Tracking Salmonella Contamination in Various Watersheds and Phenotypic and Genotypic Diversity

Abstract

Salmonella enterica is an important foodborne pathogen, and contamination of surface and ground water that may result from various human activities, such as animal production and urbanization, may contribute to the public health burden. The aims of this study was to determine the sources of Salmonella contamination in four different types of watersheds and to assess the relative contribution of multidrug-resistant strains. Eighty-six water samples collected from four different watershed systems, including those impacted by swine production (n = 12), residential/industrial (n = 34), crop agriculture (n = 12), and forestry (n = 28), were cultured for Salmonella and further characterized by serotyping, antimicrobial susceptibility testing, and pulsed-field gel electrophoresis genotyping. Salmonella prevalence was high in all four watersheds: residential/industrial area (58.8%), forestry (57.1%), crop agriculture (50%), and swine production (41.7%). Majority of the Salmonella isolates (87.1%) were pansusceptible. Multidrug resistance up to eight antimicrobials (R-type: AmStTeAxChCeKmGm) was detected in water samples that originated from swine production systems only. Serovars identified included Anatum, Gaminara, and Inverness (18.3% each) and Muenchen and Newport (8.7% each), Bredeny (7.6%), and Montevideo (6.8%). Pulsed-field gel electrophoresis analysis indicated genotypic relatedness among Salmonella recovered from residential/industrial and forestry-associated watersheds (genotypic cluster types A, C, D, E, F, G, H, and J), sites with relatively close geographic proximity. Swine-production-associated isolates were distinctly different from the others (genotypic cluster types B and I), corroborating the phenotypic findings. Overall, the findings suggest that all the various watersheds, including natural forest, remain important contributors of Salmonella contamination. While swine-production-associated water samples were not found to have a disproportionately high prevalence, it was the most important reservoir of multidrug-resistant strains.

Introduction

A pproximately 56,000 acres of shellfish beds along coastal North Carolina and many more acres nationwide are permanently closed to harvesting due to high levels of bacterial contamination (White, 2001). Many more streams in North Carolina and nationally are also listed as impaired due to fecal coliform contamination. Shellfish bed closures due to excessive levels of fecal indicator bacteria are common to many eastern and some Pacific states, causing the loss of millions of dollars of potential revenue (Little, 1989; Line et al., 2008). Fecal coliform levels determined using traditional monitoring approaches do not differentiate relative contributions of human, animal, and other sources. Therefore, phenotypic and genotypic approaches that supplement traditional methods and enable effective tracking of sources of contamination are essential.

Bacterial levels in surface water depend on a highly variable combination of factors, including terrestrial stores of fecal matter, hydrologic characteristics of storm events, and physical characteristics of the delivery area (White, 2001). In many cases, bacterial contamination and associated pollutants such as sediment that facilitate its movement are assumed to originate from human activities, particularly animal farming and urbanization (White, 2001; Line et al., 2008). Fecal bacterial indicator organisms have been proposed as useful tools in the quantification of indicators of pollution sources that are a threat to public health (White, 2001; WHO, 2008). Majority of the studies to date mainly focused on indicator organisms, and limited data exist on pathogens of public health significance. Pathogenic organisms such as salmonellae are ubiquitous and important zoonotic enteric bacteria, and shedding of these organisms in excreta of humans and animals can serve as main sources of contamination of the environment and can contribute to the pathogen load of the various watersheds by runoff from fields (Ferguson et al., 2003). The organisms can persist in the environment for long periods and are the prime pathogens transmitted through food and food products, including water and shellfish products (Baudart et al., 2000; Baloda et al., 2001). Salmonella-contaminated water and fresh produce have been implicated in a number of outbreaks in the United States and indicate the significance of environmental sources of Salmonella to human illness compared to those originating from meat and poultry products (Mohle-Boetani et al., 2002; Martinez-Urtaza et al., 2004; Haley et al., 2009; Hanning et al., 2009).

Previous data on antimicrobial resistance among foodborne pathogens, including Salmonella isolated from swine and humans, have shown that resistance to multiple antimicrobial agents has been common in isolates derived from swine production units (Gebreyes et al., 2000, 2004). Salmonella isolates from clinical human samples showed lower frequency of multidrug resistance than those of porcine origin (Gebreyes et al., 2009). Similar as well as distinct genotypic strains of Salmonella from environmental and fecal specimens in swine production units and associated environment were also detected (Gebreyes and Altier, 2002).

