Environ Monit Assess

DOI 10.1007/s10661-011-1918-9

Faecal contamination of water and sediment in the rivers

of the Scheldt drainage network
Nouho Koffi Ouattara Julien Passerat
Pierre Servais

Received: 5 May 2010 / Accepted: 27 January 2011

Springer Science+Business Media B.V. 2011

Abstract The Scheldt watershed is characterized the Scheldt and the Nethe Rivers. FIB were also
by a high population density, intense industrial quantified in representative point sources (waste-
activities and intensive agriculture and breeding. water treatment plants) and non-point sources
A monthly monitoring (n = 16) of the abundance (runoff water and soil leaching on different types
of two faecal indicator bacteria (FIB), Escherichia of land use) of faecal contamination. The com-
coli and intestinal enterococci (IE), showed that parison of the respective contribution of point
microbiological water quality of the main rivers of and non-point sources at the scale of the Scheldt
the Scheldt drainage network was poor (median watershed showed that point sources were largely
values ranging between 1.4 103 and 4.0 105 E. predominant.
coli (100 mL)1 and between 3.4 102 and 7.6
104 IE (100 mL)1 ). The Zenne River downstream Keywords Microbiological water quality
from Brussels was particularly contaminated. Glu- Rivers Faecal indicator bacteria Sediment
curonidase activity was measured in parallel and contamination Wastewater
was demonstrated to be a valid surrogate for a
rapid evaluation of E. coli concentration in the
river waters. FIB were also investigated in the Introduction
river sediments; their abundance was sometimes
high (average values ranging between 2.1 102 Polluted surface waters can contain a large vari-
and 3.3 105 E. coli g1 and between 1.0 102 ety of pathogenic microorganisms including bac-
and 1.7 105 IE g1 ) but was not sufficient to teria, viruses and protozoa. The main origin of
contribute significantly to the river water conta- these pathogenic microorganisms is the faeces
mination during resuspension events, except for of human and warm-blooded animals; they are
brought to the aquatic environments through the
release of wastewater effluents, surface runoff and
soil leaching. The human sanitary risk linked to
N. K. Ouattara J. Passerat P. Servais (B) the presence of these pathogens depends on the
Ecologie des Systmes Aquatiques, use of the water (drinking, recreational activities,
Universit Libre de Bruxelles, bathing, irrigation, shellfish harvesting) and on the
Campus de la Plaine, CP 221,
pathogen concentration in water.
Boulevard du Triomphe,
1050 Bruxelles, Belgium In aquatic systems, the detection and enumera-
e-mail: pservais@ulb.ac.be tion of all pathogenic microorganisms potentially
 Environ Monit Assess

present is impossible due to the large diversity of Sediments of aquatic environments may consti-
the pathogens, the low abundance of each species tute a reservoir for FIB. Indeed, different studies
and the absence of standardized and low-cost reported high FIB concentrations in sediments
methods for the detection of each of them. Thus, from streams and rivers (Crabill et al. 1999; Craig
for routine monitoring, faecal indicator bacteria et al. 2002a; Smith et al. 2008), lakes (An et al.
(FIB) are usually enumerated to evaluate the level 2002), estuaries and coastal areas (Craig et al.
of microbial water contamination. The abundance 2002a; Roslev et al. 2008). High concentration of
of these FIB is supposed to be correlated with the faecal bacteria in sediments is usually explained
density of pathogenic microorganisms from faecal by the longer survival of FIB in sediments than
origin and is thus an indication of the sanitary in the overlying water (Davies et al. 1995; Craig
risk associated with the various water utilisations. et al. 2004; Lee et al. 2006). A part of the present
For more than a century, total coliforms and fae- study was devoted to evaluate the FIB level in the
cal (also called thermotolerant) coliforms were sediments of the rivers of the Scheldt watershed
the main organisms used as bacterial indicators. and to assess the possible impact of sediment
Nowadays, Escherichia coli and intestinal entero- resuspension on the microbiological water quality.
cocci (IE) are the most frequently used indicators Another objective of this study was to deter-
of faecal pollution as it was demonstrated by epi- mine the main sources of the faecal contamination
demiological studies that they were better indi- of the rivers. The sources of FIB to the rivers
cators of the human risk associated with waters of the watershed were thus studied in order to
than coliforms (Edberg et al. 2000; Fewtrell and quantify and compare the point sources (outfall
Bartram 2001). Recent guidelines for assessing the of treated and untreated wastewaters) and non-
water quality required for different water uses are point sources (surface runoff and soil leaching) of
based on their abundance, i.e. the recent Euro- faecal pollution at the scale of the whole drainage
pean Union Bathing water Directive (EU 2006). network.
In the present study, abundances of E. coli and
IE were used to estimate microbiological quality.
E. coli and IE were enumerated using classical Materials and methods
culture-based methods (plate count on agar me-
dia); in addition, a direct (without cultivation) and Water and sediment sampling
rapid enzymatic method, the measurement of the
glucuronidase activity (George et al. 2000), was During the monitoring survey conducted in the
tested as a surrogate of the culture-based method scope of this study, water samples were collected
for estimating E. coli concentration. in the downstream part of the main rivers of
In the present study, the microbiological quality the Scheldt watershed. Twelve sites were inves-
of the rivers of the Scheldt drainage network was tigated monthly from March 2007 to June 2008
investigated. The Scheldt watershed, which covers (Fig. 1); a total of 16 sampling campaigns were
an area of 22,000 km2 located from the North thus performed. During these campaigns, water
of France to the BelgianDutch border (Fig. 1), samples were collected in the rivers with a plastic
is characterized by a high population density bucket from bridges, halfway between the banks.
(around 500 inhabitants/km2 ), intense industrial Some of the sampling locations were under the
activities and intensive agriculture and breeding. tide influence; during the survey, these sites were
Due to these anthropogenic pressures, one can systematically sampled during the low tide period.
assume that the microbiological water quality of Samples were stored in 1-L sterile bottles, kept at
the main rivers of the Scheldt drainage network is 4 C and analysed within a maximum of 6 h after
low. A first objective of this study was to evaluate collection.
the level of faecal contamination of the water of Sediments were collected from the upper layer
the main rivers. For this purpose, a monitoring of the river sediment (approximately 10 cm from
survey was organized in the main rivers of the the surface of the sediment) using a hand trowel.
watershed. Sediment samples were collected at the same
Environ Monit Assess

