Citation
Determining the source of fecal pollution in urban streams through microbial source tracking

Material Information

Title:
Determining the source of fecal pollution in urban streams through microbial source tracking
Creator:
Scopp, Anna Marie ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (57 pages) : ;

Thesis/Dissertation Information

Degree:
Master's ( Master of Science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Integrative Biology, CU Denver
Degree Disciplines:
Biology
Committee Chair:
Mosier, Annika C.
Committee Members:
Miller, Christopher S.
Roane, Timberley M.

Subjects

Subjects / Keywords:
Bacterial pollution of water ( lcsh )
Water quality ( lcsh )
Microbial toxins ( lcsh )
Diagnostic microbiology ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Freshwater ecosystems are routinely monitored for fecal contamination to protect human health and preserve natural biodiversity. Traditionally, cultivation of fecal indicator bacteria (FIB) such as Escherichia coli is used to signify fecal contamination in a water body; however, this approach provides no information about the source of fecal contamination. In this study, Microbial Source Tracking (MST) was used to identify potential sources of fecal pollution affecting Bear Creek and Cherry Creek (Denver, CO, USA) using high-throughput 16S rRNA gene sequencing and quantitative Polymerase Chain Reaction (qPCR) of human-associated Bacteroides 16S rRNA genes. This new, multifaceted approach utilizing both qPCR and gene sequencing suggested that human fecal pollution contributes to the water column microbial communities in Bear Creek and Cherry Creek. In Bear Creek, bacteria associated with human fecal matter were distributed evenly along the creek, suggesting that there are multiple locations where human fecal contamination is entering the creek. Human fecal bacteria increased at downstream sites in Cherry Creek, suggesting that the predominant sources of human fecal pollution occur downstream of the CHERRY-07 site. Levels of potential human fecal pollution were very low in the sediments of both creeks. Sequence analyses suggested that fecal matter from Canadian geese made a very small contribution to the overall microbial community structure in water column and sediment samples (0.4% contribution on average and across nearly half of the samples). Fecal matter from other animals (e.g., duck, dog) were not identified as significant contributing sources to the Bear Creek and Cherry Creek microbial communities. Overall, this study highlights the strengths and limitations of utilizing a multifaceted MST approach (sequencing and qPCR) in ambient waters, which few other studies have demonstrated to date. Future efforts aimed at improving our understanding of fecal pollution in Denver waterways should include seasonal sampling, direct correlations between E. coli culture counts and human-associated Bacteroides qPCR counts, and testing how long human-associated Bacteroides persist in these freshwater streams after fecal contamination.
Thesis:
Thesis (M.S.)--University of Colorado Denver, 2017.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Anna Marie Scopp.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Anna Marie Scopp. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
on10836 ( NOTIS )
1083643678 ( OCLC )
on1083643678

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

DETERMINING THE SOURCE OF FECAL POLLUTION IN URBAN STREAMS THROUGH MICROBIAL SOURCE TRACKING by ANNA MARIE SCOPP B.S., University of Kansas, 2015 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Biology Program 2018

PAGE 2

ii This thesis for the Master of Science degree by Anna Marie Scopp has been approved for the Biology Program by Annika C. Mosier, Chair Christopher S. Miller Timberley M. Roane Date: May 12, 2018

PAGE 3

iii Scopp, Anna Marie (M.S., Biology Program) Determining the Source of Fecal Pollution in Urban Streams Through Microbial Source Tracking Thesis directed by Assistant Professor Annika C. Mosier ABSTRACT Freshwater ecosystems are routinely monitored for fecal contamination to protect human health and preserve natural biodiversity. Traditionally, cultivation of fecal indicator bacteria (FIB) s uch as Escherichia coli is used to signify fecal contamination in a water body; however, this approach provides no information about the source of fecal contamination. In this study, Microbial Source Tracking (MST) was used to identify potential sources of fecal pollution affecting Bear Creek and Cherry Creek (Denver, CO USA ) using high throughput 16S rRNA gene sequencing and quantitative Polymerase Chain Reaction (qPCR) of human associated Bacteroides 16S rRNA genes. This new, multifaceted approach utiliz ing both qPCR and gene sequencing suggested that human fecal pollution contributes to the water column microbial communities in Bear Creek and Cherry Creek. In Bear Creek, bacteria associated with human fecal matter were distributed evenly along the creek, suggesting that there are multiple locations where human fecal contamination is entering the creek. Human fecal bacteria increased at downstream sites in Cherry Creek, suggesting that the predominant sources of human fecal pollution occur downstream of th e CHERRY 07 site Levels of potential human fecal pollution were very low in the sediments of both creeks. Sequence analyses suggested that fecal matter from Canadian geese made a very small contribution to the overall microbial community structure in wate r column and sediment samples (0.4% contribution on average and across nearly half of the samples). Fecal matter

PAGE 4

iv from other animals (e.g., duck, dog) were not identified as significant contributing sources to the Bear Creek and Cherry Creek microbial commu nities. Overall, this study highlights the strengths and limitations of utilizing a multifaceted MST approach (sequencing and qPCR) in ambient waters which few other studies have demonstrated to date Future efforts aimed at improving our understanding of fecal pollution in Denver waterways should include seasonal sampling, direct correlations between E. coli culture counts and human associated Bacteroides qPCR counts, and testing how long human associated Bacteroides persist in these freshwater streams af ter fecal contamination. The form and content of this abstract are approved. I recommend its publication. Approved: Annika C. Mosier

PAGE 5

v ACKNOWLEDGEMENTS Annika C. Mosier Timberley M. Roane Christopher S. Miller Sladjana Subotic Andrew Boddicker Munira Lantz Ren e Kershaw Orin C. Shanks Catherine Kelty Bhargavi Ramanthan Ben Wise Kelsey Foster Emily Corkum Adrienne Narrowe Michael Wunder Kathy Meyer LVT and Wildlife Rehabilitator Tom Parks (Sun Prairie Farms) Golden Animal Hospital Goldenview Vet Hospital Wild Animal Sanctuary Mike Nicks (The Urban Farm at Stapleton) Ellicott Wildlife Rehabilitation Center

PAGE 6

vi TABLE OF CONTENTS CHAPTER I. INTRODUCTION................................................................... ............................... 1 II. MET HODS.................................................................................................. .......... 6 Fecal Library Collection................................................................................... 6 Wastewater Sample Collection.. ....................................................................... 6 Bear Creek and Cherry Creek Sampling Site Selection ..... ............................... 7 Sample Acquisition .............. .............................................. ............................... 7 DNA Extraction ......... ........................................................ ............................... 9 DNA Sequencing........................ ............................... ....................................... 9 Sequence Analyses......................................................................................... 10 SourceTracker ................................................................................................ 11 Quantitative PCR (qPCR) of Hu man Specific Bacteroides 16S rRNA Genes .................................................................................... .................. ........ 11 III. RESULTS.............................................................................................. .............. 17 .. 17 Comparison of Total Bacterial Community Structure Among Sample Types ........................................................................... ........... ........................ 1 7 Proportional Contribution of Fecal Sources to Water Column and Sediment Communities ............. ......... ........................................................... .. 19 Potential Indicator Organisms for Fecal Contamination .......... ..................... 2 0 Quantitative PCR of Human Associated Bacteroides 16S rRNA Genes ... .... 21 IV. QUALITY CONTROL.................................................................... ..................... 35 Sequence Controls........ .................................................................................. 35 Quantitative PCR Controls............................................................................. 3 5

PAGE 7

vii V. DISCUSSION...................................................... ................................. ............... 37 VI. .............. 4 4 REFERENCES................................................................................................................ 4 4

