Color profile: Disabled

Composite Default screen

326

Prevalence of Escherichia coli O157:H7 and

Salmonella spp. in surface waters of southern
Alberta and its relation to manure sources
J.Y.M. Johnson, J.E. Thomas, T.A. Graham, I. Townshend, J. Byrne, L.B. Selinger,
and V.P.J. Gannon

Abstract: The Oldman River watershed in southern Alberta, Canada, is an extensively irrigated region in which intensive
agricultural practices have flourished. Concern over water quality in the basin has been expressed because of high levels
of enteric disease indigenous to the region. To address these concerns, we conducted a 2-year study to estimate the
prevalence of Escherichia coli O157:H7 and Salmonella spp. in surface water within the basin. This study is the first
of its kind to identify E. coli O157:H7 repeatedly in surface water collected from a Canadian watershed. Prevalence of
E. coli O157:H7 and Salmonella spp. in water samples was 0.9% (n = 1483) and 6.2% (n = 1429), respectively. While
data examined at a regional level show a relationship between high livestock density and high pathogen levels in southern
Alberta, statistical analysis of point source data indicates that predicted manure output from bovine, swine, and poultry
feeding operations was not directly associated with either Salmonella spp. or E. coli O157:H7 prevalence. However,
geography and weather variables, which are likely to influence bacterial runoff, were not considered in this model. We
also postulate that variations in time, amount, and frequency of manure application onto agricultural lands may have
influenced levels of surface-water contamination with these bacterial pathogens.
Key words: water, Salmonella, E. coli O157:H7, manure output, agriculture.
Rsum : Le bassin versant de la rivire Oldman, dans le sud de lAlberta, Canada, est une rgion fortement irrigue
o des pratiques dagriculture intensive se sont dveloppes. Des inquitude quant la qualit de leau dans le bassin
ont t mises en raison des taux levs de maladies entriques qui svissent dans la rgion. Pour rpondre ces inquitudes,
une tude dune dure de deux ans a t accomplie afin dvaluer la prvalence de Escherichia coli O157:H7 et de
Salmonella spp. dans les eaux de surface du bassin. Cette tude est la premire de ce type identifier du E. coli
O157:H7 rptition dans de leau de surface dun bassin versant canadien. La prvalence de E. coli O157:H7 et de
Salmonella spp. dans les chantillons deau tait de 0,9 % (n = 1483) et de 6,2 % (n = 1429), respectivement. Bien
que les donnes au niveau rgional qui ont t examines rvlent un rapport entre la densit leve du btail et les niveaux
levs de pathognes dans le sud de lAlberta, lanalyse statistique des donnes des sources ponctuelles indique que la
production projete de fumier provenant de lalimentation dlevages bovins, porcins ou de volaille ntait pas directe-
ment associe la prvalence de Salmonella spp. ou de E. coli O157:H7. Toutefois, les variables gographiques et cli-
matiques, qui pourraient influencer le ruissellement de bactries, nont pas t retenues dans ce modle. Nous postulons
galement que les variations de temps, de quantit et de frquence dpandage de fumier sur les terres agricoles ont pu avoir
un effet sur la contamination de leau de surface par ces bactries pathognes.

Mots cls : eau, Salmonella, E. coli O157:H7, production de fumier, agriculture.

[Traduit par la Rdaction] Johnson et al. 335

Introduction waters used by rural and urban centres. Approximately 10%

of the Oldman River watershed is irrigated agricultural land.
Southern Alberta is a semi-arid region in which the Irrigation waters are used for the production of feed needed
Oldman River drainage basin is the principal source of water to maintain intensive livestock operations, for livestock wa-
for agriculture, industry, and both drinking and recreational tering, as well as for recreation and residential purposes.

Received 23 January 2003. Revision received 9 June 2003. Accepted 10 June 2003. Published on the NRC Research Press Web site
at http://cjm.nrc.ca on 10 July 2003.
J.Y.M. Johnson,1 J.E. Thomas,2 and L.B. Selinger. Department of Biological Sciences, University of Lethbridge, Lethbridge, AB
T1K 3M4, Canada.
T.A. Graham and V.P.J. Gannon. Health Canada, P.O. Box 640, Lethbridge, AB T1J 3Z4, Canada.
I. Townshend and J. Byrne. Department of Geography, University of Lethbridge, Lethbridge, AB T1K 3M4, Canada.
1
Present address: Department of Public Health Sciences, University of Alberta, Edmonton, AB T6G 2G3, Canada.
2
Corresponding author (e-mail: thomas@uleth.ca).