sampling sites than water samples (Fig. 1) during and denitrification associated with phosphorus re-
two of the 16 sampling campaigns. Sediment sam- moval (P), and stabilisation pond. For each of
ples were placed into a sterile plastic container, these treatment types, samples were collected dur-
transported to the laboratory and kept at 4 C and ing three sampling campaigns in different WWTPs
analysed within a maximum of 6 h after collection. in the Scheldt basin excepted for the treatment
with PT followed by trickling filtration for which
only one sampling campaign was performed.
Evaluation of point sources of faecal
contamination
Evaluation of non-point sources of faecal
In order to quantify the contribution of waste- contamination
water releases to the FIB load and to determine
the treatment efficiency on FIB removal, sam- Faecal pollution brought to the rivers through
ples were collected in raw and treated waters of soil leaching or surface runoff represents the non-
various wastewater treatment plants (WWTPs) point sources; its origin can be the faeces from
located in the Scheldt watershed. The studied wild animals and grazing livestock and also the
WWTPs had a large range of treatment capacities cattle manure spread on cultivated areas. There-
(from 1,000 to 1,100,000 inhabitant equivalents) fore, the land use can have a major impact on the
and were characterized by various types of water level of microbial pollution brought to rivers by
treatment. The types of treatment considered in soil leaching and surface runoff. Quantification of
the present study were primary treatment (PT; in- faecal contamination of rivers through non-point
cluding screening, grease collection and settling), sources is relatively difficult. In this study, the
PT followed by trickling filtration, PT followed by approach proposed by George et al. (2004) and
activated sludge (AS) for carbon removal, PT fol- Garcia-Armisen and Servais (2007) was adopted
lowed by AS for carbon removal and nitrification to quantify the contribution of non-point sources
(Nit), PT followed by AS for carbon removal, of faecal pollution: FIB concentrations were mea-
nitrification and denitrification (Denit), PT fol- sured in small streams (stream order 1 or 2 ac-
lowed by AS for carbon removal, nitrification cording to the geomorphologic criteria defined by

Fig. 1 Location of
sampling sites in the
downstream part of the
main rivers of the Scheldt
watershed. Ly Lys River
at St-Martens-Leerne,
De Dendre River at
Gijzegem, Ne Nethe
River at Duffel, Dy1 Dyle
River at Gastuche, Dy2
Dyle River at Rijmenam,
Ze1 Zenne River at Lot,
Ze2 Zenne River at
Eppegem, Ze3 Zenne
River at Leest, Ru Rupel
River at Boom, Sc1
Scheldt River at Gavere,
Sc2 Scheldt River at
Uitbergen, Sc3 Scheldt
River at Temse
 Environ Monit Assess

Strahler (1957)) located in rural areas sampled particles on the plate, samples were let to settle for
upstream from any wastewater outfall so that FIB 10 min after sonication as proposed by Anderson
present in the samples resulted only from soil et al. (2005), and aliquots of the supernatant were
leaching and surface runoff. These small streams collected with a sterile syringe (50 mL) and plated
were classified on the basis of their watershed land onto CCA and CEA. E. coli and IE counts were
use: forested areas, cultivated areas and pastured expressed as CFU per gram dry weight sediment.
areas. The sediment was weighted before and after dry-
ing at 105 C for 24 h in order to determine the dry
weight-to-wet weight ratio.
Enumeration of faecal indicator bacteria in water
and sediments D-Glucuronidase activity measurement

E. coli and IE were enumerated in water samples The measurement of -d-glucuronidase activity
by standard plate counts on Chromocult coliform (an enzymatic activity specific to E. coli) has been
agar (CCA) and Chromocult enterococci agar proposed as an alternative to classical enumer-
(CEA), respectively (Merck KGaA, Darmstadt, ation methods to investigate the abundance of
Germany). These two chromogenic growth media E. coli in waters (Fiksdal et al. 1994). George
were shown to be highly specific to their corre- et al. (2000) optimised a protocol for measur-
sponding indicator bacteria. Prats et al. (2008) and ing -D-glucuronidase (GLUase) activity in river
Miranda et al. (2005) reported high percentages water using the substrate 4-methylumbelliferyl-
of specificity, respectively, for CCA enumeration -D-glucuronide (MUG) in a period as short as
of E. coli (96%) for CEA enumeration of ente- 30 min. This method was tested in this study in
rococci (98%) in water samples. CCA and CEA parallel with E. coli plate counts. The protocol
plates were incubated at 36 C for, respectively, 24 used for GLUase activity measurements was as
and 48 h. Plate counts were expressed as colony- follows (Servais et al. 2005): Briefly, the water
forming units (CFU) per 100 mL of sample. sample (10 or 100 mL) was filtered through a
Enumeration of faecal indicators by plate 0.2-m pore-size 47-mm diameter polycarbonate
counts in sediment samples required first to de- filter (Nuclepore). The filter was placed in a
tach the bacteria from the sediment particles. In 100-mL sterile DURAN flask containing 17 mL
the scope of the present study, extraction of FIB of sterile phosphate buffer (pH 6.9) and 3 mL
from sediments by sonication was optimised with of a MUG solution (55 mg of MUG (Biosynth,
the objective of detaching a maximum of bac- Switzerland), and 20 L of Triton X-100 in 50 mL
teria from the particles without impairing their of sterile water) was added to the flask (final
cultivability. Sonication using probes (Labsonic L MUG saturating concentration of 165 mg L1 ).
Braun, 220 V, 50/60 Hz) and a sonication bath The flask was incubated in a shaking water bath at
(47 kHz; 60 W) were tested. Treatments with two 44 C. Every 5 min for 30 min, a 2.9-mL aliquot of
probes (4 and 9 mm diameter), three sonication the 20 mL was put in a quartz cell with 110 L of a
times (1, 2 and 3 min) for each probe and two 1-M NaOH solution to obtain a pH of 10.7 (corre-
intensity levels were tested (60 and 100 W). Three sponding to the maximum of fluorescence of the
sonication times (1, 5 and 10 min) were tested us- methylumbelliferone (MUF)). The fluorescence
ing the sonication bath. For each treatment type, intensity of the aliquot was measured with a SFM
0.5 g of wet sediment was placed in 50 mL of 25 spectrofluorometer (Kontron AG, Zrich,
sterile Ringer solution. The higher recovery of Switzerland) at an excitation wavelength of
FIB from sediments was obtained for both indi- 362 nm and an emission wavelength of 445 nm.
cators using 4 mm sonication probe at 60 W for The production rate of MUF (picomoles of
2 min; this sonication procedure was used during MUF released per minute for 100 mL of sample
the whole study. We verified that this procedure filtered), expressing the enzymatic activity, was
of sonication did not affect the culturability of determined by a linear least squares regression of
the FIB. In order to avoid, a too high load of MUF concentration on incubation time.
Environ Monit Assess

