Abstract

Private water wells provide drinking water for an estimated 4.1 million households in Canada yet remain understudied in the context of microbial water quality or human health impacts. As there exists little systematic surveillance for enteric infections or outbreaks related to well water sources, consumers may be at risk of waterborne infectious diseases. A standard protocol in Ontario requires 200 mL of water, collected, and submitted by well owners, half of which is used to analyze for Escherichia coli and total coliforms (TCs). The aim of this study was to determine the efficacy of testing small water volumes and to survey for other contaminants in addition to bacterial indicators to inform pathogen prevalence and fecal source in drinking water wells. Samples were assessed for E. coli and TCs, by culture, and genetic markers of Bacteroides spp., Campylobacter spp., Salmonella spp., and Shiga toxin-producing E. coli, using qPCR. The source of fecal contamination varied by the geographic region and may be explained by septic tank density and underlying geology, among other factors. A small number of samples (1.9%) showed the evidence of contamination with enteric pathogens. Lastly, E. coli measured by qPCR, as opposed to culture, correlated more strongly to Bacteroides markers.

INTRODUCTION

Water consumption without fear of illness is a defining characteristic of developed nations and is often taken for granted (Hrudey et al. 2006). Although the magnitude of drinking water-related illness in developed countries is lower than that in developing regions, waterborne disease outbreaks remain a significant cause of acute gastrointestinal illness and constitute a health risk as well as a substantial economic burden that could be better managed and prevented (Krentz et al. 2013). Of particular interest is waterborne disease attributable to private drinking water wells that serve a single household. Aquifers supplying these wells can become contaminated with enteric pathogens from both human and animal feces potentially causing serious illness (Charrois 2010). In Canada, an estimated 4.1 million (∼12%) households rely on private supplies, most of which are rural groundwater sources (Murphy et al. 2016), with almost 2 million residing in Ontario (Office of the Auditor General of Ontario 2014). However, these particular water supplies are subject to little or no systematic surveillance. In Ontario, the testing of private wells is provided cost free by Public Health Ontario (PHO) at the discretion of the well owner. Well owners voluntarily submit 200 mL samples, of which 100 mL is used to test for the presence of indicator bacteria, namely Escherichia coli and total coliforms (TCs). Maier et al. (2014) found that, in Ontario, only 11–12% of owners who submitted water samples for testing over 5 years (2008–2012) met the historic guidelines of at least three tests per year in any given year. Furthermore, less than 1% met those guidelines every year, when followed over 5 years, further elevating risk to the consumer, given the lack of risk assessment by testing. The nature of private water supplies can vary greatly over time and one test per year, or less, is insufficient to determine the stability of the water quality. Furthermore, enteric pathogens in private wells, and their subsequent associated infections, may go largely undetected, leaving a significant proportion of the population vulnerable to waterborne infectious diseases.

Traditionally, private water supplies are tested for TCs and E. coli as indicators of environmental and fecal contamination. TCs include intestinal genera such as Escherichia, Enterobacter, Klebsiella, Citrobacter and Kluyvera (Figueras & Borrego 2010). TC testing serves best as an indicator of environmental contamination (i.e., derived from soils and plants) or of the integrity of a water distribution system (WHO 2011). E. coli is the most commonly employed indicator of recent fecal contamination and potential enteric pathogen presence in drinking water, given its prevalence and specificity for the gastrointestinal tract (GIT) of animals and its inability to survive and grow in aquatic environments (Saxena et al. 2015). These bacteria do not allow for the identification of the fecal source, given their ubiquity across multiple species and environments. The detection of Bacteroides 16S rRNA genes by the PCR has been through microbial source tracking (MST) methodology, as certain sequences within the 16S rRNA gene are highly host-specific (Lee et al. 2014). Furthermore, Bacteroides are relatively easy to detect in environmental waters given their high abundance in feces and, as anaerobes, do not survive outside of the GIT. Thus, detection suggests that the contamination event is recent (Lee et al. 2014). It should be noted that the persistence of E. coli, TCs, and Bacteroides markers differs from each other in environmental waters and that their persistence is highly dependent on factors such as dissolved oxygen concentration, temperature, sunlight, salinity, and predator presence (Okabe et al. 2007; Walters & Field 2009; Lee et al. 2010).

In southern Ontario, four discrete regions with an increased relative risk for E. coli contamination among private wells have been identified as seen in Figure 1 (Krolik et al. 2013). The vulnerability of these wells to E. coli contamination suggests that they are also vulnerable to pathogen contamination, particularly those spread from the feces of humans or other animals. To address this, the high-risk regions were further investigated as well as a region with high cattle density (a circled orange region in Figure 1) (Statistics Canada 2011). At its core, this study aimed to investigate the value of small water volumes obtained from convenience samples for the detection of additional markers of contamination using molecular methods, in addition to the traditional culture and quantification of TC and E. coli colonies for water potability determinations. Specific aims were to assess 100 mL water samples from E. coli and/or TC-contaminated wells for (1) the presence of molecular markers of enteric pathogens, namely Campylobacter spp., Salmonella spp. and Shiga toxin-producing E. coli (STEC), (2) to determine the predominant sources of fecal contamination as human, bovine, or other animal, and (3) to establish the nature of the relationship between the presence of these pathogens and microbial indicators, including E. coli and TCs (by culture), Bacteroides fragilis, and E. coli (by qPCR).

Figure 1

The four regions, (1) Kingston/Belleville, (2) Niagara/Hamilton, and (3) and (4) Grey/Bruce, of increased relative risk of E. coli contamination (in black) within private drinking wells and the cattle dense region as identified by the orange circle. More specifically, regions in black represent clusters of E. coli contamination of private wells. Krolik et al. (2013) used SaTScan™ software to perform spatial analysis on multiple data sets of private well water submissions from PHO Kingston and across Ontario, visualized the spatial results, and did subsequent mapping on ArcMap version 10.0. Inner points were removed to clearly compare between clusters. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/wqrj.2019.035.

