This item is only available as the following downloads:
1 DETECTION, ISOLATION, AND ENUMERATION OF SALMONELLA FROM CENTRAL FLORIDA SURFACE WATERS By RACHEL MCEGAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREME NTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Rachel McEgan
3 I would like to dedicate this work to my Mom, Dad, and Grama, and to the memory of my Oma, and Opa
4 ACKNOWL EDGMENTS I would like to gratefully and sincerely thank my major advisor, Dr. Michelle D. Danyluk, for her guidance, patience, and understanding during my time at the University of Florida. Her mentorship was, and will remain, paramount in my success. I would also like to thank my supervisory committee members, Dr. Keith Schneider, Dr. Anita Wright, and Dr. Arnold Schumann for their continuous guidance and help. Additionally, I would like to thank Gwen Lundy and Lorretta Friedrich for the constant willin gness to help; without them the completion of this project would not have been possible. Luis Marteniz and Lisseth Porao are also appreciated for their assistance. I would like to thank the USDA CSREES Specialty Crops Initiative 2008 51180 04846 for the financial support of this project. Most of all, I would like to thank my parents John and Rosemarie, for their endless support and encouragement throughout my entire schooling; for never suggesting I should give up on my dreams. Also I thank all my family and friends for to Tintin for being an endless source of amusement and for her unconditional love and
5 T ABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 12 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 2 DETECTION, IDENTIFICATION, PERSISTENC E, AND INDICATORS OF SALMONELLA IN SURFACE WATERS OF AGRICULTURAL PURPOSES: A LITERATURE REVIEW ................................ ................................ .......................... 21 Salmonella ................................ ................................ ................................ .............. 21 Outbreaks ................................ ................................ ................................ ............... 23 Contamination through Agricultural Waters ................................ ............................ 24 Antimicrobial Resistance ................................ ................................ ......................... 25 Environmental Presents ................................ ................................ .......................... 25 Chemical and Physical Water Quality Parameters ................................ ........... 27 Fresh Water ................................ ................................ ................................ ...... 28 Estuary Water ................................ ................................ ................................ ... 31 Marine Water ................................ ................................ ................................ .... 31 Biotic Sources ................................ ................................ ................................ .. 33 Fate Outside Host ................................ ................................ ................................ ... 35 Microcosm and Survival Time ................................ ................................ ................. 36 Possible Correlating Bacteria (Coliforms and E. coli ) ................................ ............. 38 Possible Correlating Factors (Chemical and Physical Water Parameters, and Environmental Conditions) ................................ ................................ ................... 46 Sample Collection ................................ ................................ ................................ ... 48 Cultural Methods ................................ ................................ ................................ ..... 49 American Public Health Association ................................ ................................ 49 Food and Drug Administration Bacteriological Analytical Manual .................. 52 International Organization for Standardization 6579 ................................ ........ 53 Variations on Standard Methods ................................ ................................ ...... 54 Water Filtration ................................ ................................ ................................ ....... 56 Normal Filtration ................................ ................................ ............................... 57 Ta ngential Flow Filtration ................................ ................................ ................. 58 Moore Swabs ................................ ................................ ................................ ... 60 Immunology Based Methods ................................ ................................ .................. 61
6 Dynabeads ................................ ................................ ................................ ....... 63 Pathatrix ................................ ................................ ................................ ........... 63 Molecular Methods ................................ ................................ ................................ 64 PCR ................................ ................................ ................................ .................. 65 Real Time PCR ................................ ................................ ................................ 67 Multiplex PCR ................................ ................................ ................................ ... 68 Risk Analysis ................................ ................................ ................................ .......... 70 3 DETECTION OF SALMONELLA SPP. FROM LARGE VOLUMES OF WATER BY MODIFIED MOORE SWABS AND TANGENTIAL FLOW FILTRATION ................. 83 Materials and Methods ................................ ................................ ............................ 85 Bacterial Strains and Preparation of Inoculum ................................ ................. 85 Collection of Water Samples ................................ ................................ ............ 86 Normal Filtration ................................ ................................ ............................... 86 Tangential Flow Filtration ................................ ................................ ................. 86 Modified Moore Swabs ................................ ................................ ..................... 87 Conventional Salmonella Detection from NFF and TFF ................................ ... 87 Immuno magnetic Separation of Salmonella during TFF Optimization. ........... 88 Detection of Salmonella During TFF Optimization ................................ ............ 88 Detection of Pathogens from MMS ................................ ................................ ... 89 Validation of MMS Conc entration ................................ ................................ ..... 89 Statistical Analysis ................................ ................................ ............................ 89 Results ................................ ................................ ................................ .................... 90 Discussion ................................ ................................ ................................ .............. 91 4 PREDICTING SALMONELLA POPULATIONS FROM BIOLOGICAL, CHEMICAL, AND PHYSICAL INDICATORS IN FLORIDA SURFACE WATER ......................... 98 Materials and M ethods ................................ ................................ .......................... 100 Water Sampling ................................ ................................ .............................. 100 Chemical and Physical Water Characteristics ................................ ................ 101 Total Aerobic Plate Count ................................ ................................ ............... 101 Coliform/ E. coli MPN ................................ ................................ ...................... 101 Salmonella Detection ................................ ................................ ..................... 102 Modified MPN for Salmonella Positive Water Samples ................................ .. 103 Weather Data Collection ................................ ................................ ................. 104 Statistical Analy sis ................................ ................................ .......................... 104 Results ................................ ................................ ................................ .................. 104 Salmonella MPN ................................ ................................ ............................. 105 Coliform and E. coli MPN ................................ ................................ ............... 105 Aerobic Plate Count ................................ ................................ ....................... 106 Air and Water Temperature ................................ ................................ ............ 106 pH, OR P, Turbidity, and Conductivity ................................ ............................. 106 Rainfall and Solar Radiation ................................ ................................ ........... 107 Discussion ................................ ................................ ................................ ............ 107
7 5 IDENTIFICATION AND CHARACTERIZATION OF SALMONELLA FROM CENTRAL FLORIDA SURFACE WATERS ................................ .......................... 121 Materials and Methods ................................ ................................ .......................... 122 Water Sample Collection and Salmonella Enrichment ................................ ... 122 Multiplex PCR Analysis ................................ ................................ .................. 123 Serogrouping ................................ ................................ ................................ .. 124 Serotyping ................................ ................................ ................................ ...... 124 Pulsed field gel electrophoresis ................................ ................................ ...... 124 Antimicrobial Susceptibility Test ing ................................ ................................ 125 Results ................................ ................................ ................................ .................. 125 Water Sample Collection ................................ ................................ ................ 125 Multiplex PCR ................................ ................................ ................................ 126 Serogrouping ................................ ................................ ................................ .. 126 Serotyping ................................ ................................ ................................ ...... 127 PFGE Analysis ................................ ................................ ............................... 128 Antimicrobial Susceptibility Testing ................................ ................................ 129 Discussion ................................ ................................ ................................ ............ 130 6 FATE OF SALMONELLA SPP. IN FLORIDA SURFACE WATERS UNDER LABORATORY CONDITIONS ................................ ................................ .............. 146 Materials and Methods ................................ ................................ .......................... 148 Waters and Microcosm Conditions ................................ ................................ 148 Bacterial Strains and Culture Preparation ................................ ...................... 149 Microbiological Analysis ................................ ................................ ................. 150 Distribution of strain survival in cocktails ................................ ........................ 151 Statistical Analysis ................................ ................................ .......................... 152 Results ................................ ................................ ................................ .................. 152 Cocktail in Each Water and Temperature Combination Over Time ................ 152 DI water ................................ ................................ ................................ 152 Sterile surface wat er ................................ ................................ ............ 153 Non sterile surface water ................................ ................................ ..... 154 EPA worst case water ................................ ................................ .......... 15 4 Fa te of Individual Serovars Compared to Cocktail Over Time ........................ 155 Serovar Survival Within Cocktail Inoculated Waters ................................ ....... 156 Effect of Wa ter Type and Temperature on the Fate of Salmonella ................. 157 DI water ................................ ................................ ................................ 157 Sterile surface water ................................ ................................ ............ 158 Non sterile surface water ................................ ................................ ..... 159 EPA worst case water ................................ ................................ .......... 159 Effect of Eukaryotes in Non Sterile Surface Water ................................ ......... 160 Discussion ................................ ................................ ................................ ............ 163 7 CONCLUSIONS AND FUTURE WORK ................................ ............................... 176 LIST OF REFERENCES ................................ ................................ ............................. 181
8 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 190
9 LIST OF TABLES Table page 2 1 Salmonella i solated from surface waters ................................ ............................ 72 3 1 Percent Salmonella positive samples of 10 L inoculated surface water, concentrated using TFF and pre enriched in lactose broth evaluated using different dete ction techniques (n = 6). ................................ ................................ 96 3 2 Percent Salmonella positive samples of 10 L inoculated non sterile saline, pre concentrated by TFF or MMS and detected by cultural means (n = 3). ........ 97 4 1 Description of 18 Central Florida surface water sampling sites including water source type, land use, and watershed as described by local water management district ................................ ................................ ......................... 114 4 2 Correlation coefficients (R 2 ) determined for Salmonella log MPN/100 ml enumerated monthly for 12 consecutive months from 18 Central Florida surface water sites vs. each of the physical, chemical, and biological water c haracteristic recorded at time of samplings, values greater than 0.4 have been bolded ................................ ................................ ................................ ...... 115 4 3 Correlation coefficients (R 2 ) with p values determined between each of the physical, chemical, and biological water characteristics for all sampling sites and months combined ................................ ................................ ...................... 116 5 1 Suggested serotypes and corresponding suggested serogroup based on multiplex PCR patterns for Salmonella is olated from Central Florida surface waters compared with actual serogroup determined by autoagglutination ....... 136 5 2 Salmonella serovars isolated from Central Florida surface waters over 12 conse cutive months compared with Salmonella top 20 most clinically common serovars ................................ ........................... 137 5 3 Salmonella serovars isolated over 12 consecutive months from Central Florida surface waters with each different multiplex PCR banding pattern for all serovars ................................ ................................ ................................ ....... 138 5 4 Antibiotic resistance profiles of Salmonella isolates from Central Florida surface waters, only isola tes resistant to greater than five of the antibiotics screened are listed ................................ ................................ ........................... 142
10 LIST OF FIGURES Figure page 4 1 Populations of Salmonella enumerated via MPN enrichments in LB, TT broth, E. coli ) enumerated via MPN method using Colisure, all reported in log MPN/100 ml (left axis), and aerobic plate counts ( ), enumerated on TSA and reported in log CFU/1 00 ml (right axis), as determined for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period. ................................ .. 117 4 2 easured using a temperature probe at the time of sampling for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period. .............. 118 4 3 right axis) measured at the time of sampling for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period. ................................ ................................ ........................ 119 4 4 sampling for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period. ................................ ................................ ...... 120 5 1 Dendrogram of representing PFGE Xba I fingerprinting of Salmonella isolates recovered from sampling site 1 6, PFGE, pulsed field gel electrophoresis ...... 143 5 2 Dendrogram representing PFGE Xba I fingerprinting of Salmonella isolates r ecovered from sampling sites 7 12. PFGE, pulsed field gel electrophoresis .. 144 5 3 Dendrogram representing PFGE Xba I fingerprinting of Salmonella isolates recovered from sampling sites 13 18. PFGE. Pulsed field gel electrophoresis. ................................ ................................ ................................ 145 6 1 Survival of Salmonella cocktail in surface waters at ( ( ) 15C ( m ean standard deviation, n=6). (A) DI water. (B) Ste rile surface water. (C) Non sterile surface water. (D) EPA worst case water. .................... 170 6 2 Survival of Salmonella Anatum in surface waters at ( ( ) 15C ( m ean standard devia tion, n=3). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water. .................... 171 6 3 Survival of Salmonella Gaminara in surface waters at ( and ( ) 15C ( m ean standard deviation, n=3). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water. ....... 172 6 4 Survival of Salmonella Newport in surface waters at ( ( ) 15C ( m ean standard deviation, n=3). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water. .................... 173
11 6 5 Composition of Salmonella cocktail in surface waters at final enumeration by spread plating as determined by serogrouping 100 colonies, Salmonella serovar Anatum (solid), Newport (crosshatched), and Gaminara (striped) ( m ean standard deviation, n=6). (A ) 28C. (B) 21C. (C) 15C. .................... 174 6 6 Survival of Salmonella cocktail in sterile surface water ( ), non sterile surface water ( ), and non ( m ean standard deviation, n=6). (A) 28C. (B) 21C. (C) 15C. ................... 175
12 LIST OF ABBREVIATIONS APHA American Public Health Association BGA Brilliant green agar BPW Buff ered peptone water BSA Bismuth sulfite agar CDC Centers for Disease Control and Prevention CFU Colony forming unit DI Deionized E. Escherichia EPA Environmental Protection Agency FDA BAM Food and Drug Administration Bacteriological Analytical Manual FSMA Food Safety Modernization Act GAPs Good Agricultural Practices H Flagellar antigen HE Hektoen enteric agar IMS Immuno magnetic separation ISO International Organization for Standardization LB Lactose broth LGMA Leafy Greens Marketing Agreement LIA Lysine iron agar MS Moore swab MMS Modified Moore swab mPCR Multiplex polymerase chain reaction MPN Most probable number NFF Normal flow filtration
13 NVSL National Veterinary Services Laboratory O Somatic antigen ORP Oxidation reduction potential PFGE Pulsed field gel electrophoresis PFU Plaque forming unit qPCR Quantitative polymerase chain reaction RV Rappaport Vassiliadis S. Salmonella SC Selenite cysteine SF Selenite F spp. species ( plural ) TFF Tangential flow filtration T GAPs Tomato Good Agricultural Practic es TSA Tryptic soy agar TSAA Tryptic soy agar supplemented with ampicilin TSAN Tryptic soy agar supplemented with nalidixic acid TSB Tryptic soy broth TSBA Tryptic soy broth supplemented with ampicilin TSBN Tryptic soy broth supplemented with nalidixic aci d TSI Triple sugar iron agar TT Tetrathionate UPB Universal Pre enrichment broth USA United States of America VBNC Viable but not culturable Vi Capsular antigen
14 XLD Xylose lysine desoxycholate agar XLT4 Xylose lysine Tegertol4
15 Abstract of Dissertation Pre sented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DETECTION, ISOLATION, AND ENUMERATION OF SALMONELLA FROM CENTRAL FLORIDA SURFACE WATERS By Rachel M cEgan May 2013 Chair: Michelle D. Danyluk Major: Food Science and Human Nutrition Salmonellosis outbreaks have been linked to consumption of contaminated fresh produce; water has been suggested as a contamination source. Screening agricultural waters f or Salmonella or determining another indicator, can aid in avoidance of produce contamination. A method was developed to screen 10L of surface water for Salmonella using a combination of tangential flow filtration, overnight enrichment, immun o magnetic s eparation, and qPCR detection. Modified Moore s wabs performed equally as well as tangential flow filtration for concentration. Monthly samples from 18 Central Florida surface water sites were evaluated for Salmonella ; 202 samples were collected, all were Salmonella positive. Aerobic plate counts, coliform/ E. coli MPNs and Salmonella MPN were also determined. Air and water temperature, pH, ORP, turbidity, and conductivity were measured. No attribute exhibited a correlation with Salmonella All 562 isol ates were grouped based on a previously described multiplex PCR (mPCR). Salmonella were identified as having the same mPCR pattern in 68 samples. Four hundred and ninety four isolates did not match the thirty most clinically common serovar patterns previ ously reported. All groupings based on mPCR pattern matched groupings based on serogroup; 19 isolates serogrouped according to mPCR suggested serovar. Isolates
16 were also characterized using serotyping, p ulsed f ield g el e lectrophoresis (PFGE) and antimic robial resistance patterns. Thirty two different serovars were identified. Three serovars were added at 8 log CFU/100ml to sterile deionized water (DI), sterile surface water, non sterile surface water, or EPA worst case water at 15, 21, or 28C. At six months DI, sterile surface water, and EPA worst case water had Salmonella ranging between 6.0 and 7.5 log CFU/100ml. Salmonella in non sterile surface waters ranged between 0.9 and 3.8 log MPN/100ml. Salmonella is prevalent in Central Florida surface wa ters. The role of indicators in determining Salmonella prevalence in surface water is limited. Direct screening of surface water for Salmonella may be necessary where Salmonella levels are high and where waters are applied close to harvest. The mPCR met hod can be used to group environmental Salmonella isolates as different or potentially the same, but has limited ability to predict serogroup. Salmonella survives extended periods of time in waters, indicating a continued risk for fresh produce contaminat ion.
17 CHAPTER 1 INTRODUCTION Outbreaks of salmonellosis have been linked to the consumption of fresh produce. Nontyphoidal Salmonella is the second leading cause of foodborne disease in the USA resulting in an estimated 1.0 million cases of illness per an num (Scallan et al. 2011). There have been a wide variety of produce items that have been linked to salmonellosis outbreaks recently including: basil, cantaloupe, lettuce, mangoes, Serrano peppers, spinach, and tomatoes (CDC 2008, Erickson et al. 2010, Kl ontz et al. 2010, Behravesh et al. 2011). The sporadic and heterogeneous nature of contamination, indicator organisms, and microbial counts in fresh produce both within and between fields makes direct screening of produce problematic (Kase et al. 2012). I t is therefore imperative to reduce the incidence of possible contamination. Irrigation waters, and other agricultural waters, have been suggested to be the major source of in field contamination of fresh produce (Jacobsen and Bech 2012). The presence of Salmonella in pre harvest water that comes into contact with, or is in close proximity to, produce production may represent a risk to produce safety. The screening of pre harvest waters for Salmonella or other biological, chemical, or physical indicator, can aid in the avoidance of produce contamination. Screening of larger volumes of surface waters will decrease the lower limit of detection and allow for a more representative sample to be analyzed. Current methods of screening for Salmonella either allo w for the use of 25 100 ml samples or do not provide quantitative data, as in the case of Moore swabs which are suspended in a body of water for a given period of time (APHA 1992, US FDA 1998). Larger sample
18 volumes would enable to acquisition of quanti tative data, which could in turn be used to generate more accurate risk assessments for microbial water quality of waters to be used in fresh produce production. Tests for indicator microorganisms are less cumbersome and time consuming, as well as typicall y less expensive, as compared to screening for the presence of Salmonella Thus, the use of indicator microorganisms is a much more economical and practical choice. However, contradictory results have been reported as to the efficacy of indicator microor ganism in predicting the presence and/or prevalence of human pathogens, such as Salmonella in surface waters (Burton et al. 1987, Rhodes and Kator 1988, Chandran and Hatha 2005, Chandran et al. 2011). Physicochemical water characteristics also have the p otential to correlate with the presence and/or prevalence of Salmonella in surface waters (APHA 1992). Physicochemical measurements have the added advantage of being able to be measured rapidly. Currently, there are no nationally required microbial water quality standards for fresh produce production. The US produce industry relies on testing 100 ml of water for indicator organisms, specifically, generic E. coli as a means to monitor water quality used for edible horticultural crop production (LGMA 2012, F DACS 2007). Testing of 100 ml for E. coli and having a geometric mean less than 126 MPN/100 ml, is a requirement only for Florida tomatoes; leafy greens producers may opt to follow this standard as a condition of belonging to the Leafy Greens Marketin g Agreement. Once isolated, the identification of Salmonella serovars is important for source tracking and outbreak investigations. The current serotyping procedure requires numerous O and H antisera for the characterization of each serovar (Jean Gilles
19 B eaubrun et al. 2012). Pulsed field gel electrophoresis (PFGE) is used to produce a DNA fingerprint for the purpose of comparing Salmonella isolates; this procedure requires five days (Ribot et al. 2006). A more time effective method for comparing and ser otyping Salmonella isolates would be beneficial. A multiplex PCR method could potentially be used to identify the most clinically common Salmonella serovars (Kim et al. 2006). Salmonella serovars isolated from surface waters are frequently not those sero vars included in the 30 most clinically relevant serovars (Kim et al. 2006, Haley et al. 2009, Patchanee et al. 2010, Rajabi et al. 2011, Strawn et al. 2013), hence alternative methods for serovar identification of clinical isolates need to be validated fo r those coming from environmental sources. The duration of survival of Salmonella in surface water constitutes a public health concern through direct consumption, indirect consumption (via irrigation of edible crops), and contact during recreation (Greene et al. 2008, Ijabadeniyi et al. 2011, Ge et al. 2012). Minimal work has focused on the duration of Salmonella survival in various types of environmental waters. Many previous works have also focused on marine or estuary waters, results from which cannot be entirely relied upon to predict survival in fresh water. Most studies have focused on clinically important serovars, such as S Typhimurium, rather than other less clinically common serovars (Rhodes and Kator 1988, Maki and Hicks 2002, Chandran and Hat ha 2005). The use solely of a local water source, sterile or non sterile, incorporates uncontrolled variables into each set of experiments that in turn make the studies difficult to compare. This work had four main objectives: (1) compare i) concentration and detection of Salmonella in large volumes (10 L) of surface water by tangential flow filtration and
20 normal flow filtration and ii) concentration of Salmonella by tangential flow filtration and Modified Moore swabs; (2) explore the relationships between the presence and/or concentration of Salmonella and biological, chemical or physical indicators in Central Florida surface water samples taken monthly for twelve consecutive months; (3) determine i) if a multiplex PCR method is able to verify that all Sal monella isolates from a single water sample are identical; ii) if a multiplex PCR method is able to predict the serogroup of environmental isolates; iii) the serovar, PFGE pattern, and antibiotic resistance profile for the Salmonella isolates determined to be different via the multiplex PCR ; (4) explore the relationship between the persistence of Salmonella spp., either individually or as a cocktail, and biological, chemical, and physical water quality characteristics, representative of surface waters, ove r six months; the suitability of EPA worst case water as a standard medium for investigating Salmonella persistence was also evaluated.
21 CHAPTER 2 DETECTION, IDENTIFICATION, PERSISTENCE, AND INDICATORS OF SALMONELLA IN SURFACE WATERS OF AGRICULTURAL PURPO SES: A LITERATURE REVIEW The presence of Salmonella in pre harvest water that comes into contact with, or is in close proximity to, produce production may represent a risk to produce safety. Direct screening of produce items is problematic. It has been r epeatedly shown for many produce items that contamination, indicator organisms, and microbial counts are all sporadic and heterogeneously distributed both within and between fields (Kase et al. 2012). Testing one produce sample may not provide an accurate representation of the overall microbial quality of the field. It is therefore imperative to reduce the incidence of possible contamination. It has previously been suggested that irrigation waters, and other agricultural waters, are the major sources of contamination of fresh produce in the field (Jacobsen and Bech 2012). The notion of pre harvest Salmonella contamination of fresh produce elicits several unanswered questions regarding prevalence, and persistence of Salmonella in surface waters, which cou ld be used for agricultural purposes. The objective of this review is to discuss the prevalence and persistence of Salmonella in surface waters, as well as methods used to determine either the presence/absence or enumeration of Salmonella from surface wat ers. Salmonella Salmonella spp. are members of the family Enterobacteriac eae ; these are mesophillic, facultatively anaerobic Gram negat ive rod sha ped bacteria. The genus Salmonella comprises two species: S. bongori and S. enterica (D'Aoust and Maurer 2007 ). The subspecies S. enterica is further differentiated into six subspecies (enterica, salamae, arizonae, diarizonae, indica, and houtenae) among which the S. enterica
22 subspecies enterica is mainly associated with human and other warm blooded vertebrates. The subspecies are differentiated on the basis of biochemical traits and genomic relatedness. Traditionally members of the genus Salmonella are clustered in serovars according to their flagellar (H) and somatic (O) anitgens, some serovars also have caps ular (Vi) antigens (Kim et al. 2006). The majority of Salmonella are motile via peritrichous flagella; however, nonflagellated variants to occur. Both S Pullorum and S. Gallinarum are nonmotle. The serovar is determined via agglutination with specific antibodies for the O, H, and Vi antigens (Kim et al. 2006, Jean Gilles Beaubrun et al. 2012). The serogroup is determined by the O antigen. Each subspecies contains numerous serovars. Currently over 2,500 Salmonella serovars have been described, but onl y about 50 serovars, a ll within the subspecies enteri ca, are common causes of infections in humans and warm blooded animals. Salmonella is currently believed to be a zoonotic foodborne pathogen, which is spread by the fecal oral route (D'Aoust and Maurer 2 007). This microorganism can enter aquatic environments directly with the feces of an infected human or animal, or indirectly via sewage discharge or agricultural land run off. Salmonella is a ubiquitous enteric pathogen with a worldwide distribution tha t comprises a large number of serovars characterized by different host specificity and distribution. Two serovars, S. Typhi and S. Paratyphi, are host adapted to infect only humans; these cause typhoid fever. Nontyphoidal Salmonella is the second leading cause of foodborne disease in the USA resulting in an estimated 1.0 million cases of illness per annum (Scallan et al. 2011), following only norovirus at 5.5 million cases of illness (Scallan et al. 2011). With an estimated 27% of salmonellosis cases req uiring hospitalization, Salmonella is the
23 estimated leading cause of hospitalization due to foodborne disease (Scallan et al. 2011). Nontyphoidal salmonellosis typically results in enterocolitis between 8 to 72 h after consumption. The disease is general ly self limiting with symptoms lasting approximately five days. Symptoms include nonbloody diarrhea and abdominal pain. Outbreaks There have been a wide variety of produce items that have caused salmonellosis outbreaks recently including: basil, cantaloup e, lettuce, mango, tomatoes, Serrano peppers, and spinach (CDC 2008, Erickson et al. 2010, Klontz et al. 2010, Behravesh et al. 2011). Fresh produce related outbreaks are typically shorter in duration, when compared to nonperishable food items, and occur with the seasonality of harvest of the particular produce item. Outbreaks can be large or small scale depending on the distribution of the produce item. Linking environmental sources of Salmonella with foodborne outbreaks is very challenging; only two suc h cases exist. In a 2005 outbreak of S Newport, Salmonella strains isolated from a pond on the Eastern shore of Virginia used for irrigation of tomato fields matched an outbreak strain recovered during the outbreak investigation (Greene et al. 2008). Th e same strain caused a 2002 outbreak as well (Greene et al. 2008), which indicates a possible long term environmental survival of Salmonella as well as suggesting the irrigation pond was a source of contamination for the tomatoes. A second outbreak where t he same S Saintpaul was isolated from an on farm source was the Serrano pepper outbreak in 2008, when the matching isolate was able to be isolated from an irrigation pond on a farm where the epidemiologically linked Serrano peppers could have been produce d (Klontz et al. 2010).
24 Although only two outbreaks have been conclusively linked back to Salmonella contamination from irrigation water many more instances potentially exist. Changes in microbial populations between incidence of contamination and outbrea k investigation, heterogeneous distribution of Salmonella within the water source, or a lack adequate sampling could account for why so few outbreaks have conclusively been linked to contamination through irrigation water. Often trace back to the farm has occurred in outbreaks involving produce, while the definitive identification of the mode of contamination is left unknown. Contamination t hrough Agricultural Waters It is widely accepted that contaminated agricultural waters are able to cause produce cont amination. The percentage of pathogen that is able to remain on, or internalize with the various produce items is not yet well established; however it is thought to occur at low levels. Evidence of pathogens internalizing in the edible portions of plants is inconsistent. When irrigating greenhouse tomatoes for 14 days with S Montevideo (7 log CFU/ml in 350 ml) in the irrigation water that was applied directly to the soil, no Salmonella was found in the red ripe tomatoes after harvest (Miles et al. 2009). Another study irrigating with S Newport at 7 log CFU/ml once every seven days, found internalization did occur, but at a low incidence (Hintz et al. 2010). During an in field study performed in South Africa irrigation water from the irrigation canal was positive for Salmonella in 33% (4/12) of 25 ml samples and 11% of broccoli and cauliflower, the crops grown in the irrigated field, were positive for Salmonella (Ijabadeniyi et al. 2011). Ge et al. (2012) noted that lettuce was able to internalize S. Typ himurium from soil irrigation; higher rates of internalization resulted
25 when drought conditions preceded irrigation. Another pathogen, E. coli O157:H7, can internalize into spinach leaves after being spray irrigated, but only at high concentrations (6 log CFU/ml) in the irrigation water (Erickson et al. 2010). Antimicrobial Resistance Concern regarding the emergence of Salmonella or other bacterial pathogens, resistant to antimicrobials is increasing. Such resistance can complicate treatment of both huma n and animal cases of the disease. The environmental occurrence of antimicrobial resistant serovars of Salmonella has been the topic of several investigations. In Morocco, a monthly survey of coastal seawater isolated nine S. Senftenberg isolates; six of these isolates were resistant to ampicillin (Setti et al. 2009). A Tunisian survey of clams harvested from coastal waters, isolated S Typhimurium, all of which were resistant to ampicillin, erythromycin, rifampicin, and tetracycline (El Mejri et al. 2012 ). Sixty three Salmonella isolates were obtained from various environmental samples from tomato farms in the Mid Atlantic Region of the U.S., all of these isolates were susceptible to amikacin, ceftriaxone, ciprofloxacin, gentamicin, nalidixic acid, ceftio fur, trimethoprim/sulfamethoxazole and streptomycin (Micallef et al. 2012). Three isolates were multidrug resistant; these three isolates came from all different farms and included two S Newport and one S Tennessee (Micallef et al. 2012). Environmental Presents The data provided in Table 2 1 demonstrates the ubiquitous nature of Salmonella in environmental waters. Salmonellae have been detected in different geographic regions throughout the world, in very diverse water sources, ranging from pristine to
26 those heavily impacted by anthroprogenic activities. Detection frequencies were extremely variable in surface waters. A surface water sampling survey has previously been performed in North Florida in the Suwannee River watershed (Rajabi et al. 2011). Her e the researchers isolated Salmonella from 96% (106/110) of water samples when three by five MPN method was used (1, 10, 50, 100, 500 ml) (Rajabi et al. 2011). Salmonella populations ranged from <18 MPN/100 ml to 5400 MPN/100 ml, the highest concentration obtained in February when samplings were performed January through April (Rajabi et al. 2011). Another geographically nearby surface water sampling survey for Salmonella to has been performed in the state of Georgia. Here, Salmonella was found in 79.2% ( 57/72) of surface water samples when a five by three MPN method was used (1, 10, 100 ml) (Haley et al. 2009). Salmonella populations ranged from <1.8 to 68.6 MPN/L, with the highest levels being obtained in August (Haley et al. 2009). In Northeast Georgi a pond water has been sampled and had Salmonella concentrations ranging from below detectable limits (<0.015 MPN/L) to 3.4 MPN/L (Jenkins et al. 2008). Sampling was conducted in August, October, December, March, and May; again, the largest concentration o f Salmonella was obtained in August (Jenkins et al. 2008). Many parameters could potentially influence the presence of Salmonella in surface waters. Such parameters as water quality, type of water (fresh, estuary, or marine), the soil surrounding the wate r source, as well as the biotic sources that come near or contact the surface water could influence the fate of Salmonella
27 Chemical and Physical Water Quality Parameters Various water quality parameters may, potentially correlate with the microbial qualit y of the water source. A diverse array of physical and chemical water quality parameters have been considered as means to predict microbial water quality. While numerous survey and microcosm studies exist, the wide array of parameters considered and reco rded often makes multiple studies difficult to compare. The most commonly reported parameter is water temperature. Water temperature has been shown to correlate significantly with microbial water quality, and specifically with Salmonella presence and/or persistence. In marine bathing waters, Salmonella had an inverse relationship to water temperature when the recorded temperatures were grouped into the following categories: <20.5C, 20.6 to 21.9C, 22.0C to 22.5C, and >22.6C (Serrano et al. 1998). Wa ter temperature also had a significant correlation with aerobic plate count, aerobic spore formers, anaerobic spore formers, and intestinal Enteroccocus (Ijabadeniyi et al. 2011). In microcosm experiments, temperature has repeatedly been an important fact or for the survival of Salmonella in environmental samples (Jacobsen and Bech 2012); in general, higher temperatures result in an increased rate of Salmonella die off (Jacobsen and Bech 2012). Other parameters that have been reported, and correlated with h igher Salmonella prevalence include: overcast skies, low or outgoing tides, rough seas, and intense turbidity (no values given) (Serrano et al. 1998). Exposure to sunlight (UV) may also affect Salmonella survival: incubating microcosms in the dark has bee n shown to slow the rate of inactivation, changing the two log reduction time from 122 h to 144 h for S Typhimurium (Bae and Wuertz 2012). Indicator organisms correlate with chemical oxygen demand (COD) (fecal and total coliforms), pH, turbidity, and rai nfall (aerobic
28 plate count, aerobic spore formers, anaerobic spore formers, and intestinal Enteroccocus ) (Ijabadeniyi et al. 2011). The use of turbidity as a means of predicting Salmonella presence or prevalence may be either flawed or not applicable to al l situations. At 20C, in filter sterilized surface water streptomycin resistant S Typhimurium remained detectable by pour plating to bismuth sulfite agar for up to 41 days; suspended particles (clay, silt, sand, organic flocculent) did not significantly affect the survival of S Typhimurium (Maki and Hicks 2002). In this experiment it was noted that the Salmonella decreased in a non linear fashion, regardless of particle size or concentration (Maki and Hicks 2002). This would indicate that turbidity mi ght not be a good indicator of pathogen presence in water. Fresh Water Fresh water has <0.5 ppt dissolved salts. Fresh surface water sources include: irrigation canals, streams, rivers, ponds, and lakes. In attempts to elucidate the prevalence as well as the parameters that affect survival of Salmonella in fresh water sources numerous surveys and microcosms have been reported along with some select parameters. It can be suggested that Salmonella has a relatively high prevalence in surface waters throughou t the world, especially in warmer climates. A South African 12 month survey of an irrigation canal and the two rivers that feed into it, determined 42% (15/36) of their 25 ml samples were positive for Salmonella when screening the samples using the ISO 65 79 method (Ijabadeniyi et al. 2011). During an environmental survey in North Carolina, 25 ml surface water samples taken from crop agricultural areas were 50% (6/12) positive for Salmonella via cultural
29 methods based on the ISO 6579 method (Patchanee et al 2010). Similarly, this survey also found surface water from a residential/industrial area to have a 58.8% (20/34) positive for Salmonella and 57.1% (16/28) of surface water samples were positive for Salmonella from a forested area (Patchanee et al. 201 0). This study also reported on the diversity of serovars found in each area; the highest diversity of serovars came from the residential/industrial area (eight serovars), followed by crop agriculture (five serovars); the least came from forestry (four se rovars) (Patchanee et al. 2010). Another US based survey was performed in the Mid Atlantic Region. Fourteen tomato farms (nine large scale, and five small scale) were sampled (irrigation pond water and sediment, groundwater, irrigation ditch water, soil, leaves, and tomatoes) periodically. Salmonella was isolated using a culture based method, which was based on the FDA BAM method; 250 ml of water was filtered through 0.45 m, 47 mm mixed cellulose ester filter, the filter pre enriched in 30 ml lactose bro th, 1 ml aliquot to 15 ml Tetrathionate Hajna broth with iodine, streaked to XLT4 and R & F Salmonella Chromogenic Plating Medium; isolates were confirmed by PCR amplification of Salmonella specific enterotoxin ( stn ) gene. Salmonella was isolated on 29 % (4/14) of the tomato farms; four of the large scale farms, and none of the small scale farms (Micallef et al. 2012). Ponds on the small scale farms were typically small, artificially created and disconnected from o ther water ways or wooded areas. P onds o n the large scale farms were typically large, deep and supported more wildlife (frogs and snakes), and received run off from wooded areas and were connected to other bodies of water (Micallef et al. 2012). Any or none of these factors could explain or par tially explain the difference in Salmonella prevalence at the two distinct farm sizes. No Salmonella was
30 isolated from the groundwater, leaves, or tomatoes (Micallef et al. 2012). Salmonella was isolated from 7.7% (3/39) of irrigation pond water samples, while irrigation pond sediment only yielded 1.8% (2/111) positive samples (Micallef et al. 2012). Irrigation ditch water yielded Salmonella in 15.4% (2/13) of samples (Micallef et al. 2012). Isolates from water or sediment samples included: S Newport, S Braenderup, S Javiana, S Lille, and S Tennessee (Micallef et al. 2012). Multiple serovars were identified from the same water sample; however, detection of various or multiple serovars in one water sample was not consistent (Micallef et al. 2012). The authors note the abundant presence of goose and deer feces in several of the fields sampled, suggesting possible biotic sources of the Salmonella Four of the six serovars isolated from the tomato farms have previously been associated with outbreaks f rom the consumption of raw tomatoes in the US and Canada (Micallef et al. 2012). It cannot be assumed that all surface waters in warm climates will have detectable Salmonella when using cultural methods. Wa ter samples (1 to 1 00 ml) taken from Spring Lake, the headwaters of the San Marcos River, TX from just below the surface and 50 cm of depth; all samples taken were negative for Salmonella when pre concentrated by centrifugation followed by ISO 6579 cultural method (Gaertner et al. 2008). The number of s amples screened during this survey was not reported. It also should not be assumed that no Salmonella were present in these waters; one cannot forget the possibility of viable but non culturable state (VBNC) as well as the possibility that the Salmonella were present, but in numbers less than 1 CFU/100 ml, as would have been the detection limit.