Suspended matter and potential contribution Results

of sediment resuspension to the faecal
contamination in water column Monitoring of faecal contamination in the main
rivers of the Scheldt drainage network
Suspended matter (SM) of water samples was
estimated as the weight of material remaining on a Faecal contamination in the river water column
Whatman GF/F membrane per volume unit after
drying of the filter for 24 h at 105 C. SM data are The faecal contamination level of the main rivers
expressed in milligrams per litre. of the watershed was monitored for 16 months by
In order to calculate the maximum potential monthly measurements at 12 locations. E. coli and
concentration of FIB due to sediment resuspen- IE abundances as well as glucuronidase activity
sion, we considered that, at each sampling site, presented similar trends from one sampling site to
the maximum resuspended sediment in the water another (Fig. 2). The contamination level of the
column corresponds to the highest SM concentra- sampling locations covered a wide range. The less
tion measured during the survey. By this approach contaminated site (Sc3) had median abundances
that considers that all SM in the river originated of 1.4 103 E. coli and 3.4 102 IE per 100 mL
from sediment resuspension and that fixes the while the most contaminated site (Ze3) had me-
SM concentration to its highest observed value, dian abundances of 4.0 105 E. coli and 7.6 104
the hypothesis made about the concentration of IE per 100 mL, representing a range of 2.5 logs for
resuspended sediment in the water column was E. coli and 2.4 logs for IE. The range between the
probably on the upper range of what could be lowest and the highest abundances measured dur-
actually observed. The concentration of potential ing the monitoring reached 4 orders of magnitude
resuspended sediment (the maximum SM concen- (e.g. from 3.7 102 to 3.8 106 CFU/100 mL for
tration observed at each sampling site) was then E. coli). Sites from the Lys River (Ly), the Nethe
multiplied with the geometric mean of the abun- River (Ne) and from the Scheldt River (Sc1, 2 and
dance of each FIB in the sediment to determine 3) were the less contaminated, while the sites from
the potential abundance of resuspended FIB. the Zenne River downstream from Brussels (Ze2
Percentage of FIB potentially originating from and 3) and from the Rupel (Ru), which receives
sediment resuspension in the water column was the water from the Zenne River, were the most
calculated as the ratio of potentially resuspended contaminated.
FIB to the geometric mean abundance of FIB in During the monitoring, a high variability was
the water column. observed in the contamination level of several
sites (Fig. 2). This abundance variability was more
Flow rate data pronounced for IE than for E. coli. The most
contaminated sites were among the most variable
Daily average flow rates data of the different stud- (sites from the Zenne River, the Dyle River and
ied rivers were obtained from the Flemish Envi- the Rupel River) but the most variable site was
ronment Agency (Vlaamse Milieumaatschappij). De, with 2.4 and 3.7 logs between the minimal and
maximal abundances of E. coli and IE, respec-
Statistical analysis tively. Comparatively, the less contaminations
sites (Sc2 and Sc3) were also the less variable, with
The probability test ( p value) associated to linear differences between minimal and maximal abun-
regression was performed using Student test, and dances of 0.7 and 1.5 logs for E. coli and IE, re-
significance was determined at 95% confidence spectively. The river flow rate is often reported in
level. A WilcoxonMannWhitney test was also the literature as a factor affecting the level of fae-
performed to compare the abundances of fae- cal contamination (Schilling et al. 2009). We inves-
cal bacteria from different non-point sources. tigated if variations in the flow rate of the rivers
These analyses were performed using statistical measured on the sampling date could explain the
software R. observed fluctuations in the contamination level
 Environ Monit Assess

7 7
a
6
6
Log (CFU (100 mL)-1)

5
5

Log (IE (100mL)-1)

4

4
3

3 2

1
2
Ly
De
Ze1
Ze2
Ze3
Dy1
Dy2
Ne
Ru
Sc1
Sc2
Sc3
0
6
0 1 2 3 4 5 6 7
b Log (E. coli (100mL)-1)

5
Fig. 3 Loglog linear regression between IE and E. coli
Log (CFU (100 mL)-1)

abundances in the main rivers of the Scheldt watershed.

4 Correlation: log (IE/100 mL) = 0.99 log (E. coli/100 mL)
0.62 (r2 = 0.89; n = 192; p < 0.01)

of the sites. The E. coli and IE abundances (in log

2
units) were positively correlated at three sites (De,
p < 0.05; Dy2 and Sc2, p < 0.01) with the flow
1 rate of the river. Flow rate variations explained
Ly
De
Ze1
Ze2
Ze3
Dy1
Dy2
Ne
Ru
Sc1
Sc2
Sc3

between 32% and 46% of the variability of the

4.5 log abundances (0.32 < r2 < 0.46). No significant
c
4.0
Log (pmol (min)-1(100 mL)-1)

3.5 5

3.0
4
Log (pmol (min-1) (100mL)-1)

2.5

2.0 3

1.5
2
1.0
Ly
De
Ze1
Ze2
Ze3
Dy1
Dy2
Ne
Ru
Sc1
Sc2
Sc3

1
Fig. 2 Box plots in log units of the abundance of E. coli
(a), intestinal enterococci (b), and the GLUase activity
(c) measured in the main rivers of the Scheldt drainage 0
network during a monthly monitoring survey (n = 16). Box 0 2 4 6 8
plots represent the median (horizontal line in the box), the
Log (E. coli (100mL)-1)
lower and upper quartiles (bottom and top box lines), the
10th and 90th percentiles (bottom and top whiskers) and Fig. 4 Loglog linear regression between GLUase activi-
the outliers (circles). Dashed lines represent the maximum ties and E. coli abundances in the main rivers of the Scheldt
admissible value of the 90th percentile for a sufficient watershed. Correlation: log (GLUase Act (picomoles per
quality of bathing waters according to the EU Directive minute per 100 mL) = 0.76 log (E. coli/100 mL) 0.79
2006/7/EC (r2 = 0.87; n = 191, p < 0.01))
Table 1 FIB abundances measured in the sediment and in the water column of the rivers of the Scheldt watershed and estimation of the potential contribution of
sediment resuspension to the FIB load of the water column
River Sampling SM E. coli Intestinal enterococci
site (minmax)a Sedimentb Water columnc Potentially Potential Sediment Water column Potentially Potential
(mg L1 ) (CFU g1 ) (CFU (100 mL)1 ) resuspendedd contributione (CFU g1 ) (CFU (100 mL)1 ) resuspended contribution
(CFU (100 mL)1 ) (%) (CFU (100 mL)1 ) (%)
Lys Ly 456 0.1
Environ Monit Assess

