FACTORS ASSOCIATED WITH CONCENTRATION S OF PERSISTENT ENTERIC MARKER S IN WATER QUALITY SAMPLES AND SEDIMENT CORES FROM THE LAKE ST. CLAIR WATERSHED By Yolanda Marie Brooks A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Microbiology and Molecular Genetics - Doctor of Philosophy 2015 ABSTRACT FACTORS ASSOCIATED WITH CONCENTRATION S OF PERSISTENT ENTERIC MARKER S IN WATER QUALITY SAMPLES AND SEDIMENT CORES FROM THE LAKE ST. CLA IR WATERSHED By Yolanda Marie Brooks M olecular methods to monitor water quality can address current and historical pollution . M olecular measurements of enteric markers from sediment s can aid in the evaluation of historic al water quality using a singular index and be used to analyze correlations to climate and human impact. However, the stability of the markers in defined storage conditions and durations is uncertain. The goals of this dissertation were to : 1) investigate how storage conditions and duration affected concentrations of enteric molecular marker s in water sample s; and 2) evaluate the correlations of historic al anthropogenic and climate variables with the deposition of persistent enteric markers in s ediment cores from the Lake St. Cl air watershed. Autoclaved water from the Red Cedar River was seeded with 10% (vol/vol) raw sewage and stored in liquid suspension (LS) or attached to a solid matrix (SM). Enterococci (ENT) 23S rDNA , Escherichia coli (EC) uidA , and Bacteroides thetataiotao micon (BT) 1 , 6 alpha - mannanase were measured with quantitative polymerase chain reaction (qPCR) in order to evaluate their persistence for up to 28 and 366 days at 4º (long term and short term studies), and 27º and 37ºC (short term study only). Five linear and non - linear best - fit models were fit to the i ndicator concentrations . P ersistence of the indicators was enhanced on SM (p < 0.001) , and decreased with time (p < 0.001) . Persistence was also dependent on i ndicator species in the short term and long term studies (p < 0.001, and p = 0.001, respectively). The least to most persistent indicator s w ere : BT < EC < ENT . The time needed for 90% decay of the indicators, T 90 , calculated with the best - fit models in the short te rm study (and long term study) ranged from 1 day for BT in LS > 28 days for ENT and EC on SM at all temperatures (and 35.8 days for EC in LS to 164 days for ENT on SM at 4°C ). At 4 ° C, the T 90 values were greater in the long term study compared to the short term study. This study suggests that storage of water samples at 4°C attached to a solid matrix can increase the persistence of markers from fecal indicators. Surface sediments from Anchor B ay, northwestern Lake St. Clair (AB), and the mouth of the Clinton River (CR) were spiked with EC and Enterococcus faecium . qPCR measurements of ENT 23S rDNA and EC uidA extracted from 17 DNA extraction methods were compared. Within each location, Kruskal - Willis tests confirmed few significant differences betwee n the concentrations of the indicators . T he optimal method included a bead beating step with a DNA sorption blocker followed by centrifugation. This method evaluated the concentrations of ENT 23S rDN A and EC uidA i n sediment cores from AB and CR, representing the years c. 1757 - 2012, and c. 1895 - 2012, respectively. EC concentrations in the AB and CR cores increased with year, and ranged from 1.42 x 10 6 to 16. 9 x 10 6 cell equivalents (CE) per g - dry wt, and 1.81 x 10 6 to 8.46 x 10 6 CE per g - dry wt, respectively. ENT concentrations in the CR core increased with year, and ranged from 3 x 10 3 to 9 9 0 x 10 3 CE per g - dry wt. The ENT concentrations in the AB core experienced two steady states: ~1 x 10 4 , and ~2 x 10 5 CE per g - dry wt during c. 1757 c. 1878, and c. 1902 c. 2010, respectively. ENT concentrations in both cores were correlated to river discharge (p = 0.046), while EC concentrations were correlated to air temperature (p = 0.018), and total nitrogen and total carbon concentrations (p = 0.038, and 0.029, respectively). Also, ENT and EC concentrations were significantly correlated to population in watershed (p = 0.003 and 0.023, respectively). This study offers a novel a management of the Clinton River watershed. Copyright by YOLANDA MARIE BROOKS 2015 v To my supportive family, both given and chosen. Thank yo u for always being there for me . vi ACKNOWLEDGMENTS I would like to acknowledge the following people that have helped me complete my dissertation: Joan Rose for her guidance during this project; Rebecca Ives for all of her help and dedication in the laboratory; and my collaborators Asli Aslan, Tiong Gim Aw, Mark Baskaran, Melissa Baustian, Jade Mitchell, Bharathi Murali, Nathaniel Ostrom and Sushil Tamrakar. Finally, I would like to thank my labmates. This dissertation would not have been possible without your help. This research was funded by the following organizations: NOAA, NSF, and the Graduate School and Department of Microbiology and Molecular Genetics at Michigan State University. vii TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ........................ x LIST OF FIGURES ................................ ................................ ................................ .................... xii CHAPTER 1. WATER POLLUTION AND WATER QUALITY ................................ .......... 1 1.1 Fecal indicator organisms ................................ ................................ ................................ ......... 2 1.1.1 Total coliforms and fecal coliforms ................................ ................................ ............... 2 1.1.2 Escherichia coli ................................ ................................ ................................ .............. 3 1.1.3 Enterococci ................................ ................................ ................................ ..................... 4 1.2 Fecal indicators and their associations to health outcomes in recreational waters ................... 6 1.3 Criteria and standards of fecal indicators concentrations in the United States ......................... 7 1.4 Microbial source tracking ................................ ................................ ................................ ......... 9 1.5 Factors affecting t he concentration of enteric bacteria in the environment ............................ 13 1.5.1 Factors in environmental waters ................................ ................................ ................... 13 1.5.2 Factors in sediments ................................ ................................ ................................ ..... 14 1.6 Scientific Needs ................................ ................................ ................................ ...................... 15 1.7 Research Objectives ................................ ................................ ................................ ................ 16 1.7.1 Goal 1 ................................ ................................ ................................ ........................... 16 1.7.2 Goal 2 ................................ ................................ ................................ ........................... 17 CHAPTER 2. SHORT TERM PERSISTENCE ENTERIC BACTERIA I N WATER SAMPLES STORED ON A SOLID MATRIX AND IN LIQUID SUSPENSION ............... 18 2.1 Introduction ................................ ................................ ................................ ............................. 19 2.2 Methods ................................ ................................ ................................ ................................ ... 21 2.2.1 Water Sample Storage ................................ ................................ ................................ .. 23 2.2.2 Sample processing and DNA extraction ................................ ................................ ....... 23 2.2.3 qPCR quantification of three indicators ................................ ................................ ....... 24 2.2.4 S tatistical Analyses and Persistence Modeling ................................ ............................ 29 2.3 Results ................................ ................................ ................................ ................................ ..... 32 2.3.1 Summary of results ................................ ................................ ................................ ....... 32 2.3.2 Comparison of concentrations of total DNA after storage at 4°C. ............................... 33 2.3.3 Model comparisons for analyzing FIB DNA persistence ................................ ............. 35 2.3.4 Comparison of the three FIB and effects of at tachment and temperature. ................... 37 2.3.5 Multiple Linear Regression Analyses. ................................ ................................ ......... 40 2.4 Discussion ................................ ................................ ................................ ............................... 44 APPENDIX ................................ ................................ ................................ ................................ ... 51 CHAPTER 3. COMPARISON OF THE LONG TERM PERSISTENCE OF ENTERIC BACTERIA ATTACHED TO A SOLID MATRIX AND IN LIQUID SUSPENSION ....... 59 3.1 Introduction ................................ ................................ ................................ ............................. 60 3.2 Methods ................................ ................................ ................................ ................................ ... 61 3.2.1 Sample preparation and storage ................................ ................................ .................... 61 viii 3. 2.2 Sample processing and DNA extraction ................................ ................................ ....... 62 3.2.3 qPCR quantification of three indicators ................................ ................................ ....... 62 3.2.4 Persistence modeling and statistical analyses ................................ ............................... 63 3.3 Results ................................ ................................ ................................ ................................ ..... 65 3.3.1 Summary of molecular marker persist ence ................................ ................................ .. 65 3.3.2 Comparison of models estimating indicator persistence ................................ .............. 66 3.3.3 Comparison of predicted T 90 and T 99 values ................................ ................................ 68 3.3.4 Comparison of genetic marker persistence at t = 31 days in the persistence models and observed data in both storage conditions of our study and Chapter 2.3.3 ............................. 69 3.3.5 Linear regression analysis ................................ ................................ ............................ 70 3.4 Discussion ................................ ................................ ................................ ............................... 72 APPENDIX ................................ ................................ ................................ ................................ ... 76 CHAPTER 4. DNA YIELDS, AND CONCENTRATIONS OF ESCHERICHIA COLI UIDA AND ENTEROCOCCI 23S rDNA IN FRESHWATER SEDIMENTS: A COMPARISON OF DNA EXTRACTION METHODS ................................ ......................... 79 4.1 Introduction ................................ ................................ ................................ ............................. 80 4.2 Methods ................................ ................................ ................................ ................................ ... 81 4.2.1 Sediment sampling ................................ ................................ ................................ ....... 83 4.2.2 DNA extraction of spiked sediment with modified methods of MoBio UltraClean®, EPA - DNA2, and Mobio PowerSoil®. ................................ ................................ .................. 84 4.2.3 q PCR analysis of E.coli uidA and enterococci 23S rDNA. ................................ ........... 86 4.2.4 Kruskal - Wallis statistical analyses of DNA yields and concentration s of E.coli and enterococci produced from the modified DNA extraction methods from AB and CR sediments ................................ ................................ ................................ ............................... 88 4.2.5 Comparison of the DNA yields and FIB concentrations produced from the modified DNA extraction methods to their respective standard method ................................ .............. 89 4.3 Results ................................ ................................ ................................ ................................ ..... 89 4.3.1 Site specific comparison of the DNA yield produced by the modified DNA extraction methods ................................ ................................ ................................ ................................ .. 89 4.3.2 Site specific comparison of the qPCR amplified concentrations of enterococci and E.coli produced by the modified DNA extraction methods. ................................ ................. 92 4.3.3 Comparisons of DNA extraction efficiency of the modified DNA extraction methods ................................ ................................ ......... 93 4.4 Discussion ................................ ................................ ................................ ............................... 94 CHAPTER 5. HISTORICAL ASSOCIATIONS OF FECAL INDICATOR CONCENTRATIONS TO ANTHROPOGENIC ACTIVITIES AND CLIMATE IN FREHSWATER SEDIMENTS ................................ ................................ ................................ . 98 5.1 Introduction ................................ ................................ ................................ ............................. 99 5.2 Methods ................................ ................................ ................................ ................................ . 100 5.2.1 Field site description, sample collection and processing ................................ ............ 100 5.2.2 DNA Extraction and qPCR measurements of enterococci 23S rDNA and E.coli uidA ................................ ................................ ................................ ................................ ............. 102 5.2.3 Sediment chronostratigraphy ................................ ................................ ...................... 102 5.2.4 Nutrient meas urements ................................ ................................ ............................... 103 5.2.5 Measurements of the anthropogenic and climate data ................................ ............... 104 ix 5.2.6 Multiple linear regression analyses ................................ ................................ ............ 106 5.3 Results ................................ ................................ ................................ ................................ ... 106 5.3.1 Sedimentation rate and sediment chr onostratigraphy ................................ ................. 106 5.3.2 Climatic measurements: air temperature and discharge rates ................................ .... 107 5.3.3 Anthropogenic attributes: estimated census population and nutrient loading in the cores. ................................ ................................ ................................ ................................ .... 112 5.3.4 Fecal indicator concentrations in sediment cores ................................ ....................... 113 5.3.5 Linear regression analyses ................................ ................................ .......................... 114 5.4 Discussion ................................ ................................ ................................ ............................. 117 APPENDIX ................................ ................................ ................................ ................................ . 124 CHAPTER 6. CONCLUSIONS ................................ ................................ ............................... 129 6.1 Goals of the research and summary of the results ................................ ................................ 130 6.2 Correlations of the results of the long term persistence study to persistent concentrations of fecal indicators in sediment cores ................................ ................................ ............................... 133 6.3 Implications of the results of this dissertation and recommendations to water quality monitoring ................................ ................................ ................................ ................................ ... 134 6.4 Implications of results to watershed management and recommendations for management actions ................................ ................................ ................................ ................................ ......... 136 BIBLIOGRAPHY ................................ ................................ ................................ ..................... 138 x LIST OF TABLES Table 1.1: Marker name, species, and gene target of a selection of human associated Bacteroides spp. markers that are used in microbial source tracking and qPCR methods. .............................. 11 Table 1.2: Sequences and length of two human associated Bacteroides spp. markers. ............... 12 Table 2.1: Fecal indicator species, targe ted gene, primer sequences, probe sequences, and amplified sequence length (bp) of the qPCR genetic markers in this study. ................................ 27 Tab le 2.2: Average efficiency, R 2 l), and DNA extraction method specific detection limits from EC - uidA, ENT - 23, and BT - am from E.coli , enterococci and B.thetataiotaomicron , respectively. ................................ ........................ 28 Table 2.3: List of best fit models represented in the datasets, their equations, and properties. ... 31 Table 2.4: Selected models, T 90 values (days) and persistence parameters for E.coli (EC), enterococci (ENT), and B.thetataiotaomicron (BT) stored at 4°, 27°, and 37°C unde r two storage conditions, in suspension (LS) or attached (SM). ................................ ................................ ......... 37 Table 2.5: Multiple linear regression analyses determined how the observed fractional persistence ratios a , were influenced by: storage time measured in hours, storage conditions b , extraction method c , genetic markers d and storage temperature e . ................................ ............... 42 Table 2.6: A list of the coefficients in the linear regressio n equation a that evaluated the correlations of the independent variables to fractional persistence b . ................................ ........... 43 Table 2.7: Comparison o f genetic markers from Bacteroidales, E.coli , and enterococci in various storage conditions, and T 90 values from our study and previous studies. ................................ ..... 47 Table S2.1: Best fit equations and their references that evaluated the persistence of the indicators in this study. ................................ ................................ ................................ ................................ .. 52 Table S2.2: The best fit models, and the equations that represented the persistence patterns of the three indicators a stored at three temperatures, and two storage conditions b . ........................ 54 Table S2.3: BIC values and the standard deviations of the 17 best - fit models analyzed for best fit of the indicators a that were stored in two conditions b at three temperatures for u p to 28 days. . 56 Table 3.1: The qPCR reaction efficiency, CT values of the lowest detected dilutions (and representative copies/5 l), and detection limits of EC - uidA, ENT - 23, and BT - am a . ................ 63 Table 3.2: Descriptions of the parameters, BIC values, predicted T 9 0 and T 99 values (days) from the models that evaluated the persistence of cell equivalents three indicators a in two storage conditions b at 4 ° C for up to 366 days. ................................ ................................ .......................... 68 xi Table 3.3: Comparison of the fractional persistence of cellular equivalents, Escherichia coli (EC), enterococci (ENT), and Bacteroides thetataiotaomicron (BT), quantified in our study and Chapter 2.3.3 at N 31 /N 0 and N 28 /N 0 , respectively. ................................ ................................ ....... 70 Table 3.4: Correlation coefficient and p - values of multiple linear regression analyses that evaluated the association of the fractional persistence of the indicators a to: storage time (days), storage condition b , and indicator species c . ................................ ................................ .................. 71 Table 3.5: A list of the coefficients of the linear regression equation a , that evaluated the correlations of the independent variables to fractional persistence b . ................................ ........... 72 Table S3.1: BIC values and the standard deviations of the 17 best - fit models that analyzed the persistence of genetic markers from three indicators a that were stored in two conditions b at 4 ° C for up to 366 day s. ................................ ................................ ................................ ........................ 77 Table S3.2: The best fit equations that were chosen to represent the persistence patterns of threes indicators a stored at three temperatures, and two storage conditions b . ................................ ....... 78 Table 4.1: Calculated detection limits of the qPCR amplified cellular equivalents of E.coli and enterococci extracted with the modified DNA extraction methods a in 20 µl qPCR reaction volumes from sediments from the Clinton River (CR) and Anchor Bay (AB). ........................... 87 Table 4.2: Within each sediment location, Anchor Bay and Clinton River, a ratio compared the average DNA yields, and concentrations of enterococci and E.coli produced from modifications ................................ .... 94 Table 5.1: Linear regressions evaluated the correlation strengths (and p - values) of climatic and anthropogenic variables a to 1) enterococci concentrations (ENT) b,c , C ENT ; and 2) E.coli concentrations (EC) b,d . ................................ ................................ ................................ ............... 115 Table 5.2: The coefficients and their standard errors of the climatic and anthropogenic varia bles in linear regression equations a that evaluated their associations to: 1) enterococci concentrations (ENT) b,c , C ENT ; and 2) E.coli concentrations (EC) b,d , C EC . ................................ ....................... 116 Table S5.1: Concentrations of Escherichia coli , enterococci, total phosphorus, % total nitrogen and % total carbon measured in the Anchor Bay sediment core. ................................ ............... 125 Table S5.2: Concentrations of Escherichia coli , enterococci, total phosphorus, % total nitrogen and % total carbon measured from the Clinton River sediment core. ................................ ........ 127 xii LIST OF FIGURES Figure 2.1: Overview of experimental design. ................................ ................................ ............ 22 Figure 2.2A - B: Total DNA concentrations a suspension (LS) and B) solid matrix (SM), and eluted from three DNA extraction methods b . .. 35 Figure 2.3A - F : Observed and predicted fractional persistence a of three indicators b stored for up (A - B) (C - D) (E - F) stor age temperatures in two storage conditions c . ................................ ................................ ................................ ................................ ... 39 Figure 3.1A - B: The observed and predicted fractional persistence a of three indic ators b stored for up to 366 days at 4°C A) in suspension (LS); or B) attached to a solid matrix (SM). ............ 66 Figure 4.1: Overview of experimental design. ................................ ................................ ............ 82 Figure 4.2A - F: A - B) DNA yields, and qPCR quantified concentrations a of C - D ) enterococci (ENT), and E - F ) Escherichia coli (EC) spiked into surface sediment from Anchor Bay (AB) and the Clinton River (CR), and extracted by modifications b of three DNA extraction methods c . .. 91 Figure 5.1: Map of the Lake St. Clair basin, and the sediment core sites labeled with stars: Anchor Bay (AB), and the Clinton River (CR). ................................ ................................ ......... 101 Figure 5.2: Cs - 137 radio isotope activity profile (dpm/g - dry wt) of the sediment cores from e mass depth of the cores (g - dry wt/cm2). ................................ ................................ ................................ ........ 107 Figure 5.3A - H: Indictors of climatic variables a (A - B) , population in the watershed b (C) , sedimentary nutrient concentrations c (D - F) , and fecal indicator concentrations d (G - H) in the Anchor Bay watershed. ................................ ................................ ................................ ............... 109 Fig ure 5.4A - H: Indictors of climatic variables a (A - B) , population in the watershed b (C) , sedimentary nutrient concentrations c (D - F) , and fecal indicator concentrations d (G - H) in the Clinton River watershed. ................................ ................................ ................................ ............ 111 Figure 5.5A - B: Illustrations of the patterns of discharge of the St. Clair River (A) and the Clinton River (B) , anthropogenic variables b , sedimentary nutrient concentrations c , and fecal indicator concentrations d from the Anchor Bay e (A) and Clinton River e (B) watersheds. ...... 120 1 CHAPTER 1. WATER POLLUTION AND WATER QUALITY 2 1.1 Fecal indicator organisms Access to clean water is a basic human right (United Nations, 2010) . However, contact with and ingestion of polluted water can cause respiratory, gastrointestinal, ear, and eye illnesses. Therefore, water quality monitoring is essential in order to mitigate the risk of waterborne disease in drinking water and recreational water. Water quality monitoring is achieved by measuring concentrations of fecal indicator organisms in water. Fecal indicators are enumerated in place of pathogens because culturing multip le pathogens is difficult and time consuming (USEPA, 2012a) . The World Health Organization (WHO) defined the following ideal characteristics of fecal indicator organisms: 1) non - pathogenic viruses and bacteria that are universally found in animal and/or human excreta; 2) do not grow in the environment; 3) have similar persistence compared to pathogens; 4) are in higher concentration s than pathogens; and 5) have methods that are easy and cost - efficient to numerate (World Health Organization, 2011) . A variety of fecal indicators are recommended by the WHO , and they include: total coliforms, fecal coliforms, sulfite - reducing clostridia, F - specific RNA bacteriophages, somatic coliphages, total bacteriophage, Bacteroides fragilis bacteriophages, Escherichia coli and enterococci (World Health Organization, 2003, 2011) . However, there is no single fecal indicator that meets all of the characteristics listed above. Below are descriptions of four fecal indi cators. 1.1.1 Total coliforms and fecal coliforms Total coliforms are a group of bacteria that are associated with fecal pollution (USEPA, 2011) . They include taxonomically unrelated species th at are gram - negative, facultative anaerobes, non - spore forming, rod shaped , grow at 36 ° C, and produce gas while fermenting 3 lactose (Madigan, Martinko, Dunlap, & Clark, 2009) . Not all total coliforms represent fecal pollution. Some species have natural habitats in the environment. Within the total coliform group, is a subset called fecal coliforms. They produce acid in lactose fermentation when grown at 44 ± 0.5°C, grow in the presence of bile salts , and are oxidase negative (Doyle & Erickson, 2006) . However, s ome species are found naturally in the environment (USEPA, 2012a) . 1.1.2 Escherichia coli Escherichia coli is a species within the fecal coliform group. It is a popular fecal indicator for recreational water and drinking water (World Health Organization, 2011) because its primary environment is in the intestinal tract of warm blooded animals (Doyle & Erickson, 2006) . Its mutualistic relationship provides the host with vitamin K 2 , while the host provides a nutrient rich and stable hab itat ( Suvarna, Stevenson, Meganathan, & Hudspeth, 1998) . Although most E.coli strains are beneficial, the re are seven pathogenic groups including: enteropathogenic, enterotoxigenic, enteroinvasive, enteroaggregative, diffusely adherent , and shiga toxin produc ing (Brenner & Farmer III, 2005) . The pathogeni c strains can cause urinary tract infections, pneumonia, respiratory illnesses and diarrhea (Centers for Disease Control, 2011) . Also, s ome pathogenic strains are multi - drug resistant (Centers for Disease Control, 2011) . Many standard methods to enumerate culturable E.coli in water have a single carbon source, glucuronide is metabolized by an enzyme specific to E.coli , D - glu curonidase (Hansen & Yourassowsky, 1984) , that is encoded by uidA (USEPA, 2002 ) . In its natural habitat, D - glucuronidase glucuronides, into alcohols and glucuronic acid for cellular metabolic energy (USEPA, 2002) . 4 1.1.3 Enterococci T he re are 28 enterococci species . They are facultative anaerobes, gram - positive, cocci, catalase - negative, form no gas with lactose f ermentation , and form pairs or short chains (Facklam, Carvalho, & Teixeira, 200 2) . Enterococci can hydrolyze esculin, and can optimally grow in media with 40% bile salts and 6.5% NaCl (Fisher & Phillips, 2009; G. Klein, Pack, Bonaparte, & Reuter, 1998) . Some enterococci naturally inhabit plant matter (Fisher & Phillips, 2009) , while others are commensal to the human oral cavity, the human vagina , and animal gastrointestinal tracts and fecal matter (Jett, Huy cke, & Gilmore, 1994) . Enterococcus faecalis and Enterococcus faecium are the first and second most common enterococci isolated from humans , respectively , but are in lesser prevalence in livestock (Franz, Holzapfel, & Stiles, 1999) . Enterococci have been associated with community - acquired and nosocomial infections like urinary tract infections, wound infections, bacteraemia, sepsis , and endocarditis (Poh, Oh, & Tan, 2006) . Enterococcal , and infect immuno - compromised people (Murray, 1990) . Treating enterococc al infections are difficult because many strains are multi - drug resistant (Ryan, Ray, Ahmad, Drew, & Plorde, 2010) . The standard methods to enumerate culturable enterococci contain two carbon/energy sources, esculin and its chromogenic analog, which select for enterococci growth (USEPA, 2009) . The incubation temperature, 41°C, also selects for its growth (USEPA, 2009) . A molecular method, Method 1611, is recommended by the United State Environmental Protection Agency (EPA) to enumerate enterococci 23S rRNA , Entero1, in water samples within 3 - 4hrs with quant itative polymerase chain reaction (qPCR, USEPA, 2012b) . The primer and 5 probe sequences are: GAGAAATTCCAAACGAACTTG, CAGTGCTCTACCTCCATCATT, and [6 - FAM] - 5' - TGGTTCTCTCCGAAATAGCTTTAGGGCTA - TAMRA, respectively (USEPA, 2012b) . 