Figure 1

The four regions, (1) Kingston/Belleville, (2) Niagara/Hamilton, and (3) and (4) Grey/Bruce, of increased relative risk of E. coli contamination (in black) within private drinking wells and the cattle dense region as identified by the orange circle. More specifically, regions in black represent clusters of E. coli contamination of private wells. Krolik et al. (2013) used SaTScan™ software to perform spatial analysis on multiple data sets of private well water submissions from PHO Kingston and across Ontario, visualized the spatial results, and did subsequent mapping on ArcMap version 10.0. Inner points were removed to clearly compare between clusters. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/wqrj.2019.035.

MATERIALS AND METHODS

Sampling sites and collection

Private water well convenience samples submitted for bacteriological analysis between 1 June and 31 August 2014 to three PHO laboratories located in Hamilton, Kingston and London, Ontario, were studied. Well water samples were strategically selected from the three ‘hotspots’ for E. coli contamination, as previously described by Krolik et al. (2013). These hotspots will be referred to as the Grey/Bruce region, the Niagara/Hamilton region, and the Kingston/Belleville region, named for their approximate geographic locations. A region of high cattle density, located in mid-southwestern Ontario, was also included, given cattle are a major reservoir of enteric pathogens (Figure 1). Well water (200 mL) was collected using PHO sample bottles (BioNuclear Diagnostics, Inc., Toronto, Canada) containing 0.8 mM sodium thiosulfate. Each water sample was transferred in coolers, to a PHO laboratory, at approximately 2–8 °C and tested within 48 h of collection. Partial vacuum filtration was used to filter each sample (100 mL) through a 0.45 μm pore size mixed cellulose esters filter (Millipore, Billerica, MA, USA). Each filter was placed on a Differential Coliform Agar (Oxoid, Nepean, Canada) containing lactose and 5-bromo-4chloro-3-indolyl-d-glucuronide (BCIG), and incubated for 18–24 h at 36°C at atmospheric conditions. After incubation, the number of presumptive E. coli and TC colony forming units (CFUs) were counted and reported as CFU/100 mL. Water samples positive for E. coli (E. coli (+)) as well as samples positive for TCs, but negative for E. coli (TC-only), were chosen for further study. All E. coli and TCs were declared based on the phenotype on the differential agar only and considered presumptive E. coli and TCs for the purpose of this paper unless stated otherwise. For the purposes of water potability testing, E. coli and TCs are routinely reported as presumptive positive without further biochemical or molecular confirmation.

Nucleic acid extraction

The remaining 100 mL of E. coli (+) samples and TC-only samples underwent a second partial vacuum filtration step. The filters were rolled and placed into 3 mL cryovials (Simport, Beloeil, Canada), lysed with NucliSENSE® easyMAG® lysis buffer (2 mL) (Biomerieux Inc., St. Laurent, Canada), and placed in a rotating drum for 1 h at 37 °C. The DNA was extracted using MagSil beads and an automated NucliSENS® easyMAG® DNA extractor (Biomerieux Inc.) to a final volume of 100 μL. Extracts were stored at −20 °C until further processing.

Quantitative PCR assays for the detection of fecal indicators, sources, and bacterial pathogens

Extracted DNA underwent qPCR using TaqMan® chemistry for biomarkers of the indicator organisms and bacterial pathogens previously mentioned. The primers, probes, and internal amplification control (IAC) sequences of the assays are described in Table 1. Custom assays were modified or created for Campylobacter spp., Salmonella spp., and stx1/stx2 genes to avoid difficulties in adapting and validating pre-existing assays to the ViiA™ 7 Real-time PCR system (Applied Biosystems) and the unique matrix private water well samples represent. Existing assays were used as previously described to detect E. coli and Bacteroides as these had been validated in several pre-existing studies which examined these indicators in well water (Krolik et al. 2014, 2016). To confirm qPCR assay efficiency and sensitivity, all sequences underwent BLAST™ analysis and were tested against available strains of ATCC organisms as indicated in the subsection for each assay. To quantitatively detect these indicator and pathogenic organisms in water samples, standard curves were created from DNA extracted from laboratory strains of the organisms in question as per the DNeasy Blood & Tissue Handbook and kit (Qiagen 2006) protocol, that is, pretreatment for Gram-negative bacteria. All assays included a set of four standards assessed in duplicate to generate a standard curve and allow quantitative analysis of samples. A no template control in which TE buffer (Invitrogen, Carlsbad, CA, USA) was added, instead of DNA, was included in each assay. All qPCR assays had amplification and detection performed in a MicroAmp® optical 96-well reaction plate sealed with optical film (Applied Biosystems, Burlington, Canada) on the ViiA™ 7 Real-time PCR System. The run method followed 50°C for 2 min, 95°C for 10 min, then 40 cycles at 95 °C for 15 s and 60 °C for 1 min.