PAGE 8

1 CHAPTER I INTRODUCTION Freshwater ecosystems are habitats of pronounced biodiversity and provide vital resources for humans aroun d the globe. In recent decades, population expansion and urbanization have increasingly stressed these freshwater ecosystems and put valuable human resources at risk. Only a small percentage of freshwater on earth is directly available for human use and ye t contamination jeopardizes this supply. A foremost threat to freshwater ecosystems in urban and agricultural settings is fecal contamination (Meays et al., 2004) Exposure to fecal matter in contaminated waters can cause several illnesses including gastroenteritis, urinary tract infections, meningitis, and ear, nose, and eye infections (Cheung et al. 1990 ; Hamuel et al., 2011) Fecal matter from humans, domestic animals, or wild animals can enter a waterway from point sources such as wastewater treatment plant (WWTP) effluent or from non point sources such as leaky septic system s or storm water runoff (Sowah et al., 2017) It is estimated that 39% of all lakes, streams, and rivers in the United States are not safe for recreational use primarily du e to fecal contamination (Shanks et al. 2016) T he Environmental Protection Agency (EPA) projects that rivers in the United States are subject to 1.3 trillion gallons of raw sewage each year (Environmental Protection Agency 2001) F ecal contaminated waters result in over four billion cases of diarrheal illness worldwide and are responsible for 1.8 million deaths e ach year (Koskey et al. 2015) Fecal pollution has broad economic impacts. Researchers have estimated that over 4.3 billion dollars is spent annually by the United S tates to mitigate freshwater pollution (Kansas State University, 2008). In addition, economic impacts occur when rivers, fisheries, and

PAGE 9

2 beaches are closed as a result of fecal contamination (Rabinovici et al., 2004) In 20 05 alone, over 20,000 days of beach advisories and closures were reported in the United States causing an extreme reduction in recreational tourism that many regions depend on for revenue (Santo Domingo et al., 2007) Treatment for gastrointestinal illnesses caused by swimming in fecal contaminated coastal waters can cost individual cities millions of dollars. It has b een estimated that in Los Angeles and Orange County alone, $21 $51 million are spent annually to treat gastroenteritis caused by recreation in contaminated coastal waters (Given et al., 2006) The risk of acquiring a waterborne illness increases as fecal contamination increases in a waterway. Thus, identifying the amount of fecal contamination in environmental waters is critical to determine the human health risk and to implement proper remediat ion strategies. Problematically, many pathogens are difficult to culture in a laboratory setting and are often distributed irregularly in water bodies making their detection challenging (Field et al., 2007) Quantification of fecal indicator bacteria (FIB; total coliforms, fecal coliforms, Escherichia coli and enterococci) is ins tead used as a proxy for determining if a waterway is contaminated with fecal matter since FIB are normal inhabitants of warm gastrointestinal tracts (Ferguson et al., 2011) Escherichia coli (one of the most common FIB) is suggested by the EPA as the best indicator of freshwa ter fecal contamination (US EPA 1986) Several culturing methods have been developed for the cost effective and feasible quantification of culturable E. coli from water samples. These methods are routinely used across the United States to monitor water quality in order to meet state and federal regulations.

PAGE 10

3 FIB are present in the gastroint estinal tracts of humans, mammals, birds, and several insects, but it is assumed that human waste is a greater threat to human health than animal waste (Gomi et al., 2014) Therefore, d etermining the source of fecal contamination is a key step in identifying and mitigating the human health risks. Problematically, however, culturing methods broadly used to quantify E. coli in waterways lack the ability to distinguish where the pollution i s coming from (e.g., human or animal fecal matter). Additionally, there is very little genetic variation among E. coli isolates from mammalian hosts, making it difficult to distinguish between host species using sequencing approaches (Dick et al. 2005) An approach known as Microbial Source Tracking (MST) has been applied to identify sources of fecal pollution using chemical, biochemical, and molecular approaches (Meays et al., 2004) Modern molecular tools (based on DNA analyses) have greatly extended the effectiveness of MST by identifying genetic markers specific to unique animal hosts (Korajkic et al., 2014) While these approaches are still in early stages of development, molecular MST has been successfully used to distinguish between sources of human and animal waste in aquatic environments including rivers, lakes, and salt water beaches (E ren et al., 2014; Fisher et al., 2015; Hamilton et al., 2010; Ishii et al., 2006; Stoeckel et al., 2011) Molecular MST studies often utilize quantitative polymerase chain reaction (qPCR) to determine the abundance of fecal associated microbes such as Bacteriodes (Ahmed et al., 2008) Bacteriodes are Gram negative, non endospore forming obligate anaerobes that are major inhabit ants of the gastrointestinal tracts of mammals (Koskey et al., 2015) and have extremely high host specificity (Layton et al., 2006) Studies have shown that human associated Bacteroides are more sensitive indicator s of human fecal pollution than E. coli because fecal anaerobes (such as Bacteroides ) are present at 1,000 fold higher densities than

PAGE 11

4 fecal colif orms (such as E. coli ) in the gastrointestinal tracts of humans (Bower et al. 2005) (Fiksdal et al. 1985) Additionally, Bacter oides survive outside of the gastrointestinal tract for much shorter periods of time than E. coli making them the ideal indicator organisms of recent contamination (Fiksdal et al. 1985; Kreader 1998) The HF183/BacR287 qPCR Taqman assay amplifying human associated Bacteroides 16S rRNA genes (Bern hard and Field 2000; Green et al., 2014) has high specificity to human fecal pollution and reproducible results (tested across 27 labs worldwide, including the EPA) (Boehm et al., 2013) Recently, molecular MST studies employing genetic sequencing meth odologies have successfully identified sources of fecal pollution and fecal pathogens in saltwater beaches, lakes, and wastewater treatment plants (Cai and Zhang 2013; Fisher et al., 2015; Neave et al., 2014) This appr oach relies upon the notion that different mammalian hosts have different microbial flora due to differences in diet, immune systems, and antimicrobial exposure (Santo Domingo et al., 2007) The fecal microbiome is determined in a variety of animals (using high throughput DNA sequencing) as a reference library and then compared to the microbiome of natural water or sediment samples. Potential sources of fecal matter in the water samples are de termined by evaluating the sequence diversity and relative abundance patterns of each organism. Fecal indicator species specific to unique animal hosts also help identify sources of contamination and can illuminate cases with several contaminant inputs aff ecting a waterbody (Kinzelman et al., 2011) The aim o f this study was to evaluate the sources of fecal pollution in Bear Creek and Cherry Creek in Denver, Colorado where fecal contamination has been historically problematic and E. coli levels frequently exceed local and state standards ( Department of Environmental Health, 2015) Specifically, the objectives of this study were to compare the

PAGE 12

5 overall bacterial community composition o f water column and sediment samples to known sources of fecal pollution and to determine the overall abundance of fecal indicators specific to human waste. This study used a multifaceted approach combining high throughput gene sequencing and quantitative P CR which few other MST studies have utilized to date. Identification of the sources of fecal pollution is the first step necessary to successfully mitigate contamination in Denver streams.

PAGE 13

6 CHAPTER II METHODS Fecal Library Collection An animal fecal library was generated from 10 different fecal sources: six dog, six Canadian goose, five pigeon, six raccoon, six mallard duck, three cat, three bat, two coyote, two squirrel, and two fox. Fecal samples were collected from June 2016 to Nove mber 2016 in sterile centrifuge tubes and immediately frozen at 20C until permanent storage at 80C. Animal fecal samples were designated AF DOG). Wastewater Sample Collection WWTP influent was us ed as a proxy for human associated fecal contamination. WWTP effluent samples were also evaluated for comparisons to the influent community structure. Four influent and three effluent WWTP samples from separate waste streams were collected using sterile Na lgene bottles submerged approximately 15 cm into the influent/effluent stream. Two influent samples were collected before the first bar screen and two influent samples directly after the first bar screen. Influent and effluent samples were immediately plac ed on ice and transferred to the lab for filtration within three hours. Samples were mixed and then 5 8 mL of each influent sample or 30 mL of each effluent sample were n, Ann Arbor, MI) and then stored at (e.g., WWTP IN)

PAGE 14

7 Bear Creek and Cherry Creek Sampling Site Selection Water column (designated collected from 18 sites along Bear Creek, 18 sites along Cherry Creek, and four sites along the South Platte River at the confluences of Bear Creek and Cherry Creek in Denver, Colorado, USA (Figure 1; Tabl e 1). Bear Creek and South Platte samples at the confluence of Bear Creek were collected on September 29, 2016 while Cherry Creek and South Platte samples at the confluence of Cherry Creek were collected on October 04, 2016. Site selection was determined b y assessing city outfall maps and previous sampling locations used by local to downst ream gradient (e.g., SED CHERRY) Samples collected upstream and downstream of the confluence of each stream with the South Platte River were designated accordingly (e.g., WC SP UP BEAR represents water column samples collected on the South Platte River up stre am of the confluence with Bear C reek). Sample Acquisition Sediment samples were collected using a sterile spatula and sterile petri dish. Prior to sampling at each site, the spatula was rinsed with 70% ethanol and wiped dry with a clean Kimwipe. The sp atula was then rinsed with sterile water to remove residual ethanol, followed by site water. The bottom of the sterile petri dish was then placed down into the sediment, open side down. The sterile spatula was slid under the dish, trapping the sediment in the petri dish. The lid was put on the petri dish and wrapped with parafilm. All sediment samples were stored on dry ice for a maximum of three hours before arriving at the lab for permanent storage at 80C.

