xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003223
The importance of benthic habitats as reservoirs of persistent fecal indicator bacteria
h [electronic resource] /
by Brian Badgley.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains X pages.
Dissertation (Ph.D.)--University of South Florida, 2009.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
ABSTRACT: Enterococci are fecal indicator bacteria (FIB) that are used worldwide for water quality assessment. However, evidence of high densities and extended survival of enterococci in sediments and submerged aquatic vegetation (SAV) has caused uncertainty about their reliability in predicting human health risks from recreational activities in environmental waters. To address the concern that sediments and SAV may harbor large reservoirs of enterococci that can affect water column concentrations, aquatic mesocosms and environmental sampling were employed to investigate patterns of enterococci densities and population structure across the Tampa Bay watershed. In mesocosm experiments and environmental samples, SAV harbored higher densities of enterococci, per mass of substrate, than sediments, and sediments harbored higher densities than water. Population structure assessed by BOX-PCR genotyping was relatively unique in each sample, although slight similarities among samples suggested grouping primarily by location rather than substrate or season. Strain diversity was highly variable, and many samples had low diversity, including nearly monoclonal structure throughout the mesocosm experiments and in several of the environmental samples. Several strains were highly abundant and cosmopolitan (found across sites, seasons, and substrates), and may represent highly naturalized and reproducing indicator bacteria populations that are not directly related to pollution events. When the enterococci densities were viewed from the perspective of the entire aquatic system, SAV-associated enterococci did not comprise a major proportion of the total population, due to the typically large differences in volume of each substrate (SAV vs. sediments vs. water). Instead, the largest proportions of enterococci were typically found in the water or the sediments, depending on the relative volume of substrate or the enterococci density associated with each substrate. Modeling results illustrate that the relative importance of each substrate in terms of FIB populations can shift dramatically over time and space due to changes such as vegetation cover, tidal cycles, and bacteria densities. Furthermore, at several sites within the watershed, estimates of sediment and bacteria resuspension from sediments were very low, suggesting that this process rarely, if ever, significantly affects water column concentrations of enterococci in the watershed.
Advisor: Valerie J. Harwood, Ph.D.
t USF Electronic Theses and Dissertations.
The Importance of Benthic Habitats as Reservoi rs of Persistent Fecal Indicator Bacteria by Brian D. Badgley A dissertation submitted in partial fulfillment of the requirements of the degree of Doctor of Philosophy Department of Integrative Biology College of Arts and Sciences University of South Florida Co-Major Professor: Valerie J. Harwood, Ph.D. Co-Major Professor: Florence I. M. Thomas, Ph.D. Susan S. Bell, Ph.D. Gordon A. Fox, Ph.D. Mya Breitbart, Ph.D. Keywords: Enterococcus clonal structure, populati on dynamics, microbial ecology, microbial resuspension Copyright 2009, Brian D. Badgley
To my Mom and Dad Â– this was made possibl e by the love, support, and opportunity that you have provided throughout my life. You always have my thanks and my love.
ACKNOWLEDGEMENTS I would like to extend my sincere gratitude to Drs. Valerie Harwood and Florence Thomas, my co-major professors, for the s upport, respect, and friendship they have shown to me during my time as a doctoral student. I also thank my other committee members, Drs. Susan Bell and Gordon Fox, fo r their insight and gui dance, and Dr. Mya Breitbart for participating as the Outside Chair of my defense. I am also deeply indebted to the numerous friends and colleagues that ha ve contributed consider able time and effort in assisting me in this re search, including Miriam Br ownell, Katrina Gordon, Neal Halstead, Gabe Herrick, Phoebe Koch, Shannon McQuaig, Alison Meyers, Bina Nayak, Mythili Penugonda, Chris Staley, Zach Staley, and Stacy Villanueva. This research was made possible in part by awards from the Na tional Water Research Institute, Aylesworth Foundation for the Advancement of Marine Science, and the Old Salt Fishing Club. Finally, I would like to acknow ledge my wife, Andrea, my son, Owen, and my daughter, Annabelle. Their love and laughter have provided endle ss support over the last five years, and I will be eternally grateful for the sa crifices they made so that I could complete my degree.
i TABLE OF CONTENTS LIST OF TABLES.............................................................................................................iii LIST OF FIGURES............................................................................................................v ABSTRACT.....................................................................................................................vi ii CHAPTER ONE Â– BAC KGROUND AND OVERVIEW OF RESEARCH.....................1 Research Goals and Chapter Objectives...............................................................22 Significance of Research.......................................................................................24 CHAPTER TWO Â– THE EFFECTS OF SUBMERGED AQUATIC VEGETATION ON THE PERSISTENCE OF ENVIRONMENTAL POPULATIONS OF ENTEROCOCCUS SPP............................................................26 Introduction...........................................................................................................26 Methods.................................................................................................................29 Experimental Mesocosms.........................................................................29 Bacterial Concentrations...........................................................................31 Enterococcus Genotyping.........................................................................33 Taxonomic Identification of Isolates........................................................34 Calculations and Statistical Analysis........................................................35 Results...................................................................................................................37 Mesocosm Conditions...............................................................................37 Population Dynamics................................................................................37 Effects of SAV..........................................................................................38 Population Structure..................................................................................49 Discussion.............................................................................................................52 CHAPTER THREE Â– THE IMPORTANCE OF SEDIMENT AND SUBMERGED AQUATIC VEGETATION AS POTENTIAL HABITATS FOR PERSISTENCE STRAINS OF ENTEROCOCCUS ACROSS A WATERSHED............................................................................................................58 Introduction...........................................................................................................58 Methods.................................................................................................................62 Sampling Sites..........................................................................................62 Environmental Sampling..........................................................................64 Genetic Typing..........................................................................................65
ii Taxonomic Identification of Isolates........................................................66 Calculations and Statistical Analysis........................................................67 Results...................................................................................................................67 Enterococci Densities................................................................................67 Clonal Structure........................................................................................73 Discussion.............................................................................................................85 CHAPTER FOUR Â– INVESTIGATING THE IMPORTANCE OF SEDIMENT AND SUBMERGED AQUATIC VEGETATION AS ENVIRONMENTAL RESERVOIRS FOR WATER QUALI TY INDICATOR BACTERIA.....................95 Introduction...........................................................................................................95 Methods...............................................................................................................102 Environmental Sampling........................................................................102 Habitat Characterization.........................................................................102 Calculations.............................................................................................105 Modeling and Sediment Resuspension Estimates...................................106 Results.................................................................................................................110 Habitat Characterization and Area Normalization..................................110 Modeling Theoretical Habitat Changes..................................................118 Resuspension Estimates..........................................................................123 Discussion...........................................................................................................133 SAV as a Reservoir.................................................................................134 Sediment as a Reservoir..........................................................................136 Predicting Reservoir Shifts.....................................................................138 Resuspension of FIB...............................................................................140 LITERATURE CITED...................................................................................................147 ABOUT THE AUTHOR.......................................................................................End Page
iii LIST OF TABLES Table 1. Mean concentration and pr oportional distribu tion of culturable enterococci in water, sediment and SAV in the vegetated mesocosms for all experiments......................................................................................................44 Table 2. Comparison of enterococci persistence between vegetated and unvegetated mesocosms........................................................................................50 Table 3. Results from a three-factor ANOVA showing significant differences in mean enterococci densities across the watershed between sites, matrices, and seasons............................................................................................................69 Table 4. Measures of Enterococcus strain diversity in each sample of site, season, and substrate..........................................................................................................76 Table 5. Measures of Enterococcus strain diversity in each sample of site and season with all three substrates comb ined (water + sediment + SAV).................82 Table 6. Results from a three-factor ANOVA showing significant differences in percentage of strains in each samp le that are unique between sites, matrices, and seasons............................................................................................84 Table 7. List of strains that occurred in four or more sa mples, with species results from 16s sequencing and a summary of their distribution among the samples..................................................................................................................88 Table 8. Key habitat characteristics meas ured at each of the freshwater sites, which were used to conv ert enterococci densities in each substrate from mass-normalized values to those normalized to landscape area.........................111 Table 9. Key habitat characteristics meas ured at each of the freshwater sites, which were used to conv ert enterococci densities in each substrate from mass-normalized values to those normalized to landscape area.........................112 Table 10. Sediment characteristics at the la rge stream, river, lake, and upper bay sites.....................................................................................................................124 Table 11. Values for key hydrodynamic parameters and resuspension of sediment and associated bacteria estimated fr om stream flow data at the large stream and river sites..........................................................................................127
iv Table 12. Values for key hydrodynamic a nd bedform parameters, and estimates for resuspension of sediment and associ ated bacteria at the lake and upper bay sites...............................................................................................................131
v LIST OF FIGURES Figure 1. Schematic representation of th e recirculating flume (~180L) that was used to created vegetated and unvegetated mesocosms for flow experiments...........................................................................................................30 Figure 2. Culturable enterococci concentr ations in unvegetated (A) and vegetated (B) mesocosms for the April experiments............................................................39 Figure 3. Culturable enterococci concentr ations in unvegetated (A) and vegetated (B) mesocosms for the May experiments.............................................................40 Figure 4. Culturable enterococci concentr ations in unvegetated (A) and vegetated (B) mesocosms for the July experiments..............................................................41 Figure 5. Culturable enterococci concentr ations in unvegetated (A) and vegetated (B) mesocosms for the August experiments.........................................................42 Figure 6. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the April experiments................................45 Figure 7. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the May experiments.................................46 Figure 8. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the July experiments..................................47 Figure 9. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the August experiments.............................48 Figure 10. Representative results fro m Enterococcus strain typing showing extremely low strain diversity detected by BOX-PCR for Enterococcus isolates recovered from th e mesocosm experiments.............................................51 Figure 11. Site locations in the Tampa Bay watershed....................................................63 Figure 12. Mean densities of culturable enterococci from water, sediment, and SAV for each site over all sampling dates............................................................70 Figure 13. Mean densities of culturable enterococci from water, sediment, and SAV for all sites grouped by each season.............................................................71
vi Figure 14. Mean enterococci densities in all substrates combined (total) and in the water at each site compared to dir ect-line distance of the site from the mouth of Tampa Bay............................................................................................72 Figure 15. Mean enterococci densities in sediment and SAV at each site compared to direct-line distance of the site from the mouth of Tampa Bay.........74 Figure 16. Accumulation curves constr ucted for unique strains found in each matrix at the river site during the (A ) spring, (B) summer, (C) fall, and (D) winter season.........................................................................................................75 Figure 17. Accumulation curv es constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the large stream site.........................................................................................................................78 Figure 18. Accumulation curv es constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the river site...................79 Figure 19. Accumulation curv es constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the lake site....................80 Figure 20. Accumulation curv es constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the upper bay site...........81 Figure 21. Non-metric multidimensional scalin g analysis of similarities in clonal structures from each sample..................................................................................86 Figure 22. Non-metric multidimensional scalin g analysis of similarities in clonal structures from each site.......................................................................................87 Figure 23. Depth profiles of mean enteroco cci densities (presented as decrease in log CFU / 100g from the shallowest dept h) in from sediment cores at the small stream, large stream, and river sites..........................................................113 Figure 24. Depth profiles of mean enteroco cci densities (presented as decrease in log CFU / 100g from the shallowest dept h) in from sediment cores at the lake, upper bay, and lowe bay sites.....................................................................114 Figure 25. Mean proportion of total number of enterococci per square meter landscape area found in water, sediment, and SAV samples at three freshwater sites around the Tampa Bay watershed.............................................116 Figure 26. Mean proportion of total number of enterococci per square meter landscape area found in water, sediment, and SAV samples at two estuarine sites in Tampa Bay..............................................................................117
vii Figure 27. The relative proportion of total enterococci found in water, sediment, and SAV in response to theoretically varying values for water depth and SAV bottom cover at the large stream site.........................................................119 Figure 28. The relative proportion of to tal enterococci f ound in water and sediment in response to theoretically varying water depth as a result of tidal fluctuations at the upper ba y site during the April 2008 sampling event....................................................................................................................121 Figure 29. Predicted increases in waterbor ne enterococci con centrations at the upper bay site resulting from theoreti cally varying values for sediment enterococci densities and the amoun t of sediment resuspended.........................122 Figure 30. Mean flow speed at the river s ite (the only site fo r which historical data were available) for all days between May 2007 and April 2008................125 Figure 31. Frequency histogram of wind di rection (grouped in bins of 10 degrees) for all available wind records at th e port of Tampa between May 2007 and April 2008...........................................................................................................128 Figure 32. Frequency histogram of wind speed (grouped in 1 km/hr categories) for all wind records that were onshore for the upper bay site (between 180 and 270 degrees) between May 2007 and April 2008........................................129 Figure 33. Frequency histogram of wind speed (grouped in 1 km/hr categories) for all wind records that were onshore for the lake site (between 60 and 190 degrees) between May 2007 and April 2008...............................................130
viii The Importance of Benthic Habitats as Reservoi rs of Persistent Fecal Indicator Bacteria Brian D. Badgley ABSTRACT Enterococci are fecal indicator bacteria (FIB) that are used worldwide for water quality assessment. However, evidence of high densities and extended survival of enterococci in sediments and submerged aqua tic vegetation (SAV) has caused uncertainty about their reliability in pr edicting human health risks fr om recreational activities in environmental waters. To address the concer n that sediments and SAV may harbor large reservoirs of enterococci that can affect wa ter column concentrations, aquatic mesocosms and environmental sampling were employed to investigate patterns of enterococci densities and population structure ac ross the Tampa Bay watershed. In mesocosm experiments and environm ental samples, SAV harbored higher densities of enterococci, per ma ss of substrate, than sedime nts, and sediments harbored higher densities than water. Population structure assessed by BOX-PCR genotyping was relatively unique in each sa mple, although slight similarities among samples suggested grouping primarily by location rather than subs trate or season. Strain diversity was highly variable, and many samples had low diversity, including nearly monoclonal structure throughout the mesocosm experiment s and in several of the environmental samples. Several strains were highly abundant and cosmopolitan (found across sites,
ix seasons, and substrates), and may represent highly naturalized and reproducing indicator bacteria populations that are not dire ctly related to pollution events. When the enterococci densities were view ed from the perspective of the entire aquatic system, SAV-associated enterococci did not comprise a major proportion of the total population, due to the typically large diffe rences in volume of each substrate (SAV vs. sediments vs. water). Instead, the largest proportions of entero cocci were typically found in the water or the sediments, depending on the relative volume of substrate or the enterococci density associated with each substrate. Modeling results illustrate that the relative importance of each s ubstrate in terms of FIB populations can shift dramatically over time and space due to changes such as ve getation cover, tidal cycles, and bacteria densities. Furthermore, at several sites w ithin the watershed, estimates of sediment and bacteria resuspension from sedi ments were very low, suggesti ng that this process rarely, if ever, significantly affects water column c oncentrations of entero cocci in the watershed.
1 CHAPTER ONE BACKGROUND AND OVERVIEW OF RESEARCH Our ability to assess the transmission risk of waterborne pathogens that results from fecal contamination in recreational and fishing wa ters is an extremely important tool in protecting public health. It is well established that swimmers and bathers in water with known impacts of sewage or other fecal contamination are at a higher risk for gastrointestinal and respiratory illnesses, as well as skin, ear and eye infections (Cabelli et al., 1982; Cheung et al., 1990; Rees et al., 1998; Wade et al., 2003). Although far fewer epidemiological studies have been done concerning contaminated beach sands, it appears that a similar correla tion exists (Bonilla et al., 2007; Heaney et al., 2009). In the United States, the Clean Water Act (1972) and the Beaches Environmental Assessment and Coastal Health Act (2002), mandate regul ar monitoring of recr eational water quality and public advisories of risks. Unfortunate ly, determining an effective standardized means of detecting and quantifying the human he alth risks that are associated with fecal pollution in all types of environmental wate rs has been a continuous challenge. One reason for the challenge is that it is difficu lt for monitoring agenci es to conduct assays that directly detect pathogens that are in troduced upon sewage contamination. While the direct monitoring for pathogens is ideal in th eory, in practice it beco mes logistically and financially prohibitive due to the wide divers ity of potential pathogens (including viruses,
2 bacteria, and protists) which would have to be monitored, as well as the facts that many pathogens are difficult and costly to culture, have no reliable molecu lar assays, or have patchy distributions or low concentr ations (Field and Samadpour, 2007). Traditionally, this problem has been appr oached by monitoring for fecal indicator bacteria (FIB), or particular bacteria that, although they are not pathogenic themselves, are reliably abundant in feces and sewage. The presence of FIB (at sufficiently high numbers) is assumed to indicate contaminati on of environmental waters by sewage or other fecal material and the likely presen ce of human pathogens. The selection of appropriate FIB has a long history, dating back to original work describing Bacillus coli (now Escherichia coli ) as ubiquitous in human feces by its namesake, Theodor Escherich (Griffin et al., 2001). Soon after, other res earchers described the coliform group (which contains E. coli ) as a suitable indicator of fecal po llution and it was recommended for this purpose in the first edition of the Amer ica Public Health AssociationÂ’s (1905) Standard Methods for the Examinati on of Water and Wastewater Afterwards, the coliforms were the dominant indicator of microbial water quality for decades (Tallon et al., 2005). Today, total coliforms are officially characteri zed as all aerobic and facultative anaerobic, non-spore forming, gram-negative, rod-shaped ba cteria that ferment lactose with gas and acid formation within 48 hours at 35.0 C (APH A, 1998). A subset of the total coliforms, which are capable of growth at 44.5 C and called the fecal, thermotolerant, or thermotrophic coliforms, were later adopt ed as a more suitable indicator.
3 Members of the genus Enterococcus constitute another importa nt group of FIB that has been used for decades. The enterococci are gram-positive cocci that typically occur singly, in pairs, or in short chains. They have a growth range of 10 to 45 C, with an optimum of 35 C, and can grow in solutions containing up to 10% NaCl. Metabolically, they are aerotolerant anaerobes that fermen t carbohydrates to l actic acid and can hydrolyze esculin in the presence of 40% bile salts (Facklam et al ., 2002). Previously, the enterococci were contained within the genus Streptococcus but recent genetic evidence has shown that sufficient difference exists between the two groups to merit a separate genus (Schleifer and Kilpperbalz, 1984). More than 20 species are now included in the genus (Facklam et al., 2002). Entero cocci can be found in the feces of most mammals and birds and, in general, they are not host sp ecific. A couple exceptions include Ent. asini in donkeys (de Vaux et al., 1998) and Ent. columbae in pigeons (Devriese et al., 1990), but it is possible th at these may simply be undersampled. Although the same species can be found in many different hosts, there can be distinct differences in the relative abundances of the di fferent species in differe nt hosts, as well as at different ages within the sa me host (Aarestr up et al., 2002). Given this wide diversity of potential FIB, including others that ha d been proposed such as Clostridium perfringens and coliphages (Fujioka and Shizumura, 1985; Payment and Franco, 1993; Gantzer et al., 1998), the US Environmental Protection Agency (1986) reviewed the performance of va rious indicator organisms and developed a standard set of recommendations for regulatory agencies. In their report, they conc luded that the best
4 indicator in fresh water was E. coli specifically, and in estuarin e and marine waters, they recommended the enterococci. They cauti oned against the use of total or fecal coliforms, as false positives may result be cause of confusion with other groups (e.g., Pseudomonas and Vibrio ) in media based tests (Griffin et al., 2001) or the ability of some members of the coliform group to survive in the environment (discussed in further detail below). Historically, the successful imp lementation of water quality monitoring programs based on FIB has led to dramatic redu ctions in waterborne disease outbreaks in many parts of the world (Leclerc et al., 2001). Furthermore, in a recent meta-analysis of 27 epidemiological studies, Wade et al (2003) concluded that, among the various potential FIB, the highest associations be tween specific FIB and risk of waterborne illness were with the EPA-recommended E. coli in freshwater and enterococci in marine water. Despite the successes of water quality monitoring programs, the use of FIB is far from a perfect solution. Decisions regarding the clos ure of water bodies that may be important for recreation and fisheries have serious ec onomic and public health impacts. Falsenegative conclusions expose swimmers and bath ers to avoidable health risks, and falsepositive conclusions result in unnecessary clos ures that can be economically costly to waterfront communities and i ndustries. One ongoing challenge in the continued use of the indicator approach is searching for and validating the Â‘idealÂ’ FIB. Typical characteristics of an ideal indicator organi sm include its occurrence in correlation with human pathogens, an inability to replicate in the environmen t, ease of identification and
5 quantification, and a correlation between its de nsity and the degree of hazard to public health (Griffin et al., 200 1; Field and Samadpour, 2007; Ishii and Sadowsky, 2008). Unfortunately, recent research has indicated th at these assumptions are often false. In fact, many studies have shown that the pres ence of FIB in envir onmental waters do not correlate well with the pres ence of pathogens, including Salmonella Campylobacter Cryptosporidium Giardia or enteroviruses (Lund, 1996; Bonadonna et al., 2002; Lemarchand and Lebaron, 2003; Harwood et al ., 2005). While there are many potential reasons for this lack of correlation, and it is lik ely to be a combination of several factors, one major problem is the assumption that, subsequent to the introduction of fecal contamination into the environment, FIB w ill exhibit similar survival dynamics as the pathogens they are being used to detect. On the contrary, many recent studies show that FIB Â– both coliforms and enterococci Â– are capab le of persisting in a culturable form for extended periods of time in a wide variety of environmental matrices after their initial introduction (see below). In contrast, pat hogens have often been shown to be very different than FIB in terms of their ability to withstand stressors and survive in the environment (Davies et al., 1995; Lund, 1996; Desmarais et al., 2002; Fujioka and Yoneyama, 2002; Nasser et al., 2007; Englebert et al., 2008). This disconnect between the survival of FIB and pathogens in the environment may undermine one of the key assumptions in the use of wate r quality indicator bacteria. Aquatic sediments have been one of the most widely and thoroughly studied environmental substrates for the persistence of FIB. Once introduced to the water, FIB
6 often quickly attach to suspended particles a nd then settle into the sediments (Auer and Niehaus, 1993; Davies and Bavor, 2000; Je ng et al., 2005). Ironically, however, the author of one of the earliest published account s of FIB in the sediments did not find it surprising that FIB were elevated and likely to be persisting in sediments (Savage, 1905). In fact, he theorized that, because of this theoretically expected phenomenon, sediment samples may offer a more uniform and historic ally accurate record of fecal contamination than single water samples, which, even then, were recognized as being highly variable in time, depending on factors such as tide and w eather. In the several decades following SavageÂ’s work, a few studies revisited sedime nts at various locations and reliably showed that FIB concentrations in sediments were well correlated with the level of pollution (Allen et al., 1953; Ritte nberg et al., 1958; Bonde, 1967; Vandonsel and Geldreich, 1971). In more recent years, however, numerous studies investigati ng the persistence of FIB in sediments have been published from a variety of environmental waters, including tropical and temperate streams (Buckley et al., 1998; Byappa nahalli et al., 2003b), temperate rivers (Tunnicliff and Brickler 1984; Obiri-Danso a nd Jones, 1999), lakes (Doyle et al., 1992; Whitman and Nevers, 2003; Ishii et al., 2007), and subtropical and temperate estuaries (Shiaris et al., 1987; Solo-Gabriele et al., 2000; Desmarais et al., 2002). Much of this recent work has focused on the likelihood that se diments can act as a reservoir of FIB that may represent a signifi cant source of water column concentrations, and several of these studies pr ovide evidence that may be the case at some sites (Crabill et al., 1999; Byappanahalli et al., 2003a; Whitma n and Nevers, 2003; Ishii et al., 2007).
7 Laboratory mesoand microcosms have al so been used extensively to study the persistence of FIB in aquatic sediments. Se veral researchers have found that the addition of sediments to mesocosms containing fres h or saltwater significantly extends the survival of FIB (Vandonsel and Geldreic h, 1971; Hood and Ness, 1982; Burton et al., 1987; Craig et al., 2004; Ande rson et al., 2005). In addi tion, although these groups of bacteria have historically been thought to reproduce only in the intestinal tracts of endothermic animals (Savageau, 1983; Leclerc et al., 2001), in some mesocosm experiments using sterilized se diments, culturable FIB have increased in number (Gerba and McLeod, 1976; Laliberte and Grimes, 1982; Davies et al., 1995; Desmarais et al., 2002). This apparent growth of FIB casts doubt on the validity of the assumptions associated with the use of water quality indicator organisms by suggesting the potential for reproducing populations in aquatic sediments, at least in the abse nce of predation and competition for resources. One specific type of aquatic sedi ment that has received particul ar attention in recent years is shoreline beach sand, specifically that por tion which is not constantly submerged, but is periodically wetted due to wave or tidal act ivity. A number of studies have found that the highest densities of FIB in cross-shore tran sects of beach sands occur in this shoreline or foreshore section of damp sand (Whitman and Nevers, 2003; Kinzelman et al., 2004; Whitman et al., 2006; Beversdorf et al., 2007; Is hii et al., 2007; Yamahara et al., 2007). Solo-Gabrielle et al. (2000) observed a similar pattern in cross-stream transects of a brackish tidal river. The use of experiment ally wetted microcosms has confirmed this
8 effect, resulting in growth of FIB immedi ately following the re-wetting event (SoloGabriele et al., 2000; Yamahara et al., 2009). While this narrow band of sediment may not provide a significant sour ce of FIB on a large scale, at the small scale it may be sufficient to elevate water column concentrations for short periods of time following changes in water level that result from tidal or wave activity (Whitman and Nevers, 2003; Kinzelman et al., 2004; Whitman et al., 2006; Yamahara et al., 2007). In addition to aquatic sediments, high densit ies of FIB can also be associated with submerged aquatic vegetation (SAV). The most well-studied example of this association is the observation of high dens ities and possible growth of E. coli and enterococci on Cladophora a highly abundant green macroalga, in Lake Michigan, USA (Byappanahalli et al., 2003b; Whitman et al., 2003; Olapad e et al., 2006; Engleb ert et al., 2008; Kleinheinz et al., 2009). In a ddition, FIB have been found to be associated with drifting marine macroalgae washed up on beaches in New Zealand (Anderson et al., 1997) and microalgal periphyton in the Great Lake s, USA (Ksoll et al., 2007). Although FIB have also been shown to be associated w ith other aquatic biota, such as plankton (Signoretto et al., 2004; 2005) a nd fish (Del Rio-Rodriquez et al., 2008; Hansen et al., 2008), these sources have not been well studied and it is currently unclear if they have any significant impact on water column concentrations. Hypotheses explaining increased survival in benthic matrices focus mostly on the availability of increased resources and pr otection from environm ental stress. For
9 example, increased survival of FIB in sedime nts has been shown to result from protection of the bacteria from predation by prot ozoans (Flint, 1987; Marino and Gannon, 1991; Davies et al., 1995), as well as protection from ultraviole t light (Davies and Evison, 1991; Fujioka and Yoneyama, 2002; Sinton et al., 200 2). In addition, SAV has been shown to increase available carbon to surrounding se diments both by causing in creased settling of suspended particulates (Posey et al ., 1993; Fonseca, 1996) and by exudation of photosynthate from the roots (Pollard and Mo riarty, 1991). Furthermore, these changes have also been shown to result in increased microbial activity (Lope z et al., 1995; Hansen et al., 2000; Karjalainen et al ., 2001), suggesting that the mi crobial inhabitants of the nearby sediments are able to use the extra resources provided by the SAV. At the larger watershed scale, several acc ounts have been published showing significant densities of culturable FIB in terrestrial matric es as well. Much of the early evidence was collected in tropical soils, where it was assumed that the warm temperatures were conducive to the success of FIB (Hardina and Fujioka, 1991; Roll and Fujioka, 1997; Fujioka et al., 1999; Desmarais et al., 2002). More recently, however, similar results have been observed in temperate soils (Zha i et al., 1995; Byappanahalli et al., 2003a), where some strains have even been found to overwinter in the frozen soils (Ishii et al., 2006). Accounts have even been published of enterococci associated with terrestrial plants (Muller et al., 2001; O tt et al., 2001) and insects (Gel dreich et al., 1964), although the widespread impacts of this association on our ability to predict human health risks is not well understood. Laboratory mesocosm ex periments have provided some evidence
10 that E. coli is capable of growth in terrestrial soils, suggesting that soil may serve not only as a reservoir for persistent bacteria, but even as a source for new cells that may be washed into water bodies during rain even ts (Byappanahalli and Fujioka, 1998; Topp et al., 2003; Ishii et al., 2006) Although soils typically contain highly variable concentrations of coliforms and enteroco cci (both within and among studies), some studies have determined that they may re present a non-point source of FIB that is significant enough to affect conc entrations in the water co lumn (Hardina and Fujioka, 1991; Roll and Fujioka, 1997; Fujioka et al., 1999). The extended persistence of FIB in the sedi ments and SAV of aquatic habitats is one likely cause of the frequently poor correla tion between FIB and pathogens that was discussed above. When FIB persist in benthic matrices, the potential for those cells to be resuspended back into the water column may lead to falsely positive conclusions regarding microbiological contamination. Resuspension might occur during any event that generates significant hydrodynamic activity such as storms, boating and shipping traffic, or high levels of activity by swimmers and bathers. Prokaryotes are typically the most easily resuspended of all benthic organism s, due to their small size and the fact that they are frequently associated with cohesi ve surficial fluff sediments (Gannon et al., 1983; Auer and Niehaus, 1993; Howell et al., 1996; Shimeta et al., 2002). The potential for resuspension is a major concern about the re liability of the indicator organism concept which has been raised repeatedly in the litera ture (Solo-Gabriele et al., 2000; Grant et al., 2001; Whitman et al., 2003; Anderson et al., 2005; Ishii and Sadowsky, 2008).