23S rRNA is a bacterial ribozyme, peptidyltransferase, in the large subunit of the ribosome (Snyder & Champness, 2007) . The gene is highly conserved and is essentia l for bacterial growth and development (Snyder & Champness, 2007) . Although, the copy number of 23S rRNA is dependent on the species and the metabolic state , the suggested copy number is four, which is the average copy number of Enteroc occus faecalis 23S rRNA (USEPA, 2012b) . The detection limit is about 27 cell equiva lents (CE) or ~10 2 Entero1 sequences per reaction (Haugland, Siefring, Wymer, Brenner, & Dufour, 2005) . There are many disadvantages of assessin g water quality with enterococci and E.coli . Their culturable methods need an incubation period between 18 - 24 hrs, which may not allow for fast response to unacceptable concentrations. Also, false - negatives and false - positives can occur with the culturable and qPCR methods. False - negatives in the culturable methods can be attributed to the lack of a functional target enzyme in subpopulations of the fecal indicators, while false - negatives obtained in qPCR methods may be attributed to qPCR inhibitors or the d etection limit. The false - positives obtained in culturable methods may be due to targeted enzyme activity from other species. There are few reported false - positive results in qPCR methods (Griffith, Weisberg, & McGee, 2003) . Another limitation of culturable and qPCR methods to enumerate E.coli and enterococci represent general pollution and do not identify the host. E.coli and enterococci can also persist and even grow in the environment (Pote et al., 2009; Wheeler Alm, Burke, & Spain, 2003) . Growth of fecal indicators in the environm ent may warrant unnecessary beach warnings and 6 closures. Additionally, concentrations of fecal coliforms and enterococci were not associated with concentrations of Salmonella spp., and Cryptosporidium spp. in wastewater effluent and freshwater (Lemarchand & Lebaron, 2003) . 1. 2 Fecal i ndicators and their associations to health out comes in recreational waters Contact with recreational waters can cause various types of adverse health outcomes affecting the following areas of the human body: gastrointestinal tract, ear, respiratory tract, eyes, and skin. During 2009 - 2010, there were 8 1 water associated disease outbreaks in the United States with a total of 1,326 cases affecting 28 states and Puerto Rico that were reported to the Centers for Disease Control (CDC, Hlavsa et al., 2014) . However, the actual number of water associated disease outbreaks in the United States is suspected to be larger as local public health departments voluntarily report outbreaks to the CDC (Blackburn et al., 2004) . Beginning i n the 1950s, epidemiological studies began to investigate the relationship of swimming in recreational waters to adverse health outcomes . For example, one study determined that s wimmers were at a higher risk to experience gastrointestinal, skin, respiratory, eye , and ear illnesses compared to non - swimmers in the following freshwater and marine beaches: Lake Michigan , Illinois; Ohio River , Kentucky; and Long Island Sound , New York (Stevenson, 1953) . Further studies have examined the dose - response relationship between health outcomes of swimmers and the w ater quality of recreational water experiencing various sources and levels of pollution. For example, culturable E.coli and enterococci were associated with the incidence of gastrointestinal illness from beaches on Lake Erie, Pennsylvania and Keystone Lake , Oklahoma (Dufour, 1984) . M eta - analys e s of freshwater and marine water determined that there wa s a 7 significant association between concentrations of enterococci to gastrointestinal illn ess inc idence (Wade, Pai, Eisenberg, & Colfor d, 2003) . Compared to concentrations of total coliforms and fecal coliforms, enterococci concentrations were better associated with the risk of gastroent eritis in marine beaches in New York polluted with point sources (Cabelli, Dufour, McCabe, & Levin, 1982) . Meta - analyses of recreational freshwater s also determined that concentrations of E.coli were shown to be a better indicator of risk of gastrointestinal illness compared to concentrations of enterococci and fecal coliforms (Prüss, 1998; Wade et al., 2003) . A comparison of qPCR methods determined that measurements of Entero1 w ere better correlated than Bacteroidetes 16S rRNA to waterborne illness incidence on beaches at Lake Michigan, Michigan and Indiana; and Lake Erie, Ohio (Wade et al., 2006) . Additionally, compared to concentrations of culturable enterococci, measurements of Entero1 had a stronger correlation to gastrointestinal illness incidence of children visiting beaches near point sources of treated sewage at Lake Michigan, Michigan a nd Indiana; and Lake Erie, Ohio (Wade et al., 2008) . In marine beaches exp eriencing non - point pollution at Mission Bay, California, the incidence of adverse health outcomes was better associated to the concentration of male - specific coliphage compared to culturable enterococci, Entero1, total coliforms and fecal coliforms (Colford et al., 2007) . The risk of adverse health outcomes reported from a marine beach in Santa Monica Bay, California, with untreated storm runoff inputs was a ssociated to concentrations of either culturable E.coli and enterococci (Haile et al., 1999) . 1.3 Criteria and standards of fecal indicators concentrations in the United States Decades of epidemiolo gical studies investigating the association between adverse health outcomes and fecal indicator concentrations have guided the formation of the water quality 8 criteria for recreational waters in the United States. The EPA established criteria for the maximu m acceptable concentrations of E.coli and/or enterococci concentrations in recreational freshwater and marine water. Recommended criteria were set to estimated illness rates of 32 or 36 cases per 1000 people that have primary contact with recreational wate rs. The recommended 30 day geometric mean for an estimated illness rate, 32/1000 (or 36/1000), is 100 colony forming units (or 126 colony forming units, CFU) of E.coli per 100ml in freshwater and 30 CFU (or 35 CFU) of enterococci per 100ml in freshwater an d marine water (USEPA, 2012c) . The criteria are enforced by the states, territories and tribal governments. Also, the state s, territories, and Native America n tribal lands can strengthen regional areas can have unique epidemiology situations that affect the association between risk of adverse health outcomes and fecal indicator concentrations in recreational waters (López - Pila & Szewzyk, 2000) . Specifically, Michigan monitors the water quality of its recreational waters at least five times during a 30 day period during peak recreational usage (May - October). Water samples are taken from waist deep water in at least three locations in a beach area. The highest acceptable fecal indicator concentratio ns for full body exposure is set at a geometric average of 130 CFU of E.coli per 100ml over a 30 day period, or 300 CFU of E.coli per 100ml for measured in a single day (MDEQ, 2012) . Unacceptable concentrations of fecal ind icators the can cause beach closures or beach advisories. In 2013, there were 413 beaches that were monitored in Michigan (MDEQ, 2013) . Only 3.3% of all of the samples taken at these beaches exceeded the water qualit y standard for a single day measurement (MDEQ, 2013) . 9 1.4 M icrobial source tracking Conventional fecal indicator organisms cannot identify the source of fecal pollution in water samples. Identifying the species of f ecal waste in recreation water is important because different origins of pollution are suggested to pose varying risks of adverse health effects (Soller, Schoen, Bartrand, Ravenscroft, & Ashbolt, 2010) . Therefore, M icrobial source tracking (MST) methods were developed in orde r to identify the source of fecal pollution. Host specific library independent methods are a subset of MST methods. These methods target genetic elements or chemical markers such as sterols, pharmaceuticals, detergents, and other chemicals associated with anthropogenic pollution (Santo Domingo, Bambic, Edge, & Wuertz, 2007) . The pres ence of host associated markers in library independent MST methods is confirmed by polymerase chain reaction (PCR), while qPCR enumerates host - specific markers. qPCR methods follow a basic protocol, which includes collection of a water sample, filtration, DNA extraction, and qPCR quantification (Shanks et al., 2010) . Once host - specific markers are identified, they undergo a validation process that determines the specificity, sensitivity, and quantities of the marker in non - host and host fecal ma tter (Harwood, Staley, Badgley, Borges, & Korajkic, 2014) . However, there is lack of evidence that any marker has been completely validated. qPCR methods to measure host - specific pollut ion are preferred over conventional methods for many reasons. qPCR methods have a low detection limit , high specificity, and quickly return results (Santo Domingo et al., 2007) . A single sample can be anal yzed for multiple indicators. The DNA extracts can be stored and used for another indicator analysis as long as DNA degradation can be accounted for (Santo Domingo et al., 2007) . qPCR methods can quantify host - specific pollution independent of viability . However, many of the markers have cross - reactivity to other species (Shanks et al., 2010) . Additionally, a negative result does 10 ontain the host - specific pollution, but that its concentration is undetectable due to the detection limit, die - off, or inhibitors (Field & Samadpour, 2007) . Additionally, the concentration s of some human - specific markers in primary sewage effluent can be as low as 1,000 copies per ng - DNA (Shanks, Kelty, Sivaganesan, Varma, & Haugland, 2009) , which can be further diluted to undetectable concentrations in recreational waters. Bacteroides spp. is a popular genus for genetic markers enumerated in lib rary - independent MST methods. Th e genus belongs to the phylum Bacteroidetes , class Bacteroidia , order Bacteroidales and family Bacteroidaceae (Boone et al. 2001). The genus is made up of 42 known obligate anaerobic gram - negative species. Bacteroides spp. live in the intestinal tract of some birds and most mammals , and are excreted in fecal matter (Madigan et al., 2009) . They are the dominant commensal organisms in the large intestine s of humans because of their ability to ferment sugars and proteins (Madigan et al., 2009) . Markers have been developed to measure human ( Shanks et al., 2009) , bovine (Layton et al., 2006) , porcine (Mieszkin, Furet, Corthier, & Gourmelon, 2009) , and bird associated pollution (Green, Dick, Gilpin, Samadpour, & Field, 2012) . Table 1 .1 outlines the name and gene target of a selection of human associated Bacteroides spp. markers. Many of the markers highlighted in Table 1 .1 target human associated Bacteroides spp. 16S rRNA (Converse, Blackwood, Kirs, Griffith, & Noble, 2009; Layton et al., 2006; Reischer, Kasper, Steinborn, Farnleitner, & Mach, 2007; Seurinck, Defoirdt, Verstraete, & Siciliano, 2005) . There are many advantages to using Bacteroide s spp. markers to measure host associated pollution. For example, Bacteroides spp. do not experience extensive growth in the environment 11 (Ballesté & Blanch, 2010) . One study suggested that fecal associated Bacteroides spp. 16S rRNA are of higher epidemiological importance because they better correlate to water borne illness incidence in four Great Lakes beaches located near treated wastewater outflows (Wade et al., 2008, 2010) . Human associated Bacteroides spp. markers such as Human - Bac1 show promise because their concentrations better predicted the presence of Salmonella spp., and E.coli O - 157 compared to concentrations of fecal coliforms and total coliforms in freshwater receiving various types and concentrations of point and non - point pollution (Savichtcheva, Okayama, & O kabe, 2007) . A description of the primer and probe sequences, and amplified sequence lengths of two human associated Bacteroides spp. markers, HF183 and BT - am, are in Table 1.2 . HF18 and BT - am further described below. Table 1 . 1 : Marker name, species, and gene target of a selection of human associated Bacteroides spp. markers that are used in microbial source tracking and qPCR methods. Marker name Gene target Reference HumM2 Bacteroides spp. h ypothetical protein, BF 3236 Shanks et al. ( 2009) HumM3 Bacteroides spp. putative RNAP sigma factor HuBac Bacteroides spp. 16S rRNA Layton et al. ( 2006) HF183 Seurinck et al. ( 2005) Human - Bac1 Okabe, Okayama, Savichtcheva, & Ito ( 2007) BacH Reischer et al. ( 2007) BFD Converse et al. ( 2009) BT - am - 1 - 6 mannanase Yampa ra - Iquise, Zheng, Jones, & Carson ( 2008) Bf Bact eroides fragilis gyrB Lee & Lee ( 2010) 12 Table 1 . 2 : Sequences and length of two human associated Bacteroides spp. markers. Marker Name - Sequence length (bp) Reference HF183 ATCATGAGTTCACATGTCCG 82 Seurinck et al. ( 2005 ) TACCCCGCCTACTATCTAATG SYBR Green I BT - am CATCGTTCGTCAGCAGTAACA 63 Yampara - Iquise et al. ( 2008 ) CCAAGAAAAAGGGACAGTGG FAM - ACCTGCTG - NFQ One popular marker is Bacteroides spp. 16S rRNA , HF183 (Seurinck et al. 2005). The species origin of HF183 is unknown (Seurinck et al., 2005) . 16S rRNA is a highly conserved gene that encodes a non - protein t hat help s shape the ribosome , and initiate protein synthesis (Madigan et al . 2009). Bacteroides spp. 16S rRNA has multiple cop ies per genome (Xu et al., 2003) , and thus improves the detection limit of HF183 . Although HF183 is associated with fecal p ollution of human origin, it was amplified from chicken , and dog feces (Seurinck et al., 2005; Shanks et al., 2010) . In freshwater microcosms spiked with human wastewater and stored at 25 ° C, HF183 persistence decreased with sediment presence and sunlight, and was enhanced at 15 ° C storage and reduced predation (Dick, Stelzer, Bertke, Fong, & Stoeckel, 2010) . HF183 decayed fas ter than cultivable E.coli in freshwater microcosms spiked with human feces (Liang et al., 2012) , which indicate s that HF183 may only represent recent fe cal pollution . However, culturable en terococci decayed faster than HF183 in marine water microcosms spiked with sewage (Walters, Yamahara, & Boehm, 2009) . It is suggested that rel iable genetic marker s for MST methods are from genes that are involved in a specific host - microbe interaction (Carson et al., 2005) , such as Bacteroides thetataiotaomicron - 1 - 6 mannanase, BT - am (Yampara - Iquise et al., 2008) . B.the tataiotaomicron is dominant in human fecal matter, but present in smaller concentrations in animals (Kreader, 1998) . Bt - am is a single genome copy number (Yampara - Iquise et al., 2008) . 13 Also, - 1 - 6 mannanase encod es a glycosylhydrolase that metabolizes plant based dietary mannose polysaccharides into mannose monosac charides for host consumption (Xu et al., 2003) . BT - am has cross reactivity in gulls, swine, and cat feces, bu t has greater host specificity than HF183 (Aslan & Rose, 2013) . Concentrations of BT - am were significantly associated to culturable E.coli and enterococci throughout wastewater treatment processes (Srinivasan, Aslan, Xagoraraki, Alocilja, & Rose, 2011) . However, few studies have evaluated its persistence in the environment. 1. 5 Factors affecting the concentration of e nteric bacteria in the environment 1.5.1 Factors in environmental waters Accurately and precisely measuring fe cal indicators in environmental waters is paramount to monitoring the water quality of recreational waters. Therefore, it is crucial to investigate the factors that affect the survival of enteric bacteria in the environment. For example , fecal coliforms su rvived longer in freshwater - sediment mesocosms spiked with wastewater , compared to enterococci, while the opposite was true for seawater - sediment mesocosms (Anderson, Whitlock, & Harwood, 2005) . The inactivation of total coliforms increased with salinity, while salinity did not affect the survival of fecal coliforms (Okabe & Shimazu, 2007) . T he host origin of enterococi and fecal coliforms affected their decay rates in freshwater and saltwater mesocosms (Anderson et al., 2005) . Salinity increased the inactivation rate of enterococci in mesocosms inoculated with dog feces, contaminated soil, and wastewater (Anderson et al., 2005) . The survival of e nterococci and E.coli decreased in the presence of autochthonous microorganisms in freshwater microcosms inoculated with raw sewage (Medema, Bahar, & 14 Schets, 1997) . The inactivation rate of E.coli in freshwater microcosms spiked with raw sewage was dependent on the strain ( Anderson et al., 2005) . Researchers have also evaluated the persistence of enteric markers in the environment. For example, the persistence of human associated Bacteroidetes 16S rRNA markers , HF183 and BacHum, were significantly different than general Bacteroides 16S rRNA , Allbac, in freshwater - sediment microcosms spiked with raw sewage (Dick et al., 2010) . The persistence of Entero1 spiked with human, dog and bovine feces, while the survival of c ulturable enterococci decreased significantly in sunlight (Bae & Wuertz, 2009) . Additionally, general Bacteroides 16S rRNA , GenBac, persisted longer in marine water than freshwater (Green, Shanks, Sivaganesan, Haugland, & Field, 2011) . 1.5.2 Factors in sediments It is commonly accepted that fecal pollution can affect the water quality of recreational waters . However, many indicators and pathogens do not decay in the water column. For example, E.coli and Salmonella spp. can excrete extracellular polymeric substances that bind to suspended particulate matter in order to facilitate deposition to benthic sediment (Droppo et al., 2009) . Thus, sediments tend to have larger concentrations of fecal indicators such as E.coli , total coliforms, and enterococci compared to the water column (Pote et al., 2009) . H igh organic carbon content and small sediment particle size were associated with increased survival of culturable E.coli ATCC 25 922 inoculated in microcosms of coastal sediments that were submersed in estuarine water and stored in various temperatures (D L Craig, Fallowfield, & Cromar, 2004) . A decrease in temperature (24 ° vs. 4 ° C) resulted in a 15 - fold increase of first - order inactivation rates of culturable E.coli in the sediment portion of water - sediment microcosm 15 spiked with dairy manure (Garzio - Hadzick et al., 2010) . Naturally occurring enterococci and E.coli decreased 2 logs in the first 30 days of storage followe d by a subsequent < 1 log decrease during 30 90 days of storage in microcosms of surface sediments obtained near an outlet from a wastewater treatment plant in Geneva, Switzerland, and stored at 4 ° C (Haller, Poté, Loizeau, & Wildi, 2009) . The results of a study that evaluated the persistence of fecal indicators in sediments suggested that disruption of sediments can resuspend enteric bacteria into the water column (Pote et al., 2009) . Another study determined that enterococci survived longer in sediment microcosms spiked with wastewater compared to E.coli and total coliforms (Haller, Amedegnato, Poté, & Wildi, 2009) . Concentrations of naturally occurring fecal coliforms in sediments were correlated to rainfall events in a marine beach and an estuary in Southern Australia (D.L. Craig, Fallowfield, & Cromar, 2002) . Concentrations of naturally occurring fecal colifor ms in sediments in a marina in Southern Australia increased in colder months with less sunlight (D.L. Craig et al., 2002) . Another study determined that E.coli in manure could leach through so ils . Despite the large wealth of knowledge of the factors that are associated with the persistence of enteric bacteria in sediments, there are no recommended criteria for the monitoring of fecal indicators in sediments. 1.6 Scientific Needs qPCR measurements of host associated and general pollution in recreational waters are being adopted by many public health departments. Currently, the storage of water samples slated qPCR is the same as conventional methods . There is a lack of knowledge of how the persistence of enteric markers is affected by storage condition s, DNA extraction methods, and storage temperature s over short term and lo ng term storage durations. Such investigations will allow for 16 insight into the parameters to effectively store water samples in order to efficiently monitor the water quality of recreational waters. It is accepted that fecal indicators persist longer in sediments than the water column. Studies have evaluated various factors that have affected the persistence of fecal indicators in sediments. However, few studies have evaluated the anthropogenic and climate factors that affect the concentrations of genera l pollution markers over larger time scales. Such investigations would allow for meaningful, multi - disciplinary exploration into what factors are significant influences to persistent fecal indicators using a singular index. 1.7 Research Objectives 1.7.1 Goal 1 Water quality monitoring with qPCR methods are not efficient because of the lack of optimization of the storage methods. Therefore, water quality samples spiked with raw sewage were stored in three temperatures, and in two storage conditions, i . e. l iquid suspension and attached to a solid matrix, for short term and long term durations up to 28 and 366 days, respectively. DNA extraction methods were evaluated in order to determine how these methods affected indicator persistence patterns. The samples were assayed for the concentrations of three enteric markers, and mathematical relati onships of the persistence of three genetic markers over time were evaluated with linear and non - linear models. The specific objective s were to: 1) i dentify the mathematical relationship s of enteric markers measured in water samples over short term and long term storage durations ; 2) describe and compare the persistence pattern of general and human - associated enteric 17 markers stored over short and long duratio ns ; and 3) e valuate how temperature, attachment to a solid surface vs. liquid suspension, DNA extraction method, indicator species , and storage duration affect ed the persistence of DNA over time . 1.7.2 Goal 2 The Lake St. Clair watershed is small and hig hly populated. As part of the Great Lakes, it has a long history of water quality and anthropogenic disruptions. Fecal indicators were spiked into surface sediment from the Lake St. Clair watershed, and eighteen DNA extraction methods were evaluated in ord er to compare DNA extraction efficiency and identify the optimal method. Additionally, few studies have evaluated the water quality in the watershed over larger time scales. Sediment cores were collected in the watershed and assayed for fecal indicators co ncentrations , and sedimentary nutrients concentrations in order to determine how climatic and anthropogenic variables we re associated with historic al fecal pollution over 100 years . The specific objectives were to: 1) compare the concentrations of fecal indicator markers spiked into sediments and extracted with various D NA extraction method s ; 2) identify the optimal DNA extraction method to extract DNA from sediments in the Lake St. Clair watershed; and 3) determine how time, climate, i.e. air temperatur e , and discharge rate , and anthropogenic attributes , i.e. nutrient loading and population, were associated with concentrations of general pollution markers over large time tables in Lake St. Clair sediment cores. 18 CHAPTER 2. SHORT TERM PERSISTENCE ENTERIC BACTERIA I N WATER SAMPLES STORED ON A SOLID MATRIX AND IN LIQUID SUSPENSION 19 2.1 Introduction Rapid and sensitive detection of fecal indicator bacteria (FIB) in drinking, recreational, ambient waters, and wastewater is critical for address ing pollution, providing ad equate treatment solutions , and estimating potential public health risk s upon exposure . Culturable methods for FIB have known limitations. For example, some FIB have no association with concentrations of Salmonella and Cryptosporidium in wastewater effluen t and freshwater (Lemarchand & Lebaron, 2003) reportin g results (Wade et al., 2006) . USEPA has published an alternative rapid method, Method 1611, which is culture independent, and uses quantitative polymerase chain reaction (qPCR) to measure enterococci (ENT) concentrations in recreational waters (USEPA, 2012b) . Also, qPCR has been used to measure Escherichia coli (EC) (Frahm & Obst, 2003) , and general Bacteroi dale s (USEPA, 2010b) in recreational water regardless of me tabolic state. The refore, the persistence of FIB cells as measured by their target genes is particularly important for recreational water monitoring. Bacteroidales 16S rRNA markers , bac - pre1, human - bac1 and pig - bac2, were shown to have increased persistenc e in environmental water microcosms as temperature decreased and salinity increased (Okabe & Shimazu, 2007) , while the persistence of Bacteroidales 16S rDNA marker s , GenBac3, BuniF2, and HF183 , decreased i n sunlight in freshwater and seawater microcosms spiked with sewage (Dick et al., 2010; Green et al., 2011) . Bacteroidales 16S rDNA , Allbac, showed increased persistence with decreased dissolved oxygen concentration and decreased river water temperatures (Ballesté & Blanch, 2010) . The persistence of ENT via the 23S rDNA w as also negatively correlated with temperature, while EC ( 23S r D NA ) decayed rapidly on either side of its optimal growth temperature (37°C) in manure (M. Klein, Brown, Ashbolt, Stuetz, & Roser, 2011) . 20 In comparative studies, Dick et al. (2010) showed that culturable EC, HF183, and BacHum had a similar persistence , with 99% reductions in concentration (T 99 ) at 2.0, 2.2, and 1.7 days, respectively, in freshwater microcosms stored at 25 ° C (Dick et al., 2010) . Additionally, BacHum and culturable ENT quantified from sewage spiked freshwater microcosms in sunlight had comparable T 90 ( remov al of 90% of initial concentration ) at 1.8, and 1.0 day, respectively (Walters et al., 2009) . The first purpose of this study was to use qPCR to describe and compare the persistence of Bacteroides thetataiotaomicron alpha mannanase (BT - am), enterococci 23S rRNA (ENT - 23), and E.coli uidA (EC - uidA) measured from sewage spiked river . In a controlled bench scale study, the second purpose was to evaluate if immobilization of cells on a solid surface such as a membrane filter may extend the persistence of FIB markers. BT - am was chosen as a target because it was shown to be specific to human fecal contamination (Aslan & Rose, 2013) . ENT - 23 was chosen because the USEPA recommended this target for monitoring water quality in recreational waters (USEPA, 2012b) . EC - uidA was selected because it ha s a species specific gene that is the basis of USEPA conventional cultiva table enumeration methods (USEPA, 2002) . The influence of storage temperature on FIB persistence over time was also analyzed . Four mimicked storage temperature at room temperature. Also, 37°C represented an extreme storage temperature such as sto rage in a car without ice during the summer in a remote. Investigating the effects of storage temperature and transit time on the persistence of genetic markers from FIB measured in water samples may improve the efficiency of sample collection and storage from remote areas or low resource settings where storage on ice is not an option or extended time in transit is necessary . 21 The specific objectives of this study were to: i) Identify the mathematical relationship between time (up to 28 days), and FIB persisten ce as measured by qPCR in sewage spiked river water stored at 4° C , 27° C and 37°C. ii) Compare how cell attachment to a solid matrix (SM) versus liquid suspension (LS) affect persistence calculated from the experimental data , and calculate the time for 90% of F IB decay (T 90 ). iii) Compare how DNA extraction methods affect the analysis of DNA persistence over time. 2.2 Methods An overview of the experimental design is shown in Figure 2. 1 and described below. 22 Figure 2 . 1 : Ov erview of experimental design. a EPA DNA extraction method , was performed with three replicates from each storage condition. VI. Indicator DNA Decay Models: Indicator concentrations within each storage condition, temperature and DNA extraction method were analyzed with best - fit models. V. DNA Quantification: Three gene markers were enumerated in all replicates. IV. DNA extraction: At each sampling time point, three DNA extraction methods were used from each storage condition and temperature. III. Sampling scheme : At 0, 0.25 , 1, 2, 5 , 7, 14 , 21 and 28 days ; in suspension (only in bold days) and attached were removed from storage . II. Storage temperatures: Divided at inital sampling II. Storage Conditions : Partioned at inital sampling. The solid matrix was a membrane filter. I. Sampling matrix 10% vol/vol sewage in autoclaved river water Liquid Suspension 4 C (on ice) six 100ml samples six 100ml samples EPA DNA Extraction Method Qiagen DNA Mini Kit Qiagen Stool Kit 27 C (room temperature) three 100ml samples a three 100ml samples a EPA DNA extaction Method Total eluted DNA B. thetaiotaomicron alpha - 1 - 6 mannase E.coli uidA Enterococci spp. 23S rRNA 37 C six 100ml samples six 100ml samples EPA DNA Extraction Method Qiagen DNA Mini Kit Qiagen Stool Kit Solid Matrix 23 2.2.1 Water Sample Storage W ater was collected from the Red Cedar River (East Lansi ng, MI) on June 6, 2011 and October 28, 2011. The water temperature, turbidity, conductivity, and pH values from the June (and October) samples were 20.5°C ( ) , 8.67 NTU ( 2.29 NTU ) , 1970 µS ( 855 ) , and 6.95 (and 7.78 ) , respectively. The water was autoclaved and seeded to a final concentration of 10% ( vol/vol ) raw sewage (City of East Lansing Sewage Treatment Plant, MI). After seeding, the samples were set up as bench scale experiments under two conditions . In t he first set , the cells were left naturally in suspension ( LS) as 100 ml aliquots in seventy - five sterile plastic bottles, and the second set consisted of 100 ml of seeded river water (SM) filtered onto 120 Nucleopore Track Etch polycarbonate membrane filters (0.4 5 , 47 mm diameter , Whatman Inc., Piscataway, NJ) . The filters were removed from the housing wet, folded in half and individually stored in sterile 50 ml centrifuge tubes , and moisture from the filter was able to collect in the tube . The LS (an d SM) samples were divided into three temperature groups , 4°, 27 , ° and 37°C , that included 30 bottles ( and 48 filters), 18 bottles ( and 24 filters), and 30 bottles ( and 48 filters) , respectively ( Figure 2. 