PAGE 15

8 Water samples were collected using sterile 1L N algene bottles submerged approximately 15 cm in creek water undergoing constant flow. Nalgene bottles were rinsed with site water three times before sample acquisition. Following sample collection, samples were immediately placed on ice and transferred to the lab for processing for filtration within three hours. Samples were not filtered in the field due to logistical and time constraints in order to sample all sites on a stream within the same day. Water column samples (62 185 mL) were filtered onto 25 m pore size Supor membrane (Pall Corporation, Ann Arbor, MI) filters in a Swinnex ( Merck Millipore, Burlington, MA) filter housing using a peristaltic pump (Series II Geopump, Geotech Environmental Equipment Denver, CO). Following sample filtration, filters were stored at 80 C until DNA extraction. Pump tubing was sterilized prior to filtering each sample by sequential rinses of the pump tubing: (1) rinsed with 250 mL of autoclaved MilliQ water, (2) recirculation of 500 mL of 10% Hydroch loric acid through the tubing for three minutes; (3) rinsed again with 250 mL of autoclaved MilliQ water; and (4) rinsed with 250 mL of site water. Three method blanks were collected for both Bear Creek and Cherry Creek (six in total). Method blanks consisted of a sample of autoclaved MilliQ water processed in the same manner as the study samples. Method blanks were used to ensure that the equipment d econtamination procedure effectively removed microbes between sampling sites and to ensure that contamination was not introduced during field and laboratory processing. Additionally, two DNA extraction blanks were evaluated to test for contamination in the kit buffers and reagents.

PAGE 16

9 Duplicate samples were collected in order to determine the precision of the sample analysis (two duplicate water column samples and two duplicate sediment samples at each creek). Duplicate sites were chosen at random. Duplicate samples were processed in the field and lab in the same manner as the rest of the samples. DNA Extraction DNA was extracted from sediment, water column, and fecal samples using the MP Biomedicals FastDNA Spin Kit for Soil (MP Biomedicals, Santa Ana, CA) a ccording to kit instructions with homogenization in the FastPrep instrument at 6.0 m/sec for 40 seconds. Approximately 500 mg of sediment or fecal matter was weighed out and placed into Lysing Matrix E tubes. Water column samples and WWTP filters were plac ed directly into Lysing Matrix E tubes. DNA extracts were stored at 80C. A Nanodrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA) was used to determine 260/280 values of select DNA extracts. The Qubit dsDNA HS and dsDNA BR Assay kits (Life Tech nologies, Carlsbad, CA) were used to determine the DNA concentration of extracts. Qubit DNA quantitation was run in duplicate to determine the average DNA concentration for each DNA extract. DNA Sequencing DNA extracts were sent to the University of Illino is Roy J. Carver Biotechnology Center, Urbana, Illinois for amplicon 250 bp paired end sequencing with the 515F Y and 926R primers for total 16S rRNA (Parada et al., 2016; Quince et al., 2011) Library preparation was completed with the Fluidigm 48.48 Acce ss Array IFC platform (Fluidigm Corporation, South San Francisco, CA) and sequencing was performed on the Illumina

PAGE 17

10 HiSeq sequencing platform ( Illumina, San Diego, CA ), as previously described (Ramanathan et al., 2017) Sequence Analyses Relative sequence abundance and diversity analyses were conducted using QIIME (Quantita tive Insights into Microbial Ecology; Caporaso et al., 2011). Paired ends were joined using the fastq join method and a specific minimum overlap score of 30 (Aronesty, 2011). Joined sequences were filtered using a Phred quality score of 20. Sequences with minimum merge length < 80 bp were discarded. The last 20 bp were removed from both ends of each sequence to remove primers Sequencing reads were clustered into operational taxonomical units (OTUs) at 97% sequence identity and assigned genus level taxonomy using the August 2013 Greengenes bacterial 16S rRNA gene database (DeSantis 2006). Final OTUs were checked for chimeras using the DECIPHER web tool with the short sequences option (Wright et al., 2012) and putative chimeras were removed from the OTU table OTUs with less than 0.05% of the sequence reads were discarded (including singletons; min_count_fraction in QIIME) (Bokulich et al., 2013). All chloroplast OTUs and OTUs present in blanks were also removed from all analyses. Alpha and beta diversity meas ures of total 16S rRNA sequences were calculated in QIIME to assess the number of observed and predicted OTUs within each sample. These analyses were completed using the core diversity command in QIIME using a sequencing depth of 13,607 (the minimum number of sequence reads across all sam ples). Correlations between community composition (OTU distribution and relative abundance) of watershed samples and of reference fecal samples were analyzed by Principal Coordinates Analysis (PCoA ). ANOSIM (Clarke 1993) statistical testing was used to determine the percent of the

PAGE 18

11 variation observed in the microbial communities that may be explained by sample type fecal type, or creek type OTU sequences were search ed against the SILVA rRNA gene database ( Quast et al., 2013) using BLAST to identify Enterobacteriaceae OTUs and potential indicator organisms for fecal co ntamination (SILVA was used for taxonomy predictions due to the higher resolution than Greengenes). The distribution of OTUs shared among and unique to (i.e., indicator organisms) reference host fecal types were assessed. SourceTracker Analysis SourceTrac ker software (Version 1.0; Knights et al., 2011) was implemented within QIIME and was used to predict the potential sources of microbial communities within the water column and sediment samples, based on sequence community profiles. Source (reference anima l and WWTP samples) and sink (Bear Creek and Cherry Creek water column and sediment) samples were designated in a mapping file. Only samples containing over 13,607 reads for the total 16S rRNA gene sequence data were considered. The program uses a Bayesian approach to estimate the proportion of each sink sample (e.g., water column) comprising taxa from a source environment (e.g., animal fecal matter) (Knights et al., 2011). Quantitative PCR (qPCR) of Human Specific Bacteroides 16S rRNA Genes qPCR assays were performed following EPA methods and based on previously reported methodology (Shanks et al. 2016a; Green et al., 2014) Reaction mixtures contained 23 L of PCR Master Mix and 2 L of DNA template ( average DNA conc entration of 27.76 ng/ L). PCR Master Mix was prepared by combining 12.5 L 1X TaqMan Environmental Master Mix version 2.0 (Applied Biosystems, Foster City, CA), 6.59 L PCR grade nuclease free water, 0.25 L 0.2 mg/mL bovine serum albumin (Sigma Aldrich, St. Louis,

PAGE 19

12 MO), 3.0 L Probe/Primer Mix, and 0.66 L Internal Amplification Control (IAC) template (151 copies/L; described below) per reaction. Probe/Primer Mix was prepared in advance by combining 500 M stock solution of HF183 forward primer (ATCATGAGT TCACATGTCCG) (Green et al. 2014) 500 M stock solut ion of BacR287 reverse primer (CTTCCTCTCAGAACCCCTATCC) (Green et al. 2014) 100 M 6 carboxyfluorescein (FAM) labeled probe (FAM CTAATGGAACGCATCCC MGB) (Green et al. 2014) 100 M VIC labeled probe (VIC AACACGCCGTTGCTACA MGB) (Green et al. 2014) and PCR grade nuclease free water for a final Primer/Probe Mix concentration of 1 M/80 nM. Multiplex reaction mixtures contained either pure DNA extracts from environmental samples or 11 to 1.45 x 10 5 gene copies of reference standard DNA in a total reaction volume of 25 L. No template controls (NTC) contained the same reaction mixture, but without sample DNA. Sediment samples were analyzed at both full concent ration and after dilution to 5 ng/L since DNA concentrations were high. All samples were run in a minimum of six qPCR reaction replicates. 2 Reaction tubes were amplified in a StepOnePlus real time PCR System (Applied Biosystems, Foster City, CA) with the following conditions: 95C for 10 minutes, then 40 cycles of 95C for 15 seconds and 60C for 1 minute. The cycling threshold (C T ) was set to 0.03 (Shanks et al. 2016a) All R 2 values were amplification efficiencies fell between 0.90 1.10. qPCR data was analyzed by determining the median value of all replicates for each sample (median was chosen because it does not assume normal distribution and is more robust when sample size is relatively low). All replicate measurements were included in the median calc ulation including values within standard range, values below detection limit but quantifiable, and zero values for replicates that did not amplify. In all cases, the reported

