Thesis (M.S.)--Worcester Polytechnic Institute. / Keywords: classification; k-nearest-neighbor (k-n-n); neural networks; linear discriminant analysis (LDA); support vector machines; microbial source tracking (MST); quadratic discriminant analysis (QDA); logistic regression. Includes bibliographical references (p. 59-61).

Residents in rural communities across Canada rely on groundwater as their main drinking water source, but the private maintenance of this source may increase the risk of fecal contamination caused by human or animal wastes. Wainfleet, a rural Ontario community, has been under an active boil water advisory for the past decade. The last study to assess groundwater quality, performed in 2007, determined that half of the 586 groundwater wells contained exceedances in total fecal coliform and E. coli counts. A critical examination of fecal contamination levels and its sources is not only necessary for maintaining public health in the township, but is also an opportunity to examine the robustness of culture-independent methods for quantifying and sourcing fecal contamination in groundwater environments across Canada. For this project, culture-based and culture-independent methods were utilized to quantify and source any fecal contaminants in Wainfleet’s groundwater. Culture counts of fecal indicator bacteria (FIB) suggested that some of the groundwater wells were receiving more fecal contamination than others, as expected based on a previous study that was conducted 10 years prior. The groundwater wells with higher E. coli counts also had higher read counts of microbes like Campylobacterales which could come from septic tanks and higher concentrations of oxidized nitrogen which can also indicate human-based fecal contamination. Finally, fecal contamination in groundwater wells with E. coli tested positive for the human Bacteroidales marker. Taken together, this study shows that fecal contamination pervades groundwater wells across the boil water advisory zone, much of which originates from leaking septic tanks and poorly-constructed groundwater wells. In this study, we have shown that a suite of protocols, from physiochemical quantification to targeted sequencing and qPCR, can be used to complement culture-based assays in quantifying and pinpointing fecal contamination in groundwater sources. / Thesis / Master of Science (MSc) / Boil water advisories are enacted when fecal contamination levels exceed provincial limits. Standard methods for quantifying fecal contamination use the culture-based detection of fecal indicator bacteria. Sequencing the 16S rRNA gene and amplifying Bacteroidales markers can also be used to identify novel fecal markers and quantify host-specific contaminants in source waters. Using culture and genetic-based methods determined that groundwater wells across Wainfleet, a Niagara township with the longest active boil water advisory in Canada, contain septic tank microbes and are primarily contaminated by leaking septic tanks. Genetic-based assays can complement culture-based detection of fecal bacteria in groundwater sources across Canada.

Water sources are susceptible to faecal contamination from animal and human pollution sources. Pollution of our waterways has significant implications on human health, especially from a pathogen perspective. Microbial source tracking (MST) is a promising field which aims to identify the sources of faecal contamination, and thereby allowing for the development of effective management strategies to minimise pollution and the impact on human health. Many of the currently used methods rely on the identification of host-specific markers within the 16S ribosomal RNA (rRNA) gene of bacteria by use of amplification techniques such as polymerase chain reaction (PCR). However, these methods can be limited by sensitivity, quantification, geographical differences and issues of cost which can limit how many markers are evaluated. Developments in DNA sequencing technologies over the past decade have led to a number of next generation sequencing (NGS) platforms which have a rapid, high throughput approach, resulting in an exponential decrease in the cost of sequencing. This has enabled the development of sequence-based metagenomics, where entire communities from environmental samples can be analysed based on their genetic material. The ability to barcode allows for analysis of multiple samples at once, reducing the cost of sequencing environmental samples even further. This is a promising technique for MST, which has had little investigation to date. The primary focus of the studies described in this thesis was to evaluate the use of NGS technology through a metagenomic approach. Roche 454 amplicon sequencing was used to sequence a 16S rRNA gene target, amplified from faecal and water samples from various sources in New Zealand. Barcode strategies were incorporated in the amplification design to allow multiple samples to be sequenced simultaneously. A proof-of-concept study initially utilised a small sequence dataset to evaluate a range of analysis tools available. Taxonomic identification and diversity measures were used to evaluate a selection of currently available tools designed for analysing metagenomic data, with the Quantitative Insights Into Microbial Ecology (QIIME) platform decided upon for further studies. A larger study, including 35 faecal samples from 13 difference sources and 10 water samples, resulted in 522,065 raw sequencing reads. Diversity results suggest three phyla, Bacteroidetes, Firmicutes and Proteobacteria, are strongly represented across all faecal sources analysed. Microbial diversity analysis using clustering techniques provided evidence of host source being the largest influence on bacterial diversity, with samples from each source generally clustering together. This technique could not be used to identify sources of contamination sources in water samples as the water samples all clustered separately from the faecal samples. More successful was the use of taxonomic classifications to determine bacteria genera that were potentially specific to one source. Water samples were screened for these genera, with six out of the ten water samples being indicators of either ruminant or human contamination. Faecal and water samples were also analysed for a selection of published 16S rRNA PCR markers, using a computational motif-based search method. Of the twenty motifs screened for, 14 were found to be relatively source-specific for ruminant, human, dog or pig faecal samples, with some cross-reactivity with chicken and possum samples. Using this method, the contamination source for six of the ten water samples was identified, with the remaining four samples found to not have enough sequences to assess with confidence. Both metagenomic strategies produced comparable results which were consistent with previous MST analysis. This project demonstrates the potential application of next generation sequencing technologies to microbial source tracking, suggesting the possibility this approach to replace existing microbial source tracking methods.

