Quantitative multi-year elucidation of fecal sources of waterborne pathogen contamination in the South Nation River basin using Bacteroidales microbial source tracking markers

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Quantitative multi-year elucidation of fecal sources of waterborne pathogen contamination in the South Nation River basin using Bacteroidales microbial source tracking markers Romain Marti a, Victor P.J. Gannon b, Cassandra Jokinen b, Martin Lanthier c, David R. Lapen c, Norman F. Neumann d,e, Norma J. Ruecker e,f, Andrew Scott a, Graham Wilkes c, Yun Zhang a, Edward Topp a,* a

Agriculture and Agri-Food Canada, 1391 Sandford Str., London, Ontario N5V 4T3, Canada Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, PO Box 640, Township Road 9-1, Lethbridge, Alberta T1J 3Z4, Canada c Agriculture and Agri-Food Canada, Eastern Cereal and Oilseed Research Centre, Ottawa, Ontario K1A 0C6, Canada d School of Public Health, University of Alberta, Edmonton, Alberta, Canada e Alberta Provincial Laboratory for Public Health, 3030 Hospital Drive NW, Calgary, Alberta, Canada f Department of Microbiology and Infectious Diseases, University of Calgary, Calgary, Alberta, Canada b

article info

abstract

Article history:

Over a seven-year period (2004e2010) 1095 water samples were obtained from the South

Received 17 October 2012

Nation River basin at multiple watershed monitoring sites (Ontario, Canada). Real-time PCR

Received in revised form

using Bacteroidales specific markers was used to identify the origin (human (10% preva-

30 January 2013

lence), ruminant (22%), pig (w2%), Canada goose (4%) and muskrat (7%)) of fecal pollution.

Accepted 2 February 2013

In parallel, the distribution of fecal indicator bacteria and waterborne pathogens (Crypto-

Available online 14 February 2013

sporidium oocysts, Giardia cysts, Escherichia coli O157:H7, Salmonella enterica and Campylobacter spp.) was evaluated. Associations between the detection of specific Bacteroidales

Keywords:

markers and the presence of fecal indicator bacteria, pathogens, and distinct land use or

Agriculture

environmental variables were evaluated. Linear correlations between Bacteroidales markers

Water quality

and fecal indicator bacteria were weak. However, mean marker densities, and the presence

Microbial source tracking

and absence of markers could be discriminated on the basis of threshold fecal indicator densities. The ruminant-specific Bacteroidales marker was the most frequently detected marker in water, consistent with the large number of dairy farms in the study area. Detection of the human or the ruminant markers were associated with a slightly higher risk of detecting S. enterica. Detection of the muskrat marker was related to more frequent Campylobacter spp. detections. Important positive associations between markers and pathogens were found among: i) total Bacteroidales and Cryptosporidium and Giardia, ii) ruminant marker and S. enterica, and iii) muskrat and Campylobacter spp. Crown Copyright ª 2013 Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ1 519 457 1470x235; fax: þ1 519 457 3997. E-mail address: [email protected] (E. Topp). 0043-1354/$ e see front matter Crown Copyright ª 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2013.02.009

2316

1.

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

Introduction

A variety of enteric bacteria, viruses and parasites carried by humans, livestock animals and wildlife pose a public health risk when they contaminate water used for drinking, recreation, or crop irrigation. Exposure to fecally-contaminated water is estimated to be at the origin of 4% of all deaths, and 6% of the total disease burden occurring worldwide (Pruss et al., 2002). Fecal pollution leads to significant economic loss including increased water treatment costs associated with poor source water quality, beach closures, and compromised shellfish fisheries (Rabinovici et al., 2004). Detecting all pathogens that can compromise water quality is not technically feasible or cost-effective (Field and Samadpour, 2007), and thus fecal indicator bacteria (e.g. Escherichia coli and enterococci) are monitored and used as pathogen surrogates to measure and mandate water quality standards (National Research Council and Committee on Indicators for Waterborne Pathogens, 2004; Yates, 2007). Fecal indicator bacteria densities are generally coherent with the fecal pollution input intensity (Dufour, 1984), but are variably correlated with detection of pathogenic microorganisms (Edge et al., 2012; Field and Samadpour, 2007; Jokinen et al., 2012; Wilkes et al., 2011). Factors that may be responsible for the apparent discrepancy between fecal indicator bacteria and fecal pathogens could include differential environmental fitness and survival characteristics, and differences in size and surface properties resulting in various sorption to organic material, settling rates, and re-mobilization potential (Payment and Locas, 2011). At a watershed scale, knowledge of the spatial and temporal distribution of potential human, agricultural and wildlife sources of fecal pollution is a prerequisite for targeting mitigation measures (USEPA, 2005). Some waterborne enteric pathogens may be informative with respect to their source origin. For example, various Cryptosporidium species and genotypes can be very host specific (Ruecker et al., 2012). Enterohemorrhagic E. coli O157:H7 are mainly carried by cattle (Bach et al., 2002). However, the origin of waterborne fecal microorganisms are now best inferred on the basis of a variety of Microbial Source Tracking (MST) methods that have recently become available (Roslev and Bukh, 2011; Santo Domingo et al., 2007; Simpson et al., 2002; USEPA, 2005). Powerful methods infer fecal source on the basis of waterborne nucleic acid markers that are host-source specific (Roslev and Bukh, 2011). Nucleic acid markers can include host mitochondrial DNA originating from sloughed intestinal epithelial cells shed in feces, viral DNA or RNA, or bacterial DNA (Roslev and Bukh, 2011). Bacterial markers of choice include members of the Bacteroidales, dominant members of the fecal flora and obligate anaerobes, a feature which obviates regrowth following shedding (Anderson et al., 2005; Fiksdal et al., 1985). Bacteroidales markers have been developed to identify fecal contamination of human (Kildare et al., 2007; Seurinck et al., 2005), livestock animal (Mieszkin et al., 2009; Okabe et al., 2007; Reischer et al., 2006) and wildlife animal (Fremaux et al., 2010; Marti et al., 2011b) origin. However, little is known about associations between these genetic markers and the distribution of human pathogens, or for that matter