Can. J. Microbiol. 49: 326335 (2003) doi: 10.1139/W03-046 2003 NRC Canada

I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:01 AM
Color profile: Disabled
Composite Default screen
Johnson et al.
Since 1991, the livestock industry in Alberta has ex-
panded significantly. Between 1991 and 2001, the number of
cattle and calves in Alberta increased from approximately
4.7 to 6.6 million (Alberta Agriculture 1991; Alberta Agri-
culture Food and Rural Development 1996; Statistics Can-
ada Online 2002). From 1996 to 2001, the total number of
pigs on Alberta farms grew from 1.7 million to more than
2.0 million and the total number of poultry from 10.6 to
13.3 million (Statistics Canada Online 2002). Much of this
activity has been centred in southern Alberta, a region that
also has one of the highest levels of enteric disease in the
province (Khakhria et al. 1996).
Waterborne outbreaks of enteric disease have resulted either
when public drinking water supplies were not adequately
treated after contamination with surface water (Barwick et
al. 2000; Jones and Roworth 1996; Kondro 2000; Swerdlow
et al. 1992) or when surface waters contaminated with en-
teric pathogens have been used for recreational purposes
(Keene et al. 1994; Kramer et al. 1996). While protozoa,
such as Giardia and Cryptosporidium (Roach et al. 1993;
Wallis et al. 1996), have been the focus of much concern in
water contamination, enteric bacteria, such as Escherichia
coli O157:H7 and Salmonella spp., which are leading causes
of gastroenteritis in Canada and the United States (Baxter
and Morton 1987; Kramer et al. 1996; Levy et al. 1998), can
also be a significant source of water contamination. From
1993 to 1998, water contaminated by E. coli O157:H7
resulted in 454 cases of illness in the United States, while
water contaminated with Salmonella species, such as Salmonella
typhimurium, infected 625 individuals and caused seven deaths.
Regions with high cattle density are prone to enteric dis-
ease (Johnson et al. 1999). Michel et al. (1999) performed
epidemiological studies that showed the incidence of human
infection with verotoxin-producing E. coli (VTEC) in On-
tario between 1990 and 1995 was related to cattle density. In
Canada, Alberta has the greatest population of cattle and one
of the highest levels of gastroenteritis resulting from infec-
tion by Salmonella spp. and E. coli O157:H7 (Khakhria et
al. 1996).
In this study, we examined the prevalence of E. coli O157:H7
and Salmonella spp. in surface waters within the Oldman
River basin in southern Alberta and explored relationships
between the prevalence of these pathogens and both animal
agriculture and human population within the region.
Materials and methods
Study sites
The Oldman River basin is a semi-arid zone located in
southern Alberta, Canada. Total precipitation in Lethbridge,
the principal urban centre (1999 population 68 712), was
231.6 mm from May to October 1999 and 130.5 mm from
May to October 2000 (information from the Agriculture and
Agri-Food Canada Lethbridge Research Centre). Water sam-
ples were collected from surface-water drainage from irri-
gated agricultural regions, the Little Bow River, the Oldman
River, sewage treatment plants, and the City of Lethbridge
storm drains. Sampling sites were located east from the
Rocky Mountains, along the Oldman River; north, along the
Little Bow River and at inflow and outflow locations of the
Lethbridge Northern Irrigation District; north of the Oldman
I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:02 AM
327
River; and west of the Little Bow River (Figs. 1, 2, and 3),
so that water samples were representative of the region. Al-
though livestock feeding facilities were situated throughout
the region with a nonhomogenous distribution, sampling
sites were selected to reflect surface-water flow and not dis-
tribution of livestock density. In 1999, 522 samples were
collected from 69 sites. During the following year, 84 sites
were sampled and 961 samples were collected, including 38
samples from the City of Lethbridge storm drain system and
six samples from six municipal sewage treatment plants
within the Oldman River basin. Twenty-three sites were
sampled regularly in both years. Sampling sites were se-
lected to assess surface water at locations of interest
(nonrandomized sampling).
Sample collection
Samples were taken either upstream from foothold or
from water at approximately 30 cm below the surface, using
a plastic container. Samples were collected in sterile, plastic
bottles; placed on ice; and transported to the laboratory for
testing.
Detection of E. coli O157:H7
Water samples were dispensed in 90-mL volumes into
clean, sterile culture bottles. To each of these bottles, we
added a 10-mL volume of 10 buffered peptone water
(BPW) (Becton Dickinson, Sparks, Md., U.S.A.) and incu-
bated the bottles at 37C for approximately 6 h. At this
point, 1 mL was withdrawn, added to 9 mL of trypticase soy
broth supplemented with 20 g/mL novobiocin (Difco, De-
troit, Mich., U.S.A.), and held at 42C for 6 2 h.
Immuno-magnetic separation using magnetic beads coated
with E. coli O157 antibody was carried out, as described by
the manufacturer (Dynal, Oslo, Norway). After the last wash
step, the anti-E. coli O157 magnetic beads were resuspended
in 50 L of wash buffer and streaked onto sorbitol MacConkey
agar containing cefixime (ctSMAC) (0.05 mg/L) and tellurite
(2.5 mg/L) (Sanderson et al. 1995). The ctSMAC plates
were incubated for 24 h at 42C and translucent colonies
tested by slide agglutination, using E. coli O157-specific an-
tibody (Difco). Positive bacteria were inoculated into 5 mL
brain heart infusion broth (Difco) and grown at 37C over-
night. Each of these bacterial cultures was tested by PCR us-
ing oligonucleotide primers that detect vt1, vt2, fliC, and
eaeAO157 genes (Gannon et al. 1997). Putative E. coli O157:H7
cultures were sent to the Health Canada Verocytotoxigenic
E. coli Reference Laboratory in Guelph, Ontario, for inde-
pendent confirmation.