In Canada, as in many other affluent nations, private well water consumers remain at risk for gastrointestinal (GI) illness due to fecal contamination of groundwater. There have been numerous documented outbreaks of GI illness related to contaminated drinking water. While the general risk to well water consumers has been established, the risk in southern Ontario is poorly understood. As a preliminary step towards understanding this risk, a study of Escherichia coli (E. coli) contamination in private well water was undertaken. Spatial scan statistics were employed to determine the extent of contamination for over 30,000 private wells in southeastern Ontario between 2008 and 2012, inclusive. This analysis revealed one large, temporally stable elevated risk region, and three significant smaller regions within it. The methodology utilized in the primary investigation was then applied to a 2012 dataset for all of southern Ontario, resulting in the identification of three regions of elevated risk. The presence of E. coli, a traditional fecal indicator organism, indicates lack of water potability. To provide knowledge regarding the origins of fecal contamination in southeastern Ontario, a molecular microbial source tracking (MST) study was undertaken. A quantitative real-time Bacteroidales PCR assay specifically targeting human, bovine, and general (specific to 10 hosts) was optimized and applied to 716 private well water samples. Almost half of the samples showed evidence of human fecal contamination, whereas only 13% contained evidence of bovine fecal contamination. Approximately one quarter of well samples tested positive for the general host Bacteroidales assay, with an additional one quarter testing negative for all MST assays. Additionally, spatial scan statistics revealed a region of human-sourced contamination, which geographically corresponded with the E. coli contamination cluster for the same study year. The presence of E. coli contamination clusters among private wells reveals an at-risk group of well water consumers. As such, public health practitioners may use this information to target well stewardship programs in higher risk regions. Humans were the predominant contributors of fecal contamination to private wells within the primary study region. These findings may enable future preventative measures by providing insight into the true origins of groundwater fecal pollution. / Thesis (Master, Pathology & Molecular Medicine) -- Queen's University, 2014-06-02 14:50:16.685

No description available.