PAGE 20

13 median was within the standard range detection limit. Error was measured using th e median absolute deviation (MAD) of the replicates for each sample. Amplification interference was monitored using an internal amplification control (IAC) reference sequence to assure no false negatives were reported (Shanks et al. 2016a; Green et al., 2014) The IAC reference DNA material contained the same primer binding sites as target sequences, however the IAC DNA sequence contained a different probe sequence and different reporter molecule (VIC). The IAC reference sequence was i ncluded in all reactions at a known concentration and amplification interference was monitored for each specific instrument run to assure accurate amplification of target sequences. If greater than one third of the replicates of a given water column or sed iment sample showed amplification interference (defined as mean of the IAC amplification in the NTC plus three times the standard deviation of the mean), sample extracts were further assessed for inhibition or competition. Evidence of inhibition or competi tion was determined using the Range of Quantification (ROQ), where the upper limit equals the mean C T of the FAM reporter for the highest standard passing the amplification interference test. Samples with a mean FAM C T exceeding the ROQ were considered to display inhibition (e.g., PCR amplification interference from substrates that persist in the filter DNA extract after DNA purification). Samples with a mean FAM C T less than the ROQ were considered to display competition (e.g., competition between PCR ampl ification of the native human associated target sequence and the IAC spike). A cross reactivity test was performed to ensure that the qPCR assay quantified human associated waste but not fecal pollution from other animals. Human specific Bacteroides 16S rRNA genes were quantified in animal fecal samples from the fecal

PAGE 21

14 reference library described above (fecal samples from six dog, five Canadian goose, six pigeon, five raccoon, six mallard duck, three cat, three bat, two coyote, two squirrel, and two f ox ) and eight human wastewater samples. Animal fecal DNA extracts were diluted with PCR grade nuclease free H 2 O to a concentration of 1 ng/L. Multiplex reaction mixtures were prepared and analyzed as described above

PAGE 22

15 Figure 1. Sampling sites for (A) Bear Creek and (B) Cherry Creek in the Denver metropolitan area (Colorado, USA). South Platte River sites upstream and downstream of the confluence with each stream are labeled as SP UP and SP DOWN, respectively. Arrows indicate the direction of the stream or river flow. Base images were modified from Google Maps (Map data Google 2017). A B

PAGE 23

16 Site Name Site Latitude Site Longitude Site Name Site Latitude Site Longitude BEAR 01 39.663375 105.093474 CHERRY 01 39.662043 104.869227 BEAR 02 39.662752 105.079908 CHERRY 02 39.66857 104.886255 BEAR 03 39.659843 105.075901 CHERRY 03 39.674655 104.893919 BEAR 04 39.659554 105.069392 CHERRY 04 39.685666 104.903894 BEAR 05 39.659006 105.067326 CHERRY 05 39.691185 104.913262 BEAR 06 39.656967 105.063212 CHERRY 06 39.697187 104.923762 BEAR 07 39.655516 105.058284 CHERRY 07 39.703428 104.930432 BEAR 08 39.654496 105.054059 CHERRY 08 39.706345 104.941062 BEAR 09 39.652621 105.04591 CHERRY 09 39.711913 104.948497 BEAR 10 39.653428 105.041612 CHERRY 10 39.715905 104.957674 BEAR 11 39.652257 105.032944 CHERRY 11 39.718303 104.972513 BEAR 12 39.651653 105.029795 CHERRY 12 39.721027 104.97952 BEAR 13 39.650706 105.026945 CHERRY 13 39.725966 104.985876 BEAR 14 39.650948 105.024513 CHERRY 14 39.729313 104.990544 BEAR 15 39.651898 105.022727 CHERRY 16 39.738722 104.997525 BEAR 16 39.651213 105.016156 CHERRY 17 39.742764 104.999866 BEAR 17 39.651153 105.015331 CHERRY 18 39.748336 105.001407 BEAR 18 39.649898 105.013679 CHERRY 19 39.754432 105.008185 SP DOWN BEAR 39.65046 105.01295 SP DOWN CHERRY 39.755004 105.007748 SP UP BEAR 39.649381 105.013462 SP UP CHERRY 39.754434 105.008408

PAGE 24

17 CHAPTER III RESULTS Historical Fecal Pollution in Denver, Colorado Fecal pollution in Bear Creek, Cherry Creek, and the South Platte River has been historically problematic as E. coli levels frequently exceed local and state standards ( Department of Environmental Health, 2015) In 2016, E. coli levels exceeded state regulatory limits on 34.5% of Cherry Creek sampling events, 30.6% of Bear Creek sampling events, and 50.0% of South Platte River sampling events (Figure 2A C). Local limits were exceeded by E. coli on 56.4% of Cherry Creek sampling ev ents, 52.8% of Bear Creek sampling events, and 66.5% of South Platte River sampling events (Figure 2A C). Comparison of Total Bacterial Community Structure Among Sample Types Sequence analysis of total 16S rRNA gene data identified 1,402 total OTUs across all samples, with 18 982 OTUs present within a single sample. Alpha rarefaction curves and richness estimates (based on Chao1) both indicated that the sequencing diversity had been saturated ( Figure 3 ). Broad taxonomic classification (class and genus level ) and relative abundance analysis showed that Actinobacteria were abundant in all water column samples ( Figure 4 ). The genus Microcystis was most abundant in Bear Creek water column samples, while the genus Synechococcus was most abundant in Cherry Creek w ater column samples. The family Comamonadaceae was present in all sediment samples and made up as large as 22% of the relative abundance in some samples. The taxonomic profile for reference fecal samples was highly variable within each group (e.g., cat sam ples) and variable among fecal types (e.g., dog, cat). Influent WWTP samples showed a high relative abundance of Acinetobacter and Comamonadaceae.

PAGE 25

18 Analyses of the number and relative abundance of OTUs in each sample showed that each sample type (reference fecal samples, creek water column samples, and creek sediment samples) had unique community structures. Weighted and unweighted UniFrac PCoA plots showed samples clustered by type (sediment, water column, or fecal) whether or not relative abundance was tak en into account (Figure 5 A and 5 B). ANOSIM statistical testing revealed that a majority of the variation in community structure was explained by sample type (weighted = 68.5%, p = 0.001; unweighted = 71.4%, p = 0.001). When comparing all reference fecal s amples (animal and WWTP influent/effluent), influent and effluent WWTP samples showed distinct clustering from each other and from other animal fecal samples whether or not relative abundance was taken into account (Figure 5C and 5 D). ANOSIM statistical te sting supported the observed pattern of the PCoA plots showing that fecal type (animal, influent, or effluent) explains some of the observed variation in fecal community structure (weighted = 43.7%, p = 0.001; unweighted = 26.3%, p = 0.024). Weighted and unweighted UniFrac PCoA plots comparing all creek samples (Bear Creek, Cherry Creek, and South Platte River) to one another showed that water column samples from each creek clustered independently ( Figure 6A and 6 B). ANOSIM statistical testing of the weigh ted UniFrac showed that 92.7% of the variation of water column microbial communities may be explained by creek type (p = 0.001), while unweighted UniFrac revealed 71.1% of the variation may be explained by creek type (p = 0.001). When comparing all sedimen t samples to one another, weighted and unweighted UniFrac PCoA plots showed some clustering of samples by creek type ( Figure 6C and 6 D), but the clustering was less strong than seen for water column samples (ANOSIM for weighted = 47.9%, p = 0.001; unweight ed = 41.3%, p = 0.001).

