ovink R.W. the northern jarrah forest, Western Australia. PhD Thesis,
& Peterson R.C. (1985) Benthic community metabolism Department of Zoology, The University of Western
in four temperate stream systems: an inter-biome Australia.
comparison and evaluation of the river continuum Eaton A.D., Clesceri L.S. & Greenberg A.E., (Eds.) (1995)
concept. Hydrobiologia, 123, 3±45. Standard Methods for the Examination of Water and
Bran J. & Luebbe G.J. (1990) Individual/simultaneous Wastewater, 19th edn. American Public Health Associa-
determination of nitrogen and/or phosphorus in BD acid tion, Washington, DC, 15 pp.
digests. Industrial Method no. 167±B590:-XX/329±74± Faith D.P., Dostine P.L. & Humphrey C.L. (1995)
74W/B, Bran & Luebbe GmbH, Norderstadt, Germany. Detection of mining impacts on aquatic macroinverte-
Bunn S.E. (1993) Riparian-stream linkages: research brate communities: results of a disturbance experiment
needs for the protection of in-stream values. Australian and the design of a multivariate BACIP monitoring
Biologist, 6, 46±51. program at Coronation Hill, Northern Territory.
Bunn S.E. (1995) Biological monitoring of water quality in Australian Journal of Ecology, 20, 167±180.
Australia: workshop summary and future directions. France R.L. (1995) Critical examination of stable isotope
Australian Journal of Ecology, 20, 220±227. analysis as a means for tracing carbon pathways in
Bunn S.E. & Boon P.I. (1993) What sources of organic stream ecosystems. Canadian Journal of Fisheries and
carbon drive food webs in billabongs? A study based Aquatic Sciences, 52, 651±656.
on stable isotope analysis. Oecologia, 96, 85±94. France R.L. (1996a) Carbon-13 conundrums: limitations
Bunn S.E. & Davies P.M. (1990) Why is the stream fauna and cautions in the use of stable isotope analysis in
of southwestern Australia so impoverished? Hydro- stream ecotonal research. Canadian Journal of Fisheries
biologia, 194, 169±176. and Aquatic Sciences, 53, 1916±1919.
Bunn S.E., Davies P.M. & Kellaway D.M. (1997) France R.L. (1996b) Stable isotopic survey of the role of
Contributions of sugar cane and invasive pasture grass macrophytes in the carbon flow of aquatic food webs.
to the aquatic food web of a tropical lowland stream. Vegetatio, 124, 67±72.
Marine and Freshwater Research, 48, 173±179. Frissell C.A., Liss W.J., Warren C.E. & Hurley M.D.
Bunn S.E., Davies P.M., Kellaway D.M. & Prosser I. (1986) A hierarchical framework for stream habitat
(1998) Influence of invasive macrophytes on channel classification: viewing streams in a watershed context.
morphology and hydrology in an open tropical low- Environmental Management, 10, 199±214.
land stream, and potential control by riparian shading. Furse M.T., Moss D., Wright J.F. & Armitage P.D. (1984)
Freshwater Biology, 39, 171±178. The influence of seasonal and taxonomic factors on the
Bunn S.E., Edward D.H. & Loneragan N.R. (1986) Spatial ordination and classification of running-water sites in
and temporal variation in the macroinvertebrate fauna great Britain and on the prediction of their macro-
of streams of the northern jarrah forest, Western invertebrate communities. Freshwater Biology, 14, 257±
Australia: community structure. Freshwater Biology, 16, 280.
67±92. Graham J.M., Lembi C.A., Adrian H.L. & Spencer D.F.