Table 1

Primers, probes and IAC sequences used in qPCR assays

Target organismPrimer/probeOligo sequence (5′ → 3′)References
Campylobacter spp. hipO-F
hipO-Pa,g
hipO-R
IAC-Pb,e
IAC-Target 
AATGCACAAATTTGCCTTATAAAAGC
ACATACTACTTCTTTATTGCTTGC
TICCATTAAAATTCTGACTTGCTAAATA
TTCGAAATGTCCGTTCGGTGGC
AATGCACAAATTTGCCTTATAAAAGCGTAT
TCGAAATGTCCGTTCGGTTGGCGCTATGAA
CAAGTCAGAATTTTAATCCIA 
Toplak et al. (2012)  
Salmonella spp. fimA-F
fimA-Pa,e
fimA-R
IAC-Pb,e
IAC-Target 
CACTAAATCCGCCGATCAAAC
CTGGGTCAATACCGTACCGCCAGC
AGTCGTATTACCAATCGCCGTAA
TTCGAAATGTCCGTTCGGTTGGC
CACTAAATCCGCCGATCAAACGTATTCGAA
ATGTCCGTTCGGTTGGCGCTATGAAGAGAT
ACGCGGTGGAACCTGGAATTACGGCGATTG
GTAATACGACT 
This study 
Shiga toxin-producing E. coli (STEC) — stx1 and stx2 toxin stx1-F
stx1-Pa,e
stx1-R
stx2-F
stx2-Pc,e
stx2-R
IAC-Pd,f
IAC-Target 
GGATAATTTGTTTGCAGTTGATGTC
CCGTAGATTATTAAACCGCCCTTCCTCTGGA
CAAATCCTGTCACATATAAATTATTTCGT
GGGCAGTTATTTTGCTGTGGA
ATGTCTATCAGGCGCGTTTTGACCATCTT
GAAAGTATTTGTTGCCGTATTAACGA
TTCGAAATGTCCGTTCGGTTGGC
GGGCAGTTATTTTGCTGTGGAGTATTCGAA
ATGTCCGTTCGGTTGGCGCTATGAAGAGAT
ACGCGGTGGAACCTGGAGTCGTTAATACGG
CAACAAATACTTTC 
This study 
Target organismPrimer/probeOligo sequence (5′ → 3′)References
Campylobacter spp. hipO-F
hipO-Pa,g
hipO-R
IAC-Pb,e
IAC-Target 
AATGCACAAATTTGCCTTATAAAAGC
ACATACTACTTCTTTATTGCTTGC
TICCATTAAAATTCTGACTTGCTAAATA
TTCGAAATGTCCGTTCGGTGGC
AATGCACAAATTTGCCTTATAAAAGCGTAT
TCGAAATGTCCGTTCGGTTGGCGCTATGAA
CAAGTCAGAATTTTAATCCIA 
Toplak et al. (2012)  
Salmonella spp. fimA-F
fimA-Pa,e
fimA-R
IAC-Pb,e
IAC-Target 
CACTAAATCCGCCGATCAAAC
CTGGGTCAATACCGTACCGCCAGC
AGTCGTATTACCAATCGCCGTAA
TTCGAAATGTCCGTTCGGTTGGC
CACTAAATCCGCCGATCAAACGTATTCGAA
ATGTCCGTTCGGTTGGCGCTATGAAGAGAT
ACGCGGTGGAACCTGGAATTACGGCGATTG
GTAATACGACT 
This study 
Shiga toxin-producing E. coli (STEC) — stx1 and stx2 toxin stx1-F
stx1-Pa,e
stx1-R
stx2-F
stx2-Pc,e
stx2-R
IAC-Pd,f
IAC-Target 
GGATAATTTGTTTGCAGTTGATGTC
CCGTAGATTATTAAACCGCCCTTCCTCTGGA
CAAATCCTGTCACATATAAATTATTTCGT
GGGCAGTTATTTTGCTGTGGA
ATGTCTATCAGGCGCGTTTTGACCATCTT
GAAAGTATTTGTTGCCGTATTAACGA
TTCGAAATGTCCGTTCGGTTGGC
GGGCAGTTATTTTGCTGTGGAGTATTCGAA
ATGTCCGTTCGGTTGGCGCTATGAAGAGAT
ACGCGGTGGAACCTGGAGTCGTTAATACGG
CAACAAATACTTTC 
This study 

F, forward primer; R, reverse primer; P, TaqMan® probe.

a6-FAM reporter.

bCAL Fluor Orange 560.

cCAL Fluor Gold 540.

dCAL Fluor Red 590.

eBHQ1 quencher.

fBHQ2 quencher.

gBHQplus quencher.

Detection of Campylobacter spp.

A TaqMan® qPCR assay targeting the hipO gene of Campylobacter was used to analyze samples for the presence of Campylobacter spp. Primers and probes were adapted from Toplak et al. (2012) with some modifications. Specifically, the hipO probe was chemically synthesized with proprietary modified nucleotides or BHQplus™ technology (Biosearch Technologies) in addition to a 3′-C nucleotide to increase the probe's Tm to 71.8 °C. A further modification to the assay was the addition of an IAC, which shares the same forward and reverse primer as the Campylobacter assay but relies on a different probe.

Seven standards were prepared from the extracted DNA of Campylobacter jejuni ATCC strain 33291 for the assay as a 10-fold serial dilution in 1× TE buffer, ranging from 1.0 × 106 to 1.0 × 100 gene copies of hipO per reaction, and tested in duplicate. The limit of detection (LOD) was approximately one gene copy per PCR, equivalent to one Campylobacter cell per reaction as hipO is a single copy gene (Leblanc-Maridor et al. 2011). This assay was tested against DNA extracted from the ATCC 33291 C. jejuni, as well as against five clinical isolates confirmed as C. jejuni by matrix-assisted laser desorption/ionization time-of-flight mass spectrometer (VITEK Mass Spectrometer, Biomerieux Inc.). Furthermore, six additional bacteria tested negative by qPCR, including Citrobacter freundii (ATCC 8090), Proteus mirabilis (ATCC 12453), E. coli (ATCC 25922), E. coli O157 (ATCC 35150), Salmonella enterica subsp. enterica serovar Typhimurium (ATCC 14028), and Salmonella enterica subsp. enterica serovar Enteritidis (ATCC 13076). Briefly, the reaction mixture (25 μL) consisted of TaqMan® environmental master mix 2.0 at 1× (Applied Biosystems), 400 nM of each forward and reverse primer (Biosearch Technologies), 100 nM of each hipO and IAC probe (Biosearch Technologies), 0.1 μL of the IAC target as a final copy number of 100, 5 μL of the DNA template, and 4.5 μL nuclease-free water.

Detection of Salmonella spp.

A TaqMan® qPCR assay targeting the fimA gene found ubiquitously in all Salmonella species was custom-designed to detect the presence of Salmonella in samples. Primer Express Software Version 3.0.1 (Applied Biosystems) was used to design primers and probes within the fimA gene, using default parameters. Seven standards were prepared from DNA extracted from Salmonella enterica subsp. enterica serovar Enteritidis (ATCC 13076) for the assay as a 10-fold serial dilution in 1× TE buffer, ranging from 1.0 × 106 to 1.0 × 100 gene copies of fimA per reaction, and tested in duplicate. The LOD was approximately one gene copy per PCR, equating to one Salmonella cell per reaction since fimA is a single copy gene (Zeiner et al. 2012). The Salmonella spp. assay was tested against Salmonella enterica subsp. enterica serovar Typhimurium and Salmonella enterica subsp. enterica serovar Enteritidis, both of which tested positive. Additionally, a number of non-Salmonella organisms tested negative including C. jejuni (ATCC 33291) and those aforementioned for the Campylobacter assay. The reaction mixture (30 μL) consisted of TaqMan® environmental master mix 2.0 at 1 × , 100 μM of each forward and reverse primer, 100 μM fimA and IAC probe, 0.1 μL of the IAC target as a final copy number of 100, 5 μL of DNA template, and 9.54 μL nuclease-free water.