2.1 102 2.7 103 0.1 101 <0.1 1.0 102 4.8 102 <1
Dendre De 8324 1.3 104 5.4 103 4.2 102 7.8 6.6 103 1.1 103 2.1 102 19.1
Zenne Ze1 10393 6.1 103 2.3 104 2.4 102 1.0 3.3 103 5.7 103 1.3 102 2.3
Ze2 1298 3.0 104 3.7 105 3.0 102 0.1 3.8 105 6.7 104 3.7 103 5.6
Ze3 19109 2.2 105 4.6 105 2.4 103 0.5 1.4 105 8.8 104 1.6 103 1.8
Dyle Dy1 12192 7.7 102 2.9 104 1.5 101 0.1 1.6 103 6.2 103 3.0 101 0.5
Dy2 9169 3.6 103 9.1 103 6.2 101 0.7 2.0 103 2.3 103 3.5 101 1.5
Nethe Ne 20113 3.3 105 2.9 103 1.5 103 51.8 1.7 105 6.8 102 7.5 102 109.6
Rupel Ru 1281 1.9 104 2.9 104 1.6 102 0.5 2.4 104 6.4 103 2.0 102 3.1
Scheldt Sc1 19125 7.1 103 2.3 103 8.9 101 4.0 7.8 103 5.2 102 9.8 101 18.9
Sc2 12113 4.0 103 3.8 103 4.5 101 1.2 2.2 103 8.3 102 2.4 101 3.0
Sc3 22124 3.5 104 1.4 103 4.4 102 32.0 1.6 104 3.5 102 2.0 102 58.2
a Minimal and maximal SM concentrations measured during the monitoring survey (n = 16)
b Geometric mean of the abundances measured in the sediments during the monitoring survey (n = 2)
c Geometric mean of the abundances measured in the water column during the monitoring survey (n = 16)
d Potential abundance of resuspended FIB as calculated by multiplying the FIB abundance in the sediments with the maximum SM concentration observed at each

sampling site
e Percentage of FIB potentially originating from sediment resuspension in the water column, calculated as the ratio of potentially resuspended FIB (d) to the geometric

mean abundance of FIB in the water column (c)

 Environ Monit Assess

correlation was found between the flow rate and trance and the outlet of various WWTPs were
the log abundances of FIB for the other sites analysed. In raw waters, the abundances ranged
(0.08 < r2 < 0.23; p> 0.05). between 9.6 105 and 2.0 107 CFU/100 mL for
Comparison of Fig. 2a, b showed that the E. coli (geometric mean 1.0 107 CFU/100 mL)
abundances of both FIB followed very similar and between 2.3 105 and 4.1 106 CFU/100 mL
trends. Accordingly, when IE numbers were plot- for IE (geometric mean 1.7 106 CFU/100 mL;
ted against E. coli numbers in a loglog scale, a Fig. 5). In the treated effluents of the WWTPs,
significant correlation (r2 = 0.89, p < 0.01) was the abundances of both FIB were significantly
found (Fig. 3). The GLUase activity, tested here reduced. The log removal of FIB due to the
as a surrogate for E. coli plate counts, followed a different types of treatment was calculated as the
trend between sampling sites very similar to the difference between the log of the concentration
one observed for E. coli abundance (Fig. 2a, c). before and after treatment. When a primary treat-
When GLUase activities were plotted against E. ment was applied alone, average log removals
coli counts in a loglog scale, a significant correla- of 0.40 and 0.38 were estimated for E. coli and
tion (r2 = 0.87, p < 0.01) was found (Fig. 4). IE, respectively; the removal is due to settling of
FIB attached to suspended matter. When PT was
followed by an activated sludge process for carbon
Faecal contamination in the river sediments
removal, average log removals were 1.8 and 1.6 for
E. coli and IE, respectively. As shown in Fig. 5,
The geometric means of E. coli and IE concentra-
when the residence time of the water in biological
tions measured in the sediments of the main rivers
processes is increased to allow nitrification and
of the watershed during two sampling campaigns
denitrification, the efficiency of faecal pollution
are presented in Table 1. These values ranged, re-
removal is also increased. In the present study,
spectively, between 2.1 102 and 2.2 105 CFU/g
the lowest concentrations after an intensive waste-
dry weight sediment for E. coli and between 1.0
water treatment were observed when a complete
102 and 3.8 105 CFU/g dry weight sediment for
treatment (primary treatment followed by AS for
IE. The higher levels of faecal contamination were
carbon and N removal and dephosphatation) was
observed in the Zenne River downstream from
Brussels area and in the Nethe River.
In this study, comparison of the abundances
of both FIB observed in the water column and 8
in the sediments of rivers was performed. The 7
abundances of E. coli and IE in the sediments did
6
Log (CFU (100 mL-1))

not follow a similar pattern of variation between

5
sampling as observed in the water column. The
abundances of FIB in the sediments in log units 4

(geometric means for two sampling campaigns) 3

were not significantly correlated with the abun- 2
dances in the water samples in log units (geomet- 1
ric means for 16 sampling campaigns).
0
PT + AS

PT + AS +

PT + AS +
PT

Nit-Denit+ P
PT + trickling
wastewater

Stabilisation
Nit-Denit
Untreated

PT + AS +
Nit

pond

Sources of faecal contamination in the rivers

filter

Point sources of faecal indicator bacteria

Fig. 5 E. coli (black) and IE (grey) abundances (log
In order to quantify the contribution of treated units) in raw and differently treated wastewaters (see
text in Materials and methods section for the symbol
wastewater to FIB loading and to investigate the
definitions of the different wastewater treatments). Data
efficiency of different wastewater treatments on are expressed as geometric mean values, and vertical bars
the removal of FIB, samples collected at the en- represent the range between minimal and maximal values
Table 2 Budget for the whole Scheldt River watershed of the point and non-point sources of faecal contamination to the rivers
Contribution of point sources
Point sources WWTP type Number of WWTPs Volume (m3 )/year E. coli Intestinal enterococci
CFU (100 mL)1 Flux (CFU s1 ) CFU (100 mL)1 Flux (CFU s1 )
PT 6 4.0 106 4.0 106 5.0 109 6.9 105 8.8 108
PT + AS 165 2.4 108 1.6 105 1.2 1010 4.2 104 3.2 109
Environ Monit Assess