other variables such as land use, weather, and hydrology (Fremaux et al., 2009; Gourmelon et al., 2010b; Jokinen et al., 2012). In 2004 we initiated a long-term surface water quality research program within a 200 km2 portion of the South Nation River basin, located in Eastern Ontario, Canada. The basin contains multiple watersheds and is characterized by mixed land use including livestock and crop production, small urban centres, rural residences, and wildlife habitat. The overall objectives of the research are to better understand at a policy-relevant scale the contributions of agriculture to water quality impairment, and validate the efficacy of agricultural better management practices (BMPs) expected to mitigate water quality impairment. Within this context, we have explored relationships between land use practices, the implementation of some BMPs, and the distribution of pathogenic organisms in surface water of the watersheds. Relationships between densities of indicator bacteria and the distribution of pathogens have been explored, waterborne microorganisms have been characterized in detail, and this information has been used to comment on risks to human health, and how these might vary according to fecal inputs (Edge et al., 2012; Lanthier et al., 2011; Lyautey et al., 2007; Ruecker et al., 2007, 2012; Sunohara et al., 2012; Wilkes et al., 2009, 2011). Anticipating future developments in molecular microbiology, DNA was extracted from water samples throughout the multi-year study, and has been archived. This rich resource can be used to retrospectively evaluate and understand the distribution of specific genes (e.g. antibiotic resistance determinants), organisms (e.g. emerging pathogens), or communities (e.g. fecal bacteria) of interest using PCR, microarray or metagenomic methods, for example. In the present study, the primary objectives were to: i) Examine if newly available source-specific qPCR markers were associated with densities of fecal indicator bacteria and the detection of waterborne pathogens; and examine relationships between Bacteroidales marker densities/detection in water and potential determining factors including variation in land use, hydrology, and weather. Taken together, this new information will help target future water quality improvement efforts within the drainage basin.

2.

Material and methods

2.1.

Site description and water sample collection

The South Nation River basin (w3900 km2) is located east of Ottawa (45 170 1700 N, 75 070 4600 W) in the province of Ontario, Canada (Fig. 1). Agricultural farming activities are common in the watershed, including dairy farming and cash and livestock production. Many of the cultivated fields in the watershed are flat and tile drained. Potential sources of fecal pollution within the study area include spring and fall application of stored manure, animals on pasture, septic systems, effluent release from municipal lagoons, and wildlife. Results presented in this study are from water samples collected at up to a total of 24 long term water sampling sites,

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

2317

Fig. 1 e Water sample site locations in the South Nation River basin study area, meteorological and stream flow gauging stations utilized in the study, and inset map indicating the approximate location of the study region in North America.

all located within a 200 km2 area, normally visited on a biweekly basis, beginning in March 2004 and ceasing in November 2010 (7 years). Due to ice, some sample sites could not be sampled in winter or late fall. Yearly sampling of all sites typically began in MarcheApril and ended in NovembereDecember. These sites are described in Wilkes et al. (2011). A total of 1095 samples (2004 n ¼ 186; 2005 n ¼ 168; 2006 n ¼ 146; 2007 n ¼ 94; 2008 n ¼ 204; 2009 n ¼ 206; 2010 n ¼ 91) were collected and are included in the analysis presented here. In any given year, 7e24 sites were visited and sampled over 2e3 seasons. Methods for sampling surface water and collecting ancillary water quality data are described in Wilkes et al. (2011).

2.2.

Microbiological analysis of water

The present study uses water quality data previously described in detail, supplemented with recent unpublished results (Lyautey et al., 2007; Ruecker et al., 2007, 2012; Wilkes et al., 2009, Wilkes et al., 2011). Indicator bacteria (E. coli, Enterococcus spp., Clostridium perfringens, total and fecal coliforms) were enumerated by membrane-filtration on semiselective differential media, and confirmed using standard methods (Lanthier et al., 2011; Wilkes et al., 2009). Methods for the enrichment, isolation and confirmation of E. coli O157:H7, Salmonella enterica, and Campylobacter spp. are described in Jokinen et al. (2010). Cryptosporidium oocysts and Giardia cysts were enumerated using USEPA Method 1623 and genotyped (Ruecker et al., 2007, 2011). Detection limits of the different methods are indicated in Table S1.

2.3.

Quantitative PCR analysis

The method used for DNA extraction from water samples (n ¼ 1095) for qPCR was the same as described in Marti et al. (2011b). The elution volume was 100 mL. PCR amplification was performed using a Bio-Rad CFX96 Real-time PCR instrument with Bio-Rad CFX Manager software,

version 2.0. All primer and probes used in this study were purchased from SigmaeAldrich (Toronto, ON) and the sequences are presented in Table S2. The primer set designed by Kildare et al. (2007) was used to measure total Bacteroidales concentrations. The markers HF183, BacR, Pig-2-Bac, CGOF1-Bac and MuBa were used to detect and quantify human, ruminant, pig (Sus scrofa), Canada goose (Branta canadensis) and muskrat (Ondatra zibethicus) fecal pollution, respectively (Fremaux et al., 2010; Marti et al., 2011b; Mieszkin et al., 2009; Seurinck et al., 2005). All markers used Taqman chemistry, except for the human marker which used SYBR Green chemistry. Mix preparation, PCR reactions and standard curves for quantification have been already described (Marti et al., 2011b). The limit of quantification of the total Bacteroidales and the specific qPCR MST markers was 1.0  104 and 1.0  103 copies per 100 mL of filtered water, respectively (Table S1) (Gourmelon et al., 2010a; Kildare et al., 2007; Marti et al., 2011a, 2011b; Mieszkin et al., 2009).

2.4.

Statistical methods and data analysis

Analyses undertaken in the present study included: i. the frequency of marker detection by sample site, ii. the seasonality in marker detection with accompanying Chi-square and Fisher’s exact tests, iii. linear correlations between marker densities and fecal indicator bacteria densities, iv. identification of fecal indicator bacteria densities (independent criteria) by which MST marker density data (dependent criteria) could be subdivided into homogenous marker density groups using regression tree analysis in CART (CART Pro 6.0 Salford Systems, San Diego, CA), v. using classification tree methods in CART to discriminate marker presence and absence data on the basis of fecal indicator bacteria densities, vi. exploratory analysis of land use-environmental variable (independent criteria) associations with marker densities (dependent criteria) using CART regression tree analysis, vii. using ManneWhitney U tests to test for significant differences in marker density

2318

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

distributions associated with the presence and absence of zoonotic pathogens, viii. the odds ratios of the odds of detecting a pathogen in water when a MST marker was present versus the odds of detecting a pathogen in water when the marker was absent in water; ix. the co-occurrence of pathogens and markers with significance identified by Fisher’s exact tests, and x. determination of marker density values that optimally split zoonotic pathogens into homogeneous presence and absence groups (i.e., CART is used in classification tree mode, targeting pathogen presence and absence, with marker densities as independent data splitting criteria). For analysis (vi.), independent land use and environmental variables used were defined in Wilkes et al. (2011); these included: mean daily temperature on day of water sampling, cropland coverage in upstream site catchment, nearest upstream distance to a pasture observation, nearest upstream distance to a barn observation, season in which the sample was collected, stream order of the sample site (a number which represents the numerical size of the watercourse sampled; see Shreve (1966)), turbidity of the water when sampled, and mean daily river discharge at Environment Canada’s gauging station on the Payne River tributary of the greater South Nation River (Fig. 1). See Wilkes et al. (2009, 2011; 2013) for more in depth discussion of CART methods.