Overall, information on the sources and diversity of Salmonella in watersheds is limited. Prompted by the paucity of data on foodborne pathogens in watershed systems, this study was conducted to determine the sources and prevalence of Salmonella in four watershed systems and to characterize isolates using phenotypic and genotypic approaches. Phenotypic (serotyping and antimicrobial susceptibility profiling) and genotypic (DNA fingerprinting using the pulsed-field gel electrophoresis [PFGE]) approaches were also conducted to characterize Salmonella isolates recovered from the four watersheds. Clonal relatedness of Salmonella isolates among the four different watershed systems was elucidated.

Methods

Sampling sites and collection methods

A total of 86 water samples were collected in duplicates (n = 172) from four different watershed systems: swine production (n = 12), residential/industrial (Jumping Run Creek [JRC]; n = 34), row-crop agriculture (open ground farm; n = 12), and forestry (Pettiford Creek; n = 28) from January 2005 to October 2007 (Fig. 1). The JRC watershed encompasses about 320 ha in the eastern part of North Carolina and is part of the White Oak River Basin. The land uses represented in the watersheds include small residential lots that contain mobile homes and recreational vehicles (24 ha), low- and medium-density single-family homes (157 ha), commercial and industrial facilities (93 ha), forest (45 ha), and mining (1 ha). A detailed description of the watersheds is given by Line et al. (2008). The swine farm watershed was located in one of the densest pork production areas (south eastern North Carolina). Lagoon water samples (n = 12) at swine production site was also additionally collected using conventional methods using a 12 feet aluminum pole with a 250 mL sampler bottle clamped to it. Lagoon-specific sites for water collections including inlet, with four different corners, and outlet were sampled by dipping a sampler bottle into the water ∼12–18 inches below the water surface. Locations and additional details on watershed characteristics and sampling techniques for the other watersheds can be found in Line et al. (2008).

FIG. 1.

Sample collection sites for the study.

One monitoring station was identified per watershed system. An automated sampler with an integrated flow meter was installed at each monitoring station. The automated samplers continuously measured water depth or stage, and converted these measurements to discharge using a stage–discharge relationship that was programmed into the machine. Flow-proportional samples were collected automatically during rainfall events and placed in 1 of 24 sampler bottles. Soon after each storm event, a technician visited the site and placed equal volume aliquots from each sampler bottle into a sterile laboratory bottle, placed on ice, and shipped the samples for overnight delivery to the laboratory for analysis. In addition, nonstorm event (grab) samples were collected quarterly throughout the year when there was sufficient nonstorm discharge to characterize base flow. These were obtained from the vertical and horizontal midpoint of flow to approximate the average concentration.

Salmonella isolation and identification

All water samples were shipped on ice overnight to the Infectious Diseases Molecular Epidemiology Laboratory, Department of Veterinary Preventive Medicine, The Ohio State University, and cultured for Salmonella isolation immediately upon arrival. Isolation and identification of Salmonella was done following conventional methods as described previously (Gebreyes et al., 2004) with some modifications. Each water sample was processed in duplicates. Briefly, 25 mL of water samples was preenriched in equal volume (25 mL) of buffered peptone water (Becton Dickinson) with a 1:1 v/v ratio and incubated at 37°C overnight. About 100 μL of the pre-enriched suspension was transferred into 9.9 mL of Rappaport–Vassiliadis broth (Becton Dickinson) and incubated at 42°C for 24 h. A loopful of the selective enrichment was plated onto XLT4 (Xylose Lysine Tergitol TM4; Becton Dickinson) plates and incubated at 37°C for 24 h. A total of 10 presumptive Salmonella colonies (5 per XLT4 plate) were selected and tested for biochemical and serological reactions. When there were less than five colonies per plate, all colonies from each plate were tested. All isolates were further subjected to biochemical testing using triple sugar iron agar and urea agar slants for confirmation and serologically using polyvalent Salmonella antisera (Poly Groups A-I and Vi antiserum; Becton Dickinson) for serogrouping purposes.