PT + AS + Nit 21 3.5 106 7.9 104 8.9 107 2.0 104 2.2 107
PT + AS + Nit-Denit 44 1.9 107 5.0 104 3.0 108 1.4 104 8.6 107
PT + AS + Nit-Denit + P 148 6.3 108 2.0 104 4.0 109 7.4 103 1.5 109
PT + trickling filter 1 2.0 106 5.4 105 3.4 108 1.2 105 7.6 107
Stabilisation pond 35 6.4 106 2.1 104 4.3 107 5.6 103 1.1 107
Total 420 9.0 108 2.2 1010 5.7 109
Contribution of non-point sources
Non-point sources Land use Surface (km2 ) E. coli Intestinal enterococci
CFU (100 mL)1 Flux (CFU s1 ) CFU (100 mL)1 Flux (CFU s1 )
Forested areas 1.5 103 4.8 101 5.4 106 3.0 101 3.4 106
Cultivated areas 8.4 103 2.2 102 1.4 108 1.9 102 1.2 108
Pastured areas 4.9 103 1.1 103 4.7 108 5.7 102 2.4 108
Total 1.5 104 6.2 108 3.6 108
 Environ Monit Assess

used (log removals of 2.7 and 2.4 for E. coli and Discussion
IE, respectively). Finally, the average efficiency of
the stabilisation ponds investigated in this study Relationship between GLUase activity
was in the same order of magnitude than for the and E. coli abundance
latter treatment (log removals of 2.67 and 2.47
for E. coli and IE, respectively). Whatever the In the present study, GLUase activity was tested
treatment applied, the waters at the outlet of the as a surrogate for a rapid evaluation of E. coli
WWTPs still contained high abundances of FIB abundance in river waters. A significant corre-
(concentrations higher than 2 104 and 5 103 lation was found (in loglog units) between the
for E. coli and IE, respectively) leading to a major enzymatic activities and the plate counts of E. coli.
impact on the receiving waters if the dilution fac- Similar correlations were reported in previous
tor is not sufficient (see Discussion section). Of studies for different types of aquatic systems: river
course, lower values in the treated effluents can waters (Farnleitner et al. 2001; Servais et al. 2005),
be obtained if a disinfection stage (UV irradiation, marine waters (Lebaron et al. 2005) and waste-
for example) is added at the end of the wastewater waters (Garcia-Armisen et al. 2005). The slope of
treatment line (George et al. 2002), but this was the regression straight line obtained in this study
not investigated in the present study as there is was lower than 1 in agreement with these previous
presently no WWTP with a disinfection unit in the studies (Garcia-Armisen et al. 2005). This indi-
Scheldt watershed. cates that the ratio of GLUase activity to cultur-
able E. coli abundance increased in less polluted
Non-point sources of faecal indicator bacteria environments. A possible explanation for this ob-
servation was suggested by George et al. (2000):
In this study, FIB abundances were measured in The higher enzymatic activities per culturable cell
samples collected in small streams located in rural in less contaminated natural waters may be due to
areas. The small streams were classified on the an underestimation of the number of FIB when
basis of the land use of their watershed: forested enumerated by culture-based methods. This un-
areas, cultivated areas and pastured areas. Table 2 derestimation may be explained by a higher pro-
presents the level of faecal contamination due portion of viable but non-culturable cells (VBNC;
to surface runoff and soil leaching for both FIB. cells presenting a detectable GLUase activity but
Data showed that small streams flowing through unable to multiply in or on the specific media used
pastures were, on average, more contaminated in culture-based methods) in low contaminated
than those flowing through cultivated areas, which waters. The higher proportion of VBNC faecal
were more contaminated than those flowing in indicator bacteria in less polluted environments
forested areas. However, a high variability was could be the result of more severe or longer envi-
observed between the minimal and the maximal ronmental stress factors such as nutrient limitation
values for each type of land use. A Wilcoxon and and enhanced solar radiation effects due to deeper
MannWhitney test performed to compare the light penetration (Servais et al. 2009). The quality
level of E. coli from different land uses showed of the correlation presented in Fig. 4 showed that
that there was a significant difference between GLUase activity measurement can be considered
pastured and cultivated areas ( p < 0.05) and also as a valid alternative for monitoring E. coli in
between pastured and forested areas ( p < 0.01). the rivers of the Scheldt watershed. This method
However, no significant difference was found be- offers the advantage to give a result in less than
tween cultivated and forested areas ( p > 0.05). 1 h while the culture-based method requires a 24-
This analysis demonstrates that the land use of the h incubation.
watershed has a significant impact on the quan-
tity of faecal microorganisms brought to rivers Relationship between E. coli and IE abundances
by surface runoff and soil leaching and thus on
the microbiological quality of the small streams E. coli and IE are now usually enumerated to eval-
flowing in rural areas. uate the microbiological quality of recreational
Environ Monit Assess