Detection of STEC toxin genes, stx1 and stx2

A multiplex TaqMan® qPCR assay targeting the stx1 and stx2 genes was custom-designed to detect the presence of STEC. Primer Express Software Version 3.0.1 was used to design primers and probes within the stx1 and stx2 genes, using default parameters. Seven standards were prepared (from DNA extracted from E. coli O157 ATCC 35150) as a 10-fold serial dilution in 1× TE buffer, ranging from 1 × 106 to 1 × 100 gene copies of both stx1 and stx2 per reaction. The LOD was approximately one gene copy of each stx1 and stx2 per PCR. This multiplex assay was tested against a laboratory strain of E. coli O157 to ensure specificity to stx1/stx2 toxin genes. Additionally, a number of non-stx1/stx2-producing organisms tested negative including a laboratory strain of non-STEC E. coli (E. coli ATCC 25922), Enterococcus faecalis (ATCC 29212), and those previously mentioned for the Campylobacter assay. It should be noted that the stx1 assay is unable to distinguish between the stx1 gene of Shigella and STEC; however, the contamination of well water by Shigella is highly unlikely, given humans are the only animal reservoir and human cases remain rare in Ontario. Accordingly, dissemination of Shigella into private water wells in Ontario is not anticipated, as would be expected for pathogens such as Campylobacter and STEC, the reservoirs of which include livestock (PHO 2018). The reaction mixture (25 μL) consisted of TaqMan® environmental master mix 2.0 at 1× , 100 μM each forward and reverse primer, 100 μM each stx1, stx2, and IAC probe, 0.1 μL of the IAC target as a final copy number of 100, 5 μL of the DNA template, and 1.9 μL nuclease-free water.

Detection of E. coli markers

A qPCR assay targeting the E. coli 23S rRNA gene sequence, adapted from Chern et al. (2011), was used to detect E. coli in samples. Standards were prepared using genomic DNA from ATCC strain 25922 and ranged from 1 × 106 to 1 × 100 cells per reaction. Subsequently, E. coli standards, to a limit of one gene copy per reaction, were assessed independently and an LOD of 2.5 gene copies per reaction or 7 cells/100 mL was repeatedly and accurately detected. The qPCR mixture and conditions are the same as detailed by Krolik et al. (2016).

Detection of Bacteroides 16S rRNA genetic markers

All samples underwent three qPCR assays for the detection of the total, human, and bovine-specific Bacteroides 16S rRNA genetic markers in order to reveal information about the source of contamination. The total Bacteroides assay targets a conserved region for 10 host-specific species including feline, bovine, deer, canine, anserine, gull, equine, human, swine, and raccoon. These MST assays were based on earlier work by Lee et al. (2010) for surface water, and Krolik et al. (2014), for private water wells and primers, probes, and reaction conditions, have been previously described. Krolik et al. (2014) validated an adjusted LOD of 10 gene copies for all three singleplex assays. Standard curves for the MST assays were created as previously described using plasmids with concentrations ranging from 6.5 × 106 to 6.5 × 100 gene copies per reaction.

Statistical analyses

The distribution of all measured variables was analyzed with histograms and goodness-of-fit tests for normality, including Kolmogorov–Smirnov, Cramer–von Mises, and Anderson–Darling tests to inform the subsequent analysis. Data sets were analyzed using Pearson's chi-squared tests to determine if an association existed between a well location and the source of fecal contamination. Finally, data sets were analyzed using Spearman rank correlation tests to evaluate the relationship of microbial indicators, E. coli and TCs, and microbial source indicators, Bacteroides 16S rRNA, to each other and to pathogen detection. Spearman correlation coefficients were interpreted based on the guidelines detailed in Table 2 (Evans 1996). All statistical analyses were performed in SAS 9.3 at a level of significance of α = 0.05.

Table 2

Interpretation of correlation coefficients for this study (Evans 1996)

Correlation coefficientInterpretation of relationship
0.2–0.39 ‘Weak’ 
0.4–0.69 ‘Moderate’ 
0.7–0.1 ‘Strong’ 
Correlation coefficientInterpretation of relationship
0.2–0.39 ‘Weak’ 
0.4–0.69 ‘Moderate’ 
0.7–0.1 ‘Strong’ 

RESULTS

Sample collection overview

In total, 557 private drinking water samples, 314 E. coli-positive and 243 TC-only, were obtained during the collection period. Of these, 96% (297 E. coli and 236 TC-only) had sufficient well location information enabling geocoding to a source well within one of the four regions. Of these, 74% (205 E. coli (+) and 190 TC-only) were associated with a unique source well. The remaining samples (92 E. coli (+) and 46 TC-only) represented 61 unique source wells (40 E. coli and 21 TC-only), indicating wells that were tested more than once over the study period. In terms of testing and statistical analyses, samples from wells that submitted more than once were treated the same as samples from unique source wells. Some samples could not be successfully extracted or tested for a final count of 519 samples (289 E. coli (+) and 230 TC-only) (Table 3).

Table 3

Total samples organized by the ‘hotspot’ location

HotspotE. coli (+) samplesTC-only samples
Grey/Bruce 132 45 
Niagara/Hamilton 68 77 
High Cattle 62 88 
Kingston/Belleville 27 20 
Total 289 230 
HotspotE. coli (+) samplesTC-only samples
Grey/Bruce 132 45 
Niagara/Hamilton 68 77 
High Cattle 62 88 
Kingston/Belleville 27 20 
Total 289 230 

Prevalence of enteric pathogens

The evidence of enteric pathogens was detected in 10 (1.9%) drinking water samples (Table 4). A genetic marker of Campylobacter, namely the hipO gene, was detected in five samples, and the stx1 gene was detected in the remaining five. Genetic markers of Salmonella spp. were not detected. At least one pathogen-positive sample came from each of the four regions of interest (Figure 2), suggesting no particular geographic trend for pathogen presence, although the sample size is small and no firm conclusions can be made. It should be noted that samples 9 and 10 were associated with the same source well and collected on the same day. Of the 10 samples, only two were from the E. coli (+) sample set, with the remaining eight being negative for culturable E. coli and positive for TCs. The measured variables associated with these samples, E. coli and TC culture results, E. coli qPCR, total, human, and bovine-specific Bacteroides 16S rRNA genetic markers do not appear to follow any particular trends. However, it should be noted that eight of the 10 samples were positive for at least one indicator of fecal contamination, either in the form of culturable E. coli and/or E. coli or Bacteroides detected by qPCR.