PAGE 26

19 Within each creek, the sediment and water column communities differed from each other (Figure 7 ). For Bear Creek, ANOSIM statistical testing of weighted UniFrac revealed that 100% of the variation in microbial communities may be exp lained by sample type (p = 0.001), while unweighted UniFrac showed that 89.0% of the variation may be explained by sample type (p = 0.001). A similar pattern was observed in Cherry Creek with water column and sediment samples clustering distinctly. Weighte d UniFrac revealed that 99.9% of the variation in microbial communities may be explained by sample type (p = 0.001) while unweighted UniFrac revealed that 99.3% of the variation may be explained by sample type (p = 0.001). Proportional Contribution of Feca l Sources to Water Column and Sediment Communities SourceTracker was used to determine the likely proportional contribution of fecal sources (reference animal and human fecal samples) to the overall microbial community structure (total 16S rRNA genes) of water column and sediment samples. SourceTracker ide ntified that human waste (based on human WWTP influent) likely contributed to the microbial communities of all 18 Bear Creek water column samples and 17 out of 18 Cherry Creek water column samples (Figure 8 B). The percent contribution of human waste to water column samples was relatively constant along the length of Bear Creek (7.1% on average for all sites). In Cherry Creek water column, the contribution of human waste increased at downstream sites: the average contribution at upstream sites (CHERRY 01 to CHERRY 07) was 1.0%, whereas the average contribution at downstream sites (CHERRY 08 to CHERRY 19) was 4.6%. The percent contribution was greatest in the South Platte water column samples (11.4% on average). For sediment samples, human waste contrib uted only a small

PAGE 27

20 proportion of the total microbial community structure across all sites (0.7% on average, ranging from 0 3.8% across all sediment samples) (Figure 9B) SourceTracker identified that fecal matter from Canadian geese likely contributed a sm all proportion to the overall microbial community structure in water column and sediment samples: 0.4% contribution on average and affected nearly half of the samples (35 out of 80 total samples). For the water column samples specifically, goose fecal matt er contributed 0.1% of the community structure on average and affected 36% of the Bear Creek and Cherry Creek water column samples. Goose fecal matter contributed a much larger percentage (5.4% on average) to the overall microbial community structure of th e South Platte water columns samples. Pigeon fecal matter contributed 0.01% of the community structure at one site (WC CHERRY 17), but no other water column or sediment samples. Fecal samples from dogs, raccoons, mallard ducks, cats, bats, coyotes, squirre ls, and fox were not identified as contributing sources to the Bear Creek or Cherry Creek water column or sediment communities. Potential Indicator Organisms for Human Associated Waste Potential indicator organisms associated with human waste were identifi ed as OTUs found in all four WWT P samples influent samples but not present in any other animal fecal sample s Overall, 17 potential indicat or OTUs were identified (Table 2 ). An additional two OTUs (OTU 1078587 and OTU 975306 ) did not meet these criteria bu t were included in the analyses because they had a high identity (100% ID over 372 bp) to sequences in the Lachnospiraceae family, which has been previously identified to include potential indicator organisms ( Eren et al. 2015; Koskey et al., 2014; Mclella n et al. 2013; Newton et al. 201 3 ) These two OTUs were found in all four WWTP influent samples, but also occurred at low

PAGE 28

21 abundance (6 78% lower relative abundance than WWTP influent samples) in 2 4 other animal fecal samples (out of 33 total). These OTUs may be considered as human associated, but not necessarily human specific. The 19 p otential indicator OTUs displayed different patterns of relative abundance across the water column samples from each cree k (Figure 10). Group 1 contained 3 OTUs that had hi ghest relative abundance only at site BEAR 16. Group 2 contained 2 OTUs that had high relative abundance at South Platte River sites. Group 3 contained 8 OTUs that had highest relative abundance at the lower Cherry Creek sites. Group 4 contained 6 OTUs that were more evenly distributed across all sites. It is important to note that further analyses would be needed in order to eliminate infrastructure associated OTUs (e.g., biofilms within sewer system pipes) from this dataset, however they likely occur at very low abundances due to the high volume and biomass of human waste in WWTP influent streams Oligotyping analyses may enhance the resolution of different ecotypes associated with each fecal source (Eren & McLellan, 2014). Quantitative PCR of Human As sociated Bacteroides 16S rRNA Genes The overall abundance of human associated Bacteroides 16S rRNA genes was determined in samples from Bear Creek and Cherry Creek. In the Bear Creek water column samples, human associated Bacteroides 16S rRNA genes amplifi ed in 11 out of 18 samples (Figure 8 A; Table 3 ). Gene copy numbers in water column samples that amplified were fairly consistent across all Bear Creek sites, ranging from 1.2 x 10 3 to 1.1 x 10 4 gene copies per 100 mL creek water. In the Cherry Creek water column samples, human associated Bacteroides 16S rRNA genes amplified in 9 out of 18 samples and showed increasing abundance towards downstream sites (Figure 8 A ; Table 3 ). Gene copy numbers were highest in WC CHERRY

PAGE 29

22 14 (2.0 x 10 5 gene copies per 100 mL cre ek water). Cherry Creek water column samples were on average over 11 times higher than Bear Creek w ater column samples. Human associated Bacteroides 16S rRNA genes amplified in all of the South Platte River samples (upstream and downstream of the confluenc e with Bear and Cherry Creeks). Abundance increased along the South Platte River from 2.0 x 10 3 to 1.8 x 10 4 gene copies per 100 mL at the confluence of Bear Creek to 8.0 x 10 5 to 1.4 x 10 6 gene copies per 100 mL at the confluence of Cherry Creek. Human a ssociated Bacteroides 16S rRNA genes did not amplify in 92% of the Bear Creek and Cherry Creek sediment samples. Amplification was only seen in two Bear Creek samples (5.6 x 10 2 copies per gram sediment at SED BEAR 08 and 7.3 x 10 3 copies per gram sediment at SED BEAR 15) and one Cherry Creek sample (2.7 x 10 3 copies per gram sediment at SED CHERRY 17) (F igure 9A ; Table 4 ). Amplification was very high in the South Platte sediment samples at the confluence with Cherry Creek (1.1 x 10 4 2.8 x 10 4 copies per gram sediment).

PAGE 30

23 Figure 2 M onthly Escherichia coli averages of (A) Cherry Creek, (B) Bear Creek, and (C) South Platte River Data adapted from the 2016 Denver Water Quality Data (Department of Public H ealth and Environment 2016)

PAGE 31

24 Figure 3 Chao1 alpha rarefaction curves for Bacterial 16S rRNA gene sequences for all samples.

PAGE 32

25 Figure 4 Relative abundance taxonomic bar graph separated by sample type where each color represents a different taxonomic group at family or genus level taxonomic resolution. The graph shows an overall trend in differences of taxonomic groups found in wastewater influent, wastewater effluent, animal fecal, sediment, and water column samples. Sediment and water column samples are ordered as shown in Figure 10 with South Platte samples following each creek.

PAGE 33

26 Figure 5 (A) Weighted and (B) Unweighted UniFr ac PCoA plot showing fecal samples (red squares ), Bear Creek, Cherry Creek, and South Platte River water column samples (orange circles ), and Bear Creek, Cherry Creek, and South Platte River sediment samples (blue triangles ). (C) Weighted and (D) Unweighte d UniFrac PCoA plot showing animal fecal samples (red triangles ), WWTP effluent (blue circles ), and WWTP influent (orange squares ).

PAGE 34

27 Figure 6 (A) Weighted and (B) Unweighted UniFrac PCoA plot showing Bear Creek water column samples (red circles ), Cherry Creek water column samples (blue triangles ), and South Platte water column samples ( orange squares ). (C) Weighted and (D) Unweighted UniFrac PCoA plot showing Bear Creek sediment samples (red circles ), Cherry Creek sediment samples (blue triangles ), and South Platte sediment samples (orange squares ).

PAGE 35

28 Figure 7 (A) Weighted and (B) Unweighted UniFrac PCoA plot showing Bear Creek sediment samples (red circles ) and Bear Creek water column samples (blue squares ). (C) Weighted and (D) Unweighted UniFrac PCoA plot showing Cherry Creek sediment samples (red circles ) and Cherry Creek water column samples (blue squares ).

PAGE 36

29 Figure 8 (A) Median number of human associated Bacteroides 16S rRNA gene copies per mL of Bear Creek and Cher ry Creek water column samples. Error bars represent the median absolute deviation calculated from the median of the replicates from each sample. (B) Percent of predicted origin from human waste (inferred from WWTP influent) in water col umn samples from Bea r Creek, Cherry Creek, and the South Platte River based on SourceTracker analyses.