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing of Salmonella isolates (n = 427) was performed using Kirby–Bauer disk diffusion method to a panel of 12 antimicrobial agents. The following BBL™ Sensi-Disc™ antimicrobial susceptibility test discs (Becton Dickinson) with their respective abbreviations and disc potencies were used: ampicillin (Am; 10 μg), amoxicillin/clavulanic acid (Ax; 20/10 μg), amikacin (An; 30 μg), ceftriaxone (Ce; 30 μg), cephalothin (Ch; 30 μg), chloramphenicol (Cl; 30 μg), ciprofloxacin (Cip; 5 μg), gentamicin (Gm; 10 μg), kanamycin (Km; 30 μg), streptomycin (St; 10 μg), sulfamethoxazole (Su; 250 μg), and tetracycline (Te; 30 μg). Results were interpreted according to the Clinical Laboratory Standards Institute recommendations (NCCLS, 2002a). All isolates that showed intermediate resistance were grouped with the susceptible strains to avoid overestimation of resistance. Control tests of Escherichia coli ATCC 25922, E. coli ATCC 35218, Enterococcus faecalis ATCC 29212, Staphylococcus aureus ATCC 25923, and Pseudomonas aeruginosa ATCC 27853 were regularly performed in accordance to the Clinical Laboratory Standards Institute standards (NCCLS, 2002b).

Serotyping

Of the total 427 Salmonella isolates, 104 were selected systematically representing various antimicrobial resistance patterns and watershed sources, and serotyped at the Ohio Department of Health, Columbus, Ohio.

PFGE genotyping

DNA fingerprinting of Salmonella isolates (n = 104) was conducted using the PFGE. PFGE was performed according to the standardized laboratory protocol recommended by the PulseNet program, Centers for Disease Control and Prevention (CDC) (Ribot et al., 2006). The PulseNet “universal” standard strain Salmonella enterica serovar Braenderup H9812 was used as a reference marker. Gel images were then transferred to Bionumerics software version 4.61 (Applied Maths NV) 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. Similarity coefficients were obtained within Bionumerics by calculating Dice coefficients. PFGE banding patterns with a similarity index >80% were grouped within the same genotypic cluster.

Statistical analysis

Descriptive and inferential statistical analysis such as frequency distribution and cross-tabulation analysis was carried out using SPSS^® statistical software packages version 16.0 (SPSS Inc.). Statistical significance was considered at the p < 0.05 levels.

Results

Salmonella prevalence

The overall prevalence of Salmonella was 54.7%, with the 95% confidence interval between 44% and 65% (Table 1). The highest proportion of Salmonella isolation was detected in water samples collected from the residential/industrial area (58.8%), followed by samples collected from forestry area (57.1%). Half of the water samples from crop agriculture (6 of 12) were also Salmonella positive. The prevalence of Salmonella in water samples associated with swine production units was relatively low with 41.7% prevalence (Table 1). There was no significant difference on the prevalence of Salmonella among water samples from the four watershed systems (p > 0.05).

Table 1.

Salmonella Prevalence, Distribution, and Antimicrobial Resistance Patterns of Serovars Among Different Sources of Watersheds

Water source (n)	Prevalence (%)	95% CI	Salmonella serovars (no. of colonies) ^a	Antimicrobial resistance patterns (n)
Swine production (12)	5 (41.7)	9–74	Anatum (19)	AmAxChCeGmKmStTe (2) AmAxChGmKmStTe (3) AmAxChCeGmKmTe (1) GmKmStSuTe (3); GmKmStTe (7) GmStSuTe (1); GmKmSt (1); StTe (1)
			Bredeney (7)	Su (3); StSu (1); Pansusceptible (3)
			Inverness (1)	AmGmKmStTe
			Mbandaka (2)	SuTe (2)
Residential/industrial (34)	20 (58.8)	41–76	Gaminara (8)	St (3) Pansusceptible (5)
			Bullbay (1)	Pansusceptible
			Give (3)	Pansusceptible (3)
			Inverness (8)	Pansusceptible (8)
			Miami (4)	Pansusceptible (4)
			Montevideo (6)	Pansusceptible (6)
			Muenchen (3)	Pansusceptible (3)
			Rough O's:d:7 (1)	Pansusceptible (1)
Forestry (28)	16 (57.1)	38–77	Gaminara (9)	St (3); Pansusceptible (6)
			Give (2)	Pansusceptible (2)
			Inverness (9)	Pansusceptible (9)
			Muenchen (6)	St (1); Pansusceptible (5)
Crop agriculture (12)	6 (50.0)	17–83	Brandenburg (1)	Pansusceptible (1)
			Gaminara (1)	Pansusceptible (1)
			Inverness (1)	Pansusceptible (1)
			Miami (2)	StTeKm (1); Pansusceptible (1)
			Newport (9)	Pansusceptible (9)
Overall (Total = 86)	47 (54.7)	44–65	104

Up to six colonies per positive sample were further characterized.