Table 4

Full profile of private water well samples that demonstrated molecular evidence of targets of interest

SampleTarget of interestTC (CFU/100 mL)E. coli (CFU/100 mL)E. coli gene copies and cell equivalents/100 mL qPCRHuman Bacteroides 16S rRNA markersBovine Bacteroides 16S rRNA markersTotal Bacteroides 16S rRNA markersHotspot
hip0/0 High cattle 
hip64 22.4/3.2 21 50 K/B 
hip0/0 15 High cattle 
hip0/0 G/B 
hip81 45 140/20 8,498 44 37,515 G/B 
stx81 9.1/1.3 High cattle 
stx0/0 1,333 N/H 
stx81 29.4/4.2 1,740 11,890 G/B 
stx22 79.8/11.4 N/H 
10 stx20 130.2/18.6 17 N/H 
SampleTarget of interestTC (CFU/100 mL)E. coli (CFU/100 mL)E. coli gene copies and cell equivalents/100 mL qPCRHuman Bacteroides 16S rRNA markersBovine Bacteroides 16S rRNA markersTotal Bacteroides 16S rRNA markersHotspot
hip0/0 High cattle 
hip64 22.4/3.2 21 50 K/B 
hip0/0 15 High cattle 
hip0/0 G/B 
hip81 45 140/20 8,498 44 37,515 G/B 
stx81 9.1/1.3 High cattle 
stx0/0 1,333 N/H 
stx81 29.4/4.2 1,740 11,890 G/B 
stx22 79.8/11.4 N/H 
10 stx20 130.2/18.6 17 N/H 
Figure 2

Location of private water wells that were associated with a water sample positive for a target of interest in each of the ‘hotspot’ regions (shaded in gray).

Figure 2

Location of private water wells that were associated with a water sample positive for a target of interest in each of the ‘hotspot’ regions (shaded in gray).

Microbial source tracking using B. fragilis

Samples were tested via three MST qPCR assays in order to determine the pre-eminent source of fecal contamination. The following MST statuses were assigned to each sample based on the established gene copy thresholds: negative for all Bacteroides (none), positive for total Bacteroides only (not human or bovine), positive for human Bacteroides, positive for bovine Bacteroides, and positive for both human and bovine Bacteroides. Among the E. coli (+) samples, a higher proportion tested positive for human-specific contamination (41%) compared with the TC-only samples (22%) (Figure 3). The TC-only group had a higher proportion of samples negative for all Bacteroides genetic markers (50%) compared with the E. coli (+) sample set (23%).

Figure 3

Distribution of microbial sources arranged for E. coli (+) and TC-only samples.

Figure 3

Distribution of microbial sources arranged for E. coli (+) and TC-only samples.

Prior work by Krolik et al. (2016) found that the same E. coli (+) samples assessed in this study were highly contaminated with human-specific contamination, especially within the Kingston/Belleville (2012 data) and Grey/Bruce regions (56% and 53%, respectively). A Chi-square test and post-hoc comparisons revealed that these two regions had significantly more wells positive for human-sourced contamination compared with the Niagara/Hamilton and High Cattle regions. The latter two regions appeared to have a higher number of wells with only total Bacteroides genetic markers detected (38% and 37%, respectively).

In addition to the E. coli (+) samples, the TC-only samples, from the same four regions, were investigated for sources of fecal contamination (Figure 4). Fewer samples were positive for human contamination than for E. coli (+) samples, and a chi-squared test determined there was no statistically significant association between the region and human-specific contamination (χ2 = 0.62, p = 0.89). The Grey/Bruce and High Cattle regions had a higher proportion of samples negative for all Bacteroides genetic markers (63% and 57%, respectively). In contrast, the Kingston/Belleville and Niagara/Hamilton regions had a lower proportion of samples with no Bacteroides genetic markers detected (30% and 39%, respectively) and a higher proportion of samples with only total Bacteroides (40% and 31%, respectively). There was a statistically significant association between regions sampled and the proportion of samples with no Bacteroides genetic markers detected (χ2 = 12.0, p < 0.01).

Figure 4

Distribution of microbial sources in TC-only samples stratified by the region. K/B, Kingston/Belleville; G/B, Grey/Bruce; N/H, Niagara/Hamilton.

Figure 4

Distribution of microbial sources in TC-only samples stratified by the region. K/B, Kingston/Belleville; G/B, Grey/Bruce; N/H, Niagara/Hamilton.

All regions had a low percentage of samples contaminated by a bovine fecal source. The High Cattle region had the lowest proportion of E. coli culture (+) samples with bovine-specific Bacteroides genetic marker contamination (2%), and only 5% of TC-only samples in the High Cattle region were positive for bovine-specific contamination. Due to the low amount of observed bovine-specific contamination, not enough data were available to statistically test association with the region, nor its relationship with a number of other measured variables.

Correlation between alternative and traditional indicators

Total Bacteroides versus E. coli

All measured variables, including E. coli CFU/100 mL, TC CFU/100 mL, E. coli cells as detected via qPCR, and total, human, and bovine-specific Bacteroides 16S rRNA genetic markers, showed non-normal skewed distributions. Therefore, Spearman rank correlations were used to analyze the relationship between variables. When data from all wells were aggregated, there was a weak but significant correlation (ρ = 0.20, p < 0.01) between total Bacteroides genetic markers and culturable E. coli (CFU/100 mL). There was also a weak but significant correlation (ρ = 0.34, p < 0.001) between total Bacteroides genetic markers and E. coli cells/100 mL as detected by qPCR.