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

344 S. E. Bunn et al.
(1995) Physiological responses to temperature and Mann K.H. (1988) Production and use of detritus in
irradiance in Spirogyra (Zygnematales, Charophyceae). various freshwater, estuarine, and marine coastal
Journal of Phycology, 31, 531±540. ecosystems. Limnology and Oceanography, 33, 910±930.
Gregory S.V., Swanson F.J., McKee W.A. & Cummins Marchant R., Mitchell P. & Norris R. (1984) Distribution
K.W. (1991) An ecosystem perspective of riparian of benthic invertebrates along a disturbed section of the
zones. Bioscience, 41, 540±551. LaTrobe River, Victoria: an analysis based on numer-
Hamilton S.K., Lewis W.M. & Sippel S.J. (1992) Energy ical classification. Australian Journal of Marine and
sources for aquatic animals in the Orinoco River Freshwater Research, 35, 355±374.
floodplain: evidence from stable isotopes. Oecologia, McCreadie J.W. & Colbo M.H. (1991) A critical examina-
89, 324±330. tion of four methods of estimating the surface area of
Hancock M.A. & Bunn S.E. (1997) Population dynamics stone substrate from streams in relation to sampling
and life history of Paratya australiensis Kemp, 1917 Simuliidae (Diptera). Hydrobiologia, 220, 205±210.
(Decapoda: atyidae) in upland rainforest streams, McKnight D.M. & Tate C.M. (1997) Canada stream: a
southeast Queensland. Marine and Freshwater Research, glacial meltwater stream in Taylor Valley, south
48, 361±369. Victoria Land, Antarctica. Journal of the North American
Harris G.P. (1994) Pattern, process and prediction in Benthological Society, 16, 14±17.
aquatic ecology. A limnological view of some general Minshall G.W., Cummins K.W., Petersen R.C., Cushing
ecological problems. Freshwater Biology, 32, 143±160. C.E., Bruns D.A., Sedell J.R. & Vannote R.L. (1985)
Harris J.H. (1995) The use of fish in ecological assess- Developments in stream ecosystem theory. Canadian
ments. Australian Journal of Ecology, 20, 65±80. Journal of Fisheries and Aquatic Sciences, 42, 1045±1055.
Hawkins C.P., Murphy M.L., Anderson N.H. & Wilzbach Minshall G.W., Petersen R.C., Cummins K.W., Bott T.L.,
M.A. (1983) Density of fish and salamanders in relation Sedell J.R., Cushing C.E. & Vannote R.L. (1983)
to riparian canopy and physical habitat in streams of Interbiome comparison of stream ecosystem dynamics.
the northwestern United States. Canadian Journal of Ecological Monographs, 53, 1±25.
Fisheries and Aquatic Sciences, 40, 1173±1185. Minshall G.W., Petersen R.C., Cushing C.E., Cummins
Hill W.R. (1996) Factors affecting benthic algae Ð effects K.W., Vannote R.L. & Sedell J.R. (1992) Stream
of light. Algal Ecology Ð Freshwater Benthic Ecosystems ecosystem dynamics of the Salmon River, Idaho: an
(eds R.J. Stevenson M.L. Bothwell & R.L. Lowe). 8th-order system. Journal of the North American Bentho-
Academic Press, San Diego. logical Society, 11, 111± 137.
Hughes J.M., Bunn S.E., Hurwood D.A., Choy S. & Mosisch T.D. & Bunn S.E. (1997) Temporal patterns in
Pearson R. (1996) Genetic differentiation among popu- stream epilithic algae in response to discharge regime.
lations of Caridina zebra (Decapoda: atyidae) in tropical Aquatic Botany, 58, 181±193.
rainforest streams, northern Australia. Freshwater Biol- Murphy M.L., Hawkins C.P. & Anderson N.H. (1981)
ogy, 36, 289±296. Effects of canopy modification and accumulated sedi-
Keithan E.D. & Lowe R.L. (1985) Primary productivity ment on stream communities. Transactions of the
and spatial structure of phytolithic growth in streams American Fisheries Society, 110, 469±478.
in the Great Smoky Mountains National Park, Tennes- Murphy J. & Riley J.P. (1962) A modified single solution
see. Hydrobiologia, 123, 59±67. method for the determination of phosphate in natural
Lambert W. (1984) The measurement of respiration. A waters. Analytica Chimica Acta, 27, 31±36.
Manual on Methods for the Assessment of Secondary Naiman R.J. (1983) The annual pattern and spatial
Productivity in Fresh Waters, 2nd edn. (Eds J.A. Down- distribution of aquatic oxygen metabolism in boreal
ing and F.H. Rigler), pp. 413±468. IBP Handbook 17, forest watersheds. Ecological Monographs, 53, 73±94.
Blackwell Scientific Publications, Oxford. Naiman R.J. & Link G.L. (1997) Organic matter dynamics
Langdon C. (1988) On the causes of interspecific in 5 subarctic streams, Quebec, Canada. Journal of the
differences in the growth-irradiance relationship for North American Benthological Society, 16, 33±39.
phytoplankton. II. A general review. Journal of Plankton Norris R.H. & Norris K.R. (1995) The need for biological
Research, 10, 1291±1312. assessment of water quality. Australian Journal of
Magurran A.E. (1988) Ecological Diversity and its Measure- Ecology, 20, 1±6.
ment. Princeton University Press, Princeton, NJ, Odum H.T. (1956) Primary production in flowing waters.
179 pp. Limnology and Oceanography, 1, 102±117.
Mann K.H. (1975) Patterns of energy flow. River Ecology. Owens M. (1974) Measurements on non-isolated natural
(Ed. A. Whitton), pp. 248±263. University of California communities in running waters. A Manual on Methods
Press, Berkeley, CA. for Measuring Primary Production in Aquatic Environ-