respect to agricultural inputs, very few samples were positive for the pig marker, consistent with the very few pig farming operations in the area (Fig. 2C). In contrast, dairy operations dominate livestock-based agricultural activity in the study area, and this land use is consistent with the ruminant marker being ubiquitously and frequently detected (Fig. 2D). With respect to wildlife, the muskrat marker was detected in more than half of sampling sites (n ¼ 14) with the most frequent detections at sites 3, 8 and 20 (Fig. 2E). Muskrat positive samples were notably clustered at sites 18e23; these sampling a series of intermittent agricultural drainage systems (Fig. 1). The Canada goose marker was infrequently detected, as only seven sites gave positive results (sites 1,6, 8, 15, 18, 22 and 24) (Fig. 2F). The Canada goose marker was detected in more than 4% of observations only at sites 8, 18, 22 and 24 (Fig. 2F). Seasonal variation in the distribution of specific markers was evaluated (Table 2). The frequency of total Bacteroidales occurrence was significantly different amongst the seasons, with slightly lower occurrence in the spring (91%) relative to the summer (96%) and fall (95%). The seasonal detection of the ruminant marker was significant at the 0.05 level. It increased in prevalence from spring to summer to fall, by 3 and 9% respectively. The frequency of detection of the human, pig, Canada goose and muskrat markers did not vary significantly with season.

3.

Results

3.2. Associations between specific MST markers and fecal indicator bacteria

3.1.

Frequency and distribution of MST marker detection

The distribution of MST markers in the 1095 water samples obtained from 2004 to 2010 was determined by qPCR (Table 1). The total Bacteroidales marker was detected in 94% of the samples. Source-specific MST markers were detected in 35% (n ¼ 381) of the samples. Among those samples that were positive, the most frequently detected marker was ruminant (22% of all water samples; n ¼ 240), followed by human (10%), muskrat (7%), Canada goose (4%) and pig (1.7%). The total Bacteroidales marker was detected at high frequency across all sites, indicating some amount of fecal input entering into, and being transported within, the study area watercourses (Fig. 2A). Only sites 1 and 21 had total Bacteroidales prevalence below 90% (w87 and w79%, respectively), all other sites had prevalence 90%. The human-specific marker was undetectable at most sites, but was detected in over 40% of the samples obtained at sites 3, 8 and 24 (Fig. 2B). With

Table 1 e The prevalence of MST markers in water samples. Bacteroidales marker Total Bacteroidales Human Pig Ruminant Muskrat Canada goose

Number of positive results

Number of negative results

% Positive per total number of samples

1028 112 19 240 74 39

67 983 1076 849 1021 1056

94 10 1.7 22 7 4

Spearman rank order correlations coefficients (rs) (Table S3) were calculated among fecal indicator bacteria and MST markers. Total Bacteroidales densities were significantly positively correlated (rs 0.16e0.24) with all indicator bacteria except C. perfringens. Ruminant markers showed similar correlation trends, but were weaker (rs 0.07e0.13). Table S4 summarizes results whereby marker presence and absence (classification tree) and marker densities (regression tree) could be discriminated, in a binary way, on the basis of a specific type of fecal indicator bacteria. Higher densities of fecal coliforms were associated with greater detection of total Bacteroidales (Table S4b); 84% of the total Bacteroidales positive water samples had fecal coliform densities >51 CFU 100 mL1 (Table S4b); 81% of samples that were positive for the ruminant marker and 91% of water samples that were positive for the muskrat marker contained Enterococcus spp. at >45.5 CFU 100 mL1 (Table S4b). In contrast, Canada goose markers were detected when fecal indicators were at lower threshold densities. Thus, 67% of samples that were positive for the Canada goose marker were detected when C. perfringens was below 5.5 CFU 100 mL1, whereas round to 33% of the Canada goose positive water samples were detected at C. perfringens densities >5.5 CFU 100 mL1. No human markers were detected at total coliform densities > 10,500 CFU 100 mL1 (n ¼ 34, Table S4b). Overall, E. coli was not as important as other fecal indicator bacteria for binary discrimination of the presence or absence of specific MST markers. Discrimination of higher and lower mean marker densities via binary splitting the marker density data on the basis of indicator bacteria densities could be achieved via CART; albeit r2 values for these simple binary results were

2319

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

Fig. 2 e Percentage of total water samples for each sampling site that had a detection of: A) Total Bacteroidales, B) Human marker, C) Pig marker, D) Ruminant marker, E) Muskrat marker, and F) Canada goose marker.

within 70 m (n ¼ 202) of pasture the average human marker density was 3.4  104 copies 100 mL1, whereas if it was farther downstream the average density was 4.8  102 copies 100 mL1 (Figure S1A). When average daily discharge was in excess of 0.10 m3 s1 (n ¼ 181) the average density was 1.1  104 copies 100 mL1, whereas if discharge was 0.10 m3 s1or lower (n ¼ 21) the average marker density was 2.3  105 copies 100 mL1, for sampling points closer to pasture. Further examination of these sampling points within a 70 m distance of pasture revealed they were all on a shared drain with homes in close proximity (<390 m). Due to a larger than expected number of human marker detections at site 24, which was considered a ‘pristine’ site with no evidence of

weak (Table S4a). For all but the goose marker, higher marker densities were associated with fecal indicator densities greater than the designated optimal fecal indicator split criteria. Interestingly, the optimal splitting indicator organisms here (upper panel, Table S4a) differed with those associated with the classification results previously described (lower panel, Table S4a).