Human-specific Bacteroides versus E. coli and TCs

No significant correlation was found between human-specific Bacteroides genetic markers and culturable E. coli (CFU/100 mL) (ρ = 0.16, p = 0.07). However, there was a significant correlation overall (ρ = 0.3, p < 0.05) when investigating the relationship between human-specific Bacteroides genetic markers and E. coli cells/100 mL as detected by qPCR.

TCs and Bacteroides

There was no significant correlation between total or human-specific Bacteroides genetic markers and TC CFU/100 mL in the TC-only sample set (data not shown).

DISCUSSION

Enteric pathogens in private drinking water wells

It was hypothesized that water from private drinking wells in previously identified high-risk regions would be vulnerable to contamination by enteric pathogens. This study explored the value of 100 mL convenience samples as a potential resource for on-going water surveillance purposes and in the context of previously established indicators of drinking water potability, namely E. coli and TCs. In total, 10 (1.9%) samples were positive for targets of interest by qPCR, including five for the hipO gene specific to Campylobacter spp. and five for the stx1 gene that encodes the Stx1 toxin produced by STEC and Shigella spp. Although low, these detections are significant, given this study was limited by a small volume of water available for testing. Moreover, for E. coli (+) samples, it was determined, via qPCR, that the human-specific Bacteroides genetic marker was detected more frequently than non-human-specific Bacteroides markers, indicating that human feces and/or sewage was the primary contaminating source (Figure 3). Although humans are capable of becoming ill with Campylobacter, Salmonella, and E. coli O157, these diseases remain relatively rare occurrences (PHO 2018), and humans are not considered natural reservoirs of these pathogens unlike food-producing animals (e.g. cattle and chickens). As such, it is not surprising that these zoonotic enteric pathogens for which this study surveyed were rarely detected. Furthermore, Campylobacter spp. and Salmonella spp. have a wide range of animal hosts (i.e., wild life, livestock, and domestic pets), and some them are outside of that which the Bacteroides assay can detect. In particular, birds are a notable source of both pathogens, and although the total Bacteroides assay can detect gull and anserine contamination, there is a gap in poultry sources. Additionally, only a very small subset of samples were positive for bovine-specific contamination (6%), and as cattle are the primary reservoir of STEC, this finding may, in part, explain the low number of stx1/stx2 detections.

Although there is vast literature indicating that surface water, wastewater, and sewage effluent are often contaminated with pathogens, there is relatively less evidence for private groundwater supplies (Hörman et al. 2004; Walters et al. 2007; Viau et al. 2011). Won et al. (2013) showed that in private wells proximal to bovine operations in northeastern Ohio, Campylobacter spp. DNA was not detected by qPCR in any samples (2 L filtered) and E. coli O157:H7 by qPCR was detected in just 4% of samples. Similarly, Schets et al. (2005) tested drinking water from private water supplies in the Netherlands and found that E. coli O157:H7 was isolated from just 2.7% of samples. Investigation of the microbial quality of private groundwater wells generally requires large volume samples (e.g., greater than 1 L) relative to less protected sources, such as surface or wastewater. Hänninen et al. (2003) studied waterborne outbreaks in Finland caused by C. jejuni in private groundwater supplies and found that the use of large sample volumes of several liters, versus the ≤1 L proposed by the International Standardization Organization (2002) for the detection of Campylobacter from drinking water, increased the chance of pathogen detection. Schets et al. (2005) discovered that E. coli O157, as detected by qPCR, was isolated more frequently from 1 L versus 100 mL sample volumes of private groundwater drinking wells.

Statistical analyses were not performed to assess the predictive value of fecal indicators as it relates to pathogen presence as only a few samples tested positive for targets of interest. Overall, eight of 10 samples in which pathogen presence was detected showed the evidence of fecal contamination, either E. coli (by culture or qPCR) or Bacteroides 16S rRNA genetic markers. However, only two were positive for culturable E. coli, with the remaining positive for culturable TC-only, suggesting that TCs remain a relevant and useful indicator of water quality. Furthermore, only four samples with the molecular evidence of pathogens were positive for E. coli by qPCR when applying the LOD of seven E. coli cells/100 mL. Numerous studies provide evidence that E. coli does not always accurately predict the presence of pathogens within aquatic environments. Won et al. (2013) found that in the samples positive for E. coli O157 by qPCR, only one out of seven was also positive for E. coli by a commercial MPN test. Wu et al. (2011) assessed 540 cases representing independent indicator–pathogen correlations from 1970 to 2009 and found when E. coli was used as the indicator organism, only 25% of cases were correlated to pathogen presence.

Microbial fecal sources of contamination in Ontario E. coli hotspot well waters

Relative to TC-only samples, a higher percentage of E. coli (+) samples (by culture) were also contaminated with Bacteroides 16S rRNA genetic markers (77% versus 50%). Interestingly, 23% of these E. coli (+) samples were negative for Bacteroides genetic markers. This apparent discrepancy may be consequent to the fact that: (1) E. coli detection is via a membrane filtration method followed by culture on differential agar, whereby E. coli are identified based on phenotype alone (color change), and not subject to further confirmatory assays. This aligns with the current testing practices for the presumptive identification of E. coli in drinking water in Ontario, (2) these samples may have fecal contamination in the form of Bacteroides that is attributable to hosts other than the 10 sources detected by the total Bacteroides assay, and (3) the Bacteroides qPCR assays may have been less sensitive for small volume water samples especially in instances where E. coli CFU/100 mL counts were low. Similarly, Krolik et al. (2014) showed that for those private drinking water samples that were positive for culturable E. coli, 23.6% were negative for all three Bacteroides genetic markers. Although the TC-only samples had less total Bacteroides contamination than E. coli (+) samples, 50% were still positive for Bacteroides. Many of these samples, although negative for culturable E. coli, were positive for E. coli by qPCR, suggesting that the samples were fecally contaminated at one time with organisms that are either non-viable or viable but non-culturable (VBNC) (Oliver 2010).