Detection of Salmonella spp.
Water samples cultured in BPW, from the pre-enrichment
medium described above, were incubated for 1624 h at
37C. The pre-enriched culture was inoculated into two sep-
arate enrichment broths, Rappaport Vassiliadis (RV) broth
and tetrathionate brilliant green (TTBG) broth. For the RV
broth, we inoculated 9.9 mL of the broth with 100 L of the
BPW culture. For the TTBG broth, we inoculated a 9.0-mL
volume with 1.0 mL of the BPW culture. Both of the inocu-
lated enrichment broths were incubated at 42C for 24 h.
Following enrichment, both the RV and the TTBG broth cul-
tures were plated onto two different selective culture plates,
2003 NRC Canada
Color profile: Disabled
Composite Default screen
328
Fig. 1. Prevalence of Escherichia coli O157:H7 in surface water and the animal manure output (AMU) in southern Alberta from 1999
to 2000; Universal Transverse Mercator (UTM) coordinates Zone 11, Datum, North American Datum (NAD) 27.
modified semi-solid Rappaport Vassiliadis (MSRV) agar
(DeSmedt and Bolderdijk 1987) and brilliant green sulpha
(BGS) agar (Becton Dickinson, Mississauga, Ont., Canada).
For inoculation onto MSRV plates, a 20-L aliquot of each
culture was dispensed onto the outside edge of each individ-
ual plate. The plates were incubated at 42C for 24 h. Motile
bacteria producing halos greater than 20 mm in diameter in
the MSRV medium were subcultured onto MacConkey
(Difco) agar by taking an inoculation loop of cells from the
leading edge of the halo and streaking onto MacConkey
plates. Following inoculation, the MacConkey plates were
then incubated at 37C for 1624 h. For testing on the BGS
agar plates, both the RV and TTBG broths were gently
mixed, and an inoculation loop of each of the cultures was
plated onto BGS agar plates. The BGS agar plates were then
incubated at 37C for 1624 h. Following incubation, three
red colonies from each of the BGS plates were plated onto
MacConkey agar plates. These MacConkey plates were then
grown at 37C for 1624 h. For both the MSRV and BGS
agar plating methods, suspect translucent colonies were
picked from each of the MacConkey plates and tested for
agglutination, using Salmonella O antiserum Poly A-I & Vi
I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:02 AM
Can. J. Microbiol. Vol. 49, 2003
(Difco). Putative Salmonella-positive colonies were further
tested using standard (Christensens) urea agar, triple sugar
iron agar, lysine iron agar slants (DAoust and Purvis 1998),
and sulphide indole motility (Difco) agar (Carter 1979).
Those colonies exhibiting biochemical responses typical of
Salmonella spp. were further tested using a PCR specific for
the detection of the Salmonella invA gene (Chiu and Ou
1996). Salmonella culture isolates were confirmed at the
Health Canada Salmonella Reference Laboratory in Guelph,
Ontario.
Prevalence
The prevalence of both E. coli O157:H7 and Salmonella
spp. in water samples was calculated as
[1] (No. samples from which pathogens were detected
total no. samples tested) 100
Mapping
Prairie Farm Rehabilitation Administration (PFRA),
Lethbridge, provided data of mapped manure output from
confined animal agriculture operations in southern Alberta.
2003 NRC Canada
Color profile: Disabled
Composite Default screen
Johnson et al.
Fig. 2. Prevalence of Salmonella spp. in surface water and the animal manure output (AMU) in southern Alberta from 1999 to 2000;
Universal Transverse Mercator (UTM) coordinates Zone 11, Datum, North American Datum (NAD) 27.
The numbers of animals for each confined livestock feeding
operation in the region were estimated from information
provided in field surveys, air surveillance photos, or expan-
sion permits. Animal manure units (AMU) were estimated
as the product of the number of animals by type for each lo-
cation, the annual excreted nitrogen mass for animal types
(Natural Resources Conservation Service 1999), and the in-
verse of the mass of nitrogen required to fertilize 1 acre of
corn for 1 year in Alberta, 73 kg/year (PFRA). To obtain
AMU estimates that correspond to water sampling sites, a
150-m2 grid of AMU values over the study area was interpo-
lated from the AMU point data for confined livestock opera-
tions (ArcView 3.0, ESRI, Redland, Fla., U.S.A.).
Interpolated values for each cell of the surface were calcu-
lated as the sum of the products of the AMU operations
within a 15-km radius and the inverse of the distance to
those operations squared. In this study, we assume that
transport of manure beyond a 15-km radius of an animal
rearing facility would be prohibitively expensive and, conse-
quently, unlikely. AMU surfaces were generated for bovine,
poultry, swine, and combined animal feeding operations.
Bovine operations included confined beef and dairy farms;
I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:03 AM
329
swine operations included farrow-to-finish, weaning, and
feeder hog farms; poultry operations included broiler, layer,
turkey, and unknown poultry farms. Sampling sites from
which more than two samples were collected and for which
there were bovine, swine, and poultry operations within
15 km were used for AMUpathogen prevalence analysis.
AMU estimates at these sampling sites were calculated as
the mean of triplicate measurements. To classify sampling
sites outside the City of Lethbridge limits and provincial
park boundaries by livestock agriculture activity, we made a
5-km2 grid of mean AMU density over the study area. The
AMU density of all grid cells were divided into four equal
intervals and classified as low, medium, high, and very high
AMU density.
Statistical methods
The apparent prevalence of E. coli O157:H7 and Salmo-
nella spp. associated with sampling sites was estimated as
the percentage of pathogen-positive samples of all samples
tested for both pathogens. The change in prevalence among
sampling sites tested in both years was assessed using a
paired sample t test. The association between the prevalence
2003 NRC Canada
Color profile: Disabled
Composite Default screen