Am, ampicillin; Ax, amoxicillin/clavulanic acid; Ce, ceftriaxone; Ch, cephalothin; Gm, gentamicin; Km, kanamycin; St, streptomycin; Su, sulfamethoxazole; Te, tetracycline; CI, confidence interval.

Antimicrobial resistance patterns

Of the total Salmonella isolates tested for antimicrobial susceptibility (n = 427), 12.9% (55/427) were resistant to one or more of the antimicrobials, and by far the majority, 72.7% (40/55), of the resistant isolates originated from swine-production-associated watershed (Table 1). All isolates tested were susceptible to the Cl, Cip, and An. Overall, 15 different antimicrobial resistance patterns (R-types) were detected, with the majority being pansusceptible (87.1%) (Table 1). Among the resistant isolates, 63.6% (35/55) exhibited multidrug resistance (arbitrarily defined as resistance to ≥3 antimicrobials) and the majority originated from water samples associated with swine. The R-types detected in Salmonella from water sample associated with swine production revealed the most diversity with 13 different resistance patterns (Table 1). In addition, highly multidrug-resistant (MDR) strains to up to eight antimicrobials were also detected from these water samples with R-types, with AmStTeAxChKmGm+ being the predominant MDR pattern. Among isolates from the other three watersheds, one isolate recovered from crop agriculture area showed MDR with resistance to St, Te, and Km (R-Type: StTeKm). The antimicrobial resistance profiles of Salmonella isolates detected in water samples from residential/industrial (n = 4) and forestry (n = 5) were limited to streptomycin resistance (R-Type: St) (Table 1). Overall, while the MDR isolates were identified mainly from water samples in swine production systems, antimicrobial resistance profiling, as a tool for subtyping, was not highly discriminatory among the isolates that originated from the various watersheds.

Serovar distributions

A total of 12 different serovars were detected among 104 Salmonella isolates serotyped (Table 1). One Salmonella isolate of residential/industrial water origin was untypable. Three serovars, Anatum, Gaminara, and Inverness (18.3% each), represented the majority (55.2%) of the isolates in the study. Serovars of high public health significance, such as serovar Muenchen and Newport (each 8.7%), Bredeny (6.7%), and Montevideo (5.8%), were also detected. Two predominant Salmonella serovars, Anatum (65.5%) and Bredeney (24.1%), were detected in water samples collected from swine production units only, whereas serovars Gaminara, Inverness, and Muenchen were primarily found in water samples related to forestry and residential/industrial areas. In addition, Salmonella serovars Newport and Montevideo, which are among the top 10 most frequent serovars reported in human salmonellosis cases in the United States (CDC, 2003), were found in water samples collected from crop agriculture and residential/industrial watershed systems, respectively. The highest diversity based on serotyping was found in water samples from residential/industrial areas (eight different serovars) compared to other water samples from crop agriculture (five different serovars), forestry (four different serovars), and swine production (four different serovars) (Table 1).

PFGE analyses

The PFGE XbaI macrorestriction banding patterns consisted of 10–13 DNA fragment bands with sizes between 33 and 1150 kb. PFGE analysis of the 104 Salmonella isolates among different watershed systems generated 10 major genotypic clusters (A–J) comprising at least four isolates with a dice coefficient index cut-off point of 80% (Fig. 2). The 28 Salmonella isolates recovered from the water sample associated with swine production showed identical DNA fingerprint profiles and clustered in only two genotypic groups (cluster B, n = 19; cluster I, n = 7). The predominant serovar in this watershed and also in cluster B was Salmonella Anatum. Salmonella isolates from water samples associated with the forestry, residential/industrial, and crop agriculture areas showed clonal relatedness to each other and were grouped under two genotypic cluster (clusters A and C), implying sharing of similar genotypes and suggesting potential cross contamination among these watersheds. The findings reveal that Salmonella serovars recovered from water samples associated with swine production system were distinctly different and clonally unrelated to isolates from other three watershed systems.