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345

 Ecosystem measures of river health 345
ments. (Ed. R.A. Vollenweider), pp. 111±119. IBP Vannote R.L., Minshall G.W., Cummins K.W., Sedell J.R.
Handbook no. 12, Blackwell Scientific, Oxford. & Cushing C.E. (1980) The river continuum concept.
Reid M.A., Tibby J.C., Penny D. & Gell P.A. (1995) The Canadian Journal of Fisheries and Aquatic Sciences, 37,
use of diatoms to assess past and present water quality. 130±137.
Australian Journal of Ecology, 20, 57±64. Wade P.M. (1994) Management of macrophytic vegeta-
Richardson K., Beardall J. & Raven J.A. (1983) Adaptation tion. The Rivers Handbook. Hydrological and Ecological
of unicellular algae to irradiance: an analysis of Principles. Volume 2. (Eds P. Calow and G.E. Petts).
strategies. New Phytologist, 93, 157±191. Blackwell Scientific Publications, Oxford.
Rosenfeld J.S. & Roff J.C. (1992) Examination of the Webster J.R. & Benfield E.F. (1986) Vascular plant
carbon base in southern Ontario streams using stable breakdown in freshwater ecosystems. Annual Review
isotopes. Journal of the North American Benthological of Ecology and Systematics, 17, 567±594.
Society, 11, 1±10. Webster J.R. & Meyer J.L. (1997) Stream organic matter
Schofield N.J. & Davies P.E. (1996) Measuring the health budgets. Journal of the North American Benthological
of our rivers. Water (May/June) 39±43. Society, 16, 5±13.
Searle P.J. (1984) The Berthelot or indophenol reaction Whitton B.A. & Kelly M.G. (1995) Use of algae and other
and its use in the analytical chemistry of nitrogen Ð a plants for monitoring rivers. Australian Journal of
review. The Analyst, 109, 549±568. Ecology, 20, 45±56.
ter Steege H. (1993) Hemiphot ± a Programme to Analyze Wicklum D. & Davies R.W. (1995) Ecosystem health and
Vegetation Indices, Light and Light Quality from Hemi- integrity. Canadian Journal of Botany, 73, 997±1000.
spherical Photographs. Tropenbos documents 3. Stichting Woodwell G.M., Whittaker R.H., Reiners W.A., Likens
Tropenbos, Wageningen. G.E., Delwiche C.C. & Botkin D.B. (1978) The biota and
Steinman A.D., McIntire C.D., Gregory S.V. & Lamberti the world carbon budget. Science, 199, 141±146.
G.A. (1989) Effects of irradiance and grazing on lotic Wright J.F. (1995) Development and use of a system for
algal assemblages. Journal of Phycology, 25, 478±485. predicting the macroinvertebrate fauna in flowing
Treadwell S.A., Campbell I.C. & Edwards R.T. (1997) waters. Australian Journal of Ecology, 20, 181±197.
Organic matter dynamics in Keppel Creek, south-
eastern Australia. Journal of the North American Bentho- (Manuscript accepted September 1998)
logical Society, 16, 58±61.

ã 1999 Blackwell Science Ltd, Freshwater Biology, 41, 333±345