31 Estuary Water An estuary is a transition zone between rivers and ocean waters; depending on water flow the salinity varies within and among estuaries. Findings f rom Salmonella survival studies in estuarian waters are often difficult to compare due to the many and widely varying water parameters that make up these water bodies. A microcosm based study on the survival of S Typhimurium in estuarian water observed a 3 log CFU/100 ml decrease in two days when the Salmonella was inoculated into raw estuarian water and held at either 20 or 30C (Chandran and Hatha 2005). The raw water in their experiment contained 5 6 log CFU/ml heterotrophic bacteria, 2 log PFU/ml bact eriophage predating Salmonella 3 log PFU/ml coliphage, and had protozoans detected by qualitative microscope observation (Chandran and Hatha 2005). Alternately, S Tennessee held in diffusion chambers in the estuarian waters of Chesapeake Bay displayed g rowth during the first three days in raw water (Rhodes and Kator 1988). The microbial quality of the raw water used in the Chesapeake Bay study was not reported, but can be assumed to be a significant factor in the observed multiplication of the Salmonell a The difference in serovars used in the respective studies many also explain some of the difference in the observations. Marine Water Marine water, while of little agricultural use, will be discussed here because of the numerous studies reported and pos sibility of the findings within being applicable to waters of agricultural significance. The salinity of marine waters varies from 3.1 to 3.8% (31 to 38 ppt). Much of the Salmonella in marine water studies that have been carried out center around bathing waters, as this is a primary form of contact (and thus health risk) related to Salmonella in marine waters.
32 Bathing waters in Spain were sampled and screened for Salmonella over a 10 week period using the ISO 6579 method (Serrano et al. 1998). Three beac h areas were sampled and ranged between 0.8 to 8.5% (of 356 samples) positive for Salmonella (Serrano et al. 1998). In Northwestern Spain, 2.5% (18/707) water samples were positive for Salmonella using the ISO 6579 method (Martinez Urtaza et al. 2004). T his study was performed over four years and was able to discern a seasonal pattern; an increased incidence of Salmonella existed from October to December (Martinez Urtaza et al. 2004). It is also noted that weather data (wind, rainfall, temperature, and h ours of sunshine) from the previous day correlated with Salmonella isolation frequency (Martinez Urtaza et al. 2004). When conditions were correlated with specific serovars, stronger correlations were obtained for S Seftenburg and S Agona (the two most frequently isolated serovars) (Martinez Urtaza et al. 2004). A survey performed in Southern Morocco found Salmonella in 4.1% (10/243) of their monthly seawater samples when 100 ml was screened using the ISO 6579 method (Setti et al. 2009). The region scre ened was noted to be populous. The same survey noted a higher prevalence in sediment from the same sampling locations; 6.8% (19/279) of 25 g sediment samples were positive for Salmonella (Setti et al. 2009). These authors found a positive correlation (lin ear relationship) to rainfall and the presence of Salmonella (Setti et al. 2009); a seasonal variation was also noted, with higher prevalence in the autumn and winter, however, no average temperature is given. This survey was unusual in that only two serov ars were obtained ( S Blockley and S Senftenberg) throughout the 24 month survey (Setti et al. 2009).
33 Four strains of S Typhimurium were isolated from clams harvested from the Tunisian coast in December 2005 and January 2006 (El Mejri et al. 2012). Thes e four strains, along with an isogenic wild type ( S Typhimurium C52) were individually inoculated into sterile sea water microcosms and held in the dark at ambient temperatures (authors do not state actual temp). The culturability of these isolates on TS A decreased progressively throughout the 30 days that they were held (El Mejri et al. 2012). The time for one log reduction (28, 27, 27, and 30 h) determined for each of the strains originally isolated from clams taken from the coast of Tunisia were all g reater than the time for one log reduction obtained for the S Typhimurium C52 strain used as a control (25 h) (El Mejri et al. 2012). Biotic Sources Numerous biotic sources of Salmonella exist, including various reptiles, cows, horses, cats, goats, emus, dogs, elk, coyotes, ground squirrels, raccoons, chipmunks, and various birds (Moore et al. 2003). Shedding of Salmonella in feces is thought to contribute to the pathogen load of watersheds by runoff from adjacent lands. Reptiles are common inhabitants bo th in and around surface waters in Florida. Reptiles (snakes, lizards, chameleons, and turtles) have a 50.5% Salmonella prevalence, when cultural methods are used to test fecal samples (Corrente et al. 2004). Of the reptiles screened by Corrente et al. ( 2004), snakes had the highest prevalence (83.3%), lizards a 50% prevalence, and turtles a 10.3% prevalence. Chameleons had 100% prevalence, however, this is reflective of only two samples taken (Corrente et al. 2004). Serovars isolated include: Aqua, Gam inara, Gbadago, Give, Havana, Minnesota, Muenster, Newport, Oldenburg, Poona and Senftenberg (Corrente et al. 2004).
34 Wild turtles (including common musk turtles, red eared sliders, Texas cooters, snapping turtles, and Guadalupe spiny soft shell, n=52) have been Salmonella positive via carapace or cloacal swabs using the ISO 6579 method (Gaertner et al. 2008). The most common serovar associated with these turtle was S Rubislaw; S Gaminara, S Newport, and S Thompson were also isolated (Gaertner et al. 20 08). Water samples from the water source where these turtles were captured tested negative for Salmonella when up to 100 ml was screened (Gaertner et al. 2008), indicating the presence of Salmonella on/with turtles in the environment does not necessarily predict the presence of Salmonella in the water habitat. American alligators have been a source of both fecal coliforms and Salmonella when cloacal swabs were taken from wild alligators (Johnston et al. 2010); captive American alligators have fecal colifo rms at a level of 9 log CFU/g (wet wt) of feces (Johnston et al. 2010). Haley et al. (2009) suggested that the dominance of S. enterica subspecies arizonae (40% of isolates) in the Little River rural watershed (in Georgia USA) indicate the local reptile po pulations are the likely source of the Salmonella isolated from the Little River watershed. Other wild, non reptilian animals can also be carriers of Salmonella Raccoons have been noted to be asymptomatic carriers of Salmonella at rates of up to 65% of individuals in urban areas or 27% of individuals in rural areas (Jardine et al. 2011). Serovars isolated from raccoon feces include both those that have and those that have not caused human outbreaks (Jardine et al. 2011); S Typhimurium the most prevent
35 serovar in rural, and S. Newport the most common serovar in urban raccoons (Jardine et al. 2011). Birds are another potential biotic source of Salmonella Seagulls along the western Mediterranean coast were 17% (31/182) positive for Salmonella when fecal samples were screened using ISO 6579 2002 method (Ramos et al. 2010). In another study performed in the Oldman river watershed (Alberta, Canada) only the serovar Rubislaw could be detected in the collected water samples; this serovar was also detected in wild bird feces, but never in the analyzed human sewage or domestic animal feces of the area (Jokinen et al. 2011). Insects are capable of being carriers of Salmonella from one location to another. Flies are known to transmit Salmonella typically via the ir sponging mouth parts, body and leg hairs, the sticky parts of their feet, fecal deposition, or vomitus (Wang et al. 2011). Flies captured near pig farms carry Salmonella at a rate of 26.4% (58/220); predominate serovars included Anatum, Choleraesuis va r. kunzendorf, and Derby (Wang et al. 2011). These same serovars, with matching PFGE patterns, were also isolated from pig feces on the nearby farms (Wang et al. 2011), thus suggesting flies as a vector. Midges can act as a carrier for S Typhimurium DT1 04 in the laboratory, becoming infected with Salmonella and transfer it to the next, otherwise sterile, aqueous environment (Moore et al. 2003). Fate Outside Host The duration of survival of Salmonella in the aquatic environment may influence the potential risk to human health; as duration of survival increases so does the time for human contact with the contaminated water source. Studying the duration of survival of Salmonella in the aquatic environment is problematic because there are numerous
36 variables that could be affecting their survival. These variables are too numerous to completely account for in microcosm based studies. In situ studies do not allow for control of variables. Environmental samplings do not account for re contamination, rather tha n solely duration of survival, when attempting to discern the fate of Salmonella in aquatic environments. During one study on the Red River of the North (between North Dakota and Minnesota), Salmonella survived three days (62 river miles) from the point of suspected contamination (sewage treatment plant) during the month of November; the survival time increased to four days (73 river miles) during January, when water temperatures are noted as extremely cold and ice cover was present (Spino 1966). From this study it is not known, however, if the Salmonella was capable of surviving for a longer duration or if they were being diluted out to lower than the limit of detection, yet still surviving. To more accurately determine duration of survival in various typ es of surface water, microcosm based studies are required to control certain parameters, as well as avoid the Salmonella from merely being diluted out beyond the limit of detection due to environmental events (river flows, rainfall, etc.) and thus not bein g detected. Microcosm and Survival Time world. This allows for the effect of one parameter to be studied whilst the others are held constant. Duration of Salmonella survival in surface waters can be evaluated using this means as it can be assured that no dilution effect or additional contamination events are occurring. The longest duration of survival reported for Salmonella in a water microcosm is seventeen years (Amel and Am ina 2008). This microcosm contained sterile seawater
37 into which the Salmonella had been inoculated, this was held at ca. 20C and allowed to dry to salt crystals. To culture the Salmonella from this microcosm an extended, three month, resuscitation in nu trient broth was required (Amel and Amina 2008). This lengthy duration of survival has not been repeated, however many microcosm experiments simply stop after a given period of time (typically months) and do not undertake such extensive resuscitations in nonselective enrichment media. In the same article, a steady decline in Salmonella population in sterile seawater, held at 20C, from 8 log CFU/ml to 1.5 log CFU/ml over a 15 month time frame (Amel and Amina 2008). Salmonella Typhimurium multiresistant DT 104 has survived at culturable levels up to 54 days in 23C sterile well water (Moore et al. 2003), although detection was intermittent between days 21 and 54 (when sampled every other day) (Moore et al. 2003). At 20C, in filter sterilized surface water streptomycin resistant S Typhimurium remained detectable by pour plating to bismuth sulfite agar for up to 41 days; suspended particles (clay, silt, sand, organic flocculent) did not significantly affect the survival of S Typhimurium (Maki and Hicks 200 2). In this study Salmonella decreased in a non linear fashion, regardless of particle size or concentration (Maki and Hicks 2002), indicating turbidity might not be a good indicator of pathogen presence in water. Microcosms established using sterile wat er do not accurately represent the real world situation. Salmonella in real world surface waters, are subject to predation by protozoa, competition for resources from other bacteria, and attack by bacterio phages. While microcosms established using filte r sterilized water would still be expected to contain bacteriophages, most of the competing bacteria, and all of the predacious
38 protozoa are removed. Sterilizing via autoclaving would inactivate protozoa, bacteria, and bacteriophages, additionally the hea t exposure would also be expected to alter some of the nutrients in the original water sample. At 20C, in filter sterilized surface water streptomycin resistant S Typhimurium remained detectable by pour plating to bismuth sulfite agar for up to 41 days; non filter sterilized surface water, with organic flocculent, showed a significantly higher decrease in Salmonella overtime, remaining culturable for 16 days; this is speculated to be due to both predation and competition with the native microflora (Maki a nd Hicks 2002). A microcosm study that investigated the survival S Typhimurium in estuary waters focusing on the effect of biological factors, dissolved organic and inorganic substances, and sunlight on the inactivation of these bacteria. Sunlight was th e most important factor for inactivation, followed by biotic factors. An overall 2.5 log CFU/ml reduction in Salmonella is seen over four days at 20C in the non sterile water; a greater decrease (3.5 log CFU/ml) was observed when the temperature was 30C (Chandran and Hatha 2005). The biotic factors were determined to comprise of a combination of predation by protozoa and bacteria, bacteriophage lysis, and competition with autochthonous microbes. Dissolved organic substances present in estuarine water, in the absence of the native biota, promoted survival and growth of S Typhimurium (Chandran and Hatha 2005). Possible Correlating Bacteria (Coliforms and E. coli ) The direct detection of pathogenic bacteria, such as Salmonella is costly and time consumin g, requiring well trained labor. It is these factors that lead to the concept of indicator organisms for detecting fecal contamination, and thus potentially human pathogens. In 1914 the coliform group was adopted by the US Public Health Service as
39 an ind icator of fecal contamination (APHA 1992). There are both total coliforms and fecal coliforms. Total coliforms are a group of bacteria belonging to the Enterobacteriaceae ; these include both aerobes and facultative anaerobes, they are Gram negative, non s pore forming, rod shaped bacteria that ferment lactose with gas production within 48 h at 35C. Included in this group are: Escherichia coli Enterobacter spp., Klebsiella spp., and Citrobacter spp. Fecal coliforms are thermotolerant coliforms that ferme nt lactose at 44.5C; these include E. coli and Klebsiella spp. Generic E. coli are also used as indicator bacteria. Escherichia coli can be differentiated from other fecal coliforms by the absence of urease and the presence of glucuronidase. It is the se differences in metabolism that products such as Colilert and Colisure (IDEXX, Westbrook, Maine) are based on. In Colilert, galactosidase from the total coliforms is used to metabolize ortho nitrophenyl galactoside (ONPG) and change it from colorles s to yellow; while E. coli uses glucuronidase to metabolize 4 methylumbelliferyl D glucuronide to D glucuronide and 4 methylumbelliferone which is fluorescent under UV light. Colisure uses the same enzyme reaction for E. coli but uses chlorophenol red D galactosidase that then has the color change going from yellow to magenta. Ideally, an indicator organism fits the following criteria: associated with the intestinal tract of warm bloode d animals; present when pathogens are present, and absent from uncontaminated samples; present in greater numbers than the pathogen; be at least equally resistant as the pathogen to environmental factors and disinfectants; does not multiply in the environm ent; be detectable by easy, rapid, and inexpensive
40 methods; and be non pathogenic. Many commercial products, such as Colisure and Colilert discussed above, have been developed to make coliforms and E. coli easy, rapid, and inexpensive to detect. It is th for coliforms and E. coli as indicator organisms for use in Florida surface waters. The correlation between coliforms o r E. coli and Salmonella still needs to be better understood. Understanding the correlation between Salmonella and indicator organisms is an important for setting scientifically valid Good Agricultural Practices (GAPs). Currently, Tomato Good Agricultural Practices (T GAPs) are the only regulatory requirement for microbial water quality of us e on edible horticultural crops (FL5G 6). T he Leafy Greens Marketing Agreement (LGMA) also contains water quality standards for waters used on edible leafy greens, ho wever, the LGMA standards are voluntary (F DACS 2007, LGMA 2012). The T GAPs regulation contains two microbial water standards, one for foliar applied waters applied near the time of harvest, and one for non foliar applied waters. Under T GAPs, foliar app lied waters are required to meet potable water quality standards; these standards are contained in 40 CFR 141.63. That is, no more than 5.0% of 40 samples may be positive for total coliforms; when fewer than 40 samples are collected in a month, no more th an one sample per month may be total coliform positive. Fecal coliform or E. coli positive samples are not permitted. The regulation for non foliar waters uses the US EPA standards for recreational waters, 40 CFR Part 131.41(c); no single sample may conta in greater than 235 generic E. coli per 100 ml and the geometric mean of all samples taken must be below 126
41 generic E. coli per 100 ml. These requirements were originally designed for bathing waters, and have not been validated for indirect consumption, as would be the case when the water is used to irrigate edible horticultural crops. The LGMA, a voluntary grower group, also relies on these standards for irrigation waters. With these standards in mind, the EPA assessed 33% of US waters in the year 2000 and found that 40% of streams, 45% of lakes, and 50% of estuaries were not clean enough to bathing water quality standards (Jacobsen and Bech 2012). This may be problematic when looking for acceptable water sources for use on edible horticultural crops. While other indictor organisms do exist and are used, including Enteroco c ci fecal Streptococci various bacteriophages, etc., this review will focus on E. coli and coliforms due to their significance in agricultural regulations and guidance documents. Th e American Public Health Association (APHA) suggests monitoring heterotrophic plate counts, as well as indicator organisms, such as total coliforms, fecal coliforms, and E. coli to determine microbial water quality (APHA 1992). The remainder of this secti on will focus on E. coli and coliforms, and their ability to predict the presence or prevalence of Salmonella Several studies have reported coliforms and/or E. coli being acceptable indicators of Salmonella presence or prevalence. Microcosm based studies have shown Salmonella dies off at a greater rate than E. coli Klebsiella pneumonia and Pseudomonas aeruginosa when a variety of sediment types were included in the microcosms (Burton et al. 1987). This would indicate good potential as indicator organis ms because the slower die off rate would result in the indicator organisms being present at greater numbers than the Salmonella ; this is a requirement for an ideal
42 indicator organism. Another microcosm based study looked at the survival of S Typhimurium in estuarian water. The authors used a starting inoculum of 6 log CFU/ml when inoculating Salmonella into the microcosms. A r apid, approximate 3 log CFU/ml inactivation of the Salmonella in raw estuarian water was observed over a two day period when hel d at either 20 or 30C (Chandran and Hatha 2005). It is noted that Salmonella declined at a faster rate than the E. coli in the same experimental conditions (Chandran and Hatha 2005). When the biotic factors were removed by filter sterilization, enhanced survival of Salmonella and E. coli was observed under the otherwise identical conditions; however, E. coli remained at a higher concentration than Salmonella (Chandran and Hatha 2005), however this is not significantly different. The raw water in their e xperiment contained 5 6 log CFU/ml heterotrophic bacteria, 2 log PFU/ml bacteriophage predating Salmonella 3 log PFU/ml coliphage, and had protozoans detected by qualitative microscope observation (Chandran and Hatha 2005). Although these examples both s uggest that coliforms and/or E. coli is capable of acting as an indicator organism for Salmonella the findings must be carefully considered because the microcosm conditions will not be perfect representations of real world occurrences, and thus may miss s ome parameters that could change the survival rate of either bacteria. It is therefore important to also consider results obtained from real world samplings. Samplings performed in South Africa have consistently high correlations between both fecal colifo rms and total coliforms and the presence of Salmonella (Ijabadeniyi et al. 2011). In cases of known human fecal contamination both fecal and total coliforms can be used as indicator organisms for Salmonella In an example going back to 1966,
43 sewage outfa ll was monitored 22 miles downstream from initial outfall in the Red River ND fecal coliforms were enumerated at 2,850 CFU/100 ml, total coliforms were enumerated at 47,600 CFU/100 ml at the furthest location where Salmonella was able to be isolated (Spin o 1966). This would suggest that either set of indicator organisms could be used for Salmonella however, this study is limited in that few samples were taken, and only takes into account an obvious source of human fecal contamination. Observations may n ot hold true when there is no obvious source of human fecal contamination, or when the bacteria are previously from other, less hospitable, sources. Numerous examples also exist where coliforms and/or E. coli was not a good indicator of Salmonella presence or prevalence. Microcosm studies have also been able to show a lack of correlation and a greater die off of Salmonella compared to E. coli when temperatures of 10C were investigated (Rhodes and Kator 1988). Sampling of ornamental fountains, which coul d be interpreted as microcosms in the real world, in Guadalajara City, Mexico showed a correlation between the presence of coliforms and generic E. coli however, a correlation was not seen in the presence / absence of Salmonella in the same water samples (Escartin et al. 2002). The lack of correlation could be explained by the difference in sampling methods for the two sets of organisms; coliforms and E. coli were determined by MPN (10, 1, 0.1 ml), whereas Salmonella was determined through enrichment of a 24 h Moore swab (Escartin et al. 2002), intuitively this samples a larger volume of water. B oth of these examples not exact representations of the real world, and can thus only be considered with careful prior thought.
44 There are real world sampling examp les that also demonstrate the potential lack of correlation between coliforms and/or E. coli with Salmonella A study in Georgia found E. coli in all surface water samples, with concentrations varying significantly by site; however, E. coli was not found to be able to correlate with levels of Salmonella determined for the same surface water samples (Haley et al. 2009). In a pond water study in Northeastern Georgia when Salmonella populations were at their highest level, both generic E. coli and fecal ente rococci were not detected from the same water sample (Jenkins et al. 2008). At other sampling times in the same study E. coli was detectable, but did not correlate with Salmonella populations. Furthermore, mineral water from Portugal allowed for S Typhi murium to persist up to 20 days longer than E. coli (Ramalho et al. 2001). This aqueous environment provided a very low nutrient level with a background microflora level of approx. 6 log CFU/ml (Ramalho et al. 2001). This last example is especially conce rning because of the long duration during which Salmonella was able to survive and E. coli the indicator organism, died off much sooner; an ideal indicator organism must be able to persist in the environment for a greater duration than the predicted patho gen. Some attempts to identify parameters that can predict whether a specific indicator can be useful have been carried out. One study looked at E. coli in sediment and water microcosms and the effect of temperature (Craig et al. 2004). Sediment type was found to affect the survival of E. coli with mainly sandy soil allowing for less survival than others with higher silt, clay, and organic carbon content (Craig et al. 2004), difference in E. coli concentration was noted at day zero sampling of sediment. Water column E. coli concentrations remained relatively constant for the initial sampling period
45 (days zero and one) (Craig et al. 2004). E. coli persisted better at 10C, than at 20 and 30C (Craig et al. 2004). In another study Salmonella originally isolated from sewage, survived over 56 days in static sterile creek water microcosms (Fish and Pettibone 1995). The Salmonella population increased from the initial 6.7 log CFU/ml to 8.1 log CFU/ml within the first 10 days; the population decreased to ba ck to 6.7 log CFU/ml after 56 days (the end of the experiment). There was no significant difference between the plate counts (spread plated on nutrient agar) and direct counts or direct viable counts (Fish and Pettibone 1995), demonstrating a lack of stre ssed cells. Escherichia coli followed a similar pattern in this experiment, with a starting population of 6.3 log CFU/ml a maximum population was reached at 10 days (7.6 log CFU/ml), then remained relatively constant through the remaining experiment (56 d ays) (Fish and Pettibone 1995). This study does demonstrate that the E. coli and Salmonella follow similar survival patterns in the surface water microcosm; however, it should be noted that in each case the bacteria were in otherwise sterile conditions (i .e. no competition or predation), a situation that is not accurately reflective of real world situations. In this case the E. coli does not demonstrate to be an ideal indicator organism because it was not consistently present at greater numbers than the Salmonella Considering the plethora of conflicting data on the ability of coliforms or E. coli to act as indicator organisms for Salmonella it may be suggested that the ability to predict Salmonella presence needs to be established for individual sets of parameters (geographic location, sources of contamination, water and sediment characteristics) under which one wishes to implement the use of indicator organisms, rather than selecting and accepting one or one set of indicator organisms universally.
46 Possi ble Correlating Factors (Chemical and Physical Water Parameters, and Environmental Conditions) In addition to microbial indicators of water quality, several chemical and physical water parameters have been suggested as a possible correlating data point to indicate the presence or prevalence of Salmonella or other pathogens. The major advantage to using a chemical or physical water quality parameter is the general ability to obtain immediate results for these parameters, rather than the typically long time delay in obtaining microbial results. The APHA states in its Standard Methods for the Examination of Water and Wastewater that physicochemical factors, such as turbidity, pH, salinity, temperature, assimilable organic carbon, dissolved oxygen, biochemical oxygen demand, and ammonia may provide useful information about contamination or the potential of water to support bacterial growth (APHA 1992). During a twelve month sampling of Georgia surface waters no significant correlations could be found between S almonella populations and any measured water parameters, these included conductivity, pH, DO, turbidity, ORP, fluorescence, and chlorophyll a (Haley et al. 2009). The only factor that the APHA recommends monitoring that did have a significant correlation with Salmonella populations was water temperature (Haley et al. 2009). It should be noted, however, that this is the only measured parameter that varied significantly over the study. It is perhaps due to the lack of variability of these parameters in the sampling location that resulted in the lack of correlations. Through the use of microcosm based experiments, numerous studies have determined sunlight to decrease the survival time in a variety of surface waters (Ramalho et al. 2001, Chandran and Hatha 20 05, Bae and Wuertz 2012,). This should
47 be expected due to the effect th at UV from sunlight has on DNA; bonds between adjacent thymine base pairs are broken and some bonds form between neighboring base pairs. Repair mechanisms do exist, however, many cell s are irreversibly inactivated from this damage. Besides sunlight, rainfall has also been recognized as a possible correlating environmental condition for Salmonella populations. In Georgia a 12 month survey found the daily rainfall volumes for three days prior to sampling had a strong positive correlation with Salmonella populations (Haley et al. 2009). Rainfall from beyond three days was not found to correlate with Salmonella populations. It is possible that rainfall can also result in a decrease in ba cterial populations in surface waters; this decreasing effect is hypothesized to result from a dilution effect that could be expected when large volume rainfall events occur. This has been suggested as a reason why fecal coliforms have been found to have no correlation with rainfall when it is known they are commonly found in tropical sediments and re suspension would be expected during rainfall events (Santiago Rodriguez et al. 2012). Rainwater itself is not typically thought of as a source of contaminat ion, although a study performed in Singapore, found rainwater to be (21/50) positive for E. coli when 100 ml samples were screened with qPCR (Kaushik et al. 2012). Many parameters and environmental conditions can affect the ability of Salmonella to persist in surface waters. It is likely that these have synergistic effects and that one parameter cannot be considered on an individual basis as a predictor of Salmonella presence of persistence. It is therefore imperative that more complex models be built usi ng greater amounts of data that is still to be gathered.
48 Sample Collection Collecting surface water samples must utilize either nonreactive borosilicate glass or plastic bottles that have been cleansed, rinsed, and sterilized (APHA 1992). The sampling con tainer must be large enough to allow for a minimum of 2.5 cm headspace to remain above the sample; this allows for mixing the water sample before examination (APHA 1992). To establish the time of day water should be sampled the APHA suggests peak bathing h ours when the waters are expected to be used for recreational purposes (APHA 1992). No other suggestion of time of day is given. This concept is reinforced by a study performed in Spain where sampling at 10:00, 13:00, and 16:00 h was performed in bathing waters and the highest level of Salmonella was detected at 16:00 h (6.7%), whereas at 10:00 and 13:00 h Salmonella positives were 3.3% and 4.4%, respectively (Serrano et al. 1998). An opposite pattern was observed for all other indicator microorganisms e numerated during the same study. While this sampling strategy may allow for Salmonella to be detected more readily, the decrease in indicator microorganism may result in a lower level of risk being assumed if only indicator microorganisms were to be enume rated. When manual sampling is being performed the APHA suggests the bottle should be held by the base and plunged, neck downward, below the surface (APHA 1992). The bottleneck should then be turned up slightly and into the current, and thus allowed to fi ll. If sampling by hand is not possible a weight may be attached to the bottle, and this lowered into the water source (APHA 1992).