FIG. 2.

Dendrogram representing PFGE-XbaI fingerprinting and antibiograms of Salmonella isolates recovered from difference sources of watersheds. Black boxes demonstrate resistance to a particular antimicrobial agent tested in the current study. For abbreviations of antimicrobials, refer to Table 1. Ten major pulsotypes (A–J) comprising at least four isolates were identified at >80% genetic similarity. PFGE, pulsed-field gel electrophoresis.

Discussion

The overall prevalence (54.7%) of Salmonella recovered from watershed systems in this study was higher than that reported in previous studies in other geographic locations. A study in Canadian watershed systems reported a 6.2% prevalence of Salmonella (Johnson et al., 2003), and a recent study in the coastal plains of south Georgia detected Salmonella in 79.2% (57/72) of the water samples collected from surface waters in a region with a history of high case rates of salmonellosis (Haley et al., 2009). In the current study, almost 60% of water samples from residential/industrial watershed area were Salmonella positive. One of the common assumptions regarding enteric pathogens such as Salmonella load in surface water is that livestock production is one of the primary sources of contamination. This study revealed that almost half of the water samples from swine production units were positive for Salmonella. While this is a relatively high proportion, samples from other sources were also found to be positive for Salmonella; more than half (57.1%) of water samples from watershed in forestry areas were also Salmonella positive.

Multidrug resistance was highly associated with water samples from swine production systems. This may be associated with the common use of antimicrobials in swine production for growth promotion and prophylactic and therapeutic purposes. Even though we found 15 different patterns of antimicrobial resistance among S. enterica serovars recovered from water samples, the majority (87.1%) were susceptible to all the 12 antimicrobials tested. In contrast, only 4 of the 49 Salmonella isolates from water samples associated with swine production were shown to be pansusceptible. In concordance with previous data, resistance to multiple antimicrobial agents has been common in Salmonella isolates derived from swine production units (Gebreyes et al., 2000, 2004; Gebreyes and Altier, 2002). The MDR strains, particularly Salmonella Anatum, exhibited resistance up to eight antimicrobials (R-Type: AmAxChCeKmGmStTe). However, the antimicrobial resistance patterns of the majority of Salmonella isolates recovered from the other three watersheds were mainly pansusceptible (97.4%; 368/378) (Table 1). We were not able to compare and contrast our results with other previous studies because data on sources, diversity, and antimicrobial resistance profiles of Salmonella from various watershed strains are very limited.

A total of 12 different serovars were found among the Salmonella isolates from the four watershed systems. Salmonella Anatum and Salmonella Bredeney were the two most predominant serovars detected from the swine production watersheds. We also analyzed additional isolates that were recovered from cattle feces grazing around the swine production units and found that ∼60% of the fecal samples collected from cattle (n = 12) were Salmonella positive. Serotyping results showed that all were identified as Salmonella Anatum, further implying that the isolates may have spread from the swine barns via recycled lagoon water spray on grazing field. Three major Salmonella serovars detected from the three watersheds, Newport (crop agriculture), Montevideo (residential/industrial), and Muenchen (forestry and residential/industrial), were among the top 10 most frequently reported Salmonella serovars from human salmonellosis cases in the United States (CDC, 2007). While none of these serovars were detected in water samples associated with swine production systems in the current study, serovar Muenchen was previously reported in swine in the same geographic area (Gebreyes and Thakur, 2005). In the current study, Salmonella Newport was detected only from crop-agriculture-associated watershed and was pansusceptible and did not appear to be related to the highly MDR Newport strains reported previously (Varma et al., 2006; Egorova et al., 2008; Greene et al., 2008; Irvine et al., 2009). Salmonella Gaminara and Inverness were detected from residential and natural forest water samples. These two serovars have been reported in humans (CDC, 2003), and serovar Gaminara is known to be mostly associated with wild animals (Gaertner et al., 2008), corroborating its common occurrence in the national-forest-associated water samples in the current study.