330 Can. J. Microbiol. Vol. 49, 2003

Fig. 3. Prevalence of Salmonella spp. and Escherichia coli O157:H7 in surface water in southern Alberta and bovine manure output
(AMU) surface estimated using inverse distance weighting from confined beef and dairy feeding operations; Universal Transverse
Mercator (UTM) coordinates Zone 11, Datum, North American Datum (NAD) 27.

of Salmonella spp. and E. coli O157:H7 and the AMU of
sampling sites was assessed using generalized linear models,
where inputs included AMU output from confined bovine,
swine, and poultry livestock operations. The prevalence of
Salmonella spp. among sampling sites followed a Poisson
distribution; thus, a Poisson regression model was used
(SAS, SAS Institute Inc., Cary, N.C., U.S.A.). The model
was
[2] ln(SALM) ln(SALMTEST) = + 1 BOVAMU
+ 2 SWNAMU + 3 PLTAMU
where SALM is the number of samples in which Salmonella
spp. was detected from a given sampling site; SALMTEST
is the number of samples tested from that site; is the inter-
cept; BOVAMU, SWNAMU, and PLTAMU are independent
variables; and 1, 2, and 3 are slopes (parameter estimates
for bovine, swine, and poultry AMU, respectively). The
prevalence of E. coli O157:H7 followed a binomial distribu-
tion, thus, the following logit regression model was used:
[3] logit(ECOLI) logit(ECOLITEST)
= + 1 BOVAMU + 2 SWNAMU
+ 3 PLTAMU
where ECOLI is the number of samples from a site positive
for E. coli O157:H7, ECOLITEST is the number of samples
tested from that site, and other variables are as defined pre-
viously. Type III likelihood ratio statistics were used to as-
sess the strength of association for each animal type, and
residual deviance statistics were used to assess good-
ness-of-fit. Logistic regression models were developed to de-
scribe the relationship between pathogen presence in water
samples and AMU associated with the site from which the
samples were collected, assuming the AMU at sites re-
mained stable over the period of time in which samples were
collected (SPSS 10.0.5, SPSS Inc., Chicago, Ill., U.S.A.).
Model fit was assessed using the Hosmer and Lemeshow
goodness-of-fit test.

2003 NRC Canada

I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:03 AM
Color profile: Disabled
Composite Default screen

Johnson et al. 331

Table 1. Percentage bacterial prevalence standard error (SE) at 10 different sampling

sites within the Oldman River Basin in 1999 and 2000, containing zero, low, medium, and
high animal manure units (AMU); in storm drains; and in urban settings.
Escherichia coli
Salmonella spp. O157:H7
Site Site classificationa n % SE % SE
1 Zero AMU density 21 19.0 8.8 4.8 4.8
2 Zero AMU density 24 13.0 6.9 8.3 5.8
3 Zero AMU density 17 18.0 9.5 0
4 Low AMU density 25 20.0 8.2 0
5 Medium AMU density 26 7.7 5.3 3.9 3.9
6 Medium AMU density 27 15.0 7.0 3.7 3.7
7 High AMU density 6 33.0 21.0 0
8 SD 4 25.0 25.0 0
9 SD 5 80.0 20.0 0
10 CL 22 4.6 4.6 9.1 6.3
Note: SD, storm drain samples; CL, sites within the City of Lethbridge that were not storm drain
samples.
a
AMU density zero, no confined livestock operations within a 5-km grid; low, mean AMU density
is between 17 and 166 within a 5-km grid; medium, AMU density between 166 and 315 within a 5-km
grid; and high, AMU density between 315 and 464 within a 5-km grid.

Results (Table 1) or (ii) downstream of high-density livestock opera-

Prevalence of Salmonella spp. and E. coli O157:H7
In 1999, water was collected every 2 weeks throughout
the Oldman River basin from May through October. Salmo-
nella spp. were isolated from 14 of 468 water samples, for a
prevalence of 3.0%. In 2000, water was collected every
week from May through September. Salmonella spp. were
isolated from 74 of 961 samples, for a prevalence of 7.7%.
This represents a 2.6-fold increase in the prevalence of Sal-
monella spp. within the Oldman River basin from 1999 to
2000. During this period, the basin received less than aver-
age rainfall for the region and was suffering from drought
conditions. Consequently, because of lack of water, not all
sites could be sampled consistently both in 1999 and in
2000; therefore, only bacterial prevalence at sites that were
consistently sampled in both 1999 and 2000 was compared.
No significant annual differences were observed in the prev-
alence of Salmonella spp. (t = 1.15; one-tailed p = 0.1319)
at these sites (data not shown). On average, for 1999 and
2000, Salmonella spp. were detected in 88 of 1429 samples,
with a prevalence ( SE) of 6.2% 0.6%. A high proportion
(26.3%) of the 38 samples from municipal storm drain sites
contained Salmonella spp., and one of these had the greatest
Salmonella spp. prevalence (80.0% 20.0%; Table 1) ob-
served in this study.
In 1999 and 2000, E. coli O157:H7 was isolated from 13
of 1483 water samples for a prevalence (mean SE) of
0.9% 0.2%. As with the Salmonella spp. data, no signifi-
cant annual differences existed in the prevalence of E. coli
O157:H7 (t = 1.451; one-tailed p = 0.0811) at sites that
were consistently sampled both years (data not shown). On
more than one occasion, E. coli O157:H7 was detected at an
irrigation outflow within the City of Lethbridge (Table 1).
However, most sites contaminated with E. coli O157:H7
were located near some form of agricultural activity (Fig. 1).
Further examination of these sites showed that they were lo-
cated (i) in areas with low to moderate livestock density
tions. In contrast, water obtained from regions with high
AMU were, generally, not contaminated with E. coli
O157:H7, although prevalence of Salmonella spp. was high
in some cases (Table 1); animals in these regions were nor-
mally confined and did not have direct access to surface wa-
ter. No samples positive for E. coli O157:H7 were found
within urban storm drain sites.
Manure production
Mean AMU values at sampling sites were 1655, 240, and
394 for bovine, swine, and poultry operations, respectively.
Figures 1, 2, and 3, respectively, are maps detailing
E. coli O157:H7 prevalence, Salmonella spp. prevalence,
and interpolated surface of AMU for bovine operations, us-
ing inverse distance weighting. Of 84 sampling sites, 10
were not within 15 km of a confined feeding operation and
had no interpolated AMU value; i.e., sites without bovine,
swine, or poultry AMU values were not included in the
AMUpathogen prevalence analysis, while the remaining 74
sites were included. Locations of high bacterial prevalence
did not appear to be situated near large manure output
sources (Figs. 1 and 2). Levels of aggregated manure output
from local bovine (Fig. 3), swine, and poultry feeding opera-
tions were not high at sampling sites that had a high bacte-
rial prevalence (interpolated swine and poultry AMU maps
not shown).
Consequently, no direct correlation between bacterial
prevalence and manure production from confined livestock
feeding operations was identified. Bovine, poultry, and
swine manure output associated with sampling sites was not
a significant predictor of Salmonella spp. or E. coli O157:H7
prevalence at sampling sites (Table 2) or among individual
samples (Table 3).
Discussion
This study is the first of its kind to isolate E. coli O157:H7