49 Cultural Methods Current protocols for enrichment culture of Salmonella from foods or water rely on a primary nonselective enrichment broth to recover stressed cells followed by secondary enrichment in selective broths. A solid medium is then required to obtain isolated colonies, which can then be biochemically, serologically, or genetically confirmed as Salmonella These me thods are laborious and require several days to complete; this is due to Salmonella being a biochemically diverse genus, thus multiple media types containing different selective and differentiating agents are required. Even with the use of selective broth s and selective and differential agars, Citrobacter and Proteus are both capable of causing false positive results in Salmonella enrichments (Gorski and Liang 2010). Multiple organizations have developed their own standard protocols which all follow the s ame general format; in the US standard Salmonella protocols have been developed by the APHA, the US Food and Drug Administration (US FDA), and the International Organization for Standardization (ISO). American Public Health Association The APHA suggested m ethods for Salmonella detection from water consist of a concentration step, enrichment, selective growth, and biochemical confirmations (APHA 1992). The concentration techniques suggested for use include swabs, diatomaceous earth, large volume samplers, a nd membrane filter technique (APHA 1992). The swab technique uses sterilized cheesecloth, which is folded and wrapped, or gauze pads being suspended in the water to be sampled for a desired (unspecified) length of time. The swab acts to concentrate parti culate matter and microorganisms from the water as it flows through and over the swab (APHA 1992). The diatomaceous earth technique requires diatomaceous earth to be added to a funnel above an absorbent pad (which
50 hold the diatomaceous earth in place); th e water sample is then vacuum filtered through the plug (APHA 1992). The APHA states that this method should be capable of filtering up to 2 L of sample water. The large volume sampler technique uses a borosilicate glass microfiber bonded with epoxy resi n as the filter through which the APHA states larger sample volumes can be passed through (a specific volume is not reported) (APHA 1992). The membrane filter technique, recommended for low turbidity waters, either 142 mm diameter or using multiple 47 mm diameter filters (APHA 1992). For higher turbidity waters, it is suggested that this method may still be used if the filter is pre coated with diatomaceous earth by filtering a suspension of the fore mention ed (APHA 1992). In either case the filter, or filters, are homogenized via blending into the enrichment medium. After the selected filter concentration method has been performed the concentrated sample needs to be selectively enriched in a growth medium t hat will suppress the growth of coliform bacteria (APHA 1992). The APHA recommends selecting two of the following enrichment media and running them in parallel: selenite cysteine (SC) broth, selenite F (SF) broth, or tetrathionate (TT) broth (APHA 1992). Selenite cysteine broth will inhibit Gram positive and nonpathogenic enterobacteria while allowing for recovery of most species of Salmonella including S. Typhi (APHA 1992). It is suggested that after incubation turbid tubes be streaked to solid media d aily, for five days, in order to optimally obtain all serovars that may be present (APHA 1992). Selenite F broth is stated to allow for optimum recovery of most Salmonella including S Typhi; however, this increased recovery is accompanied by a decrease in selectivity ( E. coli growth is not inhibited) (APHA 1992). Again, repeated streaking daily for five days
51 after incubation is recommended to better obtain all serovars present. Tetrathionate broth inhibits coliforms and Gram positive bacteria, while al lowing for the growth of most Salmonella including S Typhi (APHA 1992). The APHA states that TT broth is more selective for Salmonella than the selenite based broths, especially when incubated at 43C for 48 h (APHA 1992). Unfortunately, this broth allo ws for the optimum growth of Proteus mirabilis ; without the addition of brilliant green Proteus and Citrobacter can also grow (APHA 1992). The addition of brilliant green and incubation at 43C will inhibit the growth of some species of Salmonella includi ng S Typhi (APHA 1992). Following enrichment, a secondary differentiation on selective solid media is required. The following solid media are suggested for use: brilliant green agar (BGA), bismuth sulfite agar (BSA), xylose lysine desoxycholate agar (XLD ), and xylose lysine brilliant green agar (APHA 1992). Brilliant green agar will cause S. Typhi and a few other Salmonella to grow poorly because of the brilliant green dye content this can be aided by incubating plates for 48 h (APHA 1992). Slow lactose fermenters, such as Proteus Citrobacter, and Pseudomonas will produce colonies that resemble Salmonella (APHA 1992). On BGA, non lactose fermenting colonies can be masked by lactose fermenting ones in close proximity (APHA 1992). Bismuth sulfite agar allows for much growth of most Salmonella (APHA 1992). These plates also require 24 and 48 h incubations to better obtain the slow growing serovars. After 48 h of incubation, typical Salmonella colonies may be masked by numerous bordering colonies (APHA 1992). Proteus and some coliforms can also develop the typical black color of the Salmonella colonies due to H 2 S production (APHA 1992). Xylose lysine desoxycholate agar can be slightly toxic to fastidious Salmonella (APHA 1992). Coliforms, Proteus and some
52 Enterobacter can produce yellow colonies on this agar; depending on the starting sampling this many cause difficulties in isolating Salmonella colonies (APHA 1992). Xylose lysine brilliant green agar is recommended for marine samples as it inhibits the Proteus Enterobacter and Citrobacter (APHA 1992). After isolation on selective media, biochemical followed by serological confirmations are suggested (APHA 1992). Any combination of these methods can be used to perform a five by three MPN in order to obtain quantitative results (APHA 1992). Food and Drug Administration Bacteriological Analytical Manual The Food and Drug Administration Bacteriological Analytical Manual (FDA BAM) method contains specific instructions for the preparation of a varie ty of food types for the isolation of Salmonella (US FDA 1998). This method follows the common protocol of pre enrichment, selective enrichment, colony isolation, and colony confirmation as Salmonella From sampling to final result, this method requires five days. A detection limit of 1 2 log CFU/25 g has been established for this protocol (Delibato et al. 2011). In this method the pre enrichment media is lactose broth, universal pre enrichment broth, buffered peptone water, trypticase soy broth, or Dey Engley broth. In each case only one of these pre enrichment broths is to be used; the recommended broth is based on the composition of the food item as well as the expected background microflora. The typical pre enrichment conditions require 242 h at 35 2C. Selective enrichment for low microbial load foods uses both Rappaport Vassiliadis (RV) medium and TT broth are to be used. From the pre enriched broth 0.1 ml is transferred to 10 ml RV and 1 ml is transferred to TT broth (US FDA 1998); these are inc ubated 242 h at 420.2C and 242 h at 352C, respectively. Foods with
53 higher microbial loads require the TT broth to be incubated at 242 h at 430.2C. Only when the product is Guar gum, or S Typhi is expected is the use of SC broth recommended to r eplace the RV. Colony isolation is performed on BSA, XLD, and Hektoen enteric (HE) agar (US FDA 1998). Plates are incubated 242 h at 352C. Typical colonies for each media are identified and picked; it is recommended two or more colonies from each sele ctive agar plate be used. If no typical colonies are available, atypical Salmonella colonies should be picked. Picked colonies are inoculated to lysine iron agar (LIA) and triple sugar iron agar (TSI); tubes are incubated 242 h at 352C. Salmonella ty pically produce alkaline (red) slant and acid (yellow) butt, with or without the production of H 2 S (blackening of agar) in TSI. In LIA, Salmonella typically produce alkaline (purple) reaction in the butt of the tube; most will produce H 2 S in LIA. Once pr esumed to be Salmonella from the reactions of the slants, isolates are confirmed biochemically and serologically. International Organization for Standardization 6579 ISO met hod has a detection method equal to the 1 2 log CFU/25 g of the FDA BAM method (Delibato et al. 2011) and also requires five days. The ISO method for Salmonella enrichment is number 6579. ISO 6579:2002 method uses a 25 g sample in 225 ml Buffered peptone water (BPW) for 16 20 h at 371C; 0.1 ml to 10 ml Rappaport Vassiliadis broth with soy peptone (RVS) 24 h at 41.51C and 1 ml in 10 ml Muller Kauffman tetrathionate broth with Novobiocin (MKTTn) 24 h 371C; plating to XLD and a C hromagar of the laborato ries choice 24 h at 371C; one presumptive colony per plate is then picked and
54 streaked to solid nutrient agar, incubated 24 h at 371C, this is then used for serological and biochemical confirmation (Schonenbruecher et al. 2008). Variations on Standard M ethods Salmonella are a metabolically diverse genius of bacteria, and it is thus difficult to ensure all species and subspecies will be isolated by a particular method. This fact is reflected in the numerous differences within the standard methods for Sa lmonella isolation. To add to this difficulty, variations in sample composition and background microflora also add to the complexity of Salmonella isolation. For these reasons many variations in cultural methods for Salmonella isolation can be found in t he literature. The pre enrichment broth can be BPW, lactose broth (LB), universal pre enrichment broth (UPB), or tryptic soy broth (TSB). Buffered peptone water as a medium for pre enrichment may yield a higher number of positive results as compared to us of LB or straight enrichment with TT broth (Thomason et al. 1977). When looking at 100 soil and water samples 29 were positive for Salmonella when using BPW as a pre enrichment before TT broth, LB as a pre enrichment method gave only 24 positives, 26 wer e positive when pre enrichment was avoided and the sample inoculated straight into TT broth (Thomason et al. 1977). Additionally, S. enterica grows equally well in both UPB and BPW (Mafu and Sirois 2009). When comparing TSB, BPW, and UPB for the detectio n of S Baildon, S Cholerasuis, S Poona, S Saintpaul, and S Typhimurium from either green leaf lettuce or Roma tomato pre enrichments, TSB allowed the greatest Salmonella detection when the final real time PCR was completed (Gorski and Liang 2010). Fe wer aerobic mesophilic bacteria grew in BPW, than in TSA or UPB at the common 35C incubation temperature (Gorski and Liang 2010).
55 After the pre enrichment, a selective enrichment must be performed prior to colony isolation on solid media. Beginning enri medium does not allow some S Arizona strains to multiply, these are often associated with reptiles; especially when the Salmonella injured from environmental conditions (Madsen 1994). Use of RVS in isolating Salmonella from artificially inoculated (10 CFU/25 g) pork samples gave 100% recovery when a 24 h enrichment was used (Schonenbrucher et al. 2008). When screening environmental samples from a zoo, both animal enclosures and guest areas, significantly higher number of positives samples were obtained when the selective enrichment was performed in either RVS or MKTTn as compared to SCSC broth (Bauwens et al. 2006). For each selective enrichment, HE provided the highest number of positives, followed by XLD; BGA and semis olid Rappaport medium gave the fewest positive results (Bauwens et al. 2006). When isolating Salmonella from reptile feces, no significant difference has been noted between use of Rambach and XLD after use of either SC broth or RV broth (Corrente et al. 2 004) XLD and xylose lysine Tergitol 4 (XLT4) have been seen to have false positives from Citrobacter freundii and false negatives from S. arizonae and S. Senftenberg (Schonenbrucher et al. 2008); BGA has been shown to give false positives for C. freundii and Pseudomonas aeruginosa (Schonenbrucher et al. 2008). Salmonella Senftenberg can be a weak producer of hydrogen sulfide. Citrobacte r freundii is H2S positive (Schonenbrucher et al. 2008). While these false positives are not noted for the chromogeni c agars, S Dublin and S Derby can give false negatives (Schonenbrucher et al. 2008). Bismuth sulfite agar remains a medium of choice for isolating Salmonella
56 because, in addition to a high level of selectivity, it responds solely and most effectively to the production of extremely low levels of hydrogen sulfide, eliminating the problem of missing the biochemically atypical lactose and/or sucrose positive Salmonella Due to the numerous options for isolating Salmonella it is advised that each type of sa mple have the isolation method optimized prior to sampling projects being performed. Sample physicochemical characteristics and background microflora, as well as expected serovars need to be hypothesized prior to optimizing the isolation protocol. Water F iltration Water filtration can be used as a concentration technique prior to pre enrichment or enrichment. When concentrating bacteria by water filtration, filters can either be used to trap the bacteria within the filter matrix, or as a barrier to push t he excess water through, while retaining the bacteria on the retentate side, resulting in a smaller sample volume to move forward with. The most important advantage of filtration as a means of concentration of bacteria from water samples is the ability to handle much larger sample sizes, thus decreasing the final limit of detection. Filters possess complex and irregular structures; the pore shape, tortuosity, and connectivity will determine the dimensions of bacteria that are able to pass through the filte r. Filter pore sizes are measured based on physical measurements and mathematical extrapolations. For example the bubble point is the force necessary to extrude air through the capillary network of a wet filter. Although test bacteria are larger than 0. 2 m, some, such as Listeria monocytogenes and Hydrogenophaga pseudoflava may penetrate 0.2 m rated filters (Lee et al. 2010). Some autochthonous aquatic bacteria are able to pass through the standard 0.2 m pore size commonly used for filter sterilizat ion (Ramalho et al. 2001).
57 Extremes of process fluid composition (osmolarity, ionic strength, and viscosity) can impact bacterial passage through a filter. Osmotic shock is known to cause either shrinkage or swelling of bacterial cells, which presumably i mpacts filter passage. Fluid composition also impacts the membrane itself, for example charge shielding can occur. Some membrane materials, including aluminum oxide, cellulose acetate, and polyethersulfone, are not significantly impacted by extreme proces s fluid physicochemical characteristics (Lee et al. 2010). Both nitrocellulose acetate and polysulfone are affected more significantly (Lee et al. 2010). Use of filters to concentrate bacteria can result in reductions of the bacteria; this is may be due t o caking of the bacteria on the filter membrane (Lee et al. 2010). At lower starting concentrations in the starting fluid, lower reductions have been observed after filtering (Lee et al. 2010). Various types of filters exist for water filtration. Two fil ter categories are normal filtration, and tangential flow filtration. These will be discussed in the sections below, along with Moore swabs (MS) a specialized type of filtration used for water. Normal Filtration The term normal filtration comes from math ematics, in that the water (vector) is hitting the filter (plane) at a right (90) angle. The water can be pulled through the filter either by gravity or use of a vacuum, which is pulled under the filter membrane. A major disadvantage to normal flow filt ration is that a limited volume of water can be passed through the filter, especially in cases of high turbidity, before other components become entrapped in an on the filter causing the filter to clog. The APHA uses normal flow filtration in their standa rd water testing methods; 100 ml is passed through a sterile 0.45 m pore size, 47 mm diameter filter and the filter is applied directly to an appropriate
58 agar plate (APHA 1992). The APHA has normal flow filtration methods for coliforms or Salmonella, as well as other bacteria; in both cases 100 ml of sample water is collected and filtered through a normal flow filter as a means of concentrating the target organism. If a larger volume of water is to be analyzed, the APHA states that sterile diatomaceous e arth may be place on top of an absorbent pad over the filter, in order to act as a filter aid (APHA 1992). The normal flow filtration method has been used numerous times to survey environmental waters for Salmonella In the mid Atlantic region of the USA, up to 250 ester filter; these filters were then processed through an enrichment method (Micallef et al. 2012). This enabled Salmonella to be isolated from 7.7% (3/39) o f the pond water samples, and 15.4% (2/13) of the irrigation ditch water samples; none of the 25 ground water samples were Salmonella positive using normal flow filtration (Micallef et al. 2012). The same normal filter specifications were used to as a pre concentration step before using cultural methods to screen seawater (100 ml) from coastal waters of Agadir Morocco for Salmonella finding 4.1% (10/243) Salmonella positive samples (Setti et al. 2009). The limited volumes of water, 100 250 ml, that hav e been reported to have been screened for Salmonella using normal flow filtration may be a contributing factor to the lower percentage of Salmonella positive samples. Tangential Flow Filtration Tangential flow filtration (TFF) has the water flowing along t he filter membrane; the filter membrane is formed in a tube shape inside which the water sample flows. The water sample is pumped through the TFF unit, and is re circulated numerous times until the desired final volume has been obtained. The difference b etween the pressure on
59 the retentate side and the permeate side is referred to as the transmembrane pressure. The transmembrane pressure is what causes the water to move through the filter and to the permeate side. The tangential flow of the water sample across the filter membrane acts to push the filter cake along and off the filter membrane, thus increasing the amount of sample that can be passed through the filter. The material that the filter membrane is constructed will play a major role in the volu me of sample that can pass before the filter is irreversibly clogged. Ideally, the material should have low protein binding properties; polyethersulfone and cellulose acetate have low protein binding tendencies (Gibson and Schwab 2011). Although not used in any standard methods, TFF has been used in used in numerous situations to concentrate bacteria from large volumes of water. A cellulose triacetate TFF has been used to recover Enterococcus faecalis E. coli and Clostridium perfringens spores from 100 L surface water samples with a recovery efficiency of 70.5% (Gibson and Schwab 2011). Using TFF with a dead end (no recirculation of sample) coupled with a culture based final detection, recovers up to 93% of Enterococcus feacalis when 3 log CFU where ino culated into 100 L of tap or surface water when the turbidity was low (0.29 NTU) (Smith and Hill 2009); a higher recovery rate (78%) was seen with higher turbidity (4.3 NTU) (Smith and Hill 2009). TFF, coupled with continuous centrifugation and immuno magn etic separation followed by qPCR can detect E. coli O157:H7 at 50 MPN per 40 L of surface water with a rate of 100% detection (Mull and Hill 2009). Occasionally, studies using TFF obtain >100% recovery efficiencies, this is hypothesized to be due to disag gregation of bacterial clumps during
60 the filtration process (Smith and Hill 2009). This could also explain the relatively high recovery rate obtained when using TFF as a concentration method. Potential additional benefit of TFF is the passage of several P CR inhibiting compounds to the permeate side and thus removed from the sample. TFF using a polysulfone membrane has been combined with conventional PCR to detect human associated Bacteroidales (inoculation from sewage, actual concentrations not given), wi thout problems of PCR inhibition, due to compounds regularly found in either sewage or other waters, from both fresh and estuarine waters (Leskinen et al. 2010). The use of TFF gave the ability to sample larger volumes of water than normal filtration (due to fouling) (Leskinen et al. 2010). Moore Swabs In 1948, Moore introduced the use of cotton gauze swabs for isolating Salmonella from sewage contaminated waters. The cotton gauze was used at various locations in the sewage system in a small town in Engla nd to trace the source of paratyphoid to a single carrier. More recently, MS have been used to isolate Salmonella from surface waters and ornamental fountain waters (Spino 1966, Sears et al. 1984, Thomas et al. in press). In Santiago Chile, MS have been used to isolate Salmonella from 11% (8/76) of samples used in rivers and irrigation canals, known to have received raw sewage (Sears et al. 1984). Salmonella was isolated from 12.4 % (18/145) of MS used in ornamental fountains in Guadalajara City, Mexico (Escartin et al. 2002). Moore swabs were used in Southern Ontario to isolate Salmonella from 78.4% (91/116) river water samples collected from rivers in one watershed which was influenced by numerous recreational and agricultural activities (Thomas et al. in press). When placed in drainage lines, MS have been used to isolate E. coli O157:H7 from soil
61 and irrigation water (Ogden 2001). Moore swabs allow for a higher percentage (44/181 vs. 13/181) of positive samples, compared to 4 L of the same water bein g passed through a 0.45 m pore size membrane filter, when used to concentrate Campylobacter from river water in Chile (Fernandez et al. 2003). The APHA standard method for qualitative isolation of Salmonella from waters uses the MS which is to be suspen ded below the surface for one to three days (APHA 1992). Their method states that a longer duration of swab exposure will not increase final isolation efficacy (APHA 1992). A modification to the MS concept has also been reported. A MS is a rolled cotton gauze tied and suspended in a water sample for a minimum of 24 h, a Modified Moore swab (MMS) uses the same concept of the rolled cotton gauze, but fits it into a pipe. The water sample is pumped through the gauze in the pipe, and the bacteria are trapped within. Modified Moore swabs have been used to isolate Escherichia coli O157 and Salmonella at levels down to 1 log CFU/100 ml from 10 L of irrigation water when coupled with cultural detection methods (Bisha et al. 2011). Major advantages of the MS and MMS is that they are inexpensive, simple to use, can be used in the field rather than collecting and returning large volumes of water to the laboratory, and require minimal operator prerequisite skill base and training. The major disadvantage to MS is t he inability to give quantitative results (APHA 1992); however this can be somewhat overcome by pumping a known volume through the MMS Immunology Based Methods Rapid methods for Salmonella detection often require one to three days because a pre enrichmen t step is still required. Cultural based enrichment can be substituted by or combined with biospecific purification and concentration procedures, such as
62 immun o magnetic separation (IMS), in order to decrease analysis times (Roda et al. 2012). IMS involv es the use of polyclonal or monoclonal antibodies bound to paramagnetic particles/beads to bind target molecules or cells and remove them from the sample matrix by use of a magnet (Warren et al. 2007). The beads are paramagnetic, meaning that they only ex hibit magnetic properties when exposed to a magnetic field. Without exposure to a magnetic field the beads remain in suspension with the sample. Conventionally, these are referred to as microbeads, however, they are in the nanoparticle size range and are neither round nor bead like in shape. The main advantage of immun o based methods is that they are able to remain highly specific even in complex matrices (Roda et al. 2012), which can include environmental samples. IMS can be used to reduce time for anal ysis completion, while enhancing selectivity and sensitivity (Notzon et al. 2006). Disadvantages of immun o based methods include: lower sensitivity than molecular based detection methods, usually requiring a pre enrichment step; specificity and sensitivit y are largely dependent on the quality of the antibodies; possible difficulties in obtaining high performance antibodies for certain strains; antibody ability can be compromised if changes in expression or denaturation of target antigen occurs; sample matr ix can make the antibody unstable; and viable and non viable cells are not distinguished (Roda et al. 2012). Two commercially available IMS kits are available for Salmonella concentration: Dynabeads (Invetrogen, Grand Island, NY, USA), and Pathatrix (Matr ix MicroScience, Golden, CO, USA). Both require a sample pre enrichment, but can replace selective enrichments, thus decreasing total time for sample analysis.
63 Dynabeads Dynabeads are produced by Invetrogen (Grand Island, NY, USA) and can accommodate a 5 ml total sample volume. The initial start up costs are generally low, as only a Dynal BeadRetreaver is required (in addition to consumable supplies). Dynabeads have been coupled with qPCR to produce an assay capable of detecting Salmonella at 10 CFU/25 g from meat samples using only a 6 h pre enrichment time (total 12 h analysis time) (Notzon et al. 2006). Fat and other food debris can lead to a loss of magnetic beads during procedure (Notzon et al. 2006). Citrobacter freundii Pseudomonas spp. and Proteu s spp. may interfere with Salmonella binding to IMS beads (Colman et al. 1995). Rough forms of Salmonella may not bind to Dynabeads, theoretically due to the loss of their O specific antigen structure (Notzon et al. 2006). Additionally, Salmonella Putten, Worthington (serogroup G), Minnesota (L), and Urbana (N) will not bind with Dynabeads (Leon Velarde et al. 2009), and result in false negatives when further analysis is completed. Pathatrix The Pathatrix system is an IMS method that functions as a flow th rough immunocapture; the initial purchase of the required equipment can be rather costly. However, this set up allows for much larger sample volumes to be analyzed; previously 250 ml sample homogenates have been processed through this system (Warren et al 2007). In the Pathatrix system the sample is continuously pumped around inside sterile tubing system for a pre set 30 min time period. During this the sample continuously passes a magnet where the IMS beads are immobilized. Sample debris and other
64 pot entially inhibitory substances and cells are removed during a washing step, which occurs prior to the retrieval of the IMS beads. Studies using the Pathatrix system to isolate Salmonella from surface water, or other water types, have not previously been re ported. However, numerous studies have used the Pathatrix system to isolate Salmonella from food or environmental samples, including green tomatoes, ground beef, potato salad, poultry meats, and swabs from poultry houses (Warren et al. 2007, Leon Velarde et al. 2009, Taban and Aytac 2009). Molecular Methods The use of molecular detection methods has numerous advantages over cultural techniques. These include: decreased time to results, lower detection limits, as well as the ability to detect VBNC cells. Culture dependent methods to measure and/or detect Salmonella or fecal indicator bacteria require 18 h to five days for results to become available. The use of rapid methods, which include molecular based methods, reduces the time required for results to become available. This decrease in testing time, could reduce human health risks associated with contaminated food items or use of contaminated water for edible horticultural crop production as risk management decisions could be made in a more timely mann er. When screening water samples for Salmonella, there is a risk of underestimating the contamination level when using culture based methods because bacterial cells may enter viable but non culturable state (Jacobsen and Bech 2012). The bacteria are still present and able to cause disease, but are not capable of being grown in the laboratory. The true risk level of using the water source would thus be underestimated. The disadvantages to these types of techniques are that they often require specialized
65 t raining for laboratory personnel, and typically require costly equipment and supplies. Additionally, molecular based methods can detect both live and inactivated target organisms (Bae and Wuertz 2012). In these methods a particular molecular target is de tected as a means of determining organism presence, it is possible for the DNA of inactivated cells to remain intact enough to be detected by some molecular based methods (Bae and Wuertz 2012). False negative results can occur because of the high specific ity of these methods, if a mutation in the target organism has caused a mutation in the target molecule negative results will be obtained, when in fact the target organism is present (Malorny et al. 2003). PCR The polymerase chain reaction (PCR) is an exam ple of a molecular based method. PCR allows the specific and exponential synthesis of a targeted DNA region through the use of specifically designed primers (oligonucleotides). PCR often required pre enrichment prior to Salmonella detection (Malorny et a l, B. 2003, Gorski, and Liang 2010). Selection of the pre enrichment broth much be carefully considered. When comparing TSA, UPB and BPW for use as a pre enrichment broth prior to PCR detection of Salmonella from either lettuce or tomato samples, TSA had the least effect on PCR efficiency (Gorski and Liang 2010). Salmonella was not detected in any of the UPB samples in any of the trials; Salmonella positive samples were seen from BPW the target (Gorski and Liang 2010). UPB contains iron (III) and this is a known PCR inhibitor (Gorski and Liang 2010). The inv A gene, which encodes a protein in the Type III secretion system required for invasion of epithelial cells by Salmonella during infection, has been used previously
66 for the detection of Salmonella from w ater samples (Gaertner et al. 2008). The inv A gene has been suggested as the international standard to be used for the detection of Salmonella (Malorny et al. 2003). Using the proposed standard primers, results in a 284 base pair amplicon. inv A primers have previously been used to detect Salmonella from lettuce and tomato enrichments without producing detectable amplicons from non Salmonella bacteria (Gorski and Liang 2010). Molecular detection targets must be specific enough not to share DNA sequences with phylogenetically near neighbors. The selection of targets must take into account the fact that the microflora of enrichment culture may affect detection, and that bacterial cultures differ depending on the source of inoculum. A disadvantage associate d with PCR based detection methods is the inability to distinguish between DNA associated with live bacteria, and DNA remaining from destroyed cells. There are two ways to ameliorate this situation: use reverse transcriptase PCR (RT PCR) or quench DNA not inside a non compromised cell membrane. RT PCR uses RNA as the target, RNA is less stable than DNA and will not remain around after the cell is dead. Propidium monoazide is a DNA intercalating dye with an azide group and will penetrate only the membrane of dead cells with compromised cell walls/membranes and then bind their DNA for attach to free DNA. Upon exposure to bright visible light, the photoactive azide group on the dye is converted to a nitrene radical, which is readily attached to a carbon ato m of the DNA through a C H insertion reaction, resulting in an inhibition of amplification of DNA from dead cells or free DNA (Bae and Wuertz 2012).
67 Real Time PCR Real Time PCR (qPCR) requires less time to results than does conventional PCR, this is becaus e the amount of PCR product formed is measured as the reaction cycles progress. During the PCR reaction cycles, real time fluorescence data are collected cycle by cycle for each individual reaction (Delibato et al. 2011, Balachandran et al. 2011). Positi ve signals show an increase of the target specific fluorescence signal while the fluorescence of negative signals remains below the threshold of detection. The reaction remains in one closed tube during the entire PCR amplification and detection, thus sig nificantly reducing the potential for cross contamination and false positives (Mafu and Sirois 2009). While technical expertise is still required, the use of commercial kits can reduce the amount of personnel training required. Using a commercially availa ble kit in which all the reaction components are lyophilized into a bead ready for use in the reaction tube can reduce technical error due to excessive pipetting (Balachandran et al. 2011). The lyophilized format can also allow for a greater volume sample reaction tube allowing for increased sensitivity (Balachandran et al. 2011). Additionally, these kits contain an internal positive control that allows for monitoring for the presence of PCR inhibitors, which reduces the occurrence o f false negatives. Detection from environmental samples can often be problematic because of the presence of PCR inhibitors ( e.g humic acid, bacterial debris, complex polysaccharides, and metal ions) in qPCR assays; it is imperative to have an internal p ositive control in each reaction tube (Gibson and Schwab 2011). In previous studies to validate the MicroSEQ real time PCR kits 45 samples for each of 10 food matrixes (dry dog food, ground black pepper, raw ground beef, raw
68 chicken wings, shrimp, cantalou pe, brie cheese, dry infant formula, chocolate, or shell eggs) were spiked with Salmonella (Anatum, Newport, Typhimurium, Saintpaul, Infantis, Senftenberg, Enteritidis, Paratyphi A, or Poona) and enriched in BPW, with no false negatives being reported (Bal achandran et al. 2011). Additional work in the same study showed that this real time PCR kit was able to detect 100 different Salmonella serovars at a 3 log CFU/ml limit of detection; when screening for exclusivity 30 non Salmonella organisms were tested and only Serratia marcescens created a false positive result (Balachandran et al. 2011). Citrobacter freundii which is commonly found giving false positives in cultural based Salmonella detection methods (Schonenbruecher et al. 2008), did not generate a positive response (Balachandran et al. 2011). Real time PCR has been shown to correlate with standard culture methods when at least 3 log CFU/ml are present in the samples; when Salmonella counts are lower, qPCR is more sensitive (Mafu and Sirois 2009). I t is thus possible to decrease the limit of detection in an experiment via the use of qPCR rather than cultural methods. Multiplex PCR Multiplex PCR consists of multiple pairs of primers being used in one PCR reaction in order to target more than one gene. The genes can be from multiple organisms, such as those multiplex PCRs used to detect numerous pathogens in one reaction, or the genes can be from one organism and be used to differentiate between subspecies, such as the multiplex PCR reported by Kim et al. (2006) for the serotyping of Salmonella Kim et al. (2006) has suggested a multiplex PCR method for serotyping Salmonella ; this method consists of two five plex PCR reactions and one two plex PCR reaction. The entire method is based upon six genetic l oci from S. enterica serovar
69 Typhimurium and four loci from S. enterica serovar Typhi. This method is reported to be sensitive, reproducible, and cost effective (Jean Gilles Beaubrun et al. 2012). The Kim et al. (2006) multiplex PCR is designed to be abl e to identify the 30 most common serovars of clinical importance. The same Kim et al. (2006) multiplex PCR was confirmed against 30 serovars from an FDA reference collection, using two or three isolates per serovar; an exact band pattern match to the patte rns reported by Kim et al. (2006) was found for 29 of the serovars reported (Jean Gilles Beaubrun et al. 2012). Salmonella Oranienburg could not be matched. The same study noted serotypic variations in at least one of the isolates tested from the followi ng serovars: Montevideo, Newport, Muenchen, Hadar, Infantis, Paratyphi, Derby, and Berta. Banding pattern variations included missing expected genes or amplification of serovar specific PCR products not typical for their serotypic banding pattern (Jean Gi lles Beaubrun et al. 2012). The authors suggest that these maybe results of SNPs, insertions, or deletions at the gene loci. When using the same Kim et al. (2006) multiplex PCR reactions to serovar isolates from tomato farms, in the Mid Atlantic region of Virginia, samplings the multiplex banding patterns were not able to type 23.8% (15/63) of their isolates (Jean Gilles Beaubrun et al. 2012). The 15 isolates that could not be typed were not completely reported, however, the authors do mention they were o utside of the 30 most common serovars and give the examples of S Lille and S Tennessee (Jean Gilles Beaubrun et al. 2012). Of the 48 isolates from the tomato farm samplings, the typable serovars included multiple isolates of: S Typhimurium, S Newport, S Javiana, S Thompson, and S Braenderup (Jean Gilles Beaubrun et al. 2012). These authors also
70 report some isolates from food associated outbreaks had different banding patterns than those reported by Kim et al. (2006); these included: S Montevideo, S Seftenberg, and S Tennessee (Jean Gilles Beaubrun et al. 2012). The use of a multiplex PCR for determining serovar for Salmonella has the advantage of decreased time requirement and decreased labor costs, as fewer steps are required. Risk Analysis Prev ious attempts to assess the human health risk for surface waters contaminated with Salmonella have been hampered by a lack of quantitative data, such as concentration, as well as large enough sample volumes to be truly representative samples. It is curren tly unknown what is an acceptable microbial quality for waters that are to be used on edible horticultural crops. Many variables surround such a knowledge gap. For example, should one water quality standard be applied to all waters to be used on edible h orticultural crops, or are separate standards for irrigation, frost protection, and foliar applications, more appropriate? Do various modes of irrigation necessitate different microbial water quality standards? Furthermore, it has previously been suggest harvest setting in not realistic (Doyle and Erickson 2012). This is mostly due to the understanding that there exists a widespread distribution of foodborne pathogens in the farm environment (Doyle and E attainable then what is an acceptable level of risk, or even can there be an acceptable level of risk? Currently the Food Safety Moderni zation Act (FSMA) has directed the US FDA to establish science based minimum standards for the production of fruits and vegetables that may be consumed raw, so that the standards would minimize risks of
71 foodborne disease (Harris et al. 2012). Agricultural waters can act as sources for contam ination of fruits and vegetables, so the risk analysis type questions posed here are in imminent need of investigating. In 1989, the US EPA set a goal that all water from surface sources should not pose a risk of infection from waterborne pathogens greater than 1:10,000 per year (Stine et al. 2005). With this as a set goal, a risk assessment was performed by Stine et al. (2005) to determine the microbial water quality required for irrigation water to obtain the same rate of infection from fresh produce con sumption, specifically looking at Salmonella The horticultural items considered included: cantaloupe, iceberg lettuce, and bell peppers. These authors determined that an annual infection risk of 1:10,000 could result from furrow irrigation waters contai ning Salmonella at levels as low as 2.5 CFU/100 ml when harvest occurred the day after irrigation (representing a worst case scenario) (Stine et al. 2005). A greater understanding of risk associated with the use of surface waters on edible horticultural cr ops could be established if larger, more representative, samples could be quantitatively analyzed. To ease the labor burden on producers, indicator organisms or physicochemical indicators of Salmonella presence should be established. Of additional importa nce is an understanding of which serovars of Salmonella are predominating in surface waters relevant to edible horticultural crop production, such as the surface waters of Central Florida, and the duration of the Salmonella survival once contaminating the surface water.