An increase in the diversity of serovars was detected in residential/industrial area with eight different serovars isolated from this water source. The JRC watershed contains mixed land uses, including residential, industrial, and forested areas along streams or in wetland areas. Such mixed land use may have contributed to the diversity of sources of bacteria, including humans and domestic and wild animals, including birds. In addition, this ∼800-acre watershed has been extensively ditched to reduce the high ground water table. The ditches cause the surface and ground waters, and the bacteria with which the waters intersect, to comingle extensively before emptying from the creek into the shellfish beds of Bogue Sound. In addition, the ditches reduce the time of concentration for water draining the landscape, which allows the many sources of bacteria, whether they are from wildlife, humans, or domestic pets that are available in JRC watershed, little time for die off.

In the present study, we report the use of DNA fingerprinting (PFGE) and serotyping as important tools to track and assess the spread of Salmonella. PFGE analyses subdivided the Salmonella isolates recovered from different watersheds into 10 major genotypic clusters. It was also surprising and important to note that in some instances, different serotypes showed similar PFGE profiles. This is particularly true since in the current study, only one restriction enzyme (XbaI) was used to discriminate among the isolates. Since PFGE is a macrorestriction profiling, such occurrences may be observed. This is caused mainly because serotype difference may be caused by slight changes in somatic and flagellar epitopes that may not necessarily result in changing macrorestriction profiles. Therefore, as recommended previously, interpretation of findings is best done by looking at both phenotypic and genotypic aspects (Gebreyes et al., 2006). Multidrug resistance Salmonella strains from the swine production were clonally related and were grouped into two clusters (clusters B and I). Most of the S. enterica serovars recovered from residential/industrial and forestry showed similar PFGE profiles in diverse clusters. This suggests that the strains that are commonly detected in the residential/industry areas may be similar due to the forested areas contained within the watershed and geographic proximity.

Conclusions

This study revealed that various watersheds, including the national forest, could be important contributors of Salmonella contamination of shellfish water sanitation areas as well as other potential public health settings. We also demonstrated that the PFGE molecular typing coupled with serotyping techniques enabled to track Salmonella strains and represents a suitable tool for the epidemiological studies of Salmonella recovered from various watersheds.

Footnotes

Acknowledgments

This work was supported by a grant from the U.S. Department of Agriculture–Integrated Water Quality Program (2004–2007). We acknowledge technical assistance by members of the Infectious Diseases Molecular Epidemiology Laboratory team, Derek Coombs and Rebecca Robbins from North Carolina State University.

Disclosure Statement

No competing financial interests exist.

References

Baloda

SJB

, Christensen

, Trajcevska

. Persistence of a Salmonella enterica serovar Typhimurium DT12 clone in a piggery and in agricultural soil amended with Salmonella-contaminated slurry. Appl Environ Microbiol, 2001; 67:2859–2862.

Baudart

, Lemarchand

, Brisabois

et al. Diversity of Salmonella strains isolated from the aquatic environment as determined by serotyping and amplification of the ribosomal DNA spacer regions. Appl Environ Microbiol, 2000; 66:1544–1552.

[CDC] Centers for Disease Control and Prevention. Salmonella Surveillance: Annual Summary, 2005. Atlanta, GA: U.S. Department of Health and Human Services, CDC, 2007.

[CDC] Centers for Disease Control and Prevention. Salmonella Surveillance Summary, 2002. Atlanta, GA: U.S. Department of Health and Human Services, CDC, 2003.

Egorova

, Timinouni

, Demartin

et al. Ceftriaxone-resistant Salmonella enterica serotype Newport, France. Emerg Infect Dis, 2008; 14:954–957.

Ferguson

, Husman de

, Atavilla

et al. Fate and transport of surface pathogens in watersheds. Crit Rev Environ Sci Technol, 2003; 33:299–361.

Gaertner

, Hahn

, Rose

et al. Detection of salmonellae in different turtle species within a headwater spring ecosystem. J Wildl Dis, 2008; 44:519–526.

Gebreyes

, Altier

. Molecular characterization of multidrug resistant Salmonella enterica subsp. enterica serovar Typhimurium isolates from swine. J Clin Microbiol, 2002; 40:2813–2822.

Gebreyes

, Altier

, Thakur

. Molecular epidemiology and diversity of Salmonella serovar Typhimurium in pigs using phenotypic and genotypic approaches. Epidemiol Infect, 2006; 134:187–198.