3.3. Associations between land-use, environmental variables, and the density of MST markers The density of human marker was associated with flow path distance to pasture (Figure S1). If the sampling point was

Table 2 e The seasonal variation of MST marker detections amongst all sampling sites. Contingency table tests were used to determine significant seasonal differences. D [ positive sample, e [ no detection. *p < 0.05. Markers

Total Bacteroidales Human Pig Ruminant Muskrat Canada goose

Spring

Summer

Fall

Significance

þ

e

%þ

þ

e

%þ

þ

e

%þ

Chi-square test p-value

Fisher’s exact test p-value

334 31 7 61 17 17

32 335 359 299 349 349

91 8 2 17 5 5

385 44 7 82 35 16

17 358 395 320 367 386

96 11 2 20 9 4

278 37 4 86 22 5

16 257 290 208 272 289

95 13 1 29 7 2

0.028* 0.219 0.858 0.001* 0.080 0.111

0.033* 0.219 0.913 0.001* 0.074 0.102

2320

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

known factors in the watershed (e.g. septic, waste water treatment plant effluent, land application of municipal biosolids) that would contribute to human fecal material to the stream at such intensity, we excluded this site in the CART analyses. This minimized confounding the CART analysis in its capacity to uncover potential associations between the human marker and known sources/drivers of human fecal material. All pig marker data (n ¼ 1029) was split on the basis of nearest upstream distance to a pasture, where if the nearest pasture was within 0.25 km upstream of a sample site, the average density of the pig marker was 3.7  102 copies 100 mL1 (n ¼ 301; Figure S1B). If a pasture occurred at a distance greater than 0.25 km upstream of a sample site, the pig marker average was 3.0  101 copies 100 mL1. The variable that was found to initially split all the ruminant marker data was the percent of cropland in the watershed upstream of a sample site (Figure S1C). The majority of water samples (n ¼ 976; 95%) were associated with a value higher than 0.27 km2 of cropland per km2 of land surface. These water sampling sites had an average of 1.9  104 copies 100 mL1 of ruminant marker, whereas samples downstream of these less intensively cropped watersheds had an average of 1.9  105 copies 100 mL1. The CART selected variable splitting the >0.27 km2 per km2 data grouping, was the nearest upstream distance to a barn. Here, the majority of samples (n ¼ 884) were associated with locations where the nearest upstream distance to a barn was >0.25 km. The average ruminant marker density of this group of data was 1.1  104 copies 100 mL1. In contrast, samples taken where the nearest upstream distance to a barn was 0.25 km, had an average ruminant marker density of 9.4  104 copies 100 mL1. Finally, in conditions of greater cropland coverage and closer upstream proximity to barns when sample water turbidity was relatively higher (>24.25 NTU), the density of ruminant marker was on average higher, in relation to water taken from these areas where turbidity was relatively lower (24.25 NTU). Data for the muskrat marker (n ¼ 1029) were optimally split on the basis of mean daily temperature on day of sampling (Figure S1D). When the temperature was greater than 22.9  C the average muskrat copy number was 7.5  103 copies 100 mL1 (primarily the summer), whereas when mean daily air temperature was 22.9  C, average marker density was 1.3  103 copies 100 mL1. For samples collected during these relatively cooler air temperatures, and at sites where the nearest upstream distance to a pasture was 1.24 km, the

average marker density was 3.3  102 copies 100 mL1: the average density was 3.2  103 copies 100 mL1 for samples that were collected at sites where pasture was >1.24 km upstream. Stream order was a CART selected splitting variable for the mean daily air temperature >22.9 C group of data. Here, water obtained from watercourses that had a Shreve order greater than 34.5 had mean marker densities of 2.7  103 copies 100 mL1, whereas for the smaller stream order (34.5) group of data, the average marker density was 3.1  104 copies 100 mL1.

3.4. Associations between pathogen detection and MST marker presence or absence Amongst the 1095 water samples analysed (Table S5), Cryptosporidium oocysts were detected in 218 samples (20% of all water samples), Giardia cysts was found in 113 samples (10%), E. coli O157:H7 in 22 samples (2%), S. enterica in 35 samples (3%) and Campylobacter spp. in 358 samples (33%). Associations between the presence or the absence of the various pathogens and the specific MST markers were explored (Table 3, Tables S5 and S6). With four exceptions, there was no significant change in odds of pathogen occurrence in the presence of a marker in comparison with pathogen occurrence in the absence of a marker, as evidenced by odds ratios with confidence intervals that bracketed unity (Table 3). The three significant odds ratios greater than one were between detection of S. enterica and the human marker (odds ratio of 2.37), S. enterica and the ruminant marker (odds ratio of 2.07), and between detection of Campylobacter spp. and the muskrat marker (odds ratio of 2.43; Table 3). The odds ratio of Cryptosporidium was significantly less than 1 for (odds ratio of 0.27) for Canada goose comparisons, indicating a significant decrease in the odds of Cryptosporidium presence when the Canada goose marker was detected. ManneWhitney U tests were performed to evaluate if there were significant differences in the distribution of marker densities for when a specific pathogen was present and absent (Table S6). Only five pathogen-marker pairings were significant at the 0.05 level. Median values of total Bacteroidales were greater in the presence of Cryptosporidium (w4.2  105 copies 100 mL1), in comparison with samples lacking Cryptosporidium (w1.1  105 copies 100 mL1). CART selected marker splitting criteria amongst all possible split criteria in Table S7 are as follows: when the density of the muskrat marker was 3.6  103 copies 100 mL1,

Table 3 e Odds ratio (OR) estimate and 95% confidence interval [CI] limits of the OR estimate in brackets, summarizing the odds of a pathogen occurring when a MST marker is present, relative to the odds of a pathogen occurring when a MST marker is absent. Bold values indicate OR CI limit ranges that are greater than, or less than, unity. MST Marker Total Bacteroidales Human Pig Ruminant Muskrat Canada Goose

Cryptosporidium oocysts 1.61 [0.46e5.68] 0.91 [0.47e1.76] 0.94 [0.16e5.71] 1.44 [0.86e2.42] 0.94 [0.26e3.40] 0.27 [0.08e0.88]

Giardia cysts.

E. coli O157:H7

2.12 1.10 1.45 1.37 0.93 0.64

0.00 0.75 0.00 1.45 0.53 0.00

[0.45e9.96] [0.55e2.17] [0.24e8.82] [0.82e2.29] [0.24e3.65] [0.17e2.38]

[0.00e0.00] [0.17e3.28] [0.00e0.00] [0.56e3.76] [0.07e4.02] [0.00e0.00]

S. enterica

Campylobacter spp.

1.03 [0.24e4.45] 2.37 [1.04e5.39] 0.00 [0.00e0.00] 2.07 [1.01e4.26] 1.07 [0.32e3.58] 1.64 [0.37e7.20]

1.04 [0.57e1.89] 0.89 [0.57e1.39] 0.55 [0.10e3.02] 0.75 [0.52e1.07] 2.43 [1.40e4.24] 0.60 [0.27e1.33]

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

45% of those water samples were positive for Campylobacter spp., whereas at higher densities of the muskrat marker, the percentage of Campylobacter spp. positive samples in those samples increased to 85% (n ¼ 41). A predictive threshold value was also found for the total Bacteroidales marker with respect to the presence of Cryptosporidium. When total Bacteroidales densities were 4.7  105 copies 100 mL1 and 1.1  106 copies in water, 52% and 28% of those samples were positive for Cryptosporidium spp. and Giardia spp., respectively. Whereas at higher total Bacteroidales densities, 78% and 44% of those water samples were positive for Cryptosporidium spp. and Giardia spp., respectively. For S. enterica, the ruminant marker was the CART selected marker amongst all marker for discrimination of this pathogen: samples with ruminant marker densities above 1.5  103 had S. enterica detections rates of 9%, whereas S. enterica detections rates below1.5  103 were 4%.