72 Table 2 1. Salmonella isolated from surface waters Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Anatum Fresh Georgia Meandering streams in one watershed 333 ml 1/197 i solates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Bareilly Fresh Georgia Weather pools Moore swab 5 days in river 29/100 Enrichment in peptone, then TT broth (Thomas et al. in pre ss) Bareilly Fresh Georgia Meandering streams in one watershed 333 ml 6/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Blockley Fresh North Dakota River, ca. 22 miles dow nstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 1/6 and 0/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966 ) Blockley Marine Southwestern Morocco Coastal, Tourism and shellfish 100 ml 1/243 samplings ISO 6579 1993 (Setti et al. 2009)
73 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Braenderup Fresh North Dakota River, ca.7, 12 and 22 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 3/6 and 0/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG s ulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5 C (Spino 1966) Braenderup Fresh Mid Atlantic Region Tomato farms, pond sediment 250 ml 1/39 lactose broth, XLT4 and R & F Salmonella Chromogenic p lating m edium (Micallef et al. 2012) Braenderup Fresh Georgia Meandering streams in one watershed 333 ml 12/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Braenderup Fresh North Florida Suwannee River 1 983 ml 2/30 isolates 5x3 MPN with buffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Brandenburg Fresh North Carolina River, Crop agriculture area 25 ml 6/12 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010)
74 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Chester Fresh North Dakota River, ca. 7, 12 and 32 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 0/6, and 3/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5 C (Spino 1966) Cubana Fresh North Florida Suwannee River 1 983 ml 1/30 isolates 5x3 MPN with buffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Enteritidis Fresh North Dakota River, ca. 12, 32, and 62 miles downstr eam from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 0/6, and 3/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966) Gaminara Fresh North Carolina River, forest area 25 ml 16/28 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010) Gaminara Fresh North Carolina River, c rop agriculture area 25 ml 6/12 samplings Buffered peptone water, RV broth, XLT4 (P atchanee et al. 2010)
75 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Gaminara Fresh Georgia Meandering streams in one watershed 333 ml 3/197 isolates from 72 sa mples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Give Fresh North Carolina River, forest area 25 ml 16/28 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010) Heidelberg Fresh North Dakota River, ca. 12 and 62 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 2/6 and 1/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Gre en Agar, all incubated at 41.5C (Spino 1966) I 4,:b Fresh Georgia Meandering streams in one watershed 333 ml 4/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) I 47:z4z 23 Fresh Georgia Meandering streams in one watershed 333 ml 1/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009)
76 Table 2 1. Continued Serovar Water Type Source Area Environm ental conditions Volume Concentration Detection or isolation method Reference Infantis Fresh North Dakota River, ca. 13 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 1/6 and 0/5 sampling locations Moore s wab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966) Inverness Fresh North Carolina River, forest area 25 ml 16/28 samplings Buffered peptone water, RV broth, XLT4 (Patchanee e t al. 2010) Inverness Fresh North Carolina River, Crop agriculture area 25 ml 6/12 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010) Inverness Fresh North Florida Suwannee River 1 983 ml 9/30 isolates 5x3 MPN with buffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Ituri Fresh Democratic Republic of the Congo, Ituri region Duck feces (normal adult) Not reported Not reported Not reported (Kauffmann and Fain 1953) Javiana Fresh Mid Atlantic Regi on Tomato farms, Pond sediment 250 ml 2/39 lactose broth, XLT4 and R & F Salmonella Chromogenic Plating Medium (Micallef et al. 2012)
77 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Johann esburg Fresh North Florida Suwannee River 1 983 ml 1/30 isolates 5x3 MPN with buffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Kasenyi Fresh Democratic Republic of the Congo, Ituri region Duck feces (normal adult) Not reported Not reported Not reported (Kauffmann and Fain 1953) Kentucky Fresh North Dakota River, ca. 0 and 12 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 2/3, 0/6, and 0/5 sampling locations Moore swab he ld in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966) Lille Fresh Mid Atlantic Region Tomato farms, irrigation ditch 250 ml 1/39 lactose broth, XLT4 and R & F Salmonell a Chromogenic Plating Medium (Micallef et al. 2012) Liverpool Fresh Georgia Meandering streams in one watershed 333 ml 6/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009)
78 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Miami Fresh North Carolina River, Crop agriculture area 25 ml 6/12 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010) Mikawasima Fresh Georgia Meandering streams in one watershed 333 ml 12/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Montevideo Fresh Georgia Mea ndering streams in one watershed 333 ml 1/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Montevideo Fresh North Florida Suwannee River 1983 ml 1/30 isolates 5x3 MPN with b uffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Muenchen Fresh North Carolina River, forest area 25 ml 16/28 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010) Muenchen Fresh Georgia Meanderi ng streams in one watershed 333 ml 28/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009)
79 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Muenchen Fresh North Florida Suwannee River 1 983 ml 3/30 isolates 5x3 MPN with buffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Newport Fresh North Carolina R iver, Crop agriculture area 25 ml 6/12 samplings Buffered peptone water, RV broth, XLT4 (Patchanee et al. 2010) Newport Fresh Mid Atlantic Region Tomato farms, pond water 250 ml 4/39 lactose broth, XLT4 and R & F Salmonella Chromogenic p lating m edium (Micallef et al. 2012) Newport Fresh North Florida Suwannee River 1 983 ml 1/30 isolates 5x3 MPN with buffered peptone water enrichment, TT selective enrichment, XLD agar (Rajabi et al. 2011) Not determined Fresh Zimbabwe Nile crocodiles, concrete pond, 29 to 33 C 4.6 log CFU/ 100 ml (DPC) 2.6 log 100 ml (MPN) Spread plate to XLD, slide agglutination and biochem rxn MPN starting with RV (Madsen 1994) Not determined Mari ne San Sebastian, Spain Bathing beaches Not reported 0.8, 4.2, and 8.5% at each of three sites sample number not reported ISO 6579 (buffered peptone RVS & MKTTn) (Serrano et al. 1998)
80 Table 2 1. Continued Serovar Water Type Source Area Environmental c onditions Volume Concentration Detection or isolation method Reference Oranienburg Fresh North Dakota River, ca. 7 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 0/6, and 1/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966) Reading Fresh North Dakota River, ca. 7, 12 and 13 miles downstream from sewage treatment plant Moore swab 5 days in river, jus t below surface 0/3, 4/6 and 0/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966) Rubislaw Fresh Georgia Meandering streams in one watershed 333 m l 26/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Rubislaw Fresh North Florida Suwannee River 1983 ml 6/30 isolates 5x3 MPN with buffered peptone water enrichment, TT se lective enrichment, XLD agar (Rajabi et al. 2011)
81 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference S enterica subsp. arizonae Fresh Georgia Meandering streams in one watershed 333 ml 109/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Saintpaul Fresh North Dakota River, ca. 7, 12, 22, 32, 62, and 73 miles downstream from sewage trea tment plant Moore swab 5 days in river, just below surface 1/3, 0/6 and 5/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966) Saintpaul Fresh Georg ia Meandering streams in one watershed 333 ml 9/197 isolates from 72 samples 5x3 MPN with buffered peptone water enrichment, RV selective enrichment, XLD agar (Haley et al. 2009) Senftenberg Marine Northwestern Spain Ria 25 ml water ISO 6579 (Martinez Ur taza et al. 2004) Senftenberg Marine Southwestern Morocco Coastal, Tourism and shellfish 100 ml 9/243 samplings ISO 6579 1993 (Setti et al. 2009) Tennessee Fresh Mid Atlantic Region Tomato farms, irrigation ditch 250 ml 1/39 lactose broth, XLT4 and R & F Salmonella Chromogenic p lating m edium (Micallef et al. 2012)
82 Table 2 1. Continued Serovar Water Type Source Area Environmental conditions Volume Concentration Detection or isolation method Reference Thompson Fresh North Dakota River, ca 7, 32, 62 and 73 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 0/6, and 4/5 sampling locations Moore swab held in water 5 days, tetrathionate or SBG sulfa enrichment broth to Brilliant Green Agar, all in cubated at 41.5C (Spino 1966) Typhimurium Fresh North Dakota River, ca. 7 and 12 miles downstream from sewage treatment plant Moore swab 5 days in river, just below surface 0/3, 1/6 and 1/5 sampling locations Moore swab held in water 5 days, tetrathionat e or SBG sulfa enrichment broth to Brilliant Green Agar, all incubated at 41.5C (Spino 1966)
83 CHAPTER 3 DETECTION OF SALMONELLA SPP. FROM LARGE VOLUMES OF WATER BY MODIFIED MOORE SWABS AND TANGENTIAL FLOW FILTRATION Reliable screening of fresh produce fo r the presence of foodborne pathogens, including Salmonella is difficult for multiple reasons. Challenges include collecting representative but manageable volumes of water for analysis, how and where to collect representative samples, and the time requir ed for culture based conformations of pathogens (i.e. Salmonella up to seven days; US FDA 1998). This interval may represent a significant portion of the shelf life of often highly perishable commodities, making test and hold screening financially and l ogistically untenable. Broadly viewed, the occurrence of pathogen contamination on produce items has been determined to be sporadic and heterogeneously distributed (Kase et al. 2012), rendering even a single composite sample an unreliable representation o f the entire lot. Testing pre harvest water for Salmonella or E. coli O157:H7 that comes into contact with, or is in close proximity to, produce production may be one component of an effective strategy to implement a risk based approach to enhance produce safety. Large volume water samples can be concentrated and screened pre season and pre harvest, and may provide information for management decisions to reduce the risk of initial contamination. Current microbial water quality standards in the US produce industry rely on testing 100 ml of water for generic E. coli (F DACS 2007, LGMA 2012) rather than pathogen populations. It is likely that pathogen contamination of surface water occurs at low rates (Madsen 1994, Rhodes and Kator 1988, Escartin et al. 2002 Haley et al. 2009, Ijabadeniyi et al. 2011); large volume samples may be required to provide a more realistic representation of the risk of common pathogen presence in the body of water at levels relevant to the mode of application.
84 Various filtration me thods can be used to concentrate large volumes of water. Normal flow filtration (NFF) is the standard water testing method of the American Public Health Association (APHA, APHA 1992), where 100 ml is passed through a sterile filter, and the filter applied directly to an agar plate for enumeration of coliforms. Alternative filtration methods include Tangential Flow Filtration (TFF) where water flows across a filter membrane, Moore s wabs (MS), where rolled cotton gauze is tied and allowed to sit in a body o f water for at least 24 h, and Modified Moore s wabs (MMS) where pieces of pipe are filled with rolled cotton gauze which traps bacteria as water is pushed through the pipe. Previously, up to 100 L of surface water has been concentrated using TFF (Mull and Hill 2009, Gibson and Schwab 2011) and MS have been used to isolate Salmonella (Spino 1966, Sears et al. 1984, Escartin et al. 2002), E coli O157 (Ogden 2011), Campylobacter (Fernandez et al. 2003), and Vibrio cholera (Barrett et al. 1980). MMS have bee n used to isolate E coli O157 and Salmonella (Bisha et al. 2011). Screening for the presence of pathogens is preferred to enumeration of indicator organisms in surface waters as no strong correlations between the presence and/or concentration of pathogens and indicator or index organisms have consistently been documented (Burton et al. 1987, Chandran and Hatha 2005, Pachepsky et al. 2011). The use of PCR based methods for foodborne pathogen detection reduces testing times compared to cultural based method s (Mafu and Sirois 2009). Real Time PCR (qPCR) has the added advantage of increased speed, sensitivity and minimizing post PCR contamination (Balachandran et al. 2011; Delibato et al. 2011).
85 The objectives of this study were to compare i) concentration an d detection of Salmonella in large volumes (10 L ) of surface water by TFF and NFF and ii) concentration of Salmonella and E. coli O157:H7 by TFF and MMS. Materials and Methods Bacterial Strains and Preparation of Inoculum Two different Salmonella cocktails were used in the experiments. NFF and TFF optimization was done using Salmonella serovars Saintpaul (BAC 133); Newport (C4.2); Anatum (LRB6802); Montevideo (LJH0654) and Typhimurium (LJH0738) resistant to naladixic acid (50 mg/l). For the comparison of MMS and TFF identical strains to Bisha et al. (2011) were used; including Salmonella serovars Montevideo (MDD 22 pGFP), Poona (MDD 237 pDsRed) and Newport (MDD 314 pAMCyan) and three Escherichia coli O157:H7 stains (PVTS 016 pGFP, PVTS 087 pDsRed, and PVTS 088 pAMCyan) all transformed to express a differential fluorescent protein moiety and selectable ampicillin resistance at 20 mg/l, as previously described (Bisha et al. 2011). All isolates are available from the culture collection of Dr. Michelle D. Dany luk. Frozen stocks, stored at 80C in Tryptic soy broth (TSB; Difco, Becton Dickinson, Sparks, MD, USA) with 15% glycerol, were streaked onto TSAN or 20 mg/l ampicillin (TSAA; Sigma), and incubated 242 h at 352C. An isolated colony was transferred i nto TSB supplemented with nalidixic acid (TSBN) or ampicillin (TSBA), and incubated 242 h at 352C. A 10 l loop of the culture was transferred to TSBN or TSBA and incubated 242 h at 352C. Cultures were combined in equal parts and diluted in 0.1% pe ptone water (Difco, Becton Dickinson). Following addition to the 10 L water sample, the inoculum was distributed through the water by swirling in a 10 L carboy in a 30 cm circle for 30 s.
86 Inoculum concentrations were determined by serial dilutions in 0.1% peptone water, plating onto TSAN or TSAA and incubation 242 h at 352C. Target concentrations were 10 0 10 3 CFU/L for NFF and TFF optimization, 10 2 10 5 CFU/L for MMS and TFF comparison, and 10 0 CFU/L for MMS validation. Collection of Water Samples F or NFF and TFF optimization and MMS validation, surface water (10 L) was collected from Lake Swoope, in the town of Lake Alfred, F L USA and stored for no longer than 24 h at 4C prior to use. For MMS and TFF comparison 10 L of non sterile 0.85% saline wa s used to emulate the methodology of Bisha et al. (2011). Normal Filtration Vacuum filtration was used to attempt passage of 10 L of water, through a 0.45 size filter (Millipore, Billerica, MA, USA). Whatman Grade 41(Fisher Scientific, New Jersey, USA) filter paper and diatomaceous earth (1 g; Fisher Scientific) were eval uated separately as filtration aids. Tangential Flow Filtration The tangential flow filter used was a Mini Kros Plus Tangential Flow Filter Module (Spectrum Labs, Rancho Dominguez, CA, USA) composed of polyethersulfone with a and a surface area of 1 050 cm 2 Filter and tubing sets were sterilized prior to each experiment. All tubing was L/S 17, constructed of silicone with platinum coating (Cole Palmer, Vernon Hills, IL, USA). A KrosFlo Research II Pump was used in combinati on with Masterflex easy load pump head (Spectrum Labs). The inlet flow rate was standardized at 1 000 ml/min. The TFF was run at a trans membrane pressure of 67 Pa. The initial 10 L of water was concentrated to a final retentate volume of ca. 250 ml.
87 Mo dified Moore Swabs The MMS was constructed out of PVC pipe (12 by 4.4 cm diameter; Home Depot, Winter Haven, FL) with 2.54 cm PCV connector to a 2.54 cm male to male coupler on each end. Inside the PVC pipe, was pre sterilized 80 by 22 cm cheesecloth (Fish er Scientific; Grade #90) folded lengthwise and rolled. The 10 L water sample was passed through the MMS at a flow rate of 300 ml/min. The use of MMS here was introduced as a proof of concept of another concentration means for pathogens, Salmonella and E coli O157:H7, from water; further experiments were not performed using E. coli O145:H7. Conventional Salmonella Detection from NFF and TFF A modified conventional Salmonella isolation method (US FDA 1998) was used. For NFF, the finished filter was place d into 50 ml lactose broth (Difco, Becton Dickinson), and for TFF, a 1:1 volume of double strength lactose broth was added to the retentate. The inoculated lactose broth was allowed to incubate for 1 h at room temperature, before addition of 50 g/l nalid ixic acid or 20 g/l of ampicillin and incubation for 242 h at 352C. A 1 ml aliquot of the incubated lactose pre enrichment was transferred to tetrathionate broth (TT broth; Difco, Becton Dickinson) and Rappaport Vassiliadis R10 broth (RV broth; Difco, Becton Dickinson) and incubated at streaked onto x ylose l ysine t ergitol 4 agar (XLT4; Difco, Becton Dickinson) and Chromagar Salmonella Plus (DRG International Inc, Mou ntainside, NJ, USA). Both media were incubated for 242 h at 352C; colonies were confirmed on lysine iron agar slants (LIA; Difco, Becton Dickinson) and triple sugar iron agar slants (TSI; Difco, Becton Dickinson).
88 Immuno magnetic Separation of Salmonel la during TFF Optimization. Salmonella was captured on antibody coated paramagnetic beads using either Pathatrix (Matrix MicroScience, Golden, CO, USA) or Dynabeads (Invitrogen, Grand Island, New York, USA) microspheres. The entire lactose broth pre enric hment was used in the Pathatrix by replacing the 100 ml conical tube that came with the purchased set up, with a sterile plastic stomacher bag. Either Pathatrix or Dynabeads (50 l) were re set 30 min circulation cycle. Following IMS, beads ( and Chromagar Salmonella Plus; colonies were confirmed on LIA and TSI as described above. Detection of Salmonella During TFF Optimization UltraClean DNA kit (MoBio, Carlsbad, CA, USA). DNA was used to detect Salmonella by PCR ( invA and oriC Salmonella spp. Detection Kit (Applied Biosystems, Carlsbad, CA, USA) according to instructions in a BioRad CFX96 RealTime thermocycler. The invA and oriC primer sets were as described previously by Malorny et al. (2003). The conventional PCR reaction was optimized using overnight cultures of the Salmonella cocktail used in the TFF o ptimization approach. All reagents were obtained from the Fisher exACTGene Complete PCR kit (Fisher Scientific). Primers were used The optimized PCR conditions were: 3 min at 94C melting, followed by 30 cycles of 30 s at 94C, 30 s at 58C, 1 min at 72C; a final elongation of 5 min at 72C. Gel
89 electrophoresis on a 1.8% agarose gel with 0.5X TBE buffer. DNA was stained with ethidium bromide (1 (MultiDoc It Digital Imaging System; UVP, Upland, CA, USA). PCR amplicon size was determined by comparison to molecular size ma rkers (Hyperladder V; Bioline USA Inc., Randolph, MA, USA). Detection of Pathogens from MMS Following MMS filtration, the cheesecloth was removed and placed in a stomacher bag. Excess liquid (ca. 25 35 ml) was wrung out of the cheesecloth and passed throu size, 47 mm diameter filter using NFF vacuum filtration. The filter was placed on TSAA and incubated for 242 h at 352C. Colonies were checked for fluorescence, and confirmed by REVEAL Salmonella or E coli O157:H7 lateral flow device (Neogen Corp., Lansing, MI, USA). Validation of MMS Concentration MMS and TFF were used to concentrate a 10 0 CFU/L Salmonella spiked 10 L water sample, and retentates pre enriched in lactose broth. The entire pre enrichment was used in the Pathatrix sy stem with the Pathatrix beads (50 l); beads (10 l) were plated to XLT4 and Chromagar Salmonella Plus and typical colonies confirmed. Remaining beads (40 l) were used in DNA extraction and qPCR was performed, all as described above. Statistical Analysis Experiments evaluating TFF and detection methods were repeated six times for each dilution reported. Experiments comparing MMS to TFF were repeated in triplicate.
90 Results Results of the concentrati on and detection of Salmonella by TFF for surface water samples spiked with the five strain cocktail of Salmonella at varying concentrations are shown in Table 3 1. Detection of 1 CFU/L in 10 L was consistently obtained within two days when using the TFF immuno capture Real Time PCR method. The use of the modified conventional FDA BAM method following the TFF concentration and overnight pre enrichment only detected Salmonella in 5 out of 6 water samples when spiked at 1 CFU/L in 10 L. Taking the immuno m agnetic beads straight from the Pathatrix and spread plating them onto Chroma gar Salmonella Plus or XLT4 agar was not able to consistently detect Salmonella at spiked levels below 76 CFU/L in 10 L of surface water (Table 3 1). Inconsistent rates of detect ion were achieved when taking the immuno magnetic beads to conventional PCR. Switching to the more convenient DynaBeads brand of immuno magnetic beads also gave inconsistent rates of detection at all concentrations of Salmonella attempted. No Salmonella was detected when a 0.250 ml aliquot of the permeate was spread plate to TSAN and incubated overnight, ensuring that the filter had not unknowingly ruptured during usage. Salmonella detection in the permeate would indicate a failure of the TFF and that th e bacteria were not being concentrated. The 10 L surface water sample was not able to completely pass through the NFF filter. Approximately 500 ml (data not shown) could be passed through the filter before the filter became irreversibly clogged. The addi tion of filter aids (Whatman Grade 41 filter paper or diatomaceous earth) did not significantly improve the volume of water filtered prior to irreversible clogging. No further efforts were made to concentrate 10 L surface water samples by NFF.
91 A compariso n of MMS to TFF using non sterile saline as the starting medium and detecting the pathogen by capturing bacteria from the respective retentates on to a filter membrane and growing overnight on TSA amp is shown in Table 3 2. MMS is consistently able to det ect E. coli O157:H7, while the TFF results were more variable, with only the highest concentration (ca. 100,000 CFU/L in 10 L) of E. coli O157:H7 detected in all samples. Salmonella was detected by TFF in samples containing concentration of Salmonella. Only at the highest concentration (>100,000 CFU/L in 10 L) of Salmonella was the MMS concentration able to detect Salmonella in all the samples. Due to the poor detection of Salmonella work comparing MMS to TFF, a final set of experiments to compare these concentration protocols using Salmonella detection method s described in TFF optimization were carried out. When surface water was inoculated with the Salmonella fluorescence cocktail at 2 CFU/L, concentrated using either the MMS or TFF, and coupled with the modified conventional FDA BAM culture method, immuno m agnetic separation using the Pathatrix beads followed by culture methods or Real Time PCR, Salmonella was detected from all samples (n=6, data not shown). Using more rigorous Salmonella detection methods, both MMS and TFF concentration protocols perform e qually well in detecting low levels of Salmonella (<10 CFU/L) from spiked surface water samples. Discussion The public health impact of the use of recreational water quality standards based on testing 100 ml water used in produce production is poorly estab lished. Poor correlation between pathogens presence/a bsence and indicator organisms (Rhodes
92 and Kator 1988, Bae and Wuertz 2012) makes it is advantageous to screen for specific pathogens in agricultural waters, rather than enumerating indicator organisms. Screening for the presence of pathogens is a preferential practice over the enumeration of indicator organisms because of differing survival times and population sizes that have been observed to occur in environmental waters (Burton et al. 1987, Chandran and Hatha 2005). Reproducible detection of Salmonella in 10 L of surface water was accomplished within two days using the TFF or MMS concentration, pre enrichment in lactose broth, further concentration by immunocapture and detection by qPCR methods descr ibed. Levels of as low as 1 CFU/L in 10 L were reliably detected in spiked surface waters. Salmonella populations at levels below 1 CFU/L may be below a produce safety significance threshold level, thus the detection limit described here are believed to be low enough to meet all practical screening purposes for surface waters used in the produce industry. The proposed method detected Salmonella more consistently at low levels than the culture alone based methods of screening for Salmonella ; this may resu lt fewer false negative reports when Salmonella is present at low levels. An additional benefit of the proposed method is that only two days are required for completion, whereas the current standard culture based methods for Salmonella testing requires u p to seven days (US FDA 1998) The use of NFF for 10 L of the Florida lake surface water was not possible, as the filter irreversibly clogged prior to the sample passing completely through. While NFF is part of the APHA standard methods for screening surfac e for various pathogens (APHA 1992), the recommended volume for screening is 100 ml. Screening higher
93 volumes of water may be necessary due the likelihood of pathogens being present at l ow levels, if and when present (Rhodes and Kator 1988, Madsen 1994, Escartin et al. 2002, Haley et al. 2009, Ijabadeniyi et al. 2011). TFF has been successfully used to screen large volumes of surface water for the presence of specific pathogens. A major advantage of TFF over NFF is the flowing retentate removes the filter cake during the run; this increases the time before the filter clogs allowing greater volumes of water to be processed. Using 100 L surface water, a recovery efficiency rate of 70.5% for E. coli (starting concentration 10 to 100 CFU/L) coupled with a cul ture based assa y has previously been reported (Gibson and Schwab 2011). Others have used TFF, coupled with continuous centrifugation and immun o magnetic separation followed by qPCR to detect E. coli O157:H7 at 50 MPN per 40 L of surface water with a rat e o f 100% detection (Mull and Hill 2009). A hollow fiber ultrafiltration (similar to TFF, but with a dead end instead of recirculation) method is capable of recovering up to 93% when 3 log CFU of Enterococcus faecalis were inoculated into 100 L of tap or sur face water when the turbidity was low (0.29 NTU) (Smith and Hill 2009). With higher turbidity (4.3 NTU), the authors found a 78% recovery rate for the same starting concentration of E. faecalis (Smith and Hill 2009). The use of MS is common in water scre ening (Rhodes and Kator 1998, Madsen 1994, Escartin et al. 2002, Haley et al. 2009, Ijabadeniyi et al. 2011). MS have previously been used to isolate Salmonella from surface (Spino 1966, Sears et al. 1984) and fountain water (Escartin et al. 2002), E co li O157 from soil and field water (Ogden 2001), Campylobacter from river water (Fernandez et al. 2003), and Vibrio cholera from sewage (Barrett et al. 1980). In these examples, the MS was secured it in
94 the body of water for a duration of time, rather than a specified volume of water run through the swab. Previous studies evaluating agricultural water, and pumping water through MMS demonstrate concentration abilities of MMS. Starting concentrations of 100 CFU/L of either E. coli O157:H7 or Salmonella were detected when either MMS or continuous flow centrifugation was used to concentrate 10 L starting volume of irrigation water (Bisha et al. 2011). The authors do not state whether they attempted the detection of lower levels. However, as the concentration step was followed by a culture based method, without enrichment, rather than enrichment followed by PCR bases methods, the limit of detection may be improved. MMS was better at concentrating lower levels of both Salmonella and E. coli O157:H7, than the TF F from saline and detecting by the same culture based method described by Bisha et al (2011). This decrease in TFF performance may be due to the use of a culture based method for detection in that the cells may be damaged from circulating in the TFF syst em. Using a subsequent immuno magnetic concentration following pre enrichment of Salmonella and qPCR to detect the pathogen, both methods of concentration, MMS and TFF, performed equally in detecting Salmonella at 2 CFU/L in 10 L of surface water. Repla cing the TFF with a MMS does not appear to alter the detection of Salmonella spp. in the optimized method. The use of MMS as the concentration method may allow for additional advantages to the produce industry in that MMS are inexpensive, simple to use, c an be used in the field rather than collecting and returning large volumes of water them to the laboratory, and require minimal personnel training prior to use.
95 The method described here for screening large volumes of surface water enables larger, more rep resentative, samples to be screened for Salmonella presence. Detection of Salmonella within two days following sampling is advantageous in determining food safety risks associated with surface waters and this may be advantageous for the produce industry i n avoiding pre harvest contamination.
96 Table 3 1 Percent Salmonella positive samples of 10 L inoculated surface water, concentrated using TFF and pre enriched in lactose broth evaluated using different detection techniques (n = 6). Inoculum Concentrati on (CFU/L) Detection Technique a Culture DB/qPCR DB/PCR DB/Culture PTX/Culture PTX/qPCR PTX/PCR 1 83% 100% 17% 33% 67% 100% 50% 7.5 100% 100% 33% 33% 67% 100% 17% 76 100% 83% 67% 100% 100% 100% 50% 760 100% 100% 67% 83% 100% 100% 33% McNemar P value 0.0736 0.0736 0.0009 0.0077 0.0077 Control 0.0003 a Detection techniques included: Culture, TT broth and RV broth enrichment, streaking onto XLT4 and Chroma gar Salmonella Plus ; DB/qPCR, DynaBeads for IMS, DNA extraction, and detection by real time PCR (C t < 35); DB/PCR, DynaBeads for IMS, DNA extraction, detection by conventional PCR (band on agarose gel at expected size; Malorny et al. 2003); DB/Culture, DynaBeads for IMS, beads spread onto XLT4 and Chroma gar Salmonella Plus; PXT/Culture, Pathatrix used for IMS, beads spread to XLT4 and Chroma gar Salmonella Plus; PTX/qPCR, Pathatrix beads for IMS, DNA extraction, detection by real time PCR; PXT/PCR, Pathatrix beads for IMS, DNA extraction, detection by conventional PCR
97 Table 3 2 Percent Salmonella po sitive samples of 10 L inoculated non sterile saline, pre concentrated by TFF or MMS and detected by cultural means a (n = 3). Pathogen Inoculum Concentration (CFU/L) MMS TFF Salmonella spp. 137 67% 0% 1,366 67% 33% 13,660 67% 100% 136,604 100% 100 % E. coli O157:H7 86 100% 33% 865 67% 67% 8,652 100% 33% 86,524 100% 100% a Detection by cultural means involved the liquid retentate being passed through a size, 47 mm diameter, the filter plac ed on TSAA and incubat ed for 24 2 h at 35 2C A positive sample is noted by the presence of a fluorescing colony that also gave positive results from the appropriate REVEAL test.
98 CHAPTER 4 PREDICTING SALMONELLA POPULATIONS FROM BIOLOGICAL, CHEMICAL, AND PHYSICAL INDICATORS I N FLORIDA SURFACE WATER Water quality can be characterized by three types of analysis; index microorganisms, indicator microorganisms, and other physicochemical water presence of fecal contamination (Smith and Schaffner 2004). Index organisms represent the presence and behavior of a pathogen in a given environment (Mossel et al. 1995); one organism can be both an index and an indicator organism. Physicochemical water characteristics include turbidity, temperature, pH, or oxidation reduction potential (ORP) (APHA 1992). Physicochemical measurements have the advantage of being considerably more rapid than microbial based measurements, but such measurements may not corr elate with microbiological quality. The US produce industry currently relies on testing 100 ml of water for indicator organisms, specifically, generic E. coli as a means to monitor water quality used for edible horticultural crop production (LGMA 2012, F DACS 2007). There are a wide variety of bacteria, both groups and species, viruses, and bacterio phages that have been used or proposed for use as indicator microorganisms (Goyal et al. 1977, APHA 1992, Feng et al. 2010, Bae and Wuertz 2012). Coliforms, either total or fecal, are a common choice of indicator organism. Total coliforms are aerobic facultative anaerobic, Gram negative, non spore forming, rod shaped bacteria that produce gas upon lactose fermentation within 48 h at 35C; fecal coliforms are a su b set of coliforms that also ferment lactose at 44C. Generic E. coli has been proposed both as an indicator and index microorganism for enterohemorrhagic E. coli Salmonella spp., and Shigella spp. (Leclerc et al. 2001).