11 Unfortunately, the importance of long-term environmental persiste nce and resuspension of FIB has been difficult to quantify, in terms of both their association with human health risks and broader ecological effects. A lthough recent epidemiological studies have shown that increased exposure to beach sand ca rries increased risk of disease (Bonilla et al., 2007; Heaney et al., 2009), no correlations have been determ ined between health risks and concentrations of FIB in sediments or S AV. In fact, no standard method has yet been adopted for their detection and quantifica tion in these matrices. Typically, FIB associated with sediment or SAV are disl odged, in some way, so that they can be resuspended in sterile buffered water. This water can then be subjected to the standard membrane filtration methods used for normal water samples (APHA, 1998; USEPA, 2000). However, the methods of dislodging the cells from their substrate have varied widely, including shaking, sonication, and the us e of surfactants (Boe hm et al., 2009). Furthermore, there is no widely accepted mean s of standardizing the densities of benthic FIB, as there is for water column bacteria, which are typically standardized to concentration of colony forming units (C FU)/100 mL. Benthic bacteria are often normalized in some way to the mass of substr ate, such as CFU per 1 g or 100 g wet or dry weight of sediments (SoloGabriele et al., 2000; Desmarai s et al., 2002; Craig et al., 2004; Anderson et al., 2005; Yama hara et al., 2009), but have also been normalized to square centimeter of substrate surface area (Ksoll et al., 2007), volume of interstitial water (Buckley et al., 1998), and square me ter of landscape area (M uirhead et al., 2004; Jamieson et al., 2005).
12 As a result, it is difficult to interpret whether the densitie s reported in the literature actually represent a large reservoir of FIB that are available to be resuspended and affect water column concentrations and, ultimately, water quality monitoring. The resuspension of FIB has been directly observed to occur in managed streams in the absence of rainfall or groundwater inputs, as a re sult of both natural (Nagel s et al., 2002; Jamieson et al., 2005) and experimentally-induced (McDonald et al., 1982; Wilkinson et al., 1995; Nagels et al., 2002; Muirhead et al., 2004) periods of high flow. However, in other environmental water bodies of recreational importance that ar e not so easily constrained (such as beaches and lakes) the resuspension of benthic FIB has typically been inferred. For example, observations of relatively high E. coli concentrations in the water column have been shown to correlate with factors that cause sedi ment motion and resuspension, such as wave or tidal activity (Le Fevre and Lewis, 2003; Shibata et al., 2004; Whitman et al., 2006; Yamahara et al., 2007), dredgi ng or boating activity (Grimes, 1975, 1980; Pettibone et al., 1996; An et al., 2002), or recreational activity (C rabill et al., 1999). Water column concentrations have also been directly correlated with sediment densities through the use of time series or structural equation m odeling (Whitman and Nevers, 2003; Whitman et al., 2006). Modeling the fate and transport of FIB is an increasingly popular focus of current research, and these models can also help elucidate the importance of sediments as a potential reservoir of FIB by incorporating terms for settlin g and resuspension processes
13 and determining the resulting effect on the modelÂ’s predictive power. Through this approach, hydrodynamic information and sediment characteristics can be used to predict sediment resuspension and, in turn, offer re latively good approximations of the behavior of benthic FIB in the sediments (Bai and Lung, 2005; Jamieson et al., 2005). Unidirectional (e.g., tidal and st ream flows) and oscillator y (e.g., wave action) flow regimes both set up velocity gradients near the bottom of the water column that increase from zero at the sediment-water interface up to the mainstream velocity. The steepness of these gradients, in combination with bottom roughness that re sults from bedform elements (e.g., sediment grains, sand ripples, rocks, or organisms) establishes a shear stress that acts on the top layer of sedime nt (Denny, 1988; Soulsby, 1998). If the force of this shear stress is sufficiently strong to overcome the natural se ttling velocity of individual sediment grains, then some am ount of sediment will be resuspended and maintained in suspension (Soulsby, 1998; Le Roux, 2005). Many sets of theoretical and empirical equations exist that allow the pred iction of concentrations and transport of suspended sediment under a given set of unidir ectional or oscillatory flow regimes, which can then be used to estimate resuspensi on of the associated FIB (Bai and Lung, 2005; Jamieson et al., 2005). General terms for resuspension rates (typically based on the shear stresses and sediment qualities described above) have been incorporated into emba yment-wide models used to predict net transport of FIB (Steets and Holden, 2003; Sanders et al., 2005). In addition, a much more focused study has been publis hed that uses the Environmental Fluid
14 Dynamics Code model to specifically predict the impacts of sediment association on the settling, resuspension, transport, and persis tence of FIB (Bai and Lung, 2005). These models, however, necessarily re ly on many assumptions and ge neralities with regards to the ecology of FIB and sediment/bacteria resuspension dynamics. It is becoming increasingly clear that improving our unders tanding of how bent hic-pelagic coupling affects the population dynamics of FIB speci es is an important step towards improving our ability to predict their survival and tr ansport in environmental waters. In fact, additional data and experimentation on the beha vior of benthic FIB ha s been outlined as a distinct need for future model improvement (Bai and Lung, 2005; Pachepsky et al., 2006) It is important to note here that bacteria a ttached to SAV would behave very differently than what was described above for sediments. SAV-associated bacteria would only be returned to the water column as a result of some sort of mechan ical shearing of the bacterial cells, or the partic les to which they are attached (e.g., detritus or epiphytic algae), from the vegetative surface. In ei ther case, these dynamics are probably quite complex and different from sediment resu spension theory. To my knowledge, these processes have never been examined expe rimentally or modeled. In addition, the structural presence of the macrophytes in ve getated habitats alters the hydrodynamics themselves, by interacting with the overlying flow to create increased turbulence in the water column and in the upper levels of the submerged canopy, but greatly decreased flows within the canopy near the sediment (Gambi et al., 1990; Ikeda and Kanazawa, 1996; Ghisalberti and Nepf, 2002). This alte red flow can result in much lower shear
15 stresses at the sediment-water interface a nd probably also reduce bacterial resuspension into the water column as compared to similar flows over unvegetated sediments. This effect, coupled with potential for vegetated canopies to supply si gnificant amounts of nutrients to support bacterial gr owth, suggests that modeling th e fate and transport of FIB in highly vegetated habitats may require enti rely new model development to accurately predict these processes. Whether they are attached to sediment or ve getation, one of the major deterrents to easily interpreting the importance of be nthic reservoirs of FIB is th at indicator densities have typically been normalized per unit volume fo r water samples, but have been normalized per unit mass of substrate for sediment and SAV samples (Byappanahalli and Fujioka, 1998; Solo-Gabriele et al., 2000; Topp et al., 2003; Whitman et al., 2003; Jeng et al., 2005; Ishii et al., 2006; Ksoll et al., 2007). Normaliza tion by volume makes sense for the water column, as concentration is the most appropriate value to consider in terms of the exposure and health risks posed to swimme rs. However, when one tries to compare these values to data that are normalized to mass of sediment or SAV, the two different approaches do not offer an equal measure of water column and benthic densities, nor do they allow for a simple interpretation of the importance of benthic sources of resuspendable bacteria. One approach that bypasses this shortcoming, however, is to use a different method of normalizing bacterial de nsities and look at aquatic systems on the basis of landscape area (e.g., per m2). When bacterial densities are integrated vertically (e.g., by depth of the water column or sedime nts) to obtain total CFU per unit landscape
16 area, the resulting densities allow for direct comparison of bacter ial population sizes in water column and benthic habitats within or between water bodies (Muirhead et al., 2004; Jamieson et al., 2005). The consequences of persistent FIB in sec ondary habitats are further complicated by the effects of inherent diversity among, and even wi thin, the various species of FIB. Groups of FIB, such as the enterococci and fecal co liforms, harbor interspe cific variability among their member species. Within the coliforms, for example, E. coli typically displays a much higher association with sewage and hum an fecal material, while other members of the group such as Enterobacter spp. and Klebsiella spp., are widespread in the environment and therefore poorer predictors of fecal contamination (Leclerc et al., 2001). This finding was a major factor in prompti ng the U.S. Environmental Protection Agency to promote a shift from the use of fecal coliforms to E. coli as the preferre d water quality indicator in freshwater habitats (USEPA, 1986). Similar interspecific variability has been seen within the en terococci, although the entire group is still used as the preferred water quali ty indicator in marine waters. Often, in environmental samples, the dominant species tend to be Ent. faecalis Ent. faecium Ent. hirae and Ent. mundtii (Pinto et al., 1999; Harwood et al., 2004; Ferguson et al., 2005; Moore et al., 2008). However, among thes e species there are wide differences in associations. For example, Ent. faecalis and Ent. faecium are dominant in human feces and sewage (Ruoff et al., 1990; Manero et al., 2002; Gelsomino et al., 2003), while
17 pigmented species such as Ent. casseliflavus and Ent. mundtii are rarely associated with human sources and considered to have e nvironmental sources such as plants and waterfowl (Leclerc et al., 1996; Pinto et al., 1999; Aarestrup et al., 2002). The diversity of each group complicates their survival dynami cs and their performance as an indicator, and must be taken into account in order to improve their utility. In addition to the effects of inter specific variability, additional complexity in evaluating indicator performance results from the fact th at each individual species can also exhibit considerable intra specific variability as a result of the clonal diversity i nherent in a given population. In microbiology, the species concep t is typically agreed to be functional from an operational perspective, but the lack of sexual repr oduction and the existence of many mechanisms for genetic exchange across prokaryotic taxa make it very difficult to firmly ground the concept in theory (R ossello-Mora and Amann, 2001; Oren, 2004). Historically, microbiologists were limited to identifying microbial species by phenotypic characteristics such as morphology, physiology, and culture conditions, usually by working with isolated pure cu ltures. However, with the a dvent of molecular techniques, new parameters for species delineation have been based on genomic information, which carries the benefit of bei ng culture independent. The two most common molecular indicators of species delineation are a 70% or greater DNA-DNA hybridization for the total genome and 97% or greate r similarity of 16s rDNA gene sequence (largely because it correlates most closely w ith a 70% hybridization rate). Although the new molecular methods of taxonomy have proven to be very us eful, they are not foolproof, and it is now
18 generally agreed that using a polyphasic approach, wh ere phenotypic characteristics are used in conjunction with genomic informa tion, is the best means of determining prokaryotic relations (Stackeb randt et al., 2002). Howeve r, even with a polyphasic approach, it is clear that the criteria delineate a very broad prokaryotic species definition, especially in comparison to eukaryotes, and that intraspecific variation among cells can be quite high for important characteristics such as antibiotic resistan ce, virulence factors, and physiology (Ward et al., 1998). For FIB, this inherent variation within speci es has been a major focus of microbial source tracking (MST), which is an area of activ e research that attempts to overcome the limitations of using only the concentrations of FIB to determine risks associated with fecal pollution. The goal of MST is to disti nguish contamination that originates from various fecal sources (e.g., huma n, agricultural, or wildlife), thereby offering a means of determining when high concentrations of FI B are truly representative of human fecal pollution and pose increased health risks (Simpson et al., 2002; Field and Samadpour, 2007; Stoeckel and Harwood, 2007). A wide range of methods have been employed to differentiate between cont amination sources, including both library-dependent and library-independent approaches that try to iden tify particular microbi al strains or target genes that are specific to, or at least highly associated with, waste from particular host species (Bernhard and Field, 2000; Whitlock et al., 2002; Field et al., 2003; Seurinck et al., 2003; McQuaig et al., 2006; Shanks et al., 2009). Although many of these methods have been successfully used to determine sources of fecal FIB in recent years, the
19 implementation and interpretation of thes e methods in the environment may be complicated by the complex population dynamics and variable persistence of FIB in natural environments. Many methods have been used to differentiate between strains of FIB. Two methods that were widely used in earlie r library-dependent source track ing studies include ribotyping (Parveen et al., 1997; 1999; Ca rson et al., 2001) and antibiotic resistance analysis (ARA) (Wiggins et al., 1999; Harwood et al., 2000; Gaun et al., 2002; Whitlock et al., 2002). By creating libraries of ribotype and antibiotic re sistance profiles for nu merous strains from known sources, researchers were able to take unknown strains from environmental samples, compare them to the reference libra ry, and classify them according to source with varying degrees of success. In additi on to these two methods a number of others have been suggested, including profiles of carbon source ut ilization (Hagedorn et al., 2003), sequencing of the 16s-23s intergenic space r region (Seurinck et al., 2003), or sequencing of the -glucoronidase gene (Ram et al., 2004). The typing method that has probably been us ed the most widely and successfully, however, is repetitive sequence-based PCR (re p-PCR, also known as repetitive extragenic palindromic PCR). This method uses conven tional PCR techniques that target repetitive elements in the bacterial genome such as duplicated genes, insertion elements, transposons, and mosaic repetitive elem ents (Ishii and Sadowsky, 2009). Using total genomic DNA as a template, the amplification of these elements generates multiple
20 amplicons of various sizes that relate to the distances between each of the repeating elements along the genome, and the separa tion of those amplicons on a traditional agarose gel creates unique banding Â‘fingerprintsÂ’ for each strain. Primers targeting many different types of elements can be used, with REP, BOX, and ERIC elements being commonly used for studies of FIB (Koeuth et al., 1995; Malathum et al., 1998; Dombek et al., 2000; McLellan and Salmore, 2003; T opp et al., 2003). When compared to other techniques, rep-PCR offers clear advantages in terms of higher resolution and sample throughput (Dombek et al., 2000; Ishii and Sa dowsky, 2009), and has been shown to be reliably stable over tim e (Seurinck et al., 2003). Regardless of the method used to differ entiate strains, however, it is becoming increasingly clear that clona l diversity is an important factor in determining the population structure and dynamics of fecal FIB. Clonal diversity has been shown to be lower in environmental samples than sour ce samples (Gordon et al., 2002; Brownell et al., 2007), suggesting that select ion and changing clonal stru cture are occurring over time after the introduction of FIB to the environment. This process has also been observed directly in mesocosm experiment s, where different strains of E. coli have exhibited differential survival in environmental wa ter and sediment (Ande rson et al., 2005) and soils (Topp et al., 2003), with some strains disappearing from the population before others. In addition, high a bundances of strains not associ ated with any known source have been observed at a variety of sites, suggesting that some strains may not only survive longer than others, but may also be adapted to continued persistence or even
21 growth in environmental hab itats. For example, presumab ly naturalized strains of E. coli have been found in temperate soils (Ishii et al., 2006) and freshwater beach sands (McLellan, 2004). Similarly, extremel y high density blooms of pelagic E. coli in Australia were shown to be mostly compri sed of three different strains, even in geographically distant lakes (P ower et al., 2005). Evidence for differential survival and naturalized strains implies a hi gh level of intraspecific physiol ogical diversity that affects persistence in natural habitats, resulti ng in changing clonal structure and complex population dynamics that confound our ability to link FIB to potential sources. Furthermore, naturalized st rains have strong potential fo r decoupling any correlation between FIB concentrations and pathogen presen ce, raising concerns over the utility of the indicator paradigm. Improving our understanding of these complex population dynamics is a critical need in continuing to improve water quality monitoring and MST efforts, and also to improve our understan ding of the species concept in microbiology.
22 Research Goals and Chapter Objectives The primary goals of my doctoral work are to determine the extent to which sediments and SAV may be serving as re servoirs of persistent en terococci in the Tampa Bay watershed and to estimate the potential for resuspension of enterococci from these reservoirs to impact their utility as an indi cator of water quality. On a broader level, I hope that by investigating the persistence and distribu tion of various strains of Enterococcus I can contribute additional insight into how intraspecific variability in microbial species effects the population dynamics of the species as a whole. In addition, I hope that by looking at resuspension of b acteria from vegetated habitats I hope to provide more insight into the role that physical forces and benthic-pelagic coupling play in aquatic microbial ecology. The description, results, and discussion of the original research in my doctoral work are outlined in Chapters Two through Four. In Chapter Two, I employed experimental mesocosms to determine the effects of SAV on enterococci persistence in realistic aquatic habitats that include three different substrates Â– wate r, sediment, and SAV Â– under controlled conditions. Specificall y, my three objectives were: 1) Examine the persistence dynamics of environmental ente rococci populations (as opposed to inoculated la boratory strains) on different substrates in realistic aquatic mesocosms; 2) Experimentally determine the effect of SAV on the persistence of enterococci by simultaneously comparing enterococci persistence in paired vegetated and unvegetated mesocosms; and
23 3) Determine the effects of SAV on the population structure of enterococci by examining shifts in species a nd strain diversity over time. In Chapter Three, I simultaneously monitored concentrations and popul ation structures of enterococci in water, sediment, and SAV at several sites across the Tampa Bay watershed over an entire year. The goals of the study were four-fold: 1) Determine if high densities of benthi c enterococci occurred regularly and consistently across a variety of site s and substrates in the watershed; 2) Investigate the potential for any spatia l or temporal patterns in the water column or benthic densities; 3) Employ molecular fingerprinting te chniques to inve stigate how the Enterococcus population structure and strain diversity varies over space and time; 4) Look for any evidence of widespread or co smopolitan strains th at appear to be adapted to the environment. In Chapter Four I revisited the same sites in the Tampa Bay watershed that were sampled in Chapter 3 and I used the concept of landscape area to reexamine the relative population sizes of the enterococci found in the water, sediment, and SAV. Furthermore, I investigated the theoretical potential for be nthic substrates, such as sediment and SAV, to serve as important reservoirs of resusp endable FIB, as is often suggested in the literature. Specifically, the st udy had three important goals: 1) Identify and quantify key habitat ch aracteristics that would allow the normalization of enterococci densitie s on a landscape basis and directly compare the population sizes in water, sediment, and SAV at each site; 2) Develop a model that predicts shifts in the relative population sizes at a given site that result from theoretical change s in important habitat characteristics such as bacterial densities, wa ter depth, SAV cover, etc.;
24 3) Use historical wind and flow data at each site, in conjunction with theoretical calculations of sediment resuspension, to determine the likely effect of sediment-associated bacterial resusp ension on water quality monitoring at each site. Significance of Research Through the objectives outlined above, I aim to improve our understanding of the ecological dynamics of enterococci in the environment. This will have the applied benefit of improving our knowledge of how th e enterococci function as indicators of fecal pollution and, hopefully, how their use in this fashion can better predict the human health risks associated with waterborne disease. From th e standpoint of basic science, these results offer further insight into the ability of microbial species to adapt to new habitats and environments and the role th at clonal diversity plays in determining microbial diversity. In addition, by determini ng the relative reservoi r sizes and potential resuspension of benthic bacter ia, we gain a better understand ing of the role of physical processes and benthic-pelagic coupling in structuring microbi al ecology and affecting the fate and transport of microbial species in the environment. My investigation of enterococc i associated with environmental water, sediment, and SAV represents the first account of simultaneous ly monitoring the role that these three important environmental matrices have on th e persistence of FIB. The use of this approach in both mesocosms and field samp ling allows direct comparisons of their importance as a substrate, as well as the obs ervation of potential i ndirect effects among the different substrates from the perspective of the entire aquatic system. It also allows
25 for a more complete picture of the clonal di versity of environmental enterococci present at each of the sites. In addition, my app lication of the landscape scale for normalizing bacterial densities greatly improves our abi lity to directly compare the importance of environmental reservoirs of water quality indi cator bacteria. The us e of this approach allows targeted focus on only those reservoirs that have the pote ntial to significantly affect water column concentrations, and the model developed in Chapter Four provides a means of estimating how the population sizes in each reservoir may change as habitat characteristics change. Finally, my estimates of resuspension of FIB at select sites illustrates that bacterial densities normalized to mass of substrate are not sufficient to determine when resuspension of benthic sources of FIB may affect wa ter quality, and that additional weather and habitat char acteristics must also be known.
26 CHAPTER TWO THE EFFECTS OF SUBMERGED AQUATIC VEGETATION ON THE PERSISTENCE OF ENVIRONM ENTAL POPULATIONS OF ENTEROCOCCUS SPP. Introduction Most water quality monitoring strategies are based on the measurement of indicator organisms Â– microorganisms that indicate f ecal pollution and thus the potential for the presence of waterborne pathoge ns Â– as opposed to monitoring for individual pathogens. Bacteria belonging to the genus Enterococcus are one of the major groups used as such an indicator in many monitoring programs fo r recreational water quality (USEPA, 2000; WHO, 2001). Recent research, however, ha s provided evidence that some FIB are capable of persisting in a culturable form for extended periods in the sediments and submerged aquatic vegetation (SAV) of many secondary environmental habitats (Byappanahalli et al., 2003a; Cr aig et al., 2004; Anderson et al., 2005; Ishii et al., 2006; Englebert et al., 2008). The pe rsistence of benthic FIB (not e that we use benthic to describe bacteria associated with the bottom of aquatic habitats, in cluding sediment and vegetation, as opposed to those suspended in th e water column) is pa rticularly important because resuspension of those cells back into the water co lumn, such as might occur during storms or high recreational activity, ma y provide an erroneous signal of recent fecal contamination. If the persistence of FIB exceeds that of most pathogens in secondary habitats, elevated levels of resi dual FIB can trigger a false alarm of human
27 health risk, which has been a major concern ra ised about the reliability of the indicator organism concept (Solo-Gabriele et al., 2000; Grant et al., 2001; Whitman et al., 2003; Anderson et al., 2005; Is hii and Sadowsky, 2008). Evidence of persistent populat ions of culturable FIB has b een found in terrestrial soils (Topp et al., 2003; Ishii et al., 2006), aquatic sediments (Byappanahalli and Fujioka, 1998; Solo-Gabriele et al., 2000; Jeng et al., 2005), and attached to SAV (Anderson et al., 1997; Whitman et al., 2003; Ksoll et al., 2007 ; Kleinheinz et al., 2009). Increased environmental persistence of FIB has been experimentally correlated with decreased temperature, salinity, and sola r radiation (Davies and Evis on, 1991; Howell et al., 1996; Anderson et al., 2005). Field samples and laboratory mesocosms (using inoculated cultures) have been used to show increased persistence of cells associated with benthic sediments or SAV as compared to cells susp ended in the water column, presumably due, at least in part, to the incr eased supply of organic carbon and protection from ultraviolet radiation afforded by these habitats (She rer et al., 1992; Desm arais et al., 2002; Byappanahalli et al., 2003a; Crai g et al., 2004; Anderson et al ., 2005; Ksoll et al., 2007). To my knowledge, however, the controlled experimental setting represented by mesocosms has rarely been used to investig ate the persistence of unaltered environmental populations of FIB (Desmarais et al., 2002) and has never been used to simultaneously compare the suitability of water, sediment and SAV as refuges for persistent populations of FIB.
28 The consequences of persistent FIB in sec ondary habitats are further complicated by the effects of intragroup diversity. This diversity can allow for differential survival of certain members of FIB populations in a given set of habitat conditions (Topp et al., 2003; Anderson et al., 2005; Ishii et al., 2006), which complicates their population dynamics. Furthermore, it confounds our ability to link them to sources and potentially creates a disconnect between FIB concentration and path ogen presence in environmental waters. Many methods have been used to differentia te strains of FIB, including ribotyping and antibiotic resistance pa tterns (Parveen et al., 1997; Ha rwood et al., 2000; Anderson et al., 2006) as well as amplification of repe titive DNA sequences (Dombek et al., 2000; Johnson et al., 2004; Ishii et al., 2006), in orde r to investigate the influence of population structure on the ecology of persistent FIB. In this study, experimental mesocosms w ith continuous unidirec tional current were employed to investigate the persistence of environmental enteroco cci populations under controlled conditions over a 14 day period. Specifically, ther e were three objectives: (1) to simultaneously examine the persisten ce dynamics of environmental enterococci populations (as opposed to inocul ated laboratory or sewage st rains) in the water and on sediments and SAV; (2) to experimental ly determine the effect of SAV on the persistence of enterococci by simultaneously comparing enterococci persistence in paired vegetated and unvegetated meso cosms; and (3) to determin e the effects of SAV on the population structure of enterococci by examini ng shifts in species and strain diversity over time.
29 Methods Experimental Mesocosms Eight mesocosm experiments were conducted in total and were organized as four paired trials (May, July, and August of 2007 a nd March of 2008), with a vegetated and unvegetated replicate run side by side. Each pair of mesocosms was maintained and monitored for a two week period in an openair greenhouse at the Un iversity of South Florida Botanical Gardens, allowing exposur e to ambient air temperatures and solar radiation, but protection from rainfall. The mesocosms we re of a recirculating racetrack flume design, using an electric trolling moto r housed in a drop box at one end to power constant water flow under relatively controlled hydrodynamic conditions ( Figure 1 ). Such water movement has been found to be important to avoid anoxic conditions in the mesocosms (Harwood, unpublished data), and s imulates a flowing water body. Because I originally intended to examine the effect of water velocity on enterococci persistence, the May and August experiments were run at a high er water velocity (12 cm/s) than the April and July experiments (2 cm/s). However, due to the logistical difficulty of achieving high levels of replication with the experime ntal setup, and the fact that all measured physical conditions (i.e. dissolved oxygen, temp erature, pH, and turbidity) in high water velocity vs. low water velocity mesocosms were similar, I pooled data from all experiments for analysis. It is important to note, however, that I am not claiming that hydrodynamics do not have important biological or physicochemical effects Â– merely that this study, as conducted, was unable to detect any such responses.
30 Figure 1 Schematic representation of the recirc ulating flume (~180L) that was used to created vegetated and unvegeta ted mesocosms for flow experiments. Top panel is side view and bottom panel is top view; gray sh ading represents areas with sediment and arrows indicate direction of flow.