PAGE 37

30 F igure 9 (A) Median number of human associated Bacteroides 16S rRNA gene co pies per gram of Bear Creek, Cherry Creek and South Platte River sediment samples. Error bars represent the median absolute deviation calculated from the median of the replicates from each sample. (B) Percent of predicted origin from human waste (inferred from WWTP influent) in sediment samples from Bear Creek, Cherry Creek, and South Platte River based on SourceTracker analyses. A B

PAGE 38

31 Table 2 Potential indicator OTUs (OTUs found in human associated waste samples but no other animal fecal samples). Putative taxonomy was assigned based on the top BLAST hit against the SI LVA database.

PAGE 39

32 Figure 10 Relative abundance of potential indicator OTUs associated with human waste, based on normalized read counts within the total 16S rRNA gene sequence dataset. OTUs are grouped by patterns of relative abundance across water column sampling sites.

PAGE 40

33 Table 3 Median number of human associated Bacteroides 16S rRNA gene copies per mL of Bear Creek, Cherry Creek, and South Platte River water column samples. Median absolute deviation was calculated as a measure of error. Samp le Median copies per mL of Water Filtered Median Absolute Deviation of copies per mL of Water Filtered WC BEAR 01 44 8 WC BEAR 02 12 4 WC BEAR 03 15 7 WC BEAR 04 39 18 WC BEAR 05 19 4 WC BEAR 06 15 3 WC BEAR 07 0 0 WC BEAR 08 0 0 WC BEAR 09 0 0 WC BEAR 10 19 7 WC BEAR 11 0 0 WC BEAR 12 16 12 WC BEAR 13 0 0 WC BEAR 14 0 0 WC BEAR 15 15 3 WC BEAR 16 0 0 WC BEAR 17 113 27 WC BEAR 18 21 3 WC SP UP BEAR 179 39 WC SP DOWN BEAR 20 7 WC CHERRY 01 0 0 WC CHERRY 02 0 0 WC CHERRY 03 0 0 WC CHERRY 04 0 0 WC CHERRY 05 0 0 WC CHERRY 06 0 0 WC CHERRY 07 0 0 WC CHERRY 08 0 0 WC CHERRY 09 14 4 WC CHERRY 10 0 0 WC CHERRY 11 86 11 WC CHERRY 12 159 3 WC CHERRY 13 423 29 WC CHERRY 14 2052 116 WC CHERRY 16 303 46 WC CHERRY 17 178 25 WC CHERRY 18 332 5 WC CHERRY 19 59 12 WC SP UP CHERRY 13539 114 WC SP DOWN CHERRY 7727 605

PAGE 41

34 Table 4 Median number of human associated Bacteroides 16S rRNA gene co pies per gram of Bear Creek, Cherry Creek, and South Platte River sediment samples. Median absolute deviation was calculated as a measure of error. Sample Median copies per gram of Sediment Median Absolute Deviation of copies per gram of Sediment SED BEAR 01 0 0 SED BEAR 02 0 0 SED BEAR 03 0 0 SED BEAR 04 0 0 SED BEAR 05 0 0 SED BEAR 06 0 0 SED BEAR 07 0 0 SED BEAR 08 558 168 SED BEAR 09 0 0 SED BEAR 10 0 0 SED BEAR 11 0 0 SED BEAR 12 0 0 SED BEAR 13 0 0 SED BEAR 14 0 0 SED BEAR 15 7280 3469 SED BEAR 16 0 0 SED BEAR 17 0 0 SED BEAR 18 0 0 SED SP UP BEAR 0 0 SED SP DOWN BEAR 0 0 SED CHERRY 01 0 0 SED CHERRY 02 0 0 SED CHERRY 03 0 0 SED CHERRY 04 0 0 SED CHERRY 05 0 0 SED CHERRY 06 0 0 SED CHERRY 07 0 0 SED CHERRY 08 0 0 SED CHERRY 09 0 0 SED CHERRY 10 0 0 SED CHERRY 11 0 0 SED CHERRY 12 0 0 SED CHERRY 13 0 0 SED CHERRY 14 0 0 SED CHERRY 16 0 0 SED CHERRY 17 2738 2738 SED CHERRY 18 0 0 SED CHERRY 19 0 0 SED SP UP CHERRY 10968 3153 SED SP DOWN CHERRY 27788 2651

PAGE 42

35 CHAPTER IV QUALITY CONTROL Sequence Controls Four method blanks and two DNA extraction blanks were evaluated by sequence analysis since low levels of background contamination have been previously reported in DNA extraction and sequencing kits (Glassing et al., 2016; Salter et al., 2014). All OTUs found in blanks were removed from their respective analyses. Sequencing reproducibility was evaluated in nine duplicate samples (one WWTP influent, four water column, and four sediment samples). Relative abundan ce values of bacterial 16S rRNA gene OTUs were highly correlated for water column and WWTP influent samples (R 2 > 0.96). Sediment sample replicates were more variable with R 2 values ranging from 0.58 to 0.91 (perhaps due to heterogeneity in the sediment st ructure). Quantitative PCR Controls Six method blanks (consisting of autoclaved MilliQ water) and two DNA extraction blanks were processed during sampling of each creek to ensure sterility of equipment and no cross contamination. All blanks showed no ampl ification of human associated Bacteroides 16S rRNA genes. No template controls (NTCs) in each qPCR run showed no showed no amplification of human associated Bacteroides 16S rRNA genes. Two sites were chosen at random on each creek for water column and sedi ment biological replicates. All water column replicates showed either exact or similar (less than 15% standard deviation) gene copy numbers of human associated Bacteroides 16S rRNA genes when compared to their respective water column sample, with the excep tion of WC BEAR 18 (1.4 2.1 x 10 1 human associated Bacteroides 16S rRNA genes among duplicates)

PAGE 43

36 that likely had higher variability because the values were approaching the lower limits of detection. All sediment samples showed exact gene copy numbers of h uman associated Bacteroides when compared to their respective sediment replicate sample No water column or sediment samples showed evidence of amplification inhibition. Evidence of competition was observed in two water column samples (WC SP UP CHERRY and WC SP DOWN CHERRY) where human associated Bacteroides 16S rRNA gene copies were especially high, which suggests that gene copy numbers in these samples would be even higher without competition. A specificity test was performed to assure that the human ass ociated Bacteroides 16S rRNA qPCR assay amplified in human waste samples but not in fecal matter from other animals (e.g., dog or goose). Assay specificity was calculated by dividing the number of true negatives (reference fecal samples that are not human and test negative) by the sum of true negatives and false positives (reference fecal samples that are not human and test positive) to give a qualitative assessment. Results indicated that the assay specificity was 87.3% for our reference fecal samples, whi ch exceeded the specificity metric of 80% set as a minimum by the EPA (Shanks et al. 2016a) WWTP influent and effluent samples amplified at extremely highly levels ( 1. 1 x 10 6 3.1 x 10 9 copies/ 100 mL) confirming that the assay amplifies human specific Bact eroides 16S rRNA genes in human fecal waste

PAGE 44

37 CHAPTER V DISCUSSION Until recently, monitoring for single genetic markers has been the primary method of assessing fecal pollution in watersheds This study is one of few to utilize a community sequencing approach to investigate the sources of fecal pollution in watersheds and to further confirm sequencing results with qPCR (Newton et al. 2013; Ohad et al. 2016) W e determined potential source and abundance of fecal associated bacteria in Bear Creek, Cherry Cree k and the South Platte River, where l evels of fecal indicator bacteria often exceed regulatory limits but little had been known about the source of fecal pollution. Human fecal pollution in waterways is a significant public health threat since it is assum ed that human waste is more likely to carry human pathogens than animal waste (Eren & Mclellan et al., 2014; Soller et al., 2010) Studies have shown that microbial communities in human fecal matter vary across time and space, indicating that it is important to consider geographic location when selecting genetic markers of potential sources (Koskey et al., 2015; Yatsunenko et al. 2012) In this study, wastewater samples were collected from local wastewater treatment plants servicing the Denver metropolitan area (Colorado) to ass ure that human waste signatures were specific to the area (Mclellan et al. 2013 ; Newton et al. 2015) These influent samples were used as a proxy for human fecal pollution as they are representative of composites of tens of thousands to hundreds of thousands of Colorado residents The influent samples contained o rganisms within th e taxonomic groups Ruminococcaceae Lachnospiraceae, Bacteroidaceae and Bifidobacteriaceae which have been previously associated with human fecal matter (McLellan et al. 2011)