99 When contaminated agricultur al water is used for irrigation, pesticide and fertilizer application, and/or frost protection, the safety of fresh produce items has been affected (Miles et al. 2009, Erickson et al. 2010, Ijabadeniyi et al. 2011). In 2005 a large scale outbreak of S Ne wport from tomato consumption occurred, the trace back investigation revealed the same serovar was in the irrigation pond of the farm from which the suspect tomatoes were harvested (Greene et al. 2008). Another large outbreak in 2009 was epidemiologically linked to Serrano peppers, which came from a farm where a matching S Saintpaul was isolated from the irrigation pond (Klontz et al. 2010). Therefore, the quality of water applied to fresh produce is of utmost importance in ensuring food safety. Contradi ctory results have been reported as to the efficacy of index or indicator organisms in predicting the presence and/or prevalence of human pathogens, such as Salmonella in surface waters (Burton et al. 1987, Rhodes and Kator 1988, Chandran and Hatha 2005, Chandran et al. 2011). While any specific pathogen may not be present in the sample being tested, the presence of indicator microorganisms is meant to suggest that pathogens have a reasonable likelihood of being present (Smith and Schaffner 2004). The tes ts for indicator microorganisms are less cumbersome and time consuming, as well as typically less expensive, as compared to screening for the presence of individual pathogens. This makes the use of indicator microorganisms a much more economical and pract ical choice over screening for individual pathogens. Physicochemical water characteristics, such as turbidity, temperature, pH, and ORP have all been used to monitor water quality in rivers and lakes (APHA 1992). However, the strength of the correlation i s not discussed by the APHA in their methods
100 manual. Physical measurements like temperature and rainfall (R 2 = 0.317) have been observed to correlate with Salmonella isolation frequency in seawater (Martinez Urtaza et al. 2004). The main advantage of mon itoring a physicochemical characteristic is that nearly instantaneous results that can be obtained and used to predict water quality so that a risk management decision can be made in a timely manner. The relationships between the presence and/or concentrat ion of Salmonella and biological, physical or chemical indicators in Central Florida surface water samples taken monthly for twelve consecutive months was explored. Materials and Methods Water Sampling Eighteen surface water sites in Central Florida were s ampled monthly for twelve consecutive months starting in August 2010. Sites in rural agricultural areas, away from animal agriculture were selected, and included ponds, creeks, rivers, and canals. Sites 1 6 were located in the center of Central Florida, sites 7 12 were located in Eastern Central Florida, and sites 13 18 were located in Western Central Florida. All sites were in rural agricultural areas, away from animal agriculture. Sampling locations included two lakes, one pond, six creeks, two stream one river, and six canals. The land use and catchment watershed for each site is detailed in Table 4 1. Water was collected in 10 L sterile carboys (Nalgene, Rochester, NY, USA) fitted with 4 kg of lead weights attached to a rope. The rope was used to lower the carboy to a depth of 20 cm into the water source. Samples from all but one location were collected away from shore, using a bridge or other outcropping. Samples were collected on three separate days each month (six samples per day). All sample s were taken before solar noon (when the sun is at its highest elevation of the day), from a
101 shaded area of the water source. Chemical and physical water characteristics were measured immediately after sampling as described below. Samples were transporte d to the laboratory and stored at 4C for up to 24 h before microbial testing as described below. Chemical and Physical Water Characteristics Turbidity, temperature (air and water), pH, and oxidation reduction potential (ORP) were measured. Each measureme nt was repeated in triplicate. Turbidity was measured in Formazin Attenuation Units (FAU) using a portable colorimeter (DR/850, temperatures were measured with a portable temperature probe (SH66A, Cooper Instrument Corporation, Middlefield, CT, USA). The pH value and ORP were measured with a portable pH/ORP meter (pH6 Acorn series, Oakton, Vernon Hills, IL USA). Total Aerobic Plate Count Surface water samples were spread p lated in duplicate, onto tryptic soy agar (TSA; Difco, Becton Dickinson) following serial dilutions in 0.1% peptone water (Difco, Becton Dickinson, Park, MD, USA) to determine total aerobic counts. Plates were incubated 242 h at 352C and all bacterial colonies were enumerated by hand. Coliform/ E. coli MPN Colisure Presence Absence Snap Packs (IDEXX Laboratories, Inc., Westbrook, Maine, USA) were used to determine coliform and E coli Most Probable Numbers (MPN) in a five by three MPN configuration (10, 1, 0.1 ml dilutions). Tubes were incubated 24 h at 350.5C. A yellow color indicated coliforms and E. coli was determined by observing fluorescence using a 6 watt UV lamp. The MPN/100 ml was
102 determined from the table in Standard Methods for the Examina tion of Water and Wastewater, 18 th ed. (APHA 1992). Salmonella Detection Salmonella was determined in each water sample as previously described in Chapter 3. Briefly, each water sample was concentrated using tangential flow filtration (TFF) to a final ret entate volume of ca. 250 ml. A KrosFlo Research II Pump (vendor, location) was used in combination with Masterflex easy load pump head (Spectrum Labs, Rancho Dominguez, CA, USA). The inlet flow rate was 1 000 ml/min. A Mini Kros Plus Tangential Flow Filt er Module (Spectrum Labs) made of polyethersulfone with a pore rating of 0.2 m and a surface area of 1 050 cm 2 was used as a tangential flow filter. The TFF was run at a transmembrane pressure of 67 Pa. Double strength lactose broth (250 ml; Difco, Bect on Dickinson) was added to the retentate, which was then stored for 1 h at room temperature and incubated 242 h at 352C, to pre enrichment the sample. Pre enrichment was followed by an immuno capture of Salmonella using the Pathatrix system (Matrix Micr oScience, Golden, CO, USA). The method was modified such that the entire pre enrichment was placed in a sterile stomacher bag (17.8x30.5 cm; Fisher Scientific, New Jersey, USA; instead of the typical 100 ml conical tube). Pathatrix Salmonella capture bead s (50 l) were added to the system and used set 30 min cycle. The remaining beads (40 l) were used in a DNA extraction using the MoBio UltraClean DNA kit (MoBio, Carlsbad, CA, USA). The extracted DNA was used in real Salmonella spp. Detection Kit (Applied
10 3 Biosystems, Carlsbad, CA, USA). The real time PCR kit was used according to Time thermocycler. Modified MPN f or Salmonella Positive Water Samples A 50 ml retain sample of each water sample was removed prior to processing. A modified Salmonella MPN method was used to determine Salmonella concentration (US FDA 1998) for each retained sample. A three by thre e tube MPN was set up and dilutions were as follows: 10 ml in 10 ml double strength lactose broth, 1 ml in 9 ml single strength lactose broth, and 0.1 ml in 9 ml single strength lactose broth. These were incubated for 242 h at 352C. Selective enrichme nt was done by transferring a one ml aliquot to tetrathionate broth (TT broth; Difco, Becton Dickinson) and 0.1 ml to Rappaport Vassiliadis broth (RV broth; Difco, Becton Dickinson); selective enrichment broths were incubated 242 h at 352C and 482 h at 412C, respectively. Ten l was streaked onto XLT4 and Chromagar Salmonella Plus and incubated for 242 h at 352C following enrichment. Colonies displaying typical Salmonella phenotypes were confirmed biochemically on LIA and TSI slants as described a bove. One representative biochemically confirmed Salmonella colony from each plate was transferred to TSA and incubated for 242 h at 352C. One colony was transferred to tryptic soy broth and incubated 242 h at 352C and DNA extraction using the MoBio UltraClean DNA kit. Salmonella was genetically confirmed by PCR of the invA and oriC genes. The invA primers were GTGAAATTATCGCCACGTTCGGGCAA and TCATCGCACCGTCAAAGGAACC, giving a PCR product of 284 bp; the oriC primers were TTATTAGGATCGCGCCAGGA and AAAGAA TAACCGTTGTTCAC, giving a PCR product of 163 bp. Both primer sets were as described by (Malorny et al. 2003). All reagents were obtained from the Fisher
104 exACTGene Complete PCR kit (Fisher Scientific, NJ, USA) and primers were used at a concentration of 20 M. The PCR reaction mix was as follows: 34.75 l water; 5 l 10X PCR buffer; 1 l of each primer; 0.25 l Taq DNA polymerase; 5 l template DNA. The optimized PCR conditions were: 3 min at 94C melting, followed by 30 cycles of 30 s at 94C, 30 s at 58 C, 1 min at 72C; a final elongation of 5 min at 72C. Gel electrophoresis was done using a 1.8% agarose gel with 0.5X TBE buffer. Gels were run at 70 V for 90 min. Weather Data Collection Rainfall and solar radiation data were collected from the Florid a Automated Weather Network ( http://fawn.ifas.ufl.edu/ ) The closest weather recording station to each individual sampling site was determined and the data applied accordingly. Statistical Analysis Correlations wer e determined using JMP Pro 9. Multiple linear regression analysis of Salmonella vs. all parameters was done using SAS 9.3 (SAS Institute Inc., NC, USA) software. A p value of < 0.05 was considered significant. Results Two hundred and two samples were colle cted and analyzed over the twelve month sampling period. Fourteen samples were missed due to laboratory error or inaccessibility of the water source due to unrelated factors (e g. nearby herbicide application precluded access due to investigator safety). The correlation coefficients for each characteristic vs. Salmonella log MPN/100 ml are displayed in Table 4 2. When each R 2 value is calculated for individual sampling sites 152 of the 162 of the R 2 values fall below 0.4 indicating a lack of correlation Each characteristic will be discussed further in the appropriate section below.
105 Salmonella MPN All 202 water samples were positive for Salmonella using the TFF immuno capture r eal t ime PCR method. The detection level for the described methodology has a detection limit of ca. 1 CFU/L in 10 L (McEgan et al. 2013); so all 10 L surface water samples collected had greater than 1 CFU/L Salmonella spp. in each 10 L sample volume collected and processed. Salmonella concentrations for each site by sampling dat e can be seen in Figure 4 1. The median Salmonella concentration was 0.79 log MPN/100 ml for all samples taken; the upper quartile was 1.2, the lower quartile was 0.48 log MPN/100 ml; and 139 of 202 samples had Salmonella concentrations below 1 log MPN/ 1 00 ml. The highest concentration of Salmonella, 3.0 log MPN/100 ml, was identified in July (Fig 4 1); however, the month of sampling did not exhibit a strong correlation with the concentration of Salmonella (R 2 =0.20). Coliform and E. coli MPN Values for c oliform and E. coli log MPN/100 ml can also be seen in Figure 4 1. Many of the coliform (128/202), and some of the E. coli (17/202), MPNs were at or over log MPN/100 ml. E. coli had a median concentration of 1 .7 log MPN/100 ml, an upper quartile of 2.4 and lower quartile of 0.60 log MPN/100 ml. Neither coliform log MPN/100 ml (R 2 =0.004; Table 4 2), nor E. coli log MPN/100 ml (R 2 =0.078; Table 4 2) values were strongly correlated with the Salmonella MPN values. However, at sites 11 and 18, E. coli log MPN/100 ml resulted in higher correlations, R 2 =0.606 and R 2 =0.678,
106 respectively. Interestingly, coliform and E. coli log MPN/100 ml were not strongly correlated (R 2 =0.364; Table 4 3) with each other. Aerobic Plate Count The values for each aerobic plate count for each site can be seen in Figure 4 1. The median aerobic plate count was 3.7 log CFU/100 ml; the upper quartile was 4.0 log CFU/100 ml, and the lower quartile was 3.2 log CFU/100 ml. The two highest aerob ic plate counts, 5.2 and 5.5 log CFU/100 ml (Figure 4 1 H and 4 1 D), both occurred during the month of July. The aerobic plate count, in log CFU/100 ml, did not correlate with the Salmonella log MPN/100 ml (R 2 =0.004; Table 4 2). The aerobic plate count did not strongly correlate strongly with any of the other variables recorded; all R 2 values were <0.212, and p values ranged from 0.8620 to 0.0039. Air and Water Temperature Air and water temperatures are displayed in Figure 4 2. Not surprisingly, recorde d air and water temperatures showed the strongest correlation to each other, of any two variables (R 2 =0.680; Table 4 3). Neither the air nor water temperature correlated with the Salmonella log MPN/100 ml (R 2 =0.000 and R 2 =0.006, respectively; Table 4 2). However, when calculated on an individual site bases, site 14 (R 2 =0.405; Table 4 2) and site 15 (R 2 =0.403; Table 4 2) Salmonella concentrations showed relatively higher correlation with water and air temperatures, respectively. pH, ORP, Turbidity, and Co nductivity pH and ORP values by sampling site over time are displayed in Figure 4 3; and turbidity and conductivity by sampling site over time are displayed in Figure 4 4. None of these physical water characteristics correlated with Salmonella log MPN/100 ml, as show by the R 2 values in Table 4 2. However, when calculated by site, ORP had a
107 relatively high correlation with Salmonella concentration at site 9 (R 2 =0.424) and site 10 (0.555); while turbidity had a relatively high correlation with Salmonella c oncentration at site 4 (R 2 =0.691) and site 11 (R 2 =0.584); and conductivity had a relatively high correlation with Salmonella concentration at site 4 (R 2 =0.405) and site 17 (R 2 =0.450). pH correlations were below 0.4 at all sites. Rainfall and Solar Radiati on The average rainfall (in the previous day, week or month) before sampling did not correlate with Salmonella coliform, E. coli or aerobic plate count levels (data not shown). In addition solar radiation was also poorly correlated with Salmonella, coli forms, E. coli or aerobic plate counts levels (data not shown). Discussion Salmonella is present in Central Florida surface waters throughout the year. Surveys of surface waters for Salmonella in the South Eastern US have reported similar findings; in No rth Carolina surface water had a Salmonella prevalence of 54.7% (94/172) in 25 ml water samples (Patchanee et al. 2010), and in a Southern Georgia watershed 79.2% of water samples were Salmonella positive using a 5 replicate, 3 dilution (100, 10, 1 ml) MPN (Haley et al. 2009). A survey of the Suwannee River in North Florida reported 96% (106/110) of water samples Salmonella positive using a three replicate, five dilution (500, 100, 50, 10, 1 ml) MPN method (Rajabi et al. 2011). These findings are also sim ilar to those found by other researchers performing surface water samplings over extended periods of time at the same set of sites in more geographically diverse locations. Two rivers and one creek all in one Southern Ontario, Canada water shed had 78.4% (91/116) Salmonella positive samples when locations were sampled by 5 day Moore swabs twice monthly for 2.5 years (Thomas et al. in
108 press). Other studies have reported much lower Salmonella prevalence in surface waters. Surface waters in the Salinas Vall ey, CA, surveyed using either 3 day Moore swabs or a 100 ml sample reported 7.1% (18/252) Salmonella positive samples (Gorski et al. 2011). The prevalence of Salmonella in the coastal waters of southern Morocco during a 4 year monthly sampling of six site s was reported at 4.1% (10/243) (Setti et al. 2009). Similarly, Salmonella in surface waters collected from fruit and vegetable farms in New York state was reported at 11% (16/145) (Strawn et al. 2013). Overall, the presence of Salmonella throughout the year in Central Florida surface waters is similar to surface water surveys performed throughout the Southeastern US, especially when considered the increased sample volume used in the current study. None of the index microorganisms measured strongly correl ated with the prevalence of Salmonella in the same Central Florida surface water sample. Many of the coliform log MPN/100 ml were at or above the upper limit of detection may be responsible for the lack of correlation between Salmonella and coliform popul ations. In the cases where coliform populations were at or above the upper limit of detection, the population could have in fact been higher and this would not have been able to be reflected in the data point. All water samples in this study with colifor m counts above 3.2 log MPN/100 ml, the upper limit of detection, would by the design of the enumeration procedure be recorded as the calculations of the correlation coefficients. Numerous publications have described the lack of, or presence of weak correlations between indicator microorganisms and pathogens in surface water (Goyal et al 1977, Payment and Locas 2011). Moderate correlations were obtained when E.
109 coli or coliform data from the current study were analyzed for correlations with Salmonella populations, for individual sites (Table 4 2); however, when the data was aggregated f or all sites correlations between E. coli or coliforms and Salmonella both had R 2 < 0.1 indicating that other factors beyond those studied here influence the relationships between pathogens and indicator microorganisms. Studies that report similar results conclude that high correlations between pathogens and indicator microorganisms may be temporal, random, site specific or time specific (Payment and Locas 2011). High correlations between pathogens and indicator microorganisms often occur at point sources where the levels of indicator microorganisms and pathogens are consistently high for a longer period of time (Payment and Locas 2011), such as surface waters impacted by improperly treated sewage or runoffs from livestock farms. Our study was conducted i n rural agricultural Florida where the influence of factors such as raw sewage effluent was unexpected. The occurrence of Salmonella in the environment may be at a much higher frequency than previously assumed; Salmonella may also not be transient in natu re within surface waters as was also previously believed (Buchanan 2000). Not only did our analysis not find strong correlations between E. coli or coliform populations and Salmonella both were also not correlated with rainfall or seasons as would have be en predicted (Goyal et al. 1977). In our analysis, average rainfall, whether aggregated of 24 h, one week, or one month (prior to sampling), did not correlate with Salmonella E. coli or coliform levels. Conflicting results have been reported in several o ther studies that also attempted to determine a correlation between rainfall and pathogen prevalence. A Southern Ontario surface water study found the lowest
110 occurrence of Salmonella did coincide with the month (February) with the lowest recorded precipit ation, air temperature, and water temperature (Thomas et al. in press). A positive correlation between Salmonella prevalence and rainfall (mm/day) has been reported, without R 2 during a study in southern Morocco (Setti et al. 2009). The Morocco study an alyzed 100 ml of seawater collected monthly at six sites for a four year duration. In New York, Salmonella has been reported to have a higher prevalence in surface waters when measurable precipitation occurred within three days prior to sampling, however, this was only observed in areas of poorly drained soils (Strawn et al. 2013). The differing results could be suggestive that rainfall, does have some effect on Salmonella prevalence in surface waters, but not as a direct correlation and perhaps in a mult i factorial way including more characteristics than those noted in the current study. A study conducted in Puerto Rico attempted to correlate rainfall from 24 h, 48 h, and one week prior to water sampling for fecal coliforms (coliforms that grow at 42C (A PHA 1992); a less inclusive group compared to total coliforms which were measured in the current study) found no correlation between fecal coliforms and precipitation in any of the ten sampling sites (Santiago Rodriguez et al. 2012). These authors suggest that although fecal coliforms, commonly found in tropical sediments, may be resuspended from preceding precipitation that the rainfall may have a possible dilution effect as well on the fecal coliforms (Santiago Rodriguez et al. 2012). Some studies have indicated that turbulence caused by rainfall could resuspend pathogens or indicator microorganisms from sediments, raising their levels in the waters above by 100 or 1 000 fold, and that the depth and size of the water body dampen any rainfall
111 dilution eff ect on the indicator/pathogen levels (Goyal et al. 1977). Our study did not find a correlation between rainfall and Salmonella or E. coli and coliform populations. At no point during the water sampling procedure was the sediment disturbed. This could in dicate that sediment disturbance is not a major contributing factor to Salmonella in Central Florida surface waters. Correlations were not influenced by season, possibly due to the relatively small temperature changes in the sub tropical climate. However, in Southern Ontario and New York State, areas with much larger seasonal temperature variations, Salmonella positive samples were not significantly different between seasons (Thomas et al. in press, Strawn et al. 2013). The rise and fall in the levels of biological characteristics was random and did not also coincide with increasing temperatures in the summer and reduced temperatures in the winter. This is contrary to the finding of Gorski et al. (2011), who note a noticeable seasonal trend in the prevale nce of Salmonella in surface water samples from the Salinas Valley C A Further work is required to acquire a better understanding if and over what range water temperature would have an effect on Salmonella prevalence and whether geographic location or ot her climacteric factors also have an effect. Physicochemical water characteristics have been used to monitor the chemical and microbiological quality of drinking and recreational water. Data from some sites (Table 4 2) showed that turbidity, ORP and condu ctivity could predict the levels of Salmonella in surface water, but when aggregated across all sites none of these parameters produced acceptable correlations. Although R 2 values were low, indicating additional unmeasured factors may be exerting an influ ence.
112 When E. coli or solar radiation levels increase, these increases are correlated, significantly (P<0.05) but with a low R 2 with increases in Salmonella levels. The positive correlation of solar radiation with Salmonella may be partially explained by Salmonella being more resistant to solar radiation than other bacterial pathogens, such as E. coli Shigella flexneri and Vibrio cholera (Berney et al. 2006). Potentially leaving fewer bacteria to compete for nutrients after increases in solar radiation thus Salmonella would have access to a greater amount of nutrients as solar radiation increased and inactivated the less resistant bacteria; less nutrient competition as well as nutrients made available through the inactivation of other bacteria may be b enefitting Salmonella survival. When conductivity levels increase, there is a statistically significant (p=0.000) inverse correlation with E. coli concentration. Conductivity of surface water is an indirect measure of salinity and total dissolved solids, these have previously been suggested to be influenced by storm water and runoff (Gonzalez et al. 2012). An inverse correlation between indicator bacteria ( E. coli, enterococci, and Bacteroides ) and conductivity or salinity has also been reported for estua ry waters in eastern North Carolina (Gonzalez et al. 2012). The inverse correlation noted in the current study is perhaps also due to storm water and runoff effects, in this case the addition of the storm water or runoff in carrying in salinity or other t otal dissolved solids from the surrounding area, whilst diluting the E. coli concentrations. The sampling sites used in the current study were not influenced by seawater. Though these relationships are statistically significant, low coefficient of correl ation vales (< 1 or < 1) values indicate that other as yet unknown factors are also involved.
113 The complex nature of the index or indicator and pathogen relationship, makes predicting the levels of pathogens through index/indicator microorganisms challengi ng at best; simple, linear relationships cannot be relied upon for predicting pathogen levels from indictor populations. Logistic regression analysis could be used in this case to further develop models to better predict the levels of Salmonella using ind ex/indicator microorganisms or physicochemical water characteristics. The absence of strong correlations (low R 2 ) between index/indicator microorganisms, chemical or physical indicators with Salmonella levels limits their ability to predict the prevalence of Salmonella However significant positive relationships have been shown to occur between E. coli and Salmonella levels and between solar radiation and Salmonella levels. These may offer a fast but qualitative indication of the degree of Salmonella risk. However, until a more reliable, and quantitative, index or indicator of Salmonella in surface waters can be established, direct screening for Salmonella should occur. Finally given the dynamic, heterogonous and complex nature of surface water ecosystems the challenge remains to find an indicator parameter that has a more intimate relationship besides just being associated with the pathogen and therefore less sensitive to extraneous factors.
114 Table 4 1 Description of 18 Central Florida surface water sa mpling sites including water source type, land use, and watershed as described by local water management district Site Number Water source type Immediate area land use Percentage Watershed Agriculture Percentage Watershed Rangeland Percentage Watershed Urb an/Suburban 1 Lake Agriculture, Wetlands, Residential 33 <1 29 2 Lake Agriculture, Wetlands 33 <1 29 3 Pond Agriculture 33 <1 29 4 Creek Wetlands 33 <1 29 5 Creek Wetlands 33 <1 29 6 Stream Agriculture 32 3 20 7 Creek Agriculture 26 <1 49 8 C reek Agriculture 32 12 25 9 Creek Agriculture 32 12 25 10 Stream Agriculture 32 12 25 11 River Wetlands, Agriculture 32 12 25 12 Creek Agriculture 45 13 10 13 Canal Agriculture 55 NA* 21 14 Canal Agriculture 55 NA 21 15 Canal Agriculture 55 NA 2 1 16 Canal Agriculture 55 NA 21 17 Canal Agriculture 55 NA 21 18 Canal Agriculture 55 NA 21 *NA Not Available
115 Table 4 2 Correlation coefficients (R 2 ) determined for Salmonella log MPN/100 ml enumerated monthly for 12 consecutive months from 18 Cen tral Florida surface water sites vs. each of the physical, chemical, and biological water characteristic recorded at time of samplings values greater than 0.4 have been bolded Site E. coli (log MPN /100 ml) Coliforms (log MPN /100 ml) APC a (log CFU /100m l) Water Temp (C) Air Temp (C) pH ORP (mV) T urbidity (FAU) C onducti vity ( S/ cm ) 1 0.108 0.009 0.001 0.000 0.158 0.003 0.007 0.220 0.021 2 0.000 0.059 0.119 0.205 0.235 0.006 0.008 0.011 0.210 3 0.082 0.079 0.178 0.101 0.043 0.123 0.146 0.141 0.033 4 0.149 0.087 0.144 0.099 0.123 0.006 0.027 0.691 0.405 5 0.034 0.003 0.003 0.069 0.103 0.068 0.080 0.152 0.036 6 0.097 0.254 0.344 0.014 0.127 0.177 0.165 0.195 0.039 7 0.066 0.039 0.023 0.026 0.076 0.009 0.271 0.042 0.265 8 0.263 0.003 0.007 0.031 0.0 19 0.052 0.240 0.078 0.194 9 0.150 0.001 0.092 0.065 0.016 0.021 0.424 0.055 0.315 10 0.278 0.037 0.038 0.037 0.035 0.071 0.555 0.359 0.269 11 0.606 0.003 0.099 0.005 0.039 0.272 0.070 0.584 0.355 12 0.230 0.079 0.020 0.119 0.136 0.204 0.346 0.019 0.02 4 13 0.372 0.019 0.000 0.153 0.303 0.010 0.005 0.109 0.018 14 0.017 0.322 0.354 0.408 0.120 0.191 0.154 0.023 0.030 15 0.016 0.016 0.377 0.138 0.430 0.075 0.022 0.010 0.065 16 0.131 0.168 0.069 0.148 0.041 0.013 0.037 0.024 0.054 17 0.107 0.004 0.000 0.022 0.087 0.185 0.102 0.152 0.450 18 0.678 0.247 0.002 0.001 0.020 0.000 0.101 0.237 0.053 Overall 0.078 0.000 0.004 0.000 0.006 0.076 0.070 0.010 0.015 a Aerobic Plate Count
116 Table 4 3 Correlation coefficients (R 2 ) with p values determi ned between each of the physical, chemical, and biological water characteristics for all sampling sites and months combined Water Characteristic Air temp (C) pH ORP (mV) Turbidity (FAU) Conductivity Coliforms a E. coli a Aerobic plate count b Solar Radiation (W/m 2 ) Salmonella a Water temp (C) 0.825 c 0.208 0.0117 0.144 0.1300 0.0165 0.0125 0.0330 0.1210 (0.0000) d (0.0031) (0.8800) (0.0418) (0.0707) (0.8170) (0.0432) (0.8620) (0.6410) (0.0946) Air temp (C) 0.0127 0.0254 0.1410 0.225 0. 0307 0.0744 0.0297 0.0696 0.0538 (0.8580) (0.7450) (0.0473) (0.0017) (0.6680) (0.2990) (0.6790) (0.3250) (0.4590) pH 0.1330 0.0686 0.423 0.0479 0.0141 0.0398 0.0101 0.0506 (0.0867) (0.3370) (0.0000) (0.5040) (0.8450) (0.5790) (0.8860) (0.4880) ORP (mv) 0.0209 0.0566 0.0799 0.0499 0.0151 0.1010 0.0578 (0.7880) (0.4720) (0.3020) (0.5200) (0.8450) (0.1870) (0.4640) Turbidity (FAU) 0.0878 0.0297 0.0336 0.0819 0.0824 (0.2210) (0.0151) (0.6780) (0.6390) (0.2460) (0.2570) 0.0592 0.0568 0.1210 (0.4130) (0.0000) (0.4320) (0.0001) (0.0994) Coliforms 0.364 0.212 0.0670 0.0032 (0.0000) (0.0029) (0.3450) (0.9650) E. coli 0.0386 0.151 0.248 (0.5920) (0.0330) (0.0005) Aerobic plate count 0.0132 0.0314 (0.8520) (0.6680) Solar Radi ation (W/m 2 ) 0.165 (0.0216) a E. coli coliforms, and Salmonella reported in log MPN/100 ml b Aerobic plate count reported in log CFU/100 ml c coefficients displayed in bold are statistically significant(p < 0.05) d p values are displayed in parenthesis
117 Figure 4 1 Populations of Salmonella enumerated via MPN enrichments in LB, TT broth, and isolation on XLT4, ( E. coli ) enumerated via MPN method using Colisure, all reported in log MPN/100 ml (left axis), and aerobic plate counts ( ), enumerated on TSA and reported in log CFU/100 ml (right axis), as determined for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period log MPN/100 ml log CFU/100 ml Time (month/year)
118 Figure 4 2 temperature probe at the time of sampling for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period Temp ) Time (month/year)
119 Figure 4 3 pH left axis) and ORP right axis ) measured at the time of sampling for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period Time (month/year) pH ORP
120 Figure 4 4 right axis) measured at the time of sampling for each of eighteen Central Florida sites sampled monthly for a continuous twelve month period Time (month/year) Turbidity (FAU) Conductivity (S/cm)
121 CHAPTER 5 IDENTIFICATION AND CHARACTERIZATION OF SALMONELLA FROM CENTRAL FLORIDA SURFACE WATERS Salmonella is o ne of the major causes of enteric disease in humans, causing over 40,000 laboratory confirmed cases per yea r in the US (Scallan et al. 2011). Salmonella enterica subspecies enterica comprise more than 1,400 non typhoidal serovars, while only 30 serovars c ause the majority o f salmonellosis cases in the US (Kim et al. 2006). The identification of Salmonella serovars is important for source tracking and outbreak investigations. The genus Salmonella is divided into two species, Salmonella enterica and Salmone lla bongori Salmonella enterica is further divided into six subspecies. The most common serotyping method for Salmonella isolates involves the serological discrimination of surface polysaccharides (O), flagella (H), and capsular (Vi) antigenic propertie s (Kim et al. 2006). This method requires more than 150 O and H antisera for the characterization of each serovar (Jean Gilles Beaubrun et al. 2012). The use of a multiplex PCR method, previously reported by Kim et al. (2006) could be employed to greatly reduce the time and labor required to serotype 30 of the most clinically relevant serovars (Kim et al. 2006). This PCR method consists of two five plex PCR reactions, and is based on six genetic loci form S Typhimurium and four loci from S Typhi. Anothe r method used for the typing of Salmonella is pulsed field gel electrophoresis (PFGE). This method relies on a macrorestriction of genomic DNA, c health laboratories for comparing Salmonella isolates; it has
122 geographically diverse outbreaks (Ribot et al. 2006). The standardized method has allowed for PFGE profiles fro m numerous laboratories to be comparable, and thus much data now exists for purposes of comparison. Some disadvantages to PFGE are the time required to results (five days), technical expertise required, and costly equipment. Previously, Salmonella have be en isolated from Central Florida surface waters (Chapter 4). Environmental sampling for Salmonella has become an important research area in an effort to further understand potential sources of contamination of produce. Irrigation water has been suggested as a potential source of pre harvest contamination (Miles et al. 2009, Erickson et al. 2010, Ijabadeniyi et al. 2011). Environmental sampling studies could benefit from having a more time and cost efficient method for typing Salmonella isolates. Salmonel la serovars isolated from surface waters are frequently not those serovars included in the 30 most clinic ally relevant serovars (Kim et al. 2006, Haley et al. 2009, Patchanee et al. 2010, Rajabi et al. 2011, Strawn et al. 2013). Therefore, the present stu dy aimed to determine if the multiplex PCR method, previously reported by Kim et al. (2006), could be used to i) verify that all Salmonella isolates from a single water sample are identical; and ii) predict the serogroup or serovar of environmental Salmone lla isolates. Additionally, Salmonella isolates from Central Florida surface waters were characterized by PFGE, serotyping, and antimicrobial resistance profiles. Materials and Methods Water Sample Collection and Salmonella Enrichment Salmonella isolation from Central Florida surface waters is described in Chapter 4. Briefly, 10 L of surface water was collected from eighteen sites in Central Florida sampled for twelve consecutive months starting August 2010. Sites were grouped
123 generally into three region s based on location (1 6; 7 12; and 13 18). All sites were in rural agricultural areas, away from animal agriculture and included two lakes, one pond, six creeks, two stream, one river, and six canals. The land use and catchment watershed for each site i s detailed in Table 4 1 of Chapter 4. Salmonella isolates were obtained using a modified FDA BAM Salmonella enrichment method (Chapter 4). Up to 36 Salmonella isolates were obtained from each sample; Salmonella was genetically confirmed by PCR of the inv A and ori C genes; a total of 562 Salmonella were isolated (Chapter 4) and stored in tryptic soy broth (TSB; Difoc, Becton Dickinson, Sparks, MD, USA) supplemented with 15% glycerol at 80C. Multiplex PCR Analysis Frozen ( 80C) isolates were transferred to tryptic soy agar (TSA; Difco, Becton Dickinson, Sparks, MD, USA) and incubated for 242 h at 352C. One colony was transferred to TSB and incubated 242 h at 352C; DNA was extracted using the MoBio UltraClean DNA (MoBio, Carlsbad, CA, USA). A previou sly described multiplex PCR reaction (Kim et al. 2006) for Salmonella serotype analysis was performed on DNA from all isolates. The PCR assay uses the amplification of specific gene products that results in the production of a serovar specific banding patt ern for each of 30 top clinically important serovars of S. enterica subspecies enterica (Kim et al. 2006). This method consists of two five plex PCR reactions. The primers for each of the reactions are as described by Kim et al. (2006) with the following exceptions: the PCR reagents were all obtained from the Fisher exACTGene Complete PCR kit (Fisher Scientific). The bands are labeled as described by Jean Gilles Beaubrun (2012).