31 Water, sediment, and SAV (mostly Hydrilla verticillata ) with chronically high enterococci levels were collected from a fres hwater lake in Tampa, FL, USA and used to seed the mesocosms without any alteration, so that natural populat ions of enterococci would be included in the experiments. E ach substrate was collected and transported independently from the other in sterile c ontainers to avoid cross-contamination. Sediment and SAV were briefl y drained of superficial wate r at the collection site and transported damp to avoid changing densitie s due to desiccation or dilution into water during transport. During mesocosm set up, individual shoots of SAV (10-20 cm in length) were placed into two plexiglass plates that had b een drilled to accommodate approximately 600 shoots/m2, and one plate was placed in each side of the vegetate mesocosm to create a thick canopy. Next, sediment was gently added by hand to the working section of each mesocosm to a dept h of approximately 2 cm and care was taken to avoid burying the SAV shoots. Water wa s added to the non-work ing section of each mesocosm and slowly filled to a depth of a bout 12cm. Sediment was not resuspended in the mesocosms as a result of filling or the flow generated by the motors, so that mixing of enterococci due to experimental effects would be minimized. The mesocosms were established and the experiments began within two hours of collecti ng the material from the lake. Bacterial Concentrations Water, sediment, and SAV were sampled prior to collection from the lake, approximately two hours after the establishm ent of the mesocosms, and then 1, 2, 3, 4, 8, and 14 days
32 from start of the experiment. All samples (e xcept the first set taken after the mesocosm setup) were collected between the times of 0900 and 1100. Triplicate samples of 250 mL water and 25 g sediment were collected aseptically from each mesocosm along with triplicate samples of 25 g S AV from the vegetated mesoco sms, and were immediately placed on ice and processed in the laboratory within 4 h. The number of colony forming units (CFU) of enterococci was quantified vi a membrane filtration. Water samples (1, 10, or 100 mL of each triplicate sample) were co ncentrated by vacuum filtration directly onto 47 mm nitrocellulose membranes (F isher Scientific, Inc.) with a 0.45 m pore size and cultured at 41 C for 24 hours on mEI ag ar (Difco Laboratories) (USEPA, 2000). For sediment and SAV samples, 10 g (wet wei ght) of material was diluted in 100 ml of sterile buffered water (0.0425 g/L KH2PO4 and 0.4055 g/L MgCl2; pH = 7.2) and sonicated on ice at 14 watts for 30 seconds to dislodge and resuspend attached cells (Anderson et al 2005). Aliquots (10 or 25 mL) of the supernatant were then filtered and cultured as above. Concentrati ons are presented as CFU/100 mL for water or CFU/100 g wet weight substrate for sediment and SAV sa mples. For samples in which the analyte was not detectable, one half of the limit of detection (LOD) was used as the concentration for the purpose of presentation and statistical analysis. Limits of detection equaled 0.3 CFU/100 mL in water samples or 13 CFU/100 g in sediment and SAV samples. After counting, well-isolated colonies were picked from the mEI agar (up to a maximum of 32 isolates were saved for each substrate at each time point) and cultured overnight in Enterococcosel broth (EB, Difco Laboratories, enc.) at 37 to confirm esculin hydrolysis.
33 Glycerol was added (10% v/v) to cultures pr ior to storage at -80 C for later genetic typing. Enterococcus Genotyping Isolates from each substrate at the beginning of every experiment were selected for BOXPCR genotyping to determine the initial population structure for all of the mesocosms. In addition, one time point was selected near th e end of each experiment in the vegetated mesocosm for genotyping to de termine if the population st ructure was changing. These points were day 14 in May and day 5 in Ju ly, Aug, and March (only sediment and SAV available from the latter three). For the Ma y experiments, 15 isolates were typed from each substrate at the initial time point. Ho wever, after discovering the extremely low strain diversity (see results a nd discussion), this number wa s reduced to 6 isolates per substrate per sample to confirm that the popul ation structure had not changed, for a total of 153 isolates across al l of the experiments. Cryopreserved isolates were str eaked onto tryptic soy agar to ensure isolation of a pure culture. Isolates were then grown overnight in 1 mL of brain hear t infusion broth (BHI, Difco Laboratories) at 37 C and DNA was extracted using the GenElute Bacterial Genomic DNA kit (Sigma-Aldrich) per the manufacturerÂ’s instructions for Gram-positive bacteria. The DNA of each isolate was then typed by repetitive extragenic palindromic (REP) PCR fingerprinting us ing the BOX A2R primer (5Â’-ACG TGG TTT GAA GAG ATT TTC G-3Â’) (Koeuth et al., 1995). Twenty five L PCR reactions contained 5 L of
34 5 Gitschier Buffer (Kogan et al., 1987), 2.5 L of 10% dimethyl sulfoxide, 0.4 L bovine serum albumin (10 mg/mL), 2.0 L 10 mM dNTPs, 1.0 L Taq polymerase (5000 u/mL), 10.6 L water, 1.5 L 10 M BOXA2R primer; and 2.0 L of DNA template, containing between 10 and 40 ng/ L of DNA (Versalovic et al., 1991; Malathum et al., 1998). The PCR program included (1) initial denaturation at 95 C for 7 min; (2) 35 cycles of 90 C for 30 s, 40 C for 1 min, and 65 C for 8 min; and (3) final extension at 65 C for 16 min. The amplicons were separa ted on a 1.5 % agarose gel (90 watts for 4 hrs), stained with ethidium bromide (1% so lution) and imaged under UV light. A strain of Ent. faecalis isolated from a previous study (A nderson et al., 2005) was chosen as a PCR control because its banding pattern show ed even dispersion across the molecular weight range used for analysis (250 Â– 3000 bp). Banding patterns were analyzed for similarity with BioNumerics 4.0 software (A pplied Maths, Inc., Belgium) and confirmed by eye. Similarity was determined from Pearson correlations based upon the densitometric curves (optimization = 1%) fo r each genetic type and a dendrogram was constructed via UPGMA. Identical reactions of a control strain maintained a similarity of 84%, which was used as a critical value to establish which environmental strains were similar enough to be considered identical and the results were confirmed by eye. Taxonomic Identification of Isolates Two isolates of the dominant BOX-PCR ge notype (see results) were identified by sequencing the 16S rRNA gene. A 1,145 bp frag ment of the gene was amplified from extracted DNA via PCR using the universa l bacterial primers Eco8f (5Â’-AGA GTT TGA
35 TCM TGG CTC AG Â– 3Â’) and ECO1492RC (5'-GGT TAC CTT GTT ACG ACT T-3') (Lane, 1991). Fifty microliter PCR reactions contained 25 L JumpStart Taq polymerase (Sigma, USA), 2 L each primer, 5 L water, and 5 L DNA template. The ampli cation process included (1) initial denaturation at 94 C for 5 min; and (2) 20 cycles of 94 C for 1 min, 55 C for 1 min, and 72 C for 10 min. Amplicons were frozen and sent to a commercial laboratory for seque ncing (Macrogen, Inc., USA). Because the results of the sequencing did not offer resolution between the highly similar Ent. casseliflavus and Ent. gallinarum species, colonies were examined for the yellow pigmentation that is characteristic of Ent. casseliflavus. Calculations and Statistical Analysis All CFU data were transformed as log10 ( x + 1) to meet normality requirements prior to statistical analysis. Throughout the manuscript, the terms Â‘concentratio nÂ’ or Â‘densityÂ’ are used to define the CFU per unit volume of wa ter or mass of substrate (sediment or SAV), while the term Â‘total numberÂ’ is defined as the total number of CFU per mesocosm. Means of either value were calculated as th e mean and standard deviations of log CFU for all time points (n = 8) in a given meso cosm experiment. The total number of CFU associated with each substrate in each mesocosm was calculated by multiplying the enterococci density (CFU/100 ml or CFU/100 g) times the total water volume or the total mass of sediment or SAV in each mesocosm at each time point. Water volume and substrate mass in each mesocosm was determined at the end of each set of experiments, and corrections were made to account for the amount of material removed at each time
36 point for sample analysis. Finally, the to tal number of CFU per mesocosm at each time point was calculated simply as the sum of the total numbers of CFU per mesocosm associated with each substrate. Due to the pattern of the survival dynam ics observed in our study, which typically included a period of decline in density of en terococci over the first few days, followed by an extended period of persiste nce, two different types of decay rates were calculated. Initial decay rates for the populations in each substrate were calculated by regressing the log CFU against the sample time for each expe riment. Only those data up to the first sample in the time series with a non-detect for enterococci were us ed. Cumulative decay rates were calculated in the same manner, but using all time points for each experiment. The slope of the regression represents the de cay rate and is reported as change in log CFU/d (Davies and Evison, 1991; Craig et al., 2004). A negative decay rate represents a decrease in CFU, while a positive rate re presents an increase in CFU. Minimum persistence times are reported as the latest da y on which detectable levels of enterococci were found in a particular substrate (these va lues were converted to categorical data and analyzed non-parametrically with a Kruskal-Wallis H -test due to the lack of sample data available on some days). Values were co mpared statistically using randomized block analyses of variance and paired t -tests (SPSS, version 17.0, SPSS, Inc., USA; = 0.05).
37 Results Mesocosm Conditions Water temperature in the mesocosms was measured at each sampling during the experiment and ranged from lows of 18 C in March to highs of 30 C in July and August. The mean temperature at sampling in the March experiments (21.0 C) was significantly lower than the other three months (May = 24.8 C, July = 26.8 C, and August = 26.9 C; p <0.001, ANOVA and TukeyÂ’s pos t-hoc comparisons). Water chemistry stayed relatively constant in al l experiments with dissolved oxygen values from 9.5 to 10.5 mg/L, pH values from 7.5 to 8.5, and salinity values from 0.25 to 0.35 Â‰. Visible resuspension of sediments did not occur. The mesocosms were of a recirculating flume design, were approximately 1 m x 1.5 m in area, 15 cm deep, and held approximately 180 L (see Methods and Figure 1 ). Water current was controlled with submerged trolling motors so that there was a continuous unidirecti onal current over the substratum. Population Dynamics The population dynamics of culturable enteroco cci obtained from a fr eshwater lake over a two week period were highly variable among replicate pairs of e xperiments, each of which included one vegetated and one unvege tated mesocosm. In some form, however, evidence for extended persistence of culturable enterococci was exhibited in each of the four replicates of paired mesocosms. The May 2007 experiments exhibited very highlevel, consistent persisten ce. Enterococci densities we re initially high (water = 103
38 CFU/100 mL; sediment = 1.3 x 103 CFU/100 g; SAV = 6.3 x 104 CFU/100 g), decreased rapidly during the first day, rebounded dramatica lly at about day 5 and then continued to increase throughout the two week period ( Figure 2 ). The July 2007 experiments had much lower initial densities (approximately 102 CFU/100 mL in water and sediment and 103 CFU/100 g in SAV), and enterococci densit ies on all substrates varied over 2 Â– 3 orders of magnitude ( Figure 3 ). The August 2007 and March 2008 experiments were inbetween these extremes in terms of cell densities and consiste ncy of detection ( Figure 4 and Figure 5 ). Effects of SAV Because the enterococci measured in this st udy were obtained from a lake with a history of a persistent enterococci population, a nd because the population dynamics in the mesocosms were highly variable, the mean conc entrations of enterococci over the entire 14-day experiments were compared by a paired t -test. Mean numbers of total enterococci in vegetated vs. unvegetated mesocosms we re compared, where each data pair represented the two mesocosms sampled in one month. From the perspe ctive of the entire system, vegetated mesocosms maintained si gnificantly higher mean numbers of total enterococci (SAV+sediments+water) (9.4 x 104 CFU per mesocosm) than unvegetated mesocosms (sediments+water) (2.6 x 104 CFU per mesocosm; p = 0.05, paired t -test) when averaged over the 14-day period for each experiment, a 250% increase. It is important to note that this difference was not explained by the extra bacteria initially
39 A: B: Figure 2. Culturable enterococci concentrations in unvegetated (A) and vegetated (B) mesocosms for the April experiments. Time 0 represents samples from the source lake taken prior to mesocosm setup. Each point is the mean of three replicates s.d. 0 1 2 3 4 5 6 02468101214time (d)log CFU / 100mL or 100g 0 1 2 3 4 5 6 02468101214time (d) Water Sediment SAV
40 A: B: Figure 3. Culturable enterococci concentrations in unvegetated (A) and vegetate d (B) mesocosms for the May experiments. Time 0 represents samples from the source lake taken prior to mesocosm setup. Each point is the mean of three replicates s.d. 0 1 2 3 4 5 6 02468101214time (d)log CFU / 100mL or 100g 0 1 2 3 4 5 6 02468101214time (d) Water Sediment SAV
41 A: Figure 4. Culturable enterococci concentrations in unvegetated (A) and vegetated (B) mesocosms for the July experiments. Time 0 represents samples from the source lake taken prior to mesocosm setup. Each point is the mean of three replicates s.d. 0 1 2 3 4 5 6 02468101214time (d) Water Sediment SAV 0 1 2 3 4 5 6 02468101214time (d)log CFU / 100mL or 100g
42 A: B: Figure 5. Culturable enterococci concentrations in unvegetated (A) and vegetate d (B) mesocosms for the August experiments. Time 0 represents samples from the source lake taken prior to mesocosm setup. Each point is the mean of three replicates s.d. 0 1 2 3 4 5 6 02468101214time (d)log CFU / 100mL or 100g 0 1 2 3 4 5 6 02468101214time (d) Water Sediment SAV
43 introduced into the vegetated mesocosms with the SAV, which caused only a 10% increase over unvegetated mesocosms. Furthe rmore, when enterococci concentrations were normalized to the initial concentration measured in the mesocosms, the differences between treatments were still significant. When paired t -tests were run on a month by month basis (separate tests run for each month, samples paired by time), significant differences were seen in April, Ma y, and July, but not in August ( p = 0.04, 0.001, 0.002, and 0.15, respectively). When the enterococci densities were compared by substr ate, SAV harbored significantly higher mean densities on a per mass basis (8.6 x 102 CFU/100 g) than sediments (1.3 x 102 CFU/100 g), which, in turn, had significantly higher mean densities than water (18 CFU/100 mL) ( p < 0.001, randomized block ANOVA; pair wise comparisons tested with LSD post-hoc tests; p = 0.02 for SAV vs. sediments and p = 0.01 for sediments vs. water; Table 1 ). However, when the data are examined as the total CFU present in or on each substrate in the entire mesocosm (i.e. tota l on SAV, total in sediment, total in water; see methods for explanation of calculations ), the vegetated ca nopy did not typically harbor the largest proportion of all culturable cells in the syst em. The proportions varied by experiment; i.e., the population in the wa ter column dominated the May experiment, while in July the sediment population represen ted the largest proportion of all culturable enterococci in the mesocosm ( Figure 6-Figure 9). On average, over all experiments, the largest proportion of cells within the mesoco sms was in the sediment (50%), followed by
44 Table 1. Mean concentration and proportional distri bution of culturable enteroco cci in water, sediment and SAV in the vegetated mesocosms for all experiments. Minimum persistence time = latest day in each experiment with dete ctable levels of enterococci in that substrate. (SD = standard deviation; randomized-block ANOVAs followed by Tukey's post-hoc comparisons; S = sediment, V = SAV, W = water). CFU density (log CFU/100 mL or 100 g) % Total CFU Initial Decay Rate ( log CFU/d) Cumulative Decay Rate ( log CFU/d) Minimum Persistence (d) Mean SD Mean SD Mean SD Mean SD Mean SD Water 1.3 1.0 35 18 -0.63 0.50 -0.02 0.12 11 5.5 Sediment 2.1 0.66 50 21 -0.11 0.22 0.02 0.08 10 4.5 SAV 2.9 1.1 15 5 -0.50 0.49 -0.08 0.15 12 4.5 p-value 0.001 0.12 0.07 0.13 0.42 Post-hoc S >W; V > W N/A N/A N/A N/A
45 0.0 0.2 0.4 0.6 0.8 1.0 02468101214time (d)proportion of total CFU Water SAV Sediment Figure 6. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the April experime nts. Time 0 represents samples from the source lake taken prior to mesocosm setup. Data are presented as proportion of the total number of enterococci CFU per mesoco sm persisting in each substrate type
46 0.0 0.2 0.4 0.6 0.8 1.0 02468101214time (d)proportion of total CFU Water SAV Sediment Figure 7. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for th e May experiments. Time 0 represents samples from the source lake taken prior to mesocosm setup. Data are presented as proportion of the total number of enterococci CFU per mesoco sm persisting in each substrate type.
47 0.0 0.2 0.4 0.6 0.8 1.0 02468101214time (d)proportion of total CFU Water SAV Sediment Figure 8. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the July experime nts. Time 0 represents samples from the source lake taken prior to mesocosm setup. Data are presented as proportion of the total number of enterococci CFU per mesoco sm persisting in each substrate type.
48 0.0 0.2 0.4 0.6 0.8 1.0 02468101214time (d)proportion of total CFU Water SAV Sediment Figure 9. Total number of culturable enterococci associated with each substrate type in the vegetated mesocosms for the August experi ments. Time 0 represents samples from the source lake taken prior to mesocosm se tup. Data are presented as proportion of the total number of enterococci CFU per meso cosm persisting in each substrate type.
49 the water column (35%), and finally SAV (15%), although these differences were not significant ( Table 1 ). Indirect effects of SAV were also evident when comparing enterococci densities in the sediment and water of the vegetated mesocosm s to those of the unvegetated mesocosms. Across all replicate experiments, mean cultura ble enterococci densities in the sediment were higher in the vegetated mesocosm s than in the unvegetated mesocosms ( p = 0.05, paired t -test). Sediment-associated initial and cu mulative decay rates were also lower in vegetated mesocosms, though the differences were not significant at = 0.05 ( Table 2 ). In contrast, no significant di fference between vegetated a nd unvegetated treatments was observed in the water column, although si milar general trends were observed ( Table 2 ). When paired t -tests were run for each month (sam ples paired by time), significant differences in mean enterococci densities were seen in April, May, and nearly in July for sediments ( p = 0.01, 0.001, and 0.07, respectively), bu t only in May for water ( p = 0.001). Population Structure BOX-PCR typing of enterococci isolated from the mesocosms showed that these populations were dominated by a single genotyp e. One particular strain accounted for 96.5% of all isolates recovered from all experiments, regardless of substrate, mesocosm, or month (Figure 10). The dominant strain was identified as Ent. casseliflavus by a
50 Table 2. Comparison of enterococci persis tence between vegetated and u nvegetated mesocosms. Values for water and sediments are given. Minimum persistence time = latest da y in each experiment with dete ctable levels of enterococc i in that substrate. (SD = standard deviation; paired t-tests). CFU density (log CFU/100 g) Initial Decay Rate ( log CFU/d) Cumulative Decay Rate ( log CFU/d) Minimum Persistence (d) Mean SD Mean SD Mean SD Mean SD Water Unvegetated 0.85 0.34 -0.96 -0.74 -0.03 0.06 14 0.0 Vegetated 1.3 1.0 -0.63 0.50 -0.02 0.12 10 5.8 p-value 0.17 0.09 0.40 0.20 Sediment Unvegetated 1.7 0.33 -0.23 -0.20 -0.05 0.05 8 4.2 Vegetated 2.1 0.66 -0.11 0.22 0.02 0.08 10 4.5 p-value 0.04 0.17 0.15 0.20
51 Figure 10. Representative results from Enterococcus strain typing showing extremely low strain diversity detected by BOX-PCR for Enterococcus isolates recovered from the mesocosm experiments. Lane 2 represents an isolate from a July SAV sample, lanes 3-6, 8, and 9 are from a July water sample, and lanes 10-13, 15, and 16 are from a July sediment sample, and lane 17 is from an August SAV sample. Lanes 1, 7, 14, and 20 are 1kb ladders, and lanes 18-19 are PCR controls (positive and negative, respectively). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
52 combination of 16S rRNA sequencing and the presence of the distinct yellow pigment that is characteristic of the species. Discussion Concerns about the relationship between fecal FIB concentrations and the presence of pathogens in aquatic habitats have caused an increased focus on FIB that may persist in terrestrial soils, aquatic sediments, and SAV. In this study, the use of paired mesocosms and enterococci collected directly from an e nvironmental source (a freshwater lake in Florida) allowed us to (1) simultaneously compare the relative importance of water, sediment and SAV as substrates for the pe rsistence of enterococci in environmental waters from the perspective of the substrate a nd also from that of the entire system; (2) determine indirect effects of the presence of SAV on the persistence of enterococci in an aquatic system, and (3) examine the populati on structure and dynamics of enterococci from this lake. The results provide evidence that, while SAV can harbor relatively high densities of enterococci per unit mass in relation to other subs trates, in some cases it does not harbor a large proportion of the total enterococci present in the entire aquatic system. It can, however, have indirect effects in the system by facilitating higher densities in the sediment and in the habitat as a whole. Furthermore, the fact that the populations collected from the lake over a ten month pe riod were dominated by a single strain of Enterococcus casseliflavus strongly suggests that some enterococci strains are highly adapted to secondary environmen tal habitats and may not be reliable indicators of human health risk in subtropical waters.
53 Although the persistence of FIB in sediment s has been relatively well documented, SAV has been much less extensively studied, with only a few published accounts of the persistence of FIB on submer ged macroalgae in temperate climates (Whitman et al., 2003; Byappanahalli et al., 2007; Ksoll et al., 2007; Englebert et al., 2008; Kleinheinz et al., 2009). Our results expand upon previous findings by providing evidence of enterococci persisting on vascul ar aquatic plants in a subt ropical environment and, even more importantly, by allowing the simultaneous comparison of the importance of SAV as a substrate in relation to sediment and water from the same aquatic system. We observed mean densities of enterococci that were comp arable to previous work (Solo-Gabriele et al., 2000; Desmarais et al., 2002; Whitman et al., 2003; Ksoll et al., 2007) and were significantly higher on SAV than in the sediment or the water column. While these data certainly highlight the potential for SAV to serve as an important substrate for the persistence of enterococci, there are curre ntly no standardized units for normalizing bacterial densities to substr ates such as sediment and SAV and normalization methods have varied in the literatur e (Solo-Gabriele et al., 2000; Whitman et al., 2003; Anderson et al., 2005; Ksoll et al., 2007). As a result, direct comparisons of the relative importance of each substrate have been difficult to make across studies. In our experiments, we were able to make this comparison by calculating the total number of cells associated with each substrate in an entire mesocosm. When viewed in this manner, the apparent importance of each substrate as a reservoir fo r enterococci shifted dramatically. The high densities observed on SAV were offset by th e relatively low vegetative mass, and the
54 largest proportion on average was in the sediments and the water column. These comparisons, based on the entire mesocosms, illu strate that high densities of FIB per unit mass of vegetative substrates do not necessarily indicate cell reserv oirs that are large enough to greatly affect water quality upon resu spension into the water column. Thus, the importance of benthic rese rvoirs (including SAV) is highly dependent on the total mass of resuspendable, bacteria-harboring subs trate and its relations hip to the volume of water at a specific site. Even though the enterococci that were direct ly associated with SAV were not typically numerically dominant, the use of all three substrates in the mesocosms also allowed us to observe important indirect effects that the pr esence of SAV can have in aquatic systems. The vegetated mesocosms contained significantl y elevated total numbers of enterococci as well as significantly elevated mean dens ities of sediment-associated enterococci, whether or not the densities were normalized to the initial inoculum density to account for the effect of a 10% higher inoculum carried on SAV. These results provide substantial evidence that SAV can, in addition to serving as a substrate for enterococci, also facilitate elevated densit ies in the surrounding habitat. These results are important from the perspective of the entire system as th ey illustrate that the presence of vegetation in an aquatic habitat can signi ficantly elevate the number of enterococci that may be able to persist in the entire habitat as a whole. Possible mechanisms for th is effect include the growth and efflux (disassocia tion and entry into the water column) of SAV-associated cells (as suggested in Ksoll et al ., 2007), increased levels of organic carbon, particularly
55 in sediments, and protection from solar ra diation. Submerged macrophytes have been shown to increase available carbon to su rrounding sediments both by causing increased settling of suspended particulates (Posey et al., 1993; Fonseca, 1996) and by exudation of photosynthate from the roots (Pollard and Mo riarty, 1991). Furthermore, these changes have also been shown to result in increased microbial activity (Lope z et al., 1995; Hansen et al., 2000; Karjalai nen et al., 2001). Our ability to detect culturable enterococci throughout each 14-day experiment (even if intermittently) strongly suggests that thes e enterococci populations are capable of extended persistence in secondary habitats. While the initial decay rates observed in this study were comparable to those reported in ot her studies (Craig et al., 2004; Anderson et al., 2005), the population dynamics observed over the entire course of the experiments were highly variable over both daily and seasonal temporal scales. Unfortunately, the high variability among seasons resulted in a lower than ideal leve l of reproducibility among the replicates of our experiments, and a high level of replicati on was difficult with this experimental setup given the limitati ons of time and resources available for mesocosm sampling and maintenance. Howeve r, we do not believe that the relatively low reproducibility is simply a matter of the leve l of replication we were able to achieve. Instead, we believe that it is actually an interest ing finding in its own right, as it illustrates the extremely high level of va riability inherent in these systems over time and highlights the need for an increased focus on finer scale temporal dynamics among FIB in the environment. Such variable densities have been previously obse rved over time in the
56 field and in laboratory settings (Desmarais et al., 2002; Boehm, 2007), and has been attributed to environmental stresses such as temperature (Stephenson and Street, 1978; Howell et al., 1996; Craig et al., 2004), sali nity (Anderson et al., 1979; Anderson et al., 2005), UV radiation (Davies and Evison, 1991; Muel a et al., 2000), or grazing (Davies et al., 1995). While some authors ha ve suggested that such variab ility reflects the death and growth of these bacteria in the environment (Solo-Gabriele et al., 2000; Desmarais et al., 2002; Craig et al., 2004), we have no data to support whet her the short-term variability seen in our experiments represents changes in the number of total viable cells or changes in the culturability (Lleo et al., 1998; Muela et al., 2000) of a relatively stable number of cells. The domination of the environmental populations in these experiments by one particular strain of Ent. casseliflavus was highly surprising, especi ally considering that the experiments and collections from the lake spanned 10 months. This lake is highly impacted from stormwater via man-made st ructures, but is not known to be directly impacted by sewage or other obvious fecal sources. Ent. casseliflavus is commonly associated with waterfowl, which are pres ent at low density around the lake. Most sources of FIB, such as feces, sewage, a nd stormwater, typically exhibit much higher levels of strain diversity than we observe d here (Anderson et al ., 2005; Anderson et al., 2006; Brownell et al., 2007). It appears th at the chronically elevated levels of enterococci at this site are the result of this particular strain's ad aptation and persistence in this particular hab itat, rather than the result of a chronic influx of pollution. Power et
57 al (2005) came to a similar conclusion regarding high densities of E. coli in two Australian lakes, where they found that re curring blooms were dominated by only three distinguishable strains. All three of the E. coli strains had a group 1 capsule, which would potentially offer an ad aptive advantage to environmental strains by ameliorating stresses such as desiccati on, irradiance and predation. Ent. casseliflavus is characterized by a yellow pigment that may provide similar protection from the stress of UV radiation. Regardless, it is clear that further research on the ecology of persistent fecal FIB in secondary habitats is required to improve our understanding of how FIB populations reflect the fate and ecology of pathogens and how they can best be used as a water quality monitoring tool.
58 CHAPTER THREE THE IMPORTANCE OF SEDIMENT AN D SUBMERGED AQUATIC VEGETATION AS POTENTIAL HABITATS FOR PERSISTENT STRAINS OF ENTEROCOCCUS ACROSS A WATERSHED Introduction Most water quality monitoring strategies are based on the measurement of indicator organisms Â– microorganisms that indicate recent fecal pollution and thus the potential for the presence of waterborne pathogens Â– as opposed to monitoring for individual pathogens specifically. Bact eria belonging to the genus Enterococcus (also termed enterococci) are one of the major groups used as such an indicator in many monitoring programs for recreational water quality (U SEPA, 2000; WHO, 2001). Recent research, however, has provided evidence that some FIB are capable of persis ting in a culturable form for extended periods in the sediments and submerged aquatic vegetation (SAV) of many secondary environmental habitats (Bya ppanahalli et al., 2003a; Whitman et al., 2003; Craig et al., 2004; Anderson et al., 2005; Ishii et al., 20 06; Ksoll et al., 2007). The persistence of benthic FIB (note that we use benthic to descri be bacteria associated with the bottom of aquatic habitats, including se diment and vegetation, as opposed to those suspended in the water column) is particular ly important because resuspension of those cells back into the water column, such as mi ght occur during storms or periods of high recreational activity, may lead to false conclu sions regarding recent contamination. This
59 has been a major concern raised about the reli ability of the FIB concept (Solo-Gabriele et al., 2000; Grant et al., 2001; Whitman et al ., 2003; Anderson et al., 2005; Ishii and Sadowsky, 2008). The consequences of persistent FIB in sec ondary habitats are further complicated by the effects of inherent diversity among FIB. I ndicator groups, like the enterococci and fecal coliforms, harbor interspecifi c variability among their memb er species. For example, within the coliforms, E. coli has been found to be much more commonly associated with sewage and human fecal material, while other members of the group, such as Enterobacter and Klebsiella are much more ubiquitously found in the environment (Leclerc et al., 2001). The same is true for enterococci, where Ent. faecalis and Ent. faecium are very abundant in human feces and sewage, while Ent. casseliflavus and Ent. mundtii are typically associated with environmen tal sources, such as birds and plants (Leclerc et al., 1996; Muller et al., 2001; O tt et al., 2001; Aarestrup et al., 2002). However, in addition to this interspecifi c variability, each species can also exhibit considerable intra specific variability as a result of the strain diversity inherent in a given population. This inherent vari ation has been a major focus of microbial source tracking (MST), which is an area of active research that attempts to overcome the limitations of using solely the concentrations of FIB as the predictor of human health risk from water use. The goal of MST is to distinguish cont amination that originates from various fecal sources (e.g., human, agricultur al, or wildlife), thereby offe ring a means of determining
60 when high concentrations of FIB are truly re presentative of human or other high-risk types of fecal pollution and pose increased hea lth risks (Simpson et al., 2002; Field and Samadpour, 2007; Stoeckel and Harwood, 2007). Methods include both librarydependent and library-independent approaches that try to identify particular microbial strains and target genes that are specific to, or highly associated with, waste from particular host species. Although a variety of methods have been succe ssfully employed to discriminate between sources of FIB in recent years (Bernhard and Field, 2000; Dombek et al., 2000; Scott et al., 2005; USEPA, 2005; McQuaig et al., 2006), the implementation and interpretation of these methods in the environment may be complicated by the population dynamics of persistent FIB in natural environments. Ma ny methods have been used to differentiate among strains of FIB, including ribotyping (Car son et al., 2001; Anderson et al., 2005), antibiotic resistance patterns (Parveen et al., 199 7; Harwood et al., 2000), and amplification of repetitive DNA sequences (Dombek et al., 2000; Johnson et al., 2004). Regardless of the method employed, however, a substantial body of work indicates that interand intraspecific variability can result in differential survival of FIB strains and species in environmental habita ts (Topp et al., 2003; Anderson et al., 2005; Ishii et al., 2006). Such evidence implies that strain s probably have inherently different physiological capabilities that affect their persistence in natural habitats, resulting in complex population dynamics that confound our abil ity to link FIB to potential sources.