Bacterial water pollution is a significant problem because it is associated with reduction in the ‘quality’ of water systems with a potential impact on human health. Faecal indicator bacteria (FIB) are usually used to monitor the quality of water, and to indicate the presence of pathogens in water bodies. However, enumeration alone does not enable identification of the precise origin of these pathogens. This study aimed to monitor the quality of bathing water and associated fresh water in and out of the ‘bathing season’ in the UK, and to evaluate the use of microbial source tracking (MST) such as the host-specific based polymerase chain reaction (PCR) and quantitative PCR (qPCR) to recognize human and other animal sources of faecal pollution. The culture-dependent EU method of estimating FIB in water and sediment samples was performed on beach in the South Sands, Kingsbridge estuary, Devon, UK- a previously ‘problematic’ site. FIB were present at significant levels in the sediments, especially mud, as well as fresh water from the stream and pond flowing onto South Sands beach. However, the quality of bathing water was deemed to be ‘good’ and met with the EU bathing water directive 2006. Using MST it was possible to successfully classify the nature of the source from which the bacteria came. PCR was applied to detect the Bacteroides species 16S rRNA genetic markers from human sewage and animal faeces. All water and sediment samples displayed positive results with a general Bacteroides marker indicating the presence of Bacteroides species. Host-specific PCR showed the human Bacteroides genetic marker only in the sediment of the stream. However, limitations in the ‘types’ of probes available and in the persistence of these markers were identified. Thus, novel dog-specific Bacteroides conventional PCR and qPCR primer sets were developed to amplify a section of the 16S rRNA gene unique to the Bacteroides genetic marker from domestic dog faeces, and these were successfully used to quantify those markers in water samples at a ‘dog permitted’ and ‘dog banned’ beach (Bigbury-on-Sea, Devon, UK). Generic, human and dog Bacteroides PCR primer sets were also used to evaluate the persistence of Bacteroides genetic markers in controlled microcosms of water and sediment at differing salinities (< 0.5 and 34 psu) and temperature (10 and 17 ºC). The rates of decline were found did not differ significantly over 14 and 16 days for the water and sediment microcosms, respectively. Beach sediments which were studied in this project may act as a reservoir for adhesive FIB, and this was confirmed using fluorescence in situ hybridisation (FISH). The similarity in the persistence of these Bacteroides 16S rRNA genetic markers in environmental water and sediment suggests that viable but non-culturable (VBNC) Bacteroides spp. do not persist in the natural environment for long. Therefore, 16S rRNA genetic markers can be of value as additional faecal indicators of bathing water pollution and in source tracking. Thus, in this study MST methods were successfully used and in future applications, dog-specific primer sets can be added to the suite of host-specific Bacteroides genetic markers available to identify the source(s) of problem bacteria found on failing beaches.

During the 1980’s millions of households in Bangladesh switched from drinking surface water to private groundwater wells to reduce their exposure to fecal microorganisms. Sadly, this switch to shallow groundwater resulted in the largest example of drinking water poisoning in history, with approximately 100 million people exposed to high concentrations of naturally occurring Arsenic in the groundwater. Spatial distribution of Arsenic in the shallow aquifers tends to be patchy, so the most economical mitigation option has been lateral switching from high Arsenic wells to nearby low Arsenic wells. The recently developed Arsenic flushing conceptual model, which explains the spatial distribution of Arsenic throughout the shallow aquifers in Bangladesh, suggests however, that low Arsenic zones are recharged via coarse-grained, rapid flow pathways and therefore represent a higher risk for waterborne pathogens. The objectives of this dissertation are to evaluate new methods for sampling and detection of waterborne pathogens, while also identifying sources of fecal contamination and transport pathway(s) to private wells emplaced within the shallow aquifers. It was demonstrated that private wells are broadly contaminated with E. coli, with prevalence ranging from 30 to 70%. The fact that E. coli was detected more frequently in private wells than sealed monitoring wells (p<0.05) suggests that well construction and/or daily pumping contribute to fecal contamination of the private wells. Using DNA-based molecular fecal source tracking, contamination was demonstrated to originate from human fecal waste. Unsanitary latrines, which spill effluent onto the open ground, were demonstrated to cause elevated levels of fecal bacteria in ponds, found in every village. These ponds were demonstrated to have an influence on concentrations of fecal bacteria to at least distances of 12m into the adjacent aquifer. In a culture where latrines, private wells and ponds are frequently clustered closely together, these findings suggest that improvements in the management of human fecal waste changes in placement and construction of private wells could substantially reduce exposure of people to fecal pathogens. Fecal contamination was found to be pervasive in low Arsenic, unconfined, shallow aquifers, and therefore gains from well switching to avoid Arsenic need to be balanced with the risk of consuming waterborne pathogens.