124 Gel electrophoresis, using a 2.5% agarose gel, was run for 60 or 120 min for the STM and STY PCR products, respectively, at 0.8 mV cm 2 in 1X TBE buffer. PCR UV light. Amplicon sizes were determined by comparison to Hyperladder V molecular size marker. Serogrouping Each isolate was grown on TSA for 242 h at 352C. One isolated colony was Salmonella O Poly Antiserum (Becton Dickinson, Sparks, MD, USA) was applied and autoagglutina tion was reported after 1 min of rocking the glass slide. All isolates were serogrouped in order to determine if isolates that grouped together based on multiplex PCR banding patterns also belonged to the same serogroup. Serotyping Of the 562 Salmonella i solates, 217 were selected to represent the multiplex PCR/serogroup groupings within one sampling time/location and sent to the National Veterinary Laboratory Service (NVSL; USDA, Ames, Iowa, USA) for serotyping. Pulsed field gel electrophoresis DNA finger printing of Salmonella isolates (n=217, the same isolates that were serotyped) was conducted using pulsed field gel electrophoresis (PFGE) after DNA macrorestriction using Xba I restriction enzyme. PFGE was performed according to the standard laboratory pr otocol recommended by the PulseNet program, Centers for Disease Control and Prevention (Ribot et al. 2006). The PulseNet universal standard strain Salmonella enterica Braenderup H9812 was used as the reference marker.
125 Gel images were transferred to Bionum erics software version 6.6 (Applied Maths, NV, Austin, TX, USA) for cluster analysis. Cluster analysis was performed using the unweighted pair group method with arithmetic averages with 2.0% band position tolerances and 1.5% optimization values. Similari ty coefficients were obtained within Bionumerics by calculating Dice coefficients. PFGE banding patterns with a similarity index >80% were grouped with the same genotypic cluster. Antimicrobial Susceptibility Testing Salmonella isolates were assayed for s usceptibility to 15 antibiotics by the calibrated dichotomous sensitivity method (Bell et al. 2004). Results were interpreted according to test standards. Antimicrobial susceptibility was determined according to test standards. Isolates were grown overn ight on TSA. One colony was transferred to 1 ml of 0.85% sterile saline, 330 (Oxoid, Ogdensburg, NY, USA). Sensitest agar plates were allowed to dry 30 min before application of antibiotic discs (BBL Sensi Disc, BD Diagnostic Systems, Sparks, MD). Five antibiotic discs were appli ed per plate. Zones of inhibition were measured after incubation at 35C for 18 h. The following antibiotics were tested: amikacin (An), amoxicillin clavulanic acid (Amc), ampicillin (Am), cefoxitin (Fox), ceftriaxone (Cro), chloramphenicol (C), ciproflo xacin (Cip), gentamicin (G), imipenem (Imp), kanamycin (K), nalidixic acid (Na), streptomycin (S), sulfamethoxazole trimethoprim (STx), and tetracycline (Te). Results Water Sample Collection A total of 202 water samples were collected and analyzed over the twelve month sampling period (Chapter 4). Salmonella isolates were obtained from 165 (81.7%) of
126 the 202 water samples analyzed; a total of 562 Salmonella isolates were obtained; up to 36 isolates were obtained from one sample. Multiplex PCR The 562 isola tes were analyzed using the multiplex PCR. Salmonella from the same sample (site and time) having the same multiplex PCR pattern, were assumed to be potentially the same; Salmonella having different multiplex PCR patterns were assumed to be different. Sa lmonella were identified as having all the same multiplex PCR pattern in 68 samples (33.7%). In 52 (25.7%) of the samples two patterns were identified; in 21 (10.4%) samples three patterns were identified amongst the isolates. The remaining 24 (11.9%) sa mples had greater than three patterns identified. Serogrouping To confirm Salmonella that had all the same multiplex PCR pattern from one sample were potentially the same, serogrouping was performed (Table 5 1). When multiple isolates from the same sample had matching multiplex patterns (n=12), all isolates from the same sample matched in serogroup. In all cases where Salmonella from the same sample had matching multiplex patterns, the serogroup groupings matched the same groupings; different multiplex PC R patterns did not necessarily mean different serogroup. In each sample, all groupings based on the multiplex PCR pattern matched groupings based on serogroup. As reported by Kim et al. (2006), the multiplex PCR is able to determine the serovar of the 30 most common clinical Salmonella serovars. In the current study, 494 isolates (87.9%) did not match the thirty most clinically common isolates, according to Kim et al. (2006), serovar patterns previously reported. Of the isolates matching reported multipl ex patterns, suggested serovars included: Bovismorbificans (22),
127 Braenderup (9), Chester (1), Derby (1), Hadar (3), Meunchen (4), Newport (2), Ohio (9), Oranienburg (4), Poona (3), Saintpaul (3), Thompson (5), Typhimurium (1). Of these suggested serovars, only 31 (46.3%) serogrouped accordingly. Following serotyping by NVSL, none of the serovars matched the proposed serovar based on the multiplex PCR pattern. Serotyping One representative Salmonella isolate was sent from each multiplex PCR grouping per sa mple for serotyping. Of the starting 562 Salmonella isolates analyzed using the multiplex PCR method, 217 were sent to the NVSL for serotyping. A total of 31 Salmonella isolates were lost during freezer storage and were unable to be analyzed further. Th e frequency of isolation for each serovar obtained is shown in Table 5 2. Serovars with corresponding multiplex PCR patterns are shown in Tables 5 3. A total of 32 different serovars were isolated during the 12 month sampling from all sampling locations. The most frequently isolated serovars for all samplings, were S Muenchen (11.5%), S Rubislaw (9.5%), S Anatum (8.8%), S Gaminara (8.8%), S IV_50:z4,z23: (6.8%). In region one (sites 1 6), serovars Rubislaw, Newport, Muenchen, Gaminara, and IV_50:z4 ,z23: were the most frequently isolated; of the 32 serovars isolated, 16 were seen in region one. In region 2 (sites 7 12), serovars Muenchen, Anatum, Florida, Gaminara, IV_50:z4,z23: and Paratyphi_B_var.L tartrate+ were the most frequently isolated; of the 32 serovars isolated, 20 were seen in region 2 In region 3 (sites 13 18), serovars Rubislaw, Muenchen, Gaminara, and Braenderup were the most frequently isolated; of the 32 serovars isolated, 16 were seen in region 3
128 Considering isolates from all sampling sites, the greatest number of serovars, either nine or ten, were isolated August to December. All other months had either five or six different serovars isolated. No one serovar was isolated in all months of study from any of the sampling sites The number of different serovars isolated from any individual sampling site over the 12 month study ranged from two to 11 different serovars for the duration of the study. The average number of serovars from one site over 12 months was six serovars. Th e fewest serovars isolated from one site was site 14, a canal in a 55% agriculture land usage watershed where only two different serovars were isolated over the 12 months. The largest number of different serovars were isolated from a single site, were fro m site 10 (stream in a watershed containing 32% agriculture for land usage), site 4 (creek in a watershed containing 33% agriculture), and site 6 (stream in a watershed containing 32% agriculture); 11, 10, and 9 different serovars, respectively, were isola ted from these sites over 12 months. For the three sampling sites with the largest number of serovars isolated: S Anatum, S Gaminara, S Muenchen, and S Newport were common to these three sites. PFGE Analysis Isolates sent for serotyping were further a nalyzed using PFGE Xba I macrorestriction banding patterns. Three separate dendrograms were generated, one for each of the geographical groupings of sampling sites, (Figures 5 1 to 5 3). The dendrogram generated for sampling sites 1 6, included 62 isolate s and resulted in 10 genotypic clusters with a dice coefficient index cut off point of 80%. The dendrogram generated for sampling sites 7 12, included 85 isolates and resulted in 17 genotypic clusters with a dice coefficient index cut off point of 80%. T he dendrogram generated
129 for sampling sites 13 18, included 71 isolates and resulted in 13 genotypic clusters with a dice coefficient index cut off point of 80%. Antimicrobial Susceptibility Testing Of the 248 Salmonella isolates tested for antimicrobial su sceptibility, 24 isolates were susceptible to all antimicrobials. Forty one Salmonella isolates were resistant to streptomycin. Fifty Salmonella isolates were resistant to both streptomycin and kanamycin (data not shown). Eighteen Salmonella isolates wer e resistant to more than five of the antibiotics tested. These antibiotic resistant profiles are shown in Table 5 4. The antibiotic resistance profiles for these Salmonella isolates typically (17/18) included resistance to amoxicillin clavulanic acid and /or cephalothin. Sixteen of the multi antibiotic resistant Salmonella isolates were resistant to ampicillin and/or kanamycin; 15 of these Salmonella isolates were also resistant to cefoxitin and/or streptomycin. None of the multi antibiotic resistant Sal monella isolates showed resistance to ceftriaxone or ciprofloxacin. Ten of the multi antibiotic resistant Salmonella isolates were isolated from sampling sites 1 6; the remaining eight were evenly distributed from sites 7 18. Of the 10 isolates from sampl ing sites located in sites 1 6, four came were isolated from site 4 and four isolates were isolated from site 5 Three S Braenderup (three out of seven total isolated) and three S Florida (three out of eight total isolated) isolates were multi anitmicro bial resistant. Two isolates of the following serovars were multi antimicrobial resistant: S Bareilly (2/2), S Gaminara (2/13), S Hartford (2/5), and S Muenchen (2/17).
130 Discussion Sa lmonella was isolated from 81.7% (165/202) of water samples using the described MPN methods, this percentage is similar to that reported by Haley et al. (2009) when reporting on Salmonella isolated from surface waters in the state of Georgia were 79% (57/72) positive samples. However, the population of S almonella may be gr eater in Central Florida, as a larger water sample volume (555 ml total) that was tested by Haley et al. (2009) compared to our current method that uses 33 ml total in the modified FDA BAM MPN method. This hypothesis is furthered by a study performed in N orth Florida where 96% of samples (at 1,000 L total in an MPN) were positive for S almonella (Rajabi et al. 2011). In North Carolina, a similarly crop agriculture dominated area, Salmonella prevalence was 50% (6/12; 25 ml sample volume) (Patchanee et al. 2 010). The same North Carolina study had samples from residential/industrial areas at 58.8% (20/34) Salmonella positive, and a forested area with 57.1% (16/28) Salmonella positive (Patchanee et al. 2010). From each of the Salmonella positive samples up to 36 Salmonella isolates were obtained, this gave a total of 562 isolates to further identify and characterize. Using the multiplex PCR to group Salmonella isolates from each sample as either different or potentially the same decreased the number of isolate s to be further characterized to 248. Thus decreasing the number of isolates to less than half of what would have otherwise been required. This grouping of the Salmonella isolates was able to be confirmed using serogrouping. Thus, it was demonstrated th at the multiplex PCR method described by Kim et al. (2006) can be used to group multiple environmental Salmonella isolates from the same sample as different or potentially the same. This can
131 greatly decrease the cost and amount of time required to screen numerous Salmonella isolates resulting from an environmental sampling. The large number (494) of Salmonella isolates from the current study that did not match any of the previously reported banding patterns suggests that many of these Salmonella isolates a re not included in the 30 most common clinical serovars (Kim et al. 2006). Additionally, this would support the suggestion that the multiplex PCR, as developed, has a limited ability to predict the serovar of some environmental isolates. This notion was further supported by our findings of only 31 Salmonella isolates serogrouping according to their predicted serovar, and furthermore by having none of the predicted serovar match the serovar reported by the NVLS. This is somewhat different from work perform ed by Jean Gilles Beaubrun et al. (2012) were they were able to predict the serovar of 76.2% (48/63) of the Salmonella isolates obtained from environmental samples taken from tomato farms in the Mid Atlantic region of Virginia (Jean Gilles Beaubrun et al. 2012). The 48 successfully serotyped contained S Newport (13 isolates), S Braenderup (10 isolates), and S Typhimurium (4 isolates), these serovars were also isolated in the current study but were unable to be serotyped via the multiplex PCR method. In addition, the Virginia isolates included: S Javiana, and S Thompson; these serovars were not identified in the current study, so the multiplex PCR results cannot be compared. The seemingly increased proficiency of the multiplex PCR reported for the Vir ginia isolates may be do to the few (n=5) different serovars identified during their study. In the current study we isolated 32 different serovars, none of which were successfully serotyped using the multiplex PCR method. Of the 32 serovars isolated in t he current study, seven had
132 banding patterns reported by Kim et al. (2006) as members of the thirty top clinical serovars, however, these Salmonella isolates did not produce the expected banding pattern for any of these seven serovars. Differences in bandi ng patterns, including missing expected genes or the amplification of serovar specific PCR products not typical for the serotypic banding pattern, have previously been reported for S Montevideo, S Newport, S Muenchen, S Hadar, S Infantis, S Paratyphi S Derby, and S Berta (Jean Gilles Beaubrun et al. 2012). These authors also report more than one banding pattern for S Seftenberg and S Tennessee (Jean Gilles Beaubrun et al. 2012). Thus there may be genotypic variations in serovars that were not p reviously expected. This makes the multiplex PCR unreliable as a method for consistently predicting the serovar of environmental Salmonella isolates, including those serovars with previously reported banding patterns. This may account for some of the unm atched banding patterns obtained for our Salmonella isolates. The number of different serovars reported in our study is higher than the number of different serovars typically reported during surveys for Salmonella in surface waters (Haley et al. 2009, Patc hanee et al. 2010, Rajabi et al. 2011). This may be due in part to the greater number of sampling sites included in the current study. A previous study in a crop agri cultural area of North Carolina had five different serovars coming from 12 samplings, se rovars as reported by the Ohio Department of Health, Columbus, Ohio (Patchanee et al. 2010). The previously discussed South Georgia study reported 13 different serovars being isolated, serovars as reported by Salmonella Reference Center, Philadelphia, PA (Haley et al. 2009). Following this trend, the North Florida study
133 reported eight different serovars, serovars as reported by Salmonella Reference Center, Philadelphia, PA (Rajabi et al. 2011). The current study isolated 32 different serovars from 18 sam pling sites, serovars as reported by NVSL. The greater diversity of serovars may, in part, be a reflection of the greater distance and numerous watersheds sampled in the current study as compared with the one watershed studied by the other authors in each of which only one watershed was studied. Surface waters surrounded with crop agriculture in North Carolina isolated S Newport as their most frequently (9/14) isolated serovar; S Brandenburg (n=1), S Gaminara (n=1), S Inverness (n=1), and S Miami (n=2 ) were also isolated in the North Carolina study (Patchanee et al. 2010). With the exception of S Brandenburg, these same serovars were also isolated in the current study. Salmonella isolates from South Georgia have previously been reported to contain a high percentage (80/197 isolates) of subspecies 3 (Arizonae), while Salmonella isolates from North Florida did not contain any isolates from subspecies 3 (Haley et al. 2009, Rajabi et al. 2011). The current study did isolate serovars from subspecies 3, bu t at a much lower frequency (6/217). Additionally, the South Georgia isolated seven serovars that were isolated in the current study, these included: S Muenchen (n=28), S Rubislaw (n=26), S Braenderup (n=12), S Saintpaul (n=9), S Bareilly (n=6), S G aminara (n=3), and S Anatum (n=1) (Haley et al. 2009). In a 1998 1999 study of the North Florida river, serovars isolated included: S Inverness (n=9), S Muenchen (n=3), S Rubislaw (n=6), S Braenderup (n=2), S Mont e video (n=1), S Newport (n=1), S Jo hannesburg (n=1), and S Cubana (n=1) (Rajabi et al. 2011). Of these isolates, S Inverness, S Muenchen, S Rubislaw, S
134 Braenderup, and S Newport were isolated in the current study as well; S Muenchen and S Rubislaw were isolated frequently in the cu rrent study, 17 and 14 times respectively. Salmonella Braenderup, S Gaminara, S Inverness, S Muenchen, S Newport, and S Rubislaw were all isolated in at least two of the other studies described as well as in the current study. This may be suggestive of these serovars having an increased environmental presence as compared with other Salmonella serovars. Of these frequently environmentally isolated serovars, only S Braenderup, S Muenchen, and S Newport are also on the CDC top twenty clinical isolate s list of 2009 (CDC 2009). For top twenty list of clinical serovars for the 2009 year. These seven were S Bareilly (2), S Braenderup (7), S Muenchen (17), S Newp ort (8), S Paratyphi_B_var.L tartrate+ (5), S Saintpaul (2), and S Typhimurium var 5 (2). An increased number of different Salmonella isolates, determined by PFGE, as compared to other surface water samplings, came from the current study. When using th e same analysis parameters, and the same restriction enzyme, the current study resulted in 10, 17, and 13 genotypic clusters in each of the three regions. The previously discussed North Carolina study resulted in seven genotypic clusters using the same pa rameters (Patchanee et al. 2010). The other surface water sampling studies previously used as a comparison for the current study did not report PFGE profiles. Most of our isolates were either susceptible to all antibiotics tested or where resistant to kan amyocin and/or streptomycin. This finding could be expected because
135 both these antimicrobials are secondary metabolites of Streptomyces spp., a common soil microbe; thus the finding of resistance in environmental isolates is logical (Turpin et al. 1992, E gan et al. 2001). The higher incidence of multi antimicrobial resistance in our Salmonella isolates may be indicative of Salmonella isolates that have come from animal agriculture, as this is more expected in environments influenced by animal agriculture. This can be seen in the study by Patchanee et al. (2010), when an increased amount of antimicrobial resistance was reported for isolates in a watershed known to be influenced by pig farms, as compared to isolates from locations influenced by crop agricul ture only. The continued persistence of Salmonella in Central Florida surface waters has been demonstrated. In our twelve month study we isolated a large number (32) of different serovars of S almonella This wide variety of isolates was also reflected in the DNA fingerprinting performed using PFGE. These Salmonella isolates could be grouped as different or potentially the same using a previously reported multiplex PCR (Kim et al. 2006). As evidenced by none of the serotyping results matching the serovar predicted by the multiplex PCR banding patterns, the multiplex PCR could not be used to accurately predict the serovar of the Salmonella isolates from Central Florida surface waters.
136 Table 5 1 Suggested serotypes and corresponding suggested serogroup b ased on multiplex PCR patterns for Salmonella isolated from Central Florida surface waters compared with actual serogroup determined by autoagglutination Suggested Serovar Suggested Serogroup No. Isolates Actual Serogroup % Matching A B C D E F G Bo vismorbificans B 22 13 5 2 23 Braenderup B 9 7 1 1 78 Chester A 1 1 0 Derby A 1 1 0 Hadar B 3 1 2 0 Mu e nchen B 4 2 2 50 Newport B 2 2 0 Ohio B 9 7 1 1 78 Oranienburg B 4 2 2 50 Poona B 3 1 2 0 Sai ntpaul A 3 2 1 67 Thompson B 5 5 100 Typhimurium A 1 1 100 Other 494 57 88 46 6 9 5 17
137 Table 5 2 Salmonella serovars isolated from Central Florida surface waters over 12 consecutive months compared with Salmonella se top 20 most clinically common serovars Frequency of Isolation CDC Top 20 a Serovar Region 1 Region 2 Region 3 Total Current Study Anatum 3 7 3 13 Baildon 1 1 19 Bareilly 2 2 11 Braenderup 3 4 7 Florida 1 5 2 8 G aminara 4 4 5 13 Georgia 3 3 Give 1 1 Hartford 2 3 5 Inverness 1 1 Ituri 3 3 Lexington 1 1 Litchfield 1 1 Miami 2 3 1 6 10 Muenchen 4 8 5 17 3 Newport 5 3 8 Norwich 2 1 3 15 Paratyphi_B_var._L tartrate+ 4 1 5 Rubislaw 5 3 6 14 9 Saintpaul 2 2 2 Typhimurium var 5 2 2 Weslaco 2 2 6,8:d: 1 1 III_16:z10:e,n,x,z15 1 3 4 III_44:z4,z32: 1 1 III_60:r:z 1 1 IV Rough O:z4,z23: 1 1 IV Rough O:z4,z24: 1 1 IV_40:z4,z24: 2 2 IV_50 :z4,z23: 4 4 2 10 Rough O:y:1,5 1 1 Rough_O:d:1,7 1 1 Untypable 3 2 2 7 a CDC top 20 clinically common serovars, numbers indicate overall placement in top 20
138 Table 5 3. Salmonella serovars isolated over 12 consecutive months from Central Fl orida surface waters with each different multiplex PCR banding pattern for all serovar s Sampling Month Sampling Sites Serovar Multiplex PCR Patterns August 3, 5 Bareilly No bands, H 7, 8, 9 Florida D, BEF, CEFGH, No bands 2, 6, 12 Gaminara BEGH, D, CE FG, CEFGH, H 11 III_44:z4,z32: I, EI 12 IV_50:z4,z23: CE, BCE 13, 16 Muenchen No bands, ABCDIJ, C 1, 2, 5, 10 Newport ABCEIJ, No bands, D, CDEIJ 16 Norwich CHIJ 18 Rubislaw C, No bands September 18 Braenderup BDEI 15, 16 Florida No bands 4, 7 Gaminara ADFG, BEGH 10 III_16:z10:e,n,x,z15 BC 18 Ituri A 6, 8 Muenchen ABEIJ, ACDEHI, ACEIJ 7 Norwich BCGHIJ 14 Rough_O:1,7 BCDEFG 14 Rough_O:d:1,7 BEFG 5, 6, 13 Rubislaw H, BGH, ABGH, ACEJ October 12 50:z4,z23: ABEFGHI 7 Anatum CH BCDEG 5 Florida No bands, H 10 Gaminara BCDEFGH 8 Hartford BCDEIJ 8 Miami EGHI, CDEGHI 1 Muenchen BI, BCDEIJ 4 Newport CIJ 7 Saintpaul BCDJ 2, 4 Typhimurium var 5 No bands, CI
139 Table 5 3. Continued Sampling Month Sampling Sites Serova r Multiplex PCR Patterns November 4, 7 Anatum BCDE, BDEFGH 13, 15, 16, 17 Gaminara ADFGH, F, BCDEGH, No bands, C, J 3, 4 Georgia BEI, D 11 III_17:z10:e,n,x,z15 ABCD, ABCDE 10 Inverness ABEJ 12 IV Rough O:z4,z23: ACDEI 8 IV Rough O:z4,z24: CD EI, CEI 10, 11 IV_40:z4,z24: BCDE, ACDE 9 Miami ACDEGHI 1, 6 Rubislaw ABCDEGH December 6, 7 Anatum H, BCE 17 Baildon ABDH, ABDEH, BCH, BC, BFG, B, BCDEIJ, ACG, BE, AE 8, 10 Hartford BEJ 6 Miami BJ 10 Newport BCEJ 7 Paratyphi_B_var._L.tart rate+ BEH, BEHJ 10 Rough O:y:1,5 BEFG, BEG 7, 9, 13, 14, 15 Rubislaw BH, BCD, BC, C, D, No bands 1, 11 Untypable BEFGH, CHJ January 3, 5 IV_50:z4,z23: EJ, J, BEI, No bands, BJ 8 6,8:d: ABEI 7, 18 Anatum ABCE, ACE, BCDEIJ, CE, BCDE, BCE 7 Gam inara BEGH 9, 11, 18 Muenchen G, No bands, AE, BE 17 Litchfield BCE
140 Table 5 3. Continued Sampling Month Sampling Sites Serovar Multiplex PCR Patterns February 4, 13 Anatum ACE, No bands 4 Give AEG 18 IV_50:z4,z23: E, AE 18 Miami EFJ 10 Newport BDEIJ 8 Untypable BF March 18 Anatum BCE, No bands 3, 4 Hartford B, BDEIJ, DEIJ 17, 18 Ituri No bands, E 12 IV_50:z4,z23: No bands, AE, AEI 12 Paratyphi_B_var._L. tartrate+ AEGHI, BEGI 2 Untypable J April 10 Anatum No bands, ACE 5 Georgia C 6, 8 III_16:z10:e,n,x,z15 BCGH, CI, AB 2, 6 IV_50:z4.z23: CI, BCI, BCDEFI 3, 6, 10 Miami C, BCFI, EGHJ 4 Muenchen BCE, ABEJ 4, 6 Weslaco AC, ACEH May 15 III_60:r:z BGH 17, 18 Muenchen BDEGI, BDEJ, BEJ, I 6 Newport ABCDIJ, GI, BCDE IJ 13 Paratyphi_B_var.L tartrate+ CH 18 Rubislaw BH, BGH 1, 17 Untypable ABC, BD
141 Table 5 3. Continued Sampling Month Sampling Sites Serovar Multiplex PCR Patterns June 10, 12 Anatum ACE, No bands, A, ABCDE, ABCI 10 IV_50:z4,z23: ACDE 18 Lexington BC 9 Paratyphi_B_var._L.tartrate+ AEHI, ABEHI, DHI 12 Saintpaul ACDEGHI, ABCDEJ, CHI 13 Untypable CDH July 2, 4, 5, 18 Braenderup No bands, BEI, BEHI, BEGI, I BEF 6, 18 Gaminara ABFG, FG, F, BEFG, No bands 5, 7, 9, 10, 11 Muenchen B EI, ABCEJ, ACEJ, AC, CFIJ 7 Norwich BCGHIJ 6, 7 Rubislaw BEG, ABGH, ABCGH
142 Table 5 4 Antibiotic resistance profiles of Salmonella isolates from Central Florida surface waters, only isolates resistant to greater than five of the antibiotics screene d are listed Serovar Site Month Antimicrobial Resistance Profile Anatum 13 February AmcAmFoxCfSTe Bareilly 4 August AmcAmFoxCfKSSTxTe Bareilly 5 August AmcAmFoxCfCKNaSSTxTe Braenderup 2 July AmcAmCfImpKS Braenderup 4 July AmcAmFoxKNaSSTx Braenderup 5 July AmcAmFoxCfCKNaStxTe Florida 4 August AmcAmFoxCfImpKS Florida 5 October AmcAmFoxCfKS Florida 9 August AmcAmFoxCfKS Gaminara 15 November AmcAmFoxCfCKNaSSTxTe Gaminara 16 November AmcAmFoxCfKS Hartford 3 March AmcFoxCfKNaSTe Hartford 4 March AmcA mFoxCfCKNaSTe Muenchen 5 July AmcCfCKNaSTxTe Muenchen 11 July AmcAmCfCKNaSTxTe Rough_O:d:1,7 14 September AmcAmFoxCfGmS Rubislaw 7 July AmFoxCfKSSTxTe Saintpaul 12 June AmcAmFoxCfKS
143 Figure 5 1. Dendrogram of representing PFGE Xba I fingerprinting of Salmonella isolates recovered from sampling site 1 6, PFGE, pulsed field gel electrophoresis
144 Figure 5 2. Dendrogram representing PFGE Xba I fingerprinting of Salmonella isolates recovered from sampling sites 7 12. PFGE, pulsed field gel electrophor esis
145 Figure 5 3. Dendrogram representing PFGE Xba I fingerprinting of Salmonella isolates recovered from sampling sites 13 18. PFGE. Pulsed field gel electrophoresis.