61 Low levels of strain diversity have been obs erved at a variety of environmental sites, particularly for E. coli suggesting that some strains ma y not only survive longer, but may even be adapted to continued persistence in the environment at high numbers. For example, ubiquitous E. coli strains that are common in environmental samples and not related to known fecal sources have been observed in temperate soils (Ishii et al., 2006), freshwater beach sands (Kinzelman et al., 2004; McLellan, 2004), and on the macroalga Cladophora (Byappanahalli et al., 2007). Simila rly, extremely high density blooms of pelagic E. coli in Australia were shown to be mostly comprised of just three different strains, even in geographically distant lakes (Power et al., 2005). Such environmentally adapted strains have strong potential for decoupling FIB concentr ations and pathogen presence at these sites, raising concerns over the utility of th e indicator paradigm. In this study, the concentr ations and population structur es of enterococci were simultaneously monitored in water, sediment, and SAV at several sites across the Tampa Bay watershed over an entire year. The goals of the study were two-fold. Firstly, I wanted to determine if high densities of benthic enterococci occurred regularly and consistently across a variety of sites and substrates in the watershed, and also investigate the potential for any spatial or temporal pattern s in the waterborne or benthic densities. Secondly, because strain diversity has b een studied much more intensively for E. coli than for the enterococci, I wanted to em ploy molecular genotypi ng (Â“fingerprintingÂ”) techniques to investigate how the enterococc i population structure and strain diversity
62 varies over space and time in various wate r bodies and to look for any evidence of widespread or cosmopolitan strains that appear to be adapted to the environment. Methods Sampling Sites Four freshwater sites (two streams, one lake and one river) an d two estuarine sites (beaches in upper and lower Tampa Bay) were chosen to represent a typical range of water bodies in the Tampa Bay watershed ( Figure 11 ). The small stream site (28 1.583Â’ N, 82 11.162Â’ W) was on Spartman Branch, a first order stream surrounded by agricultural lands and subject to complete drying out during the winter dry season. The large stream site (28 4.583Â’ N, 82 15.790Â’ W) was on Flin t Creek, a third order stream that flows primarily through wooded and rura l areas and drains Lake Thonotosassa and empties into the Hillsborough River. The ri ver site (28 4.260Â’ N, 82 22.671Â’ W) was at the University of South Florida's Riverfr ont Park on the Hillsborough River, the main River through the city of Ta mpa, and downstream of a subs tantial amount of protected, undeveloped land. The lake site (28 2.918Â’ N, 82 29.828Â’ W) was on Lake Carroll, a small residential lake in West Tampa su rrounded completely by suburban housing. The two estuarine sites include the upper bay si te (27 58.141Â’ N, 82 34.522Â’ W) at Ben T Davis Beach, west of the City of Tampa in Old Tampa Bay, and the lower bay site (27 45.149Â’ N, 82 37.793Â’ W) at Lassing Park, just south of the city of St. Petersburg in the lower portion of the main stem of Tampa Bay. SAV canopies at the fr eshwater sites were dominated by Alternanthera philoxeroides (alligator weed), Egaria densa (Brazilian
63 Figure 11. Site locations in the Tampa Bay watershe d. Freshwater sites: FL = lake; FR = river; LFS = large stream; SF S = small stream. Estuarine sites: UB = upper bay; LB = lower bay. LB UB FL FR LFS SFS
64 waterweed), Hydrilla verticilata Myriophyllum aquaticum (parrot feather), and Vallisenaria Americana (eel grass) SAV at the marine sites were dominated by Halodule wrightii (shoal grass) in the upper bay and a combination of H. wrightii, Syringodium filiforme (manatee grass) and Thalassia testudinum (turtle grass) in the lower bay. All of these sites range from moderate to heavy SAV coverage and have primarily fine quartz sand bottom sediments. Environmental Sampling Water, sediment, and SAV were sampled every month from May 2007 to April 2008. Triplicate samples of 250 mL water, 25 g se diment, and 25g SAV we re collected from each site in sterile containers immediately placed on ice and processed in the laboratory within 4 h. The number of colony forming un its (CFU) of enteroco cci was quantified via membrane filtration. Water samples (1, 10, or 100 mL of each triplicate sample) were concentrated by vacuum filtration directly onto 47 mm nitrocellulose membranes with a 0.45 m pore size and cultured at 41 C for 24 h ours on mEI agar (Difco Laboratories) (USEPA, 2000). For sediment and SAV samp les, 10 g (wet weight) of material was diluted in 100 ml of steril e buffered water (0.0425 g/L KH2PO4 and 0.4055 g/L MgCl2; pH = 7.2) and sonicated on ice at 14 watts for 30 seconds to dislodge and resuspend attached cells (Anderson et al ., 2005). Aliquots (10 or 25 mL ) of the supernatant were then filtered and cultured as above. Con centrations are presen ted as log CFU/100 mL for water samples or log CFU/100 g wet weight substrate for sediment and SAV samples. After counting, well-isolated colonies were pi cked from the mEI agar (up to a maximum
65 of 32 isolates were saved each month for all s ubstrates at all sites) and cultured overnight in Enterococcosel broth (EB, Difco Laboratories, en c.) at 37 to confirm esculin hydrolysis. Glycerol was added (10% v/v) to cultures prior to storage at -80 C for later genetic typing. Genetic Typing Due to time and cost constraints only a subset of the isolates were genetically typed to investigate population structure. Four site s (large stream, river, lake, and upper bay) were chosen for the comparison and isolates fr om all substrates in one sample from each season (January, April, July, and October) we re selected for BOXPCR fingerprinting. Initially, up to 16 isolates were typed from each chosen sample, and for those sites with higher diversity ( S > N /2; S = number of unique strains, and N = number of isolates) all 32 isolates were analyzed. Cr yopreserved isolates were streak ed onto tryptic soy agar to ensure isolation of a pure culture. Isolates were then grown overnight in 1 mL of brain heart infusion broth (BHI, Di fco Laboratories) at 37 C a nd DNA was extracted using the GenElute Bacterial Genomic DNA kit (S igma-Aldrich) per the manufacturerÂ’s instructions for Gram-positive bacteria. The DNA of each isolate was typed by repeti tive extragenic palindromic (REP) PCR fingerprinting using the BOX A2R primer (5Â’-ACG TGG TTT GAA GAG ATT TTC G3Â’) (Koeuth et al ., 1995). 25 L PCR reactions contained 5 L of 5 Gitschier Buffer (Kogan et al ., 1987), 2.5 L of 10% dimethyl sulfoxide, 0.4 L bovine serum albumin
66 (10 mg/mL), 2.0 L 10mM dNTPs, 1.0 L Taq polymerase (5000 u/mL), 10.6 L water, 1.5 L 10 M BOXA2R primer; and 2.0 L of DNA template, containing between 10 and 40 ng/ L of DNA (Versalovic et al., 1991; Malathum et al., 1998). The PCR program included (1) initial de naturation at 95 C for 7min; (2) 35 cycles of 90 C for 30 s, 40 C for 1 min, and 65 C for 8 min; and (3) final extension at 65 C for 16 min. The amplicons were separated on a 1.5 % agarose gel (90 watts for 4 hrs), stained with ethidium bromide (1% solution) and imaged under UV light. Ba nding patterns were analyzed for similarity with BioNumerics 4.0 software (Applied Maths, Inc., Belgium) and confirmed by eye. Similarity was dete rmined from Pearson correlations based upon the densiometric curves (optimization = 1 %) for each genetic type and a dendogram was constructed via UPGMA. Identical reactions of a control strain maintained a similarity of 84%, which was used as a critical value to establish which environmental strains were similar enough to be considered identical and the results were confirmed by eye. Taxonomic Identification of Isolates The species of the fourteen most abunda nt BOX-PCR genotypes (see results) were identified by sequencing a 1-1.2 kb secti on of the 16S rRNA gene. The gene was amplified from extracted DNA via PCR using the universal bacterial primers ECO8F (5Â’AGA GTT TGA TCM TGG CTC AG Â– 3Â’) a nd ECO1492RC (5'-GGT TAC CTT GTT ACG ACT T-3') (Lane 1991). Fifty micr oliter PCR reactions contained 25 L JumpStart Taq polymerase (Sigma, USA), 2.5 L each primer, 5 L water, and 5 L DNA template. The ampli cation process included (1) initial dena turation at 94 C for 5min; and (2) 20
67 cycles of 94 C for 1 min, 55 C for 1 min, and 72 C for 10 min. Amplicons were frozen and sent to a commercial laboratory for seque ncing (Macrogen, Inc., USA). Because the results of the sequencing did not offer resolution between the highly similar Ent. casseliflavus and Ent. gallinarum species, colonies were examined for the yellow pigmentation that is characteristic of Ent. casseliflavus Calculations and Statistical Analysis All CFU data were transformed as log10 (x + 1) to meet normality requirements prior to statistical analysis. Percentage data were tr ansformed as arcsin-squa re root (x) to meet normality requirements. Values were comp ared statistically us ing three-factor ANOVA comparing site, substrate, and seas on (SPSS, version 17.0, SPSS, Inc., USA; = 0.05). Accumulation curves and rarefaction estimat es were conducted using ECOSIM software (Gotelli and Entsminger, 2009). Similarities in clonal structure were compared via ordination. Non-metric multidimensional scal ing diagrams were constructed based upon sample by sample matrices of Bray-curtis di stances for all samples divided by substrate and each sample of site and season with all substrates combined for comparison ( PRIMER-E with PERMANOVA +, Ivybridge, UK). Results Enterococci Densities Mean enterococci densities, when examined over all months and averaged across all sites, were significantly higher on SAV (2.5 x 103 CFU / 100g) than sediments (1.0 x 103
68 CFU/100 g), which were in turn significantly higher than those found in water (1.3 x 102 CFU/100 mL) ( Table 3 and Figure 12 ; three-way ANOVA, F = 51.7, p < 0.001). When examined individually by site, densities on both sediments and SAV were significantly elevated over waterborne de nsities at all six sites ( Figure 12 ; post-hoc paired t-tests, p values ranging from 0.013 to < 0.001). At the freshwater sites, mean densities in SAV were also significantly higher th an those found in sediments ( Figure 12 ; post-hoc paired t -tests; p -values for the small stream, large stre am, river, and lake sites were 0.002, 0.02, 0.03, and 0.05, respectively). At the estuarin e sites, however, SAV densities were not significantly different from densities in the sediments ( Figure 12 ; post-hoc paired t -tests; p -values for the upper and lower bay site s were 0.24, and 0.09, respectively). No significant differences were found when dens ities were compared across seasons with substrates combined in each si te or examined individually ( Table 3 and Figure 13 ; threefactor ANOVA, F = 1.6, p = 0.19). In addition, mean enterococci densities were also signifi cantly affected by site ( Table 3 and Figure 12 ; three-factor ANOVA, F = 12.4, p < 0.001). Sites further up in the watershed (such as the stream and river sites) exhibited higher densities than those further downstream in the watershed (such as the bay sites). This was evident from examination of the homogenous subsets of sites ge nerated from post-hoc comparisons ( Table 3 ), as well as regression analysis comparing mean site densities agains t the distance between each site and the mouth of Tampa Bay (considered the most downstream exit from the watershed) ( Figure 14 ; p < 0.001, r2 = 0.98). When examined by substrate at each site,
69 Table 3. Results from a three-fact or ANOVA showing significant differences in mean enterococci densities across the watershed betw een sites, matrices, and seasons. Overall model significance: F = 263.2; p < 0.001. Factor Mean Std Dev Std Error F p Subsets Site: 12.39 <0.001 Small Stream 3.4 0.88 0.13 a Large Stream 3.2 0.83 0.13 a, b River 3.0 0.98 0.13 b, c Lake 2.8 1.14 0.13 b, c Upper bay 2.5 0.86 0.14 c, d Lower bay 2.1 1.03 0.13 d Substrate: 51.77 <0.001 Water 2.1 0.79 0.09 a Sediment 3.0 0.90 0.10 b SAV 3.4 0.97 0.09 c Season: 1.60 0.19 Spring 2.7 1.06 0.11 a Summer 3.0 0.93 0.11 a Fall 2.8 0.96 0.11 a Winter 2.9 1.17 0.11 a
70 0 1 2 3 4 5 Small StreamLarge StreamRiverLakeUpper BayLower BaySitelog CFU / 100 mL or 100g Sediment SAV Water Figure 12. Mean densities of culturable enterococci from water, sediment, and SAV for each site over all sampling dates. (CFU = colony forming units; error bars indicate standard deviations; n=12)
71 0 1 2 3 4 5 SpringSummerFallWinterSeasonlog CFU / 100 mL or 100 g Sediment SAV Water Figure 13. Mean densities of culturable enterococci from water, sediment, and SAV for all sites grouped by each season (CFU = co lony forming units; error bars indicate standard deviations; n=18)
72 1 1.5 2 2.5 3 3.5 4 4.5 1020304050607080 Distance to Mouth of Tampa Bay (km)Mean Enterococcus density (log CFU/100 mL or 100 g) total water Linear (water) Linear (total) Figure 14. Mean enterococci densities in all substr ates combined (total) and in the water at each site compared to direct-line distance of the site from the mouth of Tampa Bay. Regression for total: F = 197.4; p < 0.001; r2 = 0.98. Regression for water: F = 33.3; p = 0.004; r2 = 0.89.
73 the same effect was true for water ( Figure 14 ; p = 0.004, r2 = 0.89) and SAV (Figure 15; p < 0.001, r2 = 0.97), but not for sediments (Figure 15; p = 0.06, r2 = 0.63). Clonal Structure Clonal structure was highly variable among all samples, regardless of site, substrate, or season. Accumulation curves are shown for i ndividual samples at the river site as an example and illustrate a wide range of values for strain richness across seasons and substrates ( Figure 16 ). For individual samples, valu es for the Shannon-Wiener diversity index ranged from 0 to 2.8, and richness estima tes for samples that could be rarified to N = 14 ranged from 1 to 13 strains ( Table 4 ). When all three substrates were combined for each site, every site displayed a range of diversity throughout the year, as evidence by accumulation curves ( Figure 17-Figure 20). In these samples with substrates combined for each site, values for the Shannon-Weiner diversity index ranged from 0.2 to 3.4, and richness estimates for samples that could be rarified to N = 35 ranged from 3 to 26 strains ( Table 5 ). However, the patterns were not simila r. For example, at the river and upper bay sites, the winter sample showed the highe st diversity, whereas fa ll was the highest at the lake site and summer was the highest at the small stream site. As a result, there were no significant effects of site, season, or substrate in determin ing the strain richness of a given site, as shown by a three-factor ANOVA conducted on sample richness values rarefied to N = 14 ( F = 1.09; p = 0.41).
74 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 1020304050607080 Distance to Mouth of Tampa Bay (km)Mean Enterococcus density (log CFU/100 mL or 100 g) sediment SAV Linear (SAV) Figure 15. Mean enterococci densities in sedime nt and SAV at each site compared to direct-line distance of the site from the mout h of Tampa Bay. Regre ssion for sediment: F = 6.8; p = 0.06; r2 = 0.63. Regression for SAV: F = 136.2; p < 0.001; r2 = 0.97.
75 Figure 16 Accumulation curves constructed for unique strains found in each matrix at the river site during the (A) spring, (B) summe r, (C) fall, and (D) winter season. The dashed line represents the unity line. 0 2 4 6 8 10 12 14 16 18 20 22 048121620242832Individual isolatesUnique strains Water Sediment SAV 0 2 4 6 8 10 12 14 16 18 20 22 048121620242832Individual isolates 0 2 4 6 8 10 12 14 16 18 20 22 048121620242832Individual isolatesUnique strains 0 2 4 6 8 10 12 14 16 18 20 22 048121620242832Individual isolates A C B D
76 Table 4 Measures of Enterococcus strain diversity in each sample of site, season, and substrate. N = number of isolates typed; S = number of strains observed; % unique = number of strains unique to that sample; Rarefaction (14) = estimated strains after rarefaction to 14 isolates (aster isk indicates less than 14 isolat es originally typed for that sample); JÂ’ = SimpsonÂ’s diversity index; HÂ’ = Shannon-Weiner diversity index (loge). Site Season Substrate N S % unique Rarefaction (14) J' H' Large Fall Sediment 13 7 62 0.89 1.73 Stream SAV 10 5 40 0.84 1.36 Water 14 8 57 8.0 0.92 1.91 Spring Sediment 7 7 43 1.00 1.95 SAV 15 11 33 10.3 0.91 2.17 Water 30 6 93 4.1 0.70 1.25 Summer Sediment n.d. n.d.n.d. n.d. n.d. n.d. SAV 27 21 48 12.1 0.96 2.94 Water n.d. n.d.n.d. n.d. n.d. n.d. Winter Sediment 16 3 94 2.8 0.42 0.46 SAV 16 6 81 5.4 0.63 1.12 Water 30 19 57 11.3 0.96 2.83 River Fall Sediment 11 6 100 0.86 1.54 SAV 27 17 41 10.3 0.92 2.60 Water 31 15 42 9.2 0.90 2.43 Spring Sediment 13 3 0 3.0 0.49 0.54 SAV 7 4 14 0.83 1.15 Water 16 3 0 2.8 0.42 0.46 Summer Sediment 29 13 52 9.0 0.93 2.38 SAV 15 7 53 6.8 0.93 1.81 Water 16 5 31 4.6 0.74 1.19 Winter Sediment 31 12 65 8.4 0.91 2.27 SAV 16 9 25 8.2 0.88 1.93 Water 27 21 41 12.1 0.96 2.94 Lake Fall Sediment 15 10 93 9.5 0.94 2.15 SAV 10 6 40 0.84 1.50 Water 16 15 69 13.2 0.99 2.69 Spring Sediment 15 8 33 7.8 0.96 1.99 SAV 10 5 10 0.91 1.47 Water 13 5 8 0.91 1.46 Summer Sediment 14 1 0 1.0 0.00 0.00 SAV 14 2 7 2.0 0.37 0.26 Water 14 2 0 2.0 0.37 0.26 (cont.)
77 Table 4 (cont.) Site Season Matrix N S % unique Rarefaction (14) J' H' Lake Winter Sediment 29 12 79 8.6 0.93 2.32 SAV 15 1 0 1.0 0.00 0.00 Water 12 11 75 0.99 2.37 Upper Fall Sediment 9 7 89 0.97 1.89 bay SAV 16 8 50 7.3 0.82 1.72 Water 16 6 0 5.6 0.84 1.51 Spring Sediment 5 4 60 0.96 1.33 SAV 30 21 60 10.6 0.89 2.70 Water 15 4 7 3.8 0.52 0.72 Summer Sediment 14 3 100 0.60 0.66 SAV 15 6 13 5.7 0.80 1.43 Water 16 3 6 2.8 0.42 0.46 Winter Sediment 27 23 81 12.8 0.98 3.07 SAV 7 4 57 0.92 1.28 Water 33 18 42 8.9 0.82 2.38
78 0 10 20 30 40 50 010203040506070Individual isolatesUnique strains Winter Spring Summer Fall Figure 17 Accumulation curves constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the large stream site. The dashed line represents the unity line.
79 0 10 20 30 40 50 01020304050607080Individual isolatesUnique strains Winter Spring Summer Fall Figure 18 Accumulation curves constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the river site. The dashed line represents the unity line.
80 0 10 20 30 40 50 0102030405060Individual isolatesUnique strains Winter Spring Summer Fall Figure 19 Accumulation curves constructed for unique strains found in all three matrices (water + sediment + SAV) for each season at the lake site. The dashed line represents the unity line.
81 0 10 20 30 40 50 010203040506070Individual isolatesUnique strains Winter Spring Summer Fall Figure 20 Accumulation curves constructed for unique strains found in all three matrices (water + sediment + SAV) for each se ason at the upper bay site. The dashed line represents the unity line.