4.

Discussion

In the present study the frequency of detection of specific fecal sources were ranked ruminant > human > muskrat > Canada goose > pig (Table 1). Dairy and beef cattle are by far the most dominant livestock type within the study area. Animals grazing on pasture, animals excreting in farm yards, and application to cropland of solid and slurry manure represent potential sources of surface water exposure to fecal bacteria of cattle origin. Roadside surveys indicate a median of 0.45 barns per km2 in regions 10 km upstream of the sites, a rather intensive number of operations per unit area in relation to pristine sites with densities of 0. The ubiquity of dairy and beef operations within the study area and the detection of Cryptosporidium andersoni in surface water in the same study area (Ruecker et al., 2007), is consistent with the widespread detection of the ruminant MST marker (Fig. 2D). Interestingly, with respect to human sources of fecal contamination, site 3 and site 24 represented the sites with the greatest human marker prevalence. Site 3 has homes along the watercourse which may have contributed to septic inputs, but site 24 has no residential land uses in the upstream catchment, whatsoever. It is a forested-wetland site where human inputs could only have come from sources such as campers, hunters, or illegal dumping of holding tanks in or near the watercourse (e.g. recreation vehicles, other stored fecal waste in from waste tanks). Hence, while formative pollution mechanisms cannot be deciphered from the data, the human marker did serve to identify at least, human activity in otherwise natural undeveloped land in the watershed. Human marker hits at other sites can also be explained by municipalities in the study area, Casselman, Crysler and St-Albert, each treating its own municipal waste and discharging it in spring and fall. As noted already, there are numerous rural residences along water courses with septic systems, and the distribution of the human MST marker across study sites (Fig. 2B) is associated with rural development (Figure S1A), and clustered at sites downstream of urbanized areas (Fig. 1). To underscore potential septic inputs, the CART analysis revealed that elevated densities (excluding site 24) of the human marker were associated with pasture within 70 m of the sampling site

2321

(Figure S1A). Samples within this grouping were all from sites 15, 22, and 23 (Fig. 1). These sites are in close proximity to homes situated along small watercourses located in rural pasture settings (a common occurrence in the region). These detections are likely due to septic system discharge into watercourses that receive little dilution, as no municipal biosolids or combined sewer outflows occurred at or upstream of these locations. With respect to wildlife, muskrats were frequently observed in the study area (Marti et al., 2011b) as were voles, and observations are clustered at sites 18e23 where a series of agricultural ditches draining small catchments were sampled (Fig. 1; (Lyautey et al., 2007)). Muskrat marker densities were most pronounced in summer, consistent with observed vole activity and muskrats having multiple summer litters, on lower stream order ditches/streams which provide necessary habitat required for survival. The highest prevalence of the goose marker was associated with site 24, the forested-wetland area where geese may accrue, especially during migrating season. The pig marker, which was infrequently detected, was most prevalent at site 18, a site that is variably impacted by runoff from fields that have received liquid swine manure amendments from time to time. Overall, the frequency and distribution of the source-specific MST markers is generally consistent with land use within the study area. Only about a third of all the water samples evaluated could be ascribed a fecal source origin on the basis of an MST marker. There are large numbers of wildlife within the study area, voles and pigeons for example, for which we currently lack MST tools. In future studies it will be important to develop and implement markers that can target these sources. In general, it is becoming clear that studies evaluating fecal inputs within a given catchment must take into careful consideration wildlife that are prominent within the area (Graczyk et al., 2008; Marti et al., 2011b; Muirhead et al., 2011; Ruecker et al., 2007). Another plausible reason for the high number of MST unknowns in the present study is the method detection limit. Extraction of DNA from larger water volumes and devising means of eliminating PCR inhibitors are examples of steps that will incrementally improve detection limits (Leskinen et al., 2009; Schriewer et al., 2011). The density of MST markers and the density of fecal indicator bacteria were moderately to poorly correlated, with the exception of C. perfringens (Table S3). All other Spearman’s rank order correlations were weak. However, CART analysis specified clear indicator thresholds whereby higher marker densities were associated with higher indicator densities (Table S4). A study undertaken in the Oldman river drainage basin at a spatial scale and intensity of sampling (n ¼ 242) comparable to the present study reported similar observations (Jokinen et al., 2012). On the other hand, Fremaux et al. (2009) analysed the correlation between human, ruminant and pigspecific Bacteroidales markers with the distribution of fecal indicator and pathogenic bacteria and found a higher prevalence for human (HF183) and ruminant markers when the E. coli concentration was above 100 most probable number 100 mL1. Gourmelon et al. (2010b) found that below 5  102 E. coli 100 mL1, no MST markers were detected (Gourmelon et al., 2010b). In an urban creek and coastal area, higher densities of the human marker HF183 were associated with higher