1 ) . The storage temperatures were measured daily. T he average daily temperatures with one standard error ± 2°C (room temperature in the dark), and 4 ± 2°C (storage in a cooler with ice packets replaced every 24 hrs). 2.2.2 Sample processing and DNA extraction Cells from t he liquid suspension (LS) sample s were filtered using 0.45 um Nucleopore Track Etch membrane filters after days 0, 0.25, 2, 5, 14 and 28 days from all storage temperatures. Cells were recovered and DNA was extracted in the same way as the SM samples. 24 Fifteen solid matrix (SM) filters were also processed on days 0, 0.25, 1, 2, 5, 7, 14, 21 and 28. At the two extreme temperatures (4° C and 37°C) , duplicate filters were folded four times aseptically with sterile forceps, and placed into the extraction tub es for the following methods: Qiamp Stool DNA Mini Kit® , STOOL ( Qiagen Inc, Valencia, CA); Qiamp DNA Mini Kit , MINI ( Qiagen Inc, Valencia, CA) ; and USEPA Method 161 1, EPA - DNA (USEPA, 2012b) . Triplicate filters from 27°C were folded four times aseptically and placed into extraction tubes of EPA - DNA. A preliminary elution was incorporated for MINI i n order to detach DNA from the filter by first transferring the filter into a sterile tube with 50 ml phosphate buffered saline (PBS) and were vortexed for 2 min at 3200 rpm. The supernatant was centrifuged for 20 min at 8000 x g and 48 ml of the supernata nt was discarded. The remaining supernatant (2 ml) was transferred into a EPA - respective ly. The MINI, EPA - DNA and STOOL final DNA elution volumes were 400, 350 and 200 µL, respectively. Total DNA concentrations were measured with Nano - Drop ND - 1000 ( Thermo Fisher Scientific Inc, Waltham, MA ) for all DNA extracts before storage at - 80°C. 2. 2. 3 qPCR quantification of three indicators Concentrations of EC - uidA, ENT - 23 and BT - am were measured in all DNA extracts with qPCR using the Roche LightCycler® 480 Instrument (Roche Applied Science, Indianapolis, IN). When inhibition was noted in the qPCR results (i.e. analytical replicates differed in marker diluted 5x and rerun. Each qPCR reaction volume, including the standards, was 20 µl (5 µl - sample and 15 µl qPCR reagents). The reagents included 10 µl LightCycler 480 Probes 25 MasterMix (Roche, Indianapolis, IN), indicator probes sequences, reverse and forward primers (Table 2. 1) , 2 mg/ml bovine serum album (BSA), 1 mM MgCl 2 (EC - uidA only), and sufficient nucleas e free water to bring reaction volume to 20 The BSA concentrations in BT - am, ENT - 23 and EC - The BT - am qPCR program was previously described (Srinivasan et al., 2011) . The EC - uidA qPCR program started with a 10 min cycle at 95°C, then 40 cycles of 30 sec at 95°C, 30 sec at 58°C, and 10 sec at 72°C . The ENT - 23 qPCR program started with a 10 min cycle at 95°C then 40 cycles of 10 sec at 95°C, 30 sec at 60°C, and 15 sec at 72°C. Duplicates of positive controls (ATCC cells or DNA extracts depending on the assay), method blan ks (sterile phosphate buffered saline water as template for each storage time interval), and no template control (nuclease free water as template) were analyzed in each qPCR run. All samples were analyzed in duplicates. The indicator concentrations were calculated from their respective qPCR standard curves. Genomic DNA was extracted from overnight cultures of Enterococcus faecalis ATCC strain 19433 and E.coli ATCC strain 15597 using MINI. The eluted DNA was serially diluted 1:10 to create 6 dilution steps for the ENT - 23 and EC - uidA standard curves. The standard curve for B. thetataiotaomicron was prepared from genomic DNA (ATCC, 29148) with 1:10 serial dilutions of the DNA to create 6 dilution steps. Each dilution was analyzed in triplicate. A new standard curve run was prepared after every four qPCR runs. The average efficiencies, r 2 values , and the threshold cycle values of the lowest detected amplification (and representative copies/rxn) of the qPCR standard curves of ENT - 23, EC - uidA and BT - am are descri bed in Table 2. 2 . EC - uidA and BT - am have one copy per genome. Therefore, each quantified sequence represented one E.coli and B. thetataiotaomicron cellular equivalent (CE) , respectively. There 26 are an estimated average four copies of ENT - 23 per Enterococcus faecalis cell (USEPA, 2010a) , and represented one CE in this study . The qPCR analytic al replicates from each sample were averaged and converted into cell equivalents (CE) per 100 ml - water sample. Specifically, the qPCR measurements of the 5 l template were a fraction of the elution volume. The concentration of the marker in copies/1 l wa s multiplied to the elution volume, which in the case of EPA - DNA and STOOL represented the 100 ml sample. In the case of MINI, it represented 10 ml of the sample, and then the concentration in the elution volume was multiplied by 10 to represent the 100 ml sample. 27 Table 2 . 1 : Fecal indicator species, targeted gene, primer sequences, probe sequences, and amplified sequence length (bp) of the qPCR genetic markers in this study. FIB Species Gene Primer and probe sequences (5' - 3') Primer (and probe) concentration in qPCR reaction matrix Amplified sequence size (bp) Reference B.thetataiotaomicron (BT) - 1 - 6 mannanase (BT - am) CATCGTTCGTCAGCAGTAACA CCAAGAAAAAGGGACAGTGG FAM - CAGCAGGT - NFQ a 0.3 µM (0.1 µM ) 63 Yampara et al. (2008) enterococci (E NT ) 23SrRNA (ENT - 23) AGAAATTCCAAACGAACTTG CAGTGCTCTACCTCCATCATT 6FAM - GGTTCTCTCCGAAATAGCTTTAGGGCTA - TAMRA 0.5 µM (0.2 µM ) 91 Frahm and Obst (2003) E.coli (EC) uidA (EC - uidA) CAATGGTGATGTCAGCGTT ACACTCTGTCCGGCTTTTG 6FAM - TTGCAACTGGACAAGGCACCAGC - BBQ 0.5 µM (0.2 µM ) 163 Srinivasan et al (2011) a Roche Universal Probe Libraries (UPL) Probe 62 . 28 Table 2 . 2 : Average efficiency, R 2 l ), and DNA extraction method specific detection limits from EC - uidA, ENT - 23, and BT - am from E.coli , enterococci and B.thetataiotaomicron , respectively. Genetic marker (Species) R 2 values Average efficiency (95% confidence interval) CT value (representative copies/5 l ) DNA extraction method specific detection limits (and indicator persistence as measured by Log N/N 0 ) a B T - am (BT) 0.96 95.7 % (± 5.00 % ) 37.60 (183) MINI: 3.05 x 10 3 copies/100 ml (Log N/N 0 = - 1.56) EPA - DNA: 2.67 x 10 2 copies/100 ml b (Log N/N 0 = - 2.31) STOOL: 1.53 x 10 2 copies/ 100 ml (Log N/N 0 = - 2.29) ENT - 23 (ENT) 0.98 97.33 % ( ± 2.88% ) 36.31 (3.72) MINI: 7.16 x 10 3 copies/100 ml (Log N/N 0 = - 3.46) EPA - DNA: 6.27 x 10 2 copies/100 ml b (Log N/N 0 = - 3.87) STOOL: 3.58 x 10 2 copies/ 100 ml (Log N/N 0 = - 1.83) EC - uidA (EC) 0.93 97 % ( ± 1.50% ) 35.85 (8.95) MINI: 1.46 x 10 5 copies/100 ml (Log N/N 0 = - 1.61) EPA - DNA: 1.28 x 10 4 copies/100 ml b (Log N/N 0 = - 2.49) STOOL: 7.32 x 10 3 copies/100 ml (Log N/N 0 = - 1.91) a The detection limits from a 100 ml water sample are based on the Qiagen Qiamp Mini Kit (MINI), USEPA Crude DNA Extraction Method (EPA - DNA) and Qiagen Qiamp DNA Stool Mini Kit (STOOL) final elution volume and the initial concentrated sample for MINI. b DNA extracts from EPA - DNA were diluted 5x prior to qPCR, the BT - am, ENT - 23 and EC - uidA detection limits were increased to 1.34x10 3 , 3.14x10 4 , and 6.4x10 4 representative copies/1 l , respectively. 29 2. 2.4 Statistical Analyses and Persistence Modeli ng A two - way ANOVA and post - IL) were performed on the total DNA extraction method and/or attachment status affected total DNA concentra tions over time. The low temperature experiments were used to minimize the effects of other variables. Total DNA concentrations from all 4 ° C samples were transformed to ng/ 100 ml - sample. Subsequently, total DNA concentrations within each treatment were transformed into log ( C t /C o ), where C t was the DNA concentration after t days of storage and C o was the DNA concentration at initial sampling (t = 0 days). After evaluation of the total DNA concentrations and linear regression analysis, the EPA - DNA eluted samples were chosen to evaluate storage conditions and temperatures. Explanation of why EPA - DNA was chosen is given in the Results section. Reporting the non - detect data at their detection limit was chosen because it cannot be reasonably assumed that a value below the detection limit equals zero concentration of the indicator. Additionally, the USEPA has advised that non - detect data be reported at the detection limit (USEPA, 1991) in order to best represent the measurements. The fractional persistence of the target concentration at time, t, to initial concentration , Log 10 (N/N 0 ), where N was the CE/ 100 ml - sample after t days in storage, and N o was the CE/ 100 ml - sample at t = 0 was calculated for each experimental condition (attachment status), temperature, and indicator. These data were fit to established decay models using maximum likelihood estimation in a modelling program in R 3.0.1 (R Development Core Team, 2013) that was provid ed by Drs. Kyle Enger and Jade Mitchell. Since there are few studies to inform the selection of appropriate decay models for qPCR CE data, seventeen models that are commonly 30 used to model bacterial decay under various conditions were initially evaluated (h ttp://qmrawiki.canr.msu.edu) . Table S2.1 highlights the equations and references of all models employed in the data analysis tool. Using the Bayesian information criterion (BIC) , these models were narrowed down to five that produced adequate fits across th e data sets in order to produce tractable comparative information. BIC values were used to evaluate the best fit models because it takes into account the number of observed data points, the number of parameters of the model, the observed data, and the maxi mum likelihood of the function. The evaluated best fit models were: first - order exponential decay model (Chick, 1908) , biphasic exponential decay (Corradini, Normand, & Peleg, 2007) , two - stage (Juneja, Huan g, & Marks, 2006) , log - logistic (Juneja, Marks, & Mohr, 2003) , and Gompertz 3 - parameter (Gil, Miller, Brandão, & Silva, 2011) . Table 2. 3 describes these models and their properties. Differences less than 2 in BIC values are not considered strong evidence for model selection. Therefore, if the smallest BIC was < 2 units from the BIC of the biphasic model, then the biphasic model was chosen. The predicted time needed for 1 log 10 reducti on (T 90 ) was also calculated by substituting - 1 .0 = Log 10 (N/N 0 ), and solving for t in each equation. The standard error values of the T 90 values were not calculated as they were not provided in the model. . 31 Table 2 . 3 : List of best fit models represented in the datasets, their equations , and properties . Equation Name Equation a Eqn. Properties Reference First - order exponential decay model (ep) Linear, negative slope Chick ( 1908 ) Biphasic exponential decay model (bi3) for 0 t 28 d - 0.01 - 0.12 (14. 0) LS bi3 * 13.5 12. 4 0.1 9 0. 2 0 (16.8 ) ENT SM bi3 2.7 > 28 d - 0.02 - 0.31 (20. 4 ) LS e p 11. 5 > 28 d 0.06 - - BT SM bi3 6.0 27.0 - 0.04 - 0.2 3 (11.8 ) LS gz3 11.2 9.6 - 2.48 0.27 5.9 27 EC SM bi3 * 5.2 > 28 d 0.08 0.07 (14 . 0) LS bi3 * 6.5 7. 1 0.78 0.69 (2.4) ENT SM jm1 2.6 > 28 d 0.0 4 0.24 - LS bi3 * - 1.8 8.4 1.9 8 1.8 3 (0.6 ) BT SM bi3 - 25. 8 1 8.0 0.21 0.17 (9.4 ) LS bi3 0.3 1.8 1.3 2 1.2 7 (4.1 ) 37 EC SM e p 27.9 > 28 d 0.17 - - LS jm2 12.0 6.0 - 0.7 8 1.66 - ENT SM bi3 * 13.9 > 28 d 0.02 - 0.09 (14.0 ) LS bi3 - 10.5 6.5 0.50 0.32 ( 3.6 ) BT SM bi3 * 19.5 3.2 0.7 2 0.66 (5. 3 ) LS bi3 * 20.0 1.0 2. 30 2.23 (2.1 ) a first - order exponential decay (ep) , biphasic exponential decay (bi3), two - stage (jm1), log - logistic (jm2) , and Gompertz 3 - parameter (gz3). b Value of the Bayesian information criterion (BIC) for the model. *Denotes i f the BIC value of bi3 was ± 2 of the smallest reported BIC value of the available models. c The breakpoint, t = x, is starting point when the slope of the biphasic exponential decay equation (bi3), , included both decay constants , k 1 and k 2 . d The experiment duration was 28 days. 2. 3. 4 Comparison of the three FIB and effects of attachment and temperature. FIB persisted longer when attached (SM) than in suspension (LS) for all indicators and temperatures ( Figure 2. 3 A - F ) . The least persistent indicator was BT regardless of the storage condition or temperature. The most to least persistent markers in LS for all temperatures were: ENT > EC > BT. 38 The T 90 values were calculated from all models ( Table 2. 4 ). Overall, the data show ed that larger T 90 values were observed for CE stored on a SM , although the effect was modulated by temperature. At 4°C, ENT had T 90 > 28 days in LS and on a SM . EC and BT had T 90 > 28 , and 27 days, respectively , on a SM , and in LS , the T 90 = 12.4 , and 9.7 days, respectively. At 27°C , the T 90 > 28 days for EC, and ENT on a SM , while BT had T 90 = 18 days. If the samples were stored in LS at 27°C, T 90 = 8.4, 7.1 and 1.8 days for ENT, EC and BT, respectively. A t 37°C , ENT, EC and BT on a SM had T 90 > 2 8, > 28, and 3.2 days, respectively, and when in LS T 90 was 6.5, 6.0, and 1.0 days, respectively. 39 Figure 2 . 3 A - F : Observed and predicted fractional persistence a of thre e indicators b stored for up to 28 days at C (A - B) C (C - D) , (E - F) storage temperatures in two storage conditions c . -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 0 10 20 30 Log (Nt/No) Days in storage A. 4 C LS BT (actual) ENT (actual) EC (actual) BT (gz3) ENT (ep) EC (bi3) -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 0 10 20 30 Log (Nt/No) Days in storage B. 4 C SM BT (actual) ENT (actual) EC (actual) BT (bi3) ENT (bi3) EC (bi3) -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 0 10 20 30 Log (Nt/No) Days in storage C. 27 C LS BT(actual) ENT (actual) EC (actual) BT (bi3) ENT (bi3) EC (bi3) -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 0 10 20 30 Log (Nt/No) Days in storage D. 27 C SM BT (actual) ENT (actual) EC (actual) BT (bi3) ENT (jm1) EC (bi3) -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 0 10 20 30 Log (Nt/No) Days in storage E. 37 C LS BT (actual) ENT (actual) EC (actual) BT (bi3) ENT (bi3) EC (jm2) -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 0 10 20 30 Log (Nt/No) Days in storage F. 37 C SM BT (actual) ENT( actual) EC (actual) BT (bi3) ENT (bi3) EC (ep) 40 a The persistence is measured in log (N/N o ) , where N is the concentration of the indicator after t days in storage, and N 0 i s the initial concentration of the indicator. b The diamond , and labels represent the observed BT, ENT, and EC log (N/N o ) persistence, respectively. The data points below the detection limit are filled in. Duplicate replicates fo averages of each time point are displayed . In the treatments, the indicators were measured in triplicate , and the averages we re displayed in the graphs along with bars that represent one standard error . The lines represent the following best fit models: first - order exponential decay (ep) , exponential decay (bi3) , two - stage (jm1), log - logistic (jm2) , and Gompertz 3 - parameter (gz3). c The graphs on the left side represent in su spension storage condition (LS), while the right represents attached storage condition (SM). 2. 3. 5 Multiple Linear Regression Analyses. Linear regression analyses evaluated the statistical significance of storage time, extraction method, indicator type, temperature, and storage condition on the fraction of persistence of the indicators, Log 10 ( N/N 0 ) ( Table 2. 5 ). The three analyses, complete dataset (n = 585), indicator specific (n = 195 for each indicator), and temperature specific (n = 117 for 4 ° and 37 ° C; and n = 234 for 27 ° C) sub - datasets showed that the independent variables did not exhibit colinearity, thus indicating statistical independence. A nalysis of the complete dataset determined tha t increasing storage time and storage in LS decreased fractio nal persistence (p < 0.001). The coefficients of the linear regression equation s and their standard errors for each dataset are listed in Table 2.6 , while the R 2 values of the equations are listed in Table 2.5 . The fractional persistence of EC and ENT data sets were significantly affected by time (p EC < 0.001; p ENT < 0.001), extraction method (p EC = 0.047; p ENT < 0.001), and storage condition (p EC = 0.027; p ENT = 0.022). Additionally, the fractional persistence of BT was significantly affected by time (p BT < 0.001), attachment (p BT < 0.001), and temperature (p BT = 0.001). For the 4 ° C dataset, time (p 4C = 0.001) significantly affected the fractional persistence. For 27 ° C and 37 ° C sub - datasets, time 41 (p 27C < 0.001; p 37C < 0.001), attachment (p 27C < 0.001; p 37C < 0.001), and type of indicator (p 27C < 0.001; p 37C < 0.001) significantly affected fractional persistence. 42 Table 2 . 5 : Multiple linear regression analyses determined how the observed fractional persistence ratios a , were influenced by: storage time measured in hours, storage conditions b , extraction method c , genetic markers d and storage temper ature e . and p - value) Data set (sample size) R 2 Storage time - S Extraction method - E Storage attachment - C Storage temperature - T Indicator - I All data (n = 585) 0.227 - 0.393 (p < 0.001 ) 0.056 (p = 0.142 ) - 0.189 ( p < 0.001 ) - 0.033 (p = 0. 385 ) 0.174 ( p < 0.001 ) (n = 234) 0.058 - 0.220 ( p 4C = 0.001 ) 0.042 (p 4C = 0.522 ) - 0.070 (p 4C = 0.28 8) N/A 0.049 (p 4C = 0.452 ) (n = 117) 0.677 - 0.605 (p 27C < 0.001 ) N/A - 0.463 ( p 27C < 0.001 ) N/A 0.308 ( p 27C < 0.001 ) (n = 234) 0.407 - 0.52 0 ( p 37C < 0.001 ) 0.108 ( p 37C = 0.048 ) - 0.219 ( p 37C < 0.001 ) N/A 0.276 ( p 37C < 0.001 ) BT (n = 195) 0.37 4 - 0.492 (p BT < 0.001 ) - 0. 067 (p BT = 0.260 ) - 0.284 ( p BT < 0.001 ) - 0.2 0 3 ( p BT = 0.001 ) N/A ENT (n = 195) 0.26 6 - 0.365 ( p ENT < 0.001 ) 0.324 ( p ENT < 0.001 ) - 0.147 ( p ENT = 0.022 ) 0.056 (p ENT = 0. 386 ) N/A EC (n = 195) 0.1 65 - 0.350 ( p EC < 0.001 ) - 0.138 ( p EC = 0.047 ) - 0.153 ( p EC = 0.02 7 ) 0.046 (p EC = 0.50 8) N/A a log N/N 0 (normalized to cell/100 ml - sample), where b in suspension - LS or attached - SM c Qiagen Mini - MINI, EPA Method 1611 - EPA - DNA, and Qiagen Stool - STOOL d EC - uidA, ENT - 23, and BT - am e 43 Table 2 . 6 : A list of the coefficients in the linear regression equation a that evaluated the correlations of the independent variables to fractional persistence b . Coefficients of the independent variables (and their standard errors) of the linear regression equations Data set (sample size) Storage time - S Extraction method - E Storage attachment - C Storage temperature - T Indicator - I Y intercept - b 0 All data (n = 585) - 0.104 (0.010) 0.225 (0.134) - 1.012 (0.201) - 0.008 (0.007) 0.552 (0.120) - 1.378 (0.495) (n = 234) - 0.028 (0.008) 0.064 (0.99) - 0.427 (0.167) N/A 0.076 (0.100) - 0.785 (0.383) (n = 117) - 0.052 (0.005) N/A - 0.813 (0.094) N/A 0.321 (0.056) 0.157 (0.183) (n = 234) - 0.057 (0.006) 0.107 (0.073) - 0.508 (0.122) N/A 0.374 (0.073) - 0.839 (0.288) BT (n = 195) - 0.015 (0.004) - 0.101 (0.089) - 0.633 (0.132) - 0.015 (0.004) N/A 0.585 (0.288) ENT (n = 195) - 0.045 (0.008) 0.546 (0.108) - 0.368 (0.159) 0.005 (0.005) N/A - 1.143 (0.349) EC (n = 195) - 0.034 (0.007) - 0.186 (0.093) - 0.311 (0.140) 0.003 (0.005) N/A 0.211 (0.303) a , b 44 2.4 Discussion Our study compared the persistence of CE of three naturally occurring FIB using qPCR from sewage spiked into river water, and the influence of attachment to a solid surface at various temperatures. Bacterial cells affixed to a membrane filter instead of remaining in a liquid suspension exhibited significantly slowe r CE and DNA target degradation r egardless of indicator species or storage tempera ture. Similarly, culturable FIB immobilized onto soil particles survive d longer in aquatic environments (Wheeler Alm et al., 2003) . A ttachment to solid surface s possibly protects the cell wall from degradation, and therefore , its DNA target , yet the mechanism, type of matrix, and strength of attachment have not been fully explored. H istorically, a first - order linear exponential model was commonly used t o describe the rate of inactivation using cultivation (Liang et al., 2012) . M ore recently, nonlinear models have provided a better fit to bacterial persistence (Coroller, Leguerinel, Mettler, Savy, & Mafart, 2006) . Our study is one of the few that have evaluated various linear and non - linear models to best describe the persistence patterns of FIB CE using qPCR in water under various conditions. L inear mode ls, particularly the biphasic exponential decay (bi3 ) and first - order exponential decay (ep), represented the majority of the treatments in our study ( Table 2. 4 ) , indicating that markers measured in water samples . Biphasic decay has been observed previously two - staged , best described by bi3 in our study, was previously coined to describe data that fit two first - order linear equations with two unique slopes (Crane & Moore, 1986) . Two - staged models also described the DNA persistence of Bacteroidales, Salmonella enterica , enterococci, and E.coli in manure amended so (Rogers et al., 2011) . Klein et al. (2011) combined two first - order 45 addressed the persistence of an EC marker in composted manure at 37 ° C (M. Klein et al., 2011) . Rogers et al. (2011) offered reasons for the biphasic behavior: 1) microbes die off at a rapid rate until the carrying capacity of the environment is approached ; and 2) the true presence of two sub - po pulations with different decay rates. More persistence studies using molecular methods in various matrices are needed in order to better analyze the biphasic nature of CE and DNA persistence. Our study also included gz3 , jm1, and jm2 models that evaluated BT in LS at 4°C, ENT on SM at 27°C, and EC in LS at 37°C, respectively . The se models are non - linear survival curves frequently used to describe thermal inactivation. Jm1 was derived on the assumption that a cell must be hit a number of times , k2 , and the probability of being hit is described by the function , exp ( - k2 x t) (Juneja et al., 2006) . Jm2 wa s used to describe survival data with an indication that as time increases , decay begins to slow , which can be witnessed in d Gz3 is an empirical model , and has been shown to be capable of quantifying behavior under non - isothermal conditions (Gil et al., 2011) . Gz3 does not assume a constant decay rate , but it can be used to model decay rates that change over time related to variable temperature. The models in our study were used to calculate the days needed to achieve 90% reduction ( T 90 ) in cellular equivalent concentrations. Our values were compared to values from previous studies that measured the persistence of Bacteroides , enterococci and E. coli in liquid microcosms, and on solid matrices ( Table 2. 7 ). Previous research also confirmed t hat enterococci markers persisted longer than Bacteroides spp. markers (Rogers et al., 2011) . Bacteroides spp. markers are relatively less persistent than other markers ( Table 2. 7 ) . The shortened persistence may be a result of the inability of Bacteroides spp. to grow in the 46 environment (Kreader, 1998) , or cell degradation due to its sharpened death rate in aerobic conditions. EC persistence in our study agreed with a previous study of E.coli uidA , Eco, in (M. Klein et al., 2011) . E nterococci 23S rRNA , Entero1, had a T 90 = 9 days in 7% ra (Walters et al., 2009) . Similarly, T 90 > 32 days was reported for Entero1 in sewage spiked beach sand drained with seawater at 22°C (Yamahara, Sassoubre, Goodwin, & Boehm, 2012) . Increased enterococci persistence on substrates like b each sand have indicated that these substrates may act as aquatic FIB reservoirs (Yamahara, Walters, & Boehm, 2009) . Human associated Bacteroidales 16S rRNA , HF183, had a calculated T 9 0 = 2.59 days , thus faster degradation, in 5% sewage spiked river water stored in a suspension stored in the dark at slightly higher temperature, 14 °C , compared to our study (Gilpin et al., 2013) . 47 Table 2 . 7 : Comparison of genetic markers from Bacteroidales, E.coli , and enterococci in var ious storage conditions, and T 90 values from our study and previous studies. E.coli genetic markers T 90 (days) Target gene (marker name) Storage condition Storage Reference 56 23S rRNA (EPA - EC23S) Beef manure amended soil, 80% field capacity moisture 25 Rogers et al. (2011) >28 d uidA (EC - uidA) SM c : 10% (vol/vol) SPRW 4 Our study 27 27 uidA ( Eco) Compost manure 20 Klein et al. (2011) 13.19 uidA (EC - uidA) SM: 10% (vol/vol) SPRW 37 This study 12.35 LS b : 10% (vol/vol) SPRW 4 7.08 LS: 10% (vol/vol) SPRW 27 6.5 uidA ( Eco) Compost manure 37 Klein et al. (2011) 6.00 uidA (EC - uidA) LS: 10% (vol/vol) SPRW 37 Our study 1.71 uidA ( Eco) Compost manure 50 Klein et al. (2011) Enterococci genetic markers T 90 (days) Target gene (marker name) Storage condition Storage Reference >32 d Enterococci 23S rRNA ( Entero1) Sewage spiked beach sand drained with seawater 22 Yamahara et al. (2012) >28 d Enterococci ( 23S rRNA, ENT - 23 ) LS: 10% (vol/vol) SPRW 4 Our study SM: 10% (vol/vol) SPRW 4 27 37 8.35 LS: 10% (vol/vol) SPRW 27 6.46 LS: 10% (vol/vol) SPRW 37 4.34 Enterococci ( 23S rRNA, Entero1) Beef manure amended soil, 80% field capacity moisture 25 Rogers et al. (2011) Bacteroidales spp. genetic markers T 90 (days) Target gene (marker name) Storage condition Storage Reference 27.00 B.thetaiotaomicron alpha - mannanase ( BT - am) SM: 10% (vol/vol) SPRW 4 Our study 48 Table 2.7 . >24 d human associated Bacteroidales 16S rRNA ( HF183 ) LS: 10% (vol/vol) SPRW 4 Seurinck et al. (2005) 17.96 B.thetaiotaomicron alpha - mannanase ( BT - am) SM: 10% (vol/vol) SPRW 27 Our study 10 human associated Bacteroidales 16S rRNA ( HF183 ) LS: 10% (vol/vol) SPRW 12 Seurinck et al. (2005) 9.60 B.thetaiotaomicron alpha - mannanase ( BT - am) LS: 10% (vol/vol) SPRW 4 Our study 5.35 Bacteroidales 16S rRNA ( GenBac3) Beef manure amended soil, 80% field capacity moisture 25 Rogers et al. (2011) 3.22 B.thetaiotaomicron alpha - mannanase ( BT - am) SM: 10% (vol/vol) SPRW 37 Our study 2.59 human associated Bacteroidales 16S rRNA ( HF183 ) LS: 5% SPRW 14 Gilpin et al. (2013) 1.75 B.thetaiotaomicron alpha - mannanase ( BT - am) LS: 10% (vol/vol) SPRW 27 Our study 1.00 B.thetaiotaomicron alpha - mannanase ( BT - am) LS: 10% vol/vol) SPRW 37 Our study a SPRW = sewage spiked river water b LS = in suspension c SM = attached d T 90 value was larger than experiment duration . There are persistence studies that differed from our results. Rogers et al. (2011) found that E.coli ( 23S rRNA ) w as more persistent than enterococci ( 23S rRNA ) in 80% moisture beef O ur study determined that T 90 = 10 days for BT at 4°C in suspension was less than the calculated T 90 > 24 days for HF183 at 4 ° C in 10% (vol/vol) sewage spiked freshwater microcosms (Seurinck et al., 2005) . The extended HF183 persistence could be due to the variability of abiotic and/or biotic factors in the freshwater and/or sewage or the gene 49 itself. Roger et al (2011) also observed that T 90 values from enterococci and E.coli 23S rRNA measured from beef manure amended soil with 80% moisture stored at 25 ° C were less than half of the T 90 values in both 27 ° C and 37 ° C storage conditions in ENT and EC in our study, excluding EC at 27 ° C SM. The decreased persistence in Rogers et al. (2011) could have been due to the storage of the manure at - 20°C for up to six months before application o n soil , thus decreasing the biotic nature of the manure . The variation of persistence of markers in our st udy and previous investigations indicates that there is currently a lack of understanding of the mechanisms of intra - and inter - species CE/ DNA persistence. The accuracy and precision of evaluation s of DNA persistence can be improved with extraction metho ds that elute high quantity and quality DNA. Our ENT and EC persistence data ( Table 2. 5 ) agree d with previous research that determined DNA extraction methods produced highly variable qPCR derived marker concentrations due to biased recovery (Inceoglu, Hoogwout, Hill, & van Elsas, 2010) . The differences in fractional persistence by extraction method in our study could be due to increased resistance to lysis treatments d ue to the peptidoglycan structure in enterococci (Mahalanabis, D o, ALMuayad, Zhang, & Klapperich, 2010) . O ur results supported the use of physical shearing DNA extraction methods without a column purification step (i.e. EPA - DNA ) as an efficient approach for gram negative and positive organisms, and environmental wat er samples (Tang, Gao, Zhu, Chao, & Qin, 2009) . Further exploration of the observed persistence data in our study showed that after 1 day of storage on a SM at 4 ° C, the observed fractional persistence of BT, ENT and EC from EPA - DNA w as 1.29, 1.04, and 0.94, respectively. At 27 ° C, the observed fractional persistence of BT, ENT and EC stored for 1 day on a SM and extracted with EPA - DNA were 0.76, 0.73, and 1.12, 50 respectively. These results indicate d that storage on SM for up to 1 day at 4 ° C , and 27 ° C c ould maintain the CE 2 31% difference, respectively , from initial measurements . Our results support filtering water and transporting samples to the laboratory on a membrane to increase CE persistence. Storage for 28 days on a solid matrix at low temperature (i.e. 4 ° C) p redicted decreases of BT, ENT, and EC concentrations by 92, 83 and 73%, respectively , of initial CE concentrations . Overall, at higher temperatures, 27° C and 37°C, and on a solid matrix, a 90% decrease in CE concentrations of BT occu rre d until 11 and 4 day s of storage, respectively. Our data suggest that water samples could be extended up to 27 days without a < 90% decrease of CE concentrations of BT, ENT and EC if the samples were initially filtered and stored on a solid matrix at low temperature. 