82 Table 5. Measures of Enterococcus strain diversity in each samp le of site and season with all three substrates combined (water + sediment + SAV). N = number of isolates typed; S = number of strains observed; % unique = number of strains unique to that sample; Rarefaction (35) = estimated strain s after rarefaction to 15 isolates; J Â’ = SimpsonÂ’s diversity index; H Â’ = Shannon-Weiner diversity index (loge). Site Season N S % unique Rarefaction (35) J' H' Large Fall 37 18 54 17.3 0.89 2.58 stream Spring 52 23 69 16.8 0.80 2.52 Summer 27 21 48 21.0 0.96 2.94 Winter 62 28 73 18.9 0.85 2.83 River Fall 69 36 51 22.4 0.90 3.21 Spring 36 7 3 6.9 0.43 0.83 Summer 60 24 47 18.1 0.89 2.84 Winter 74 37 47 23.3 0.93 3.37 Lake Fall 41 30 71 26.4 0.95 3.25 Spring 38 11 18 10.7 0.87 2.10 Summer 42 3 2 2.7 0.20 0.22 Winter 56 23 57 17.4 0.86 2.70 Upper Fall 41 21 39 18.9 0.90 2.74 bay Spring 50 27 44 19.7 0.81 2.68 Summer 45 9 38 8.2 0.76 1.67 Winter 67 42 60 24.5 0.88 3.31
83 Dominant Strains and Species Among all samples, 277 individual strains were identified. Of these, 227 were found in only a single sample (i.e., not found in samp les from any other sites, substrates, or seasons). Of the 227 unique strains, the vast majority of them (199) were found as singletons or doubletons. There were a few strains, however, that occurred in higher numbers even if only in one sample. For exam ple, five of the unique strains were highly dominant (recovered more than 10 times) in the sample in which they were found (four samples from the large stream and one from the upper bay). The percentage of unique strains in a given sample (those not found in any other sample) was also highly variable, and was found to range from 0 to 100% across all samples, or from 3 to 73% when the samples were grouped by site ( Table 4 and Table 5). The percent uniqueness of a sample did vary significantly between the substrates ( Table 6 ). On average, sediment samples had a significantly higher perc entage of unique strains (65 %) than did samples of water or SAV (29% and 33% respectively) (t hree-way ANOVA on arcsin-square root transformed data; F = 4.76, p = 0.014.) In addition, the effect of season on percent uniqueness was very nearly significant, driven by higher average percentages of unique strains found in samples from fall and wint er (58% and 57% respectively) than in samples from spring and summer (25% and 31% respectively) (three-factor ANOVA on arcsin-square root transformed data; F = 2.81, p = 0.053). Overall, however, the large number of unique strains among the samples made the overall similarities between the sites relatively low, and so few, if any patterns, could be
84 Table 6 Results from a three-factor ANO VA showing significant differences in percentage of strains in each sample that ar e unique between sites, matrices, and seasons. Overall model significance: F = 3.02; p = 0.01. Data were arcsin-square root transformed prior to analysis, and are back-transformed here. Factor Mean Std Dev Std Error F p Subsets Site: 1.83 0.16 Large Stream 61 6.2 1.4 a River 36 16.5 1.1 a Lake 27 20.6 1.1 a Upper bay 46 19.4 1.1 a Substrate: 4.76 0.014 Water 29 16.3 0.9 a Sediment 65 21.0 0.9 b SAV 33 7.8 0.8 a Season: 2.81 0.053 Spring 25.0 14.4 1.1 a Summer 31.2 20.9 1.4 a Fall 58.1 14.8 1.1 a Winter 57.1 11.8 1.1 a
85 discerned. Using non-metric multidimensi onal scaling, no recognizable groupings of similar clonal structure were observed among the samples with respect to substrate or season (data not shown). There was some sli ght grouping of clonal st ructure according to site ( Figure 21 and Figure 22 ), particularly at the rive r and upper bay sites, which separated slightly. However, the relatively high stress values associated with these analyses make this interpretation weakly s upported at best. Among th e fifty strains that were found in more than one sample, 14 were found in four or more different samples, and were found on two or more substrates, at two or more sites, and in two or more seasons, suggesting at least some degree of a cosmopolitan nature ( Table 7 ). Results of 16s sequencing show that three of these strains are Ent. casseliflavus three are Ent. faecalis three are Ent. faecium three are Ent. hirae and two are Ent. mundtii Beyond these fourteen most abundant strains, nine st rains were found in three or more samples, and the remaining 27 strains were found in only two samples. Three strains in particular were found in at least 10 (>20%) samples and appear to represent extremely widespread strains. Strains J35 ( Ent. faecium ) and J8 ( Ent. mundtii ) were particularly abundant, accounting for 7.4% and 9.0% of all isol ates, respectively. And although J45 ( Ent. hirae ) was somewhat less abundant (2.3% of isolates), it was the only strain to be found at least once on all three substrates, at each site, and in each season. Discussion While a significant number of studies have employed field sampling to investigate enterococci densities in environmental waters, se diments, and SAV, this work is the first
86 Figure 21 Non-metric multidimensional scaling analysis of similarities in clonal structures from each sample (water, sediment, and SAV plotted separately for each site in each season). Symbols are coded by site. Similarities are based upon the resemblance matrix of Bray Curtis distances. Site: large stream X river lake upper bay
87 Figure 22 Non-metric multidimensional scaling analysis of similarities in clonal structures from each site (water, sediment, and SAV are combined and plotted for each site in each season). Symbols are coded by site. Similarities are based upon the resemblance matrix of Bray Curtis distances. Site: large stream X river lake upper bay
88 Table 7 List of strains that occurred in four or more samp les, with species results from 16s sequencing and a summary of their dist ribution among the samples. Strains are presented in order of descending number of sa mples in which they were found. Isolates = number of isolates of each stra in recovered and percent of to tal isolates typed; Samples = number of samples in which the strain was found and the percent of total samples typed. Strain Species Isolates Samples Seasons Sites Substrates n % n % J35 faecium 59 7.4 13 28.3 3 4 3 J8 mundtii 71 9.0 11 23.9 4 3 3 J45 hirae 18 2.3 10 21.7 4 4 3 J41 faecium 12 1.5 6 13.0 4 3 3 J43 hirae 7 0.9 6 13.0 3 3 3 J3 casseliflavus 8 1.0 5 10.9 3 3 2 J18 faecalis 6 0.8 5 10.9 3 2 2 J4 casseliflavus 10 1.3 4 8.7 3 2 3 J9 mundtii 8 1.0 4 8.7 2 3 2 J20 faecalis 8 1.0 4 8.7 3 3 3 J6 faecalis 5 0.6 4 8.7 3 2 3 J31 casseliflavus 5 0.6 4 8.7 3 2 2 J42 faecium 4 0.5 4 8.7 2 2 2 J49 hirae 4 0.5 4 8.7 3 3 2
89 to directly compare all three matrices via simultaneous measurements across a watershed. In terms of clonal structure, much of the previous work has been conducted solely on E. coli in a limited geographic rang e (mostly in the US Great La kes region), and this work provides much needed data on clonal structure in Enterococcus from a variety of water bodies and substrates in a dist inctly different watershed. Several important conclusions are evident from this study. Firstly, on av erage across the watershed, SAV harbored the highest densities of enterococci, followed by sediments and then the water column. Secondly, mean enterococci densities changed significantly as a function of relative location within the watershed; the highest dens ities occurred at the sites furthest upstream in the watershed, and decreased as the locati on of the site moved further downstream in the watershed (i.e., closer to the mouth of Tampa Bay). Thirdly, clonal richness and diversity of Enterococcus populations varied widely and grouped slightly by site, but with no clear relationships to substrate or season. And finally, several strains were recovered from multiple samples, with three strains in particular (one each of Ent. faecium Ent. mundtii and Ent. hirae ) that were highly abundant and cosmopolitan. While elevated densities of FIB have been previously observed in sediments (SoloGabriele et al., 2000; Craig et al., 2004; Anderson et al., 2005) and SAV (Anderson et al., 1997; Whitman et al., 2003; Ksoll et al., 2007) rela tive to the water co lumn, this study is the first to directly compare all three subs trates simultaneously. Averaged across the watershed, SAV harbored significantly higher de nsities of enterococci than sediment or water, suggesting that it serves as a highly suitable substrate fo r the persistence of
90 enterococci in these habitats. Theoreticall y, this is not surpri sing that enterococci introduced into the water column would associate with SAV, given the intimate contact between SAV and the water column as well as the increased access to resources and the protection from UV radiation. In E. coli rapid association with SAV has been observed in mesocosms (Englebert et al., 2008; Kl einheinz et al., 2009), and Byappanahali et al. (2003b) observed growth of E. coli on algal ex udate, offering additional evidence of its suitability as a substrate for persistence and possible growth. There is some evidence in this study that this trend may not be as strong in the estuarine sites as in the freshwater sites, as the mean enterococci densities in sediments were nearly equal to SAV in the upper bay site and greater than SAV in the lower bay site. Possible mechanisms for this difference include the effects of salinity or differences between ma rine and freshwater aquatic plants, but the data in this study do not offer any c onclusions on these theories. Regardless, in highly vegetated sites, it is cl ear that SAV can be an important matrix for the persistence of water quality FIB. Furthermore, as the major bulk of work investigating this phenomenon has been limite d geographically to the Great lakes region of the United States (Byappa nahalli et al., 2003b; Whitman et al., 2003; Ksoll et al., 2007; Englebert et al., 2008; Kleinheinz et al., 2009), th ese data help extend these findings to a broader scale and to a wider variety of habitats. Across the watershed, sites further upstream had significantly highe r mean enterococci densities (all substrates and seasons combined) than those downstream (i.e., closer to the mouth of Tampa Bay). Although a variety of st udies have examined densities of FIB in a
91 variety of habitats in the t ypical watershed, including soil s (Hardina and Fujioka, 1991; Desmarais et al., 2002), stream s (Buckley et al., 1998; Byappa nahalli et al., 2003a), rivers (Tunnicliff and Brickler, 1984; Obiri-Danso and Jones, 1999), lakes (Doyle et al., 1992; Whitman and Nevers, 2003), estuaries (Shiaris et al., 1987; Solo-Gab riele et al., 2000), and open ocean beaches (Anderson et al ., 1997; Boehm, 2007), very few have simultaneously monitored multiple habitat types throughout a watershed. Roll and Fujioka (1997) investigated the potential for surrounding so ils to serve as a non-point source of FIB in freshwater streams. Ther e also have been a few studies on smaller watersheds (e.g., coastal streams with outfalls near a recreational beach) that have taken a watershed approach and found higher densitie s of FIB in upstream or downstream sites (Byappanahalli et al., 2003a; Boehm et al., 2004 ; Whitman et al., 2006). In one of these studies, Byappanahali et al. (2003a) observed an opposite trend, with waterborne E. coli concentrations generally increasing as stream order increased. Steet s and Holden (2003) used a modeling approach to compare a co astal wetland to a nearby beach, concluding that elevated densities in the wetland constituted a source of contamination for the beach. In this study, however, by covering such a wide diversity of habitat types and substrates over such a broad of a geographic range, I was able to observe a strikingly consistent increase in mean enterococci densities as sites are situated upstream in the watershed, regardless of substrate. Although I used di stance from a fixed point in the watershed (i.e., the mouth of Tampa Bay) as a determin ation of location in the watershed, this is only a convenient proxy for watershed location. The true causes of the relationship are
92 most likely to be other factors that vary along the same gradient. For example, sites further upstream in a watershed tend to have of much smaller ratio of water volume to shoreline influence, which allows little poten tial for dilution and ma kes these waters the most intimately connected and easily affect by terrestrial sources of FIB. Furthermore, the small, narrow waterways typical of the ups tream sites tend to have more shading from riparian vegetation, offering protection from harmful UV radiation, which has been shown to quickly inactivate waterborne FIB (Davies and Evison, 1991; Muela et al., 2000; Sinton et al., 2002; Schultz-Fademrecht et al., 2008). In terms of population structure, there seems to be little associati on between the clonal diversity or structure of the Enterococcus populations sampled in this study and cofactors such as the season or substrate from which the sample was taken. Diversity was highly variable, from almost zero in some samples to Shannon-Weiner diversity values of just below three, which is similar to valu es found in another study of enterococci in Florida (Brownell et al., 2007), but much lower than values observed for E. coli in lake sediments and algae (McLellan, 2004; Bya ppanahalli et al., 2007). While some of these differences may be due to differences between the tw o taxonomic groups, they are also likely a function of sampling effort. In this study, typi ng effort at the indivi dual sample level was limited to allow for increased sample cove rage, and it is clear from the accumulation curves that considerable increases in richness and diversity could be expected in some of the samples from increased sampling effort. Therefore, it is impossible to say whether
93 clonal diversity was actually significantly lower than that found by McLellan (2004) and Byappanahali et al. (2007). Ordination showed a slight te ndency for some of the sites, particularly the river and upper bay sites, to cluster separately based upon clonal st ructure. Environmental E. coli isolates were found to group very strongly by s ite and matrix in a study of temperate soils and lake water (Ishii et al., 2006), and by year in sampled patches of Cladophora algae (Byappanahalli et al., 2007). In both of these studies, groupings of clonal structure were much more highly separated, but this is likely due to the fact that these groupings resulted from principal component analyses based upon the densiometric curves of individual isolate fingerprints. In c ontrast, the NMDS analyses in this study are based upon BrayCurtis distances of the overall clonal stru cture in each sample, which is entirely dependent on the presence and absence of each strain, and does not account for high degrees of relatedness between two different strains. Kinzelman et al. (2004) found very little clustering of E. coli in Lake Michigan between site s or substrates and concluded that accumulation of strains, rather than long-term pers istence and replication of particular strains were the main cause of th e observed population struct ure. In this study, given the wide range of diversity values a nd the broad spatial and temporal coverage, it may be that samples represented a mix of so me population structures that are dominated by recently accumulated strains and others that represent long-term, persistent strains.
94 A fair number of strains were observed that were fairly cosmopolitan in nature and may represent naturalized strains that are widespr ead in the environment. The species mixture of these strains, including Ent. casseliflavus faecalis faecium hirae and mundtii are very similar to other studies of species structure of envir onmental enterococci (Pinto et al., 1999; Harwood et al., 2004; Ferguson et al., 2005). The species indicate a mix of those typically associated with humans ( faecalis and faecium ) and those that are typically considered more environmental in origin ( casseliflavus and mundtii ). The relatively high abundance of the Â‘enviro nmentalÂ’ species of Enterococcus compared to that found in human sewage adds weight to the theory th at many of these environmental enterococci can not be traced to recent in fluxes of fecal contamination into the environment. Three strains (J35, J8, and J45) were particularly abundant and cosmopolitan compared to all the others, and appear likely to be natura lized strains that are persisting throughout the watershed. The recovery of these strains from such a wide variety of sites, substrates, and seasons was highly surprisi ng and indicates the need for increased research into the ecology of persistent FIB. Understanding the factors that allow these strains to persist and probably replicate in the environment will greatly improve our ability to discriminate between instances where bacteria such as E. coli and the enterococci are acting reliably as indicators of fecal pollution and instances where they are merely abundant in the environment and other measures of human health risks must be employed.
95 CHAPTER FOUR INVESTIGATING THE IMPORTANCE OF SEDIMENT AND SUBMERGED AQUATIC VEGETATION AS ENVIRONM ENTAL RESERVOIRS FOR WATER QUALITY INDICATOR BACTERIA Introduction Fecal contamination in natural water bodies th at are used for shell fishing and recreation poses a significant risk to public health. It is well estab lished that swimmers and bathers who use beaches, lakes, and rivers with known sewage contamination are at a higher risk for gastrointestinal and resp iratory illnesses as well as skin, ear and eye infections (Cabelli et al., 1982; Cheung et al., 1990; Rees et al., 1998 ; Wade et al., 2003). The ongoing challenge has been the development of a suitable means of detecting fecal contamination, as well as quantifying its ma gnitude and extent, and then accurately gauging the health risks associated with contacting contaminated waters. Ideally, managing agencies would directly monitor fo r pathogens that are known to result from contamination by sewage or other fecal s ources. In practice, however, directly monitoring water bodies for the presence of human pathogens is prohibitively difficult because there is such a wide diversity of pot ential pathogens (including viruses, bacteria, and protists), many are difficult and costly to culture, many have no reliable molecular
96 assays, and many have patchy dist ributions or continue to pose significant health risks at low concentrations (Field and Samadpour, 2007). The historical approach to this problem has b een the use of FIB. FIB include specific (or specific groups of) bacteria that, although they are not path ogenic themselves, reliably occur in high numbers in feces and sewage Their presence in recreational waters, therefore, is used to indicate contaminati on by sewage or other fecal material and the likely presence of human pathogens. The main groups of FIB used today in developed areas such as the United States and the Eu ropean Union include fecal coliforms, or a specific member of that group, Escherichia coli in fresh water and the genus Enterococcus in estuarine and marine waters (USEPA, 1986, 2000; WHO, 2001). While these indicators have shown good correlation with sewage contamination and risks of waterborne illness (Wade et al., 2003; Zmirou et al., 2003) there are many assumptions that must hold true for the indicator concep t to work optimally. While various published review articles provide extended discussion re garding these assumptions (Griffin et al., 2001; Field and Samadpour, 2007; Ishii a nd Sadowsky, 2008), one assumption is of particular importance with regard to this study: FIB must co-occur with human pathogens in order to accurately indicat e a human health risk. Unfortunately, recent research has indicated that this assumption is often false. Many studies have shown that the presence of FIB do not correlate well with th e presence of pathogens, including Salmonella Campylobacter Cryptosporidium Giardia or enteroviruses (Lund, 1996; Bonadonna et al., 2002; Lemarchand and Lebar on, 2003; Harwood et al., 2005).
97 There are many possible reasons for this lack of correlation, but one major problem is the assumption that, subsequent to sewage contamination in the environment, FIB will exhibit survival dynamics that are similar to t hose of the pathogens they are being used to detect. In truth, FIB Â– including coliforms, E. coli and enterococci Â– are capable of persisting in a culturable form for extended periods in a wi de variety of environmental matrices after their initial in troduction. Such matrices incl ude terrestrial soils (Topp et al., 2003; Ishii et al., 2006), aquatic sedi ments (Byappanahalli and Fujioka, 1998; SoloGabriele et al., 2000; Jeng et al., 2005), a nd attached to submerged aquatic vegetation (SAV) (Byappanahalli et al., 2003b; Whitman et al., 2003; Ksoll et al., 2007). The extended persistence of FIB, particularly in the sediments and SAV of recreational waters, is one highly likely cause of the detec tion of FIB in the absence of pathogens and the poor correlation between the two. Persis tent FIB in benthic matrices, such as sediments and SAV, may be reintroduced into the water column whenever sediments get resuspended, such as during storms or high recreational activity. B acteria are typically the most easily resuspended of benthic organi sms due to their small size and association with cohesive surficial fluff sediments (A uer and Niehaus, 1993; Howell et al., 1996; Shimeta et al., 2002; Jeng et al., 2005). And because we do not yet have a good understanding of whether many pathogens are sim ilarly able to persis t in the environment or are regularly resuspended from sediment s, it is difficult to predict whether these resuspended FIB reflect a real human health risk. This has been a major concern about the reliability of the indicator organism con cept that has been raised repeatedly in the
98 literature (Solo-Gabriele et al., 2000; Grant et al., 2001; Whitman et al., 2003; Anderson et al., 2005; Ishii and Sadowsky, 2008). Unfortunately, the importance of extended persistence and resuspension of FIB has been difficult to quantify in terms of its effects on human health risks. Although there have been recent epidemiological studies show ing that increased exposure to beach sand carries increased risk of dis ease (Bonilla et al., 2007; Heaney et al., 2009), no correlations have been determined between health risks and concentrations of FIB in sediments or SAV. Nor is there a standard method that has been adopted for their detection and quantification. In managed stre ams, resuspension of FIB has been observed to occur as a result of both natural (Nagels et al., 2002; Jamieson et al., 2005) and experimentallyinduced (McDonald et al., 1982; Wilkinson et al ., 1995; Nagels et al ., 2002; Muirhead et al., 2004) periods of high flow, in the abse nce of rainfall and (presumably) groundwater inputs. However, in the vast majority of other recreational water bodies that are not so easily constrained (such as beaches and lake s) the resuspension of benthic FIB has typically been inferred. For example, obser vations of relatively high water column E. coli counts have been shown to correlate with factors that cause sediment erosion and resuspension, such as wave or tidal activity (Le Fevre and Lewis, 2003; Shibata et al., 2004; Whitman et al., 2006; Yamahara et al., 2007) and increased boa ting activity (An et al., 2002), or have been correlated with sedime nt densities through the use of time series or structural equation modeling (Whitman and Nevers, 2003; Whitman et al., 2006).
99 Another approach to elucidating the importan ce of sediments as a pot ential reservoir of FIB has been to incorporate resuspension pr ocesses into the modeling of FIB fate and transport. Because benthic bacteria are typi cally adhered to sedime nt particles (Auer and Niehaus, 1993; Howell et al., 1996; Davi es and Bavor, 2000), hydrodynamic information and sediment characteristics can be used to predict sediment resuspension and offer relatively good approximations of the behavior of benthic FIB in the sediments (Bai and Lung, 2005; Jamieson et al., 2005). Both unidirec tional (e.g., tidal and stream flows) and oscillatory (e.g., wave action) flow regimes set up velocity gradients along the bottom that increase from zero at the sea floor up to the mainstream veloci ty. The steepness of these gradients, in combination with bottom roughness that results from bedform elements (e.g., sediment grains and sand ripples) establishes a shear st ress that acts on the sediment water interface (Denny, 1988; Soulsby, 1998). If the for ce of this shear stress is sufficiently strong to overcome the natural settling velocity of individual sediment grains, some amount of sediment will be maintained in suspension (Soulsby, 1998; Le Roux, 2005). As a result, sets of theoretical and empirical equations allow the prediction of concentrations and transport of suspended sediment under un idirectional a nd oscillatory flow patterns, which can then be used to esti mate similar processes for the associated FIB (Bai and Lung, 2005; Jamieson et al., 2005). General terms for resuspension rates (based on critical shear stresses resulting in sediment resuspension) have been incorporat ed into embayment-wide models used to predict net transport of FIB (Steets and Holden, 2003; Sanders et al., 2005), and a much
100 more detailed model has been published that uses the Environmental Fluid Dynamics Code model to specifically pr edict the fate and persistence of sediment associated fecal bacteria (Bai and Lung, 2005). These models, however, necessarily use very broad brush approaches with regards to resuspension dyna mics, and it is becoming increasingly clear that understanding how benthic-pelagic coupl ing affects the population dynamics of species of FIB is very important to predicti ng their survival and transport in receiving waters. In fact, continued data and experi mentation on the behavior of benthic FIB has been outlined as a distinct need for fu ture model improvement (Bai and Lung, 2005; Pachepsky et al., 2006). I believe that one of the major deterrents to our ability to readily interpret the importance of benthic reservoirs of FIB is that their densities have typically been normalized per mass or volume of substrate (CFU/100 mL for water or CFU/g for solid substrates such as sediments and SAV) (Byappanahalli and Fujioka, 1998; Solo-Gabriele et al., 2000; Topp et al., 2003; Whitman et al., 2003; Anders on et al., 2005; Jeng et al., 2005; Ishii et al., 2006). Such normalization makes sense for the water column, as the concentration is the most appropriate value to consider in terms of human health risk. However, FIB concentrations that are normalized to volume of water do not represent a direct comparison to concentrations normalized to mass of sediment or SAV, and this discrepancy does not allow for a simple in terpretation of the importance of benthic sources of resuspendable bacter ia. It is possible, however, to use a different method of normalizing bacterial densities, and analyze da ta from aquatic systems on the basis of
101 landscape area (e.g., per m2), which allows direct compar ison bacterial population sizes in waterborne and benthic samples (Mui rhead et al., 2004; Ja mieson et al., 2005). In this study I revisited the same sites in the Tampa Bay watershed that were sampled in Chapter 3 and used the concept of landscape area to reexamine the relative population sizes of the enterococci found in the water, sediment, and SAV samples to determine their relative magnitude. Furthermore, I want ed to further investigate the potential for benthic substrates, such as sediment and SAV, to serve as important reservoirs of resuspendable FIB, as is often suggested in the literature. The study had three specific objectives: (1 ) To identify and quantify key habitat characteristics that would allow the normaliza tion of enterococci de nsities on a landscape basis and allow direct compar ison of the population sizes in water, sediment, and SAV at each site; (2) develop a simple model that wi ll predict shifts in the relative population sizes at a given site that resu lt from changes in important ha bitat characteristics such as bacterial densities, water depth, SAV cover, etc. ; (3) use historical wind and flow data at each site, in conjunction with theoretical ca lculations of sediment resuspension, to determine the likely effect of sediment-associ ated bacterial resuspension on water quality monitoring at each site. This research w ill contribute to a better understanding of the effects of sediment resuspension on the fate, tr ansport, and performance of FIB within the Tampa Bay watershed, based on realistic, timeaveraged values for data collected on enterococci densities.
102 Methods Environmental Sampling Data for bacterial concentrations are from Ch apter 3. Briefly, water, sediment, and SAV at four freshwater and two estuarine site s in the Tampa Bay watershed were sampled monthly from May 2007 to April 2008 ( Figure 11 ). Samples were placed on ice immediately after collection and processed in the laboratory the same day. Water samples were vacuum filtered directly onto 0.45 m nitrocellulose membranes and cultured at 41 C for 24 hours on mEI agar (Difco Laboratories) supplemented with nalidixic acid (USEPA, 2000). Sediment and vegetation samples were diluted 1:10 (w/v) in phosphate buffered water and sonicated at 19 watts for 30 seconds to dislodge and resuspend attached cells (Anderson et al., 2005). Aliquots of the water were then filtered and cultured by standard membrane f iltration methods (USEPA, 2000). Final concentrations are presented as CFU/100 mL water or CFU/100 g wet weight substrate. After counting, colonies were picked from the mEI agar and cultured overnight in Enterococcosel broth (EB, Difco Laboratories, enc.) at 37.5 to confirm identification. Habitat Characterization In order to convert the densities measured in Chapter 3 to a landscape scale, the following habitat characteristics were measured: wate r depth, sediment density, the depth of sediment containing FIB, the biomass density of SAV in a vegetated bed, and the percent coverage of SAV beds over the entire aqua tic bottom. Habitat characterization for all sites was conducted in July of 2008. Alt hough some values, such as SAV biomass,
103 cover, and water depth, are expected to va ry over seasonal cycles, midsummer in the Tampa Bay watershed typically represents the period with the most rainfall and the highest amount of SAV growth. As a result, these valu es were taken in July under the assumption that they would represent relati vely maximal values for SAV biomass and water depth over an annual cycle. However, additional water depth data were obtained at the stream sites in July 2009 during a pe riod of exceptionally heavy rainfall and high water levels, and are presented as an extreme case in terms of water depth. Water depth at each site was determined as the mean of 10 random measurements. At stream and river sites, these ten values were obtained from a transect across the channel to characterize the entire wa ter body at the location of samp ling. At the lake site, ten locations were selected randomly on a bathymet ric map averaged to get a mean depth for the entire lake. Finally, at the bay sites, only the local depth was used, which was characterized by ten random measurements that were taken in the vicinity of the sampling location, during the mean tidal hieght. At th e river and lake sites, I also obtained estimates for the local shoreline depth in th e same manner in an attempt to characterize the immediate sampling area as a contrast to using the entire water body. To determine sediment density, five sediment samples from each site were collected in 50 mL centrifuge tubes. The sediment was allowe d to settle in each tube during transport, and upon return to the laboratory, the overlying water was poured off prior to analysis. The volume and mass of the remaining sedime nt in each sample was recorded and the
104 wet density was calculated, as this was the unit or normalization used in Chapter 3. Next, each sediment sample was dried at 80 C for 24 hours, and the dry mass was measured for each sample so that a dry density could be obtained as well. To determine the depth to which FIB were detectable in the sediments, three replic ate sediment cores of approximately 2.5 cm in diameter and 20 cm in length were taken at each site. Upon removal, each core was longitudinally divided in to six sections (3 cm long Â– the deepest 2 cm were discarded due to disturbance from sampling), which were homogenized and analyzed for enterococci densities. The sa mples were analyzed using the same methods as those described above. The enterococci density in each sediment sample was calculated (CFU / 100 gww), and the cutoff depth was determined to be the last depth at which enterococci densities were within at l east one order of magnitude of those found at the surface of the core (e.g., see Figure 23 and Figure 24). Densities at deeper depths, which were beyond an order of magnitude lower than surface densities, were assumed to be numerically insignificant in terms of total population size. To determine the amount of vegetative biom ass in an SAV bed, five quadrats (0.0625 m2) were thrown haphazardly into SAV patches at each site. The emergent portion of all SAV within the quadrat was removed down to bare sand and all excess water was allowed to drain for approximately 30 s. Next the mass was immediately measured and the mean of all replicates for each site was used as the typical wet biomass of SAV. Finally, the percent coverage of SAV beds (as opposed to ba re sediment) over the entire
105 bottom was determined via visual estimati on at each site (Dethier et al., 1993; Fourqurean et al., 2001; McDonald et al., 2006; Bell et al., 2008). Calculations The values obtained from the habitat charact erizations were used, along with the mean enterococci densities for each substrate at each site observed in Chapter 3 ( Figure 12 ), in the following formulas to recalculate enteroco cci densities at the landscape scale. For waterborne bacterial concentrations, CFU/m2 was calculated as a function of CFU/100 mL and water depth: CWL = 104 CWV dW (1) where CWL = waterborne bacterial concentr ation normalized to landscape (CFU/m2); CWV = waterborne bacterial concentration normalized to volume (CFU/100 mL); and dW = water depth (m). For sediment-associated b acterial densities, CFU/m2 was calculated as a function of CFU/100 g, sediment density and sediment depth: CSL = 104 CSM DS dS (2) where CSL = bacterial concentration in sedi ment normalized to landscape (CFU/m2); CSM = bacterial concentration in sediment normalized to sediment mass (CFU/100 g); DS = sediment density (g wet weight/cm3); and dS = sediment depth (m).