146 CHAPTER 6 FATE OF SALMONELLA SPP. IN FLORIDA SURFACE WATERS UNDER LABORATORY CONDITIO NS Salmonella is an important agent of gastroenteritis in humans. Salmonella can enter surface water from a variety of sources including: untreated or partially treated wastewaters, agricultural runoff, wastes from domestic and wild animals (Dechesne and Soyeux 2007, Haley e t al. 2009, Bolton et al. 2012) The long term survival of these bacteria in surface water resources constitutes a public health concern through direct consumption, indirect consumption (via irrigation of edible crops), and contact dur ing recreation (Greene et al. 2008, Ijabadeniyi et al. 2011, Ge et al. 2012). Indirect consumption of Salmonella through the use of contaminated irrigation water on edible horticultural crops has previously caused two known outbreaks: one associated with t omatoes and one associated with Serrano peppers. In both outbreaks, the outbreak serovar was isolated from irrigation ponds on the implicated farms (Greene et al. 2008, Behravesh et al. 2011). The causative agent in the tomato borne outbreak was Salmonel la Newport (Greene et al. 2008). A previous tomato borne outbreak, from the same implicated farm, occurred several years prior and involved the same serovar. It is possible that the S Newport survived this whole time in the environment, and possibly in the surface water. A review by Jacobsen et al. (2012) suggests that irrigation water may be one of the major sources of contamination of fresh produce (Jacobsen and Bech 2012). Currently, only one regulatory microbial water quality standard exits for wate r used on edible horticultural crops, specifi cally Florida grown tomatoes (F DACS 2007). Irrigation water is required to meet recreational water quality standards outlined in 40 CFR 131.41(c), that is water samples must not have E. coli populations exceedi ng a
147 geometric mean (n=5) of 126/100 ml and a single sample must not have an E. coli population exceeding 235/100 ml. However, there is a lack of scientific evidence exists to support the concept that E. coli populations can indicate overall water microbi al safety. Minimal work has focused on the duration of Salmonella survival in various types of environmental waters. Many previous works have also focused on marine or estuary waters, results from which cannot be entirely relied upon to predict survival i n fresh water. Most studies have focused on clinically important serovars, such as S Typhimurium, rather than other less clinically common serovars (Rhodes and Kator 1988, Maki and Hicks 2002, Chandran and Hatha 2005). The use solely of a local water so urce, sterile or non sterile, incorporates uncontrolled variables into each set of experiments that in turn make the studies difficult to compare. Chemical, physical, and biological water characteristics may affect Salmonella survival. The use of a standa rdized, and representative surface water should be included to control chemical and physical water quality characteristics between microcosm based experiments. There exists a previously described representative surface water developed by the US Environmen tal Protection Agency (EPA) for the testing of camping water purifiers (US EPA 1987). This water is chemically defined and can be reproducibly prepared in laboratory using few materials. Use of such a standard medium would make discrete experiments compa rable to each other. Relying solely on sterile water can overlook some potentially important biological variables that would be present in real world situations. In non sterile waters Salmonella would be exposed to competition from both autochthonous and
148 anthropogenic bacteria, predation from protozoa, and attack by bacterio phages. These variables need to be included in microcosm based survival studies in order to more accurately represent the real world. The relationship between the persistence of Salmon ella spp., either individually or as a cocktail, and biological, physicochemical water quality characteristics, representative of surface waters, over six months was explored. The suitability of EPA worst case water as a standard medium for investigating Salmonella persistence was also evaluated. Materials and Methods Waters and Microcosm Conditions Four waters were used to set up the microcosms: sterile deionized water, sterile surface water, non sterile surface water, and EPA worst case water. EPA worst case water is a defined medium originally developed for testing point of use microbiological water purifiers (US EPA 1987). The EPA worst case water was prepared using deionized water, 1500 mg/l dissolved sea salts (Sigma Aldrich, St Louis, MO), 10 mg/l humic acid (Sigma Aldrich), and fine test dust (Powder Technology Inc., Burnsville, MN) to 30 FAU. The final pH of the EPA worst case water was 5.3. Surface water was collected in a 10 L sterile autoclavable carboy (Nalgene, Rochester, NY, USA) fitted wit h 4 kg of lead weights all attached to a rope used to lower the carboy into the water source. The carboy was lowered to the water from a manmade structure that protrudes ca. 3 m from the shore; water was collected from approx. 20 cm below the surface. Th e water source was a fresh water pond in Polk County, Florida that is surrounded with non animal agricultural lands. All waters listed as sterile were sterilized, in 1 L portions, in an autoclave at 121 C, 15 psi, for 15
149 minutes. Water was held at 4 C fo r no more than 24 h, before sterilization and/or being dispensed into the microcosms. Measured as described in Chapter 3, the non sterile surface water had a turbidity of 9 FAU, and a pH of 7.69; the sterile surface water had a turbidity of 9 FAU, and a p H of 7.37. To evaluate the effect of protozoal grazing, non sterile surface water was supplemented with cycloheximide (LeJeune et al. 2001) (20 mg/1 ml; MP Biomedicals Inc, Solon, OH, USA) 24 h prior to the start of the experiment to destroy protozan. Mi crocosms, 100 ml, were prepared and stored in 120 ml sterile polypropylene cups (Oxford, Mansfield, MA, USA). A separate cup was prepared for each sampling time to remove the chance of contamination during successive samplings. Three storage temperatures were selected: 15C, 21C, and 28C to represent Central Florida average temperatures during winter, spring/fall, and summer (see Chapter 3). For each temperature, water was stored at the appropriate temperature for 24 h prior to the beginning of the exp eriment to ensure the water was at the correct temperature before inoculation. The microcosms were held stagnant, protected from light, at the appropriate constant temperature for up to six months. At each sampling time one microcosm was removed and samp led; unused portions were disposed of. Bacterial Strains and Culture Preparation Salmonella serovars used were initially isolated from surface water in Central Florida (see Chapter 4); serovars used included: Antatum (MDD 48), Newport (MDD 59), and Gaminar a (MDD 50); available from the culture collection of Dr. MichelleD Danyluk. Serovars were stored at 80 C in Tryptic Soy Broth (TSB; Difco, Becton Dickinson, Sparks, MD, USA), supplemented with 15% glycerol, until time of use. Serovars were streaked to Xylose Lysine Tergitol 4 agar (XLT4; Difco, Becton
150 Dickinson) and incubated 18 h at 352C. One colony was picked and inoculated into TSB and incubated 18 h at 352C; 10 l was transferred to fresh TSB and incubated 18 h at 352C. Cells were washed twi ce in 0.1% peptone water (Difco, Becton Dickinson). Prior to use in experiments cultures were diluted in 0.1% peptone water to ca. 8 log CFU/ml prior to being inoculated into the pre dispensed water. Serovars were either used separately or prepared as an equal part cocktail. A separate inoculum was prepared for each repetition. Three repetitions were performed for the single serovar microcosms; six repetitions were performed for the cocktail microcosms. Microbiological Analysis One microcosm from each w ater/temperature/inoculum combination was enumerated at 0, 6, 12, and 24 h, and on days 2, 3, 5, and 7 days. Deionized water, sterile surface water, non sterile surface water, and EPA worst case water microcosms were additionally enumerated at 14 days, an d at monthly increments thereafter until six months. Salmonella populations were enumerated by removing a one milliliter aliquot from the microcosm and preparing serial dilutions in 0.1% peptone water. Appropriate dilutions (0.1 ml) were spread plated, in duplicate, onto XLT4. Plates were incubated 242 h at 352C and all typical Salmonella colonies were counted by hand. If Salmonella populations were predicted to be below 2.3 log CFU/ml, one ml was spread plated onto four XLT4 (250 l per plate) to inc rease the limit of detection; these plates were incubated and counted in the manner described above. Once Salmonella populations were predicted to be below 1 CFU/ml, a three by four tube MPN was used to enumerate cells; dilutions were as follows: 10 ml wat er in 10 ml double strength lactose broth, 1 ml water in 9 ml single strength lactose broth, and
151 0.1 ml water in 9 ml single strength lactose broth. The remaining water sample (60 ml) was pre enriched in 30 ml triple strength lactose broth in the original polypropylene cup. MPN tubes were incubated for 242 h at 352C. One ml aliquot of the pre enrichment were transferred to tetrathionate broth (TT broth; Difco, Becton Dickinson); TT broth streaked onto XLT4 and incubated for 242 h at 352C. Following incubation, colonies displaying typical Salmonella phenotypes were confirmed biochemically on lysine iron agar slants (LIA; Difco, Becton Dickinson), and triple sugar iron agar slants (TSI; Difco, Becton Dickinson). MPN values were calculated using the FDA BAM table (US FDA 1998). Coliform and E. coli populations were also enumerated from non sterile waters. A 1 ml aliquot was taken from the selected microcosm and serial dilutions were prep ared in 0.1% peptone water. Appropriate dilutions were spread plated, in duplica te, onto Chrom agar ECC (DRG, Mountainside, NJ, USA). Plates were incubated 242 h at 352C and all typical coliform (mauve) and E. coli (blue) colonies were counted by hand. If populations were predicted to be below 2.3 log CFU/ml, one ml was spread plated onto four C hrom agar ECC (250 l per plate) to increase the limit of detection to 1 CFU/ml; these plates were incubated and counted in the manner described above. Distribut ion of strain survival in cocktails Once Salmonella populations in cocktail microcosms reached the limit of detection for spread plating, 100 colonies from each temperature/water combination were transferred to t ryptic s oy a gar (TSA; Difco, Becton Dickinso n). Plates were incubated 242 h at 352C. Salmonella serovars in the cocktail were distinguished by serogroup. The serogroup of each of the 100 colonies was determined by scraping a
152 One drop of Salmonella O Antiserum Poly A, B, or C (Difco, Becton Dickinson) was added and the slide rocked back and forth for 1 min; autoagglutination was noted. Statistical Analysis Salmonella and coliform enumeration data from each of the replicates on each sampling time from all microco sms were log transformed and plotted against time. JMP Pro 9.0 (SAS, Cary, NC, USA) was used to determine statistical significances using P value was set at 0.05. Results Cocktail in Each Water and Temperature Com bination Over Time The fate of Salmonell a in various surface waters was evaluated over a six month period. Salmonella was inoculated either as a cocktail or individual serovars, into four types of surface waters and holding at three different incubation temperatures (Figures 6 1 to 6 4). At time zero, no significant differences (P>0.05) between Salmonella populations existed between water types or temperatures used in this experiment. At all but one time point (discussed below), there was no difference in the behavior between the individual serovars and the cocktail ; only the cocktail data will be discussed in detail. Unless otherwise stated, the data below are in regards to Salmonella populations from the cocktail inoculated microcosms. DI water When h eld at 15C, Salmonella populations in DI water were not significantly different from each other for the first 14 days, with an average cocktail population of 7.60.2 log CFU/100 ml (Figure 6 1A). Following one month of storage at 15C, the Salmonella pop ulation significantly declined, by 0.9 log CFU/100 ml, compared to the previous 14 days; no significant change in Salmonella populations occurred between
153 one and six months in the 15C DI water. The average Salmonella population from one to six months was 6.90.5 log CFU/100 ml. At 21C Salmonella populations in DI water increased between six and 12 h from 7.50.3 log CFU/100 ml to 8.0 0.1 log CFU/100 ml; this was a slight, but statistically significant increase (Figure 6 1A). Salmonella populations then remained statistically the same until a decrease at four months of incubation; at four months Salmonella populations were 7.70.1 log CFU/100 ml and did not significantly change for the remaining time. The six month Salmonella population level was not si gnificantly different from that at time zero or 6 h of incubation. At 28C Salmonella populations in DI water remained constant until the one month enumeration with an average population of 8.50.2 log CFU/100 ml (Figure 6 1A). At two months of incubation Salmonella populations had decreased to 8.10.04 log CFU/100 ml; Salmonella populations decreased again at three months to 7.20.5 log CFU/100 ml, then remained unchanged until the final six month enumeration. Sterile surface water Sterile surface water a t 15C had Salmonella populations remain stable for the beginning two months; Salmonella populations were at 7.80.3 log CFU/100 ml (Figure 6 1B). Salmonella populations decreased significantly between two and three months, to 7.60.9 log CFU/100 ml, then remained stable through the duration of the experiment. At 21C Salmonella populations remained stable for a longer duration, populations remained at an average of 8.10.4 log CFU/100 ml and did not decrease significantly until five months (Figure 6 1B). Salmonella populations continued to remain stable at 7.50.5 log CFU/100 ml between five and six months. At 28C Salmonella populations in sterile surface water were more variable over time, yet
154 remained within the range of 8.70.1 log CFU/100 ml and 7. 30.1 log CFU/100 ml throughout the six month incubation (Figure 6 1B). The final six month Salmonella population in 28C sterile surface water was 7.00.4 log CFU/100 ml. Non sterile surface water Non sterile surface water at 15C Salmonella populations remained constant between time zero and one day, averaging 7.80.1 log CFU/100 ml (Figure 6 1C). Salmonella populations then continuously decreased at each sampling time, requiring MPN enrichment at three months and thereafter. At six months, Salmonella populations in non sterile surface water at 15C, had reached 2.70.8 log MPN/100 ml. Salmonella populations behaved similarly at 21C and 28C, reaching 3.80.7 log MPN/100 ml and 0.9 log MPN/100 ml, respectively, at six months (Figure 6 1C). Coliform po pulations started at 7.60.8 log CFU/100 ml at all three temperatures of incubation. Coliform populations increased significantly (P<0.05) at 12 h to 8.30.1 log CFU/100 ml, after this time point the coliform populations continually decreased. At four mo nths of incubation coliforms were undetectable in all three incubation temperatures (limit of detection 1 log CFU/100 ml). No significant difference existed in coliform populations between the different temperatures of storage at any time of enumeration f or the six month experiment (data not shown). Escherichia coli was not detected at any point during the experiment. EPA worst case water Salmonella populations in EPA worst case water at 15C had no significant difference from time zero until day three wi th an average of 7.60.3 log CFU/100 ml (Figure 6 1D). At day five an increase to 8.00.3 log CFU/100 ml had occurred. Salmonella populations did not change again until four months. At four months
155 Salmonella populations had decreased, significantly, to 7.70.1 log CFU/100 ml; Salmonella populations in EPA worst case water at 15C then did not significantly change through to the final six month enumeration. Salmonella populations in EPA worst case water at 21C increased from 7.90.1 log CFU/100 ml to 8.3 0.1 log CFU/100 ml, at 12 h of incubation (Figure 6 1D). Salmonella populations then remained statistically unchanged through three months. At four months Salmonella populations had decreased, to 8.00.0 log CFU/100 ml, and then remained at this level f or the duration of the experiment. When EPA worst case water was incubated at 28C, Salmonella populations had increased by the first enumeration time point (6 h), from 7.90.1 log CFU/100 ml to 8.70.3 log CFU/100 ml (Figure 6 1D). Salmonella populations then remained statistically not different until a decrease at four months occurred; Salmonella populations had deceased to 7.11.4 log CFU/100 ml. Salmonella populations were not significantly different for the remained of the experiment. Fate of Individ ual Serovars Compared to Cocktail Over Time At time zero, no significant differences (P<0.05) between Salmonella populations existed between the cocktail or individual serovars used in this experiment (Figures 6 1. 6 2, 6 3, 6 4). The only point where a s ignificant difference existed between serovars was at five months, where S. Anatum populations were significantly higher than S. Gaminara populations (Figures 6 2 and 6 3); S. Newport and the cocktail populations were not significantly different from each other or S. Anatum or S. Gaminara populations. This significant difference was not seen at the six month enumeration.
156 Serovar Survival Within Cocktail Inoculated Waters In order to evaluate if all Salmonella serovars behaved the same in the cocktail inocu lated microcosms, or if there was variability amongst the serovars, 100 colonies from the final enumeration by spread plating were serogrouped. The percentage of each serovar in the final Salmonella population was determined (Figure 6 5). The last enumer ations by plate count for non sterile surface waters were completed at two months, therefore serogro uping for these samples was performed at this time point. All other water types were able to be enumerated by plating at six months; serogrouping on these samples was performed at this point. The percentage of each serogroup is shown in Figure 6 5. At two months non sterile surface water at 28C, S. Anatum, S Newport and S Gaminara represented 48% (6.5), 49% (10), and 3% (3.8) of colonies recovered, r espectively (Figure 6 5A); at 21C S Anatum, S Newport, and S Gaminara represented 71% (13), 25% (11), and 4 (3) of colonies recovered, respectively (Figure 6 5B); at 15C S Anatum, S Newport, and S Gaminara represented 32% (26), 42% (20), and 2 5% (20) of colonies recovered, respectively (Figure 6 5C). At six months, Salmonella serovars Anatum or Gaminara dominated the recovered colonies in DI water at 28C (Figure 6 5A); three repetitions had Salmonella populations that were predominately (100 89%) S Gaminara, and three repetitions had predominately S Anatum (100 89%). Only one co lony from DI water at 28C serogrouped as S Newport. In DI water at 21C, Salmonella populations were predominately (74%20) S Gaminara (Figure 6 5B); the remaini ng colonies were S. Anatum (11% 10) and S Newport (15% 14). In DI water at 15C S Gaminara was predominate in all but one repetition (Figure 6 5C), in which S. Anatum dominated
157 (84%). With the one S Anatum dominated microcosm removed, the recovered colonies from DI water at 15C were 15% (12) S Anatum, 16% (15) S Newport, and 50% (34) S Gaminara. At six months in sterile surface water at 28C, Salmonella populations were 70% (14) S Anatum, 15% (10) S Newport, and 16% (14) S Gaminara (Figu re 6 5A). At 21C sterile surface water, Salmonella populations consisted of 61% (13) S Anatum, 32% (14) S Newport, and 5% (7) S Gaminara (Figure 6 5B). At 15C sterile surface water Salmonella populations consisted of 29% (32) S Anatum, 20% (25 ) S Newport, and 50% (32) S Gaminara. At six months EPA worst case water at 28C had Salmonella populations predominated by S Anatum (80% 16); the other recovered colonies were 14% (16) S Newport and 7% (7) S Gaminara (Figure 6 5A). At 21C, EPA worst case water Salmonella populations were predominately (63% 19) S Anatum, the other recovered colonies were S Newport (27% 9), and S Gaminara (11% 14; Figure 6 5B). At 15C EPA worst case water Salmonella populations consisted of 57% (10) S An atum, 23% (10) S Newport, and 19% (12) S Gaminara colonies (Figure 6 5C). Effect of Water Type and Temperature on the Fate of Salmonella DI water Over six months, Salmonella populations in DI water, EPA worst case water, and sterile surface water were either not significantly different or significantly different at only one time point, then returning to not significantly different at the subsequent sampling times (Figure 6 1). At day three, Salmonella populations in non sterile surface water were signi ficantly lower than those in DI water (Figure 6 1A and 6 1C); Salmonella
158 populations in non sterile surface water remained significantly lower than those in DI water for the remainder of the study. In DI water, Salmonella populations at 15C and 21C were significantly lower than those samples incubated at 28C (Figure 6 1A). Over day ones through seven, the lowest Salmonella populations in DI water were at 15C; after day seven no difference existed between 15C and 21C. At 12 h, and there after, DI wat er at 28C had statistically lower Salmonella populations than the other water types at the same temperature. Sterile surface water Salmonella populations in sterile surface water were not significantly different from Salmonella populations in DI water or EPA worst case water for the first five days (Figure 6 1B). At day three, Salmonella populations in sterile surface water were significantly higher than those in non sterile surface water. From day seven to three months, Salmonella populations in EPA wor st case water and sterile surface water were not significantly different from each other, but were higher than Salmonella populations in either DI water or non sterile surface water. After four months, sterile surface water Salmonella populations were not significantly different from EPA worst case water or DI water Salmonella populations. Salmonella populations in sterile surface water at 15C and 21C were significantly lower than those incubated at 28C (Figure 6 1B). From 12 h to seven days, Salmonell a populations were lowest at 15C for sterile surface water; after seven days there was no significant difference in Salmonella populations in sterile surface water between 15C and 21C.
159 Non sterile surface water Salmonella populations in non sterile surf ace water were not significantly different from those in DI water or sterile surface water up to 12 h (Figure 6 1); Salmonella populations in non sterile surface water were significantly higher than those in EPA worst case water at 6 h. At day three of in cubation, and there after, the non sterile surface water had significantly lower populations of Salmonella compared to the other water types. From 12 h to day two, Salmonella populations were lowest at 15C incubation for non sterile surface water (Figure 6 1C). At day three and beyond, the lowest Salmonella populations occurred in non sterile surface water at 28C or 21C, followed by non sterile water at 15C. EPA worst case water EPA worst case water had the lowest populations of Salmonella of all the water types, at 6 h and 12 h (Figure 6 1D). Between 24 h and seven days, EPA worst case water Salmonella populations were not significantly different from those in DI water or sterile surface water. By day three, Salmonella populations in EPA worst case water were significantly higher than those in non sterile surface water. At day seven, EPA worst case water and sterile surface water had Salmonella populations not significantly different from each other, but were higher than Salmonella populations in ei ther DI water or non sterile surface water; this same pattern existed to, and including, three months. After three months, EPA worst case water Salmonella populations were not significantly different from sterile surface water or DI water. Salmonella popu lations at 15C and 21C were significantly lower than those samples incubated at 28C (Figure 6 1D). At 12 h through to, and including, seven
160 days, Salmonella populations were lowest at 15C incubation for EPA worst case water. At 14 days and beyond, no significant difference existed between Salmonella populations at 15C and 21C. Effect of Eukaryotes in Non Sterile Surface Water At time zero there was no difference between Salmonella populations in sterile surface water, non sterile surface water, or n on sterile surface water with cycloheximide added, at any of the temperatures used in this experiment, 15C, 21C, or 28C. Figure 6 6 displays the survival of Salmonella in sterile surface water, non sterile surface water, and non sterile surface water w ith cycloheximide added over seven days. At 6 h all water samples held at 28C had Salmonella populations statistically higher than those at the other two temperatures. Non sterile surface water with cycloheximide had Salmonella populations of 7.20.2 lo g CFU/100 ml at 28C, 6.70.2 log CFU/100 ml at 21C, and 6.30.1 log CFU/100 ml at 15C. At 12 h the waters held at 21 or 28C were not significantly different; microcosms held at 15C had statistically lower (P<0.05) Salmonella populations. Non sterile surface water with cycloheximide had Salmonella populations of 7.10.1 log CFU/100 ml at 28C, 6.90.1 log CFU/100 ml at 21C, and 6.30.2 log CFU/100 ml at 15C. Differences based on water types did not develop until one day of incubation, at this time s terile surface water had statistically lower Salmonella populations at 21C and 15C; no significant difference existed for waters incubated at 28C. At day one, sterile surface water had Salmonella populations of 7.60.1 log CFU/100 ml at 28C, 6.00.2 l og CFU/100 ml at 21C, and 6.1 0.3 log CFU/100 ml at 15C; non sterile surface water with cycloheximide had Salmonella populations of 7.50.1 log CFU/100 ml at 28C, 7.40.1 log CFU/100 ml at 21C, and 7.00.1 log CFU/100 ml at 15C; non sterile
161 surface w ater had Salmonella populations of 7.10.3 log CFU/100 ml at 28C, 7.30.3 log CFU/100 ml at 21C, and 6.80.1 log CFU/100 ml in 15C. The same differences based on water type and temperature continued to exist for two and three days of incubation. At fiv e days of incubation, non sterile surface water microcosms had statistically lower populations of Salmonella ; at this time the sterile surface water and the non sterile surface water with cyclohexmide were not statistically different. At five and seven da ys incubation microcosms held at 28C had lower Salmonella populations than microcosms at 21 or 15C, which were not significantly different. These microcosms were held for a total of seven days. By the seventh day sterile surface water at the three incub ation temperatures had the statistically highest Salmonella populations; sterile surface water had Salmonella populations of 8.20.2 log CFU/100 ml at 28C, 7.70.7 log CFU/100 ml at 21C, and 7.21.0 log CFU/100 ml at 15C. These were followed by non ste rile surface water with cycloheximide at 15C with Salmonella populations of 6.50.2 log CFU/100 ml, and non sterile surface water at 15C and 21C with Salmonella populations of 6.10.4 log CFU/100 ml and 5.70.6 log CFU/100 ml, respectively. The lowest Salmonella populations were from the non sterile surface water with cycloheximide at 28C and 21C with Salmonella populations of 4.50.5 log CFU/100 ml and 5.00.2 log CFU/100 ml, and non sterile surface water at 28C with Salmonella populations of 5.20. 6 log CFU/100 ml. Coliforms in non sterile water with or without cycloheximide at all three temperatures were not different at time zero; coliform populations started at 6.40.1 log CFU/100 ml. No E. coli was detected at time zero, or at any time during t his
162 experiment. By 12 h of incubation at 15C coliform populations decreased to 6.30.1 log CFU/100 ml; while those held at 21 or 28C had an increase in coliform populations, averaging 7.00.1 log CFU/100 ml. After 24 h there was no difference in colifo rm populations for any of the variables attempted, coliform populations averaged 6.60.3 log CFU/100 ml. Coliform populations remained not significantly different from each other throughout all the variables until day five. At day five non sterile surfac e water with cycloheximide incubated at 15C was significantly lower, 4.80.3 log CFU/100 ml, than non sterile surface water with cycloheximide incubated at 28C, which had a coliform population of 4.70.0 log CFU/100 ml; all other variable combinations we re not different from each other. The same difference existed at seven days (the final enumeration of this experiment). Overtime, the coliform populations in both non sterile water types incubated at 21C significantly decreased (P<0.05) between day zero and day seven; by day seven coliform populations in non sterile surface water were 5.20.2 log CFU/100 ml and coliform populations in non sterile surface water with cycloheximide were 5.30.5 log CFU/100 ml. Coliform populations in both non sterile water types incubated at 28C, had a significant decrease (P<0.05) by day 2; by day two, coliform populations in non sterile surface water were 5.80.3 log CFU/100 ml and coliform populations in non sterile surface water with cycloheximide were 6.00.1 log CFU/1 00 ml. Coliform populations in both non sterile water types at 15C experienced no significant difference over the seven day incubation; by day seven, coliform populations in non sterile surface water were 5.70.5 log CFU/100 ml and coliform populations i n non sterile surface water with cycloheximide were 6.50.6 log CFU/100 ml.
163 Discussion The three microcosm temperatures were selected to represent average winter, spring/fall, and summer temperatures in Central Florida. Previous work has elucidated that i n Central Florida, air temperature can be used to predict surface water temperatures (Chapter 4). Ten year average air temperatures at 60 cm from three local weather recording stations maintained by the Florida Automated Weather Network (http://fawn.ifas. ufl.edu) were used to determine the three microcosm temperatures used in this study. January, February and March temperatures were used to represent winter; April, May, October, November, and December to represent spring/fall; June, July, August and Septe mber were used to determine summer. These month groupings gave the smallest standard deviations for the average temperatures (data not shown). The beginning enumeration time points, up to five days, 15C incubation temperature supported the lowest Salmon ella populations. During winter months in Florida air temperatures can go below freezing, when this occurs it is common practice to spray crops with water in order to prevent freezing of the crops. This practice is known as frost protection. As 15C is representative of the winter months in Central Florida, this can be interpreted as beneficial for a decrease in risk of using surface waters for frost protection of edible horticultural crops. Although this practice should still be limited because Salmone lla has been demonstrated to survive for a long duration of time in such waters. EPA worst case water should be considered a suitable standard medium for investigating Salmonella persistence in surface waters. The inclusion of a standard medium in microco sm based experiments would aid in allowing multiple studies to be compared more fairly than when the test medium is not standardized. EPA worst case
164 water was evidenced here to support both individual serovars and serovars within a cocktail at the same le vel or better than sterile surface water at all three temperatures evaluated. EPA worst case water was not different from other water types until three days at which point the non sterile water had lower Salmonella populations. An interesting addition to EPA worst case water as a standard for water based microcosm work would be the development a standard background biota (protozoa, bacteria, and bacterio phages). However, much work would need to be performed and evaluated to determine a standard biota. At five days of incubation EPA worst case water was not different from the other sterile waters used in this experiment. After seven days EPA worst case water was supporting a higher Salmonella population than all the other waters. Growth of Salmonella Ten nessee has previously been reported in diffusio n chamber based experiments performed using estuary waters at or above 18C, in both with and without natural biota present (Rhodes and Kator 1988). Additionally, Salmonella (not serotyped), originally isolat ed from sewage, has been observed to increase from 6.7 log CFU/ml to 8.1 log CFU/ml in sterile creak water within 10 days, and continue to survive up until 56 days which marked the end of the experiment (Fish and Pettibone 1995). This would suggest that E PA worst case water is, in fact, representing the worst case in terms of Salmonella survival. At four months EPA worst case water incubated at 21C had the highest Salmonella populations, of all the variable combinations, this is disadvantageous in terms of risk assessment as 21C was the average temperature established to represent spring and fall water temperature; spring and fall are the seasons in which most Florida produce is grown.
165 For the six month experiment, in the non sterile surface waters colif orm populations decreased, from 7.50.9 log CFU/100 ml at time zero which was not significantly different from time zero Salmonella populations, to below the detection limit of 1 CFU/ml at four months; this occurred prior to the Salmonella populations decr easing to this level. The coliform population becoming undetectable before Salmonella populations would suggest that coliforms do not make for good indicator organisms for Salmonella in surface waters representative of Central Florida surface waters. Thi s is especially true when the contamination event is expected to have occurred months prior to sampling being performed In this event, testing for total coliforms would classify the water as meeting potable water standards and the water permitted for use for foliar application near time of harvest for tomato plants. The serovars chosen for these experiments were S. Anatum, S Newport, and S Gaminara. Each of these isolates were originally isolated from Central Florida surface waters (Chapter 5). Each s erovar classifies into a separate serogroup so that colonies identified at final enumeration can be distinguished based on serogroup. Salmonella. Anatum was selected because of its prevalence of detection in the 12 month Central Florida surface water surv ey (Chapter 5). Salmonella Newport was selected as it has been associated with multiple tomato borne outbreaks where S. Newport was isolated from an irrigation pond on the implicated tomato farm (Greene et al. 2008). Salmonella Gaminara was selected beca use, although it has caused few produce borne outbreaks, it was associated with a fresh orange juice outbreak borne outbreak in the state of Florida.
166 No significant difference existed between the serovars of Salmonella used in the inoculum or in the enumer ation of Salmonella populations originally from single serovar inoculums compared to the cocktail prepared from equal volumes of each serovar. When colonies from the final six month enumeration of Salmonella populations in sterile surface water or EPA wor st case water, held at either 21 or 28C, S. Anatum was the dominate serovar. A more even distribution was obtained when these waters were held at 15C. The predominance of S Anatum is not entirely surprising because of its high level of isolation durin g the Central Florida surface water survey (Chapter 5). The predominance, in some cases >80% of the colonies, of one serovar may suggest that surface waters, under set conditions, are able to support a certain level of Salmonella regardless of the serova r, but when mixed serovars are present one will predominate the population. Colonies isolated from DI water, at all three temperatures, were predominantly S Gaminara, with very poor S Newport survival. This may suggest that S Newport does not compete w ell with other Salmonella present, but is equally capable of survival in no nutrient water (as is evidenced by the lack of significant difference between single serovar microcosms). The survival of all the serovars from this experiment surviving up to, an d including, six months in the DI water is comparable to the a previous study were Salmonella Typhimurium DT 104 survived 54 days in 23C sterile well water, after which no more enumerations were attempted (Moore et al. 2003). Colonies isolated from non st erile surface water were serogrouped at two months, the last enumeration via spread plating. All serogrouping was performed on colonies from spread plating, and not the enrichment MPNs, as either the pre
167 enrichment or selective enrichment could conceivabl y select for one serovar over the others, and thus skew the serogrouping results. At two months at both 28 and 21C, microcosms had a low percentage of colonies serogroup as S Gaminara; S Anatum and S Newport percentages were not significantly differen t from each other. The difference in the trend of serovar recovery is likely due to one of two causes or a combination: bacteriophage present that are better able to attack S. Gaminara, and/or S Gaminara are grazed upon at a greater rate by protozoa. Th e phage hypothesis could be supported further by the lack of difference in Salmonella population levels for the first 24 h in non sterile water that was then followed by a continual decrease in population level over the remaining six months. The delay in population decrease could suggest the attachment and infection phase, prior to cell lysis by bacteriophage attack. We cannot suggest that if bacteriophage were present, that they were only capable of attacking S Gaminara, due to the lack of significant d ifferences among the Salmonella populations in the individual serovar microcosms. A cocktail of three bacteriophages, isolated from sewage, has been used to reduce Salmonella Kentucky, S Anatum, and S Zanzibar populations in treated wastewaters (Turki e t al. 2012). A significant decrease in Salmonella after 8 h in treated wastewater with the bacteriophage cocktail was achiev ed, however, incubation was at 30C or 37C. and a phage titer of 4 log PFU/ml (Turki et al. 2012). The higher incubation temperat ures may explain some of why the current study would find a significant decrease only after 24 h where the highest incubation temperature was 28C. In EPA worst case water bacteriophages have significantly reduced Salmonella populations at 6 h when incuba ted at 35C (McLaughlin et al. 2008); this again is a higher temperature than used in the current
168 bacteriophage lysis at a significantly slower rate (McLaughlin et al. 2008). T his could aid in explaining the slower and prolonged (six months) potential phage inactivation in the current study non sterile surface water. To further elucidate which biotic factors were playing major roles in determining Salmonella populations in non s terile surface water cycloheximide was added to the non sterile surface water to inhibit eukaryotic cells, thus inactivating protozoa. By day five of incubation all non sterile surface water microcosm had significantly lower Salmonella populations than di d the non sterile surface water with cycloheximide; sterile surface water had significantly higher Salmonella populations than either of the non sterile waters. This would suggest that protozoa grazing accounted for only part of the decrease in Salmonella populations in the non sterile surface water microcosms. Similar results were obtained by Chandran et al. (2011) when comparing S Paratyphi survival in non sterile and autoclaved lake water. With a starting concentration of 8 log CFU/ml in both water ty pes, a decrease to 1 log CFU/ml in non sterile lake water was obtained, while autoclaved lake water resulted in a 6 log CFU/ml Salmonella population after the same 27 day period (Chandran et al. 2011). Similarly, streptomycin resistant S Typhimurium in filter sterilized surface water at 20C for 41 days, while the same water used non sterile had S Typhimurium surviving 16 days; as determined by pour plating to bismuth sulfite agar (Maki and Hicks 2002). In the seven day experiment comparing non sterile surface water with and without cycloheximide coliform populations were not significantly different for any combination of water type and temperature at time zero. After 12 h a decrease in
169 coliform populations was noted in waters stored at 15C, while incr eases in coliform populations were noted for waters incubated at 21 or 28C. Both of these events suggest that coliforms may not be a suitable indicator organism for Salmonella in surface waters; at 15C a significant decrease was noted in coliform popula tions prior to a significant decrease in Salmonella populations, and at 21 and 28C an increase in coliform populations was observed. The increase in coliform populations is suggestive of growth of coliforms in the surface waters, and the ideal indicator organism should not be capable of growth in the medium to be evaluated. Growth of E. coli has previously been noted in sterile creak water microcosms 10 days into the incubation (Fish and Pettibone 1995). In conclusion, our results indicate that Salmonel la either individually or as a cocktail, are able to survive long term, up to six months, in various types of surface waters. The use of sterile waters in microcosms has been over estimating the survival duration of Salmonella in surface waters. EPA wor st case water is a suitable standard medium for studying the survival of Salmonella in surface waters. Furth er investigations should be performed for other human pathogens in water.
170 Figure 6 1 Survival of Salmonella cocktail in surface waters at ( ) 28C, ( ) 21C, and ( ) 15C ( m ean standard deviation, n=6). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water.
171 Figure 6 2. Survival of Salmonella Anatum in surface waters at ( ) 28C, ( ) 21C, and ( ) 15C ( m ean standard deviation, n=3). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water.
172 Figure 6 3. Survival of Salmonella Gaminara in surface waters at ( ) 28C, ( ) 21C, and ( ) 15C ( m e an standard deviation, n=3). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water.
173 Figure 6 4 Survival of Salmonella Newport in surface waters at ( ) 28C, ( ) 21C, and ( ) 15C ( m ean standard de viation, n=3). (A) DI water. (B) Sterile surface water. (C) Non sterile surface water. (D) EPA worst case water.