TC-only sample analysis revealed that all regions, with the exception of Kingston/Belleville, had a high proportion of TC-only samples that were negative for the presence of Bacteroides genetic markers. The Kingston/Belleville region included a small data set and was excluded for this particular analysis. The Grey/Bruce and High Cattle regions have a significantly higher percentage of samples with no Bacteroides genetic markers detected compared with the Niagara/Hamilton region. This may suggest that wells with TC-only contamination in these two regions are less susceptible to future or past fecal contamination compared with TC-only wells from the Niagara/Hamilton region. This may be due to well integrity and structure, frequent testing and treatment, and appropriate distance from septic tanks as well as different characteristics of the hydrogeological system (Tallon et al. 2005).

Relationship of an alternative to traditional fecal indicators

Given the low prevalence of E. coli (by culture and qPCR) in samples positive for genetic markers of enteric pathogens, the relationship between E. coli and the alternative indicator, Bacteroides, was investigated. A weak but significant correlation between total Bacteroides and culturable E. coli was observed, which is supported by similar investigations. Lee et al. (2010) determined that total Bacteroides exhibited a weak correlation to E. coli (r2 = 0.22) in Ontario surface water samples. However, the relationship between total Bacteroides genetic markers and E. coli (qPCR) was stronger than to culturable E. coli, a trend that has been previously documented for other water sources, but rarely for private drinking well waters. Sauer et al. (2011) saw a stronger correlation of E. coli as detected by qPCR to total Bacteroides (ρ = 0.464) than to culturable E. coli (ρ = 0.336) in Wisconsin stormwater samples.

It is known that the persistence of E. coli differs from that of Bacteroides markers in environmental waters and is highly dependent on various factors. Okabe & Shimazu (2007) demonstrated that decay rates for the Bacteroides genes were greater at high temperatures and lower between 4 and 12 °C, whereas TC and fecal coliform (FC) decay rates, measured by culture, were lower at temperatures above 10 °C and higher at 4 °C. Further differences may be explained by the different methods of detection, that is, culture for E. coli and qPCR for Bacteroides genetic markers. Culture will only detect viable cells, whereas qPCR detects viable cells, VBNC and DNA from dead cells. Furthermore, the culture method used in this study is based upon the phenotypic identification of E. coli colonies. Although the detection of β-d-glucuronidase activity is generally a good indicator of E. coli, this enzyme has been reported in other organisms, including Salmonella, Shigella, Bacteroides, and Clostridium (Cabral 2010). Therefore, culture tests may be accounting for more or less E. coli than is actually present, affecting the correlation with total Bacteroides genetic markers.

Human-specific Bacteroides genetic markers and E. coli as detected by qPCR, but not by culture, were significantly correlated. Sauer et al. (2011) showed that although total Bacteroides genetic markers were correlated with E. coli (culture and qPCR), human-specific markers exhibited non-significant correlation with culturable E. coli and Enterococci and weak correlation with E. coli by qPCR (ρ = 0.16). Okabe et al. (2007) demonstrated that in river water samples, total Bacteroides genetic markers showed a moderate level of correlation (r2 = 0.49) with FC, but human-specific Bacteroides gene markers showed no significant correlation with FC (r2 = 0.12) which was thought to be a consequence of multiple sources of fecal contaminants other than human.

The weak correlation of E. coli to human-specific Bacteroides genetic markers is partially due to the inherent differences between traditional culture methods and qPCR and the differences in survival of E. coli to Bacteroides as mentioned above. However, it is also thought that the strength of correlation may be weak because E. coli is not indicative of human contamination specifically. Strains of E. coli are prominent within the GIT of the majority of warm-blooded animals. Therefore, a number of sources besides humans, including pets, domestic livestock, and various wildlife could be responsible for E. coli contamination in private water wells.

Total and human-specific Bacteroides 16S rRNA genetic marker analyses in the context of correlation to TCs showed non-significant correlation. This was to be expected given TCs are not strictly a fecal indicator but also an indicator of environmental contamination and demonstrates well vulnerability.

CONCLUSIONS

This study aimed to investigate the feasibility of using private water well samples, submitted to the provincial laboratory for bacterial testing (convenience samples) for the purpose of pathogen surveillance, and to examine the findings in the context of various markers of fecal contamination for select enteric pathogens. Accordingly, the relationship between well location, fecal indicators, and pathogen presence was assessed. This study did not explore other drivers of well water quality, such as well screen depth, well age, depth to water, land use, septic system density, geology, or others. Current studies in our laboratory are examining these potential drivers of contamination for private water wells in Ontario. The detection of genes of various enteric pathogens was low and may, in part, be explained by small water volumes, but our findings, and those of others, might further be explained by the fact that human, not bovine, fecal contamination is responsible for the majority of private water well contamination in the regions studied. Future studies should consider including a similar assessment, but expanding the pathogens investigated to those most frequently associated with human fecal sources (e.g. norovirus) and with a comparative analysis to larger volumes in order to explore the limit of detection in the context of water volume as a future aim. If the results are comparable, this would further validate the use of small volume waters as adequate for pathogen monitoring.

Given that private drinking water samples are frequently impacted by fecal contamination, especially from humans, attention must be paid to improving barriers for fecal transport between wells, septic, or household waste management systems, and the environment. The evidence of both Campylobacter and STEC contamination was found, albeit in few samples, and is typically associated with cattle or livestock, demonstrating that multiple sources need to be considered and addressed, even though infrequently. The presence of target DNA for fecal markers and pathogens also suggests recent or past contamination, confirming the vulnerability of the well to contamination by these organisms.

In developed nations, clean and safe drinking water is the expected standard, and the presence of enteric pathogens is considered unacceptable (Safe Water Drinking Act 2002; WHO 2011). Many enteric pathogens have a very low infectious dose and if ingested may cause infection, especially in immunocompromised individuals or those persons with first exposure (naïve) to the drinking water source; for example, visitors.

This study further demonstrates that fecal indicators, in addition to E. coli, and detection methods, beyond culture, could be considered both at the laboratory and the regulatory level, for private water well testing in Ontario. In this investigation, E. coli did not accurately predict the presence of enteric bacterial pathogens within those samples tested, and the majority of samples with enteric pathogen presence had TC contamination only. Furthermore, E. coli, as detected by standard culture methodology, did not strongly correlate with other indicators of fecal contamination, notably species-specific Bacteroides. However, a larger scale study looking at additional regions of interest, with larger sample size, against which these findings could be compared could be considered in future as a validation step.