waters (EPA 1999; EU 2006). In this study, the than in wastewaters, this could partly explain the
abundances of both indicators were measured in lower ratio values in small streams waters with
parallel in river water samples (Fig. 2) and in regards to ratio values for WWTP effluents. The
sediments (Table 1). For the water samples, sig- E. coli/IE ratios observed in river waters and in
nificant correlation was found between E. coli WWTP effluents were not significantly different
and IE numbers in loglog units (r2 = 0.89, ( p > 0.05). This could suggest that the FIB in
p < 0.01; Fig. 3). Significant correlations between the main rivers were mainly from human origin
these two FIB were already reported in the liter- and brought to rivers by wastewaters. The average
ature (Kinzelman et al. 2003; Roslev et al. 2008; E. coli/IE ratio found in the main rivers of the
Kirschner et al. 2009). Additionally, in the present Scheldt watershed was quite close to the mean
study, a significant correlation was also found be- value reported by Kirschner et al. (2009) for the
tween E. coli and IE numbers in loglog units for rivers of the Danube watershed (5.8).
the sediment samples (r2 = 0.86, p < 0.01; data The average ratio of E. coli to IE was also cal-
not shown). A similar relationship was already culated for the sediment samples collected in the
reported by Roslev et al. (2008) for stream and main rivers of the watershed; the value (1.89) was
estuarine sediment samples. significantly lower than the one estimated for the
Several studies have shown that the composi- corresponding overlying water columns. A longer
tion of the faecal flora was different in animal survival of IE in the sediments in comparison to
and human faeces; for example, the proportion of E. coli (Craig et al. 2002b) could explain this ob-
faecal coliforms compared to faecal streptococci servation. Roslev et al. (2008) also reported lower
and the proportion of E. coli to IE were higher E. coli/IE ratios in sediments than in overlying
in human faeces than in animal faeces (Feachem river and estuarine samples.
1975; Geldreich 1976). Thus, some authors
(Jagals et al. 1995) have suggested that the ratio of
E. coli to IE can be used to determine the source Faecal contamination of the rivers of the Scheldt
(human or animals faeces) of faecal pollution in drainage network
aquatic systems. In this study, the E. coli/IE ra-
tios were calculated for runoff and soil leaching In order to qualify the microbiological water qual-
waters (small streams sampled upstream from any ity of the main rivers of the watershed observed
point source), main rivers and WWTP effluents. during our monitoring, we compared the mea-
The average of the E. coli/IE ratios estimated sured E. coli and IE abundances with the require-
for WWTPs effluents (5.24) was the highest, fol- ments of the new EU Directive for bathing water
lowed by the ratio for the main rivers (5.12); it quality (EU 2006). We considered the 90th per-
was much lower for the small streams (1.63). The centiles of the E. coli and IE abundances at each
Wilcoxon and MannWhitney test was performed location and compared them to the sufficient
to compare the E. coli/IE ratios observed in rivers, values set out in the Directive (i.e. the mini-
WWTP effluents and stream waters. The ratios mal quality requirement)9.0 102 and 3.3
of E. coli to IE observed in WWTP effluents 102 CFU/100 mL for E. coli and IE, respectively.
were significantly higher ( p < 0.05) than those They were higher at every location since the less
observed in the small streams. This result seems contaminated site Sc3 displayed 90th percentiles
in agreement with the origin of the faecal contam- of 3.0 103 and 1.1 103 CFU/100 mL for E. coli
ination. Soil leaching and runoff brought to small and IE, respectively (Fig. 2a, b). This indicated
streams FIB mainly from animal origin while in that none of the sampling sites had a microbiolog-
WWTP effluents FIB are predominantly from hu- ical water quality sufficient for bathing activities
man origin. However, the ratio of E. coli to IE val- and that the main rivers of the watershed have
ues can also be influenced by the residence time globally a poor microbiological water quality.
of faeces in the environment as IE survive longer The sites from the Zenne River downstream
than E. coli (Craig et al. 2002b). As we can assume from Brussels (Se2 and 3) were heavily contami-
that the residence time is longer in small streams nated: The abundances of E. coli and IE were in
 Environ Monit Assess

the same order of magnitude than those usually is particularly high in the case of the impact of
measured in treated wastewaters (Fig. 5). The Brussels treated wastewaters on the Zenne River.
Zenne River, a tributary of the Dyle River, has
a watershed of 1,011 km2 characterized by agri-
culture activities in its upstream part and an im- Impact of sediment resuspension
portant urbanization in its downstream part. The on the microbiological water quality of the rivers
population density in the watershed is very high
(on average 1,259 inhabitants km2 ) and most of Sediments have been recognised as in-stream
the inhabitants live in the city of Brussels. The store of faecal contamination that can be mo-
Zenne River crosses this city from south to north bilised during hydrological events (rapid increase
over a distance of about 20 km and receives the of the flow rate, floods) or other sediment-
sewage from two WWTPs: the Brussels South disturbing events (Muirhead et al. 2004). Sev-
WWTP (360,000 inhabitant equivalents) and the eral studies have shown that resuspension of
Brussels North WWTP (1.1 millions inhabitant faecally contaminated sediments can deteriorate
equivalents). The annual average discharge of the microbiological quality of the overlying water
the Zenne River upstream from Brussels is column (Crabill et al. 1999). During this study,
3.93 m3 s1 (average for the 20072008 period) sediments of the main rivers of the watershed
while the flow released by the two WWTPs of were shown to contain between 102 and 105 FIB/g
Brussels is of the same order of magnitude (av- dry weight. In order to determine the possible
erage for the 20072008 period3.85 m3 s1 ). contribution of sediment resuspension to the fae-
This means that, on average, the Zenne River cal contamination in the rivers of the watershed,
water downstream from Brussels is roughly half we compared the FIB abundance that could result
composed of treated wastewaters, this proportion from sediment resuspension with the FIB abun-
being higher during the low flow periods of the dance measured in the water column during the
river. In addition, during rain events, combined monitoring survey. FIB that could have originated
sewer overflows (CSOs) occur in the Brussels area from sediment resuspension according to this cal-
and release in the river a mixture of untreated culation represented in general a low proportion
wastewater and surface runoff water. of the FIB found in the water column (Table 1).
Similar degradations of river microbiological Because the ratio of E. coli to IE was lower in
quality due to the discharge of treated urban the sediment than in the water, the contribution
wastewater effluents from large cities were of sediment resuspension to the FIB load of the
already reported in the literature. For example, water was higher for IE than for E. coli. Potential
a large increase of faecal indicator bacteria resuspended E. coli represented less than 1% of
concentration was observed in the Thames River the E. coli in the water column at six sites out
downstream from London (Tryland et al. 2002) of 12 and between 1% and 10% at four sites.
where E. coli concentration can reach up to 105 For IE, these contribution levels were observed at
E. coli/100 mL. An important increase of faecal two and six sites, respectively. The contribution of
coliforms was reported by Servais et al. (2007) in sediment resuspension to the water contamination
the Seine River downstream from the Parisian was potentially high in only two sites: Sc3 (32%
area. A moderate increase in coliform counts was and 58% of the total E. coli and IE, respectively)
observed in the Danube River downstream from and Ne (52% and 110% of the total E. coli and
Vienna (Hoch et al. 1996) and large increases IE, respectively; Table 1). Therefore, except for
downstream from Budapest and Bucarest these two sites, sediment resuspension was not
(Kirschner et al. 2009). The importance of such estimated to be a major cause of water quality
quality decrease is directly related to the ratio degradation in the rivers of the Scheldt water-
of the effluents and the river flow rates and to shed, unlike what was observed in other studies in
the FIB concentration in the treated effluent which the water column was far less contaminated
depending on the type of treatment; this ratio (Crabill et al. 1999; Nagels et al. 2002).
Environ Monit Assess