PAGE 45

38 Recently, community based MST approaches have gained popularity because they allow for the identification of multiple contributing sources as opposed to single marker based methods designed to detect the presence of one type of fecal pollution (Cao et al. 2013) (Unno et al. 2010) (Cao et al. 2011) (Dubinsky et al. 2012; Wu et al. 2010) H ere, h igh throughput gene sequencing data suggested the presence of human fecal pollution in the Bear Creek, Cherry Creek, and South Platte water column samples. In Bear Creek and Cherry Creek SourceTracker software identified that human waste (based on influent WWTP samples as a proxy) contributed an average of 5.8% of the overall mic robial community composition in water column samples (Figure 8B). Additionally, potential human indicator OTUs were found in all of the water column samples, but with varyin g patterns of relative abundance across the sites (Figure 10). Groups displaying different relative abundance patterns may be indicative of different human derived fecal sources. For instance, Group 2 with highest relative abundance in South Platte river s ites may be associated with WWTPs since this is the only stream with a direct WWTP influence. The high relative abundance of Group 3 at lower Cherry Creek sites (similar to the qPCR data trends) may be associated with a fecal signature unique to the reach such as urban storm sewers. Other molecular MST studies also employed similar sequencing methodologies to successfully identify sources of fecal pollution and fecal pathogens in salt water beaches, lakes, and wastewater treatment plants (Cai & Zhang, 2013; Fisher et al., 2015; Neave et al., 2014). SourceTracker identif ied r aw sewage, storm water and treated effluent as contributing sources to diverse water systems including recreational freshwater, drinking water systems, estuaries and coastal waters (Brown et al. 2017; Henry et al. 2016; Liu et al. 2018; Staley et al. 2018)

PAGE 46

39 Previous s tudies using qPCR assays to target human associated fecal pollution confirmed human fecal contamination in creeks, lakes, bays, and marine estuaries (W Ahmed et al. 2007; Warish Ahmed et al. 2010; Bower et al. 2005; Caprais et al. 2007; Chase et al. 2012; Newton et al. 2011, 2013; Peed et al. 2011; Snyder, Molina, and Georgacopo ulos 2017) Here, h uman associated Bacteroides 16S rRNA genes were quantified with qPCR in Bear Creek, Cherry Creek, and South Platte water column samples at the confluence of Bear Creek and Cherry Creek, providing evidence of human associated fecal pol l ution in these systems (Figure 8 A). Abundance was relatively constant along the length of Bear Creek, compared to Cherry Creek where abundance increased significantly at downstream sites. Additional sampling of storm water outfalls and septic systems coul d facilitate the process of further pinpointing the precise location of human fecal pollution in Bear Creek and Cherry Creek. Sampling these creeks at multiple time points will be critical to determine patterns of human associated Bacteroides abundance over monthly and seasonal scales. S pecies within the genus Bacteroides have 4 7 16S rRNA gene copies per cell based on the ribosomal RNA operon copy number database ( rrnDB ; (Stoddard et al. 2015) ) Therefore, human associated Bacteroides qPCR results do not directly compare to cell counts. When considering 16S rRNA gene copy numbers (based on the average of 5.625 copies per cell) for a rough approximation of cell numbers, water column samples averaged 1.1 x 10 4 h uman associated Bacteroides in Bear Creek, Che rry Creek, and the South Platte River. 16S rRNA gene copy number should be considered when evaluating qPCR results in a regulatory setting or in human health risk assessments. Quantitative estimates suggest that Bacteroides levels of 860 16S rRNA gene cop ies per 100 mL water may pose a threat to human health (Soller et al. 2010) suggesting that the

PAGE 47

40 levels of Bacteroides measured in this study should be further evaluated as a potential health risk. Overall, Bear Creek and Cherry Creek sediment samples appeared to contain less human fecal pollution than water column samples. Human associated Bacteroides 16S rRNA genes were quantified with qPCR in only five sediment samples across all creeks (Figure 9A ; Table 3). So urceTracker analyses with 16S rRNA gene sequencing data suggested that human fecal pollution did contribute to the overall bacterial community structure of many creek sediment samples, but only a very small percentage (Figure 9B) Human associated Bactero ides 16S rRNA genes amplified at exceptionally high levels in all WWTP influent and effluent samples (1. 1 x 10 6 3.1 x 10 9 copies/ 100 mL ). High levels of human associated Bacteroides 16S rRNA genes have been previously found in raw sewage, treated effluen t, sewage overflow septic wastewater, and urban storm water (Ahmed et al., 2016; Sercu et al., 2009; Bower et al. 2005) The potential for h uman associated Bacteroides in treated WWTP effluent has health, ecological, and regulatory implications in effluent impacted streams and should be evaluated further Bacteroides are considered ideal indicator organisms of recent fecal contamination because they survive outside of the gastrointestinal tract for much shorter periods of time than E. coli (Fiksdal et al. 1 985; Kreader 1998). Nonetheless, prior studies showed that human associated Bacteroides can persist in water samples for up to ~six da ys (Cao et al. 2017; Mattioli et al. 2017; Zimmer Faust et al. 2017) Additionally, the persistence of free DNA released from lysed cells has been shown to affect the accuracy of molecular based enumeration methods (Seidel, Strathmann, and Nocker 2017; Walters, Yamahara, and Boehm 2009; Young et al. 2007) Without accounting for persistence or fr ee DNA, the qPCR

PAGE 48

41 values presented here may overestimate the concentration of viable Bacteroides cells ; nonetheless, the results still likely reflect relatively recent human fecal contamination. F ecal pollution from other animals (e.g., dog, cat) did not se em to be significantly contributing to the microbial community structure of Bear Creek and Cherry Creek water column and sediment samples based on s equence analyses (SourceTracker and diversity analyses ) SourceTracker did, however, identify that fecal ma terial from Canadian geese likely contributed a very small portion (<0.5% on average) of the overall bacterial community structure of the water column samples affecting 43% of sites. Similarly, a study using high throughput gene sequencing and SourceTracke r software also identified goose fecal matter as a secondary contributing source to a Lake Superior Estuary (Brown et al. 2017) Pigeon fecal matter represent ed 0.01% of the total community in only one sample (WC CHERRY 17). It is important to note that SourceTracker has shown more limited ability to identify secondary contaminant sources as compared to primary contaminant sources (Staley et al. 2018) indicating further testing may be necessary to confirm secondary sources of contamination in Denver waterways Additionally, a higher degree of intragroup variability observed in many animal fecal types may also have limit ability to identify these animals as contributing sources (Brown et al. 2017) These animal fecal results should be validated with additional testi ng, such as qPCR based MST. qPCR based MST for non human sources of animal fecal matter (e.g., dogs) are challenging since most animal specific qPCR assays are non specific or target animals not relevant in urban settings (Green et al., 2014, Ohad et al., 2016, Boehm et al., 2013). However, some studies have reported success with a goose specific qPCR assay (Fremaux et al., 2010). Future efforts should evaluate the specificity and sensitivity of this qPCR assay, and then apply the

PAGE 49

42 assay to water column samp les from Bear Creek and Cherry creek in order to confirm the presence of goose associated fecal bacteria in the streams. Interestingly, a local study on Upper Fountain Creek, Colorado inferred that birds may be the primary source of fecal pollution in the creek since human and ruminant animal fecal sources were not significant (Stoeckel et al., 2011)

PAGE 50

43 CHAPTER VI CONCLUSION Until recently, few studies have utilized both high throughput gene sequencing and qPCR as a community based MST approach in ambient waters. Here, we demonstrated that the combination of methods is transferrable to other systems and can strengthen MST conclusions. T he multifaceted approach showed that human fecal pollution likely contributes to the water column microbial communities in Bear Creek and Cherry Creek. Bear Creek showed lower levels of human associated Bacteroides 16S rRNA genes distributed evenly along the creek, suggesting that there are multiple locations where human fecal contamination is entering the creek (since Bacteroides abundance would otherwise be dilut ed over a gradient of several miles). Abundance of human associated Bacteroides 16S rRNA genes increased at downstream sites in Cherry Creek, suggesting that the primary sources of fecal pollution occur downstream of site CHERRY 07. Human fecal pollution c an enter waterways from a variety of sources including but not limited to leaky septic systems, urban storm water runoff, and wastewater treatment plant effluent discharge points. Human associated fecal pollution is a serious threat to human health, and th us further efforts are needed to address this issue in Denver waterways.