Microbial water quality is generally assessed using fecal indictor organisms; however host-specific microbial source tracking (MST) methodologies can be employed to differentiate sources of fecal pollution. The central goal of this research was to develop and validate a QPCR assay for the quantification of two human-specific polyomaviruses (HPyVs) in environmental water samples. These viruses are prevalent worldwide and produce lifelong, asymptomatic viruria in immunocompetent individuals. A Taqman® quantitative PCR (QPCR) assay based on the conserved T-antigen of two HPyVs (JCV and BKV) was developed and optimized (Chapter 2). HPyVs were detected in a high proportion of human-associated waste samples (e.g. sewage) and were not detected in animal excrement samples (Chapter 2). The effects of ultraviolet radiation, temperature, and salinity on the persistence of HPyVs in water were reported in Chapter 3. Laboratory studies analyzing the effects of various UV doses, temperatures, and/or salinities demonstrated high doses of UV were required to significantly decrease the detection of HPyVs DNA and salinity stabilized pure cultures of HPyVs virus particles at high temperatures (25°C and 35°C). Solar radiation as well as potential predation from microorganisms in sewage significantly reduced the persistence of HPyVs DNA in outdoor mesocosm studies (Chapter 3). An improved method to extract human polyomavirus (HPyVs) DNA from environmental water samples was developed, and the recoveries were larger and more consistent over a range of DNA concentrations as compared to the standard protocol (Chapter 4). In the California beaches study (Chapter 4), the presence of HPyVs by either QPCR or PCR had a high degree of matching results with the adenoviruses (83-91%), Methanobrevibacter smithii marker (82-92%) and moderate degree of matching results with the human-associated Bacteroidales spp. marker (57-80%) (Chapter 4). HPyVs were detected in the presence of various pathogens including: Giardia spp., Cryptosporidium spp., Vibrio spp., enteroviruses, and noroviruses (Chapter 5). The presence of HPyVs in relatively high concentrations of sewage and the specificity of HPyVs combined with the relatively conservative persistence of HPyVs when exposed to environmental conditions and the correlation of HPyVs with pathogens demonstrates that this assay is a useful MST method to detect human sewage.

Siesta Key Beach, located on the Gulf Coast of Florida, is frequently mentioned among the top ten beaches in the US. In summer 2004, high levels of indicator bacteria caused health warnings to be posted, and a storm drainage system was implicated as a possible source of microbial contamination. A study was initiated to determine whether indicator bacteria that persisted in the stormwater system could contribute to high microbial loads in receiving waters. Two sampling events, one within 48 hours of a rain event and the other during dry conditions, were conducted. Water and sediment samples were taken at various sites from the storm drainage system to the beach. Fecal coliforms and Enterococcus spp. were enumerated, and genotypic fingerprints of E. coli and Enterococcus spp. were generated by BOX-PCR. Diversity of E. coli and Enterococcus populations was calculated with the Shannon-Weiner diversity index. Similarity of E. coli and Enterococcus populations was calculated with the population similarity coefficient. After the rain event, levels of fecal coliforms and Enterococcus spp. were high in sediments and exceeded the regulatory standard for all water samples. In dry conditions, levels were lower in water samples, but still high in sediment samples. Significantly greater population diversity was observed in the rain event compared to the dry event for both E. coli and Enterococcus populations, and greater population similarity was vi observed in dry conditions. Enterococcus population diversity was significantly higher in untreated sewage and the Siesta Key rain event when compared to dry conditions, and to a site on the Myakka River (no known human input or urban stormwater runoff). Siesta Key populations in dry conditions were most similar to Myakka, and sewage was the least similar to all other populations. Increased population similarity for E. coli and Enterococcus spp. during dry conditions suggests that a portion of the population is composed of “survivor” isolates. Persistence of survivor isolates in the storm drainage system, where urban runoff can sit for days, suggests a reservoir for indicator bacteria that can be flushed through the system to the Gulf, causing high levels of indicator bacteria in receiving waters.

A library of 18 coliform bacteria strains was obtained from different sampling points in the drinking water distribution system in Lexington, KY, over a three month period in 2006. To investigate the cause of the coliform occurrence we conducted a microbial source tracking study using phenotypic (API 20E, Biolog, and Vitek) and genotypic (pulsed field gel electrophoresis (PFGE) and ribotyping) analyses to determine the degree of genetic variation among isolates. Characterization of isolates by PFGE and ribotyping showed that coliform events in the distribution system were related and a regrowth problem may exist due to biofilm formation. The ability of a persistent Enterobacter cloacae strain to adhere and form biofilm was found to depend on environmental conditions such as temperature, pipe material, soiled surface, chlorine and nutrient levels with higher temperatures and nutrient levels promoting adherence. Considerable variation in adherence and biofilm formation was observed among representative Enterobacter isolates.