174 Figure 6 5. Composition of Salmonella cocktail in surface waters at final enumeration by spread plating as determined by serogrouping 100 colonies, Salmonella serovar Anatum (solid), Newport (crosshatched), and Gaminara (striped) ( m ean standard deviation, n=6). (A) 28C. (B) 21C. (C) 15C.
175 Figure 6 6. Survival of Salmonella cocktail in sterile surface water ( ), non sterile su rface water ( ), and non sterile surface water with cycloheximide ( ) ( m ean standard deviation, n=6). (A) 28C. (B) 21C. (C) 15C.
176 CHAPTER 7 CONCLUSIONS AND FUTURE WORK The work presented here establishes that Salmonella is prevalent in Central Flori da surface waters. First, a method to screen large volumes of surface water for Salmonella was established. Using this method, water samples from various surface water sources throughout Central Florida were screened for Salmonella To better understand the Salmonella serovars present in Central Florida surface waters, isolates from this study were identified and characterized using a variety of procedures. The method to screen 10 L of surface water for the presence of Salmonella uses a combination of ta ngential flow filtration, overnight enrichment, immuno magnetic separation, and qPCR detection. This method had a limit of detection of 1 CFU/L in 10 L of surface water within two days. Modified Moore swabs performed equally as well as tangential flow fi ltration for concentration. The use of normal flow filtration was eliminated due to irreversible fouling of the filter prior to the entire sample volume being passed through the filter. The development of this method is significant in that it is now poss ible to screen large volumes of surface water for the presence of Salmonella in a more time efficient manner. The two days time requirement is an important improvement in that risk management decisions can now be made based on more timely data. The monthl y use, for 12 consecutive months, of the developed method to evaluate surface water samples from 18 sites throughout Central Florida demonstrated the continued presence of Salmonella in Central Florida surface waters. All 202 water samples collected were Salmonella positive using the developed concentration and detection method. When making risk management decisions, especially in relation to
177 irrigation, crop protection, and other foliar applications, the likely presence of Salmonella in the surface water s should be taken into account. An alternative water source, such as ground water, could be considered. However, a substantial survey of Salmonella in Central Florida ground water should be conducted prior to this practice being put into place. Furtherm ore, the fate of Salmonella once being sprayed onto horticultural crops should be further investigated with attempts to replicate real world conditions as close as possible. After the 10 L water samples were determined to be Salmonella positive, Salmonella MPNs were carried out to provide quantitative data. This data is important for the accurate generation of risk assessments. Of the 202 water samples, 165 (81.7%) yielded MPN values above the lower limit of detection, and thus Salmonella isolates for fur ther analysis. This quantitative data should be used to generate risk assessments for the potential use of surface waters on or near edible horticultural Salmonella may preclude the use of Central Florida surf ace waters on edible horticultural crops. This presents a potential feasibility issue for producers in Central Florida, especially during times when frost protection is required. At time of sampling, air and water temperature, pH, ORP, turbidity and condu ctivity were recorded. In addition to Salmonella each water sample was also analyzed for aerobic plate count, and E. coli /coliform MPNs. None of the water characteristics had a strong correlation with Salmonella prevalence. In the Central Florida area, indicator or index organisms may not be able to be reliable indicators of potential Salmonella presence in surface waters. The present study does not support
178 the current practice of relying on E. coli as an indicator of acceptable microbial water quality as is currently required by the Tomato Good Agricultural Practices and suggested by the Leafy Greens Marketing Agreement (F DACS 2007, LGMA 2012). Hence, direct screening for the presence of Salmonella is recommended for surface waters that are to be u sed in the production of edible horticultural crops that are not expected to undergo a kill step. Risk assessments should be generated to determine which uses of, and at what times, surface waters represent substantial risk to consumer health when present as a source of contamination. A previously reported (Kim et al. 2006) multiplex PCR reaction was capable of determining if all Salmonella isolates from one surface water sample were likely the same or different; the same multiplex PCR was not able to accu rately predict the serogroup or serovar of the Salmonella isolated from surface water. This will reduce the time required to screen numerous Salmonella isolates from one surface water sample, but will not aid in the identification of which serovars are pr esent in the water sample. Conventional serotyping and Pulsed Field Gel Electrophoresis should continue to be used for serovar analysis. In microcosm based experiments, Salmonella Anatum, Gaminara, and Newport (all previously isolated from surface water) were able to survive six months in all water types and temperatures evaluated. This extended duration of survival needs to be considered when determining the likelihood of a particular body of water being contaminated with Salmonella With this evidence it should now be considered that once a body of water has been found to have Salmonella the Salmonella will be present for an extended duration. The effect of flowing water on the duration of
179 Salmonella survival should also be examined. If it is now kno w that Salmonella can survive in surface waters for up to six months, investigations into how far surface water could potentially travel in that time and the potential sources of contamination it would encounter along the way. EPA worst case water provided support Salmonella survival equally well, or better than sterile surface water. EPA worst case water should be used in further microcosm based studies on the survival of Salmonella so that future studies are m ore readily comparable. As demonstrated by the significantly lower survival of Salmonella in non sterile surface waters, compared to other sterile waters, the other microflora present has a significant effect on the survival of Salmonella in surface waters This work was not able to determine cause of this decreased survival. It is believed that the decreased survival was not due to protozoa grazing, as inactivating eukaryotic cells did not increase Salmonella survival in non sterile surface waters. Furt her work is required to determine the cause of decreased survival. Through better understanding of why the decrease in Salmonella populations occurred in non sterile surface waters could aid in determining potential methods of Salmonella inactivation with in surface waters. A variable not examined during this set of experiments is the potential effect of sediments, and their disturbance, to the survival of Salmonella in surface waters. Further experiments should be conducted to include a standardized sedim ent within the microcosms. The extended survival time, along with the continued presence, of Salmonella in Central Florida surface waters indicates that Salmonella may be a geonotic, rather than
180 zoonotic, organism. This is especially relevant in the South eastern US, where several studies have also reported high levels of Salmonella persistence (Haley et al. 2009, Patchanee et al. 2010, Rajabi et al. 2011). As no seasonal or temperature correlations were obtained in the present study, further work is requi red to better understand why Salmonella presence is greater in the Southeastern US than in other geographic areas which typically report ca. 10% Salmonella positive water samples (Gorski et al. 2011, Strawn et al. 2013, Thomas et al. in press). If Salmonel la is considered a geonotic organism, implying that it is understood to Salmonella should be re evaluated. As an alternative risk assessments nee d to be generated, using quantitative data, to determine best practices for fresh produce production. These risk assessments then need to be used to make appropriate risk management decisions to determine best practices for fresh produce food safety. Thi s is through the use of accurate data and, risk assessments that could best minimize the potential of more fresh produce borne salmonellosis outbreaks from occurring.
181 LIST OF REFERENCES Amel, D. and Amina, B. (2008) Resuscitation of seventeen year stress ed Salmonella Typhimurium. Int J Ocean Hyrdobiol 37 69 82. APHA (1992) Standard Methods for the Examination of Water and Wastewater, 18 th ed. American Public Health Association, Washington, DC. Bae, S. and Wuertz, S. (2012) Survival of host associated Bac teroidales cells and their relationship with Enterococcus spp., Campylobacter jejuni Salmonella enterica Serovar Typhimurium, and Adenovirus in freshwater microcosms as measured by propidium monoazide quantitative PCR. Appl Environ Microbiol 78 922 932. Balachandran, P., Cao, Y., Wong, L., Furtado, M.R., Petrauskene, O.V. and Tebbs, R.S. (2011) Evaluation of Applied Biosystems MicroSEQ (R) Real Time PCR System for detection of Salmonella spp. in Food. J AOAC Int 94 1106 1116. Barrett, T.J., Blake, P.A., Morris, G.K., Puhr, N.D., Bradford, H.B. and Wells, J.G. (1980) Use of Moore swabs for isolating Vibrio cholerae from sewage. J Clinical Microbiol 11 385 388. Bauwens, L., Vercammen, F., Bertrand, S., Collard, J. and De Ceuster, S. (2006) Isolation of S almonella from environmental samples collected in the reptile department of Antwerp Zoo using different selective methods. J Appl Microbiol 101 284 289. Behravesh, C.B., Mody, R.K., Jungk, J., Gaul, L., Redd, J.T., Chen, S., Cosgrove, S., Hedican, E., Swe at, D., Chavez Hauser, L., Snow, S.L., Hanson, H., Nguyen, T., Sodha, S.V., Boore, A.L., Russo, E., Mikoleit, M., Theobald, L., Gerner Smidt, P., Hoekstra, R.M., Angulo, F.J., Swerdlow, D.L., Tauxe, R.V., Griffin, P.M., Williams, I.T. (2011) 2008 outbreak of Salmonella Saintpaul infections associated with raw produce. N Engl J Med 364 918 927. Bell, S. M., B. J. Gatus, J. N. Pham, and D. L. Rafferty. 2004. Antibiotic susceptibility testing by the CDS method, 3rd ed. Avail able at: http://web.med.unsw.edu.au/pathology cds/cdsc2004.htm Accessed 25 August 2012 Berney, M., Weilenmann, H. ., Simonetti, A. and Egli, T. (2006) Efficacy of solar disinfection of Escherichia coli Shigella flexne ri Salmonella Typhimurium and Vibrio cholerae J Appl Microbiol 101 828 836. Bisha, B., Perez Mendez, A., Danyluk, M.D. and Goodridge, L.D. (2011) Evaluation of modified Moore swabs and continuous flow centrifugation for concentration of Salmonella and E scherichia coli O157:H7 from large volumes of water. J Food Prot 74 1934 1937.
182 Bolton, D.J., O'Neill, C.J. and Fanning, S. (2012) A Preliminary Study of Salmonella Verocytotoxigenic Escherichia coli/Escherichia coli O157 and Campylobacter on Four Mixed F arms. Zoonoses Public Health 59 217 228. Buchanan, R. (2000) Acquisition of microbiological data to enhance food safety. J Food Prot 63 832 838. Burton, G.A., Gunnison, D. and Lanza, G.R. (1987) Survival of pathogenic bacteria in various freshwater sedim ents. Appl Environ Microbiol 53 633 638. CDC (2008) Outbreak of Salmonella Serotype Saintpaul infections associated with multiple raw produce items United States, 2008. MMWR 57 October/01. Chandran, A. and Hatha, A.A.M. (2005) Relative survival of Esch erichia coli and Salmonella Typhimurium in a tropical estuary. Water Res 39 1397 1403. Chandran, A., Varghese, S., Kandeler, E., Thomas, A., Hatha, M. and Mazumder, A. (2011) An assessment of potential public health risk associated with the extended survi val of indicator and pathogenic bacteria in freshwater lake sediments. Int J Hyg Environ Health 214 258 264. Coleman, D., Nye, K., Chick, K. and Gagg, C. (1995) A comparison of immunomagnetic separation plus enrichment with conventional Salmonella culture in the examination of raw sausages. Lett Appl Microbiol 21 249 251. Corrente, M., Madio, A., Friedrich, K.G., Greco, G., Desario, C., Tagliabue, S., D'Incau, M., Campolo, M. and Buonavoglia, C. (2004) Isolation of Salmonella strains from reptile faeces a nd comparison of different cultural media. J Appl Microbiol 96 709 715. Craig, D.L., Fallowfield, H.J. and Cromar, N.J. (2004) Use of microcosms to determine persistence of Escherichia coli in recreational coastal water and sediment and validation with in situ measurements. J Appl Microbiol 96 922 930. D'Aoust, J.Y. and Maurer, J. (2007) Salmonella Species. In Food microbiology fundamentals and frontiers ed. Doyle, M.P. and Beuchat, L.R. pp. 187 236. Washington D.C.: ASM Press. Dechesne, M. and Soyeux, E. (2007) Assessment of source water pathogen contamination. J Water Health 5 39 50. Delibato, E., Fiore, A., Anniballi, F., Auricchio, B., Filetici, E., Orefice, L., Losio, M.N. and De Medici, D. (2011) Comparison between two standardized cultural methods and 24 hour Duplex SYBR Green Real Time PCR assay for Salmonella detection in meat samples. New Microbiologica 34 299 305. Doyle, M.P. and Erickson, M.C. (2012) Opportunities for mitigating pathogen contamination during on farm food production. Int J Food Microbiol 152 54 74.
183 Egan, S., Wiener, P., Kallifidas, D. and Wellington, E. (2001) Phylogeny of Streptomyces species and evidence for horizontal transfer of entire and partial antibiotic gene clusters. Antonie Van Leeuwenhoek Intl J of General and Molec ular Microbiol 79 127 133. El Mejri, S., El Bour, M., Boukef, I., Al Gallas, N., Mraouna, R., Got, P., Troussellier, M., Klena, J. and Boudabbous, A. (2012) Influence of marine water conditions on Salmonella enterica serovar Typhimurium survival. J Food S af 32 270 278. Erickson, M.C., Webb, C.C., Diaz Perez, J.C., Phatak, S.C., Silvoy, J.J., Davey, L., Payton, A.S., Liao, J., Ma, L. and Doyle, M.P. (2010) Surface and internalized Escherichia coli O157:H7 on field grown spinach and lettuce treated with spr ay contaminated irrigation water. J Food Prot 73 1023 1029. Escartin, E.F., Lozano, J.S., Garcia, O.R. and Cliver, D.O. (2002) Potential Salmonella transmission from ornamental fountains. J Enviro Health 65 9 12. Feng, F., Goto, D. and Yan, T. (2010) Eff ects of autochthonous microbial community on the die off of fecal indicators in tropical beach sand. FEMS Microbiol Ecol 74 214 225. Fernandez, H., Otth, L. and Wilson, M. (2003) Isolation of thermotolerant species of Campylobacter from river water using two collection methods. Arch med vet 35 95 97. Fish, J.T. and Pettibone, G.W. (1995) Influence of freshwater sediment on the survival of Escherichia coli and Salmonella sp. as measured by three methods of enumeration. Letters Appl Microbiol 20 277 281. F DACS, Florida Department of Agriculture and Consumer Services (2007) Tomato best practices manual. A guide to tomato good agricultural practices (T GAP) and tomato best management practices (T BMP). Available at http://www.freshfromflorida.com/fs/TomatoBestPractices.pdf (accessed 10 Jun 2012). Gaertner, J.P., Hahn, D., Rose, F.L. and Forstner, M.J. (2008) Detection of Salmonellae in different turtle species within a headwater spring ecosystem. J Wildlife Dis 44 519 526. Ge, C., Lee, C. and Lee, J. (2012) The impact of extreme weather events on Salmonella internalization in lettuce and green onion. Food Res Int 45 1118 1122. Gibson, K.E. and Schwab, K.J. (2011) Tangential flow ultraf iltration with integrated inhibition detection for recovery of surrogates and human pathogens from large volume source water and finished drinking water. Appl Environ Microbiol 77 385 391.
184 Gonzalez, R.A., Conn, K.E., Crosswell, J.R. and Noble, R.T. (2012) Application of empirical predictive modeling using conventional and alternative fecal indicator bacteria in eastern North Carolina waters. Water Res 46 5871 5882. Gorski, L. and Liang, A. (2010) Effect of enrichment medium on the real time detection of S almonella enterica from lettuce and tomato enrichment cultures. J Food Prot 73 1047 1056. Goyal, S.M., Gerba, C.P. and Melnick, J.L. (1977) Occurrence and distribution of bacterial indicators and pathogens in canal communities along the Texas coast. Appl Environ Microbiol 34 139 149. Greene, S.K., Daly, E.R., Talbot, E.A., Demma, L.J., Holzbauer, S., Patel, N.J., Hill, T.A., Walderhaug, M.O., Hoekstra, R.M., Lynch, M.F. and Painter, J.A. (2008) Recurrent multistate outbreak of Salmonella Newport associa ted with tomatoes from contaminated fields, 2005. Epidemiol Infect 136 157 165. Haley, B.J., Cole, D.J. and Lipp, E.K. (2009) Distribution, diversity, and seasonality of waterborne Salmonellae in a rural watershed. Appl Environ Microbiol 75 1248 1255. Hi ntz, L.D., Boyer, R.R., Ponder, M.A., Williams, R.C. and Rideout, S.L. (2010) Recovery of Salmonella enterica Newport introduced through irrigation water from tomato (Lycopersicum esculentum) fruit, roots, stems, and leaves. HortScience 45 675 678. Ijabad eniyi, O.A., Debusho, L.K., Vanderlinde, M. and Buys, E.M. (2011) Irrigation water as a potential preharvest source of bacterial contamination of vegetables. J Food Safety 31 452 461. Jacobsen, C.S. and Bech, T.B. (2012) Soil survival of Salmonella and tr ansfer to freshwater and fresh produce. Food Research International 45 557 566. Jardine, C., Reid Smith, R.J., Janecko, N., Allan, M. and McEwen, S.A. (2011) Salmonella in raccoons ( Procyon lotor ) in southern Ontario, Canada. J Wildlife Dis 47 344 351. J ean Gilles Beaubrun, J., Cheng, C., Chen, K., Ewing, L., Wang, H., Agpaoa, M.C., Huang, M.J., Dickey, E., Du, J.M., Williams Hill, D.M., Hamilton, B., Micallef, S.A., Rosenberg Goldstein, R.E., George, A., Joseph, S.W., Sapkota, A.R., Jacobson, A.P., Tall, B.D., Kothary, M.H., Dudley, K. and Hanes, D.E. (2012) The evaluation of a PCR based method for identification of Salmonella enterica serotypes from environmental samples and various food matrices. Food Microbiol 31 199 209. Jenkins, M.B., Endale, D.M. a nd Fisher, D.S. (2008) Most probable number methodology for quantifying dilute concentrations and fluxes of Salmonella in surface waters. J Appl Microbiol 104 1562 1568.
185 Johnston, M.A., Porter, D.E., Scott, G.I., Rhodes, W.E. and Webster, L.F. (2010) Isol ation of faecal coliform bacteria from the American alligator ( Alligator mississippiensis ). J Appl Microbiol 108 965 973. Jokinen, C., Edge, T.A., Ho, S., Koning, W., Laing, C., Mauro, W., Medeiros, D., Miller, J., Robertson, W., Taboada, E., Thomas, J.E. Topp, E., Ziebell, K. and Gannon, V.P.J. (2011) Molecular subtypes of Campylobacter spp., Salmonella enterica and Escherichia coli O157:H7 isolated from faecal and surface water samples in the Oldman River watershed, Alberta, Canada. Water Res 45 1247 1257. Kase, J.A., Borenstein, S., Blodgett, R.J. and Feng, P.C.H. (2012) Microbial quality of bagged baby spinach and romaine lettuce: Effects of top versus bottom sampling. J Food Prot 75 132 136. Kauffmann, F. and Fain, A. (1953) Three new Salmonella ty pes ( S Ituri, S Kasenyi, and S Niarembre) from the Belgian Congo (Occurrence of Salmonella types in ducks). Acta Path 32 513 515. Kaushik, R., Balasubramanian, R. and de la Cruz, A.A. (2012) Influence of air quality on the composition of microbial path ogens in fresh rainwater. Appl Environ Microbiol 78 2813 2818. Kim, S., Frye, J.G., Hu, J., Fedorka Cray, P.J., Gautom, R. and Boyle, D.S. (2006) Multiplex PCR based method for identification of common clinical serotypes of Salmonella enterica subsp Enter ica. J Clin Microbiol 44 3608 3615. Klontz, K.C., Klontz, J.C., Mody, R.K. and Hoekstra, R.M. (2010) Analysis of tomato and jalapeno and Serrano pepper imports into the United States from Mexico before and during a national outbreak of Salmonella serotype Saintpaul infections in 2008. J Food Prot 73 1967 1974. Leclerc, H., Mossel, D., Edberg, S. and Struijk, C. (2001) Advances in the bacteriology of the Coliform Group: Their suitability as markers of microbial water safety. Annu Rev Microbiol 55 201 234. Lee, A., McVey, J., Faustino, P., Lute, S., Sweeney, N., Pawar, V., Khan, M., Brorson, K. and Hussong, D. (2010) Use of Hydrogenophaga pseudoflava penetration to quantitatively assess the impact of filtration parameters for 0.2 micrometer pore size filter s. Appl Environ Microbiol 76 695 700. LeJeune, J.T., Besser, T.E., Merrill, N.L., Rice, D.H. and Hancock, D.D. (2001) Livestock drinking water microbiology and the factors influencing the quality of drinking water offered to cattle. J Dairy Sci 84 1856 1 862. Leon Velarde, C.G., Zosherafatein, L. and Odumeru, J.A. (2009) Application of an automated immunomagnetic separation enzyme immunoassay for the detection of Salmonella enterica subspecies enterica from poultry environmental swabs. J Microbiol Methods 79 13 17.
186 Leskinen, S.D., Brownell, M., Lim, D.V. and Harwood, V.J. (2010) Hollow fiber ultrafiltration and PCR detection of human associated genetic markers from various types of surface water in Florida. Appl Environ Microbiol 76 4116 4117. LGMA, Leafy Greens Marketing Agreement (2012) Commodity specific food safety guidelines for the production and harvest of lettuce and leafy greens. Available at http://www.caleafygreens.ca.gov/sites/default/files/01.20.12%20CALGMA%20GA Ps%20 %20metrics.pdf (accessed 13 Jun 2012). Madsen, M. (1994) Enumeration of Salmonellae in crocodile pond water by direct plate counts and by the MPN technique. Wat Res 28 203 5 2037. Mafu, A.A. and Sirois, S. (2009) Real time PCR as a tool for detection of pathogenic bacteria on contaminated food contact surfaces by using a single enrichment medium. J Food Prot 72 1310 1314. Maki, R.P. and Hicks, R.E. (2002) Salmonella Typhimu rium survival and viability is unaltered by suspended particles in freshwater. J Enviro Quality 31 1702 1709. Malorny, B., Hoorfar, J., Bunge, C. and Helmuth, R. (2003) Multicenter validation of the analytical accuracy of Salmonella PCR: Towards an intern ational standard. Appl Environ Microbiol 69 290 296. Martinez Urtaza, J., Saco, M., de Novoa, J., Perez Pineiro, P., Peiteado, J., Lozano Leon, A. and Garcia Martin, O. (2004) Influence of environmental factors and human activity on the presence of Salmon ella serovars in a marine environment. Appl Environ Microbiol 70 2089 2097. McEgan, R., Rodrigues, C.A.P., Sbodio, A., Suslow, T.V., Goodridge, L.D. and Danyluk, M.D. (2013) Detection of Salmonella spp. from large volumes of water by Modified Moore Swabs and Tangential Flow Filtration. Letters Appl Microbiol 56 88 94 Micallef, S.A., Goldstein, R.E.R., George, A., Kleinfelter, L., Boyer, M.S., McLaughlin, C.R., Estrin, A., Ewing, L., Beaubrun, J.J., Hanes, D.E., Kothary, M.H., Tall, B.D., Razeq, J.H., Jose ph, S.W. and Sapkota, A.R. (2012) Occurrence and antibiotic resistance of multiple Salmonella serotypes recovered from water, sediment and soil on mid Atlantic tomato farms. Environ Res 114 31 39. Miles, J.M., Sumner, S.S., Boyer, R.R., Williams, R.C., La timer, J.G. and McKinney, J.M. (2009) Internalization of Salmonella enterica serovar Montevideo into greenhouse tomato plants through contaminated irrigation water or seed stock. J Food Prot 72 849 852. Moore, B.C., Martinez, E., Gay, J.M. and Rice, D.H. (2003) Survival of Salmonella enterica in freshwater and sediments and transmission by the aquatic midge Chironomus tentans (Chironomidae: Diptera). Appl Environ Microbiol 69 4556 4560.
187 Mossel, D., Corry, J., Struijk, C. and Baird, R. (1995) Essentials of the microbiology of foods: a textbook for advanced studies. Chirchester, England: John Wiley and Sons. Mull, B. and Hill, V.R. (2009) Recovery and detection of Escherichia coli O157:H7 in surface water, using ultrafiltration and real time PCR. Appl Enviro n Microbiol 75 3593 3597. Notzon, A., Helmuth, R. and Bauer, J. (2006) Evaluation of an immunomagnetic separation real time PCR assay for the rapid detection of Salmonella in meat. J Food Prot 69 2896 2901. Ogden, I.D. (2001) The fate of Escherichia coli O157 in soil and its potential to contaminate drinking water. Int J Food Microbiol 66 111 117. Pachepsky, Y., Shelton, D.R., McLain, J.E.T., Patel, J., and Mandrell, R.E. (2011) Irrigation waters as a source of pathogenic microorganisms in produce: a rev iew. Advances in Agronomy 113 73 138. Patchanee, P., Molla, B., White, N., Line, D.E. and Gebreyes, W.A. (2010) Tracking Salmonella contamination in various watersheds and phenotypic and genotypic diversity. Foodborne Pathog Dis 7 1 1120. Payment, P. and Locas, A. (2011) Pathogens in w ater: Value and l imits of c orrelation with m icrobial i ndicators. Ground Water 49 4 11. Rajabi, M., Jones, M., Hubbard, M., Rodrick, G. and Wright, A.C. (2011) Distribution and genetic diversity of Salmonella enterica in the Upper Suwannee River. Int J Microbiol 2011 Rajala, R.L. and Heinonen Tanski, H. (1998) Survival and transfer of faecal indicator organisms of wastewater effluents in receiving lake waters. Wat Sci Tech 38 191 194. Ramalho, R., Afonso, A., Cunha, J., Tei xeira, P. and Gibbs, P.A. (2001) Survival characteristics of pathogens inoculated into bottled mineral water. Food Control 12 311 316. Ramos, R., Cerda Cuellar, M., Ramirez, F., Jover, L. and Ruiz, X. (2010) Influence of refuse sites on the prevalence of Campylobacter spp. and Salmonella serovars in seagulls. Appl Environ Microbiol 76 3052 3056. Rhodes, M.W. and Kator, H. (1988) Survival of Escherichia coli and Salmonella spp. in estuarine environments. Appl Environ Microbiol 54 2902 2907.
188 Ribot, E., Fai r, M., Gautom, R., Cameron, D., Hunter, S., Swaminathan, B. and Barrett, T. (2006) Standardization of pulsed field gel electrophoresis protocols for the subtyping of Escherichia coli O157:H7 Salmonella and Shigella for PulseNet. Foodborne Pathog Dis 3 59 67. Roda, A., Mirasoli, M., Roda, B., Bonvicini, F., Colliva, C. and Reschiglian, P. (2012) Recent developments in rapid multiplexed bioanalytical methods for foodborne pathogenic bacteria detection. Microchim Acta 178 7 28. Santiago Rodriguez, T.M., Tre mblay, R.L., Toledo Hernandez, C., Gonzalez Nieves, J.E., Ryu, H., Santo Domingo, J.W. and Toranzos, G.A. (2012) Microbial quality of tropical inland waters and effects of rainfall events. Appl Environ Microbiol 78 5160 5169. Scallan, E., Hoekstra, R.M., Angulo, F.J., Tauxe, R.V., Widdowson, M., Roy, S.L., Jones, J.L. and Griffin, P.M. (2011) Foodborne Illness Acquired in the United States Major Pathogens. Emerg Infect Dis 17 7 15. Schonenbrucher, V., Mallinson, E.T. and Bulte, M. (2008) A comparison of s tandard cultural methods for the detection of foodborne Salmonella species including three new chromogenic plating media. Int J Food Microbiol 123 61 66. Schonenbruecher, V., Mallinson, E.T. and Buelte, M. (2008) A comparison of standard cultural methods for the detection of foodborne Salmonella species including three new chromogenic plating media. Int J Food Microbiol 123 61 66. Sears, S.D., Ferreccio, C., Levine, M.M., Cordano, A.M., Monreal, J., Black, R.E., D'Ottone, K. and Rowe, B. (1984) The use of Moore swabs for isolation of Salmonella typhi from irrigation water in Santiago, Chile. J Infectious Dis 149 640 642. Serrano, E., Moreno, B., Solaun, M., Aurrekoetxea, J.J. and Ibarluzea, J. (1998) The influence of environmental factors on microbiologic al indicators of coastal water pollution. Wat Sci Tech 38 195 199. Setti, I., Rodriguez Castro, A., Pata, M.P., Cadarso Suarez, C., Yacoubi, B., Bensmael, L., Moukrim, A. and Martinez Urtaza, J. (2009) Characteristics and dynamics of Salmonella contaminat ion along the coast of Agadir, Morocco. Appl Environ Microbiol 75 7700 7709. Smith, S. and Schaffner, D.W. (2004) Indicator organisms in meat. In Encyclopedia of Meat Science ed. Anonymous pp. 773 779. London, UK: Elsevier Science. Smith, C.M. and Hill, V .R. (2009) Dead end hollow fiber ultrafiltration for recovery of diverse microbes from water. Appl Environ Microbiol 75 5284 5289. Spino, D.F. (1966) Elevated temperature technique for the isolation of Salmonella from streams. Appl Microbiol 14 591 596.
189 Stine, S., Song, I., Choi, C. and Gerba, C. (2005) Application of microbial risk assessment to the development of standards for enteric pathogens in water used to irrigate fresh produce. J Food Prot 68 913 918. Strawn, L.K., Fortes, E.D., Bihn, E.A., Nigh tingale, K.K., Grohn, Y.T., Worobo, R.W., Wiedmann, M. and Bergholz, P.W. (2013) Landscape and meteorlogical factors affecting prevalence of three food borne pathogens in fruit and vegetable farms. Appl Environ Microbiol 79 588 600. Taban, B.M. and Aytac, S.A. (2009) Application of magnetic immuno polymerase chain reaction assay for detection of Salmonella spp. in chicken meats. European Food Research and Technology 229 Thomas, J.L., Slawson, R.M. and Taylor, W.D. Salmonella serotype diversity and seasona lity in urban and rural streams. J Appl Microbiol In Press Thomason, B.M., Dodd, D.J. and Cherry, W.B. (1977) Increased recovery of Salmonellae from environmental samples enriched with buffered peptone water. Appl Environ Microbiol 34 270 273. Turki, Y., Ouzari, H., Mehri, I., Ben Ammar, A. and Hassen, A. (2012) Evaluation of a cocktail of three bacteriophages for the biocontrol of Salmonella of wastewater. Food Res Int 45 1099 1105. Turpin, P., Dhir, V., Maycroft, K., Rowlands, C. And Wellington, E. (19 92) The effect of Streptomyces species on the survival of Salmonella in soil. FEMS Microbiol Ecol 101 271 280 US Code of Federal Regulations (2012) Title 40 Protection of the Environment. Available at http://ecfr.gpoaccess.gov/cgi/t/text/text idx?c=ecfr&sid=4069b93e5cf3d85662170ccf95ca91b2&rgn=div8&view=text&nod e=40:22.214.171.124.126.96.36.199&idno=40 (acce ssed 13 July 2012). US EPA, United States Environmental Protection Agency (1987) Guide standard and protocol for testing microbiological water purifiers. USEPA Task Force Report 39. US FDA, United States Food and Drug Administration (1998) Salmonella In: Bacteriological Analytical Manual, Chapter 5 Available at http://www.fda.gov/Food/ScienceResearch/LaboratoryMethods/BacteriologicalAn alyticalManualBAM/ucm070149.htm. (accessed 12 Dec 2011). Wang, Y., Chang, Y., Chuang, H., Chiu, C., Yeh, K., Chang, C., H suan, S., Lin, W. and Chen, T. (2011) Transmission of Salmonella between swine farms by the housefly (Musca domestica). J Food Prot 74 1012 101.
190 BIOGRAPHICAL SKETCH Rachel was born and raised in Ontario, Canada. She completed her B.Sc. in n utrition and f ood s ciences, with distinction, at the University of Alberta and her M.Sc. in f ood s cience at the University of Guelph. Rachel was awarded the Alumni Fellowship to pursue a Ph.D. in f ood s cience at the University of Florida. Rachel served for two years on the Student Professional Development Board for the International Association for Technologists College Bowl Team Captain. Previously, Rachel has been employed as an Allergen Pr ogram Coordinator for Nestle, and has been an Adjunct Lecturer for Conestoga College, Canada. harvest environment. She has specialized in Salmonella in the surface water environment, but wi shes to extend her research to understanding Salmonella in other pre harvest environments. Upon completion of the Ph.D. program Rachel intends to seek a teaching position at the community college level.