106 For SAV-associated bacterial densities, CFU/m2 was calculated as a function of CFU/100 g, SAV biomass in vegetated patches, and proportion of bottom with vegetated cover: CVL = 10 CVM BV PV (3) where CVL = bacterial concentration in S AV normalized to landscape (CFU/m2); CVM = bacterial concentration in SAV normalized to mass (CFU/100 g); BV = SAV biomass in a vegetated patch (kg/m2); and PV = proportion of SAV cover over entire bottom. Finally, the total CFU/m2 for the entire system at each site was simply calculated as the sum of each substrate (eqs. 1-3): CTL = CWL + CSL + CVL (4) where CTL = total bacterial density in the system normalized to landscape area. After the calculation of total densities, the relativ e population sizes for each substrate were calculated by dividing that substrateÂ’s population size by the total population size. Modeling and Sediment Re suspension Estimates Four sites (large stream, river, lake, and uppe r bay) were chosen to theoretically explore the effect of sediment resuspension on water borne concentrations of FIB. Resuspension estimates were calculated according to the methods outlined in Soulsby (1998) for determining the concentration of suspended marine sands under currents and waves. Sediments at each of the sites in the Tampa Bay watershed were dominated by noncohesive sand, allowing the use of these me thods. For the stream and river sites, estimates were based upon predictions of resuspension under a unidirectional current
107 based upon bulk water flow velocity, as th ese sites were too narrow to have any appreciable wave generation. Alternatively, for the lake a nd bay sites, estimates were based upon predictions of resuspension under wa ve-generated oscillatory flow, as the lake site is too small to have any significan t unidirectional circulation and the bay site is well sheltered from tidal currents. For resuspension estimates under both unidire ctional and oscillatory flow, input data were needed for water temperature, salinit y, depth, and median sediment grain size. Mean values for water temperature and salin ity obtained during the monthly sampling at each site (Chapter 3) were used. For sedime nt grain size, samples were collected from each site during the habitat characterization de scribed above and were sorted with sieves in the laboratory. At each site, five replicate samples of surface sediments (top 3 cm) were collected and homogenized into a single re presentative sample for each site. Upon return to the laboratory, sedime nt grains were successively washed through a series of metal sieves with pore sizes of 1.0, 0.5, 0.25, 0.125, and 0.063 mm. Each size fraction was washed into pre-combusted ceramic crucible s, and the filtrate from the smallest sieve was captured and triplicate a liquots were filtered onto pre-co mbusted glass fiber filters (Whatman GF/F, 0.7 m pore size). All sediment samples were dried at 60 C for 24 hours and then weighed to determine dry mass. Next, the samples were combusted at 450 C for 4 hours to burn off the organic portion, and the samples were weighed again to determine the percent organic matter. Fina lly, the cumulative proportions (by mass) of each size fraction were plotted and values for d10, d16, d50, d84, and d90 (corresponding to
108 the 10th, 16th, 50th, 84th, and 90th percentiles of grain diameter ) were extrapolated from the graphs (Soulsby, 1998). For the river and stream sites, water flow sp eed was measured directly during periods of extremely high flow in July 2009 to obtain an upper limit for realistic flow speeds. Mainstream current velocity was determined by measuring the time for neutrally buoyant particles suspended in the wate r to travel a known distance. Three sets of ten replicate particles were timed at different locations within the channel to obtain a mean flow speed. These flow speeds were then compared to and supplemented with archived data for the river site available from the Un ited States Geological Survey (station # 02303330). No USGS station was available for the stream site. For the lake and upper bay sites, mean wave period ( Tm) and significant wave height ( Hs) were estimated from historical wind data acco rding to established methods outlined in the Coastal Engineering Manual ( CEM ) (USACE, 2002). Wind reco rds were obtained from the National Oceanic and Atmospheric Admi nistrationÂ’s National Buoy Data Center (station OPTF1 Â– Old Port, Tampa, FL) for the entire year between May 2007 and April 2008. The complete wind records were then filtered for each site to include only the subset that could be considered onshore (and thus wave-generating). For the lake site, this included wind with directions between 60 and 190 degrees, and for the upper bay site this included records between 180 and 270 degr ees. Next, a frequenc y histogram of wind speeds for each site-specific subset was gene rated and used to create a table of wind
109 speeds that represent maximum, 0.2, 0.5, 1st, 2nd, 5th, 10th, 25th, and 50th percentiles during the sampling period. Finally, these wind speed s, in combination with a mean fetch distance (measured from ar eal photographs) were compared to nomograms in the CEM to estimate representative values for Tm and Hs, with consideration of limits for shallowwater wave generation. Because the calculations for estimating resusp ension are too lengthy to cover in detail here (Soulsby, 1998), only a brief overview wi ll be given. For resuspension under currents, the mainstream flow speed, Ub, was used to calculate a depth averaged flow speed, The median sediment grain size ( d50) was then used to determine the settling velocity, ws, the bottom skin-friction shear stress, os, and the critical bottom shear stress cr. If os > cr, meaning that some amount of sediment motion and resuspension was occurring, then a profile of se diment concentration by water depth was calculated. This profile was then integrated over the entire wa ter column to determine the total amount of sediment resuspended per square meter area. By multiplying the resuspended sediment mass by the bacteria density (CFU / 100g) in the sediment for each site, the total number of resuspended bacteria (CFU / m2) was calculated. Finall y, by dividing by the volume of water at each site per m2, the predicted increase in waterb orne bacteria concentrations (in CFU / 100 mL) was determined. The process outlined for resuspension under waves is similar in principle, although the calcula tions vary somewhat. Firstly, the JONSWAP wave spectrum was used (as opposed to monoc hromatic waves) for each combination of wave height and period obtained from th e wind records. The standard deviation, Urms,
110 and amplitude, Uw, of the bottom orbital velocity in a typical wave cycle were then calculated. Similar to the equations for curr ents, sediment grain size and density were used to determine ws and cr. The presence of bedforms was determined, as well as their height, r, and wavelength, r, if present. Next, the total bottom shear stress, b, and the rough bed friction factor, fwr, were determined and compared to cr as before to determine if there was sediment motion. Finally, the co ncentration profile and the resulting amount of suspended sediment and bacteria were calculated as explained above. Results Habitat Characterization and Area Normalization Water depths at the sites ranged from extremel y shallow (<50 cm) at the stream sites, to moderately shallow at the bay sites (<1 m w ith 0.5 Â– 1.0 m tides), to relatively deeper water at the river and lake sites (2 Â– 3 m). Sediments were primarily sand and the densities were highly consistent at all of the sites (~ 1.9 gww/cm3) except for the two stream sites, which included a higher proporti on of organic material and were slightly less dense (~ 1.6 gww/cm3) ( Table 8 and Table 9). Depth profiles of enterococci densities at the six sites gene rally showed a decline in dens ity with increasing sediment depth ( Figure 23 and Figure 24). Cutoff depths (the depth beyond which at least an order of magnitude decline in densities wa s observed) were highl y variable, ranging from 3 cm at the small stream site to 15 cm at the lower bay site ( Figure 23 and Figure 24, Table 8 and Table 9 ). SAV biomass within vegetated patches was higher at the freshwater sites, with values of about 2.5 kg/m2 at the bay sites and 2.9 to 10.5 kg/m2 at
111 Table 8. Key habitat characteristics measured at each of the freshwater sites, which were used to convert enterococci densities in each substrate from mass-normalized values to t hose normalized to landscape area. Sediment depth is the depth to which enterococci densities were found to be within one orde r of magnitude of surface dens ities. SAV density is the biomass density of SAV withi n a vegetated patch, and SAV % cover is the tota l percentage of aquatic bottom covered by ve getated patches as opposed to bare sand Enterococci densities (CFU / m2) Water depth (cm) Sediment Depth (cm) Sediment Density (gww / cm3) SAV density (kg / m2) SAV % cover Water Sediment SAV Total Small Stream 36 3 1.63 2.9 80 1.84 Â• 106 1.65 Â• 106 2.52 Â• 105 3.74 Â• 106 Small Stream, high 98 3 1.63 2.9 80 4.95 Â• 106 1.65 Â• 106 2.52 Â• 105 6.85 Â• 106 Large Stream 17 12 1.66 10.5 60 6.90 Â• 105 2.92 Â• 106 3.66 Â• 105 3.97 Â• 106 Large Stream, high 140 12 1.66 10.5 60 5.58 Â• 106 2.92 Â• 106 3.66 Â• 105 8.86 Â• 106 River 190 6 1.92 5.4 10 3.70 Â• 106 1.06 Â• 106 2.37 Â• 104 4.79 Â• 106 River Bank 36 6 1.92 5.4 95 6.74 Â• 105 1.06 Â• 106 2.37 Â• 104 1.43 Â• 106 Lake 320 6 1.90 5.4 95 2.97 Â• 106 7.61 Â• 105 1.62 Â• 105 3.89 Â• 106 Lake Shore 53 6 1.90 5.4 90 5.22 Â• 105 7.61 Â• 105 1.62 Â• 105 1.44 Â• 106
112 Table 9. Key habitat characteristics measured at each of the freshwater sites, which were used to convert enterococci densitie s in each substrate from mass-normalized values to t hose normalized to landscape area. Sediment depth is the depth to which enterococci densities were found to be within one orde r of magnitude of surface dens ities. SAV density is the biomass density of SAV withi n a vegetated patch, and SAV % cover is the tota l percentage of aquatic bottom covered by ve getated patches as opposed to bare sand Enterococci densities (CFU / m2) Water depth (cm) Sediment Depth (cm) Sediment Density (gww / cm3) SAV density (kg / m2) SAV % cover Water Sediment SAV Total Upper Bay, mid-tide 80 9 1.91 2.4 40 2.72 Â• 105 1.61 Â• 106 6.40 Â• 103 1.84 Â• 106 Upper Bay high tide 30 9 1.91 2.4 40 4.42 Â• 105 1.61 Â• 106 6.40 Â• 103 2.06 Â• 106 Upper Bay, low tide 130 9 1.91 2.4 40 1.02 Â• 105 1.61 Â• 106 6.40 Â• 103 1.72 Â• 106 Lower Bay, mid-tidle 75 15 1.92 2.6 80 2.01 Â• 105 1.45 Â• 106 3.24 Â• 103 1.66 Â• 106 Lower Bay, high tide 125 15 1.92 2.6 80 6.7 Â• 105 1.45 Â• 106 3.24 Â• 103 1.79 Â• 106 Lower Bay, low tide 25 15 1.92 2.6 80 3.35 Â• 105 1.45 Â• 106 3.24 Â• 103 1.52 Â• 106
113 0 3 6 9 12 15 18 -4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.0log CFU / 100g decrease in Enterococcus densitySediment depth (cm) Small Stream Large Stream River Figure 23. Depth profiles of mean enterococci de nsities (presented as decrease in log CFU / 100g from the shallowest depth) in from sediment cores at the small stream, large stream, and river sites; n = 3 at each depth. The vertical dashed line represents a one order of magnitude decrease from surface leve ls, which was used as the cutoff depth for the calculations of sedime nt depth in the models.
114 0 3 6 9 12 15 18 -4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.0log CFU / 100g decrease in Enterococcus densitySediment depth (cm) Lake Upper bay Lower bay Figure 24 Depth profiles of mean enterococci densities (presented as decrease in log CFU / 100g from the shallowest depth) in from sediment cores at the lake, upper bay, and lowe bay sites. Each core was 18 cm deep and divided into 6 sections that were each 3 cm long; n = 3 at each depth. The vertical dashed line represents a one order of magnitude decrease from surface levels, wh ich was used as the cutoff depth for the calculations of sediment depth in the models.
115 the freshwater sites. SAV cover over the en tire bottom was also hi ghly variable, ranging from 10% at the river site to almost complete cover at the lake site ( Table 8 and Table 9). Total enterococci densities normalized to lands cape area at all of the sites were surprisingly consistent, on the order of 106 CFU/m2. Generally, sites with lower sediment enterococci densities normalized to mass (e .g., bay sites), were compensated by having a relatively deeper depth of colonization in th e sediment, resulting in total population sizes that were comparable to sites with higher sediment enterococci densities per mass (e.g., stream sites). The landscape-normalized densitie s at each site were highest in water and sediments, ranging from 105 106 CFU/m2, while the densities in SAV were consistently lower, ranging from 105 at the stream sites to 103 at the bay sites ( Table 8 and Table 9 ). When the numbers of enterococci on each s ubstrate were reexamined as relative proportions of the total populat ion of enterococci at a give n site, the results differed between the freshwater and estuarine sites ( Figure 25 and Figure 26). At the freshwater sites, water depth seemed to be the major f actor in determining re lative population sizes of enterococci in each substrate. In situa tions with shallow water depths (such as during periods of low stream flow, or when consider ing only the shoreline at the lake or river site), the sediment population was similar to or even greater than the waterborne population. However, during peri ods of high water in the st reams, or when accounting for the entire volume of water at the lake or river site, the water borne population became the dominant portion ( Figure 25 ). Conversely, at the estuarin e sites, the populations in the sediments were consistently dominant, typically accounting for 80-90% of the total
116 0 0.2 0.4 0.6 0.8 1Small Stream Small Stream High Large Stream Large Stream High RiverRiver Bank LakeLake ShoreSiteRelative proportion of total CFU / m2 Vegetation Sediment Water Figure 25. Mean proportion of total number of enterococci per square meter landscape area found in water, sediment, and SAV samp les at three freshw ater sites around the Tampa Bay watershed obtained from mont hly samples between May 2007 and April 2008. The two columns for each stream site represent normal and extreme high water depths. The two columns for the river and lake sites represent values for the entire water body vs. those if consideration is constraine d to the nearshore banks only. CFU = colony forming unites.
117 0 0.2 0.4 0.6 0.8 1 Upper Bay Upper Bay Low Upper Bay High Lower Bay Lower Bay Low Lower Bay HighSiteRelative proportion of total CFU / m2 Vegetation Sediment Water Figure 26. Mean proportion of total number of enterococci per square meter landscape area found in water, sediment, and SAV samp les at two estuarine sites in Tampa Bay obtained from monthly samples between Ma y 2007 and April 2008. The three columns for each site represent changing proportions fo r varying water depths at mid-, low, and high tide leves, respectively. CFU = colony forming unites.
118 enterococci load regardless of changing water levels during the tidal cycle ( Figure 26 ). SAV was consistently found to harbor sma ll to negligible frac tions of the total enterococci, ranging from a maximum of 9.2% at the large stream site down to 0.2% at the lower bay site ( Figure 25 and Figure 26 ). Modeling Theoretical Habitat Changes After examining the relative popu lation size of enterococci in each substrate at each site, the model was used to theoretically vary key habitat characteri stics along a realistic continuum to determine how much the relative population sizes may pot entially shift. As a result of my initial interest in the impor tance of SAV as a substrate for enterococci, I chose to use the model to investigate the large stream site, which had the highest proportion of total enterococci associated with SAV. Figure 27 shows the results of modeling changes in the relativ e population size of enteroco cci in the sediments vs. the water column, as a result of theoretically varying water depth a nd SAV coverage. The relative proportion of total enterococci pr edicted to be on SAV ranged from 0% to approximately 18% at the shallowest depths and full coverage. The proportion in the water column along the same gradient was pred icted to range from approximately 58% to 0%, respectively. So at the shallow depths sometimes found in this stream (15-20 cm), the proportion of enterococci found in SAV wa s predicted to exceed that found in the water if the SAV coverage were to approach 100%. However, due to high densities of enterococci in the sediment, SAV was never predicted to harbor a dominant proportion of total enterococci under any conditi ons. At all levels of wa ter depth and SAV coverage,
119 Figure 27 The relative proportion of total enterococci found in water, sediment, and SAV in response to theoretically varying valu es for water depth and SAV bottom cover at the large stream site. Shades of gray and associated sideba r indicate corresponding proportion value on the z-axis. CFU = colony fo rming units; waterborne bacteria = 4.0 x 102 CFU/100 mL; sediment bacteria = 1.5 x 103 CFU/100 g; sediment depth = 12 cm; sediment density = 1.66 g/cm3; SAV bacteria = 5.8 x 103 CFU / 100 mL; SAV biomass = 10.5 kg / m2. Water Sediment SA V
120 the SAV-associated population was always c onsiderably smaller than the size of the sediment-associated population, even though the mass-normalized densities in SAV at this site averaged nearly 104 CFU/100 g. The model was also used to predict how relative enterococci population sizes would change at the upper bay site as a result of normal tidal fluctuations in water depth. Even though the vast majority of enterococci were found in the sediments on average at the bay sites, Figure 28 represents one particular sample (A pril 2008) in which the proportions of enterococci found in water and sediment were relatively equal. As a result, tidal fluctuations were predicted to cause dramatic shifts between states where the water and sediment alternated as the dominant proporti on of total enterococci. As expected, SAV consistently harbored a very small fraction of the total enterococci. As a result, the degree to which the resuspension of sedime nts could potentially effect waterborne concentrations of enterococci was predicted to be highly dependent on the tidal cycle, and therefore shift rapidly in time. Finally, the idea of sediment resuspension at the upper bay site was explored more thoroughly in Figure 29 The model was used to predict the increase in waterborne enterococci concentrations as a result of the resuspension of various amounts of sediment under a range of sediment enterococci densities. At low sediment densities (102 CFU/100 g) there were never enough enterococci resusp ended to exceed the waterborne regulatory limit of 104 CFU/ 100 mL with th e model limited to 10 cm of to tal sediment depth. At
121 Figure 28 The relative proportion of total enterococci found in water and sediment in response to theoretically varying water depth as a result of tidal fluc tuations at the upper bay site during the April 2008 sampling ev ent. Waterborne bacteria = 1.1 x 102 CFU/100 mL; sediment bacteria = 2.2 x 102 CFU/100 g; sediment depth = 9 cm; sediment density = 1.92 g/cm3; SAV bacteria = 1.2 x 103 CFU/100 g; SAV biomass = 2.4 kg / m2; SAV cover = 40%. Water Sediment
122 Figure 29. Predicted increases in waterborne ente rococci concentrations at the upper bay site resulting from theoretically varying values for sediment enterococci densities and the amount of sediment resuspended. (SAV ne gligible and not shown.) CFU = colony forming units; initial waterborne bacteria = 1.5 log CFU/100 mL; water depth = 80 cm; sediment density = 1.92 g/cm3; regulatory limit = 104 CFU/100 mL. Regulatory limit ( 104 CFU/100 mL )
123 moderate sediment densities (103 CFU/100 g), resuspension of approximately 3 cm of sediment was required to cause an exceedance, while at relatively high densities (104 CFU/100 g), an exceedance was predicted after resuspension of only 0.5 cm of sediment. Resuspension Estimates Sediment characteristics at al l of the sites were very similar. Estimated values for d50 ranged from 0.18 to 0.26 mm with corresponding values of 1.95 to 2.5. Sediments at the river, lake, and upper bay si tes were characterized as fine sand, while the sediment at the large stream site was ch aracterized as medium sand ( Table 10 ). The sediments were relatively homogenous and well sorted, with d10 values ranging from 0.09 to 0.13 mm and d90 values ranging from 0.24 to 0.45 mm. Th e sediments contained a very low amount of organic material at all sites, rangi ng from 1-2% at the rive r and stream sites to a fraction of a percent at the la ke and bay sites. Settling velo cities were all similar, at approximately 0.03 m/s ( Table 10 ). Current velocities at the stream and river site s were very low. Direct measurements taken at the two sites during the hi gh flow period of July 2009 averaged 0.32 m/s at the stream site and 0.25 m/s at the river si te. Examination of historic da ta for the river site revealed that the site typically expe rienced much lower flows duri ng the sampling period, less than 0.1 m/s ( Figure 30 ). This flow rate was only exceeded a few times during the summer of 2007, and then once during a peri od of extremely high flow in the later winter of 2008 when the flow reached approximately 0.4 m/s. Resuspension estimates for these flows at
124 Table 10. Sediment characteristics at the large stream, river, lake, and upper bay sites. Diameters are reported in millimeters, and values for are based on d50. d10 d16 d50 d84 d90 % organic Classification Settling velocity (m/s) Large Stream 0.12 0.15 0.26 0.42 0.45 1.95 2.1 medium sand 0.039 River 0.09 0.09 0.18 0.27 0.36 2.5 1.2 fine sand 0.024 Lake 0.09 0.12 0.18 0.23 0.24 2.5 0.03 fine sand 0.024 Upper Bay 0.13 0.15 0.23 0.40 0.44 2.1 0.2 fine sand 0.028
125 0 0.1 0.2 0.3 0.4 0.5 050100150200250300350400DayMean flow speed (m/s) Figure 30 Mean flow speed at the ri ver site (the only site for which historical data were available) for all days between May 2007 and April 2008.
126 the two sites were extremely low. Bottom shear stress ( b) only exceeded the critical stress for threshold motion at the absolu te highest flows of 0.4 m/s and above ( Table 11 ). Even in these cases, however, resuspenion of both sediment and associated bacteria were minimal. If the mean values for enterococci densities in sediment were used for each site, resuspenion was negligible at all flow speeds, resulting in suspended bacterial loads that were several orders of magnitude belo w what would be necessary to affect water quality monitoring. Even if the maximum obser ved densities were used at each site, the highest bacterial load that was predicted from resuspension was an increase of 1 CFU/100 mL at the river site, and this only occurred under an extremely high theoretical flow speed that was never actually observed ( Table 11 ). Examination of the wind records for the enti re year showed wind from all directions during some part of the year, with the mo st dominant directions being easterly and northeasterly, and smaller peaks in freque ncy coinciding with northwesterly and southerly winds ( Figure 31 ). In total, winds were considered wave-generating approximately 45% of the time for the lake site and 16% of the time for the upper bay site. In general, wind speeds were predominantly below 20-25 km/hr, with occasional records of higher speeds, up to maxima at each site of approximately 60 km/hr. On average, the lake site received slightly higher onshore wind speeds than the upper bay site, as judged by its relatively broader hi stogram shape and highe r percentile values ( Figure 32 and Figure 33, Table 12 ). The resulting predic tions for wave generation, however, varied between the two sites due to the different fetch lengths, which were
127 Table 11 Values for key hydrodynamic parameters and resuspension of sedi ment and associated bacteria estimated from stream flow data at the large stream and river sites. U = mainstream water flow speed; cr = threshold bottom shear stress for sediment motion; b = estimated bottom shear stress; sediment resuspende d = dry weight of sediment in suspension per m2 area; sediment bacteria = FIB density in the sediment (the two different values used represent the mean and the maxi mum observed at each site); and bacteria resuspended = concentration of FIB added to water column through resuspension. Site Water depth (m) Temp (C) U (m/s) Bedforms CD cr (N/m2) b (N/m2) Sediment Resuspended (kg/m2) Sediment bacteria (CFU/100 g) Bacteria resuspended (CFU/100 mL) 1.4 23 0.32 ripples 8.07 Â• 10-3 0.16 0.16 0.0 0.0 0.0 Large Stream 0.40 ripples 8.07 Â• 10-3 0.16 0.24 2.4 Â• 10-4 1.5 Â• 103 3.4 Â• 10-4 6.8 Â• 104 1.6 Â• 10-2 0.50 ripples 8.07 Â• 10-3 0.16 0.38 6.5 Â• 10-3 1.5 Â• 103 9.3 Â• 10-3 6.8 Â• 104 4.3 Â• 10-1 River 1.9 23 0.25 ripples 7.07 Â• 10-3 0.14 0.08 0.0 0.0 0.0 0.40 ripples 7.07 Â• 10-3 0.14 0.20 6.3 Â• 10-5 9.1 Â• 102 4.2 Â• 10-5 1.3 Â• 105 5.9 Â• 10-3 0.50 ripples 7.07 Â• 10-3 0.14 0.32 2.3 Â• 10-3 9.1 Â• 102 1.5 Â• 10-3 1.3 Â• 105 2.1 Â• 10-1
128 0 1 2 3 4 5 6 7 8 9 04080120160200240280320Wind Direction (degrees)Percentage of total time Figure 31. Frequency histogram of wind direction (g rouped in bins of 10 degrees) for all available wind records at the port of Tampa between May 2007 and April 2008. The solid vertical lines represent the records considered to be onshore winds for the upper bay site (between 180 and 270 degrees) and the dashed lines represent the limits for the lake site (between 60 and 190 degrees).
129 0 1 2 3 4 5 6 05101520253035404550556065Wind speed (km/hr)Precentage of total records Figure 32. Frequency histogram of wind speed (g rouped in 1 km/hr categories) for all wind records that were onshore for the upper bay site (between 180 and 270 degrees) between May 2007 and April 2008.
130 0 1 2 3 4 5 6 7 8 05101520253035404550556065Wind Speed (km/hr)Percentage of total records Figure 33 Frequency histogram of wind speed (g rouped in 1 km/hr categories) for all wind records that were onshore for the lake site (between 60 a nd 190 degrees) between May 2007 and April 2008.
131 Table 12. Values for key hydrodynamic and bedform para meters, and estimates for resuspension of sediment and associated bacteria at the lake and upper bay sites. FIB densitie s (CFU/100 g) at the lake site: mean = 6.6 Â• 102 and max = 1.5 Â• 104. FIB densities (CFU/100 g) at the upper bay site: mean = 9.3 Â• 102 and max = 2.9 Â• 103. cr = 0.14 N/m2 at the lake site and 0.15 N/m2 at the upper bay site; r = amplitude of sand ripples; r = wavelength of sand ripples; b = estimated bottom shear stress; sediment resuspended = dry weight of sediment in suspension per m2 area; and bacteria resuspended = concentrat ion of FIB added to water column through resuspension. Bacteria resuspended (CFU/100 mL) Wind frequency (%) Wind Speed (km/hr) r (cm) r (cm) b (N/m2) Sediment suspended (kg/m2) mean max Lake max 61 1.2 7.3 0.61 1.6 Â• 10-1 1.5 Â• 10-1 3.3 0.2 35 0.7 3.7 0.24 8.4 Â• 10-3 7.7 Â• 10-3 1.7 Â• 10-1 0.5 32 0.6 3.5 0.22 6.0 Â• 10-3 5.5 Â• 10-3 1.2 Â• 10-1 1 29 0.5 2.7 0.19 2.9 Â• 10-3 2.6 Â• 10-3 5.8 Â• 10-2 2 26 0.4 2.3 0.17 1.9 Â• 10-3 1.7 Â• 10-3 3.8 Â• 10-2 5 21 0.3 1.8 0.13 0.0 0.0 0.0 10 18 0.3 1.5 0.12 0.0 0.0 0.0 25 13 0.1 0.8 0.05 0.0 0.0 0.0 50 9 0.1 0.5 0.04 0.0 0.0 0.0 Upper bay max 59 2.7 19.8 1.22 1.37 1.8 5.5 2 33 2.7 19.8 1.22 1.37 1.8 5.5 5 28 2.4 18.3 1.29 1.19 1.6 4.8 10 24 2.7 18.6 1.00 0.95 1.3 3.8 25 19 2.7 16.9 0.74 0.51 0.7 2.7 50 12 1.9 11.1 0.40 0.07 0.1 0.3
132 found to average approximately 12 km at the upper bay site and less than 1 km at the lake site ( Table 12 ). As a result, even the strongest wi nds resulted in waves of only 0.3 m in height with a period of 1.3 s at the lake site, generating or bital velocities of 0.27 m/s. Once the wind speed dropped below the sec ond percentile (< 26 km/hr) vales for wave height and orbital velocity we re both below 0.1 m and m/s, respectively. In contrast, all of the wind records at the upper bay site down to the s econd percentile (> 33 km/hr, occurring on 27 different days in the year) were predicted to generate waves of the maximum possible height and period possible fo r the 1 m depth resulting in relatively high orbital velocities of 0.53 m/s. Even wind speeds down to 12 km/hr (occurring 50% of the total time, at least once on 159 days of the year) generated modest orbital velocities of 0.22 m/s and above ( Table 12 ). Regardless of the wind speed or orbital velocity, however, predicted bacterial resuspension due to wind-generated waves at the lake site was found to be negligible. Wind speeds below 26 km/hr (which occurred 98 % of the time) failed to even initiate sediment motion. Even at the highest wi nd speed, and assuming the highest observed sediment densities of enterococci, an incr eased bacterial load of only 3 CFU/100 mL was predicted ( Table 12 ). Resuspension at lower wind speeds was several orders of magnitude below any level that might have a significant impact on concentrations in the water column. At the bay site, the wave action was found to generate much more sediment resuspension on a regular basis, often one to two orders of magnitude more than the lake site at a given wind speed. Signifi cant resuspension of sediments even occurred
133 water column. At the bay site, the wave action was found to generate much more sediment resuspension on a regular basis, often one to two orders of magnitude more than the lake site at a given wind speed. Signifi cant resuspension of sediments even occurred at wind speeds as low as 19 km/hr, which occurred on approximately 30% of the total days during the year. However, the rela tively low mean and maximum densities of enterococci observed in the sediments at that site still caused predictions of resuspended bacteria to be very low. The resuspension of bacteria at the upper bay site was predicted to typically result in an increase of only 1-5 CFU/100mL at this site ( Table 12 ). Discussion Much attention has been paid to the potential for FIB that pe rsist in aquatic sediments and vegetation to be resuspended into the water column and thereby nega tively impact their reliable use as an indicator of fecal pollu tion (Solo-Gabriele et al., 2000; Grant et al., 2001; Whitman et al., 2003; Anderson et al., 2005; Ishii and Sadowsky, 2008). Unfortunately, the historical tendencies for co llecting and reporting data on densities of FIB in water, sediment, and SAV have made th e analysis of their im pact on water quality difficult. By quantifying key habitat character istics at my research sites, and utilizing historical data and theoretic al equations for sediment re suspension, in this study I was able to directly examine the relative size of benthic reservoirs of enterococci in the Tampa Bay watershed and determine their potential to significantly affect water column concentrations. The results can be summari zed in four key findings. Firstly, SAV, even at highly vegetated sites, always harbored the smallest percentage (between 0% and 18%)
134 of the total number enterococci in the system and is predicted to have minimal potential to affect concentrations in the water column. Secondly, sedi ment can harbor a relatively large fraction of the total reservoir of enterococci, depending on the water depth, FIB densities, and the depth of sediment cont aining FIB at each site. Thirdly, a simple modeling exercise shows the potential for the relative size of sediment and water column enterococci populations to shift dramatically as key habitat ch aracteristics, such as water depth and SAV bottom coverage vary in space and time. And finally, realistic estimates of sediment resuspension illustrate that, ev en though the reservoir of sediment-associated enterococci may be numerically dominant, current and wave cond itions in Tampa Bay and the surrounding watershed would rarely, if ever, cause an amount of sediment resuspension sufficient to significantly aff ect water column concentrations of FIB. SAV as a Reservoir One of the most interesting results of the habitat charact erization was to discover the relatively small and unimportant (at least numer ically) proportion of total enterococci in the habitat that were found in SAV. This result was unexpected cons idering that the sites in this study were specifically chosen for their high vegetated biomass and that high FIB densities in SAV on a per gram basis have been consistently reported in this study ( Figure 12 ) and others (Whitman et al., 2003; Olap ade et al., 2006). This discrepancy illustrates the importance of properly consid ering how FIB densities are normalized and compared before judging the relative importa nce of environmental reservoirs of FIB. Even though the traditional mass unit (1 g or 100 g) of SAV typically contains orders of
135 magnitude greater densities of enterococci than the traditional volum e unit (100 mL) of water, there are typically orde rs of magnitude more volume units of water than mass units of SAV in a square meter of aquatic habitat. At all of the sites in this study, the sheer volume of water in the entire system, relative to the mass of SAV, more than offset the high enterococci densities in S AV, making it a minor reservoi r in the vast majority of samples. It is important to note, however, that I am discussing numerical importance Â– the potential for a large number of FIB associated with one substrate to shift to another substrate and significantly cha nge the density in the new su bstrate (e.g., bacteria being released from SAV into the water and affecti ng water column concentrations.) I am not implying that SAV are generally unimportant as a refuge for FIB. As others have suggested, SAV may also serve an important ro le as a substrate for FIB growth, not just persistence (Byappanahalli et al., 2003b; Ksoll et al., 2007). In this case, we would need to be able to compare the rates of FIB gr owth on SAV and flux to the water column to rates of mixing and dilution with in the water column. If th e rates are sufficiently high, FIB in SAV could have signifi cant effects on FIB concentrations in the water column, even though SAV is a proportionally minor reser voir at any given instant. Unfortunately, although the above studies have reported gr owth of FIB on SAV, a lack of detailed estimates of growth kinetics and flux rate s between SAV and the water column prohibit any realistic estimates of the importance of th is process. More data on FIB growth in SAV will be highly valuable in predicting the importance of SAV as a reservoir of FIB
136 and the impact of SAV on the success of using FIB to determine risks to human health in environmental waters. Sediment as a Reservoir My calculations show that sediment is poten tially a much more important reservoir of enterococci than SAV at these sites. Depending on the site conditions, the relative proportion of enterococci in the sediments might be much higher than, approximately equal to, or much lower than that found in th e water column. The outcome at a given site was largely driven by the rela tive FIB densities in sediments and water, as well as the depths of the water column and the cutoff dept h of the sediments that harbor FIB. In other words, the relative volum e of water and habitable sedi ments in an aquatic system, combined with the relative densities of FI B contained in each, typically determined which substrate held the larges t population. This was particular ly true at the freshwater sites, where changes in water depth could quickly alter the re lative volumes such that a shift in dominance between the sediment and waterborne populations could be observed. At depths approaching 1m and above, the volume of the water column typically became large enough to cause it to hold the dominant proportion of enterococci in a given freshwater site. It is important to consider that changing wate r depths at a given site will also probably be accompanied by changes in many of these other values, depending on the time scale. Longer term changes in water depth (weeks to seasons) at a given site would be
137 accompanied by long-term changes in FIB densities, SAV biomass, and possibly the depth of FIB colonization in the sediments. But even short-term (hours to days) changes in water depth at freshwater sites that occur as a result of recent rain events would likely be accompanied by changes in FIB concentratio ns in the water column. For example, FIB concentrations in the water tend to incr ease after a rain event as a result of runoff from non-point sources on land or stormwater systems (Reeves et al., 2004; Ahn et al., 2005; Brownell et al., 2007). However, because these sources are finite, other studies have observed increased FIB concentrations on ly during the early pe riods of rain event (i.e, the rising limb of the storm hydrograph), followed by relatively lower concentrations once the system has been flushed for some ti me (McDonald et al., 1982; Nagels et al., 2002; Muirhead et al., 2004; Jamieson et al., 200 5). In either case, it is clear that estimating changes in FIB concentration th at correlate with water depth are highly dependent on several spatial a nd temporal factors. In my approach, I believe that using the mean values for such estimates gives at least a good first orde r approximation of the importance of sediments as FIB reservoirs. At the estuarine sites, sediments were c onsistently found to c ontain the numerically dominant enterococci populati on, even when water depth was allowed to vary over the entire tidal cycle, up to the deepest depths of ~ 1.5 m. This dominance seems to be driven by the relatively deep co lonization of the sediments (9 Â– 15 cm before a log unit decay in density) and the low concentrations found in the water column (typically well below 102 CFU/100 mL). Data regarding the dept h distribution of FIB in sediments is
138 sparse and highly variable. Whitman et al. (2006) reported si gnificant colonization (defined similarly as in this study) of E. coli in Lake Michigan beach sands to deeper depths, down to about 30 cm. Meanwhile, ente rococci concentrations in marsh sediments along the California coastline were significant only in the top 1 cm (Grant et al., 2001). While many factors may affect depth dist ributions of culturable FIB (e.g., wave or current action, organic content, sediment grain size), it is cl ear that such estimates are critical to improving attempts at modeling fate and transport of FIB in receiving waters. For example, in areas where FIB are limited onl y to the shallowest sediment depths (such as the small stream site in this study) care must be taken not to overestimate the total number of FIB present in the sediments a nd available for resuspension and transport. Predicting Reservoir Shifts The model developed for predicting shifts in re lative population sizes at a given site that result from changing habitat or microbial conditions proved to be a valuable and illustrative tool. It is important to note that sufficient data are not available to use this model to precisely predict water quality conditio ns at a particular point in space and time, nor was that the intent behind its development. Rather, my intent was to use it as a tool to look at broad shifts in relative sizes of FIB populati ons among the substrates. By doing so, it helped to determine particular ranges of conditions at a given site where the numbers of FIB bacteria in the benthos are large relative to those in the water column, and thus represent a potentially important reserv oir in the system that could affect water quality monitoring. For example, as explaine d above, one of the ini tial conclusions that
139 became evident from the habitat characteriz ation was that SAV rarely, if ever, was predicted to be a numerically dominant reservoi r of FIB at any of the sites. Through the use of the model at the large st ream site, I was able to determine that, at the typical FIB densities for this site, SAV-associated popul ations of FIB would only be significantly larger than water column populations at the very shallowest depths (<15 cm) and with a high percentage of vegetated bottom cover. Outside of these conditions, and barring a drastic change in relative FIB densities, S AV-associated enterococci would simply not be present in dominant enough numbers in the sy stem to drastically affect water column concentrations, no matter what proportion mi ght be resuspended. Furthermore, even when the population in SAV exceeds that in th e water, both of them are dwarfed by the proportion found in the sediments, suggesting that the role of SAV as a reservoir is still relatively minor in comparison. In terms of sediments, the model offers a si milarly valuable perspective on determining conditions under which bacteria-laden sediments will at least have the potential to affect waterborne concentrations at a given site. As stated a bove, the relative sizes of FIB populations in water and sediments at many of the sites were hi ghly dependent on the depths of water and sediment available to FI B as substrate. And as was just discussed, the use of the model allowed a rough approxima tion of where this transition takes place (dominance shifting from sediments to the wa ter) at a given site under a given set of conditions. Such determination gives a much better idea of what pe rcentage of the time bacteria-laden sediments may potentially impact water quality monitori ng at a particular
140 site of concern. Regardless of the densities th at may be present in sediments, if there are not sufficient numbers to dramatically affect the concentration in wa ter, they may be of little concern, at least in terms of determin ing the fate and transport of FIB. (The presence of pathogens also possibly persisting in sediments is, of c ourse, another question altogether.) Even at the upper bay site, where the se diment-associated population was always dominant under average conditions, the mode l was used to show that, in certain conditions, this may not always hold true. Ho wever, even more interestingly, the model illustrates that there are times when a shif t in dominance may occur not only on longer time scales (as relative concentrations of FIB change), but even on very short time scales (as tidal depth changes). Although short-term variability (i.e., hours) has been shown to occur in populations of FIB both in experi mental chambers (Desmarais et al., 2002; Chapter 2) and in the field (Boehm, 2007), it is not at all well understood. Short-term variability in habitat conditions will also affect the relative importance of various FIB populations in environmental waters, and th ese dynamics associ ated with habitat variability may even exacerbate the inherent population dynamics of the FIB themselves. Resuspension of FIB The lack of significant resuspension of bacter ia predicted for any of the sites using flow and wind data was a surprising and important conclusion of this study. The theoretical approach to sediment suspension worked well at these sites for at least two reasons.