51 A PPENDIX 52 Table S 2 . 1 : Best fit e quations and their references that evaluated the persistence of the indicators in this study . Model Equation a Reference First - order exponential decay (ep) Chick (1908) One parameter logistic (lg1) Kamau, Doores, & Pruitt ( 1990) Two parameter logistic (lg2) Peleg ( 2006) Exponential damped decay (epd) Cavalli - Sforza, Menozzi, & Strata (1983) Two - stage (jm1) Juneja et al (2006) Log - logistic (jm2) Juneja et al (2003) Gompertz (gz) Wu, Hung, & Tsai ( 2004) Weibull (wb) Coroller et al. ( 2006) Log - normal (ln) ) (Aragao, Corradini, Normand, & Peleg, 2007) Gamma (gam) Hogg & Craig ( 1978) Biphasic exponential decay with preset breakpoint at 72 hrs (bi) Carret et al. (1991) Biphasic exponential decay (bi3) Carret et al. (1991) Double exponential decay (dep) Peleg ( 2006) Gompertz 3 - parameter (gz3) Gil et al. (2011) Gompertz - Makeham (gzm) Gavrilov & Gavrilov ( 1991) 53 Table S2.1 ( ). Sigmoid - A (sA) ) Peleg ( 2006) Sigmoid - B (sB) ) Peleg ( 2006) a K n = decay constant; t = time in hours; N/N 0 = persistence ratio. 54 Table S 2 . 2 : The best fit models, and the equations that represent ed the persistence patterns of the thre e indicators a stored at three temperature s , and two storage condition s b . Temp (°C) Indicator Storage condition Model c Equation d 4 EC SM bi3 * for 0 t < 14.0 : for t 14.0 : LS bi3 * for 0 t < 16.8 : for t 16.8 : ENT SM bi3 for 0 t < 20.4 : for t 20.4 : LS e p BT SM bi3 for 0 t < 11.8 : for t 11.8 : LS gz3 27 EC SM bi3 * for 0 t < 14.0 : for t 14.0 : LS bi3 * for 0 t < 2.4 : for t 2.4 : ENT SM jm1 LS bi3 * for 0 t < 0.6 : for t 0.6 : 55 Table S2.2 ( ) . BT SM bi3 for 0 t < 9.4 : for t 9.4 : LS bi3 for 0 t < 4.1 : for t 4.1 : 37 EC SM e p LS jm2 ENT SM bi3 * for 0 t < 14.0 : for t 14.0 : LS bi3 for 0 t < 3.6 : for t 3.6 : BT SM bi3 * for 0 t < 5.3 : for t 5.3 : LS bi3 * for 0 t 60 days in microcosms containing sediment covered with fresh water spiked with sewage effluent and stored at 10°C (Pote et al., 2009) . Another study determined that molecular markers, ENT 23S rDNA ( Entero1 ) , EC 23S rDNA (EPA - EC23S) and general Bacteroidales 16S rDNA (GenBac3), in swine amended soil stored at 10°C had T 90 = 62, 10, and 52 days, respectively (Rogers et al., 2011) . However, t here are few studies that have reported the persi stence of FIB in various storage conditions o n longer time scales . 61 Previously, we reported that the persistence of cellular equivalents of genetic markers from EC - uidA, ENT - 23, and human specific Bacteroides thetataiotaomicron alpha - mannanase (BT - am) via qPCR measurements from sewage spiked river water stored at 4 ° C for up to 28 days was significantly affected by attachment to a membrane, time in storage, and type of indicator ( Chapter 2 ) . We also determined that the biphasic decay model predi cted that the T 90 of BT - am stored attached on a membrane filter (T 90 = 27 days) was twice as large as the predicted value in liquid suspension (T 90 = 13 days) at 4°C. The above study sparked the inte rest into the further investigation of the persistence of FIB on a longer time scale. Therefore, the purpose of this investigation was to: a) Describe the long term persistence (up to 366 days) of three enteric molecular markers, Bacteroides thetaiotaomicron 1,6 alpha - mannanase , enterococci 23S rDNA , and E.coli ui dA, stored in liquid suspension or attached to a solid matrix, at a low temperature (4°C). b) Compare the types of models and their parameters to data from previous short term persistence analyses. 3.2 Methods The methods are described in detail in Chapter 2.2 . Below is a brief an overview of the methods. 3. 2.1 Sample preparation and storage Water from the Red Cedar River was collected on May 1, 2013 and autoclaved . The river water was spiked with 10% (vol/vol) raw sewage from the East Lansing Wastewater T reatment . Attached samples (SM) were prepared by filter ing 100ml of the spiked river water, and 62 producing 39 separately seeded Nucleopore Track Etch polycarbonate membrane filters (0.45 , 47 mm diameter , Whatman Inc., Piscataway, NJ ) . Additiona lly, 4.2 L of 10% (vol/vol) sewage spiked river water was divided into 14 liquid suspension samples (LS). All samples were stored in the dark at 4°C. 3. 2.2 Sample processing and DNA extraction Triplicate samples from SM, and one LS sample bottle were rem oved on days 0, 31, 61, 93, 128, 155, 187, 219, 248, 279, 306 , 337 , and 366 ( SM only) days. Then, LS samples were membrane filtered in 100 ml increments onto 3 separate membrane filters . DNA extraction from all samples was performed using the EPA DNA Crude Extraction Method (USEPA , 2012b) . The concentration s of the total eluted DNA (ng/ l ) was recorded , and the samples were stored at - 80°C. T he EPA DNA Crude Extraction Method (EPA - DNA) was chosen because it was determined in Chapter 2 that the concentrations of total eluted DNA were statistically la rger than two commercial methods, and had the best detection limit. 3. 2.3 qPCR quantification of three indicators Molecular markers from Bacteroides thetataiotaomicron alpha mannanase (BT - am), Escherichia coli uidA (EC - uidA), and Ent erococcus spp. 23S rDNA (ENT - 23) were chosen because they were previously used to measure general and human specific (BT - am only) pollution (Frahm & Obst, 2003; Srinivasan et al., 2011; Yamahara et al., 2012) . The primers, probes, amplified sequence length, and qPCR protocols for the markers are described in Table 2. 1 . Preparation of the qPCR reagents and the standard curve s were completed as described in Chapter 2.2 .3 . For all standard curves, the R 2 was greater than 0.96. 63 Table 3 . 1 : The qPCR reaction efficiency, CT values of the lowest detected dilution s ( and representative copies/5 l ), and detection limits of EC - uidA, ENT - 23, and BT - am a . Genetic marker (Species) E fficiency Average l owest detected (representative copies/5 l ) D etection limits (and indicator persistence as measured by Log N/N 0 ) b B T - am (BT) 97% 36.13 (82 ) 1.34 x 10 3 copies/100 ml (Log N/N 0 = - 1.71 ) ENT - 23 (ENT) 100% 34.62 ( 19.7 ) 6.90 x 10 3 copies/100 ml (Log N/N 0 = - 3. 77 ) EC - uidA (EC) 102% 38.60 ( 1.87 x 10 3 ) 6.51 x 10 5 copies/100 ml (Log N/N 0 = - 2.69) a The genetic markers EC - uidA, ENT - 23 and BT - am represent E scherichia coli (EC) , enterococci (ENT), and B acteroides thetataiotaomicron (BT), respectively. b DNA extracts from EPA - DNA were diluted 5x prior to qPCR, the detection limits of BT, EC, and ENT were transformed to 2.68 x 10^2, 1.30 x 10^5, and 1.38 x 10^3 CE/ 100 ml - water sample . 3. 2.4 Persistence modeling and statistical analyses The statistical and modeling analyses considered all data points. Any time points that were represented by three non - detect replicates were evaluated at the detection lim it in the persistence modeling and statistical analyses, as advised by the USEPA (USEPA, 1991) . Further description of the number of non - detects in the datasets are discussed in the Results Section. As described in Chap ter 2.2.4 , a model fitting tool in R (R Development Core Team, 2013) was used f or data analysis to choose one models out of 17 models. The full list of models are listed in Table S2.1 . Best fit was analyzed with the Bayesian Information Criterion (BIC). The BIC value takes into The model with the smallest BIC value within each dataset was chosen. However, if the smallest BIC value was < 2 of bi3, then bi3 was chosen. T he following models were chosen as best fits for the persistence data : biphasic exponential decay, bi3 (Carret, Flandrois, & Lobry, 1991) ; and log - logistic model, jm2 (Juneja, Marks, Mohr, et al., 2003) . Descriptions of the parameters of best fit 64 models are outlined in Table 2.3 . The parameters of the models were used to calculate the T 90 , and T 99, the time required for 1 and 2 log 10 reductions of the indicator co ncentration , respectively. The predicted T 90 (or T 99 ) , the time needed for 1 log 10 (or 2 log 10 ) reduction wa s also calculated by substituting - 1.0 (or - 2.0) = Log 10 (N/N 0 ), and solving for t in each equation. The standard errors of the T 90 and T 99 values were not provided in the modeling program and were not calculated. - test was performed in order to determine if the initial and final concentrations of cellular equivalents (CE) per 100 ml - water sample of BT, ENT, and EC on a SM were s tatistically different. This statistical analysis was evaluated with SPSS 22.0 (SPSS, Inc., Chicago, IL). M ultiple linear regression analys e s was performed in order to determine how the fractional persistence was affected by time, storage condition and ind icator species i n the following datasets : a) the fractional persistence of replicates from all indicators, and b) fractional persistence specific to the indicator. The analyses were performed with SPSS 22.0 (SPSS, Inc., Chicago, IL) with the following equa tion: with N/N 0 = fractional persistence; b 0 = log (N/N 0 ) intercept ; b with a letter subscript represent ed the coefficients of the following independent variables: S = time in storage (0 - 366 days) ; C = storage condition (LS or SM) ; and I = indicator (BT, EC, and ENT). The dependent variable , fractional persistence, d (N, CE/ 100 ml - water sample ) at t days in relation to its initial concentration (N 0 ) . The class variables transformed t o numerical designations starting at 1 for the class variable closes to the beginning of the alphabet, and increasing by one integer for each class within the variable. The independent variables that were 65 not significantly associated to fractional persiste nce were not removed from regression equation in order to fully explain the associations of the variables to the fractional persistence. 3.3 Results 3.3.1 Summary of molecular marker persistence In total, 72 samples were analyzed by qPCR for each indicator after storage for up to 333, and 366 days in LS, and SM, respectively. At t = 187, and 337 days, there were instances of inhibition (defined as technical duplicates from qPCR differing by > 1 log) from samples that measured EC - uida on a SM , and ENT - 23 in LS, respectively. Overall, 55, 24, and 0% of the samples were below the detection limit for BT - am, EC - uida, and ENT - 23, respectively. A t t 61 days, all samples that measured BT in LS were non - dete cts, except for one sample at t = 219 days ( Figure 3.1A ). The average concentration for BT at t = 219 days was above the detection limit, but was within one standard error of the detection limit ( Figure 3.1A ). The standard error bars for ENT in LS and all indicators on a SM showed that variability of the measurements of the indicator concentration increased with time in storage ( Figure 3.1A - B ). There was a n observable increase in the observed concentration of BT, and E C (but not ENT) in the SM samples after 64, and 279 days, respectively ( Figure 3.1B ). 66 Figure 3 . 1 A - B: The observed and predicted fractional persistence a of three indicators b stored for up to 366 days at 4°C A) in suspension (LS); or B) attached to a solid matrix (SM). a Fractional persistence was m easured as Log (N/N 0 ), where N is concentration of the indicator after t days in storage normalized to cell equivalents (CE) per 100 ml - water sample N 0 is the initial indicator concentration . b represent the observed fractional persistence of Bacteroides thetataiotaomicron ( BT ) , enterococci ( ENT ) , and Escherichia coli ( EC ) , respectively. The lines represent the persistence models for each indicator in the storage conditions. The evaluated persistence models represented were biphasic exponential decay (bi3), and log - logistic model (jm2). Filled in data points illustrate the data that were below the detection limit, and represent 10 the indic ator specific detection limit. The error bars represent one standard error. 3.3.2 Comparison of models estimating indicator persistence Table 3.2 outlines the model parameters for each indicator in the LS and SM storage conditions along with the BIC val ues, and the T 90 and T 99 values predicted from the persistence models . The BIC values of all of the best fit models analyzed in this study are in Table S3.1 . A non - linear model, log - logisti c (jm2), was fit to ENT in LS, while a linear model , b iphasic expon ential decay (bi3) was fit to the remainder of the persistence datasets , ENT on a SM , and BT and EC in both conditions. The slopes (decay rates) of the models in Figure 3.1A - B -4 -3 -2 -1 0 1 0 100 200 300 400 Log (N/No) Time in storage (days) A. In liquid suspension BT EC ENT ENT model (jm2) EC model (bi3) BT model (bi3) -4 -3 -2 -1 0 1 0 100 200 300 400 Log (N/No) Time in Storage (days) B. Attached to a solid matrix BT EC ENT ENT model (bi3) EC model (bi3) BT model (bi3) 67 indicated that their rates of decay were not constant. M odel selection and shape of equations measuring BT and EC in both conditions were affected by non - detect samples. The models also calculated increases in the fractional persistence of BT, EC and ENT on a SM after 71, 253, and 279 days, respectively, in storage (Figure 3.1B ) . At t = 366 days, the models predicted that the fractional persistence of BT and EC on a SM ( log N/N 0 = - 0.126, and - 0.116, respectively) increase d to close to the initial concentration (Figure 3.1B ) . - test analyzed the concentrations (CE/100 ml - wa ter sample) of ENT, EC, and BT on a SM between the initial and final (t = 366 days) time points (n = 6 for each indicator), and determined that the final concentration of ENT was significantly less than its initial concentration (p = 0.006), while the fina l and initial concentrations of BT and EC were not significantly different (p = 0.73, and 0.71, respectively). 68 Table 3 . 2 : Descriptions of the parameters, BIC values, predicted T 90 and T 99 values (days) from the models that evaluated the persistence of cell equivalents three indicators a in two storage conditions b at 4 ° C for up to 366 days. Indicator Storage condition Model c BIC value Predicted T 90 (days) Predicted T 99 (days) Model param eters BT SM bi3 41.78 46.36 , 183.34 d N/A e k 1 = 0.05; k 2 = 0.06; bpt f = 71.23 LS bi3 * 41.24 40.76 N/A e k 1 = 0.06; k 2 = 0.06; bpt = 64.00 EC SM bi3 57.18 94.46 , 327.05 d 188.91 k 1 = 0.02; k 2 = 0.08; bpt = 253.06 LS bi3 * 34.37 35.80 71.59 k 1 = 0.06; k 2 = 0.06; bpt = 78.25 ENT SM bi3 35.05 164.03 , 345.24 d N/A e k 1 = 0.01; k 2 = 0.04; bpt = 279.00 LS jm2 34.90 62.93 138.94 k 1 = - 10.34; k 2 = 3.03 *Denotes i f the BIC value of bi3 was ± 2 of the smallest reported BIC value of the available models. a Indicators included: Bacteroides thetatiaotaomicron , BT; enterococci, ENT; and Escherichia coli , EC. b Conditions included liquid suspension, LS; or attached to a solid matrix, SM. c The persistence models were biphasic exponential decay (bi3) and log - logistic decay (jm2). d The models predicted an increase in indicator concentration that allowed for a second T 90 value. e The persistence model did not predict a 99% decrease in indica tor concentration during the experiment duration. f bpt is the break point value, x, in the bi3 equation in Table 2 .2 . 3.3.3 Comparison of predicted T 90 and T 99 values The T 90 and T 99 values, time in days that wa s needed in order to reduce the indicator population by 90, and 99%, respectively, were calculated from the persistence models ( Table 3.2 ) . The equations of the best fit models are in Table S3.2 . The relative order of persistence in LS based on the T 90 values was: EC < BT < ENT, while the relative order on a SM was: BT < EC < ENT. A T 99 value for BT in LS was not calculated because the detection limit was larger 69 than the concentration needed to convey > 99% reduction of the initial BT concentration (Figure 3.1A ) . The T 99 val ues for ENT and BT on a SM were not calculated because the predicted fractional persistence in each dataset was not reduced to < 1% of the original concentration during the experiment duration (Figure 3.1B ) . In Figure 3.1B, the persistence models of the in dicators, BT, E C, and E NT, on SM predicted that their concentrations decreased to < 10% of the initial concentrations after 46.4, 94.5, and 164.0 days, respectively ( Table 3.2 ) . After decreasing to < 10% of their initial concentrations, the predicted conce ntrations of BT, E C, and E NT on a SM increased to > 10% of the original concentration, creating two time points where the concentration of the indicators was 10% of the initial concentration (Figure 3.1B ) . The second T 90 values for BT, E C, and E NT on a SM were : 183.3, 327. 1 , and 345.2 days, respectively ( Table 3.2 ) . Additionally, EC on a SM had two predicted T 99 values, 188.9, and 28 3.0 days ( Table 3.2 ) . 3.3.4 Comparison of genetic marker persistence at t = 31 days in the persistence models and observed data in both storage conditions of our study and Chapter 2.3.3 A comparison of the observed and predicted N/N 0 at t = 31 and 28 days for ENT, EC and BT in both storage conditions was investigated in order to compare the datasets in this experiment with a p revious short term study ( Chapter 2.3.3 ) , respectively ( Table 3.3 ) . In our study, the largest observed (and predicted) N 31 /N 0 = 0.78 was from ENT stored in LS (N 31 /N 0 = 0.65; ENT on a SM) while the smallest observed N 31 /N 0 = 0.26 was from BT in LS (N 31 /N 0 = 0.14; EC on LS), resp ectively. Also in our study, t here was < 10% error, and > 50% error between the predicted and observed N 31 /N 0 in ENT on a SM (predicted and observed N 31 /N 0 = 70 0.65, and 0.67, respectively) , and EC in LS (predicted and observed N 31 /N 0 = 0.14 and 0.29, respectively) , respectively ( Table 3. 3 ) . The observed N/N 0 from Chapter 2.3.3 and this study at t = 28 and 31 days, respectively, were compared. There was < 50% difference of the following indicators: ENT on a SM and LS, and EC on a SM. The remaining indicators, EC in LS, and BT on a SM and LS, experienced > 50% difference . The predicted N/N 0 from Chapter 2.3.3 and our study at t = 28 and 31 days, respectively, was compared (Table 3.3 ) , and there was > 50% error for all indicators except ENT on a SM, which experienced < 50% error. Table 3 . 3 : Comparison of the fractional persistence of cellular equivalents, Escherichia coli (EC), enterococci (ENT), and Bacteroides thetataiotaomicron (BT), quantified in our study and Chapter 2.3.3 at N 31 /N 0 and N 28 /N 0 , respectively. Indicator Storage condition Observed N x /N 0 a Predicted N y /N 0 a Our study x = 31 days Chapter 2.3.3 x = 28 days Our study y = 31 days Chapter 2.3.3 y = 28 days ENT SM 0.67 0.27 0.65 0.27 LS 0.78 0.18 0.49 0.19 EC SM 0.66 0.01 0.47 0.17 LS 0.29 0.05 0.14 0.05 BT SM 0.33 0.09 0.21 0.08 LS 0.26 0.003 0.17 0.003 a N 0 is the initial concentration of the indicator normalized to cell equivalents (CE) per 100 ml - water sample, and N 28 and N 31 were the indicator concentrations after 28 and 31 days in storage (CE/100 ml - water sample), respectively. The samples were stored in suspension (LS) or attached to a solid matrix (SM). 3.3.5 Linear regression analysis Results of l inear regression analyses of all samples (n = 225) in our study , and the indicator specific data sets, ENT, EC, and BT (n = 72) including the correlation c oefficients of the independent variables, significance of correlation coefficient, and their corresponding R 2 values are outlined in Table s 3.4 and 3.5 . The linear regression analysis of all of the samples had R 2 = 71 0.343 , which was larger than the R 2 value for ENT and BT specific datasets (R 2 = 0.244, and 0.152, respectively). The largest R 2 was 0.609, and was from the ENT specific dataset. All of the independent variables including, time in storage (p < 0.01), indicator species (p = 0.001 , dataset of all samples only ), and storage condition (p < 0. 05 ) significantly affected fractional persistence of all of the linear regression analyses ( Table 3.4 ). T he analys e s also determined the following relative order of indicator persistence: BT < EC < ENT. Table 3 . 4 : Correlation coefficient and p - values of m u ltiple linear regression analyse s that evaluated the association of the frac tional persistence of the indicators a to: storage time (days), storage conditio n b , and indicator species c . - value) Data set (n = sample size) R 2 Storage time (S) Storage attachment (C) Indicator (I) All data (n = 216) 0.343 - 0.005 (p < 0.001) 0.543 (p < 0.001) - 0.217 (p = 0.001) BT (n = 72) 0.152 - 0.002 (p = 0.008) 0.382 (p = 0.014) N/A EC (n = 72) 0.244 - 0.004 (p < 0.001) 0.431 (p = 0.033) N/A ENT (n = 72) 0.609 - 0.008 (p < 0.001) 0.818 (p < 0.001) N/A a The evaluated data sets included : all data, BT only, EC only, and ENT only. b Conditions included liquid suspension, LS; or attached to a solid matrix, SM. c Indicators included: Bacteroides thetatiaotaomicron , BT; enterococci, ENT; and Escherichia coli , EC. 72 Table 3 . 5 : A list of the coefficients of the linear regression equation a , that evaluated the correlations of the independent variables to fractional persistence b . Coefficients of the independent variables (and their standard errors ) of the linear regression equations Data set (sample size) Storage time - S Storage attachment - C Indicator - I Y intercept - b 0 All data (n = 216) - 0.005 (< 0.001) 0.543 (0.109) - 0.217 (0.066) - 1.125 (0.232) BT (n = 72) - 0.002 (0.001) 0.382 (0.152) N/A - 1.482 (0.265) EC (n = 72) - 0.004 (0.001) 0.431 (0.198) N/A - 1.664 (0.346) ENT (n = 72) - 0.008 (0.001) 0.818 (0.176) N/A - 1.531 (0.306) a b 3.4 Discussion Our study is one of the first to investigate and compare the long term persistence (up to 366 days) of three genetic markers, enterococci 23S rRNA (ENT - 23), E . coli uidA (EC - uidA), and B. thetataiotaomicron alpha - mannanase (BT - am) in an attached state and in liquid suspension at 4 °C. Our study is an extension of Chapter 2 , which measured and modeled the short term persistence (up to 28 days) of the above markers in various temperatures. During our study, 31 and 23% of measurements BT on a SM, and EC on a SM were below the detection limit , respectively, while EC in LS and BT in LS produced 21 and 72%, respectively, of their samples that were below their respective detection limits (Figure 3.1A - B) limits were substituted for these ti me points. It is acknowledged that the best - fit models of BT in LS (and on a SM) and EC in LS (and on a SM) were affected by non - and t =219 and 248 (t = 219 , 248, and 279 ) days in storage, respectively. Two best fit models were chosen to represent the persistence patterns of the indicators in two storage conditions, liquid suspension and attachment to a solid matrix. The biphasic 73 exponential decay model was fit to all datasets except one, ENT on a SM, whi ch was f it to log - logistic model (jm2). The selection of bi3 to fit a persistence patter n for BT and EC was not dependent on storage condition. Also, the persistence pattern of the three indicators in LS was described with bi3 . The overwhelming selection o f bi3 as the best fit model for five of the six conditions may indicate that the persistence of genetic markers independent on their source organisms removed from water samples has varying decay rates that is best modeled with a biphasic exponential decay model. However, model selection was influenced by samples that were below the detection limit. This is especially true for BT in LS. There were only three time points where was above the d etection limit. Selection of this m odel , bi3, was evaluated by replacing the time points that were below detection with the detection limit. Therefore, the model and its parameters may not accurately depict the persistence pattern of the dataset. Also, there were two time points during the storage of the SM samples, where the concentration of the three indicators was at 10% of the original concentration of the indicators. These values were reported as T 90 . The first value represented a reduction to 10% of the original concentration of the g enetic marker, and the second value represented regrowth of the markers to 10% of the original population. Reporting both of these values is important because the data demonstrates that genetic marker from FIB in water samples can be measured for months. T herefore, these measurements may cause regulatory consequences for TMDL remediation efforts. During our experiment, the persistence decay rates of the models were not constant (Figure 3. 1A - B) . This is consistent with all of the 4 ° C persistence models in Chapter 2.3.3 , except ENT 74 in LS which was modeled with the linear exponential decay model. The variability in the decay seen in these studies may indicate that rate of marker decay may be affected by time in storage. The T 90 values from our study were simi lar with previous studies that measured genetic markers from E.coli , enterococci and Bacteroid al es spp. in water samples. For example, in the 4°C SM condition, our study and Chapter 2.3.3 determined that BT was the least persistent indicator. Marti et al. (2011) also determined that general Bacteroidales 16S rDNA , Allbac, and pig associated Bacteroidales 16S rDNA , Pig - 1 - Bac, and Pig - 2 - Bac, experienced T 90 > 43 days when stored in 4 ° C microaerophilic microcosms of freshwater spiked with pig manure, which was consistent with BT in LS (T 90 = 40.76 days), and SM (T 90 = 46.36 days) in our study. There were also disagreements between the T 90 values of the indicators in our study compared to previous studies. Marti et al. (2011) determined that pig associated Bacteroidales 16S rDNA marker, Pig - 1 - Bac, stored at 4 ° C in aerobic and microaerophilic microcosms of diluted pig manure spiked to river water had T 90 = 22, and 10.3 days, respectively, which was smaller than in BT in LS and SM (T 90 = 40.76, and 46.36 days, respectively) in our study. Based on T 90 values in our study, EC in LS was the least persistent indicator. However, the least persistent indicator in Chapter 2.3.3 was BT in LS . When comparing the observed N/N 0 of EC and BT in LS at t = 31 days in our stu dy (0.29 and 0.26, respectively, Table 3.3 ), it can be verified that EC in LS in our study persisted longer than BT in LS. This coupled with the discrepancy between the T 90 values of BT in LS and SM at 4 ° C from Chapter 2.3.3 (T 90 = 27, and 9.6 days, respec tively) and in our study (T 90 = 40.76, and 46.36 days, respectively) indicate that experiment duration can affect the persistence of these genetic markers. During our experiment, the concentration of ENT, EC, and BT on a SM increased, which was consistent with a previous study that showed that there was an increase in the concentrations 75 of culturable enterococci, and general and pig - specific Bateroidales genetic markers in pig manure spiked river water stored at 4 ° C in microaerophilic conditions (Marti et al., 2011) . Also, the concentration of Allbac increased over two weeks in microcosms consisting of 28% moisture sand spiked with raw sewage (Eichmiller, Borchert, Sadowsky, & Hicks, 2014) . The above findings may indicate that the clumping of cells attached to a matrix may have created anoxic or microaerophilic micro - environments, which may have decreased the DNA decay of BT, or even supported its growth in our study. Additionally, there was not a statistical difference be tween the initial and final concentrations of EC and BT on a SM, which was consistent with a previous investigation that determined that culturable EC can grow in mescocosms of sewage spiked sediment stored at 25 ° C (Byappanahalli, Roll, & Fujioka, 2012) . These results indicate that attachment to solid particles may offer a mechanism to enhance the persistence of enteric bacteria in the environment. The significance of the variables of time, indicator species, and storage condition to the fractional persistence of the indicators in our study was compared to Chapter 2.3.3 . Both studies showed that storage time significantly affected indicator persistence in the 4°C treatments (p = 0.0 01, Chapter 2.3.3 ; and p < 0.001, our study). However, our study determined that attachment to a solid matrix significantly affected persistence. The results in our study indicated that low temperatures and attachment to a solid matrix can increase the per sistence of genetic markers of E.coli, B.thetaiotaomicron , and enterococci for up to 34 days before a 90% decrease in the concentration. 76 APPENDIX 77 Table S 3 . 1 : BIC values and the standard deviations of the 17 best - fit models t hat analyze d the persistence of genetic markers from three indicators a that were stored in two conditions b at 4 ° C for up to 366 days. BIC values (and standard deviations) Model c BT ENT EC LS SM LS SM LS SM ep 50.80 (1.63) 57.08 (1.78) 43.80 (1.22) 42.82 (1.03) 53.59 (1.83) 66.06 (2.52) lg1 52.75 (1.77) 58.03 (1.85) 46.69 (1.38) 44.88 (1.12) 56.02 (2.03) 67.17 (2.63) lg2 51.79 (1.53) 67.46 (1.00) 46.60 (1.24) 44.63 (1.00) 55.37 (1.78) 68.13 (2.47) epd 172.95 (1.00 95.62 (1.00) 35.92 (0.79) 113.85 (1.00) 296.34 (1.00) 217.74 (1.00) jm1 42.57 (1.05) 49.91 (1.00) 42.23 (1.03) 42.02 (0.91) 41.53 (1.00) 64.53 (2.15) jm2 42.19 (1.03) 58.63 (1.35) 34.90 (0.76) 40.91 (0.87) 38.25 (0.87) 64.12 (2.12) gz 40.57 (0.96) 51.77 (1.00) 36.52 (0.81) 40.35 (0.85) 34.11 (0.75) 63.16 (2.04) wb 42.36 (1.03) 51.91 (1.32) 39.94 (0.94) 41.52 (0.88) 42.29 (0.93) 64.65 (2.34) ln 42.28 (1.03) 51.78 (1.32) 37.62 (0.85) 41.16 (0.88) 39.1 (0.91) 64.23 (2.13) gam 42 . 7 0 (1.05) 51.82 (1.32) 41.52 (1.00) 41.93 (0.90) 41.33 (0.99) 64.85 (2.18) bi 38.98 (0.89) 39.22 (1.33) 39.47 (0.93) 41.65 (0.89) 32.12 (0.68) 62.87 (1.75) bi3 41.24 (0.89) 41.78 (0.81) 37.94 (0.78) 35.05 (0.63) 34.37 (0.67) 57.18 (1.47) dep 41.99 (0.92) 49.32 (1.16) 37.19 (0.75) 45.65 (0.10) 49.60 (0.10) 67.44 (2.18) gz3 40.85 (0.88) 55.73 (0.10) 36.83 (0.74) 45.40 (0.96) 32.41 (0.62) 63.83 (1.90) gzm 55.77 (1.63) 62.21 (1.78) 37.32 (0.75) 47.95 (1.03) 34.20 (0.66) 71.19 (2.52) sA 43.78 (0.99) 50.14 (1.12) 55.42 (5.15) 68.86 (2.30) 79.25 (4.34) 81.64 (3.77) sB 43.46 (0.10) 55.84 (0.10) 39.46 (0.10) 45.81 (0.10) 34.93 (0.10) 84.02 (0.10) a The indicators were: Bacteroides thetataiotaomicron (BT) , enterococci (ENT) , and Escherichia coli (EC) . b The storage conditions were: liquid suspension (LS) and attached to a solid matrix (SM). c The models evaluated were: first - order exponential decay (ep), one parameter logistic (lg1), two parameter logistic (lg2), exponential damped (epd), two - stage (jm1) , log - logistic (jm2), Gompertz (gz), Weibull (wb), log - normal (ln), Gamma (gam), biphasic exponential decay with preset breakpoint at 3 days (bi), biphasic exponential decay (bi3), double exponential decay (dep), Gompertz 3 - parameter (gz3), Gompertz - Makeha m (gzm), Sigmoid - A (sA), and Sigmoid - B (sB). Bi was included in the R programming platform, but its BIC - determined. 