2322

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

densities of fecal indicator bacteria, but in contrast, low densities of the human marker were a poor predictor of fecal indicator density (Flood et al., 2011; Sercu et al., 2009). Factors such as differential persistence in water will rapidly erode the association between the Bacteroidales and indicator bacteria (Dick et al., 2010), resulting in Bacteroidales not being effective at monitoring fecal pollution other than in cases of recent contamination events (Rogers et al., 2011; Tambalo et al., 2012). Associations between MST markers and the presence or absence of specific pathogens were explored in the present study (Table 3, Table S5 and S6). Regarding pathogen-marker associations, there are three instances where the odds ratios and 95% confidence intervals were greater than 1, and related Fisher’s exact tests were significant; instances where multiple statistical comparisons suggest that there are relevant increases in pathogen occurrence in the presence of MST markers in comparison with groups lacking the MST marker. Water containing S. enterica was w2.4 times more likely to contain the human marker, and w2.1 times more likely to contain the ruminant marker, than not. This predictive association between human fecal pollution and S. enterica is in agreement with Savichtcheva et al. (2007). The persistence characteristics of Salmonella Typhimurium and the human (HF183) and ruminant markers were similar in microcosms (Walters and Field, 2009). Water containing Campylobacter spp. was 2.4 times more likely to contain the muskrat marker than not. These results suggest that humans and ruminants are a source of S. enterica, and muskrats are a source of Campylobacter spp., that are found in surface water. Several studies have shown the potential of livestock, human, avian and mammalian wildlife to shed S. enterica into surface water (reviewed in Levantesi et al., 2012). Ruminant fecal pollution has been associated with a higher detection level of S. enterica in water, but neither human marker HF183 nor ruminant marker were correlated with thermotolerant Campylobacter, or E. coli shiga toxin producers (Fremaux et al., 2009). In the study area poorly operating septic systems do occur and are problematic as evidenced by direct observation of leakage into watercourses, and frequent complaints of local residents to mandated water quality authorities. Of the 35 water samples that were positive for S. enterica, the most frequent serotype detected was I:4,5,12:b:- (12 water samples) followed by Kentucky (6 water samples). Specific serotypes generally don’t exhibit host-specificity, so their identity is not informative with respect to their fecal source at a watershed scale (Jokinen et al., 2011; Levantesi et al., 2012). On the basis of the human and ruminant fecal marker-Salmonella association identified, mitigating both potential sources of human and bovine contamination is advisable (Table S6). Campylobacteriosis is the most prevalent cause of gastrointestinal illness in Canada (Galanis, 2007). In the present study muskrat fecal contamination was associated with the detection of Campylobacter spp. in water. Muskrats can be significant reservoirs for Campylobacter, for example over 47% of 189 muskrat fecal samples sampled in central Washington were positive for Campylobacter jejuni (Pacha et al., 1985). In the experimental study area, creation of physically-protected riparian zones that are sheltered from cattle access supported the notable proliferation of mammalian wildlife including voles, mice and muskrats (Sunohara et al., 2012). Voles were

observed in the same location; further investigation will be needed in order to see if these animals could be also responsible of fecal input. However, a dominant genotype of Cryptosporidium obtained from the study area is associated with muskrats (Ruecker et al., 2011, 2012). These results highlight the complexity of zoonotic pathogen reservoirs when considering fecal pollution in open systems such as watersheds. Land uses that enhance wildlife habitat could increase the proportion of fecal inputs from wildlife sources. There can be a quandary between, on the one hand, the desirable benefits in ecological services (e.g. biodiversity, filtering of nutrients and sediments) that wildlife habitat can offer, and on the other hand the potential public health consequences of fecal pollution accompanying the proliferation of wildlife. Recent studies associating Cryptosporidium species/genotypes and zoonotic pathogens in water suggest that risks associated with livestock inputs may be more problematic than those associated with wildlife (Wilkes et al., 2013). These complex interactions will need to be considered by policy makers and water quality managers when considering options appropriate for source water protection in a given area (McAllister and Topp, 2012). Surprisingly, Canada goose marker was negatively associated with Cryptosporidium spp. Yet avian sourced Cryptosporidium was also found to be of low relative prevalence in the region which is surprising considering the region is a fly zone for geese (Ruecker et al., 2012; Wilkes et al., 2013). This waterfowl can be a vector of numerous human pathogens including Cryptosporidium parvum and hominis (Graczyk et al., 2008; Kassa et al., 2004; Zhou et al., 2004). The marker CGFO1-bac has the same threshold detection than the other qPCR MST markers and it is highly specific and sensitive (Fremaux et al., 2010). However, the low number of positive results for the Canada goose marker and the difference of threshold detection between Cryptosporidium oocysts and the qPCR MST markers may have impacted the power and direction of the statistical results. More heavy polluted sites by Canada geese should be investigated in order to confirm observations and results.

5.

Conclusions

In a 7-year study analysing over 1000 water samples obtained from a mixed activity watershed, detection of S. enterica in water was associated with ruminant and human fecal pollution, and detection of Campylobacter spp. associated with muskrat fecal pollution. Detection of Cryptosporidium oocysts, Giardia cysts and E. coli O157:H7 were not associated with a specific fecal source. Two thirds of water samples were not associated with a specific fecal source directly attributable by the use of Bacteroidales markers suggesting that unidentified sources of fecal pollution are significant, or that the method detection limit requires improvement. Linear correlations between Bacteroidales markers and fecal indicator bacteria were weak.

Acknowledgements This study was funded by AAFC’s National Water Quality Surveillance Research Initiative (NWQSRI) through an

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

agreement with Health Canada, through the Sustainable Agriculture Environmental Systems (SAGES) program, the Watershed Evaluation of Beneficial Management Practice (WEBs) program, and the Alberta Water Research Institute. We thank the South Nation Conservation Authority, and the laboratory and field personnel in AAFC, PHAC, and APLPH for their excellent assistance. The OIE Salmonella Reference Laboratory of PHAC in Guelph, Ontario provided serotyping results. We thank three anonymous reviewers whose insightful comments improved the manuscript.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2013.02.009.

references

Anderson, K.L., Whitlock, J.E., Harwood, V.J., 2005. Persistence and differential survival of fecal indicator bacteria in subtropical waters and sediments. Applied and Environmental Microbiology 71 (6), 3041e3048. Bach, S.J., McAllister, T.A., Veira, D.M., Gannon, V.P.J., Holley, R.A., 2002. Transmission and control of Escherichia coli O157: H7 e a review. Canadian Journal of Animal Science 82 (4), 475e490. Dick, L.K., Stelzer, E.A., Bertke, E.E., Fong, D.L., Stoeckel, D.M., 2010. Relative decay of Bacteroidales microbial source tracking markers and cultivated Escherichia coli in freshwater microcosms. Applied and Environmental Microbiology 76 (10), 3255e3262. Dufour, A.P., 1984. Health Effects Criteria for Fresh Recreational Waters. U.S. Environmental Protection Agency, Cincinnati. Edge, T.A., El-Shaarawi, A., Gannon, V., Jokinen, C., Kent, R., Khan, I.U., Koning, W., Lapen, D., Miller, J., Neumann, N., Phillips, R., Robertson, W., Schreier, H., Scott, A., Shtepani, I., Topp, E., Wilkes, G., van Bochove, E., 2012. Investigation of an Escherichia coli environmental benchmark for waterborne pathogens in agricultural watersheds in Canada. Journal of Environmental Quality 41 (1), 21e30. Field, K.G., Samadpour, M., 2007. Fecal source tracking, the indicator paradigm, and managing water quality. Water Research 41 (16), 3517e3538. Fiksdal, L., Maki, J.S., LaCroix, S.J., Staley, J.T., 1985. Survival and detection of Bacteroides spp., prospective indicator bacteria. Applied and Environmental Microbiology 49 (1), 148e150. Flood, C., Ufnar, J., Wang, S., Johnson, J., Carr, M., Ellender, R., 2011. Lack of correlation between enterococcal counts and the presence of human specific fecal markers in Mississippi creek and coastal waters. Water Research 45 (2), 872e878. Fremaux, B., Boa, T., Yost, C.K., 2010. Quantitative real-time PCR assays for sensitive detection of Canada goose-specific fecal pollution in water sources. Applied and Environmental Microbiology 76 (14), 4886e4889. Fremaux, B., Gritzfeld, J., Boa, T., Yost, C.K., 2009. Evaluation of host-specific Bacteroidales 16S rRNA gene markers as a complementary tool for detecting fecal pollution in a prairie watershed. Water Research 43 (19), 4838e4849. Galanis, E., 2007. Campylobacter and bacterial gastroenteritis. Canadian Medical Association Journal 177 (6), 570e571. Gourmelon, M., Caprais, M.P., Le Mennec, C., Mieszkin, S., Ponthoreau, C., Gendronneau, M., 2010a. Application of library-independent microbial source tracking methods for