141 Firstly, all of the sites in this study, even th e inland freshwater sites, were dominated by well sorted quartz sand grains with very little organic co ntent. The mechanics of resuspension of non-cohesive sediments are much better under stood and theoretical predictions tend to be simpler and more accu rate than for the organic rich, cohesive sediments found in many other areas. S econdly, the hydrodynamics at each site is predominantly driven by only one process (either unidirectional current or windgenerated waves), which could be estimated from available historical data with relative ease. As a result, we were able to obtain realistic estimates for suspended sediment concentrations at four of the sites over an entire range of hydrodynamic conditions during the sampling year. By coupling these estima tes with the frequency with which they occurred during the year, I was able to ge nerate an in-depth picture of how the resuspension of sediments laden with FIB might affect the apparent microbial water quality at a range of sites around the Tampa Bay watershed. There are several reasons that may explain th e discrepancy between our results and those of other studies in which resuspension from bottom sediments was found to be a significant source of FIB in the water column. Firstly, sediment resuspension is highly dependent on bottom shear stresses and sedi ment grain size, which can vary widely between different habitats and different hydrodynamic regimes (Soulsby, 1998). For example, the upper bay and lake sites, like most inshore waters along the Florida Gulf coast, are highly fetch and depth limited, whic h prohibits the generati on of large waves. As a result, bottom shear stresses are more limite d at these sites than at sites from other
142 studies, such as Lake Michigan and Calif ornia beaches, that may experience much steeper bottom slopes and larger wave acti on (Whitman and Nevers, 2003; Whitman et al., 2006; Yamahara et al., 2007). And simila rly, the non-cohesive quartz sand found at all of the sites in this study responds very differently to hydrodynamic forces than the much finer, cohesive silt a nd mud found beneath many freshw ater streams and estuarine marshes (Grant et al., 2001; Jamieson et al., 2005; Sanders et al., 2005), which may also account for much of the difference. Secondly, based upon the data from our previous sampling efforts, we were able to use realistic data regarding CFU densities in wate r and sediments at all of our sites. On average, the values in the sediment (CFU /100 g) were only about 1 to 1.5 orders of magnitude higher than those in the water column (CFU/100 mL) (Chapter 3). As a result, our sites exhibited a relatively less concentrated reservoir of sediment-associated bacteria than other modeling efforts have assumed. Fo r example, in what is probably the most intensive modeling effort look ing at the role of sediments in affecting the fate and transport of FIB, Bai and Lung (2005) appl y their model to a case study conducted by Muirhead et al. (2004) looking at resuspension in a stream bed during experimental flood stages. In their model, they assume an initial FIB concentration of 108 CFU / m2 in the sediments and 0 CFU in the water. While th ese input conditions re sult in their model fitting that particular case study very well, this represents densities of sedimentassociated bacteria that were approximately two orders of magnitude greater than what was found on average at the sites in this study. And while sediment densities did
143 approach these levels during th ree different sampling events in our study (once at each of the stream sites and once at the river s ite), they occurred when water column concentrations were already well in exceedance of the regulatory limit, suggesting that further resuspension was no longer as great a concern as it would be at less-polluted sites. In order for a water quality exceedance to be caused by sediment resuspension alone at any of the sites in this st udy, I calculated that, even at the maximum predicted water flows, the sediment concentrations would have to be at least 5 x 104 or 5 x 105 CFU/100 g at the upper bay and lake sites, respectively, and approximately 107 CFU/100 g at the stream and river sites. These densities were ne ver seen at any of thes e sites, regardless of the accompanying water column concentration. A final factor that is important to consider regarding conclusions about the importance of FIB resuspension is the water depth upon which the study focuses. In this study, I chose to model the site with water depths of 1-2 m because that is the true nature of the sites from which the original CFU data were colle cted. Furthermore, predicting sediment resuspension at significantly shallower dept hs, primarily where wave breaking and wave swash are involved, introduce an entirely new level of hydrodynamic complexity which is beyond the scope of this study. However, a very rough estimation of FIB resuspension in shallow water can be made from the data in this study, illust rating that there is potential for sediment-associated FIB to dominate water column concentrations, particularly at the upper bay si te. If we assume suspended sediment loads that are similar to those predicted in Table 12 but constrain the water to a much shallower depth that is
144 typical of the shoreline swash zone (e.g., 10-20 cm), the resulting bacterial load would be much more highly concentrated and becomes significant when compared to the regulatory standard of 104 CFU/100 mL (e.g., con centrating the same number of bacteria for the highest wind speeds at the upper bay site into 10 cm water depth results in an increase in enterococci concentrations of approximately 55 CFU / 100 mL). Therefore, in shallow depths along the extreme shorelin e (which is where water quality samples are often taken in practice), we can not rule out the possibility that resuspension from sediments may cause elevated waterborne conc entrations of FIB a nd potentially incorrect conclusions regarding microbial water quality. However, it is important to remember that such concentration of bacteria-l aden sediments in extremely sha llow depths is likely to be quickly mixed by wave action into the deeper water column and dilu ted to insignificant levels (Boehm, 2003; Boehm et al., 2005; Yamahara et al., 2007). Although I found SAV to be an insignificant re servoir of FIB in this study and omitted it from my resuspension estimates, that may not be the case at all sites. There are some key differences that the presence of SAV w ould have on resuspension. Firstly, SAVassociated bacteria would not be easily mode led by sediment resuspension, as they could only be resuspended as a result of mechanical shearing of the bacteria or the particles to which they are attached (e.g., detritus or epi phytic algae) from the surface of the SAV. Obviously, the physics of this process are like ly to be quite different from sediment resuspension and, to my knowledge, have never been examined theoretically or experimentally. Furthermore, in vegetated habitats the macrophytes interact with the
145 overlying flow to create increased turbulence in the water column and in the upper levels of the submerged canopy (Gambi et al., 1990; Ikeda and Kanazawa, 1996; Ghisalberti and Nepf, 2002), which can alte r the shear stress and likely also bacterial resuspension into the water column as compared to sediment alone. This effect would make the hydrodynamics that affect the SAV, and also any sediment beneath it, much more complex and difficult to model. Finally, it is important to not e that weather and hydrology ar e not the only drivers of sediment resuspension, especially in waters of recreational impor tance. Recreational activities themselves can cause increased susp ended sediment loads in the water column, particularly at the local s cale. Swimming and wading activity, for example, has been found to correlate with periods of high FIB concentrations in recr eational waters (Cheung et al., 1991; Crabill et al., 1999; Phillip et al., 2009). While the users themselves may act as a source (Elmir et al., 2007), it is also likely that they ar e disturbing FIB-laden sediments and causing resuspension into the water column, as was found experimentally by Phillip et al. (2009). Furthermore, it is widely known that recreational boating, particularly in shallow waters, can resuspe nd sediment from both the bottom (Garrad and Hey, 1987; Beachler and Hill, 2003; Lenzi et al., 2005) and the shoreline (Bauer et al., 2002) of water bodies, although it may not alwa ys be a significan t amount relative to natural mechanisms of resuspension such as waves and currents (Anthony and Downing, 2003). However, if areas of high boat traffi c overly sediments with high concentrations of FIB, this may constitute a significant source of FIB in the water column (An et al.,
146 2002). Resuspension of sediments from boats is most likely to be a bigger factor in areas with fine, cohesive sediments (< 60 M) and with larger boats traveling below planing speeds (Gucinski, 1982; B eachler and Hill, 2003).
147 LITERATURE CITED Aarestrup, F.M., Butaye, P., and Witte, W. (2002) Nonhuman reservoirs of enterococci. In The Enterococci: Pathogenesis, Molecu lar Biology, and Antib iotic Resistance Gilmore, M.S., Clewell, D.B., Cour valin, P., Dunny, G.M., Murray, B.E., and Rice, L.B. (eds). Washington, D. C.: ASM Press, pp. 55-99. Ahn, J.H., Grant, S.B., Surbeck, C.Q., Digiacomo, P.M., Nezlin, N.P., and Jiang, S. (2005) Coastal water quality impact of stormwater runoff from an urban watershed in southern California. Environmental Science & Technology 39 : 59405953. Allen, L.A., Grindley, J., and Brooks, E. (1953) Some chemical and bacterial characteristics of bottom deposits from lakes and estuaries. Journal of Hygiene 51 : 185-194. An, Y.J., Kampbell, D.H., and Breidenbac h, G.P. (2002) Escherichia coli and total coliforms in water and sediments at lake marinas. Environmental Pollution 120 : 771-778. Anderson, I.C., Rhodes, M., and Kato r, H. (1979) Sublethal Stress in Escherichia coli function of salinity. Applied and Environmental Microbiology 38 : 1147-1152. Anderson, K.L., Whitlock, J.E., Harwood, V.J., and K (2005) Persistence and differential survival of fecal indicator bacteria in subtropical waters and sediments. Applied and Environmental Microbiology 71 : 3041-3048. Anderson, M.A., Whitlock, J.E., and Harwood, V. J. (2006) Diversity and distribution of Escherichia coli genotypes and antibiotic resistance phenotypes in feces of humans, cattle, and horses. Applied and Environmental Microbiology 72 : 69146922. Anderson, S.A., Turner, S.J., and Lewis, G. D. (1997) Enterococci in the New Zealand environment: Implications for water quality monitoring. Water Science and Technology 35 : 325-331. Anthony, J.L., and Downing, J.A. (2003) Phys ical impacts of wind and boat traffic on Clear Lake, Iowa, USA. Lake and Reservoir Management 19 : 1-14.
148 APHA (1905) Standard Methods for the Examin ation of Water and Wastewater Washington, D. C.: American Public Health Association. APHA (1998) Standard Methods for the Examin ation of Water and Wastewater Washington, D. C.: American Public Health Association. Auer, M.T., and Niehaus, S.L. (1993) Mode ling Fecal-Coliform Bacteria .1. Field and Laboratory Determination of Loss Kinetics. Water Research 27 : 693-701. Bai, S., and Lung, W.S. (2005) Modeling sedi ment impact on the transport of fecal bacteria. Water Research 39 : 5232-5240. Bauer, B.O., Lorang, M.S., and Sherman, D.J. (2002) Estimating boat-wake-induced levee erosion using sediment suspension measurements. Journal of Waterway Port Coastal and Ocean Engineering-Asce 128 : 152-162. Beachler, M.M., and Hill, D.F. (2003) Stir ring up trouble? Resuspension of bottom sediments by recreational watercraft. Lake and Reservoir Management 19 : 15-25. Bell, S.S., Tewfik, A., Hall, M.O., and F onseca, M.S. (2008) Evaluation of seagrass planting and monitoring techniques: Impli cations for assessing restoration success and habitat equivalency. Restoration Ecology 16 : 407-416. Bernhard, A.E., and Field, K.G. (2000) A PCR assay to discriminate human and ruminant feces on the basis of host differences in Bacteroides-Prevotella genes encoding 16S rRNA. Applied and Environmental Microbiology 66 : 4571-4574. Beversdorf, L.J., Bornstein-Forst, S.M., and McLellan, S.L. (2007) The potential for beach sand to serve as a reservoir for Escherichia coli and the physical influences on cell die-off. Journal of Applied Microbiology 102 : 1372-1381. Boehm, A.B. (2003) Model of microbial transport and inactiv ation in the surf zone and application to field measurements of to tal coliform in Northern Orange County, California. Environmental Science & Technology 37 : 5511-5517. Boehm, A.B. (2007) Enterococci concentrations in diverse coastal environments exhibit extreme variability. Environmental Science & Technology 41 : 8227-8232. Boehm, A.B., Shellenbarger, G.G., and Pa ytan, A. (2004) Groundw ater discharge: potential association with fecal indi cator bacteria in the surf zone. Environmental Science & Technology 38 : 3558-3566. Boehm, A.B., Keymer, D.P., and Shellenbarge r, G.G. (2005) An analytical model of enterococci inactivation, grazing, and transpor t in the surf zone of a marine beach. Water Research 39 : 3565-3578.
149 Boehm, A.B., Griffith, J.F., Mcgee, C., Edge, T.A., Solo-Gabriele, H.M., Whitman, R.L. et al. (2009) Fecal indicator bacteria enumeration in beach sand: a comparison study of extraction methods in medium to coarse sands. Journal of Applied Microbiology Bonadonna, L., Briancesco, R., Ottaviani, M ., and Veschetti, E. (2002) Occurrence of Cryptosporidium oocysts in sewage effl uents and correlation with microbial, chemical and physical water variables. Environmental Monitoring and Assessment 75 : 241-252. Bonde, G.J. (1967) Pollution of a marine environment. Journal Water Pollution Control Federation 39 : R45-&-R45-&. Bonilla, T.D., Nowosielski, K., Cuvelier, M., Hartz, A., Green, M., Esiobu, N. et al. (2007) Prevalence and distribution of fecal indicator organisms in South Florida beach sand and preliminary assessment of health effects associated with beach sand exposure. Marine Pollution Bulletin 54 : 1472-1482. Brownell, M.J., Harwood, V.J., Kurz, R.C., Mc Quaig, S.M., Lukasik, J., and Scott, T.M. (2007) Confirmation of putative stormwater impact on water quality at a Florida beach by microbial source tracking methods and structure of indicator organism populations. Water Research 41 : 3747-3757. Buckley, R., Clough, E., Warnken, W., and W ild, C. (1998) Coliform bacteria in streambed sediments in a subtropica l rainforest conservation reserve. Water Research 32 : 1852-1856. Burton, G.A., Gunnison, D., and Lanza, G.R. ( 1987) Survival of pathogenic bacteria in various fresh-water sediments. Applied and Environmental Microbiology 53 : 633638. Byappanahalli, M., Fowler, M., Shively, D., and Whitman, R. (2003a) Ubiquity and persistence of Escherichia coli in a Midwestern coastal stream. Applied and Environmental Microbiology 69 : 4549-4555. Byappanahalli, M.N., and Fujioka, R.S. (1998) Evidence that tropical soil environment can support the growth of Escherichia coli Water Science and Technology 38 : 171-174. Byappanahalli, M.N., Shively, D.A., Nevers, M.B., Sadowsky, M.J., and Whitman, R.L. (2003b) Growth and survival of Escherichia coli and enterococci populations in the macro-alga Cladophora (Chlorophyta). FEMS Microbiology Ecology 46 : 203211.
150 Byappanahalli, M.N., Richard, L.W., Shivelya D.A., John, F., Ishii, S., and Sadowsky, M.J. (2007) Population stru cture of cladophora-borne Escherichia coli in nearshore water of lake Michigan. Water Research 41 : 3649-3654. Cabelli, V.J., Dufour, A.P., McCabe, L.J., and Levin, M.A. (1982) Swimming-associated gastroenteritis a nd water quality. American Journal of Epidemiology 115 : 606616. Carson, C.A., Shear, B.L., Ellersieck, M.R., a nd Asfaw, A. (2001) Identification of fecal Escherichia coli from humans and animals by ribotyping. Applied and Environmental Microbiology 67 : 1503-1507. Cheung, W.H., Chang, K.C., and Hung, R.P. ( 1991) Variations in microbial indicator densities in beach waters and health-relat ed assessment of bathing water quality. Epidemiology and Infection 106 : 329-344. Cheung, W.H., Chang, K.C., Hung, R.P., and Kl eevens, J.W. (1990) Health effects of beach water pollution in Hong Kong. Epidemiology and Infection 105 : 139-162. Crabill, C., Donald, R., Snelling, J., Foust, R., and Southam, G. (1999) The impact of sediment fecal coliform reservoirs on seasonal water quality in Oak Creek, Arizona. Water Research 33 : 2163-2171. Craig, D.L., Fallowfield, H.J., and Cromar, N. J. (2004) Use of macrocosms to determine persistence of Escherichia coli in recreational coastal water and sediment and validation with in situ measurements. Journal of Applied Bacteriology 96 : 922930. Davies, C.M., and Evison, L.M. (1991) Sunlight and the survival of enteric bacteria in natural waters. Journal of Applied Bacteriology 70 : 265-274. Davies, C.M., and Bavor, H.J. (2000) The fate of stormwater-associ ated bacteria in constructed wetland and water pol lution control pond systems. Journal of Applied Microbiology 89 : 349-360. Davies, C.M., Long, J.A.H., Donald, M., and Ashbolt, N.J. (1995) Survival of fecal microorganisms in marine and fresh-water sediments. Applied and Environmental Microbiology 61 : 1888-1896. de Vaux, A., Laguerre, G., Divies, C., and Pre vost, H. (1998) Enterococcus asini sp. nov. isolated from the caecum of donkeys (Equus asinus). International Journal of Systematic Bacteriology 48: 383-387. Del Rio-Rodriquez, R.E., Inglis, V., and Millar, S.D. (2008) Survival of Escherichia coli in the intestine of fish. Aquaculture Research 28 : 257-264.
151 Denny, M.W. (1988) Biology and the Mechanics of the Wave-Swept Environment Princeton, NJ: Princeton University Press. Desmarais, T.R., Solo-Gabriele, H.M., and Pa lmer, C.J. (2002) Influence of soil on fecal indicator organisms in a tidally in fluenced subtropical environment. Applied and Environmental Microbiology 68 : 1165-1172. Dethier, M.N., Graham, E.S., Cohen, S., a nd Tear, L.M. (1993) Visual Versus RandomPoint Percent Cover Estimations Objective Is Not Always Better. Marine Ecology-Progress Series 96 : 93-100. Devriese, L.A., Ceyssens, K., Rodrigues, U.M., and Collins, M.D. (1990) EnterococcusColumbae, a Species from Pigeon Intestines. Fems Microbiology Letters 71 : 247252. Dombek, P.E., Johnson, L.K., Zimmerley, S. T., and Sadowsky, M.J. (2000) Use of repetitive DNA sequences and the PCR to differentiate Escherichia coli isolates from human and animal sources. Applied and Environmental Microbiology 66 : 2572-2577. Doyle, J.D., Tunnicliff, B., Kramer, R., Kuehl, R., and Brickler, S.K. (1992) Instability of fecal-coliform populations in waters and bottom sediments at recreational beaches in Arizona. Water Research 26 : 979-988. Elmir, S.M., Wright, M.E., Abdelzaher, A., Solo-Gabriele, H.M., Fleming, L.E., Miller, G. et al. (2007) Quantitative evaluation of bacteria releas ed by bathers in a marine water. Water Research 41 : 3-10. Englebert, E.T., McDermott, C., and Klei nheinz, G.T. (2008) Impact of the alga Cladophora on the survival of E. coli Salmonella and Shigella in a laboratory microcosm. Journal of Great Lakes Research 34 : 377-382. Facklam, R.R., Carvahallo, M.S., and Te ixeira, L.M. (2002) History, taxonomy, biochemical characteristics, biochemi cal charasteristics, and antibiotic susceptibility testing of enterococci. In The Enterococci: Pathogenesis, Molecular Biology, and Antibiotic Resistance Gilmore, M.S., Clewell, D.B., Courvalin, P., Dunny, G.M., Murray, B.E., and Rice, L.B. (e ds). Washington, D. C.: ASM Press, pp. 1-54. Ferguson, D.M., Moore, D.F., Getrich, M.A., and Zhowandai, M.H. (2005) Enumeration and speciation of enterococci found in marine and inter tidal sediments and coastal water in southern California. Journal of Applied Bacteriology 99 : 598-608. Field, K.G., and Samadpour, M. (2007) Fecal source tracking, the indicator paradigm, and managing water quality. Water Research 41 : 3517-3538.
152 Field, K.G., Bernhard, A.E., and Brodeur, T.J. (2003) Molecular approaches to microbiological monitoring: fecal source detection. Environmental Monitoring and Assessment 81 : 313-326. Flint, K.P. (1987) The long-term survival of Escherichia coli in river water. Journal of Applied Bacteriology 63 : 261-270. Fonseca, M. (1996) The role of seagrasses in nearshore sedimentary processes: a review. In Estuarine Shores: Evolution, Environments and Human Alterations Nordstrom, K., and Roman, C.T. (eds). London: Wiley, pp. 261-286. Fourqurean, J.W., Willsie, A., Rose, C.D., a nd Rutten, L.M. (2001) Spatial and temporal pattern in seagrass community compositi on and productivity in south Florida. Marine Biology 138 : 341-354. Fujioka, R., Sian-Denton, C., Borja, M., Ca stro, J., and Morphew, K. (1999) Soil: the environmental source of Escherichia coli and enterococci in Guam's streams. Journal of Applied Microbiology 85 : 83S-89S-83S-89S. Fujioka, R.S., and Shizumura, L.K. (1985) Cl ostridium-Perfringens, a Reliable Indicator of Stream Water-Quality. Journal Water Pollution Control Federation 57 : 986992. Fujioka, R.S., and Yoneyama, B.S. (2002) Sun light inactivation of human enteric viruses and fecal bacteria. Water Science and Technology 46 : 291-295. Gambi, M.C., Nowell, A.R.M., and Jumars P.A. (1990) Flume observations on flow dynamics in Zostera marina (eelgrass) beds. Marine Ecology Progress Series 61 : 159-169. Gannon, J.J., Busse, M.K., and Schillinger, J.E. (1983) Fecal coliform disappearance in a river impoundment. Water Research 17 : 1595-1601. Gantzer, C., Maul, A., Audic, J.M., and Schwar tzbrod, L. (1998) Dete ction of infectious enteroviruses, enterovirus genomes, somatic coliphages, and Bacteroides fragilis phages in treated wastewater. Applied and Environmental Microbiology 64 : 43074312. Garrad, P.N., and Hey, R.D. (1987) Boat Tra ffic, Sediment Resuspension and Turbidity in a Broadland River. Journal of Hydrology 95 : 289-297. Gaun, S.K., Xu, R.L., Chen, S., Odumeru, J., and Gyles, C. (2002) Development of a procedure for discriminating among Escherichia coli isolates from animal and human sources. Applied and Environmental Microbiology 68 : 2690-2698.