Respective contribution of point and non-point At the scale of the whole watershed, point
sources to river faecal contamination sources (WWTPs effluents) appeared therefore
to contribute 35 and 15 times more than non-
In order to compare the global contributions of point sources to the flux of E. coli and IE to the
point and non-point sources to the faecal con- rivers. Moreover, discharges of untreated waste-
tamination of the Scheldt drainage network, we water were neglected in this budget. This could
estimated the fluxes of FIB emitted at the scale of result in an underestimation of the point source
the whole watershed by both types of source. For contribution since a small part of the population
point sources, we estimated the flux of FIB dis- in the watershed is connected to a sewer system
charged by the WWTPs located in the watershed. but not to a WWTP. In addition, during strong
For year 2008, 420 WWTPs were inventoried in rain events CSOs can occur, releasing in rivers
the watershed along with their treatment type and a mixture of urban runoff water and wastewater.
their annual treated volume. For each WWTP, the Due to the absence of data on discharged volumes
annual discharge was then multiplied by an aver- of untreated wastewater and CSOs, our estimate
age FIB concentration in function of its treatment of the flux of FIB emitted by point sources should
type. The geometric means of the abundances be considered as a minimum. If release of un-
measured in this study at the outlet of WWTPs treated wastewater and CSOs would be taken into
for different types of treatment (Fig. 5) were used account, this would reinforce the predominance of
for this calculation. The FIB fluxes of all WWTPs point sources over non-point sources.
were then summed to estimate the global contri- Finally, it should be kept in mind that this
bution of point sources; it represented a flux of global budget does not give any information on
2.2 1010 E. coli/s and 5.7 109 IE/s (Table 2). the local impact that non-point sources can have
For non-point sources, the surface occupied by on the microbiological quality of small rivers.
the different land uses in the watershed was first
evaluated. These surfaces were derived from the
CORINE Land Cover database (Bossard et al. Conclusions
2000): forested, pastured, cultivated and urban ar-
eas represented, respectively, 7%, 26%, 39% and The microbial pollution of the main rivers of
25% of the total watershed surface (21,863 km2 ). Scheldt basin was assessed by monitoring faecal
As almost all urban areas in the watershed are indicator bacteria (E. coli and enterococci) con-
equipped with combined sewers, runoff on urban tamination in water and sediments. This study
areas was assumed to be collected by sewers and demonstrated the poor microbiological quality
treated in WWTPs; it was thus included in our of the main rivers of the Scheldt watershed.
estimation of the FIB flux from point sources. For Sediments of studied rivers contained high lev-
the other types of surface, the flux of runoff or els of FIB; however, the FIB concentrations in
soil leaching water emitted to the watercourses at sediments were not sufficient to contribute sig-
the scale of the whole watershed was estimated by nificantly to river water contamination during re-
multiplying the specific discharge of the Scheldt suspension events excepted for the Scheldt and
watershed (7.4 L s1 km2 ) with the total surface the Nethe. The quantification of point and non-
of each land use in the watershed. These water point sources in the Scheldt watershed allowed the
fluxes were multiplied by a FIB concentration comparison of their respective contribution to the
specific to each land use (average abundances faecal contamination of the rivers of the Scheldt
measured in this study in the small streams from drainage network. This comparison showed that,
forested, pastured or cultivated areas (Table 2)). at the scale of the Scheldt watershed, point
The FIB fluxes obtained for the three types of sources were largely predominant in comparison
land use were finally summed to get the global to non-point sources. This finding indicates that
contribution of non-point sources; it was estimated to improve the microbiological water quality of
at 6.2108 E. coli/s and 3.6108 IE/s (Table 2). the rivers, the faecal contaminants released by
 Environ Monit Assess

the WWTPs should be reduced. This can be done freshwater sediments. Applied and Environmental Mi-
by adding, at the end of the treatment line in crobiology, 61(5), 18881896.
Edberg, S. C., Rice, E. W., Karlin, R. J., & Allen, M. J.
WWTPs, a treatment stage specifically devoted to
(2000). Escherichia coli: The best biological drinking
disinfection. UV treatment can be used for this water indicator for public health protection. Journal
purpose; it has been shown that UV treatment of Applied Microbiology, 88, 106S116S.
as final stage of wastewater treatment allowed EPA (1999) United States Environmental Protection
Agency, 1999. Action plan for beaches and recre-
increasing the Log removal by 2 to 3 U (Servais
ational waters. EPA 600-R-98/079. Washington, DC:
et al. 2007). EPA.
EU (2006). Directive 2006/7/EC of the European Parlia-
ment and of the Council of 15 February 2006 concern-
Acknowledgements This study was performed in the ing the management of bathing water quality. Of f icial
scope of the project Tracing and Integrated Modeling Journal of the European Union, 64, 3751.
of Natural and Anthropogenic effects on Hydrosystems: Farnleitner, A. H., Hocke, L., Beiwl, C., Kavka, G. G.,
The Scheldt River Basin and Adjacent Coastal North Zechmeister, T., Kirschner, A. K. T., et al. (2001).
Sea (TIMOTHY), an Interuniversity Attraction Pole Rapid enzymatic detection of Escherichia coli conta-
(IAP6.13) funded by the Belgian Federal Science Policy mination in polluted river water. Letters in Applied
Office. N.K Ouattara benefits from a doctoral grant from Microbiology, 33(3), 246250.
Ivory Cost Government. The authors thank the Vlaamse Feachem, R. (1975). An improved role for faecal coliform
Milieumaatschappij for providing the flow rate data. to faecal streptococci ratios in the differentiation be-
tween human and non-human pollution sources. Water
Research, 9(7), 689690.
Fewtrell, L., & Bartram, J. (2001). Water quality: Guide-
lines, standards and health. World Health Organization
References Water Series (pp. 2021). London: IWA.
Fiksdal, L., Pommepuy, M., Caprais, M. P., & Midttun, I.
An, Y. -J., Kampbell, D. H., & Briedenbach, G. P. (2002). (1994). Monitoring of fecal pollution in coastal waters
Escherichia coli and total coliforms in water and by use of rapid enzymatic techniques. Applied and
sediment at lake marinas. Environmental Pollution, Environmental Microbiology, 60(5), 15811584.
120(3), 771778. Garcia-Armisen, T., Lebaron, P., & Sevais, P. (2005).
Anderson, K. L., Whitlock, J. E., & Harwood, V. J. -D-Glucuronidase activity assay to assess viable Es-
(2005). Persistence and differential survival of fecal cherichia coli abundance in freshwaters. Letters in Ap-
indicator bacteria in subtropical waters and sediments. plied Microbiology, 40(4), 278282.
Applied and Environmental Microbiology, 71(6), Garcia-Armisen, T., & Servais, P. (2007). Respective con-
30413048. tributions of point and non point sources of E. coli
Bossard, M., Feranec, J., & Otahel, J. (2000). CORINE and enterococci in a large urbanised watershed (the
land cover technical guide: Addendum 2000. Copen- Seine River, France). Journal of Environmental Man-
hagen: European Environment Agency. agement, 82(4), 512518.
Crabill, C., Donald, R., Snelling, J., Foust, R., & Southam, Geldreich, E. E. (1976). Fecal coliform and fecal strepto-
G. (1999). The impact of sediment fecal coliform coccus density relationships in waste discharges and
reservoirs on seasonal water quality in Oak creek, receiving waters. Critical Reviews on Environmental
Arizona. Water Research, 33(9), 21632171. Control, 6, 349369.
Craig, D. L., Fallowfield, H. J., & Cromar, N. J. (2002a). George, I., Anzil, A., & Servais, P. (2004). Quantification
Enumeration of faecal coliforms from recreational of fecal coliforms inputs to aquatic systems through
coastal sites: Evaluation of techniques for the sepa- soil leaching. Water Research, 38(3), 611618.
ration of bacteria from sediments. Journal of Applied George, I., Crop, P., & Servais, P. (2002). Fecal coliforms
Microbiology, 93(4), 557565. removal in wastewater treatment plants studied by
Craig, D. L., Fallowfield, H. J., & Cromar, N. J. (2002b). plate counts and enzymatic methods. Water Research,
Comparison of the decay rates of faecal indicator or- 36(10), 26072617.
ganisms in recreational coastal water and sediment. George, I., Petit, M., & Servais, P. (2000). Use of enzymatic
Water Science and Technology: Water Supply, 2(3), methods for rapid enumeration of coliforms in fresh-
131138. waters. Journal of Applied Microbiology, 88(3), 404
Craig, D. L., Fallowfield, H. J., & Cromar, N. J. (2004). Use 413.
of microcosms to determine persistence of Escherichia Hoch, B., Berger, B., Kavka, G., & Herndl, G. (1996).
coli in recreational coastal water and sediment and Influence of wastewater treatment on the micro-
validation with in situ measurements. Journal of Ap- bial ecology of a large, temperate river systemthe
plied Microbiology, 96(5), 922930. Danube River. Hydrobiologia, 321(3), 205218.
Davies, C. M., Long, J. A. H., Donald, M., & Ashbolt, N. J. Jagals, P., Grabow, W. O. K., & de Villiers, J. C. (1995).
(1995). Survival of fecal microorganisms in marine and Evaluation of indicators for assessment of human and
Environ Monit Assess