2003 NRC Canada

I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:03 AM
Color profile: Disabled
Composite Default screen

332 Can. J. Microbiol. Vol. 49, 2003

Table 2. Associations between bacterial prevalence in surface water from sampling sites in
southern Alberta and manure output.
Type III likelihood
ratio test
Model Variable Estimate SE 2 p Error deviance/df
SALM Intercept 2.7698 0.5373 1.55
BOVAMU 0.0001 0.0001 0.2491 0.6177
SWNAMU 0.0005 0.0011 0.2267 0.6340
PLTAMU 0.0001 0.0008 0.0064 0.9365
ECOLI Intercept 4.8072 1.7548 0.6412
BOVAMU 0.0002 0.0003 0.2126 0.6448
SWNAMU 0.0036 0.0048 0.7043 0.4013
PLTAMU 0.0014 0.0022 0.4125 0.5207
Note: Poisson and binomial distribution models of Salmonella spp. and Escherichia coli O157:H7
prevalence, respectively, as functions of inverse distance weighting of regional manure output from
confined livestock feeding operations. SALM and ECOLI are the models of prevalence of Salmonella
spp. and E. coli O157:H7, respectively. BOVAMU, PLTAMU, SWNAMU are independent variables. is
the slope or parameter estimate for bovine, poultry, and swine animal manure units, respectively. df,
degrees of freedom; SE, standard error.

Table 3. Associations between bacterial presence and manure output (animal manure units) associated with surface-water samples of
southern Alberta.
Model likelihood Hosmer and Lemeshow
ratio test goodness-of-fit test
Bacteria Manure type B Wald p (df = 1) 2 p (df = 3) 2 p (df = 8)
Salmonella spp. Bovine 0.000 0.270 0.603 0.724 0.868 29.589 <0.001
Swine 0.001 0.231 0.630
Poultry 0.000 0.007 0.934
Escherichia coli O157:H7 Bovine 0.000 0.008 0.928 2.358 0.501 8.085 0.425
Swine 0.005 1.333 0.248
Poultry 0.000 0.002 0.922
Salmonella spp. All types 0.000 0.169 0.681 0.172 0.678 14.165 0.078
Escherichia coli O157:H7 All types 0.000 0.627 0.428 0.581 0.446 17.030 0.030
Note: df, degrees of freedom; B, logistic regression coefficient; Wald, statistic to test significance of B.

in surface waters collected from a Canadian watershed. Sys-
tematic sampling of surface water within the Oldman River
basin in southern Alberta shows that it is often contaminated
with E. coli O157:H7 and Salmonella spp.
The presence of E. coli O157:H7 and Salmonella spp. in
water presents a risk to human health when the untreated wa-
ter is consumed directly or indirectly. Our results identify
E. coli O157:H7 and Salmonella spp. contamination in water
from surface drainage channels, some of which flows directly
into the Oldman and Little Bow Rivers and from storm drains
and sewage treatment plants within the Oldman River basin.
While much of the public concern over these pathogens is re-
lated to the consumption of contaminated food, concern has
also been expressed over the contamination of water used for
both drinking and recreational activities (Arvanitidou et al.
1997; Polo et al. 1998; Sargeant et al. 2000).
Michel et al. (1999) have shown that the risk of human in-
fection by VTEC can be related to cattle density. In Canada,
Alberta has the greatest population of cattle and one of the
highest levels of gastroenteritis resulting from infection by
Salmonella spp. and E. coli O157:H7 (Khakhria et al. 1996).
Between 1991 and 1994, up to 77% of 205 children less
than 15 years of age from Alberta that were diagnosed with
hemolytic uremic syndrome (HUS) were likely infected with
E. coli O157:H7 (Rowe et al. 1998). From 1987 to 1991, the
mean annual rate of VTEC infection for the province was
12.1 cases per 100 000 compared with 2.0 per 100 000 in
Scotland during the same period (Waters et al. 1994). Some
of these cases might be the result of water contamination. In
our study, we repeatedly isolated E. coli O157:H7 from sur-
face waters within the Oldman River basin. Individuals who
consumed this untreated water or used it for recreational
purposes would be expected to be at risk of infection (Pay-
ment et al. 1997; OConnor 2002).
Khaleel et al. (1980) have suggested that a high density of
intensive animal rearing facilities could facilitate the con-
tamination of surface water, resulting in large reservoirs for
pathogenic bacteria. In 1992 in Swaziland, E. coli O157:H7
caused one of the largest outbreaks of bloody diarrhoea in
the world (Effler et al. 2001). At the end of a 3-month
drought, more than 40 000 people distributed over several
hundred kilometres downstream of an agricultural district
were reported ill after runoff from heavy rains contaminated
food and water.