2323

identifying the sources of faecal contamination in coastal areas. Water Science and Technology 61 (6), 1401e1409. Gourmelon, M., Caprais, M.P., Mieszkin, S., Marti, R., Wery, N., Jarde, E., Derrien, M., Jadas-Hecart, A., Communal, P.Y., Jaffrezic, A., Pourcher, A.M., 2010b. Development of microbial and chemical MST tools to identify the origin of the faecal pollution in bathing and shellfish harvesting waters in France. Water Research 44 (16), 4812e4824. Graczyk, T.K., Majewska, A.C., Schwab, K.J., 2008. The role of birds in dissemination of human waterborne enteropathogens. Trends in Parasitology 24 (2), 55e59. Jokinen, C., Edge, T.A., Ho, S., Koning, W., Laing, C., Mauro, W., Medeiros, D., Miller, J., Robertson, W., Taboada, E., Thomas, J.E., Topp, E., Ziebell, K., Gannon, V.P.J., 2011. Molecular subtypes of Campylobacter spp., Salmonella enterica, and Escherichia coli O157:H7 isolated from faecal and surface water samples in the Oldman River watershed, Alberta, Canada. Water Research 45 (3), 1247e1257. Jokinen, C.C., Edge, T.A., Koning, W., Laing, C.R., Lapen, D.R., Miller, J., Mutschall, S., Scott, A., Taboada, E.N., Thomas, J.E., Topp, E., Wilkes, G., Gannon, V.P., 2012. Spatial and temporal drivers of zoonotic pathogen contamination of an agricultural watershed. Journal of Environmental Quality 41 (1), 242e252. Jokinen, C.C., Schreier, H., Mauro, W., Taboada, E., IsaacRenton, J.L., Topp, E., Edge, T., Thomas, J.E., Gannon, V.P.J., 2010. The occurrence and sources of Campylobacter spp., Salmonella enterica and Escherichia coli O157:H7 in the Salmon River, British Columbia, Canada. Journal of Water and Health 8 (2), 374e386. Kassa, H., Harrington, B.J., Bisesi, M.S., 2004. Cryptosporidiosis: a brief literature review and update regarding Cryptosporidium in feces of Canada geese (Branta canadensis). Journal of Environmental Health 66 (7), 34e40. 45. Kildare, B.J., Leutenegger, C.M., McSwain, B.S., Bambic, D.G., Rajal, V.B., Wuertz, S., 2007. 16S rRNA-based assays for quantitative detection of universal, human-, cow-, and dogspecific fecal Bacteroidales: a Bayesian approach. Water Research 41 (16), 3701e3715. Lanthier, M., Scott, A., Zhang, Y., Cloutier, M., Durie, D., Henderson, V.C., Wilkes, G., Lapen, D.R., Topp, E., 2011. Distribution of selected virulence genes and antibiotic resistance in Enterococcus species isolated from the South Nation River drainage basin, Ontario, Canada. Journal of Applied Microbiology 110 (2), 407e421. Leskinen, S.D., Harwood, V.J., Lim, D.V., 2009. Rapid dead-end ultrafiltration concentration and biosensor detection of enterococci from beach waters of Southern California. Journal of Water and Health 7 (4), 674e684. Levantesi, C., Bonadonna, L., Briancesco, R., Grohmann, E., Toze, S., Tandoi, V., 2012. Salmonella in surface and drinking water: occurrence and water-mediated transmission. Food Research International 45 (2), 587e602. Lyautey, E., Lapen, D.R., Wilkes, G., McCleary, K., Pagotto, F., Tyler, K., Hartmann, A., Piveteau, P., Rieu, A., Robertson, W.J., Medeiros, D.T., Edge, T.A., Gannon, V., Topp, E., 2007. Distribution and characteristics of Listeria monocytogenes isolates from surface waters of the South Nation River watershed, Ontario, Canada. Applied and Environmental Microbiology 73 (17), 5401e5410. Marti, R., Mieszkin, S., Solecki, O., Pourcher, A.M., HervioHeath, D., Gourmelon, M., 2011a. Effect of oxygen and temperature on the dynamic of the dominant bacterial populations of pig manure and on the persistence of pigassociated genetic markers, assessed in river water microcosms. Journal of Applied Microbiology 111 (5), 1159e1175. Marti, R., Zhang, Y., Lapen, D.R., Topp, E., 2011b. Development and validation of a microbial source tracking marker for the