78 Table S 3 . 2 : The be st fit equations that were chosen to represent the persistence patterns of threes indicators a stored at three temperatures, and two storage conditions b . Indicator Storage condition Model c Equation d BT SM bi3 for 0 t < 1709.5 : for t 1709.5 : LS bi3 * for 0 t < 1536 : for t 1536 : EC SM bi3 for 0 t < 6073.4 : for t 6073.4 : LS bi3 * for 0 t < 1878 : for t 1878 : ENT SM bi3 for 0 t < 6696 : for t 6696 : LS jm2 a The indicators included: Bacteroides thetatiaotaomicron , BT; enterococci, ENT; and Escherichia coli , EC. b The storage conditions included: liquid suspension, LS; or attached to a solid matrix, SM. c first - order exponential decay (ep) , biphasic exponential decay (bi3), and log - logistic (jm2). d Time, t, is measured in hours. *Denotes i f the BIC value of bi3 was ± 2 of the smallest reported BIC value of the available models. 79 CHAPTER 4. DNA YIELDS, AND CONCENTRATIONS OF ESCHERICHIA COLI UIDA AND ENTEROCOCCI 23S rDNA IN FRESHWATER SEDIMENTS: A COMPARISON OF DNA EXTRACTION METHO DS 80 4.1 Introduction Freshwater sediments provide a long term reservoir for culturable fecal indicator bacteria (FIB), enterococci (ENT) and Escherchia coli ( EC, Thevenon et al., 2012) . These sediments can influence the pollution inputs into recreational wate rs (Pote et al., 2009) . Culture independent methods like quantitative polymerase chain reaction (qPCR) methods were developed by the United State Envi ronmental Protection Agency (USEPA) in order to address the impacts of sewage in environmental waters. These methods can monitor ENT in beaches, and microbial source tracking with human specific Bacteroides thetataiotaomicron (USEPA, 2010b, 2012b) . However, obtaining precise and accurate measurement s of pollutions from sediment has remained a challenge. Improving DNA extraction efficiency is essential for characterizing pollution levels in sediments. Many of the current DNA extraction methods have variable and distinct yields as well as detection li mits for sediments (Xiong, Xie, Wang, & Niu, 2014) . Inconsistent DNA yields can occur from incomplet e cell lysis, or losses during the purification step (Kirk et al., 2004; Sharma et al., 2007) . The concentration of organic matter and clay density of sediments can also affe ct DNA yields (Lloyd, Macgregor, & Teske, 2010) by adsorb ing DNA (Hoshino & Matsumoto, 2007) . DNA extraction efficiency also affects downstream applications. Methods that produce s mall DNA fragments, do not complete ly lys e cells, or elute DNA bound to inhibitors can reduce the qPCR ampl ification of specified genetic marker s (B ürgmann, Pesaro, Widmer, & Zeyer, 2001; Inceoglu et al., 2010; Sharma et al., 2007) . Few studies have investigated how DNA extraction protocols may affect FIB measurements in sediments. Such research would give public health officials a better understa nding of the accumulated pollution in aquatic ecosystems, and provide a better analysis of changes in water 81 quality. Therefore, the purpose of this study was to compare the DNA yields and qPCR measurements of spiked E.coli uidA (EC - uidA) and enterococci 23 S rDNA (ENT - 23) using six modifications of three DNA extraction methods from surface sediments in the Lake St. Clair watershed. E.coli and enterococci were chosen because they represent gram - positive and gram - negative FIB, respectively. Additionally, EC - ui dA is specific to E.coli , and has been used previously to measure E.coli in water samples (Srinivasan et al., 2011) . ENT - 23 was chosen because it was recommended by the USEPA to monitor pollution in recreational waters (USEPA, 2012b) . The specific objectives of this study were to: a. Compare the DNA yields , and qPCR measurements of E.coli uidA , and enterococci 23S rDNA spiked into sediments that were extracted by six modifications of three DNA extraction protocols . b. Identify the DNA extraction protocol and modification that produced the largest DNA yield , and qPCR measurements of E.coli uidA , and enterococci 23S rDNA spiked into sediments from the Lake St. Clair watershed . 4.2 Methods An overview of the experiment design is illustrated in Figure 4. 1 . Each heading in Figure 4. 1 represents a sub - section described below. 82 Figure 4 . 1 : Overview of experimental design. a This method is based on EPA Method 1611 for water samples. b Completed at bead beating step. c Completed as an initial step. d Only applicable for modified methods of MoBio UltraClean® a nd EPA - DNA2. 5. Stastical comparison of DNA yields, and concentrations of E.coli and enterococci from the modified DNA extraction methods from sediments from Anchor Bay and Clinton River. 4. qPCR analysis of E.coli uidA and enterococci 23S rDNA d . 3. DNA extraction and measurements of DNA yields of spiked sediment with modified DNA extraction methods, MoBio UltraClean®, EPA - DNA2 a , and Mobio PowerSoil® (5 replicates per modified method). Manufacterer's protocol (M) G2, DNA sorption blocker b (G) 1min sonification b (S) 0.01% Tween 80 b (W) 2x mass c,c (T) 1:10 (g - sediment:vol - dH 2 O) elution c (E) 2. Sediment Spike: E.coli and enterococci. Sediment was spiked with 15x concentrated overnight cultures E.coli and enterococci : 20ul cells per 1g - sediment (wet wt). 1. Sediment sampling. Sediment was collected from two locations in the St. Clair River watershed, Clinton River (CR) and Anchor Bay (AB). 83 4.2.1 Sediment sampling Lake St. Clair is a small lake in the North American Great Lakes (max depth = 6.4 m, surface area = 1114 km 2 ). It connects Lake Huron to Lake Erie. Surface sediment was removed from Anchor Bay in Northwest Lake St. Cla ir , and the mouth of the Clinton Rive ) on August 21, 2012. The average percent dry weight of the surface sediments collected from AB and CR were 62 and 40%, respectively. The two sites were chos en because of their historic al differences in anthropogenic pollution. The Clinton River was named an USEPA Area of Concern in 1986 because of excess nutrient loading, fecal pollution, and habitat loss (Esman, 2007) . The Anchor Bay watershed has historically been rural with recent increases in population (Baustian et al., 2014) . The sediments from AB (and CR) had the following carbon, nitrogen, and phosphorus concentrations: 80.2 (34.8), 3.2 (1.6), and 279.6 ug/g - dry wt (437.2 ug/g - dry wt), respectively. The sediment samples were stored in a cooler on dry ice during transport, and then at - 2.1 Sediment spike with enterococci and E.coli . The sediments were thawed at 4°C. Once thawed, the sediments had a slurry consistency. Overnight cultures of E.coli ATCC strain 15597 and Enterococcus faecalis ATCC strain 19433 were mixed together after two washings with sterile dH 2 0, and conce ntrated 15x. Then, 20 µl of the solution was spiked in to 1 g (wet wt ) of surface sediments from AB and CR. The sediments were briefly vortexed to initiate mixing and stored overnight in the dark at 4°C to facilitate adhesion to sediment particles. In order to address the background levels of ENT and EC, samples of un - seeded sediments from AB and CR (negative controls) were also stored overnight at 4°C. Positive controls (20 µl of the spiking solution) were stored overnight at 4 ° C. 84 4.2.2 DNA extrac tion of spiked sediment with modified methods of MoBio UltraClean ® , EPA - DNA2, and Mobio PowerSoil ®. All DNA extraction methods in our study included a bead beating step, and are listed as follows : MoBio PowerSoil ® (MB - PS , MoBio Laboratories, Inc., Carlsbad , CA ), Mobio UltraClean ® (MB - UC , MoBio Laboratories, Inc., Carlsbad, CA ) , and a method (EPA - DNA2) based on USEPA Method 1611 (USEPA, 2012b) . MB - PS and MB - UC required 0.25 g (wet - UC was chosen because it was shown to achieve DNA yields similar to laborious home - made DNA extractio n methods , and it had a low detection limit for F. tularensis in soils (Whitehouse & Hottel, 2007) . MB - PS was chosen because downstream applications showed increased species richness in agricultural and clay soils (Inceoglu et al., 2010) . EPA - DNA2 was developed for this study. Procedures for EPA - DNA2 are outlined as follows: 1 g (wet mass) of sediment was added to a 2 ml centrifuge tube and cent rifuged at 10,000x g for 30 s to remove excess liquid. Then, 0.3 g of sterilized and acid washed glass beads with diameter 212 - 300 µm (Sigma - Aldrich Corp., St. Louis, MO), and 500 µl of AE buffer (Qiagen, Inc., Valencia, CA) were added. The sediment was ho mogenized in Mini - BeadBeater 8 (Biospec Products, Inc., Bartlesville, OK) for 50 s. The sediment was centrifuged for 1 min at 12,000x g, and the supernatant was recovered. An additional 200 µl of AE buffer was added to the sediment, and it was homogenized for 10 s. The sediment was centrifuged again for 1min at 12,000x g. The supernatant was combined with the previously recovered supernatant. The combined supernatant was centrifuged for 5 min at 12,000x g. The supernatant was recovered and contained the elu ted DNA. 85 T he following modifications (and their abbreviations) were applied to each DNA extraction method : , no modification ( M ) ; 2) addition of 0.45 g of DNA sorption blocker reagent , G2 (GEUS, Copenhagen, Denmark) ( G ); 3 ) sonif ication (FS220H, Fisher Scientific, Waltham, MS) of a microcentrifuge tube containing the sediment for 1 min in dH 2 O at room temperature ( S ); 4) addition of 0.01% Tween 80 to bead beating matrix ( W ); 5 ) 2x mass of sediment ( T , only applied to MB - UC and EPA - DNA2 ); and 6) 1:10 elution of 1g - sediment (wet mass) per 10 ml sterile dH 2 O (Boehm et al., 2009) , and membrane filtered with a polycarbonate filter ( E , 0.45 . Modifications #2 - 4 were completed before the bead beating step, while modifications #5 - 6 were the initial steps of the DNA extraction methods. Within each modified DNA extraction method, there we re five replicates of spiked sediment from AB and CR. Three replicates of negative controls from AB method. Three replicates of the positive controls, the spiking sol utions, were evaluated with the elution volumes of EPA - DNA2, MB - UC, and MB - PS were: 700, 50, and 100 µl, respectively. All of the modifications, except G2, have bee n used in previous culturable bacterial enumeration studies to remove cells from various matrices (Boeh m et al., 2009; Downey, Da Silva, Olson, Filliben, & Morrow, 2012; Natvig, Ingham, Ingham, Cooperband, & Roper, 2002; Thevenon et al., 2012) . G2 is a DNA sorption blocker reagent with glass beads that is manufactured by the Geological Survey of Denmark and Greenland (GEUS). G2 was analyzed in our study because DNA yields from low biomass sediments increased by > 1000x when G2 was added to MoBio PowerLyzer PowerSoil® (MoBio Laboratories, Inc., Carlsbad, CA ) (Jacobsen, Nielsen, & Bælum, 2012) . The DNA yield of each replicate of our study was 86 measured with a UV spectrophotometer at OD 260 , and transformed to ng /g - dry wt. The eluates were stored at - 80 ° C. The average theoretical DNA yields of the positive controls, 20 µl of the spiking solution, produced from the - DNA2, MB - UC and MB - PS were normalized to ng /g - dry wt and represented a spiked sample of sediment from AB (and CR), and were 6.48 x 10 4 (1.02 x 10 5 ), 1.20 x 10 3 (1.88 x 10 3 ), and 5.09 x 10 2 ng/g - dry wt (7.98 x 10 2 ng/g - dry wt), respectively. The average DNA yields measured from of the negative controls, unseeded - DNA2, MB - UC, and MB - PS from AB (and CR) were 6.63 x 10 4 (3.06 x 10 5 ), 1.58 x 10 4 (3.68 x 10 4 ), and 1.17 x 10 4 ng/g - dry wt (2.2 x 10 4 ng/g - dry wt), respectively. 4.2.3 qPCR analysis of E.coli uidA and enterococci 23S rDNA . Measurements of the spiked EC and ENT in the replicates of each modified DNA extraction method were measured via genetic markers from enterococci 23S rDNA (ENT - 23) and E.coli uidA (EC - uidA) with qPCR using the Roche LightCycler® 480 Instrument (Roche App lied Science, Indianapolis, IN) . The primer and probe sequences and qPCR protocols are described in Table 2. 1 . Genomic DNA of overnight cultures of E.coli ATCC strain 15597 and Enterococcus faecalis ATCC strain 19433 was extracted using Qiamp DNA Mini Kit ® (Qiagen Inc, Valencia, CA) . The standard curves of ENT - 23 and EC - uidA calculated their respective indicator concentrations. The genomic DNA was se rially diluted in a 1:10 dilution series with at least six steps. The average efficiencies, r 2 values, and CT value of the lowest detected dilution step of the standard curves for ENT - 23 (and EC - uidA) were: 95% (96%), 0.99 (0.95), 35.07 (37.15), 87 respective ly. A new standard curve was made after every 4 qPCR runs. The theoretical detection limits of the markers , and the DNA extraction method ( Table 4.1) . One modification, 2x mass (T), affected the detection limit s of MB - UC and EPA - DNA2 ( Table 4.1) . Table 4 . 1 : Calculated detection limits of the qPCR amplified cellular equivalents of E.coli and enterococci extracted with the modified DNA extraction methods a in 20 µl qPCR reaction volumes from sediments from the Clinton River (CR) and Anchor Bay (AB). Kit a Sample mass (g - wet wt) b Indicator c Detection limits d Clinton River (CE/g - dry wt) Anchor Bay (CE/g - dry wt) MB - PS 0.25 ENT x 10 2 x 10 2 EC 5.61 x 10 3 8.79 x 10 3 MB - UC 0.25 ENT 7.17 x 10 1 1.12 x 10 2 0.5 3.58 x 10 1 5.62 x 10 1 0.25 EC 2.80 x 10 3 4.40 x 10 3 0.5 1.40 x 10 3 2.20 x 10 3 EPA - DNA2 1 ENT 2.51 x 10 2 3.93 x 10 2 2 x 10 2 x 10 2 1 EC 9.81 x 10 3 1.54 x 10 4 2 x 10 3 x 10 3 a The DNA extraction methods were: MoBio PowerSoil® (MB - PC), MoBioUltraClean® (MB - UC), and an EPA Method 1611 based method (EPA - DNA2). b The modification, 2x mass (T), affe cted the detection limits of MB - UC and EPA - DNA2. c The indicator organisms were: enterococci (ENT) and E.coli (EC). d The detection limits were transformed into cell equivalents (CE) per g - dry wt. One amplified copy of EC - uidA represented one CE of EC (Srinivasan et al., 2011) , while four amplified copies of ENT - 23 represented one CE of ENT (USEPA, 2012b) . The concentrations of ENT - 23 and EC - uidA from the analytical du plicates were averaged, and transformed to CE/g - dry wt. The average ENT concentrations of the positive controls, spiking - DNA2, and MB - UC were transformed to CE/g - dry wt, and were 2.74 x 10 8 (1.75 x 10 8 ), and 1.07 88 x 10 7 CE/g - dry wt (6.81 x 10 6 CE/g - dry wt), respectively. The average concentrations of EC calculated from the positive controls, spiking solutions, for AB (and CR) sediment were protocols of EPA - DNA2, and MB - UC, and were 1.45 x 10 9 (9.27 x 10 8 ), and 7.64 x 10 8 CE/1 g - dry wt (4.87 x 10 8 CE/1 g - dry wt), respectively. The average EPA - DNA2, and M B - 3.93 x 10 2 - the detection limit, 2.51 x 10 2 - the detection limit), and 1.46 x 10 5 CE/g - dry wt (2.76 x 10 4 CE/g - dry wt), respectively. The protocols of EPA - DNA2, and MB - UC from AB (and CR) were 4.18 x 10 6 (1.24 x 10 7 ), and 1.15 x 10 7 CE/g - dry wt (7.24 x 10 4 CE/g - dry wt), respec tively. 4.2.4 Kruskal - Wallis statistical analyses of DNA yields and concentrations of E.coli and enterococci produced from the modified DNA extraction methods from AB and CR sediments In each sediment location, three n on - parametric - Wallis tests, were performed with SPSS 22.0 ( SPSS, Inc., Chicago, IL ) in order to determine if the modified DNA extraction methods affected 1) the DNA yields; 2) EC concentrations; and 3) ENT concentrations . Within each location, the statistical analysis first ranked the DNA yields, EC concentrations, and ENT concentrations of each replicate from the modified DNA extraction methods. The rankings were averaged per each modified DNA extraction method. Then, the ranked mean s were evaluated for significant differences (p < 0.05) using a pairwise comparison from the Kruskal - Wallis test. The modified MB - PS methods were excluded from the statistical evaluation of the qPCR amplified ENT and EC concentrations due to low DNA yields , with 89 further explanation in the Results 4.3.1 . The Kruskal - Wallis test was chosen for all analyses because the ANOVA assumption of in the sample distributions . 4.2.5 Comparison of the DNA yields and FIB concentrations produced from the modified DNA extraction methods to the ir respective standard method Within each sediment location, the average DNA yields, and ENT and EC concentrations produced from the modifications of EPA - DNA2, MB - UC, and MB - PS (DNA yields only) were compared Y X /Y M , where Y M was the extraction method, and Y X was the average DNA yield, or ENT and EC concentration prod uced from x modification of a DNA extraction method. The subscripts of the modifications were: W was 0.01% Tween 80 , T was 2x mass; S was sonication ; E was 1:10 elution; and G was G2 DNA sorption blocker . 4.3 Results 4.3.1 Site specific comparison of the DNA yield produced by the modified DNA extraction methods The average DNA yields from the spiked sediments from AB (and CR) produced across all - DNA2, MB - UC, and MB - PS were: 7.79 x 10 5 (9 .97 x 10 5 ), 3.37 x 10 4 (1.39 x 10 4 ), and 6.48 x 10 3 ng/g - dry wt (8.74 x 10 3 ng/g - dry wt), respectively. EPA - DNA2 + G produced t he largest average DNA yields from CR and AB sediments, and were 5.22 x 10 6 and 4.36 x 10 6 ng/g - dry wt, respectively ( Figure 4.2 A - B ). 90 Of the three DNA extraction methods, EPA - DNA2 had the largest average DNA yields within each modification ( Figure 4.2A - B) . The modified MB - PS methods had the lowest average DNA yields, the exceptions were MB - PS +S in CR, and MB - PS +G in AB and CR se diments, which had yields larger than the respective modifications of MB - UC ( Figure 4.2A - B) . When comparing the modifications, the average DNA yield from AB sediment using EPA - DNA2 + G was significantly larger than MB - UC +E, +S and MB - PS +E, +S (p < 0.05) . For CR sediment , the average DNA yield from EPA - DNA2 + G was also significantly larger than MB - PS + E, +S, +W, and MB - UC +S, +W (p < 0.05) . 91 Figure 4 . 2 A - F : A - B) DNA yields, and qPCR quantified concentrations a of C - D ) enterococci (ENT), and E - F ) E scherichia coli (EC) spiked into surface sediment from Anchor Bay (AB) and the Clinton River (CR), and extracted by modifications b of three DNA extraction methods c . 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 M W T S E G DNA yield (ng/g - dry wt) A. DNA yields: AB sediment MB-PS MB-UC 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 M W T S E G DNA yield (ng/g - dry wt) B. DNA yields: CR sediment MB-PS MB-UC EPA-DNA2 1E+0 1E+2 1E+4 1E+6 1E+8 1E+10 1E+12 1E+14 M W T S E G Enterococci conc. (CE/g - dry wt) A. spiked ENT conc. in AB sediments MB-UC EPA-DNA2 1E+0 1E+2 1E+4 1E+6 1E+8 1E+10 1E+12 1E+14 M W T S E G Enterococci conc. (CE/g - dry wt) B. spiked ENT conc. in CR sediments MB-UC EPA-DNA2 1E+0 1E+2 1E+4 1E+6 1E+8 1E+10 1E+12 1E+14 M W T S E G E.coli conc. (CE/g - dry wt) C. spiked EC conc. in AB sediments MB-UC EPA-DNA2 1E+0 1E+2 1E+4 1E+6 1E+8 1E+10 1E+12 1E+14 M W T S E G E.coli conc. (CE/g - dry wt) D. spiked EC conc. in CR sediments MB-UC EPA-DNA2 92 Figure a The DNA yields were reported in ng/g - dry wt, and the qPCR amplified concentrations of E. coli and enterococci were normalized to cell equivalents (CE) per g - dry wt. b The modifications were: (M); 0.01% Tween 80 (W); 2x mass (T); sonication (S); 1:10 elution (E); and G2 DNA sorption blocker (G). c The DNA extraction methods were: MB - PS (MoBio PowerSoil®), MB - UC (MoBio UltraClean®), and EPA - DNA2 (based on USEPA Method 1611). There were five replicates per each modified DNA extraction method. The error bars represent one standard error. 4.3.2 Site specific c omparison of the qPCR amplified concentrations of enterococci and E.coli produced by the modified DNA extraction methods . DNA extracts from the modified MB - PS methods were not used to evaluate the concentrations of EC and ENT in sediments from AB and CR , because of its comparatively low DNA yields ( Figure 4.2 A - B ). The average ENT concentrations from AB (and CR) sediments across all of the modified methods, i - DNA2 and MB - UC were: 2.57 x 10 9 (2.09 x 10 9 ), and 1.03 x 10 8 CE/g - dry wt (8.12 x 10 7 CE/g - dry wt), respectively. Within each modification, EPA - DNA2 provided the largest average ENT concentrations from CR and AB sediments, except MB - UC +M (2.71 x 10 8 CE/g - dry wt) which was slightly larger than EPA - DNA2+M (1.52 x 10 8 CE/g - dry wt) in CR ( Figure 4.2C - D ). Using EPA - DNA2 + G , t he largest average ENT concentrations in CR and AB sediments were measured to be 5.04 x 10 9 and 7.46 x 10 9 CE/g - dry wt, respectively ( Figure 4.2C - D ), and were significantly larger t han MB - UC+W, +E, +T, and +S (p < 0.05) . The average EC yields from AB (and CR) sediments across all of the modifications of EPA - DNA2 and MB - UC methods, including the x 10 10 (5.30 x 10 10 ), and 2.26 x 10 10 CE/g - dry wt (2.1 x 10 10 CE/g - dry wt), respectively. Again, t he largest average concentration of EC from CR and AB sediments were measured from EPA - DNA2 +G 93 (1.17 x 10 11 CE/g - dry wt), and EPA - DNA2 +E (9.89 x 10 10 CE/g - dry wt) , respectively ( Figure 4.2E - F) . The average EC concentration in AB sediment extracted with EPA - DNA2 +E was significantly larger than EPA - DNA2 +M, +T (p < 0.05) . In CR, t he concentration of EC extracted with EPA - DNA 2 +G was significantly larger than EPA - DNA2 +T, and MB - UC +W and +S (p < 0.05) . 4. 3.3 Comparisons of DNA extraction efficiency of the modified DNA extraction methods to the ir respective The ratio, Y X / Y M , compared the DNA yields, and ENT and EC concentrations produced from the modifications of EPA - DNA2, MB - UC, and MB - PS (DNA yields only) to their (Table 4.2) . Therefore, Y x /Y M > 1.0 indicated that the modification, x , produced a larger DNA yield or FIB concentration than the of MB - PS and MB - UC (Table 4.2) . Specifically, the DNA yields and FIB concentratio ns measured in CR sediments were reduced dramatically after the modifications were applied to MB - PS (DNA yields only) and MB - UC (Table 4.2) . However, 0.01% Tween 80, 1:10 elution and G2 improved the extraction efficiency of the standard method of EPA - DNA2 from AB and CR sediments (Table 4.2) . 94 Table 4 . 2 : Within each sediment location, Anchor Bay and Clinton River, a ratio compared the average DNA yields, and concentrations of enterococci and E.coli produced from modifications DNA extraction Method a Site b Output c Ratio: Y X /Y M d W E T G S MB - PS AB DNA yield 1.25 0.29 - 0.75 0.15 CR 0.22 0.26 - 0.73 0.07 MB - UC AB DNA yield 6.03 0.14 2.67 0.23 1.64 CR 0.17 0.24 0.55 0.28 0.04 AB ENT conc. 0.15 0.07 0.35 0.60 0.10 CR 0.05 0.18 0.28 0.90 0.42 AB EC conc. 0.89 0.46 0.46 0.40 0.15 CR 0.11 0.66 1.35 0.54 0.38 EPA - DNA2 AB DNA yield 1.41 0.85 0.63 59.18 0.38 CR 1.11 0.45 0.70 24.83 0.33 AB ENT conc. 3.79 4.14 0.27 7.47 1.87 CR 3.79 4.14 0.27 7.47 1.87 AB EC conc. 2.28 4.24 0.40 5.63 1.69 CR 2.28 4.24 0.40 5.63 1.69 a The DNA extraction methods were: MB - PS (MoBio PowerSoil®), MB - UC (MoBio UltraClean®), and EPA - DNA2 (based on USEPA Method 1611). b The sites were Anchor Bay (AB) and Clinton River (CR). c The outputs were: DNA yields (transformed to ng/g - dry wt), enterococci concentration (ENT, transformed to CE/g - dry wt), and E.coli concentra tions (EC, transformed to CE/g - dry wt). d Y X was the DNA yield or FIB concentration of the modification, x . The modifications were: 0 .01% Tween 80 (W) , 2x mass (T), sonication (S) , 1:10 elution (E) , and G2 DNA sorption blocker (G). Y M was the DNA yield or 4.4 Discussion The ability to improve DNA extraction efficiency and amplify FIB markers such as ENT - 23 and EC - uidA from sediments is an ongoing pursuit. Our study examined two sediment locations, the mout h of the Clinton River, and Anchor Bay in Northwest Lake St. Clair . Sediments from AB had twice the concentration of carbon and nitrogen compared to CR, while CR had twice the concentration of phosphorus than measured in AB. Previous research has determine d that DNA 95 extraction efficiency decreased in soils with large organic content (Zhou, Bruns, & Tiedje, 1996) . However, within each sedi ment location of our study, modifications of the DNA extraction methods overall resulted in minimal differences based on location in extraction efficiency of the DNA yields, and FIB concentrations. Our results suggested that two of the investigated DNA ex traction methods, MB - UC and EPA - DNA2, overcame some of the challenges impeding the extraction of DNA in sediments - UC performed better than its modifications, which indicated that optimization was not generally needed. Within EPA - DNA2, the modifications, 1:10 elution, sonication, 0.01% Tween 80 and G2, resulted in larger ENT and EC concentrations when compared to its standard method for AB and CR sediments (Table 4.2) . A previous comparison of e ighteen methods to measure culturable ENT and EC in sands also determined that significant differences (p < 0.05) in the measurements were mainly restricted to methods that used blending versus shaking (Boehm et al., 2009) . This suggests that the variations in sampling protocols account for minimal variation of ENT and EC measurements in sands and sediments. The number of steps of a DNA extraction method may have s ome impact on its efficiency. There were only three steps in the standard method for EPA - DNA2, while MB - UC and MB - PS had five and six steps, respectively. It has also been shown that DNA purification with a silica MB - UC and MB - PS decreased DNA yields in sediments by 9 - 55% when compared to the same methods that purified DNA with an acrylamide gel (Lloyd et al., 2010) . Preferential binding of humic acids and other organic compounds to a silica column was presumed to be one of the mechanisms that decreased DNA 96 extraction efficiency (Zhou et al., 1996) , which may have accounted for the decrease in DNA extraction efficiency in MB - UC and MB - PS compared to EPA - DNA2 in our experiment . - UC had > 2 log increase of the naturally occurring ENT concentration when compared to the standard method of EPA - DNA2. These results are different from the seeded samples, and may indicate that the indigenous populations of ENT have a distinct extraction efficiency compared to the spiked concentr ations used in this study. The differential extraction efficiency may be due to increased resistance to disrupt the cell wall with bead beating, as was observed in a previous study of viable but not culturable Enterococcus faecalis in microcosms of sterili zed freshwater stored at 4 ° C for 16 days (Signoretto, del Mar Lleo, Tafi, & Canepari, 2000) . In our study, a novel DNA sorption blocker reagent, G2, produce d larger DNA yields, and FIB concentrations from AB and CR sediments ( Figure 4.2A - F ). G2 contains a proprietar y blocking reagent and glass beads. It is produced by The Geological Survey of Denmark and Greenland (GEUS). It was previously shown to help significantly increase (p < 0.05) qPCR measurements of rpoB from clay subsoil with low biomass (Jacobsen et al., 2012) . MB - UC used with G2 was shown to increase the DNA extraction efficiency of Dehalococcoides spp. in clay till by 3 logs (Bælum et al., 2014) . As p revious research also suggested, G2 may act as an adsorption - site competitor like spiked salmon sperm DNA, which increased DNA yields ex tracted from soils (Paulin & Nicolaisen, 2013) . The time and money needed to extract indicator DNA from each of the modified methods was also compared. DNA extraction using MB - UC and MB - PS can be co mpleted in 30 min, while EPA - - UC, MB - PS and EPA - DNA2 were 3, 5, and 2 $USD, respectively. The additional costs for the 97 materials for the modifications was < 1 $USD, except G2, which added an additional 7 $USD. The additional time required for the modifications were < 10 min. Thus, EPA - DNA3 with G2 was an easy, and time efficient method to extract FIB concentrations from sediments. 98 CHAPTER 5. HISTORIC AL ASSOCIATIONS OF FECAL INDICATOR CONCENTRATIONS TO ANTHROPOGENIC ACTIVITIES AND CLIMATE IN FREHSWATER SEDIMENTS 99 5.1 Introduction T he United States has a fragmented history of w ater quality standard s . C ulturable total coliforms were first regulated in drinking water in 1914 , and were used to measure ambient water quality until the 1950s (Wolf, 1972) . By the 1960s , fecal coliforms were used to monitor wastewater discharges, and then to guide the safety of recreational waters (Dufour, 2001) . Since 1986, culturable Escherichia coli (EC) and /or enterococci (ENT) have been the indicators of choice for monitoring recreational water s (USEPA, 2003) . Recently, quantitative polymerase chain reaction (qPCR) measurements of ENT 23S r D NA and Bacteroid al es 16S rDNA were recommended by the USEPA to monitor water quality at beaches (USEPA, 2010b, 2012b) . Current methods primarily measure fecal ind icator bacteria (FIB) in the water column. However, qPCR measurements of cell equivalents (CE) and c ulturable concentrations of ENT and EC revealed higher concentrations in sediment s and benthic sand, respectively (Eichmiller, Hicks, & Sadowsky, 2013; Wheeler Alm et al., 2003) . T here was a ~5 - log increase of CE of ENT in the top 11 cm of sediment and sand cores obtained from Lake Superior when compared to culturable forming units ( CFUs ) (Eichmiller et al., 2013) . It appears that culturable ENT and EC stabilized (< 1 - log reduction) in s ediment microcosms stored at 10°C for 60 days , indicating that sediments have the potential to be long term reservoirs of FIB (Pote et al., 2009) , and their resuspension can negatively impact water quality (D L Craig et al., 2004) . Scientific questions regarding water quality trends over large time scales and their associations to climate change and human interactions have remained unanswered. An approach which measures FIB in sediment cores could be used to evaluate water quality ch anges over longer timescales. For example, a previous study observed v iable concentrations of EC and ENT in sediments deposited up to 245 ± 45 years ago , and their observed concentrations increased 100 when eutrophic conditions were observed in Switzerland (Thevenon et al., 2012) . Therefore, the purpose of our study was to evalua te the relationship between the c oncentrations of enterococci 23S rDNA and E.coli uidA in sediment cores from t wo sub - watershed of a lake system to climatic and anthropogenic variables over fairly long time scales . Two locations representing different sub - watersheds in the Lake St. Clair watershed were chosen. Anchor Bay, Northwestern Lake St. Clair represented historic al agricultural lands that have transitioned to developed areas, while the mouth of the Clinton River in Western Lake St. Clair represented an increase in developed lands (J. A. Fry et al., 2011) . Sediment co res were taken from both locations, and assessed fo r changes in sedimentary concentrations of total nitrogen, total al anthropogenic activities were also evaluated, and included human population in Clinton River and Lake St. Clair watershed s. Finally, historic al climate variables were also obtained for inclusion in the analysis, and included river discharge, and air temperature. 