153 Geldreich, E.E., Clark, H.F., and Huff, C.B. (1964) A study of pollu tion indicators in a waste stabilization pond. Journal Water Pollution Control Federation 36 : 13721379. Gelsomino, R., Vancanneyt, M., Cogan, T.M., a nd Swings, J. (2003) Effect of raw-milk cheese consumption on the entero coccal flora of human feces. Applied and Environmental Microbiology 69 : 312-319. Gerba, C.P., and McLeod, J.S. (1976) Effects of sediments on survival of Escherichia coli in marine waters. Applied and Environm ental Microbiology 32 : 114-120. Ghisalberti, M., and Nepf, H.M. (2002) Mi xing layers and coherent structures in vegetated aquatic flows. Journal of Geophysical Research Oceans 107 Gordon, D.M., Bauer, S., and Johnson, J. R. (2002) The genetic structure of Escherichia coli populations in primary and secondary habitats. Microbiology-Sgm 148 : 15131522. Gotelli, N.J., and Entsminger, G.L. (2009). EcoSim: null models software for ecology, version 7. URL http://garyentsminger.com/ecosim.htm. Grant, S.B., Sanders, B.F., Boehm, A.B., Re dman, J.A., Kim, J.H., Mrse, R.D. et al. (2001) Generation of enterococci bacteria in a coastal saltwater marsh and its impact on surf zone water quality. Environmental Science & Technology 35 : 2407-2416. Griffin, D.W., Lipp, E.K., McLa ughlin, M.R., and Rose, J.B. (2001) Marine recreation and public health microbiology: Quest for the ideal indicator. Bioscience 51 : 817825. Grimes, D.J. (1975) Release of sedi ment-bound fecal coliforms by dredging. Applied Microbiology 29 : 109-111. Grimes, D.J. (1980) Bacteriological wate r quality effects of hydraulically dredging contaminated upper mississippi river bottom sediment. Applied and Environmental Microbiology 39 : 782-789. Gucinski, H. (1982) Sediment Suspension a nd Resuspensoion From Small Craft Induced Turbulence. In. Washington, D. C.: U. S. Environmental Protection Agency, p. 61. Hagedorn, C., Crozier, J.B., Mentz, K.A., B ooth, A.M., Graves, A.K., Nelson, N.J., and Reneau, R.B. (2003) Carbon source utiliz ation profiles as a method to identify sources of faecal pollution in water. Journal of Applied Microbiology 94 : 792799.
154 Hansen, D.L., Clark, J.J., Ishii, S., Sadows ky, M.J., and Hicks, R.E. (2008) Sources and sinks of Escherichia coli in benthic and pelagic fish. Journal of Great Lakes Research 34 : 228-234. Hansen, J.W., Udy, J.W., Perry, C.J., Dennis on, W.C., and Lomstein, B.A. (2000) Effect of the seagrass Zostera capricorni on sediment microbial processes. Marine Ecology Progress Series 199 : 83-96. Hardina, C.M., and Fujioka, R.S. (1991) Soil -the environmental source of Escherichia coli and enterococci in Hawaii streams. Environmental Toxicology and Water Quality 6 : 185-195. Harwood, V.J., Whitlock, J., and Withington, V. (2000) Classification of antibiotic resistance patterns of indicator bacter ia by discriminant analysis: Use in predicting the source of fecal cont amination in subtropical waters. Applied and Environmental Microbiology 66 : 3698-3704. Harwood, V.J., Delahoya, N.C., Ulrich, R.M., Kramer, M.F., Whitlock, J.E., Garey, J.R., and Lim, D.V. (2004) Molecular confirmation of Enterococcus faecalis and E. faecium from clinical, faecal and environmental sources. Letters in Applied Microbiology 38 : 476-482. Harwood, V.J., Levine, A.D., Scott, T.M., Ch ivukula, V., Lukasik, J., Farrah, S.R., and Rose, J.B. (2005) Validity of the indi cator organism paradigm for pathogen reduction in reclaimed water and public health protection. Applied and Environmental Microbiology 71 : 3163-3170. Heaney, C.D., Sams, E., Wing, S., Marshall, S., Brenner, K., Dufour, A.P., and Wade, T.J. (2009) Contact with beach sand among beachgoers and risk of illness. American Journal of Epidemiology 170 : 164-172. Hood, M.A., and Ness, G.E. (1982) Survival of Vibrio cholerae and Escherichia coli in estuarine waters and sediments. Applied and Environmental Microbiology 43 : 578-584. Howell, J.M., Coyne, M.S., and Cornelius, P.L. (1996) Effect of sedi ment particle size and temperature on fecal bacteria morta lity rates and the fecal coliform/fecal streptococci ratio. Journal of Environmental Quality 25 : 1216-1220. Ikeda, S., and Kanazawa, M. (1996) Three-di mensional organized vortices above flexible water plants. Journal of Hydraulic Engineering 122 : 634-640. Ishii, S., and Sadowsky, M.J. (2008) Escherichia coli in the environment: Implications for water quality and human health. Microbes and Environments 23 : 101-108.
155 Ishii, S., and Sadowsky, M.J. (2009) Appli cations of the rep-PCR DNA fingerprinting technique to study microbial di versity, ecology and evolution. Environmental Microbiology 11 : 733-740. Ishii, S., Ksoll, W.B., Hicks, R.E., and Sa dowsky, M.J. (2006) Pres ence and growth of naturalized Escherichia coli in temperate soils from la ke superior watersheds. Applied and Environmental Microbiology 72 : 612-621. Ishii, S., Hansen, D.L., Hicks, R.E., and Sadowsky, M.J. (2007) Beach sand and sediments are temporal sinks and sources of Escherichia coli in lake superior. Environmental Science & Technology 41 : 2203-2209. Jamieson, R.C., Joy, D.M., Lee, H., Kostaschuk, R., and Gordon, R.J. (2005) Resuspension of sediment-associated Escherichia coli in a natural stream. Journal of Environmental Quality 34 : 581-589. Jeng, H.W.C., England, A.J., and Bradford, H. B. (2005) Indicator organisms associated with stormwater suspended part icles and estuarine sediment. Journal of Environmental Science and Health Part A-Toxic/Hazardous 40 : 779-791. Johnson, L.K., Brown, M.B., Carruthers, E. A., Ferguson, J.A., Dombek, P.E., and Sadowsky, M.J. (2004) Sample size, libra ry composition, and genotypic diversity among natural populations of Escherichia coli from different animals influence accuracy of determining s ources of fecal pollution. Applied and Environmental Microbiology 70 : 4478-4485. Karjalainen, H., Stefansdotti r, G., Tuominen, L., and Kairesalo, T. (2001) Do submersed plants enhance microbial activity in sediment? Aquatic Botany 69 : 1-13. Kinzelman, J., McLellan, S.L., Daniels, A. D., Cashin, S., Singh, A., Gradus, S., and Bagley, R. (2004) Non-point source polluti on: Determination of replication versus persistence of Escherichia coli in surface water and sediments with correlation of levels to readily measurable environmental parameters. Journal of Water and Health 2 : 103-114. Kleinheinz, G.T., Coenan, A., Zehms, T., Preedit, J., Leewis, M.C., Becker, D., and McDermott, C. (2009) Effect of aquatic macrophytes on the survival of Escherichia coli in a laboratory microcosm. Lake and Reservoir Management 25 : 149-154. Koeuth, T., Versalovic, J., and Lupski, J.R. (1995) Differential s ubsequence conservation of interspersed repetitive Streptococcus pneumoniae BOX elements in diverse bacteria. Genome Research 5: 408-418.
156 Kogan, S.C., Doherty, M., and Gitschier, J. (1987) An improved method for prenatal diagnosis of genetic diseases by an alysis of amplified DNA sequences -Application to Hemophilia-A. New England Journal of Medicine 317 : 985-990. Ksoll, W.B., Ishii, S., Sadowsky, M.J., and Hi cks, R.E. (2007) Presence and sources of fecal coliform bacteria in epilithic periphyton communities of lake superior. Applied and Environmental Microbiology 73 : 3771-3778. Laliberte, P., and Grimes, D.J. (1982) Survival of Escherichia coli in lake bottom sediment. Applied and Environmental Microbiology 43 : 623-628. Lane, D.J. (1991) 16S/23S rRNA sequencing. In Nucleic Acie Techniques in Bacterial Systematics Stackebrandt, E., and Goodfellow, M. (eds). Chichester, United Kingdom: Wiley, pp. 115-148. Le Fevre, N.M., and Lewis, G.D. (2003) The role of resuspension in enterococci distribution in water at an urban beach. Water Science and Technology 47 : 205210. Le Roux, J.P. (2005) Grains in motion: A review. Sedimentary Geology 178 : 285-313. Leclerc, H., Devriese, L.A., and Mossel, D.A.A. (1996) Taxonomical changes in intestinal (faecal) enteroco cci and streptococci: Consequences on their use as indicators of faecal contamination in drinking water. Journal of Applied Bacteriology 81 : 459-466. Leclerc, H., Mossel, D.A.A., Edberg, S.C., and Struijk, C.B. (2001) Advances in the bacteriology of the Coliform Group: Thei r suitability as markers of microbial water safety. Annual review of Microbiology 55 : 201-234. Lemarchand, K., and Lebaron, P. (2003) Occurrence of Salmonella spp. and Cryptosporidium spp. in a French coastal watershed: relationship with fecal indicators. Fems Microbiology Letters 218 : 203-209. Lenzi, M., Finoia, M.G., Persia, E., Comandi, S., Gargiulo, V., Solari, D. et al. (2005) Biogeochemical effects of disturbance in shallow water sediment by macroalgae harvesting boats. Marine Pollution Bulletin 50 : 512-519. Lleo, M.D., Tafi, M.C., and Cane pari, P. (1998) Nonculturable Enterococcus faecalis cells are metabolically active and capable of resuming active growth. Systematic and Applied Microbiology 21 : 333-339. Lopez, N.I., Duarte, C.M., Vallespinos, F., Romero, J., and Alcoverro, T. (1995) Bacterial activity in NW Mediterranean seagrass ( Posidonia oceanica ) sediments. Journal of Experimental Marine Biology and Ecology 187: 39-49.
157 Lund, V. (1996) Evaluation of E-coli as an i ndicator for the presence of Campylobacter jejuni and Yersinia enterocolitica in chlo rinated and untreated oligotrophic lake water. Water Research 30 : 1528-1534. Malathum, K., Singh, K.V., Weinstock, G. M., and Murray, B.E. (1998) Repetitive sequence-based PCR versus pulsed-fiel d gel electrophoresis for typing of Enterococcus faecalis at the subspecies level. Journal of Clinical Microbiology 36 : 211-215. Manero, A., Vilanova, X., Cerda-Cuellar, M ., and Blanch, A.R. (2002) Characterization of sewage waters by biochemical fingerprinting of Enterococci. Water Research 36 : 2831-2835. Marino, R.P., and Gannon, J.J. (1991) Survival of Fecal-Coliforms and Fecal Streptococci in Storm-Drain Sediment. Water Research 25 : 1089-1098. McDonald, A., Kay, D., and Jenkins, A. (1982) Generation of fecal and total coliform surges by stream flow manipulation in the absence of normal hydrometerological stimuli. Applied and Environmental Microbiology 44 : 292-300. McDonald, J.I., Coupland, G.T., and Kendric k, G.A. (2006) Underwater video as a monitoring tool to detect change in seagrass cover. Journal of Environmental Management 80 : 148-155. McLellan, S.L. (2004) Genetic diversity of Escherichia coli isolated from urban rivers and beach water. Applied and Environmental Microbiology 70 : 4658-4665. McLellan, S.L., and Salmore, A.K. (2003) Evid ence for localized bact erial loading as the cause of chronic beach closi ngs in a freshwater marina. Water Research 37 : 27002708. McQuaig, S.M., Scott, T.M., Harwood, V.J., Farrah, S.R., and Lukasik, J.O. (2006) Detection of human-derived fecal polluti on in environmental waters by use of a PCR-based human polyomavirus assay. Applied and Environmental Microbiology 72 : 7567-7574. Moore, D.F., Guzman, J.A., and Mcgee, C. (2008) Species distribution and antimicrobial resistance of enterococci isolated from surface and ocean water. Journal of Applied Microbiology 105 : 1017-1025. Muela, A., Garcia-Bringas, J.M., Arana, I., a nd Barcina, I. (2000) Th e effect of simulated solar radiation on Escherichia coli : The relative roles of UV-B, UV-A, and photosynthetically active radiation. Microbial Ecology 39 : 65-71.
158 Muirhead, R.W., Davies-Colley, R.J., Donnis on, A.M., and Nagels, J.W. (2004) Faecal bacteria yields in artificial flood events: quantifying in-stream stores. Water Research 38 : 1215-1224. Muller, T., Ulrich, A., Ott, E. M., and Muller, M. (2001) Identif ication of plant-associated enterococci. Journal of Applied Microbiology 91 : 268-278. Nagels, J.W., Davies-Colley, R.J., Donnis on, A.M., and Muirhead, R.W. (2002) Faecal contamination over flood events in a pastor al agricultural stream in New Zealand. Water Science and Technology 45 : 45-52. Nasser, A.M., Telser, L., and Nitzan, Y. (2007) Effect of sunlight on the infectivity of Cryptosporidium parvum in seawater. Canadian Journal of Microbiology 53 : 1101-1105. Obiri-Danso, K., and Jones, K. (1999) Di stribution and season ality of microbial indicators and thermophilic campylobacters in two freshwater bathing sites on the River Lune in northwest England. Journal of Applied Microbiology 87 : 822-832. Olapade, O.A., Depas, M.M., Jensen, E. T., and McLellan, S.L. (2006) Microbial communities and fecal indicator bacteria associated with Cladophora mats on beach sites along Lake Michigan shores. Applied and Environmental Microbiology 72 : 1932-1938. Oren, A. (2004) Prokaryote dive rsity and taxonomy: current stat us and future challenges. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 359 : 623-638. Ott, E.M., Muller, T., Muller, M., Franz, C., Ulrich, A., Gabel, M., and Seyfarth, W. (2001) Population dynamics and antagonistic potential of enterococci colonizing the phyllosphere of grasses. Journal of Applied Microbiology 91 : 54-66. Pachepsky, Y.A., Sadeghi, A.M., Bradford, S.A., Shelton, D.R., Guber, A.K., and Dao, T. (2006) Transport and fate of manure -borne pathogens: Modeling perspective. Agricultural Water Management 86 : 81-92. Parveen, S., Portier, K.M., Robinson, K ., Edmiston, L., and Tamplin, M.L. (1999) Discriminant analysis of ribotype profiles of Escherichia coli for differentiating human and nonhuman sources of fecal pollution. Applied and Environmental Microbiology 65 : 3142-3147. Parveen, S., Murphree, R.L., Edmiston, L., Ka spar, C.W., Portier, K.M., and Tamplin, M.L. (1997) Association of multiple-antib iotic-resistance profiles with point and nonpoint sources of Escherichia coli in Apalachicola bay. Applied and Environmental Microbiology 63 : 2607-2612.
159 Payment, P., and Franco, E. (1993) Clostrid ium-Perfringens and Somatic Coliphages as Indicators of the Efficiency of Dr inking-Water Treatment for Viruses and Protozoan Cysts. Applied and Environm ental Microbiology 59 : 2418-2424. Pettibone, G.W., Irvine, K.N., and Monahan, K.M. (1996) Impact of a ship passage on bacteria levels and suspended sediment characteristics in the Buffalo River, New York. Water Research 30 : 2517-2521. Phillip, D.A.T., Antoine, P., Cooper, V., Fran cis, L., Mangal, E., Seepersad, N. et al. (2009) Impact of recreation on recreati onal water quality of a small tropical stream. Journal of Environmental Monitoring 11 : 1192-1198. Pinto, B., Pierotti, R., Canale, G., and R eali, D. (1999) Characterization of 'faecal streptococci' as indicators of faecal pollu tion and distribution in the environment. Letters in Applied Microbiology 29 : 258-263. Pollard, P.C., and Moriarty, D.J.W. (1991) Organic-Carbon Decomposition, Primary and Bacterial Productivity, and Su lfate Reduction, in Tropica l Seagrass Beds of the Gulf of Carpentaria, Australia. Marine Ecology-Progress Series 69 : 149-159. Posey, M.H., Wigand, C., and Stevenson, J.C. (1993) Effects of an Introduced Aquatic Plant, Hydrilla-Verticillata, on Benthi c Communities in the Upper Chesapeake Bay. Estuarine Coastal and Shelf Science 37 : 539-555. Power, M.L., Littlefield-Wyer, J., Gordon, D.M., Veal, D.A., and Slade, M.B. (2005) Phenotypic and genotypic charact erization of encapsulated Escherichia coli isolated from blooms in two Australian lakes. Environmental Microbiology 7 : 631-640. Ram, J.L., Ritchie, R.P., Fang, J.W., Gonzal es, F.S., and Selegean, J.P. (2004) Sequencebased source tracking of Escherichia coli based on genetic diversity of betaglucuronidase. Journal of Environmental Quality 33 : 1024-1032. Rees, G., Pond, K., Johal, K., Pedley, S., and Rickards, A. (1998) Microbiological analysis of selected coastal bathing wa ters in the UK, Greece, Italy and Spain. Water Research 32 : 2335-2340. Reeves, R.L., Grant, S.B., Mrse, R.D., Oan cea, C.M.C., Sanders, B.F., and Boehm, A.B. (2004) Scaling and management of fecal indicator bacteria in runoff from a coastal urban watershed in southern California. Environmental Science & Technology 38 : 2637-2648. Rittenberg, S.C., Mittwer, T., and Ivler, D. (1958) Coliform bacteria in sediments around three marine sewage outfalls. Limnology and Oceanography 3 : 101-108.
160 Roll, B.M., and Fujioka, R.S. (1997) Sources of faecal indicator bacteria in a brackish, tropical stream and their imp act on recreational water quality. Water Science and Technology 35 : 179-186. Rossello-Mora, R., and Amann, R. (2001) The species concept for prokaryotes. FEMS Microbiology Reviews 25 : 39-67. Ruoff, K.L., Delamaza, L., Murtagh, M.J., Sp argo, J.D., and Ferraro, M.J. (1990) Species identities of enterococci isolated from clinical specimens. Journal of Clinical Microbiology 28 : 435-437. Sanders, B.F., Arega, F., and Sutula, M. (2005) Modeling the dry-weat her tidal cycling of fecal indicator bacteria in surface waters of an intertidal wetland. Water Research 39 : 3394-3408. Savage, W.G. (1905) Bacteriological examinati on of tidal mud as an index of pollution of the river. Journal of Hygiene 5 : 146-174. Savageau, M.A. (1983) Escherichia coli habitats, cell types, a nd molecular methods of gene control. American Naturalist 122 : 732-744. Schleifer, K.H., and Kilpperb alz, R. (1984) Transfer of Streptococcus faecalis and Streptococcus faecium to the genus Enterococcus Nom Rev as Enterococcus faecalis Comb-Nov and Enterococcus faecium Comb-Nov. International Journal of Systematic Bacteriology 34 : 31-34. Schultz-Fademrecht, C., Wich ern, M., and Horn, H. (2008) The impact of sunlight on inactivation of indicator microorganisms bot h in river water and benthic biofilms. Water Research 42 : 4771-4779. Scott, T.M., Jenkins, T.M., Lukasik, J., a nd Rose, J.B. (2005) Potential use of a host associated molecular marker in Enteroco ccus faecium as an index of human fecal pollution. Environmental Science & Technology 39 : 283-287. Seurinck, S., Verstraete, W., and Siciliano, S. D. (2003) Use of 16S-23S rRNA intergenic spacer region PCR and repetitive extrag enic palindromic PCR analyses of Escherichia coli isolates to identify nonpoint fecal sources. Applied and Environmental Microbiology 69 : 4942-4950. Shanks, O.C., Kelty, C.A., Sivaganesan, M ., Varma, M., and Haugland, R.A. (2009) Quantitative PCR for genetic markers of human fecal pollution. Applied and Environmental Microbiology Sherer, B.M., Miner, J.R., Moore, J.A., a nd Buckhouse, J.C. (1992) Indicator bacterial survival in stream sediments. Journal of Environmental Quality 21 : 591-595.
161 Shiaris, M.P., Rex, A.C., Pettibone, G.W., Keay, K., McManus, P., Rex, M.A. et al. (1987) Distribution of in dicator bacteria and Vibrio parahaemolyticus in sewagepollluted intertidal sediments. Applied and Environmental Microbiology 53 : 1756-1761. Shibata, T., Solo-Gabriele, H.M., Fleming, L. E., and Elmir, S. (2004) Monitoring marine recreational water quality using multiple mi crobial indicators in an urban tropical environment. Water Research 38 : 3119-3131. Shimeta, J., Amos, C.L., Beaulieu, S.E., and Ashiru, O.M. (2002) Sequential resuspension of protists by accelerating tidal flow: Implications for community structure in the benthic boundary layer. Limnology and Oceanography 47 : 11521164. Signoretto, C., Burlacchini, G., Pruzzo, C., and Canepari, P. (2005) Persistence of Enterococcus faecalis in aquatic environments vi a surface interactions with copepods. Applied and Environmental Microbiology 71 : 2756-2761. Signoretto, C., Burlacchini, G., Lleo, M.D., Pr uzzo, C., Zampini, M., Pane, L. et al. (2004) Adhesion of Enterococcus faecalis in the nonculturable state to plankton is the main mechanism responsible for persiste nce of this bacterium in both lake and seawater. Applied and Environm ental Microbiology 70 : 6892-6896. Simpson, J.M., Santo Domingo, J.W., and R easoner, D.J. (2002) Microbial source tracking: state of the science. Environmental Science & Technology 36 : 52795288. Sinton, L.W., Hall, C.H., Lynch, P.A., a nd Davies-Colley, R.J. (2002) Sunlight inactivation of fecal i ndicator bacteria and bacteriophages from waste stabilization pond effluent in fresh and saline waters. Applied and Environmental Microbiology 68 : 1122-1131. Solo-Gabriele, H.M., Wolfert, M.A., Desmarai s, T.R., and Palmer, C.J. (2000) Sources of Escherichia coli in a coastal subtropical environment. Applied and Environmental Microbiology 66 : 230-237. Soulsby, R. (1998) Dynamics of Marine Sands London: Thomas Telford, Ltd. Stackebrandt, E., Frederiksen, W., Garrity, G. M., Grimont, P.A.D., Kampfer, P., Maiden, M.C.J. et al. (2002) Report of the ad hoc committee for the re-evaluation of the species definition in bacteriology. International Journal of Systematic and Evolutionary Microbiology 52 : 1043-1047. Steets, B.M., and Holden, P.A. (2003) A mech anistic model of runoff-associated fecal coliform fate and transpor t through a coastal lagoon. Water Research 37 : 589608.
162 Stephenson, G.R., and Street, L.V. (1978) Bacterial variations in streams from a southwest Idaho rangeland watershed. Journal of Environmental Quality 7 : 150156. Stoeckel, D.M., and Harwood, V.J. (2007) Perfor mance, design, and analysis in microbial source tracking studies. Applied and Environmental Microbiology 73 : 2405-2415. Tallon, P., Magajna, B., Lofranco, C., and Leung, K.T. (2005) Micr obial indicators of faecal contamination in water: A current perspective. Water Air and Soil Pollution 166 : 139-166. Topp, E., Welsh, M., Tien, Y.C., Dang, A., Lazarovits, G., Conn, K., and Zhu, H. (2003) Strain-dependent variability in growth and survival of Escherichia coli in agricultural soil. FEMS Microbiology Ecology 44 : 303-308. Tunnicliff, B., and Brickler, S.K. (1984) R ecreational water-quality analyses of the Colorado River in Grand Canyon. Applied and Environm ental Microbiology 48 : 909-917. USACE (2002) Coastal Engineering Manual. In Washington, D. C.: U. S. Army Corps of Engineers. USEPA (1986) Ambient Water Quality Criteria for Bacteria 1986. In: Office of Water, U. S. Environmental Protection Agency. USEPA (2000) Improved Enumeration Met hods for the Recreational Water Quality Indicators: Enterococci and Escheria coli In. Washington, D. C.: Office of Water, U. S. Environmental Protection Agency. USEPA (2005) Microbial Source Tracking Guide Document. In. Washington, D. C.: National Risk Management Laboratory, Un ited States Environmental Protection Agency. Vandonsel, D.J., and Geldreich, E. E. (1971) Relationships of Salmonella to fecal coliforms in bottom sediments. Water Research 5 : 1079-+-1079-+. Versalovic, J., Koeuth, T., and Lupski, J. R. (1991) Distributi on of repetitive DNA sequences in Eubacteria a nd application to fingerprint ing of bacterial genomes. Nucleic Acids Research 19 : 6823-6831. Wade, T.J., Pai, N., Eisenberg, J.N.S., and Co lford, 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 : 1102-1109.
163 Ward, D.M., Ferris, M.J., Nold, S.C., and Ba teson, M.M. (1998) A natural view of microbial biodiversity within hot spring cyanobacterial mat communities. Microbiology and Molecular Biology Reviews 62 : 1353-+. Whitlock, J.E., Jones, D.T., and Harwood, V.J. (2002) Identification of the sources of fecal coliforms in an urban watershed using antibiotic resistance analysis. Water Research 36 : 4273-4282. Whitman, R.L., and Nevers, M.B. (2 003) Foreshore sand as a source of Escherichia coli in nearshore water of a Lake Michigan beach. Applied and Environmental Microbiology 69 : 5555-5562. Whitman, R.L., Nevers, M.B., and Byappana halli, M.N. (2006) Examination of the watershed-wide distribution of Escherichia coli along southern Lake Michigan: an integrated approach. Applied and Environmental Microbiology 72 : 7301-7310. Whitman, R.L., Shively, D.A., Pawlik, H., Nevers, M.B., and Byappanahalli, M.N. (2003) Occurrence of Escherichia coli and enterococci in Cladophora ( Chlorophyta ) in nearshore water and beach sand of Lake Michigan. Applied and Environmental Microbiology 69 : 4714-4719. WHO (2001) Bathing Water Qua lity and Human Health: Faecal Pollution. In. Geneva, Switzerland: World Health Organization, p. 58. Wiggins, B.A., Andrews, R.W., Conway, R.A ., Corr, C.L., Dobratz, E.J., Dougherty, D.P. et al. (1999) Use of antibiotic resistance analysis to identify nonpoint sources of fecal pollution. Applied and Environmental Microbiology 65 : 3483-3486. Wilkinson, J., Jenkins, A., Wyer, M., a nd Kay, D. (1995) Modeling fecal coliform dynamics in streams and rivers. Water Research 29 : 847-855. Yamahara, K.M., Walters, S.P., and Boehm, A.B. (2009) Growth of Enterococci in Unaltered, Unseeded Beach Sands Subjected to Tidal Wetting. Applied and Environmental Microbiology 75 : 1517-1524. Yamahara, K.M., Layton, B.A., Santoro, A.E ., and Boehm, A.B. (2007) Beach sands along the California coast are diffuse sources of fecal bacteria to coastal waters. Environmental Science & Technology 41 : 4515-4521. Zhai, Q., Coyne, M.S., and Barnhisel, R.I. ( 1995) Mortality rates of fecal bacteria in subsoil amended with poultry manure. Bioresource Technology 54 : 165-169. Zmirou, D., Pena, L., Ledrans, M., and Letert re, A. (2003) Risks associated with the microbiological quality of bodi es of fresh and marine water used for recreational purposes: summary estimates based on published epidemiological studies. Archives of Environmental Health 58 : 703-711.
ABOUT THE AUTHOR Brian Douglas Badgley was born in Hilliard, Ohio. In 1995, he received his B.S. in zoology from the University of Georgia, in Athens, Georgia. Afterward, he taught environmental education for a year at th e Jekyll Island 4-H Center in Jekyll Island, Georgia, before later return ing to obtain his M.S. degr ee in Marine, Estuarine, and Environmental Science from the University of Maryland in College Park, Maryland, which he received in 2001. His thesis fo cused on the uptake of di ssolved nitrate as a source of nitrogen for symbiotic corals, whic h he researched at the Bermuda Biological Station in St. GeorgeÂ’s, Berm uda. During his final year at the University of Maryland, Brian was awarded a one-year Knauss Ma rine Policy Fellowship from NOAAÂ’s Sea Grant Program, which he spent with the Na tional Estuarine Research Reserve System serving as a program officer for the mid-A tlantic Reserves and working on other policy initiatives such as expanding the Reserve System. Following this, Brian worked as the Coastal Training Program Coordinator at the Rookery Bay National Estuarine Reserve in Naples, Florida, where he developed pr ofessional training programs for local environmental professionals that focus on coas tal resources. Finall y, in 2004 Brian began the Ph.D. program in Biology at the University of South Florida, under the advisement of Drs. Florence Thomas and Valerie Harw ood, where he then completed his doctoral degree.