PAGE 51

44 REFERENCES Dependent and Library Independent Methods to Identify Human Faecal Pollution in Non Sewered Water Research 41: 3771 79. Letters in Applied Microbiology 46(2): 237 42. Associated Bacteroides Canadian Journal of Microbiology 56(10): 838 45. Genes of Bacteroides and Related Taxa for Identifying Sewage Pollution in Water 8(231): 1 27. Ruminant Feces on the Basis of Host Differences in Bacteroides Prevotella Genes Encoding 16S Applied and Environmental Microbiology 66(10): 4571 74. One Microbial Source Tracking Methods: A Twenty Water Research 47(18): 6812 28. http://dx.doi.org/10.1016/ j.watres.2012.12.046. Lake Michigan and Determination of Their Relationship to Escherichia Coli Densities Using Standard Microbiological Methods Detection of Genet ic Markers of Fecal Applied and Environmental Microbiology 71(12): 8305 13. Throughput DNA Sequencing Approach for Determining Sources of Fecal Bacteria in a Lake Superior Estua Environmental Science and Technology 51(15): 8263 71. Treatment Plants by a High Environmental Science and Technology 47: 543 3 41. Environmental Science & Technology 45(17): 7195 7201. http://pubs.acs.org/doi/abs/10.1021/es201118r. Water Research 47(18): 6862 72.

PAGE 52

45 2017. Determination of DNA Based Fecal Marker Aging Characteristics for Use in Quantitative Microbial Source Tracking of Two Library Independent Microbial Source Tracking Methods To Identify Sources Applied and Environmental Micr obiology 73(15): 4857 66. Identify Human and Ruminant Sources of Faecal Pollution in an Ephemeral Florida Journal of Applied Microbiology 113(1364 5072): 1396 1406. Epidemiology and infection 105(1): 139 62. Parametric Multivariate Analyses of Changes in Community Australian Journal of Ecology 18: 117 43. Department of Environmental Health. 2015. 2015 Denver Water Quality Update 2016 Water Quality Reports https://www.denvergov.org/content/denvergov/en/environmental health/environmental quality/water quality.html. Reveal Genetic Markers for Fecal Source Ide Applied and Environmental Microbiology 71(6): 3184 91. Environmental Science and Technology 46(8): 4340 47. Environ https://www.epa.gov/sites/production/files/2015 10/documents/csortc full.pdf. Eren, A. Murat, and Sandra L. McL Trends in Microbiology (12): 697 706. Microbial Sourc e Tracking: Methods, Applications, and Case Studies 379 97. http://link.springer.com/10.1007/978 1 4419 9386 1.

PAGE 53

46 Water Researc h 41(16): 3517 38. Bacteroides Spp. Applied and environmental microbiology 49(1): 148 150. http://aem.asm.org/content/49/1/148.short. Using Oligotyping Reveals Potential Human Fecal Indicators in Multiple Taxonomic Applied and Environme ntal Microbiology 81(20): 7023 33. Health Cost Estimates of Contaminated Coastal Waters: A Case Study of Gastroenteritis Environmental Sci ence and Technology 40(16): 4851 58. Tracking in Water by next Generation Sequencing Technologies Using Host Specific Environmental Sci ence and Technology 48(16): 9616 23. Time PCR Assay for Applied and Environmental Microbiology 80(10): 3086 94. Ha Water Research 44(18): 5463 73. http://dx.doi.org/10.1016/j.watres.2010.06.058. Hamuel, James Doughari, Alois Ndakidemi Pat rick, Susan Human Izanne, and Benade O157: H7 Escherichia Coli Isolates Obtained from Water and Wastewater Samples in African Journal of Biotechno logy 10(64): 14160 68. http://academicjournals.org/journal/AJB/article abstract/F24DB4036269. Water Research 93: 242 53. htt p://dx.doi.org/10.1016/j.watres.2016.02.029. Ishii, Satoshi, Winfried B. Ksoll, Randall E. Hicks, and Michael J. Sadowsky. 2006. Applied and Environmental Microbiology 72(1): 612 21.

PAGE 54

47 Microbial Source Tracking: Methods, Applications, and Case Studies 301 12. http://link.springer.com/10.1007/978 1 4419 9386 1. Indicator Bacteria and Microbial Source Tracking Genetic Markers in the Upper App lied and Environmental Microbiology 80(13): 3952 61. Environmental Microbiology Reports 114(4): 587 92. Detectable Bacteroides Distasonis from Human Applied and Environmental Microbiology 64(10): 4103 5. Based Real T ime PCR Assays for Estimation of Total, Human, and Bovine Fecal Pollution in Applied and Environmental Microbiology 72(6): 4214 24. System in an Unchlorinated D rinking Water System by SourceTracker Using Microbial Water Research 138. https://www.sciencedirect.com/science/article/pii/S0043135418302392. Mattioli, Mia Catharine, Lauren M. Sassoubre, Todd L. Russell, and Alexandria B. Boehm. Sourced Microbial Source Tracking Markers and Fecal Water Research 108: 106 14. http://dx.doi.org/10.1016/j.watres.2016.10.066. f Sewage Derived Environmental Microbiology 12(2): 378 92. Environmental Microbiology 15(8): 2213 2 7. Journal of Environmental Management 73(1): 71 79. s to Microbial Source Tracking in Tropical MicrobiologyOpen 3(6): 860 74.

PAGE 55

48 App lied and Environmental Microbiology 77(19): 6972 81. Microbial Ecology 65(4): 1011 23. mBio 6(2): 1 9. Set for Identifying 7(February): 1 8. with PCR Based Methods for Better Characterization of Diffuse Sources of Human Environmental Science & Technology 45: 5652 59. Data Processing and Web Nucleic Acids Research 41: 590 96. Offs of Swim Closures at Environmental Science & Technology 38(10): 2737 45. http://pubs.acs.org/doi/abs/10.1021/es034905z. Ramanathan, Bhargavi, Andrew M. Boddicker, Timberley M. Roane, and Annika C. Mosier. n Sediments Impacted by Acid Mine Frontiers in Microbiology 8(November): 1 16. http://journal.frontiersin.org/article/10.3389/fmicb.2017.02136/full. Santo Domingo, Jorge W., Dustin G. Bambic, Thomas A. Edge, and Stefan Wuertz. 2007. Source Tracking? Towards a Strategic Framework for Environmental Water Research 41: 3539 52. Live Dead Distinction in qPCR Based Micro Journal of Microbiological Methods 140(June): 23 31. http://dx.doi.org/10.1016/j.mimet.2017.06.013. Drains Are Sources of Human Fecal Polluti on during Dry Weather in Three Urban Environmental Science & Technology 43(2): 293 98. Associated Fecal Source Identification Quantitative Real Applied and Environmental Microbiology 82(9): 2773 82.

PAGE 56

49 Applied and Environmental Microbiology 82(9): 2773 82. Snyder, Blake J, Marirosa Molina, and Ourania Georgaco AND TRENDS OF E COLI, HUMAN ASSOCIATED BACTEROIDES AND Waters Impacted by Human and Non Water Research 44: 4674 91. http://dx.doi.org/10.1016/j.watres.2010.06.049. Wa ter Research 108: 330 38. http://dx.doi.org/10.1016/j.watres.2016.11.007. Fecal Pollution in Recreational Freshwater: A Double Environmental Scie nce & Technology : acs.est.7b05401. http://pubs.acs.org/doi/10.1021/acs.est.7b05401. Nucleic Acids Research 43(D1): D593 98. Quantitative Evaluation of Fecal Contamination Potential by Human and Ruminant Water Research 45: 3225 44. http://dx.doi.org/10.1016/j.watres.2011.03.037. Environmental Science and Technology 44(20): 7777 82. US EPA. United States Environmental Protection Agency,Office of Water Regulations and Standards Criteria and Standards Division (January): EPA440/5 84 002. Nucleic Acid Markers of Health Relevant Organisms in Seawater Microcosms: Water Research 43(19): 4929 39. http://dx.doi.org/10.1016/j.watres.2009.05.047. Communities Leads to Alternative Community PLoS ONE 5(6).

PAGE 57

50 Microbiome Viewed across Age and Nature 486(7402): 222 27. Journal of Endodontics 33(12): 1417 20. Zimmer shwater Sediment Characteristics on the Persistence of Fecal Indicator Bacteria and Genetic Markers within a Southern Water Research 119: 1 11.