animal faecal pollution of surface run-off. Water Sci- Prats, J., Garcia-Armisen, T., Larrea, J., & Servais, P.
ence and Technology, 31(56), 235241. (2008). Comparison of culture-based methods to enu-
Kinzelman, J., Ng, C., Jackson, E., Gradus, S., & merate Escherichia coli in tropical and temperate
Bagley, R. (2003). Enterococci as indicators of lake freshwaters. Letters in Applied Microbiology, 46(2),
Michigan recreational water quality: Comparison of 243248.
two methodologies and their impacts on public health Roslev, P., Bastholm, S., & Iversen, N. (2008). Relationship
regulatory events. Applied and Environmental Micro- between fecal indicators in sediment and recreational
biology, 69(1), 9296. waters in Danish estuary. Water Air and Soil Pollution,
Kirschner, A. K. T., Kavka, G., Velimirov, B., Mach, 194(14), 1321.
R. L., Sommer, R., & Farnleitner, A. H. (2009). Mi- Servais, P., Garcia-Armisen, T., George, I., & Billen,
crobiological water quality along the Danube River: G. (2007). Fecal bacteria in the rivers of the Seine
Integrating data from two whole-river surveys and drainage network (France): Sources, fate and mod-
a transnational monitoring network. Water Research, elling. Science of the Total Environment, 375(13),
43(15), 36733684. 152167.
Lebaron, P., Henry, A., Lepeuple, A. S., Pena, G., & Servais, P., Garcia-Armisen, T., Lepeuple, A. S., &
Servais, P. (2005). An operational method for real- Lebaron, P. (2005). An early warning method to detect
time monitoring of E. coli in bathing waters. Marine faecal contamination of river waters. Annals of Micro-
Pollution Bulletin, 50(6), 652659. biology, 55(2), 151156.
Lee, C. M., Lin, T. Y., Lin, C.-C., Kohbodi, G. A., Bhatt, Servais, P., Prats, J., Passerat, J., & Garcia-Armisen, T.
A., Lee, R., et al. (2006). Persistence of fecal indicator (2009). Abundance of culturable versus viable
bacteria in Santa Monica Bay beach sediment. Water Escherichia coli in freshwaters. Canadian Journal of
Research, 40(14), 25932602. Microbiology, 55(7), 905909.
Miranda, J. M., Franco, C. M., Vzquez, B. I., Fente, C. A., Schilling, K. E., Zang, Y. K., Hill, D. R., Jones, C. S.,
Barros-Velzquez, J., & Cepeda, A. (2005). Evalu- & Wolter, C. F. (2009). Temporal variations of Es-
ation of Chromocult enterococci agar for the isola- cherichia coli concentrations in large Midwestern
tion and selective enumeration of Enterococcus spp. in river. Journal of Hydrology, 365, 7985.
broilers. Letters in Applied Microbiology, 41(2), 153 Smith, J., Edwards, J., Hilger, H., & Steck, R. T. (2008).
156. Sediment can be a reservoir for coliform bacteria re-
Muirhead, R. W., Davies-Colley, R. J., Donnison, A. M., & leased into streams. Journal of General and Applied
Nagels, J. W. (2004). Faecal bacteria yields in artificial Microbiology, 54(3), 173179.
flood events: Quantifying in-stream stores. Water Re- Strahler, A. N. (1957). Quantitative analysis of watershed
search, 38(5), 12151224. geomorphology. Transactions of the American Geo-
Nagels, J. W., Davies-Colley, R. J., Donnison, A. M., physical Union, 38(6), 913920.
& Muirhead, R. W. (2002). Faecal contamination Tryland, I., Surman, S., & Berg, J. D. (2002). Monitoring
over flood events in a pastoral agricultural stream in faecal contamination of the Thames estuary using semi
New Zealand. Water Science and Technology, 45(12), automated early warning system. Water Science and
4552. Technology, 46(3), 2531.