5.2 Methods 5.2.1 Field site description, sample collect ion and processin g The Clinton River and Anchor Bay sub - watersheds within the Lake St. Clair watershed, and are 1968 and 443 km 2 , respectively. , Figure 5. 1 ), were chosen because they have distinct historical records of environmental perturbation (Healy, Chambers , Rachol, & Jodoin, 2008) . Also, these sites were reported to be sediment accumulation zones, and thus are useful sites for the reconstruction of historical changes (International Joint Commission & Upper Great Lakes Connecting Channels Study, 1988) . At each site, a pontoon boat with a Pneumatic 101 Vibracore Core Sampler with 14, 000 vibrations per min ( Gr eat Lakes Environmental Center, Traverse City, MI) removed five sediment cores using sterile acetate butyrate tubes with an inside diameter of 9.5 cm. Surface sediment was removed with a petite ponar sampler. Surface sediment and t he cores were stored vertically on dry ice during transport , and subsequently stored vertically at - 80°C. O ne frozen sediment core from each site was aseptically cut into 2 cm v ertical sections ( core length: AB = 86 cm and CR = 58 c m) , and stored at - 80°C . There were a total of 43 and 29 sediment sections from the AB and CR cores, respectively. Figure 5 . 1 : Map of the Lake St. Clair basin, and the sediment core sites labeled with stars: Anchor Bay (AB), a nd the Clinton River (CR). 102 5.2.2 DNA Extraction and qPCR measurements of enterococci 23S rDNA and E.coli uidA The AB and CR s ediment sections homogenized with EPA - DNA2 with G2 (Geological Survey of Denmark and Greenland, GEUS, Coppenhagen, Denmark). In triplicate, the total DNA of 1 g (wet weight) of each sediment section was extracted using the method EPA - DNA2 +G2, outlined in Chapter 4.2.2 . Enterococci 23 S rRNA (ENT - 23) and E . coli uidA (EC - uidA) were measured using qPCR. The primer probe sequences, qPCR protocols and reaction matrices are outlined in Table 2.1 . The genome copy number of EC - uidA was one, while the estimated copy number of ENT - 23 was an aver age of four copies per Enterococcus faecalis cell (Srinivasan et al., 2011; USEPA, 2012b) . The concentrations of ENT - 23 and EC - uidA were transformed to cell equivalents (CE) per g - dr y wt. The EC - uidA and ENT - 23 concentrations in the DNA extracts were calculated from their respective qPCR standard curves. The ENT - 23 and EC - uidA standard curves were produced from genomic DNA extracted from overnight cultures of Enterococcus faecalis ATC C strain 19433 and E.coli ATCC strain 15597, respectively, using Qiamp DNA Mini Kit (Qiagen Inc, Valencia, CA). The genomic DNA was serially diluted (1:10) to create at least 6 dilution steps for a standard curve. A new standard curve was made after every four qPCR runs. The average efficienc y , r 2 value , and threshold cycle value of the lowest quantified concentration in the standard curve, and its representative copies/rxn of ENT - 23 (and EC - uidA) were : 102% ( 99% ) , 0.98 ( 0.99 ), 34.44 ( 36.05 ), and 19.7 copies/rxn (1860 copies/rxn), respectively. 5.2.3 Sediment c hronostratigraphy Approximately, ~ 30 g (wet weight) of each sediment section was dried at hrs . The percent water content in each sediment section was determined, and used to calcul ate the porosity and commutative mass depths (J weda & Baskaran, 2011) . From the measured vertical 103 profiles of Cs - 137 and excess Pb - 210 activity , t he approximate year s of deposition w ere calculated as previously outlined (Jweda & Baskaran, 2011) . The activities of total 210 Pb, 226 Ra, and 137 Cs were measured using high - resolution gamma - ra y spectrometer, and the details are given in Jweda and Baskaran (2011). Details on the excess 210 Pb ( 210 Pb xs ) and 137 Cs - based chronologies are discussed in Baustian et al. (in prep.). From the peak 137 Cs activity corresponding to 1963, the chronology of t he sedimentary layers were established. There was an overall agreement between 210 Pb xs and 137 Cs based chronologies. 5.2.4 Nutrient m easurements Percent total C (% tC) and total N (% tN) were measured with modifications of a previously described method (B. Fry, 2007) . Briefly, t he samples we re acidified in a 10% 1 M HCl solution overnight at 50°C to dissolve carbonates. The samples were added to a three series system: EuroVector EA3000 CHNS elemental analyzer (EuroVector S.p.A, Milan, Italy), a GC+ O 2 column scrubber system , and an IsoPrime IRMS stable isotope ratio mass spectrometer (Elementar Americas, Inc., Mt. Laurel, New Jersey). The first combustion furnace consisted of Cr 2 O 3 , CoO , and quartz chips. It remained at 1060°C to allow for CO 2 SO 2 , H 2 O, and N 2 combustion prod ucts to form. The H 2 O was scrubbed out in a glass water trap. The samples passed to a second furnace at 650°C that consisted of Cu and quartz chips. The combustion gases passed to a GC column in a 10 cm long glass oxygen trap. Mass spectrometry measured N 2 and CO 2 sequentially from peak height s of 28 , and 44 g/mol , respectively , in the GC column ( 3 m long ) A7627 - 1G, Sigma Aldrich, St. Louis, MO) of varying masses were also prepared. Linear regression analys i s of the standards was calculated using Microsoft Excel (Microsoft Corp., Redwood, Washington, USA) in order to accurately calculate the N and 104 C concentrations in the sediment samples . Concentrations were calculated as % total N and % total C based on the mass of the dried sedi ment. Concentrations of total P (tP) were measured with previously outlined methods (American Publ ic Health Association, 1998; Andersen, 1976) . Briefly, the sediment ( 0.15 - 0.2 g ) was ashed at 550°C for 2 hr in a muffle furnace. The cooled sediment residues were washed with 15 ml 1 M HCl , and boiled for 15 min. The samples were diluted with dH 2 O to 100 ml. The solutions were analyzed for PO 4 - 3 P using a previously described perchloric acid method . Briefly, 50 ml of the sample was added to a flask with one drop of phenolphthalein. If the sample turned red, then enough drops of 5 M H 2 HO 4 were added to dilute the color. A reaction matrix of the following solutions was made: 0.1 M C 6 H 8 O 6 , 5M H 2 HO 4 , 0.005M K(SbO)C 4 H 4 O 6 × ½ H 2 O and 0.043 M (NH 4 ) 6 Mo 7 O 24 × 4H2O. Of this reaction matrix, 8ml was added to the samples and mixed. The samples rested for 10 30 min and the absorbance was measured at 880 nm using Genesys 10S UV - Vis spectrophotometer (Thermo Fisher Scientific Inc, Waltman, MA). A calibration standard of serially diluted phosphate solutions was analyzed with Microsoft Excel (Microsoft Corp., Redwood, Washington, USA) , and used to of tP . 5.2.5 Measurements of the a nthropogenic and climate data The average monthly temperature (°C) was measured at Selfridge Air National Guard Base (42 ° 36 N, 82 ° W ) during 1937 2012 (Midwestern Regional Climate Center, 2014) . The base represen ted the air temperatures of Anchor Bay and Clinton River because of its close proximity to the two sites. The monthly average discharge (m 3 /sec) from St. Clair River ( 1900 - 2012 ) w as previously published (Hunter & Croley, T.E., 1993) , with data continuously added on 105 http://www.glerl.noaa.gov/data/arc/hydro/mnth - hydro.html . The St. Clair River was chosen to represent Anchor Bay because it contributes the majority of the water discharged into Anchor Bay. The average daily discharge (m 3 /sec) of the Clinton River was measured during 1935 - 2012 from the USGS weather station 04165500 ( upstream of CR (USGS, 2014) . These measurements were averaged over a calendar year. The data concerning the human population in the communities of the Clinton River and Anchor Bay watersheds during 1900 - 2010 were gathered from the U.S. Census Bureau (SEMCOG, 2002) . C ommunities in the watersheds were identifi ed with having 50% of their area within the watershed boundaries . A linear regression was performed on the census data gathered from the Anchor Bay and Clinton River watersheds during 1900 2010 , in order to estimate the annual population in the respect ive watersheds, and the linear regression equations were : ; , respectively , where P x wa s the estimated population in the Clinton River or Anchor Bay watershed, and Y wa s the year. Each core section represented multiple calendar years. Therefore, the average air temperature, river discharge from the St. Clair River and the Clinton River, and estimated population data points were calculated for each section of the core within the rang e of years represented in each core section in order to develop a dataset representative of the time frames represented in the cores. 106 5.2.6 Multiple linear regression analys e s Multiple linear regression analyses evaluated two datasets: ENT concentrations, C ENT , deposited during c.1932 - 2012 in both cores; and EC concentrations, C EC , deposited during c.1932 - 2012 in the AB core and during c.1951 - 2012 in the CR core, and their associations to the following indep endent variables : EC (when C ENT was the dependent variable) or ENT (when C EC was the dependent variable) concentrations , C EC or C ENT , respectively; calendar year, Y ; sedimentary nutrient (P, N, and, C) concentration s (ug/g - dry wt), H , N , and A , respectively; watershed population , P ; river discharge , D ; air temperature , T ; and site, S . The ENT and EC linear regression e quations were : ; , respectively. Also, b 0 was the y - intercept , and b with a letter subscript represented the correlation coefficient of each independent variable. In the EC dataset, there was a lag of three sediment core sections in the discharge rate , which translate d to ~6 years. The class variable, site, had two outputs, AB and CR, which were transformed the numbers to 1 and 2, respectively. Independen t variables that were non - significant were not removed from the regression model. 5.3 Results 5.3.1 Sedimentation rate and sediment chronostratigraphy It appears that there was little vertical mixing of sediment in the Cs - 137 profile ( Figure 5.2 ) . T he ma ss accumulation rate s (and linear sedimentation rates) in the AB and CR cores were 0.35 107 g/cm 2 (0.39 cm/yr ), and 0.42 g/cm 2 (0.67 cm/yr) respectively. Therefore, the AB and CR cores represented approximately 255 and 117 years, respectively . Uniform sedimentation rate over the whole core was assumed over this time period which cannot be validated by other methods. Figure 5 . 2 : Cs - 137 radio isotope activity profile (dpm/g - dry wt) of the sediment cores from of the cores (g - dry wt/cm2). 5.3.2 Climatic measurements: a ir temperature and discharge rates T he average annual air temperatures were averaged to represent the time interval estimated in each sediment section, and ranged from 8.4° - 10.27°, and 7.9° - 10.2° C for AB and CR, respectively (Figures 5.3 A and 5.4A, respectively ) . From 1932 to 1951 and from 1978 to 2012, the air temperatur e increased, while it decreased during 1951 to 1978 ( Figure 5.3 A and 5.4A ) . The annual average river discharge representative of AB and CR was averaged within the time intervals estimated in each sediment section. The range of discharge in St. Clair River ( 4 . 5 2 x 10 3 - 5 . 8 4 x 10 3 m 3 /s) was larger than the Clinton River (0.65 x 10 1 - 3 .2 3 x 10 1 m 3 /s, Figures 5.3 B and 5.4B, respectively ) . T he discharge in the Clinton River increased , then decreased, and 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.00 20.00 40.00 60.00 80.00 Cs - 137 conc. (dpm/g) Cummulative mass depth (g/cm 2 ) Cs - 137 radio isotope activity in the sediment cores 108 then increased again from c. 1934 c.1945; c.1945 c.1963; and c.1963 2012, respectively ( Figure 5.4 B) . T he discharge in St. Clair River decreased c. 1971 - 2012 ( Figure 5.3 B). 109 Figure 5 . 3 A - H: Indictors of climatic variables a (A - B ) , population in the watershed b (C) , sedimentary nutrient concentrations c (D - F) , and fecal indicator concentrations d (G - H) i n the Anchor Bay watershed . a A ir temperature at Selfridge Air National Guard Base, Mt. Clemens, MI ( 1937 - 2012 ), and B) discharge (m 3 /s) in the St. Clair River ( 1900 - 2012 ). The annual average air temperature, and 110 discharge were averaged within the time interval estimated in each sediment section of the Anchor Bay core. b P o pulation in the Anchor Bay watershed was e stimated from census data ( 190 0 - 2010 ) . The human population in the watershed was averaged within the time interval estimated in each sediment section of the Anchor Bay core. c Sedimentary nutrient concentrations in the Anchor Bay core (c. 1760 - 2012) were described with: D) total P (ug/g - dry wt), E) % total C , and F ) % total N . d Fecal indicator concentrations in the Anchor Bay core were measured via G) E scherichia coli (EC), and H) Ent erococci (ENT) concentrations ( cell equivalents/ g - dry wt , c. 1760 - 2012). The line s represent the ENT and EC detection limit s in the Anchor Bay core sections . Data points that are not filled indicate samples that were below the detection limit . The error bars represent one standard error. Table S5.1 gives a further description of E.coli and enterococci concentrations in the cores, respectively. 111 Figure 5 . 4 A - H: Indictors of climatic variables a (A - B ) , population in the watershed b (C) , sedimentary nutrient concentrations c (D - F) , and fecal indicator concentrations d (G - H) in the Clinton River watershed . 112 a A ir temperature at Selfridge Air National Guard Base, Mt. Clemens, MI ( 1937 - 2012 ), and B) discharge (m 3 /s) in the Clinton River (1935 - 2012 ). The annual average air temperature, and discharge were averaged within the time interval estimated in each sediment section of the Clinton River core. b P o pulation in the Clinton River watershed was e stimated from censu s data ( 190 0 - 2010 ) . The human population in the watershed was averaged within the time interval estimated in each sediment section of the Clinton River core. c Sedimentary nutrient concentrations in the Clinton River core (c.1895 - 2012) were described with: D) total P (ug/g - dry wt), E) % total C , and F ) % total N . d Fecal indicator concentrations were measured via G) Escherichia coli (EC), and H) e nterococci (ENT) concentrations (cell equiva lents/ g - dry wt, c.1895 - 2012). The line s repre sent the ENT and EC detection limit s in the Clinton River core sections . Data points that are not filled indicate samples that were below the detection limit . The error bars represent one standard error. Table S5.2 gives a further description of E.coli and enterococci concentrations in the cores, respectively. 5.3.3 Anthropogenic attributes: estimated census population and nutrient loading in the cores. The estimated human populations in the Anchor Bay and Clinton River watersheds (1900 - 2010) were averaged to represent the time intervals in each sediment section. In 1900 (and 2010), the population density per km 2 in AB and CR watersheds was: 52.76 and 51.32 people per km 2 , respectively (1.84 x 10 3 and 1.55 x 10 3 people per km 2 , respectively). The Cli nton River watershed had a larger total population compared to the Anchor Bay watershed (Figures 5.3C and 5.4 C) , while the estimated growth rates in both watersheds were similar. Since 1900, the estimated populations in both watersheds have increased by ~1 - log (Figures 5.3C and 5.4 C) . The nutrient concentrations from the CR core steadily increased towards present day ( Figures 5.4D - F ). The tP concentrations in the AB and CR cores were similar until c.1966 (~ 1.5 x 10 2 ug/g - dry wt; Figures 5.3D and 5.4D, respectively ), and thereafter doubled in CR core ( Figure 5.4D ). The range of the % tC in the AB core was larger (6.2 x 10 - 1 140 x 10 - 1 % tC) compared to the CR core (9.3 x 10 - 1 59.3 x 10 - 1 % tC, Figures 5.3E and 5.4E, respectively ), 113 while the range of the % tN in the AB core was smaller (3.0 x 10 - 2 66 x 10 - 2 % tN) compared to the CR core (5.0 x 10 - 2 30 x 10 - 2 % tN, Figures 5.3F and 5.4F ). Starting in c.1856, the % tC and % tN in the AB core began to increase > 1 - log until they reached their largest values in c.1896 (140 x 10 - 1 % tC, and 6.6 x 10 - 1 % tN, respectively), and then began to decrease by > 1 - log until the concentrations returned to the pre - 1856 values at c.1932 ( Figures 5.3E - F ). The % tC (and % tN) was lower in the CR core than in the AB co re during two intervals: pre - 1925 (pre - 1918), and post - 2007 (post - 2009; Figures 5.4E - F ). The concentrations of sedimentary nutrients measured in AB and CR are detailed in Tables 5.1 and 5.2 , respectively. 5.3.4 F ecal indicator c oncentrations in sediment c ores ENT - 23 and EC - uidA concentrations were determined in each section of the AB and CR cores (129 and 87 samples in total, respectively). In the AB (and CR) sediment sections, 9% (13%) and 72% (41%) of the samples were below the detection limits of ENT - 2 3 and EC - uidA, respectively, and those ENT and EC concentrations were reported at their detection limits ( Figures 5.3G - H and 5.4G - H ). The oldest detectable ENT (and EC) concentrations in the AB and CR cores were deposited in c.1760 (c.1776), and c.1911 (c. 1924), respectively ( Figures 5.3G - H and 5.4G - H ). The EC concentrations in the CR and AB cores increased towards present day in both sites, and ranged from 0.14 x 10 7 to 1.69 x 10 7 CE/g - dry wt, and 1.8 x 10 6 to 8.5 x 10 6 CE/g - dry wt, respectively ( Figures 5 .3G and 5.4 G ). The ENT concentrations were at steady - state in the AB core during the following time intervals: c.1760 - c.1860 (~0.01 x 10 6 CE/g - dry wt), and c.1910 c.2003 (~1.0 x 10 6 CE/g - dry wt, Figure 5.3H ). ENT concentrations in CR increased towards present day, and ranged from ~0.03 x 10 5 to 9.9 x 10 5 CE/g - dry wt 114 ( Figure 5.4H) . The EC and ENT concentrations in the AB and CR cores are further detailed in Tables 5.1 and 5.2, respectively . 5.3.5 Linear regression analyses Multiple linear regression analyses evaluated ENT concentrations (deposited during c.1932 - 2012 in both cores), and EC concentrations (deposited during c.1932 - 2012 in the AB core and during c.1951 - 2012 in the CR core), and their associations to the following independent variab les: year, site, EC or ENT concentration in the core sections ( opposite of the dependent variable ), river discharge, air temperature, sedimentary nutrient concentrations ( % t C, % t N, and t P), and estimated population in the watersheds. The R 2 values (and s ample size) for the ENT and EC datasets were 0.86 (n = 36), and 0.72 (n = 33), respectively (Table 5.1 ) . Also, the coefficients of the linear regression equation and their standard errors are listed in Table 5.2 . The ENT concentration s in both cores were significantly correlated to the discharge in the Clinton River and the St. Clair River (p = 0.046), and the estimated population in the Clinton River and Anchor Bay watershed s (p = 0.003, Table 5. 1 ) . The EC concentration s in both c ores were significantly associated with t N and t C concentrations (p = 0.038 , and 0.029 , respectively ), estimated population in the Clinton River and Anchor Bay watershed s (p = 0.023), and air temperature (p = 0.018, Table 5.1 ). ENT and EC concentrations were not significantly cor related to each other ( Table 5.1 ) , nor were the FIB concentrations significantly different between the sites ( Table 5.1 ) . 115 Table 5 . 1 : Linear regressions evaluated the c orrelation strengths (and p - values) of climatic and anthropogenic variables a to 1) enterococci concentrations (ENT) b ,c , C ENT ; and 2) E.coli concentrations (EC) b, d . Correlation strength of the independent variables (p - value) Data set (sample size) R 2 P Y H N A T D S C C ENT (n = 37) 0.860 0.570 (p = 0.003) 0.779 (p = 0.483) 0.471 (p = 0.306) 0.421 (p = 0.145) 0.483 (p = 0.492) 0.358 (p = 0.804) 0.168 (p = 0.046) - 0.161 (p = 0.175) 0.791 e (p = 0.079) C EC (n = 33) 0.724 0.478 (p = 0.023) 0.613 (p = 0.498) 0.430 (p = 0.289) 0.269 (p = 0.038) 0.408 (p = 0.029) 0.214 (p = 0.018) 0.068 g (p = 0.058) - 0.046 (p = 0.117) 0.715 f (p = 0.230) a Escherichia coli concentrations (when C ENT was the dependent variable), C EC or enterococci concentrations (when C EC was the dependent variable), C ENT ; calendar year, Y; nutrient ( total P, % N, and % C) concentrations in the cores, H, N, and A, respectively; population in the Anchor Bay and Clinton River watershed s, P; St. Clair River an d Clinton River discharge , D; air temperature, T; and site, S. b Enterococci (C ENT ), and E.coli (C EC ) concentrations were normalized to cell equivalents per g dry wt. c Linear regression analysis evaluated core sections deposited during c. 1932 - 2012 in both cores . d Linear regression analysis evaluated core sections deposited during c. 1932 - 2012 and during 1951 - 2012 in the AB and CR core , respectively , C EC . e Measured t he coefficient and significance of the independent variable, C EC . f Measured t he coefficient and significance of the independent variable, C ENT . g There was a ~ 6 year lag in the discharge rates in the E.coli dataset. 116 Table 5 . 2 : T he coefficients and their standard errors of the climatic and anthropogenic variables in l inear regression equations a that evaluated the ir associations to: 1) enterococci concentrations (ENT) b,c , C ENT ; and 2) E.coli concentrations (EC) b,d , C EC . Coefficient values (b n , where n is the coefficient) of the independent variables ( standard error of the coefficient s ) Data set a (sample size) P Y H N A T D S C b 0 C ENT (n = 37) 0.000 (0.000) 0.008 (0.011) - 0.003 (0.003) 0.822 (0.547) - 0.016 (0.023) 0.077 (0.306) 0.001 (0.001) 3.977 (2.850) 0.682 (0.373) - 25.252 (20.173) C EC (n = 33) 0.000 (0.000) - 0.004 (0.006) 0.002 (0.002) - 0.589 (0.267) 0.024 (0.010) - 0.349 (0.137) 0.001 e (0.000) 2.309 (1.417) 0.116 (0.088) 19.466 (10.509) a The equations were: ; . Their variables: Escherichia coli concentrations (when C ENT was the dependent variable), C EC or enterococci concen trations (when C EC was the dependent variable), C ENT ; calendar year, Y; nutrient (P,N, and, C) concentrations in the cores, H, N, and A, respectively; watershed population in Anchor Bay and Clinton River, P; St. Clair River and Clinton River disch arge , D; air temperature, T; and site, S. b Enterococci (C ENT ), and E.coli (C EC ) concentrations were normalized to cell equivalents per g dry wt. c in core sections deposited during 1932 - 2012 in both cores d a re deposited in core sections during 1932 - 2012 in the AB core and during 1951 - 2012 in the CR core e There was a ~ 6 year lag in the discharge rates in the E.coli dataset. 117 5.4 Discussion Previous investigations have measured chemical contaminants , and nutrients in sediment cores in order to reconstruct large scale assessments of historic al land management practices (Furl & Meredith, 2011; Kaushal & Binfor d, 1999) . Our study is one of the first studies to evaluate the associations of anthropogenic attributes , and climate variables to FIB concentrations in sediment cores o ver ~100 to 250 years, contrasting two locations in the Lake St. Clair watershed nea r the mouth of the Clinton River (CR), and Anchor Bay (AB) in Northwest ern Lake St. Clair. The Lake St. Clair watershed is important to the Great Lakes basin as a key connector between the Upper and Lower Great Lakes, as well as home to > 4 million people, and is a drinking water source for metropolitan Detroit and southwestern Ontario (Baustian et al., 2014) . Summaries of the anthropogenic, climatic, nutrient loading a nd FIB data in both locations is illustrated in Figures 5.5A - B . Analysis of the AB core determined that the bottom most sections (c.1760 c.1800) were deposited when the watershed was a trading outpost and a Native American settlement. During this time, n utrient and FIB measurements were at steady state ( Figures 5.3D - H and 5.4A ). Starting in 1766, small scale logging of the area began and the timber was rafted on the St. Clair River to Detroit (Crampton, 1921) . Wide scale logging, timber processing, and transport began after the Swampland Act of 1850 (Unknown, 1883) , which allowed settlers to obtain wetlands at no cost if they were drained and developed (Fishbeck, Thompson, Carr & Huber, 2007) . Subsequently, the Anchor Bay watershed was deforested by the early 20 th Century (Crampton, 1921) . The effects of deforestation, destruction of wetlands, and the booming logging industry were seen in the increasing nutrient and enterococci concentrations until around the start of the 20 th Century ( Figures 5 .3D - F and 5.5A ). Also, the 118 C:N ratio before the Swamplands Act of 1850 was ~11:1 (Baustian et al., in prep. ) , i ndicating that the tC and tN were mainly from aquatic sources. After the policy change, the C:N ratio increased to 21.85 (Baustian et al., in prep. ) , which indicated that the tC and tN were from terrestrial sources (Meyers & Ishiwatari, 1993) . Similarly, a previous study also observed increases in organic matter and C:N values measured in sediment cores from the Great Lakes that coincided with deforestation (Meyers, 2007) . The formation of the second steady state concentration of ENT at 1 - log larger than the pre - deforestation levels, and the return to the pre - deforestation levels of % tC, and % tN in the AB core began in c.1896 ( Figures 5.3E - G and 5.5A) could have been the result of forest re - growth or widespread agri cultural practices that buffered nutrient loading and wastewater/fecal pollution inputs into Anchor Bay. Similarly, the nutrients concentrations in a sediment core from Lake Ontario decreased following massive deforestation events (Hodell & Schelske, 1998) , and could be the result of re - forestation acti vities that buffered runoff. Increases in EC and tP concentrations in the AB core starting c.1890 and c.1849 ( Figures 5.3 D , 5.3 G , and 5.5A ), respectively, could be related to the increase in population of the watershed, which may be related to wastewater p roduction. The sedimentary % tN and % tC returned to a cyclical steady state during c.1932 c.1992, which occurred during the wide scale (Fishbeck, Thompson, Carr & Huber, 2007) ( Figures 5.3 E - F , and 5.5A ). Such development did not allow for natural buffers to retain run off from nonpoint inputs of wastewater and fertilizers, which may have accounted for the cyclic nature of the nutrient loading. In 1972, The National Pollutant Discharge Elimination System (NDPES) began regulating P loading in the Great Lakes (Harrington - Hughes, 1978) , and during this period, its concentration was in a cyclical steady state ( Figure s 5.3D and 5.5A ), which can be 119 partially attributed to erosion in th e riverbed caused by dredging and increased air temperatures that lead to higher evaporation rates (Associated Press, 20 07) ( Figure s 5.3B and 5.5A ) . The ENT and EC concentrations began t o decrease in 2003, which could be the result of programs initiated in 2002 in St. Clair and Macomb counties to identify and fix failing septic systems (St. Clair County Health Department, 2009) ( Figures 5.3 G - H ; and 5.5A ) . 120 Figure 5 . 5 A - B: Illustrations of the patterns of discharge of the St. Clair River (A) and the Clinton River (B) , anthropogenic variables b , sedimentary nutrient concentrations c , an d fecal indicator concentrations d from the Anchor Bay e ( A) and Clinton River e ( B) watersheds. a The NDPES is the National discharge pollutant elimination system. b Anthropogenic variables included human population in watershed, and policy changes . c Sedimentary nutrient measurements included: total phosphorus, % total C, and % total N . d The measurements included: E scherich ia coli a nd enterococci cell equivalents. 