10.

Gebreyes

, Davies

, Morrow

et al. Antimicrobial resistance of Salmonella isolates from swine. J Clin Microbiol, 2000; 38:4633–4636.

11.

Gebreyes

, Thakur

. Multidrug resistant Salmonella enterica serovar Muenchen from pigs and humans and potential interserovar transfer of antimicrobial resistance. Antimicrob Agents Chemother, 2005; 49:503–511.

12.

Gebreyes

, Thakur

, Davies

et al. Trends in antimicrobial resistance, phage types and integrons among Salmonella serotypes from pigs, 1997–2000. J Antimicrob Chemother, 2004; 53:997–1003.

13.

Gebreyes

, Thakur

, Dorr

et al. Occurrence of spvA virulence gene and clinical significance for multidrug-resistant Salmonella strains. J Clin Microbiol, 2009; 47:777–780.

14.

Greene

, Stuart

, Medalla

et al. Distribution of multidrug-resistant human isolates of MDR-ACSSuT Salmonella Typhimurium and MDR-AmpC Salmonella Newport in the United States, 2003–2005. Foodborne Pathog Dis, 2008; 5:669–680.

15.

Haley

, Cole

, Lipp

. Distribution, diversity, and seasonality of waterborne salmonellae in a rural watershed. Appl Environ Microbiol, 2009; 75:1248–1255.

16.

Hanning

, Nutt

, Ricke

. Salmonellosis outbreaks in the United States due to fresh produce: sources and potential intervention measures. Foodborne Pathog Dis, 2009; 6:635–648.

17.

Irvine

, Gillespie

, Smyth

et al. Investigation of an outbreak of Salmonella enterica serovar Newport infection. Epidemiol Infect, 2009; 19:1–8.

18.

Johnson

, Thomas

, Graham

et al. Prevalence of Escherichia coli O157:H7 and Salmonella spp. in surface waters of southern Alberta and its relation to manure sources. Can J Microbiol, 2003; 49:326–335.

19.

Line

, White

, Kirby-Smith

et al. Fecal coliform export from four coastal North Carolina areas. J Am. Water Res Assoc, 2008; 44:606–617.

20.

Little

. The economy of rain and the Tillamook imperative. J Soil Water Conserv, 1989; 44:199–202.

21.

Martinez-Urtaza

, Saco

, de Novoa

et al. Influence of environmental factors and human activity on the presence of Salmonella serovars in a marine environment. Appl Environ Microbiol, 2004; 70:2089–2097.

22.

Mohle-Boetani

, Werner

, Polumbo

. Outbreak of Salmonella serotype Kottbus infections associated with eating alfalfa sprouts-Arizona, California, Colorado, and New Mexico, February–April 2001. MMWR Morb Mortal Wkly Rep, 2002; 51:7–9.

23.

[NCCLS] National Committee for Clinical Laboratory Standards. Performance Standards for Antimicrobial Disk and Dilution Susceptibility Tests for Bacteria Isolated from Animals; Approved Standard, M31-A2, 2nd. Wayne, PA, 2002a.

24.

[NCCLS] National Committee for Clinical Laboratory Standards. Performance Standards for Antimicrobial Susceptibility Testing; 12^th Informational Supplement, M100-S12. Wayne, PA, 2002b.

25.

Ribot

, Fair

, Gautom

et al. Standardization of pulsed-field gel electrophoresis protocols for the subtyping of Escherichia coli O157:H7, Salmonella, and Shigella for PulseNet. Foodborne Pathog Dis, 2006; 3:59–67.

26.

Varma

, Marcus

, Stenzel

et al. Highly resistant Salmonella Newport-MDR AmpC transmitted through the domestic US food supply: a FoodNet case-control study of sporadic Salmonella Newport infections, 2002–2003. J Infect Dis, 2006; 194:222–230.

27.

White

. Integrated watershed-based molecular and hydrologic monitoring techniques to assess pathogen loading. 2001. www.usawaterquality.org/conferences/2004/abstracts/watershed/nancy white.pdf. 2008 July 2(Online.)

28.

[WHO] World Health Organization. Guidelines for Drinking-Water Quality: Vol. 1, Recommendations, 3rd. Geneva, Switzerland, 2008.