2003 NRC Canada

I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:03 AM
Color profile: Disabled
Composite Default screen
Johnson et al.
Michel et al. (1999) have demonstrated an association be-
tween human VTEC infections and cattle density in Ontario,
Canada. They suggested that regional manure spreading and
water contamination might link animal density, a group-level
risk factor, and human infection, an individual-level outcome
(Valcour et al. 2002). Livestock manure is a reservoir for
both Salmonella spp. and E. coli O157:H7 and is an impor-
tant source of surface-water contamination. Cattle are recog-
nized as reservoirs for E. coli O157:H7 (Dean-Nystrom et al.
1999), and cattle manure has been implicated as the source
of contamination in the Walkerton outbreak that occurred in
Ontario, Canada (Kondro 2000).
The Lethbridge region of southern Alberta is noted for
high cattle density (Fig. 3) as well as for one of the highest
levels of gastroenteritis in Canada, resulting from infection
by Salmonella spp. and E. coli O157:H7 (Khakhria et al.
1996). This observation is similar to that of Michel et al.
(1999), whose work was done using geographical scaling at
the census consolidated subdivision level. In our study, both
Salmonella spp. and E. coli O157:H7 were identified in sur-
face-water samples throughout the Lethbridge region, and it
is likely that cattle contributed to this contamination. How-
ever, when data from our study were examined using a
smaller geographical scale capable of identifying potential
point sources, we found that surface water obtained close to
large-scale animal feeding operations was not necessarily
more contaminated with zoonotic bacteria than were sam-
ples obtained far away from these operations; i.e., pathogen
contamination of surface water, not including groundwater,
was not necessarily related to animal density when potential
point sources of contamination were considered.
We note that producers in southern Alberta have been
working to implement best management practices that mini-
mize the risk of water contamination within the region. Ani-
mals often are confined and do not have direct access to
surface water. Our data suggest that these practices may
have a positive effect on water quality; i.e., no direct associ-
ation of concentrated livestock facilities with pathogen prev-
alence was detected in this study, suggesting that good
agricultural management practices can mitigate surface-water
contamination within agricultural regions. These observa-
tions, however, do not preclude that the spreading of manure
without sufficient treatment (e.g., composting) on fields may
have contributed to the water contamination observed. Cows
on pasture, wild deer, reptiles, and birds are reservoirs of
pathogens, such as Salmonella spp. and E. coli O157:H7,
and may also have contributed to water contamination
(Bradley et al. 2001; Cizek et al. 1994; Fischer et al. 2001;
Sargeant et al. 2000).
In our study, Salmonella spp. were more prevalent than
E. coli O157:H7 in water samples. Because Salmonella
spp. are present at a low level in beef and dairy herds from
western Canada (Sorenson 2000; Van Donkersgoed et al.
1999), it is likely that other sources also contributed to the
water contamination that we observed. Van Donsel and
Geldreich (1971) have identified pigs as reservoirs for Sal-
monella spp. and note that they may contribute to sur-
face-water contamination when runoff occurs. Domestic and
wild birds may also have been the source of some of the
bacterial isolates obtained, e.g., 42% of Salmonella spp. sub-
mitted to the Health Canada Salmonella Reference Labora-
I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:04 AM
333
tory in 1999 were isolated from poultry (Health Canada
2000).
In our study, a high number of the City of Lethbridge
storm drains were contaminated with Salmonella spp.
(Table 1). We note that much of the industry and a part of
the population within the City of Lethbridge are engaged in
agricultural or related activities. It is not clear what contribu-
tion agriculture surrounding the city made to the observed
contamination. Further studies are needed to identify sources
of contamination and to determine how city management
practices within agricultural regions might differ from those
in large urban centres.
The regional prevalence of E. coli O157:H7 was similar in
our study to that reported by McGowan et al. (1989) in their
study of surface water; however, the numbers of Salmonella
spp. we detected were similar to that seen in some studies
but were lower than that seen in others (Anonymous 1992;
Efstratiou et al. 1998; Polo et al. 1998). We note that the
prevalence of E. coli O157:H7 and Salmonella spp. through-
out the Oldman River basin was not uniform (Figs. 1, 2, and
3). While this variation in prevalence may, in part, be the re-
sult of various land management practices, other factors,
such as time of sampling, geography, and weather, could
also influence these findings and other factors in any model
developed.
We attempted to model the effect of local AMU on water
contamination. However, the predictive strength of our
model was limited. Incorporation of other variables, such as
bacterial transport and survival, into the model might im-
prove it. The velocity and direction of surface-water drain-
age, the irrigation water flow, and the surface wind at time