This study and future-related studies serve to inform water testing and maintenance practices for both the private water well owners and source drinking water stewards, as well as further enhance the understanding of well water quality implications and impacts on humans, animals, and the environment, from both the public health and policy perspectives.

ACKNOWLEDGEMENTS

The authors thank the employees from the Kingston, Hamilton and London Public Health Ontario laboratory locations for assisting with the collection of water samples. A special thanks to Lewis Tomalty and Gerald Evans for their contributions and support. The authors have no conflicts of interest to declare.

REFERENCES

REFERENCES
Cabral
J. P.
2010
Water microbiology. bacterial pathogens and water
.
Int. J. Environ. Res. Public Health
7
(
10
),
3657
3703
.
Charrois
J. W.
2010
Private drinking water supplies: challenges for public health
.
Can. Med. Assoc. J.
182
(
10
),
1061
1064
.
Chern
E. C.
,
Siefring
S.
,
Paar
J.
,
Doolittle
M.
,
Haugland
R. A.
2011
Comparison of quantitative PCR assays for Escherichia coli targeting ribosomal RNA and single copy genes
.
Lett. Appl. Microbiol.
52
(
3
),
298
306
.
Evans
J. D.
1996
Straightforward Statistics for the Behavioural Sciences
.
Brooks/Cole Publishing
,
Pacific Grove, CA
,
USA
.
Figueras
M. J.
,
Borrego
J. J.
2010
New perspectives in monitoring drinking water microbial quality
.
Int. J. Environ. Res. Public Health
7
(
12
),
4179
4202
.
Hänninen
M. L.
,
Haajanen
H.
,
Pummi
T.
,
Wermundsen
K.
,
Katila
M. L.
,
Sarkkinen
H.
,
Miettinen
I.
,
Rautelin
H.
2003
Detection and typing of Campylobacter jejuni and Campylobacter coli and analysis of indicator organisms in three waterborne outbreaks in Finland
.
Appl. Environ. Microbiol.
69
(
3
),
1391
1396
.
Hörman
A.
,
Rimhanen-Finne
R.
,
Maunula
L.
,
Bonsdorff
C. H.
,
Torvela
N.
,
Heikinheimo
A.
,
Hänninen
M. L.
2004
Campylobacter spp., Giardia spp., Cryptosporidium spp., Noroviruses, and Indicator organisms in surface water in Southwestern Finland, 2000–2001
.
Appl. Environ. Microbiol.
70
(
1
),
87
95
.
Hrudey
S. E.
,
Hrudey
E. J.
,
Pollard
S. J.
2006
Risk management for assuring safe drinking water
.
Environ. Int.
32
(
8
),
948
957
.
International Standardisation Organisation (ISO)
2002
Water Quality – Detection and Enumeration of Thermophilic Campylobacter
.
Committee Draft ISO/CD 17995
.
International Standardisation Organisation
,
Geneva
,
Switzerland
.
Krentz
C. A.
,
Prystajecky
N.
,
Isaac-Renton
J.
2013
Identification of fecal contamination sources in water using host-associated markers
.
Can. J. Microbiol.
59
(
3
),
210
220
.
Krolik
J.
,
Maier
A.
,
Evans
G.
,
Belanger
P.
,
Hall
G.
,
Joyce
A.
,
Majury
A.
2013
A spatial analysis of private well water Escherichia coli contamination in southern Ontario
.
Geospat. Health
8
(
1
),
65
75
.
Krolik
J.
,
Evans
G.
,
Belanger
P.
,
Maier
A.
,
Hall
G.
,
Joyce
A.
,
Guimont
S.
,
Pelot
A.
,
Majury
A.
2014
Microbial source-tracking analysis of E. coli contaminated private well waters in southeastern Ontario
.
J. Water Health
12
(
2
),
348
357
.
Office of the Auditor General of Ontario
2014
Annual Report, Section 3.12 Source Water Protection
.
(accessed June 2017).
Public Health Ontario (PHO)
2018
Monthly Infectious Diseases Surveillance Report, Reportable Disease Cases by Month, 2017
.
(accessed June 2019)
.
Safe Water Drinking Act
2002
S.O. 2002, c. 32. Available from: https://www.ontario.ca/laws/statute/02s32 (accessed June 2016)
.
Saxena
G.
,
Bharagava
R. N.
,
Kaithwas
G.
,
Raj
A.
2015
Microbial indicators, pathogens and methods for their monitoring in water environment
.
J. Water Health
13
(
2
),
319
339
.
Schets
F. M.
,
During
M.
,
Italiaander
R.
,
Heijnen
L.
,
Ritjes
S. A.
,
van der Zwaluw
W. K.
,
Husman
A. M. R.
2005
Escherichia coli o157:H7 in drinking water from private water supplies in the Netherlands
.
Water Res.
39
(
18
),
4485
4493
.
Statistics Canada
2011
Census of Agriculture, Farm and Farm Operator Data-95-640-XWE
.
Available from: http://www.statcan.gc.ca/eng/ca2011/index (accessed June 2017).
Tallon
P.
,
Magajna
B.
,
Lofranco
C.
,
Leung
K. T.
2005
Microbial indicators of fecal contamination in water: a current perspective
.
Water Air Soil Pollut.
166
(
1
),
139
166
.
Toplak
N.
,
Kovač
M.
,
Piskernik
S.
,
Možina
S. S.
,
Jeršek
B.
2012
Detection and quantification of Campylobacter jejuni and Campylobacter coli using real-time multiplex PCR
.
J. Appl. Microbiol.
112
(
4
),
752
764
.
Viau
E. J.
,
Goddwin
K. D.
,
Yamahara
K. M.
,
Layton
B. A.
,
Sassoubre
L. M.
,
Burns
S. L.
,
Tong
H. I.
,
Wong
S. H. C.
,
Lu
Y.
,
Boehm
A. B.
2011
Bacterial pathogens in Hawaiian coastal streams-associations with fecal indicators, land cover, and water quality
.
Water Res.
45
(
11
),
3279
3290
.
World Health Organization (WHO)
2011
Guidelines for Drinking-Water Quality
, 4th edn.
World Health Organization
,
Geneva
,
Switzerland
.
(accessed June 2016).