121 e The Anchor Bay and Clinton River sediment cores spanned the years: c.1760 - 2012 and c.1895 - 2012, respectively. The large sedimentation rate in the CR core could be due to higher sediment supply (per unit volume of water discharge) from the Clinton River compared to St. Clair River. Overall, the river discharge, and FIB and nutrient concentrations in the CR core have increased since c.1895 (and since c.1934 for river discharge only, Figures 5.4B, and 5.4D - H ). The increase of nutrient and FIB concentrations in the Clinton River were previously attributed to failing septic systems, storm water, and runoff (Environmental Conuslting and Technology, 2007) . The sedimentary tP profile suggest that its loading began to increase in c.1966 (F igures 5.4D and 5.5B) . However, the rate of tC and tN loading decreased in c.1962 ( Figures 5.4E - F) , and ENT concentrations stabilized during c.1969 c.1987 ( Figure 5.4H, and 5.5B) , perhaps due to the elimination of raw sewage inputs and regulation of poin t source pollution instituted by National Pollutant Discharge Elimination System in 1972 (Michigan D epartment of Environmental Quality, 1988) . The discharge at the mouth of the Clinton River decreased in 1951 ( Figures 5.4B, and 5.5B) as a result of the construction of a spillway to prevent flooding (Michigan Department of Environmental Quality, 1988) ( Figure s 5.3D - H , and 5.5B ). The Clinton River was first listed as an EPA Area of Concern in 1988, and a remedial action plan was drafted in order to remediate the watershed. Revitalization efforts of the Clinton River may have decreased the C:N ratio c.1987 (Baustian et al., in prep. ) . However, delisting of the Clinton River has yet to be accomplished, and is witnessed in the increasing FIB and n utrient loading in the core since c.1987 ( Figures 5.4D - H, and 5.5B ) . Our study is one of the first to measure genetic markers from ENT and EC in sediment s that were deposited in the Lake St. Clair watershed > 200 yrs ago . Previously, culturable ENT and 122 EC were enumerated from 60 cm beneath the surface sediment in Lake Geneva, Switzerland (Thevenon et al., 2012) . Their and our results further suggest that sediments are long term reservoirs for FIB . T he sensitivity of EC - uidA resulted in incomplete profile s of the EC concentrations in the cores. However, w hen detected, the EC concentrations were larger than ENT ( Figures 5.3G, and 5.4G ) , which was echoed in measurements of culturable ENT and EC from sediment core s obtained from Lake Geneva (Thevenon et al., 2012) . Our study evaluated the statistical associations of anthropogenic ( human population in watershed, and nutrient concentration) , and cl imate ( river discharge , and air temperature) variables to FIB concentrations in both cores. The ENT and EC concentrations in both cores were significantly correlated to population (p = 0.003, and 0.023, respectively) . A previous study also reported that urbanized areas of a marine estuary experienced larger fecal coliform concentrations (Kelsey, Porter, Scott, Neet, & White, 2004) . EC concentrations were significantly associated to % tC and % tN (p = 0.029, and 0.038, respectively ) in our study. Similarly, culturable EC concentrations in sediments increased with organic matter and N concentrations (Haller, Amedegnato, et al., 2009) . The air temperature was negatively associated with EC concentrations in the cores (p = 0.018). EC 23S rDNA demonstrated longer persistence in bee f manure amended soil stored at 10 ° C than at 25 ° C (Rogers et al., 2011) . ENT concentrations were significantly correlated to river discharge in our study (p = 0.046 ) , suggesting that ENT attach ment to suspended particles facilitate s their movement to the benthos (Jeng, England, & Bradford, 2005) . Therefore, increased discharge partially made up of wastewater effluent, discharges from combined sewer overflow events, and excessive non - point runoff from flooding co uld increase ENT concentrations in surface sediments. T he lack of 123 correlation between ENT and EC concentrations could be due to the type of pollution inputs and their distinct decay rates . Our data provided evidence that paleolimnological investigations c an determine associations between climate, anthropogenic attributes, regulation, and water quality over long time scales. Sediment core studies can be tools to analyze historical fecal pollution as markers of anthropogenic transformation of the area, water shed management and shape future practices. 124 APPENDIX 125 Table S 5 . 1 : Concentrations of Escherichia coli , enterococci, total phosphorus, % total nitrogen and % total carbon measured in the Anchor Bay sediment core . Depth (cm) Year deposited a E.coli conc. (CE/g - dry wt) b Enterococci conc. (CE/g - dry wt) b Total Phosphorus conc. (ug/g - dry wt) % total Nitrogen % total Carbon 2 2011 x 10 6 c 4.01 x 10 4 - 3.16 x 10 - 1 8.02 x 10 0 4 2009 x 10 6 c 1.43 x 10 5 2.80 x 10 2 2.69 x 10 - 1 6.71 x 10 0 6 2006 5.38 x 10 6 2.14 x 10 5 3.01 x 10 2 2.26 x 10 - 1 5.67 x 10 0 8 2003 8.46 x 10 6 8.35 x 10 4 2.82 x 10 2 1.21 x 10 - 1 5.23 x 10 0 10 1999 3.20 x 10 6 1.33 x 10 5 2.16 x 10 2 6.18 x 10 - 2 1.40 x 10 0 12 1992 3.26 x 10 6 1.65 x 10 5 2.40 x 10 2 5.09 x 10 - 2 7.49 x 10 - 1 14 1986 4.23 x 10 6 1.53 x 10 5 2.85 x 10 2 5.96 x 10 - 2 9.83 x 10 - 1 16 1978 4.66 x 10 6 1.49 x 10 5 2.01 x 10 2 3.69 x 10 - 2 9.96 x 10 - 1 18 1971 3.17 x 10 6 1.34 x 10 5 1.73 x 10 2 2.99 x 10 - 2 7.34 x 10 - 1 20 1963 6.76 x 10 6 1.86 x 10 5 2.15 x 10 2 5.50 x 10 - 2 1.53 x 10 0 22 1955 x 10 6 c 1.06 x 10 4 2.38 x 10 2 4.23 x 10 - 2 8.68 x 10 - 1 24 1947 2.51 x 10 6 1.65 x 10 5 2.33 x 10 2 3.59 x 10 - 2 6.64 x 10 - 1 26 1940 3.14 x 10 6 1.40 x 10 5 1.80 x 10 2 5.39 x 10 - 2 1.03 x 10 0 28 1932 3.12 x 10 6 1.22 x 10 4 2.05 x 10 2 4.79 x 10 - 2 7.82 x 10 - 1 30 1926 1.86 x 10 6 1.21 x 10 5 2.12 x 10 2 8.48 x 10 - 2 1.68 x 10 0 32 1919 2.36 x 10 6 9.56 x 10 3 2.16 x 10 2 1.34 x 10 - 1 2.56 x 10 0 34 1913 1.95 x 10 6 1.16 x 10 5 1.59 x 10 2 1.14 x 10 - 1 2.41 x 10 0 36 1907 2.04 x 10 6 1.22 x 10 5 1.66 x 10 2 9.50 x 10 - 2 2.15 x 10 0 38 1902 2.59 x 10 6 1.07 x 10 5 1.43 x 10 2 1.71 x 10 - 1 3.36 x 10 0 40 1898 x 10 6 c 7.37 x 10 4 1.75 x 10 2 4.08 x 10 - 1 7.44 x 10 0 42 1896 3.21 x 10 6 4.03 x 10 4 1.93 x 10 2 6.57 x 10 - 1 1.44 x 10 - 1 44 1894 2.68 x 10 6 c 4.35 x 10 4 1.81 x 10 2 5.53 x 10 - 1 1.04 x 10 - 1 46 1890 1.75 x 10 6 c 2.44 x 10 4 2.02 x 10 2 4.13 x 10 - 1 8.15 x 10 0 48 1887 2.13 x 10 6 2.82 x 10 4 2.17 x 10 2 4.53 x 10 - 1 7.60 x 10 0 50 1883 1.61 x 10 6 c 2.04 x 10 4 2.44 x 10 2 3.43 x 10 - 1 5.85 x 10 0 52 1878 1.47 x 10 6 c 1.46 x 10 4 2.06 x 10 2 2.31 x 10 - 1 3.94 x 10 0 54 1873 1.56 x 10 6 c 2.29 x 10 4 2.19 x 10 2 2.95 x 10 - 1 4.19 x 10 0 56 1868 1.39 x 10 6 c 3.30 x 10 4 1.68 x 10 2 8.29 x 10 - 2 1.27 x 10 0 126 Table S 5.1 ( ) . 58 1862 6 c 2.66 x 10 4 1.28 x 10 2 2.09 x 10 - 1 2.48 x 10 0 60 1856 1.28 x 10 6 c 5.45 x 10 3 1.07 x 10 2 8.37 x 10 2 9.62 x 10 - 1 62 1849 1.19 x 10 6 c 1.28 x 10 4 1.40 x 10 2 8.36 x 10 - 2 9.08 x 10 - 1 64 1842 1.18 x 10 6 c 1.09 x 10 4 1.51 x 10 2 7.47 x 10 - 2 8.01 x 10 - 1 66 1835 1.16 x 10 6 c 1.14 x 10 4 1.91 x 10 2 7.63 x 10 - 2 7.49 x 10 - 1 68 1827 1.14 x 10 6 c 3.05 x 10 4 1.93 x 10 2 8.42 x 10 - 2 8.79 x 10 - 1 70 1820 1.13 x 10 6 c 9.51 x 10 3 1.24 x 10 2 6.31 x 10 - 2 6.70 x 10 - 1 72 1812 1.12 x 10 6 c 1.03 x 10 4 1.96 x 10 2 6.47 x 10 - 2 7.01 x 10 - 1 74 1805 x 10 6 c 4.12 x 10 4 1.98 x 10 2 6.86 x 10 - 2 6.97 x 10 - 1 76 1797 x 10 6 c 1.15 x 10 4 1.77 x 10 2 6.43 x 10 - 2 7.33 x 10 - 1 78 1789 x 10 6 c 1.07 x 10 4 1.81 x 10 2 8.25 x 10 - 2 8.65 x 10 - 1 80 1781 1.81 x 10 6 x 10 3 c 2.21 x 10 2 6.62 x 10 - 2 7.07 x 10 - 1 82 1774 2.12 x 10 6 7.39 x 10 3 1.83 x 10 2 5.57 x 10 - 2 6.21 x 10 - 1 84 1765 x 10 6 c 4.11 x 10 3 2.06 x 10 2 9.22 x 10 - 2 1.01 x 10 0 86 17 60 x 10 6 c 1.74 x 10 4 2.80 x 10 2 8.55 0 x 10 - 2 1.50 *10 0 a Year of deposition was estimated from the Cs - 137 radioactivity in each sediment section. b The enterococci and E.coli concentrations were normalized to cell equivalents (CE) per gram - dry wt. c The measurement was below the detection limit, and reported value is the detection limit specific to the sediment section. 127 Table S 5 . 2 : Concentrations of Escherichia coli , enterococci, total phosphorus, % total nitrogen and % total carbon measured from the Clinton River sediment core. Depth (cm) Year deposited a E.coli conc. (CE/g - dry wt) b Enterococci conc. (CE/g - dry wt) b Total Phosphorus conc. (ug/g - dry wt) % total Nitrogen % total Carbon 2 2010 16.9 x 10 6 5.93 x 10 5 4.37 x 10 2 1.64 x 10 - 1 3.48 x 10 0 4 2007 8.87 x 10 6 9.94 x 10 5 3.45 x 10 2 2.77 x 10 - 1 5.93 x 10 0 6 2005 5.56 x 10 6 5.20 x 10 5 3.63 x 10 2 2.07 x 10 - 1 4.34 x 10 0 8 2002 3.72 x 10 6 4.36 x 10 5 3.70 x 10 2 1.71 x 10 - 1 3.61 x 10 0 10 1999 4.87 x 10 6 4.08 x 10 5 3.08 x 10 2 2.07 x 10 - 1 4.77 x 10 0 12 1996 3.23 x 10 6 3.75 x 10 5 3.67 x 10 2 2.47 x 10 - 1 5.71 x 10 0 14 1993 4.49 x 10 6 3.97 x 10 5 3.20 x 10 2 1.55 x 10 - 1 3.53 x 10 0 16 1990 3.50 x 10 6 3.10 x 10 5 3.47 x 10 2 2.25 x 10 - 1 5.34 x 10 0 18 1987 4.36 x 10 6 7.39 x 10 4 3.71 x 10 2 1.78 x 10 - 1 4.33 x 10 0 20 1984 3.30 x 10 6 5.45 x 10 4 2.09 x 10 2 1.68 x 10 - 1 3.45 x 10 0 22 1981 x 10 6 c 5.48 x 10 4 3.41 x 10 2 1.60 x 10 - 1 2.45 x 10 0 24 1978 3.01 x 10 6 5.02 x 10 4 4.10 x 10 2 2.99 x 10 - 1 4.44 x 10 0 26 1975 x 10 6 c 1.50 x 10 4 2.99 x 10 2 1.76 x 10 - 1 3.33 x 10 0 28 1972 3.48 x 10 6 5.40 x 10 4 3.02 x 10 2 1.59 x 10 - 1 2.81 x 10 0 30 1969 2.33 x 10 6 4.81 x 10 4 2.97 x 10 2 1.41 x 10 - 1 2.37 x 10 0 32 1966 2.50 x 10 6 4.74 x 10 4 2.16 x 10 2 1.58 x 10 - 1 3.15 x 10 0 34 1963 3.29 x 10 6 1.61 x 10 4 2.75 x 10 2 1.60 x 10 - 1 3.24 x 10 0 36 1960 2.78 x 10 6 1.85 x 10 4 2.22 x 10 2 1.09 x 10 - 1 1.91 x 10 0 38 1956 1.92 x 10 6 1.37 x 10 4 1.69 x 10 2 1.39 x 10 - 1 2.46 x 10 0 40 1951 1.31 x 10 6 6.63 x 10 3 2.16 x 10 2 5.93 x 10 - 2 1.56 x 10 0 42 1945 x 10 6 c 3.41 x 10 3 2.54 x 10 2 5.04 x 10 - 2 9.93 x 10 - 1 44 1940 x 10 6 c 3.48 x 10 3 2.17 x 10 2 5.26 x 10 - 2 1.13 x 10 0 46 1934 1.42 x 10 6 4.29 x 10 3 2.50 x 10 2 8.22 x 10 - 2 1.43 x 10 0 48 1929 1.97 x 10 6 1.28 x 10 4 2.39 x 10 2 6.32 x 10 - 2 9.23 x 10 - 1 50 1924 1.73 x 10 6 7.13 x 10 3 1.94 x 10 2 9.02 x 10 - 2 1.61 x 10 0 52 1917 x 10 6 c x 10 3 c 1.30 x 10 2 5.18 x 10 - 2 8.24 x 10 - 1 128 Table S 5.2 ( ) . 54 1911 6 c 5.69 x 10 3 1.15 x 10 2 1.36 x 10 - 2 3.35 x 10 - 1 56 1902 1.01 x 10 6 3 c 1.35 x 10 2 1.11 x 10 - 2 1.72 x 10 - 1 58 1895 6 c 3 c 4.37 x 10 2 1.96 x 10 - 2 3.19*10 - 1 a Year of deposition was estimated from the Cs - 137 radioactivity in each sediment section. b The enterococci and E.coli concentrations were normalized to cell equivalents (CE) per gram - dry wt. c The measurement was below the detection limit, and reported value is the detection limit specific to the sediment section . 129 CHAPTER 6. CONCLUSIONS 130 6.1 Goals of the research and summary of the results The goal of this dissertation was to investigate the factors that we re associated with c oncentrations of persistent fecal indicators measured from water quality sample s in various storage schemes and from sediment cores from the Lake St. Clair watershed. The first research project investigated the short term and long term persistence of qPCR measurements of three fecal indicators in water samples stored in various conditions. The second study compared the concentrations of total eluted DNA and qPCR measurements of two fecal indicators extracted with modifications to three DNA extraction methods. The last study investigated the associations of anthropogenic attributes and climate variables to qPCR measurements of persistent fecal i ndicators measured with the optimal DNA extraction method identified in the second experiment. In the first research project, two bench scale studies investigated how indicator species, storage condition, time, and temperature (short term study only) were associated with the short term and long term persistence of naturally occurring Escherichia coli , Bacteroides th etataiotaomicron , and enterococci in water samples stored up to 28 and 366 days, respectively. Raw sewage (10% vol/vol) was spiked into autocla ved river water and stored in two conditions, containers in liquid form (liquid suspension), and membrane filters that were membrane filtered in 100 ml increments at initial sampling (solid matrix). The short term and long term studies included 4 C storage temperatures maintained with ice packs and refrigeration, respectively, while the short term study also included 27 and 37 C storage temperatures. The persistence of the indicators was evaluated from qPCR measurements of Escherichia coli uidA, Bacteroide s thetataiotaomicron alpha - mannanase , and enterococci 23S rDNA that were transformed into 131 concentrations of cellular equivalents. The persistence patterns of the indicators were mathematically evaluated with sixteen linear and non - linear best fit models. O f those models, five linear and nonlinear models were selected. The most popular model was biphasic exponential decay, representing 75% of the treatments in the long term and short term studies. From the best fit models, t he general order of least to most persistent indicator s were : B.thetataiotaomicron < E.coli < enterococci . I ndicator persistence significantly increased when attached to a solid matrix compared to storage in liquid suspension. Increased storage time was also associated with the shortened p ersistence of indicators. Persistence was significantly affected by indicator species . In the short term study, t he time needed for 90% reduction of the initial concentrations of the indicators, T 90 , was estimated from their best fit models, and ranged from 1.0 day ( B.thetataiotaomicron i n liquid suspension at > 28 days (enterococci and E.coli attached to a solid matrix at all temperatures). In the long term study, the best fit models predicted T 90 values that ranged from 35.8 days for E.coli in liquid suspension to 164.0 days for enterococci attached to a solid matrix. The T 90 values of the indicators of the long term study were longer than the short term study, and may be a result of storage cooling strategies i.e. storage on ice vs. refrigerat ion, and/or storage duration. Comparisons of concentrations of total eluted DNA, and cellular equivalents of qPCR measurements of E.coli uidA and enterococci 23S rDNA in surface sediment from Lake St. Clair and the Clinton River spiked with E.coli and Enterococcus faec alis were evaluated with six modifications of three DNA extraction methods, MoBio UltraClean®, MoBio PowerSoil®, and EPA - DNA2 (based on USEPA Method 1611) protocol; a DNA sorption blocker, G2 ; sonication; 1:10 elution step; double recommended 132 sample mass; and addition of 0.01% Tween 80. Overall, there were few significant differences in total eluted DNA, and cellular equivalent concentrations of E.coli and enterococci extracted using the modif ied DNA extraction methods from the two locations. Modifications to the MoBio methods decreased concentrations of total eluted DNA and the indicators (MoBio UltraClean only), while modifications to EPA - DNA2 generally resulted in larger concentrations of to tal eluted DNA and the indicators. The optimal method was determined to be EPA - DNA2 modified with G2, and was used in the investigation below. Sediment cores were collected from the mouth of the Clinton River and from Anchor Bay, northwestern Lake St. Cla ir, in order to investigate how anthropogenic attributes (human population in the watershed, and vertical concentrations of total P, total C, and total N) and climate variables (air temperature and river discharge) were associated to vertical concentration s of cellular equivalents of E.coli uidA and enterococci 23S rDNA. Radio - isotope activity of Cs - 137 and Pb - 210 determined that the Clinton River (CR) and Anchor Bay (AB) cores represented the years, 1757 - 2012 and 1895 - 2012, respectively. The re were steady state concentrations of enterococci in the AB core at ~0.1 x 10 5 and ~2 x 10 5 cel l equivalents (CE) per g - dry wt duri ng 1757 - 1878, and 1902 - 2010, respectively . The temporal change of enterococci concentrations in the AB core trended with increas ed sedimentary nutrient loading. Enterococci concentrations in the CR core increased steadily with time, and ranged ~ 0.03 x 10 5 - 9.94 x 10 5 CE/g - dry wt. The E.coli concentrations in the CR and AB cores increased with time, and ranged 0.14 x 10 7 1.69 x 1 0 7 CE /g - dry wt, and 0.18 x 10 7 0.85 x 10 7 CE /g - dry wt, respectively. Statistical analyses of the core sections of the AB and CR during 1932 - 2012 (and 1951 - 2012 for E.coli concentrations measured in the CR core) determined that concentrations of enterococci were 133 significantly associated with human population in the watershed, and river discharge, while E.coli concentrations were significantly associated with human populatio n in the watershed, air temperature, and sedimentary concentrations of total nitrogen and total carbon. The historical record of the St. Clair watershed illustrated that changes in sedimentary nutrient loading, and fecal indicator concentrations occurred a fter watershed policy enactments and watershed management practices like the Swampland Act of 1860, the Clean Water Act of 1972, naming the Clinton River an EPA Area of Concern, and development of the shore lands surrounding the mouth of the Clinton River and Lake St. Clair. 6.2 Correlations of the results of the long term persistence study to persistent concentrations of fecal indicators in sediment cores One of the objectives of the bench scale persistence studies was to determine if the persistence of f ecal indicators was significantly different in the two storage conditions, liquid suspension and attached to a solid matrix. The solid matrix not only represented a novel storage scheme for water quality samples, it was also a representative of the possibl e attachment of fecal indicators to sediment particles in the environment. The results of the long term study determined that attachment to a surface such a membrane filter was associated with the extended persistence of three fecal indicators. The results supported previous research that determined that attachment to sediment particles could be one of the mechanisms that allows for extended persistence of fecal indicators in sediments. However, one of the limitations of the long term study is the lack of t hree dimensional interaction between the indicators and the abiotic and biotic factors in the environment, which is possible in sediments. 134 Another objective of the long term study was to measure the persistence of fecal indicators over a large time scale. The results in this dissertation indicated that the persistence of fecal indicators attached to a solid matrix supported the existence of highly persistent populations of fecal indicators in sediments at a low temperature (4 C). The temperature in vertical sections of the sediments of Lake St. Clair is unknown. However, the storage temperature was chosen because 4 C was considered to be the lowest temperature of the benthic portion of Lake St. Clair during the winter months. 6.3 Implications of the result s of this dissertation and recommendations to water quality monitoring The Clean Water Act ensures that all navigable waters are safe for swimming and fishing. The biological safety of recreational waters is monitored with culturable concentrations of gene ral fecal indicators, E.coli and enterococci. Such monitoring has reduced the risk of adverse health outcomes of swimmers. However, there is still much progress to be made to ensure that standardized water quality monitoring methods can accurately and prec isely estimate the health risk associated with swimming in recreational waters. Research has recently indicated that qPCR measurements of enterococci 23S rDNA were better associated with the risk of waterborne illness incidence than its culturable concentr ations. The results in of this dissertation demonstrated that enterococci 23S rDNA and E.coli uidA can persist for extended periods of time in various environmental conditions. These persistent populations may not accurately predict the presence or concent ration of waterborne pathogens. Also, previous research has suggested that qPCR measurements of human associated fecal indicators may better reflect the risk of waterborne illness incidence. The observations in this dissertation indicated that concentratio ns 135 of a human associated indicator, B.thetataiotaomicron , reflected recent pollution inputs. The use of both general and human associated indicators may give better representation of recent and persistent fecal pollution. Therefore, i t is recommended that qPCR measurements of human associated pollution indicator, B.thetataiotaomicron , and qPCR measurements of general pollution indicators, E.coli and enterococci, should be used in conjunction to better evaluate the risk of waterborne illness incidence in rec reational waters . The ability of water quality monitoring to accurately measure fecal indicator concentrations is dependent on the validity of its methods. The standardized transportation scheme for water quality monitoring was optimized for culturable me asurements of fecal indicators. Briefly, water samples are stored in closed containers in liquid suspension. Samples can spend up to 6 hrs in transport at 4 C (on ice) before analysis in the laboratory. The transportation time limit constrains the number o f samples and sites that can be sampled. The results from this dissertation illustrated that qPCR measurements of three naturally occurring fecal indicators can spend at least 9 days on ice and at least 35 days in refrigeration in liquid suspension before 90% removal of the initial concentrations. Additionally, ice can be a limiting factor for water sample storage in low resource settings. However, this dissertation estimated that the time needed for 90% removal of initial concentrations of the three indica tors in liquid suspension in storage temperatures, 27 º and 37 º C, does not occur until after at least 1 day in storage. Therefore, it is recommended that water samples in liquid suspension can be stored for up to 1 day in storage temperatures > 4 C, and wat er samples in liquid suspension can be stored at 4 C (on ice) for up to 9 days before 90% removal of concentrations of E.coli , enterococci, and B.thetataiotaomicron measured with qPCR. It is also recommended that water samples can be stored long term if 136 refrigerated at 4 C for up to 35 days before 90% removal of the initial concentrations of E.coli , enterococci, and B.thetataiotaomicron measured with qPCR. The validity of the standard methods to access the biological safety of recreational waters is also dependent on the storage matrix of water samples. The USEPA standardized methods mandate that water samples are stored in liquid suspension in a closed container before enumeration of culturable fecal indicators. Storage in liquid suspension can limit sam pling efficiency because the samples occupy more space and mass. This dissertation compared the persistence of qPCR measurements of naturally occurring E.coli uidA , enterococci 23S rDNA , and B.thetataiotaomicron alpha - mannanase in water samples stored in l iquid suspension and attached to a solid matrix (membrane filter). The results from this dissertation predicted that the indicators persisted up to 2.8x longer when attached to a solid matrix than in liquid suspension when stored at 4 º C for up to 28 days a nd 366 days. At elevated storage temperatures, i.e. 27 º and 37 º C, the persistence of the indicators in water samples was between 3.2x - 10x longer when attached to a solid matrix for up to 28 days. Therefore, it is recommend ed that water samples be filtere d at initial sampling in order to increase the persistence of indicators measured with qPCR, while decreasing the space and mass of water samples during transportation to a laboratory . 6.4 Implications of results to watershed management and recommendati ons for management actions Standardized water quality monitoring methods do not take into account the potential fecal pollution in the sediments of recreational waters. Previous research has observed that fecal 137 indicators and fecal pathogens are highly con centrated in the sediments, which indicate that sediments can be reservoirs of fecal pollution. The results of this dissertation have illustrated that concentrations of fecal indicators, E.coli and enterococci, and sedimentary nutrients like total nitrogen , total carbon, and total phosphorus, can be measured in sediments deposited > 200 years ago. Thus, vertical measurements of nutrients and fecal indicators in sediment cores can be used to measure historical impacts specific to the watershed, including pop ulation, land use, and policy enactments. Such research would allow for better identification of the strengths and deficien cies regarding watershed management and facilitate the determination of sustainable management practices . Therefore, it is recommende d that the historic al water quality of the Great Lakes Region should be surveyed via sediment core students in order to better understand how changes in water chemistry and quality are associated to previous management practices and policies . Such studies will help guide future watershed management practices and policies. 138 BIBLIOGRAPHY 139 BIBLIOGRAPHY American Public Health Association. (1998). 4500 - P Phosphorus. In Standard Methods for the Examination of Water and Wastewater . American Public Health Association, American Waterworks Association, Water Environment Federation. Andersen, J. (1976). An ignition method for determination of total phosphorus in lake sediments. Water Research , 10 (4), 329 331. doi:10.1016/0043 - 1354(76)90175 - 5 Anderson, K. L., Whitlock, J. E., & Harwood, V. J. (2005). Persistence and differential survival of fecal indicator bacteria in subtropical waters and sediments. Applied and Environmental Microbiology , 71 (6 ), 3041 8. doi:10.1128/AEM.71.6.3041 - 3048.2005 Aragao, G. M. F., Corradini, M. G., Normand, M. D., & Peleg, M. (2007). Evaluation of the Weibull and log normal distribution functions as survival models of Escherichia coli under isothermal and non isotherma l conditions. International Journal of Food Microbiology , 119 (3), 243 57. doi:10.1016/j.ijfoodmicro.2007.08.004 Aslan, A., & Rose, J. B. (2013). Evaluation of the host specificity of Bacteroides thetaiotaomicron alpha - 1 - 6, mannanase gene as a sewage marker . Letters in Applied Microbiology , 56 (1), 51 6. doi:10.1111/lam.12013 Associated Press. (2007, August 14). Report: Dredging Water Causes Huge Great Lakes Water Loss. MLive Media Group . Traverse City, MI. Retrieved from http://blog.mlive.com/chronicle/2007/ 08/report_dredging_causes_huge_gr.html Bae, S., & Wuertz, S. (2009). Rapid decay of host - specific fecal Bacteroidales cells in seawater as measured by quantitative PCR with propidium monoazide. Water Research , 43 (19), 4850 9. doi:10.1016/j.watres.2009.06.0 53 C. S. (2014). The impact of bioaugmentation on dechlorination kinetics and on microbial dechlorinating communities in subsurface clay till. Environmental Po 1987) , 186 , 149 57. doi:10.1016/j.envpol.2013.11.013 Ballesté, E., & Blanch, A. R. (2010). Persistence of Bacteroides species populations in a river as measured by molecular and culture techniques. Applied and Environmental Microb iology , 76 (22), 7608 16. doi:10.1128/AEM.00883 - 10 14 0 B. (n.d.). Paleo - environmental examination of human - driven ecosystem change in Lake St. Clair region of Laure ntian Great Lakes basin. In Prep. J. B. (2014). A one hundred year review of the socioeconomic and ecological systems of Lake St. Clair, North America. Journa l of Great Lakes Research , 40 (1), 15 26. doi:10.1016/j.jglr.2013.11.006 (2004). Surveillance for waterborne - disease outbreaks associated with drinking water United States, 2001 2002. Morbidity and Mortality Weekly Report. , 53 (SS - 8 ), 23 45. Boehm, A. B., Griffith, J. F., Mcgee, C., Edge, T. A., Solo - Weisberg, S. B. (2009). Faecal indicator bacteria enumeration in beach sand: a comparison study of extraction methods in medium to coarse sands. Journa l of Applied Microbiology , 107 (5), 1740 50. doi:10.1111/j.1365 - 2672.2009.04440.x - term persistence and leaching of Escherichia coli in temperate maritime soils. Applied a nd Environmental Microbiology , 76 (5), 1449 55. doi:10.1128/AEM.02335 - 09 Brenner, D. J., & Farmer III, J. J. (2005). Order XIII: Enterobacteriales. In N. D. Krieg, J. T. Staley, & G. M. Garrity (Eds.), . New York, N ew York: Springer. Bürgmann, H., Pesaro, M., Widmer, F., & Zeyer, J. (2001). A strategy for optimizing quality and quantity of DNA extracted from soil. Journal of Microbiological Methods , 45 (1), 7 20. doi:10.1016/S0167 - 7012(01)00213 - 5 Byappanahalli, M. N., Roll, B. M., & Fujioka, R. S. (2012). Evidence for occurrence, persistence, Microbes and Environments / JSME , 27 (2), 164 70. Retrieved from http://www.ncbi.nlm.nih.gov /pubmed/22791049 Cabelli, V. J., Dufour, A. P., McCabe, L. J., & Levin, M. A. (1982). Swimming - associated gastroenteritis and water quality. American Journal of Epidemiology , 115 (4), 606 16. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/7072706 Carret, G., Flandrois, J. P., & Lobry, J. R. (1991). Biphasic kinetics of bacterial killing by quinolones. The Journal of Antimicrobial Chemotherapy , 27 (3), 319 27. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2037537 Carson, C. A., Christiansen, J. M., Yamp ara - Iquise, H., Benson, V. W., Baffaut, C., Davis, J. V, 141 feces. Applied and Environmental Microbiology , 71 (8), 4945 9. doi:10.1128/AEM.71.8.4945 - 4949.2005 Cavalli - Sforza, L. T., Menozzi, P., & Strata, A. (1983). A model and program for study of a tolerance curve: application to lactose absorption tests. International Journal of Bio - Medical Computing , 14 (1), 31 41. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6687465 C enters for Disease Control. (2011). Escherichia coli O157:H7 and other Shiga toxin - producing Escherichia coli (STEC). Retrieved from http://www.cdc.gov/nczved/divisions/dfbmd/diseases/ecoli_o157h7/ Chick, H. (1908). An Investigation of the Laws of Disinfec tion. Journal of Hygiene , 8 (1), 92 158. S. B. (2007). Water quality indicators and the risk of illness at beaches with nonpoint sources of fecal contamin ation. Epidemiology (Cambridge, Mass.) , 18 (1), 27 35. doi:10.1097/01.ede.0000249425.32990.b9 Converse, R. R., Blackwood, A. D., Kirs, M., Griffith, J. F., & Noble, R. T. (2009). Rapid QPCR - based assay for fecal Bacteroides spp. as a tool for assessing fecal contamination in recreational waters. Water Research , 43 (19), 4828 37. doi:10.1016/j.watres.2009.06.036 Coroller, L., Leguerinel, I., Mettler, E., Savy, N ., & Mafart, P. (2006). General model, based on two mixed weibull distributions of bacterial resistance, for describing various shapes of inactivation curves. Applied and Environmental Microbiology , 72 (10), 6493 502. doi:10.1128/AEM.00876 - 06 Corradini, M. G., Normand, M. D., & Peleg, M. (2007). Modeling non - isothermal heat inactivation of microorganisms having biphasic isothermal survival curves. International Journal of Food Microbiology , 116 , 391 399. doi:10.1016/j.ijfoodmicro.2007.02.004 Craig, D. L., Fallowfield, H. J., & Cromar, N. J. (2002). Enumeration of faecal coliforms from recreational coastal sites: evaluation of techniques for the separation of bacteria from sediments. Journal of Applied Microbiology , 93 (4), 557 565. doi:10.1046/ j.1365 - 2672.2002.01730.x Craig, D. L., Fallowfield, H. J., & Cromar, N. J. (2004). Use of microcosms to determine persistence of Escherichia coli in recreational coastal water and sediment and validation with in situ measurements. Journal of Applied Microb iology , 96 (5), 922 30. doi:10.1111/j.1365 - 2672.2004.02243.x Crampton, E. J. (1921). History of the St. Clair River . St. Clair, MI: St. Clair Republican. Crane, S. R., & Moore, J. A. (1986). Modeling enteric bacterial die - off: A review. Water, Air, & Soil P ollution , 27 (3 - 4), 411 439. doi:10.1007/BF00649422 142 Dick, L. K., Stelzer, E. a, Bertke, E. E., Fong, D. L., & Stoeckel, D. M. (2010). Relative decay of Bacteroidales microbial source tracking markers and cultivated Escherichia coli in freshwater microcosms. Applied and Environmental Microbiology , 76 (10), 3255 62. doi:10.1128/AEM.02636 - 09 Downey, A. S., Da Silva, S. M., Olson, N. D., Filliben, J. J., & Morrow, J. B. (2012). Impact of processing method on recovery of bacteria from wipes used in biological surf ace sampling. Applied and Environmental Microbiology , 78 (16), 5872 81. doi:10.1128/AEM.00873 - 12 Doyle, M. P., & Erickson, M. C. (2006). Closing the Door on the Fecal Coliform Assay. Microbe , 1 (4), 162 163. Droppo, I. G., Liss, S. N., Williams, D., Nelson, T., Jaskot, C., & Trapp, B. (2009). Dynamic Existence of Waterborne Pathogens within River Sediment Compartments. Implications for Water Quality Regulatory Affairs. Environmental Science & Technology , 43 (6), 1737 1743. doi:10.1021/es802321w Dufour, A. P. ( 1984). Health Effects Criteria for Fresh Recreational Waters . Washington, DC. Retrieved from http://www.epa.gov/nerlcwww/documents/frc.pdf Dufour, A. P. (2001). Discussion of indicator thresholds. Cincinnati, OH. Eichmiller, J. J., Borchert, A. J., Sadowsk y, M. J., & Hicks, R. E. (2014). Decay of genetic markers for fecal bacterial indicators and pathogens in sand from Lake Superior. Water Research , 59 , 99 111. doi:10.1016/j.watres.2014.04.005 Eichmiller, J. J., Hicks, R. E., & Sadowsky, M. J. (2013). Distr ibution of genetic markers of fecal pollution on a freshwater sandy shoreline in proximity to wastewater effluent. Environmental Science & Technology , 47 (7), 3395 402. doi:10.1021/es305116c Environmental Conuslting and Technology, I. (2007). Water Quality Sampling and Analysis Final Report . Clinton Township, MI. Esman, L. A. (2007). The Michigan Department of Environmental Quality Biennial Remedial Action Plan Update for Clintron River Area of Concern . Lansing, MI. Retrieved from http://www.epa.gov/greatlak es/aoc/clintonriver/pdfs/2007_ClintonRvrRAP.pdf Facklam, R. R., Carvalho, M. G. S., & Teixeira, L. M. (2002). History, Taxonomy, Biochemical Characteristics, and Antibiotic Susceptibility Testing of Enterococci. In M. S. Gilmore, D. B. Clewell, P. Courvali n, G. M. Dunny, B. E. Murray, & L. B. Rice (Eds.), The Enterococci: Pathogenesis, Molecular Biology, Antibiotic Resistance, and Infection Control, (p. 450). Washington, DC: ASM Press. Field, K. G., & Samadpour, M. (2007). Fecal source tracking, the indicat or paradigm, and managing water quality. Water Research , 41 (16), 3517 38. doi:10.1016/j.watres.2007.06.056 143 Fishbeck, Thompson, Carr & Huber, I. (2007). Anchor Bay Watershed Plan 2006 . Grand Rapids. Retrieved from http://michigan.gov/documents/deq/ess - nps - wmp - anchor - bay_209078_7.pdf Fisher, K., & Phillips, C. (2009). The ecology, epidemiology and virulence of Enterococcus. Microbiology (Reading, England) , 155 (Pt 6), 1749 57. doi:10.1099/mic.0.026385 - 0 Frahm, E., & O bst, U. (2003). Application of the fluorogenic probe technique (TaqMan PCR) to the detection of Enterococcus spp. and Escherichia coli in water samples. Journal of Microbiological Methods , 52 (1), 123 31. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12 401234 Franz, C. M., Holzapfel, W. H., & Stiles, M. E. (1999). Enterococci at the crossroads of food safety? International Journal of Food Microbiology , 47 (1 - 2), 1 24. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10357269 Fry, B. (2007). Coupled N, C and S stable isotope measurements using a dual - column gas chromatography system. Rapid Communications in Mass Spectrometry , 21 (5), 750 756. doi:10.1002/rcm.2892 Completion of the 2006 National Land Cover Database for the conterminous United States. Photogrammetric Engineering and Remote Sensing , 77 , 858 566. Furl, C. V, & Meredith, C. A. (2011). Mercury accumulation in sediment cores from three Washington state la kes: evidence for local deposition from a coal - fired power plant. Archives of Environmental Contamination and Toxicology , 60 (1), 26 33. doi:10.1007/s00244 - 010 - 9530 - 5 Garzio - Hadzick, A., Shelton, D. R., Hill, R. L., Pachepsky, Y. A., Guber, A. K., & Rowland , R. (2010). Survival of manure - borne E. coli in streambed sediment: effects of temperature and sediment properties. Water Research , 44 (9), 2753 62. doi:10.1016/j.watres.2010.02.011 Gavrilov, L. A., & Gavrilov, N. S. (1991). The Biology of Life Span: A Qua ntitative Approach . New York, New York: Harwood Academic Publisher. Gil, M. M., Miller, F. A., Brandão, T. R. S., & Silva, C. L. M. (2011). On the Use of the Gompertz Model to Predict Microbial Thermal Inactivation Under Isothermal and Non - Isothermal Conditions. Food Engineering Reviews , 3 (1), 17 25. doi:10.1007/s12393 - 010 - 9032 - 2 Gilpin, B. J., Devane, M., Robson, B., Sunlight Inactivation of Human Polymerase Chain Reaction Markers and Cultured Fecal Indicators in River and Saline Waters. Water Environment Research , 85 (8), 743 750. doi:10.2175/106143012X13560205 144290 Green, H. C., Dick, L. K., Gilpin, B., Samadpour, M., & Field, K. G. (2012). Genetic markers for rapid PCR - based identification of gull, Canada goose, duck, and chicken fecal contamination 144 in water. Applied and Environmental Microbiology , 78 (2), 503 10. doi:10.1128/AEM.05734 - 11 Green, H. C., Shanks, O. C., Sivaganesan, M., Haugland, R. A., & Field, K. G. (2011). Differential decay of human faecal Bacteroides in marine and freshwater. Environmental Microbiology , 13 (12), 3235 49. doi:10.1111/j.1462 - 292 0.2011.02549.x Griffith, J. F., Weisberg, S. B., & McGee, C. D. (2003). Evaluation of microbial source tracking methods using mixed fecal sources in aqueous test samples. Journal of Water and Health , 1 (4), 141 151. Haile, R. W., Witte, J. S., Gold, M., Cre The health effects of swimming in ocean water contaminated by storm drain runoff. Epidemiology (Cambridge, Mass.) , 10 (4), 355 63. Retrieved from http://www.jstor.org/stable/3703553?origin=JSTOR - pdf H aller, L., Amedegnato, E., Poté, J., & Wildi, W. (2009). Influence of Freshwater Sediment Characteristics on Persistence of Fecal Indicator Bacteria. Water, Air, and Soil Pollution , 203 (1 - 4), 217 227. doi:10.1007/s11270 - 009 - 0005 - 0 Haller, L., Poté, J., Loizeau, J. - L., & Wildi, W. (2009). Distribution and survival of faecal indicator bacteria in the sediments of the Bay of Vidy, Lake Geneva, Switzerland. Ecological Indicators , 9 (3), 540 547. doi:10.1016/j.ecolind.2008.08.001 Hansen, W., & Yourassowsky, E. (1984). Detection of beta - glucuronidase in lactose - fermenting members of the family Enterobacteriaceae and its presence in bacterial urine cultures. Journal of Clinical Microbiology , 20 (6), 1177 9. Retrieved from http://www.pubmedcen tral.nih.gov/articlerender.fcgi?artid=271541&tool=pmcentrez&rend ertype=abstract Harrington - Hughes, K. (1978). Great Lakes Water Quality: A Progress Report. Journal (Water Pollution Control Federation) , 50 (8), 1886 1888. Retrieved from http://www.jstor.org/ stable/25040367 Harwood, V. J., Staley, C., Badgley, B. D., Borges, K., & Korajkic, A. (2014). Microbial source tracking markers for detection of fecal contamination in environmental waters: relationships between pathogens and human health outcomes. FEMS M icrobiology Reviews , 38 (1), 1 40. doi:10.1111/1574 - 6976.12031 Haugland, R. A., Siefring, S. C., Wymer, L. J., Brenner, K. P., & Dufour, A. P. (2005). Comparison of Enterococcus measurements in freshwater at two recreational beaches by quantitative polymera se chain reaction and membrane filter culture analysis. Water Research , 39 (4), 559 68. doi:10.1016/j.watres.2004.11.011 Healy, D. F., Chambers, D. B., Rachol, C. M., & Jodoin, R. S. (2008). Water quality of the St. Clair River, Lake St. Clair, and their U. S. tributaries, 1946 2005: U.S. Geological Survey 145 Scientific Investigations Report 2007 5172 . Reston, VA. Retrieved from http://pubs.usgs.gov/sir/2007/5172/pdf/sir2007 - 5172_web.pdf Hlavsa, M. C., Roberts, V. A., Kahler, A. M., Hilborn, E. D., Wade, T. J., Backer, L. C., & Yoder, J. S. (2014). Recreational Water Associated Disease Outbreaks United States, 2009 2010. Morbidity and Mortality Weekly Report. Surveillance Summaries , 63 (01), 6 10. Retrieved from http://www.cdc.gov/mmwr/preview/mmwrhtml/mm6301a2. htm?s_cid=mm6301a2_w Hodell, D. A., & Schelske, C. L. (1998). Production, sedimentation, and isotopic composition of organic matter in Lake Ontario. Limnology and Oceanography , 43 (2), 200 214. doi:10.4319/lo.1998.43.2.0200 Hogg, R. V., & Craig, A. T. (1978 ). Introduction to Mathematical Statistics (4th ed.). New York, New York: MacMillian. Hoshino, Y. T., & Matsumoto, N. (2007). DNA - versus RNA - based denaturing gradient gel electrophoresis profiles of a bacterial community during replenishment after soil fu migation. Soil Biology and Biochemistry , 39 (2), 434 444. Hunter, T. S., & Croley, T.E., I. (1993). Great Lakes Monthly Hyrdologic Data, NOAA Data Report ERL GLERL . Springfiled, Virginia. Inceoglu, O., Hoogwout, E. F., Hill, P., & van Elsas, J. D. (2010). E ffect of DNA extraction method on the apparent microbial diversity of soil. Applied and Environmental Microbiology , 76 (10), 3378 82. doi:10.1128/AEM.02715 - 09 International Joint Commission, & Upper Great Lakes Connecting Channels Study. (1988). Upper Great Lakes Connecting Channels Study: Volume II - Final Report . Retrieved from http://nepis.epa.gov/EPA/html/DLwait.htm?url=/Exe/ZyPDF.cgi?Dockey=20017E8O.PDF Recovery f rom Subsoil Clay Sediments >1.000 Times. In ISME14 . Coppenhagen, Denmark. Jeng, H., England, A., & Bradford, H. (2005). Indicator Organisms Associated with Stormwater Suspended Particles and Estuarine Sediment. Journal of Environmental Science and Health, Part A , 40 (4), 779 791. doi:10.1081/ESE - 200048264 Jett, B. D., Huycke, M. M., & Gilmore, M. S. (1994). Virulence of enterococci. Clinical Microbiology Reviews , 7 (4), 462 78. doi:10.1128/ CMR.7.4.462 Juneja, V. K., Huang, L., & Marks, H. (2006). Approaches for Modeling Thermal Inactivation of Foodborne Pathogens Kinetic Analysis of Bacterial Inactivation by Heat - Approach. Symposium A Quarterly Journal In Modern Foreign Literatures , 235 251. Juneja, V. K., Marks, H. M ., & Mohr, T. (2003). Predictive Thermal Inactivation Model for Effects of Temperature , Sodium Lactate , NaCl , and Sodium Pyrophosphate on Salmonella 146 Serotypes in Ground Beef. Applied and Environmental Microbiology , 69 (9), 5138 5156. doi:10.1128/AM.69.9. 5138 - 5156.2003 Juneja, V. K., Marks, H. M., Mohr, T., Heiri, O., Lotter, A. F., Lemcke, G., & Heinri, O. (2003). Predictive Thermal Inactivation Model for Effects of Temperature , Sodium Lactate , NaCl , and Sodium Pyrophosphate on Salmonella Serotypes in Ground Beef. Applied and Environmental Microbiology , 69 (9), 5138 5156. doi:10.1128/AM.69.9.5138 - 5156.2003 Jweda, J., & Baskaran, M. (2011). Interconnected riverine lacustrine systems as sedimentary repositories: Case study in southeast Michigan using 210Pb and 137Cs - based sediment accumulation and mixing models. Journal of Great Lakes Research , 37 (3), 432 446. doi:10.1016/j.jglr.2011.04.010 Kamau, D. N., Doores, S., & Pruitt, K. M. (1990). Enhanced thermal destruction of Listeria monocytogenes and Staphyloc occus aureus by the lactoperoxidase system. Applied and Environmental Microbiology , 56 , 2711 2716. Kaushal, S., & organic matter sources , and historical deforestation in Lake Pleasant , Massachusetts , USA. Journal of Paleolimnology , 22 (4), 439 442. doi:10.1023/A:1008027028029 Kelsey, H., P orter, D. E., Scott, G., Neet, M., & White, D. (2004). Using geographic information systems and regression analysis to evaluate relationships between land use and fecal coliform bacterial pollution. Journal of Experimental Marine Biology and Ecology , 298 (2 ), 197 209. doi:10.1016/S0022 - 0981(03)00359 - 9 Kirk, J. L., Beaudette, L. A., Hart, M., Moutoglis, P., Klironomos, J. N., Lee, H., & Trevors, J. T. (2004). Methods of studying soil microbial diversity. Journal of Microbiological Methods , 58 (2), 169 88. doi: 10.1016/j.mimet.2004.04.006 Klein, G., Pack, A., Bonaparte, C., & Reuter, G. (1998). Taxonomy and physiology of probiotic lactic acid bacteria. International Journal of Food Microbiology , 41 (2), 103 25. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/970 4860 Klein, M., Brown, L., Ashbolt, N. J., Stuetz, R. M., & Roser, D. J. (2011). Inactivation of indicators and pathogens in cattle feedlot manures and compost as determined by molecular and culture assays. FEMS Microbiology Ecology , 77 (1), 200 10. doi:10. 1111/j.1574 - 6941.2011.01098.x Kreader, C. A. (1998). Persistence of PCR - detectable Bacteroides distasonis from human feces in river water. Applied and Environmental Microbiology , 64 (10), 4103 5. Retrieved from http://www.pubmedcentral.nih.gov/articlerender .fcgi?artid=106613&tool=pmcentrez&rend ertype=abstract Layton, A., McKay, L., Williams, D., Garrett, V., Gentry, R., & Sayler, G. (2006). Development of Bacteroides 16S rRNA gene TaqMan - based real - time PCR assays for estimation of total, 147 human, and bovine f ecal pollution in water. Applied and Environmental Microbiology , 72 (6), 4214 24. doi:10.1128/AEM.01036 - 05 Lee, C. S., & Lee, J. (2010). Evaluation of new gyrB - based real - time PCR system for the detection of B. fragilis as an indicator of human - specific fec al contamination. Journal of Microbiological Methods , 82 (3), 311 8. doi:10.1016/j.mimet.2010.07.012 Lemarchand, K., & Lebaron, P. (2003). Occurrence of Salmonella spp. and Cryptosporidium spp. in a French coastal watershed: relationship with fecal indicato rs. FEMS Microbiology Letters , 218 (1), 203 209. doi:10.1111/j.1574 - 6968.2003.tb11519.x Liang, Z., He, Z., Zhou, X., Powell, C. A., Yang, Y., Roberts, M. G., & Stoffella, P. J. (2012). High diversity and differential persistence of fecal Bacteroidales popul ation spiked into freshwater microcosm. Water Research , 46 (1), 247 57. doi:10.1016/j.watres.2011.11.004 Lloyd, K. G., Macgregor, B. J., & Teske, A. (2010). Quantitative PCR methods for RNA and DNA in marine sediments: maximizing yield while overcoming inhibition. FEMS Microbiology Ecology , 72 (1), 143 51. doi:10.1111/j.1574 - 6941.2009.00827.x López - Pila, J. M., & Szewzyk, R. (2000). Estimatin g the infection risk in recreational waters from the faecal indicator concentration and from the ratio between pathogens and indicators. Water Research , 34 (17), 4195 4200. doi:10.1016/S0043 - 1354(00)00197 - 4 DNA extraction procedure: a critical issue for bacterial diversity assessment in marine sediments. Environmental Microbiology , 8 (2), 308 20. doi:10.1111/j.1462 - 2920.2005.00896.x Madigan, M. T., Martinko, J. M., Dunlap, P. V., & Clark, D. P. (2009). Wastew ater Treatment, Water Purification, and Waterborne Pathogens. In Brock Biology of Microorganisms (12th ed., pp. 1025 1042). San Francisco, CA: Pearson Education, INC. Mahalanabis, M., Do, J., ALMuayad, H., Zhang, J. Y., & Klapperich, C. M. (2010). An integ rated disposable device for DNA extraction and helicase dependent amplification. Biomedical Microdevices , 12 (2), 353 9. doi:10.1007/s10544 - 009 - 9391 - 8 Marti, R., Mieszkin, S., Solecki, O., Pourcher, A. - M., Hervio - Heath, D., & Gourmelon, M. (2011). Effect of oxygen and temperature on the dynamic of the dominant bacterial populations of pig manure and on the persistence of pig - associated genetic markers, assessed in river water microcosms. Journal of Applied Microbiology , 111 (5), 1159 75. doi:10.1111/j.1365 - 26 72.2011.05131.x MDEQ. (2012). Water Quality Parameters. Retrieved from http://www.michigan.gov/deq/0,1607,7 - 135 - 3313_3682 - 10416 -- ,00.html 148 MDEQ. (2013). Michigan beach monitoring year 2013 annual report . Lansing, MI. Retrieved from http://www.michigan.gov/d ocuments/deq/wrd - beach - 2013annualreport_454449_7.pdf?20141105114400 Medema, G., Bahar, M., & Schets, F. M. (1997). Survival of cryptosporidium parvum, escherichia coli, faecal enterococci and clostridium perfringens in river water: influence of temperature and autochthonous microorganisms. Water Science and Technology , 35 (11 - 12), 249 252. doi:10.1016/S0273 - 1223(97)00267 - 9 Meyers, P. A. (2007). The Interactions Between Sediments and Water . (B. Kronvang, J. Faganeli, & N. Ogrinc, Eds.) The Interactions Between Sediments and Water . Dordrecht: Springer Netherlands. doi:10.1007/978 - 1 - 4020 - 5478 - 5 Meyers, P. A., & Ishiwatari, R. (1993). Lacustrine organic geochemistry an overview of indicators of organic matter sources and diagenesis in lake sediments. Organic Geoch emistry , 20 (7), 867 900. doi:10.1016/0146 - 6380(93)90100 - P Michigan Department of Environmental Quality. (1988). Clinton River Remedial Action Plan . Lansing, MI. Retrieved from http://www.epa.gov/glnpo/aoc/clintonriver/pdfs/1988_Clinton River RAP.pdf Midwes tern Regional Climate Center. (2014). Cli - MATE: MRCC Application Tools Environment. Retrieved from http://mrcc.isws.illinois.edu/CLIMATE/Station/Annual/StnAnnualBTD.jsp Mieszkin, S., Furet, J. - P., Corthier, G., & Gourmelon, M. (2009). Estimation of Pig Fec al Contamination in a River Catchment by Real - Time PCR Using Two Pig - Specific Bacteroidales 16S rRNA Genetic Markers. Applied and Environmental Microbiology , 75 (10), 3045 3054. doi:10.1128/AEM.02343 - 08 Murray, B. E. (1990). The life and times of the Entero coccus. Clinical Microbiology Reviews , 3 (1), 46 65. doi:10.1128/ CMR.3.1.46 Natvig, E. E., Ingham, S. C., Ingham, B. H., Cooperband, L. R., & Roper, T. R. (2002). Salmonella enterica Serovar Typhimurium and Escherichia coli Contamination of Root and Leaf V egetables Grown in Soils with Incorporated Bovine Manure. Applied and Environmental Microbiology , 68 (6), 2737 2744. doi:10.1128/AEM.68.6.2737 - 2744.2002 Okabe, S., Okayama, N., Savichtcheva, O., & Ito, T. (2007). Quantification of host - specific Bacteroides - Prevotella 16S rRNA genetic markers for assessment of fecal pollution in freshwater. Applied Microbiology and Biotechnology , 74 (4), 890 901. doi:10.1007/s00253 - 006 - 0714 - x Okabe, S., & Shimazu, Y. (2007). Persistence of host - specific Bacteroides - Prevotella 16S rRNA genetic markers in environmental waters: effects of temperature and salinity. Applied Microbiology and Biotechnology , 76 (4), 935 44. doi:10.1007/s00253 - 007 - 1048 - z 149 - extraction of microbial DNA and RNA in adsorptive soils. , 63 , 37 49. doi:10.1016/j.soilbio.2013.02.007 Peleg, M. (2006). Advanced quantitative microbiology for foo predicting growth and inactivation . Boca Raton, FL: Taylor and Francis. Poh, C. H., Oh, H. M. L., & Tan, a L. (2006). Epidemiology and clinical outcome of enterococcal bacteraemia in an acute care hospital. The Journal of Inf ection , 52 (5), 383 6. doi:10.1016/j.jinf.2005.07.011 Pote, J., Haller, L., Kottelat, R., Sastre, V., Arpagaus, P., & Wildi, W. (2009). Persistence and growth of faecal culturable bacterial indicators in water column and sediments of Vidy Bay, Lake Geneva, Switzerland. Journal of Environmental Sciences (China) , 21 (1), 62 9. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/19402401 Prüss, A. (1998). Review of epidemiological studies on health effects from exposure to recreational water. International Journal of Epidemiology , 27 (1), 1 9. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9563686 R Development Core Team. (2013). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from www.r - project.org Reischer, G. H., Kasper, D. C., Steinborn, R., Farnleitner, A. H., & Mach, R. L. (2007). A quantitative real - time PCR assay for the highly sensitive and specific detection of human faecal influence in spring water from a large alpine catchment area. Letters in Applied Microbiology , 44 (4), 351 6. doi:10.1111/j.1472 - 765X.2006.02094.x Rogers, S. W., Donnelly, M., Peed, L., Kelty, C. a, Mondal, S., Zhong, Z., & Shanks, O. C. (2011). Decay of bacterial pathogens, fecal indicators, and real - time quant itative PCR genetic markers in manure - amended soils. Applied and Environmental Microbiology , 77 (14), 4839 48. doi:10.1128/AEM.02427 - 10 Ryan, K., Ray, C. G., Ahmad, N., Drew, W. L., & Plorde, P. (2010). Streptococci and Enterococci. In Sherris Mecical Micro biology (5th ed., p. 1026). New York, New York: McGraw - Hill Medical. Santo Domingo, J. W., Bambic, D. G., Edge, T. a, & Wuertz, S. (2007). Quo vadis source tracking? Towards a strategic framework for environmental monitoring of fecal pollution. Water Resea rch , 41 (16), 3539 52. doi:10.1016/j.watres.2007.06.001 Savichtcheva, O., Okayama, N., & Okabe, S. (2007). Relationships between Bacteroides 16S rRNA genetic markers and presence of bacterial enteric pathogens and conventional fecal indicators. Water Resear ch , 41 (16), 3615 28. doi:10.1016/j.watres.2007.03.028 150 SEMCOG. (2002). Historical Population and Employment by Minor Civil Division . Detroit, MI. Retrieved from http://library.semcog.org/InmagicGenie/DocumentFolder/HistoricalPopulationSEMI.pdf Seurinck, S., Defoirdt, T., Verstraete, W., & Siciliano, S. D. (2005). Detection and quantification of the human - specific HF183 Bacteroides 16S rRNA genetic marker with real - time PCR for assessment of human faecal pollution in freshwater. Environmental Microbiology , 7 ( 2), 249 59. doi:10.1111/j.1462 - 2920.2004.00702.x Shanks, O. C., Kelty, C. a, Sivaganesan, M., Varma, M., & Haugland, R. a. (2009). Quantitative PCR for genetic markers of human fecal pollution. Applied and Environmental Microbiology , 75 (17), 5507 13. doi:1 0.1128/AEM.00305 - 09 a. (2010). Performance of PCR - based assays targeting Bacteroidales genetic markers of human fecal pollution in sewage and fecal samples. Env ironmental Science & Technology , 44 (16), 6281 8. doi:10.1021/es100311n Quantification of functional genes from procaryotes in soil by PCR. Journal of Microbiological Methods , 68 (3), 445 52. doi:10.1016/j.mimet.2006.10.001 Signoretto, C., del Mar Lleo, M., Tafi, M. C., & Canepari, P. (2000). Cell Wall Chemical Composition of Enterococcus faecalis in the Viable but Nonculturable State. Applied and Environmental Microbiol ogy , 66 (5), 1953 1959. doi:10.1128/AEM.66.5.1953 - 1959.2000 Snyder, L., & Champness, W. (2007). Molecular Genetics of Bacteria (3rd ed.). Washington, DC: ASM Press. Soller, J. A., Schoen, M. E., Bartrand, T., Ravenscroft, J. E., & Ashbolt, N. J. (2010). Est imated human health risks from exposure to recreational waters impacted by human and non - human sources of faecal contamination. Water Research , 44 (16), 4674 91. doi:10.1016/j.watres.2010.06.049 Srinivasan, S., Aslan, A., Xagoraraki, I., Alocilja, E., & Ros e, J. B. (2011). Escherichia coli, enterococci, and Bacteroides thetaiotaomicron qPCR signals through wastewater and septage treatment. Water Research , 45 (8), 2561 72. doi:10.1016/j.watres.2011.02.010 St. Clair County Health Department. (2009). An Evaluati on of Anchor Bay Watershed Planning Activities and Monitoring Data . Port Huron, MI. Stevenson, A. H. (1953). Studies of bathing water quality and health. American Journal of Public , 43 (5 Pt 1), 529 38. Retrieved from http://w ww.pubmedcentral.nih.gov/articlerender.fcgi?artid=1620266&tool=pmcentrez&ren dertype=abstract 151 Suvarna, K., Stevenson, D., Meganathan, R., & Hudspeth, M. E. S. (1998). Menaquinone (Vitamin K2) biosynthesis: Localization and characterization of the menA gene from Escherichia coli menaquinone (Vitamin K2) biosynthesis: Localization and characterization of the menA gene from Escherichia coli. Microbiology , 180 (10). Tang, X., Gao, G., Zhu, L., Chao, J., & Qin, B. (2009). DNA extraction procedure affects organic - a ggregate - attached bacterial community profiles from a shallow eutrophic lake. Canadian Journal of Microbiology , 55 (6), 776 82. doi:10.1139/w09 - 026 Thevenon, F., Regier, N., Benagli, C., Tonolla, M., Adatte, T., Wildi, W., & Poté, J. (2012). Characterizatio n of fecal indicator bacteria in sediments cores from the largest freshwater lake of Western Europe (Lake Geneva, Switzerland). Ecotoxicology and Environmental Safety , 78 , 50 6. doi:10.1016/j.ecoenv.2011.11.005 United Nations. (2010). The human right to wa ter and sanitation A/Res/64/292 . Geneva, Switzerland: United Nations. Unknown. (1883). History of St. Clair County, Michigan, containing an account of its settlement, growth, development and resources, its war record, biographical sketches, the whole prece ded by a history of Michigan . Chicago, IL: A.T. Andreas and Co. Retrieved from http://name.umdl.umich.edu/ARX2236.0001.001 USEPA. (1991). Technical Support Document for Water Quality - Based Toxics Control . Washington, DC. USEPA. (2002). Escherichia coli ( E . coli ) in Water by Membrane Filtration Using Modified Escherichia coli Agar ( Modified mTEC ) . Science And Technology . Washington, DC. USEPA. (2003). Bacterial water quality standards for recreational waters (freshwater and marine wa ters) status report . Water . Washington, DC. USEPA. (2009). Method 1600: Enterococci in Water by Membrane Filtration Using membrane - Enterococcus Indoxyl - alpha - D - Glucoside Agar . Environmental Protection . Washington, DC. USEPA. (2010a). Enterococci in Water by TaqMan ® Quantitative Polymerase Chain Reaction (qPCR) Assay . Environmental Protection . Washington, DC. USEPA. (2010b). Method B: Bacteroidales in Water by TaqMan® Quantitative Polymerase Chain Reaction (qPCR) Assay . Washington, DC. Retrieved from http://water.epa.gov/scitech/methods/cwa/bioindicators/upload/methodb2010.pdf USEPA. (2011). Drinking Water Pathogens and Their Indicators: A Reference Source. Retrieved from http://www.epa.gov/enviro/html/icr/gloss_path.html 152 USEPA. (2012a) . 5.11 Fecal Bacteria. Retrieved from http://water.epa.gov/type/rsl/monitoring/vms511.cfm USEPA. (2012b). Method 1611: Enterococci in Water by TaqMan Quantitative Polymerase Chain Reaction (qPCR) Assay . Environmental Protection . Washington, DC. Retrieved f rom http://water.epa.gov/scitech/methods/cwa/bioindicators/upload/Method - 1611 - Enterococci - in - Water - by - TaqMan - Quantitative - Polymerase - Chain - Reaction - qPCR - Assay.pdf USEPA. (2012c). Recreational Water Quality Criteria . Washington, DC. Retrieved from http://wa ter.epa.gov/scitech/swguidance/standards/criteria/health/recreation/upload/RWQC2 012.pdf USGS. (2014). USGS Surface - Water Annual Statistics for the Nation. Retrieved from http://waterdata.usgs.gov/nwis/annual?referred_module=sw&site_no=04165500&por_0416 5500 _2=892169,00060,2,1934,2014&year_type=W&format=html_table&date_format=YY YY - MM - DD&rdb_compression=file&submitted_form=parameter_selection_list (2008). High sens itivity of children to swimming - associated gastrointestinal illness: results using a rapid assay of recreational water quality. Epidemiology (Cambridge, Mass.) , 19 (3), 375 83. doi:10.1097/EDE.0b013e318169cc87 Wade, T. J., Calderon, R. L., Sams, E., Beach, M., Brenner, K. P., Williams, A. H., & Dufour, A. P. (2006). Rapidly measured indicators of recreational water quality are predictive of swimming - associated gastrointestinal illness. Environmental Health Perspectives , 114 (1), 24 8. Retrieved from http://ww w.pubmedcentral.nih.gov/articlerender.fcgi?artid=1332651&tool=pmcentrez&ren dertype=abstract Wade, T. J., Pai, N., Eisenberg, J. N. S., & Colford, J. M. (2003). Do U.S. Environmental Protection Agency water quality guidelines for recreational waters prevent gastrointestinal illness? A systematic review and meta - analysis. Environmental Health Perspectives , 111 (8), 1102 9. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1241558&tool=pmcentrez&ren dertype=abstract Wade, T. J., Sams, E., Rapidly measured indicators of recreational water quality and swimming - associated illness at marine beaches: a prospective cohort study. Scie nce Source , 9 (1), 66. doi:10.1186/1476 - 069X - 9 - 66 Walters, S. P., Yamahara, K. M., & Boehm, A. B. (2009). Persistence of nucleic acid markers of health - relevant organisms in seawater microcosms: implications for their use in assessing risk in recreational w aters. Water Research , 43 (19), 4929 39. doi:10.1016/j.watres.2009.05.047 153 Wheeler Alm, E., Burke, J., & Spain, A. (2003). Fecal indicator bacteria are abundant in wet sand at freshwater beaches. Water Research , 37 (16), 3978 82. doi:10.1016/S0043 - 1354(03)003 01 - 4 Whitehouse, C. A., & Hottel, H. E. (2007). Comparison of five commercial DNA extraction kits for the recovery of Francisella tularensis DNA from spiked soil samples. Molecular and Cellular Probes , 21 (2), 92 6. doi:10.1016/j.mcp.2006.08.003 Wolf, H. M. (1972). The coliform count as a measure of water quality. In R. Mitchell (Ed.), Water Pollution Microbiology (1st ed.). New York, New York: Wiley - Interscience. World Health Organization. (2003). Guidelines for safe recreational water environments, Volume 1: Costal and Freshwaters. Geneva, Switzerland: WHO Press. Retrieved from http://whqlibdoc.who.int/publications/2003/9241545801_ch4.pdf World Health Organization. (2011). Guidelines for Drinking - water Quality. WHO chronicle (4th ed., Vol. 38). Geneva, Swit zerland: WHO Press. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/15806952 Wu, J. - W., Hung, W. - L., & Tsai, C. - H. (2004). Estimation of parameters of the Gompertz distribution using the least squares method. Applied Mathematics and Computation . doi:10.1 016/j.amc.2003.08.086 Xiong, W., Xie, P., Wang, S., & Niu, Y. (2014). The effect of depth on microbial communities from mesotrophic lake sediments as measured by two DNA extraction methods. Journal of Freshwater Ecology , 1 14. doi:10.1080/02705060.2014.915587 (2003). A genomic view of the human - Bacteroides thetaiotaomicron symbiosis. Science (New York, N.Y.) , 299 (5615), 2074 6. doi:1 0.1126/science.1080029 Yamahara, K. M., Sassoubre, L. M., Goodwin, K. D., & Boehm, A. B. (2012). Occurrence and persistence of bacterial pathogens and indicator organisms in beach sand along the California coast. Applied and Environmental Microbiology , 78 ( 6), 1733 45. doi:10.1128/AEM.06185 - 11 Yamahara, K. M., Walters, S. P., & Boehm, A. B. (2009). Growth of enterococci in unaltered, unseeded beach sands subjected to tidal wetting. Applied and Environmental Microbiology , 75 (6), 1517 24. doi:10.1128/AEM.02278 - 08 Yampara - Iquise, H., Zheng, G., Jones, J. E., & Carson, C. A. (2008). Use of a Bacteroides thetaiotaomicron - specific alpha - 1 - 6, mannanase quantitative PCR to detect human faecal pollution in water. Journal of Applied Microbiology , 105 (5), 1686 93. doi:1 0.1111/j.1365 - 2672.2008.03895.x 154 Zhou, J., Bruns, M. A., & Tiedje, J. M. (1996). DNA recovery from soils of diverse composition. Applied and Environmental Microbiology , 62 (2), 316 22. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid= 167800&tool=pmcentrez&rend ertype=abstract