of sample collection were not acknowledged within the
model but could strengthen it by providing a link between
the effect of local confined livestock operations and local
water contamination (these data were not available for this
study). Environmental conditions at time of sampling, such
as water temperature, pH, and nutrient concentration, could
also be used to assess the potential for bacterial survival in
water. These variables could influence bacterial concentra-
tion in water and may have influenced the ability of our
model to identify an association between local manure load
and water contamination. In this study, we also assumed that
transport of manure beyond a 15-km radius of an animal
rearing facility would be prohibitively expensive and conse-
quently, unlikely; this however, may not be the case.
Acknowledgements
This study was funded by Health Canada, the University
of Lethbridge, and the Alberta Agriculture Research Insti-
tute. Water sample collection was by staff of Alberta Envi-
ronment (Water Management Division), Lethbridge Northern
Irrigation District, Alberta Agriculture (Irrigation Branch),
Chinook Health Region, and the City of Lethbridge. R.
Hardin, Prairie Farm Rehabilitation Administration, pro-
vided the animal manure unit data. A. Senthilselvan, Univer-
sity of Alberta, assisted in the statistical modelling. We also
thank the Health Canada Salmonella Reference Laboratory
(Guelph, Ont.) and the CFIA Lethbridge Laboratory (Ani-
mal Diseases Research Institute) for their support of the pro-
2003 NRC Canada
Color profile: Disabled
Composite Default screen
334
ject. Members of the research team are part of the Canadian
Water Network.
References
Alberta Agriculture. 1991. 1991 census of agriculture for Alberta.
Alberta Agriculture, Edmonton, Alberta.
Alberta Agriculture Food and Rural Development. 1996. 1996 cen-
sus of agriculture for Alberta. Alberta Agriculture, Food and
Rural Development, Edmonton, Alberta.
Anonymous. 1992. Hospital outbreak of Salmonella enteriditis in-
fection Ontario. Can. Commun. Dis. Rep. 18(8): 5760.
Arvanitidou, M., Papa, A., Constantinidis, T.C., Danielides, V., and
Katsouyannopoulos, V. 1997. The occurrence of Listeria spp.,
and Salmonella spp. in surface waters. Microbiol. Res. 152(4):
395397.
Barwick, R.S., Levy, D.A., Craun, G.F., Beach, M.J., and
Calderon, R.L. 2000. Surveillance for waterborne-disease out-
breaks United States, 19971998. MMWR, 49(4): 121.
Baxter, D.N., and Morton, S. 1987. An outbreak of S. indiana in a
maternity unit: implications for control policies. Community
Med. 9(4): 365371.
Bradley, T., Angulo, F., and Mitchell, M. 2001. Public health edu-
cation on Salmonella spp. and reptiles. J. Am. Vet. Med. Assoc.
219(6): 754755.
Carter, G.R. 1979. Diagnostic procedures in veterinary bacteriol-
ogy and mycology. Charles C. Thomas Publisher, Springfield,
Ill.
Chiu, C.H., and Ou, J.T. 1996. Rapid identification of Salmonella
serovars in feces by specific detection of virulence genes, invA
and spvC, by an enrichment broth culturemultiplex PCR com-
bination assay. J. Clin. Microbiol. 34(10): 26192622.
Cizek, A., Literak, I., Hejlicek, K., Treml, F., and Smola, J. 1994.
Salmonella contamination of the environment and its incidence
in wild birds. Zentralbl. Veterinaermed. Reihe B. 41(5): 320327.
DAoust, J.Y., and Purvis, U. 1998. Isolation and identification of
Salmonella from foods. Health Protection Branch Method
MFHPB-20. Health Protection Branch.
Dean-Nystrom, E.A., Bosworth, B.T., OBrien, A.D., and Moon,
H.W. 1999. Bovine infection with Escherichia coli O157:H7. In
Escherichia coli O157:H7 in farm animals. Edited by C.S. Stew-
art and H.J. Flint. CABI Publishing, New York. p. 5158.
DeSmedt, J.M., and Bolderdijk, R.F. 1987. Dynamics of Salmo-
nella isolation with modified semi-solid Rappaport Vassiliadis
medium. J. Food Prot. 50: 658661.
Effler, E., Isaacson, M., Arntzen, L., Heenan, R., Canter, P.,
Barrett, T., Lee, L., Mambo, C., Levine, W., Zaidi, A., and Grif-
fin, P.M. 2001. Factors contributing to the emergence of Esche-
richia coli O157 in Africa. Emerg. Infect. Dis. 7(5): 812819.
Efstratiou, M.A., Mavridou, A., Richardson, S.C., and Papadakis,
J.A. 1998. Correlation of bacterial indicator organisms with Sal-
monella spp., Staphylococcus aureus and Candida albicans in
sea water. Lett. Appl. Microbiol. 26(5): 342346.
Fischer, J.R., Zhao, T., Doyle, M.P., Goldberg, M.R., Brown, C.A.,
Sewell, C.T., Kavanaugh, D.M., and Bauman, C.D. 2001. Exper-
imental and field studies of Escherichia coli O157:H7 in
white-tailed deer. Appl. Environ. Microbiol. 67(3): 12181224.
Gannon, V.P., DSouza, S., Graham, T., King, R.K., Rahn, K., and
Read, S. 1997. Use of the flagellar H7 gene as a target in multi-
plex PCR assays and improved specificity in identification of
enterohemorrhagic Escherichia coli strains. J. Clin. Microbiol.
35(3): 656662.
I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:04 AM
Can. J. Microbiol. Vol. 49, 2003
Health Canada. 2000. 1999 Annual report of the Health Canada
Salmonella reference laboratory. Health Canada Salmonella Ref-
erence Laboratory.
Johnson, R.P., Wilson, J.B., Michel, P., Rahn, K., Renwick, S.A.,
Gyles, C.L., and Spika, J.S. 1999. Human infection with
verocytotoxigenic Escherichia coli associated with exposure to
farms and rural environments. In Escherichia coli O157:H7 in
farm animals. Edited by C.S. Stewart and H.J. Flint. CABI Pub-