2324

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 2 3 1 5 e2 3 2 4

detection of fecal pollution by muskrats. Journal of Microbiological Methods 87 (1), 82e88. McAllister, T.A., Topp, E., 2012. Role of livestock in microbiological contamination of water: commonly the blame, but not always the source. Animal Frontiers 2 (2), 17e27. Mieszkin, S., Furet, J.P., Corthier, G., Gourmelon, M., 2009. Estimation of pig fecal contamination in a river catchment by real-time PCR using two pig-specific Bacteroidales 16S rRNA genetic markers. Applied and Environmental Microbiology 75 (10), 3045e3054. Muirhead, R.W., Elliott, A.H., Monaghan, R.M., 2011. A model framework to assess the effect of dairy farms and wild fowl on microbial water quality during base-flow conditions. Water Research 45 (9), 2863e2874. National Research Council and Committee on Indicators for Waterborne Pathogens (Ed.), 2004. Indicators for Waterborne Pathogens. National Academies Press, Washington DC. Okabe, S., Okayama, N., Savichtcheva, O., Ito, T., 2007. Quantification of host-specific Bacteroides-Prevotella 16S rRNA genetic markers for assessment of fecal pollution in freshwater. Applied Microbiology and Biotechnology 74 (4), 890e901. Pacha, R.E., Clark, G.W., Williams, E.A., 1985. Occurrence of Campylobacter jejuni and Giardia species in muskrat (Ondatra zibethica). Applied and Environmental Microbiology 50 (1), 177e178. Payment, P., Locas, A., 2011. Pathogens in water: value and limits of correlation with microbial indicators. Ground Water 49 (1), 4e11. Pruss, A., Kay, D., Fewtrell, L., Bartram, J., 2002. Estimating the burden of disease from water, sanitation, and hygiene at a global level. Environmental Health Perspectives 110 (5), 537e542. Rabinovici, S.J.M., Bernknopf, R.L., Wein, A.M., Coursey, D.L., Whitman, R.L., 2004. Economic and health risk trade-offs of swim closures at a lake Michigan beach. Environmental Science and Technology 38 (10), 2737e2745. Reischer, G.H., Kasper, D.C., Steinborn, R., Mach, R.L., Farnleitner, A.H., 2006. Quantitative PCR method for sensitive detection of ruminant fecal pollution in freshwater and evaluation of this method in alpine karstic regions. Applied and Environmental Microbiology 72 (8), 5610e5614. Rogers, S.W., Donnelly, M., Peed, L., Kelty, C.A., Mondal, S., Zhong, Z., Shanks, O.C., 2011. Decay of bacterial pathogens, fecal indicators, and real-time quantitative PCR genetic markers in manure-amended soils. Applied and Environmental Microbiology 77 (14), 4839e4848. Roslev, P., Bukh, A.S., 2011. State of the art molecular markers for fecal pollution source tracking in water. Applied Microbiology and Biotechnology 89 (5), 1341e1355. Ruecker, N.J., Braithwaite, S.L., Topp, E., Edge, T., Lapen, D.R., Wilkes, G., Robertson, W., Medeiros, D., Sensen, C.W., Neumann, N.F., 2007. Tracking host sources of Cryptosporidium spp. in raw water for improved health risk assessment. Applied and Environmental Microbiology 73 (12), 3945e3957. Ruecker, N.J., Hoffman, R.M., Chalmers, R.M., Neumann, N.F., 2011. Detection and resolution of Cryptosporidium species and species mixtures by genus-specific nested PCR-restriction fragment length polymorphism analysis, direct sequencing, and cloning. Applied and Environmental Microbiology 77 (12), 3998e4007. Ruecker, N.J., Matsune, J.C., Wilkes, G., Lapen, D.R., Topp, E., Edge, T.A., Sensen, C.W., Xiao, L., Neumann, N.F., 2012. Molecular and phylogenetic approaches for assessing sources of Cryptosporidium contamination in water. Water Research 46 (0), 5135e5150.

Santo Domingo, J.W., Bambic, D.G., Edge, T.A., Wuertz, S., 2007. Quo vadis source tracking? towards a strategic framework for environmental monitoring of fecal pollution. Water Research 41 (16), 3539e3552. Savichtcheva, O., Okayama, N., Okabe, S., 2007. Relationships between Bacteroides 16S rRNA genetic markers and presence of bacterial enteric pathogens and conventional fecal indicators. Water Research 41 (16), 3615e3628. Schriewer, A., Wehlmann, A., Wuertz, S., 2011. Improving qPCR efficiency in environmental samples by selective removal of humic acids with DAX-8. Journal of Microbiological Methods 85 (1), 16e21. Sercu, B., Van De Werfhorst, L.C., Murray, J., Holden, P.A., 2009. Storm drains are sources of human fecal pollution during dry weather in three urban southern California watersheds. Environmental Science and Technology 43 (2), 293e298. Seurinck, S., Defoirdt, T., Verstraete, W., Siciliano, S.D., 2005. Detection and quantification of the human-specific HF183 Bacteroides 16S rRNA genetic marker with real-time PCR for assessment of human faecal pollution in freshwater. Environmental Microbiology 7 (2), 249e259. Shreve, R.L., 1966. Statistical law of stream numbers. Journal of Geology 74, 17e37. Simpson, J.M., Santo Domingo, J.W., Reasoner, D.J., 2002. Microbial source tracking: state of the science environ. Science and Technology 36 (24), 5279e5288. Sunohara, M.D., Topp, E., Wilkes, G., Gottschall, N., Neumann, N., Ruecker, N., Jones, T.H., Edge, T.A., Marti, R., Lapen, D.R., 2012. Impact of riparian zone protection from cattle on nutrient, bacteria, F-coliphage, and loading of an intermittent stream. Journal of Environmental Quality 41 (4), 1301e1314. Tambalo, D.D., Fremaux, B., Boa, T., Yost, C.K., 2012. Persistence of host-associated Bacteroidales gene markers and their quantitative detection in an urban and agricultural mixed prairie watershed. Water Research 46 (9), 2891e2904. USEPA, 2005. Microbial Source Tracking Guide, Document EPA600/R-05/064, p. 131pp.. Office of Research and Development, Washington, DC. Walters, S.P., Field, K.G., 2009. Survival and persistence of human and ruminant-specific faecal Bacteroidales in freshwater microcosms. Environmental Microbiology 11 (6), 1410e1421. Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Medeiros, D., Neumann, N., Ruecker, N., Topp, E., Lapen, D.R., 2009. Seasonal relationships among indicator bacteria, pathogenic bacteria, Cryptosporidium oocysts, Giardia cysts, and hydrological indices for surface waters within an agricultural landscape. Water Research 43 (8), 2209e2223. Wilkes, G., Edge, T.A., Gannon, V.P., Jokinen, C., Lyautey, E., Neumann, N.F., Ruecker, N., Scott, A., Sunohara, M., Topp, E., Lapen, D.R., 2011. Associations among pathogenic bacteria, parasites, and environmental and land use factors in multiple mixed-use watersheds. Water Research 45 (18), 5807e5825. Wilkes, G., Ruecker, N.J., Neumann, N.F., Gannon, V.P., Jokinen, C., Sunohara, M., Topp, E., Pintar, K.D., Edge, T.A., Lapen, D.R., 2013. Spatiotemporal analysis of Cryptosporidium species/genotypes and relationships with other zoonotic pathogens in surface water from mixed-use watersheds. Applied and Environmental Microbiology 79 (2), 434e448. Yates, M.V., 2007. Classical indicators in the 21st century e far and beyond the coliform. Water Environment Research 79 (3), 279e286. Zhou, L., Kassa, H., Tischler, M.L., Xiao, L., 2004. Host-adapted Cryptosporidium spp. in Canada geese (Branta canadensis). Applied and Environmental Microbiology 70 (7), 4211e4215.