lishing, New York. pp. 147168.
Jones, I.G., and Roworth, M. 1996. An outbreak of Escherichia
coli O157 and campylobacteriosis associated with contamina-
tion of a drinking water supply. Public Health, 110(5): 277282.
Keene, W.E., McAnulty, J.M., Hoesly, F.C., Williams, L.P., Jr.,
Hedberg, K., Oxman, G.L., Barrett, T.J., Pfaller, M.A., and
Fleming, D.W. 1994. A swimming-associated outbreak of hem-
orrhagic colitis caused by Escherichia coli O157:H7 and
Shigella sonnei. N. Engl. J. Med. 331(9): 579584.
Khakhria, R., Woodward, D., and Johnson, W. 1996. Salmonella,
Shigellae, pathogenic E. coli, Campylobacter and Aeromonas
identified in Canada: annual summary 1994. National Labora-
tory for Enteric Pathogens, LCDC, Health Canada.
Khaleel, R., Reddy, K.R., and Overcash, M.R. 1980. Transport of
potential pollutants in runoff water from land areas receiving an-
imal wastes: a review. Water Res. 14: 421436.
Kondro, W. 2000. E. coli outbreak deaths spark judicial inquiry in
Canada. Lancet, 355(9220): 2058.
Kramer, M.H., Herwaldt, B.L., Craun, G.F., Calderon, R.L., and
Juranek, D.D. 1996. Surveillance for waterborne-disease out-
breaks United States, 19931994. MMWR, 45(SS-1): 133.
Levy, D.A., Bens, M.S., Craun, G.F., Calderon, R.L., and
Herwaldt, B.L. 1998. Surveillance for waterborne-disease out-
breaks United States, 19951996. MMWR, 47(SS-5): 134.
McGowan, K.L., Wickersham, E., and Strockbine, N.A. 1989.
Escherichia coli O157:H7 from water. Lancet, I(8644): 967968.
Michel, P., Wilson, J.B., Martin, S.W., Clarke, R.C., McEwen,
S.A., and Gyles, C.L. 1999. Temporal and geographical distribu-
tions of reported cases of Escherichia coli O157:H7 infection in
Ontario. Epidemiol. Infect. 122(2): 193200.
Natural Resources Conservation Service. 1999. Agricultural waste
management field handbook. United States Department of Agri-
culture, Washington, D.C.
OConnor, D.R. 2002. Report of the Walkerton inquiry: the events
of May 2000 and related issues. Ontario Ministry of the Attor-
ney General.
Payment, P., Siemiatycki, J., Richardson, L., Renaud, G., Franco,
E., and Prevost, M. 1997. A prospective epidemiological study
of gastrointestinal health effects due to the consumption of
drinking water. Int. J. Environ. Health Res. 7: 531.
Polo, F., Figueras, M.J., Inza, I., Sala, J., Fleisher, J.M., and
Guarro, J. 1998. Relationship between presence of Salmonella
and indicators of faecal pollution in aquatic habitats. FEMS
Microbiol. Lett. 160(2): 253256.
Roach, P.D., Olson, M.E., Whitley, G., and Wallis, P.M. 1993.
Waterborne Giardia cysts and Cryptosporidium oocysts in the
Yukon, Canada. Appl. Environ. Microbiol. 59(1): 6773.
Rowe, P.C., Orrbine, E., Lior, H., Wells, G.A., Yetisir, E., Clulow,
M., and McLaine, P.N. 1998. Risk of hemolytic uremic syn-
drome after sporadic Escherichia coli O157:H7 infection: results
of a Canadian collaborative study. Investigators of the Canadian
Pediatric Kidney Disease Research Center. J. Pediatr. 132(5):
777782.
Sanderson, M.W., Gay, J.M., Hancock, D.D., Gay, C.C., Fox, L.K.,
and Besser, T.E. 1995. Sensitivity of bacteriologic culture for
2003 NRC Canada
Color profile: Disabled
Composite Default screen
Johnson et al.
detection of Escherichia coli O157:H7 in bovine feces. J. Clin.
Microbiol. 33(10): 26162619.
Sargeant, J.M., Gillespie, J.R., Oberst, R.D., Phebus, R.K., Hyatt,
D.R., Bohra, L.K., and Galland, J.C. 2000. Results of a longitu-
dinal study of the prevalence of Escherichia coli O157:H7 on
cow-calf farms. Am. J. Vet. Res. 61(11): 13751379.
Sorenson, O. 2000. A survey to estimate the prevalence of Salmo-
nella spp., and Johnes disease in Alberta dairy herds. 99M461.
Alberta Agricultural Research Institute.
Statistics Canada Online. 2002. Available from http://www.statcan.
ca/english/Pgdb/Economy/Census/econ106j.htm; http://www.statcan.
ca/english/Pgdb/Economy/Census/econ109j.htm; http://www.statcan.
ca/english/Pgdb/Economy/Census/econ105j.htm. [updated Sept.
2002].
Swerdlow, D.L., Woodruff, B.A., Brady, R.C., Griffin, P.M.,
Tippen, S., Donnell, H.D., Jr., Geldreich, E., Payne, B.J., Meyer,
A., Jr., Wells, J.G., et al. 1992. A waterborne outbreak in Mis-
souri of Escherichia coli O157:H7 associated with bloody diar-
rhoea and death. Ann. Intern. Med. 117(10): 812819.
I:\cjm\cjm4905\W03-046.vp
July 8, 2003 10:33:04 AM
335
Valcour, J.E., Michel, P., McEwen, S.A., and Wilson, J.B. 2002.
Associations between indicators of livestock farming intensity
and incidence of human Shiga toxin-producing Escherichia coli
infection. Emerg. Infect. Dis. 8(3): 252257.
Van Donkersgoed, J., Graham, T., and Gannon, V. 1999. The prev-
alence of verotoxins, Escherichia coli O157:H7, and Salmonella
in the feces and rumen of cattle at processing. Can. Vet. J. 40(5):
332338.
Van Donsel, D.J., and Geldreich, E.E. 1971. Relationships of Sal-
monellae to fecal coliforms in bottom sediments. Water Res. 5:
10791087.
Wallis, P.M., Erlandsen, S.L., Isaac-Renton, J.L., Olson, M.E.,
Robertson, W.J., and Van Keulen, H. 1996. Prevalence of
Giardia cysts and Cryptosporidium oocysts and characterization
of Giardia spp. isolated from drinking water in Canada. Appl.
Environ. Microbiol. 62(8): 27892797.
Waters, J.R., Sharp, J.C., and Dev, V.J. 1994. Infection caused by
Escherichia coli O157:H7 in Alberta, Canada, and in Scotland:
a five-year review, 19871991. Clin. Infect. Dis. 19(5): 834843.
2003 NRC Canada