Plant-Pathogenic Oomycetes, Escherichia coli Strains, and Salmonella spp. Frequently Found in Surface Water Used for Irrigation of Fruit and Vegetable Crops in New York State

Abstract

In the United States, surface water is commonly used to irrigate a variety of produce crops and can harbor pathogens responsible for food-borne illnesses and plant diseases. Understanding when pathogens infest water sources is valuable information for produce growers to improve the food safety and production of these crops. In this study, prevalence data along with regression tree analyses were used to correlate water quality parameters (pH, temperature, turbidity), irrigation site properties (source, the presence of livestock or fowl nearby), and precipitation data to the presence and concentrations of Escherichia coli, Salmonella spp., and hymexazol-insensitive (HIS) oomycetes (Phytophthora and Pythium spp.) in New York State surface waters. A total of 123 samples from 18 sites across New York State were tested for E. coli and Salmonella spp., of which 33% and 43% were positive, respectively. Additionally, 210 samples from 38 sites were tested for HIS oomycetes, and 88% were found to be positive, with 10 species of Phytophthora and 11 species of Pythium being identified from the samples. Regression analysis found no strong correlations between water quality parameters, site factors, or precipitation to the presence or concentration of E. coli in irrigation sources. For Salmonella, precipitation (≤0.64 cm) 3 days before sampling was correlated to both presence and the highest counts. Analyses for oomycetes found creeks to have higher average counts than ponds, and higher turbidity levels were associated with higher oomycete counts. Overall, information gathered from this study can be used to better understand the food safety and plant pathogen risks of using surface water for irrigation.

INTRODUCTION

Water is necessary for crop production and used for irrigation, pesticide applications, freeze protection, and other agricultural purposes. For crop production, many produce growers in the United States use surface water from a variety of sources, which may include ponds, creeks, streams, lakes, and canal ways. A recent study of New York State fruit and vegetable growers responding to a survey found that 57% use surface water for irrigation and 18% use surface water for mixing of topical/pesticide sprays (1). Surface water sources are considered to be at high risk for pathogen contamination because they are open to many routes by which both human food-borne illness and plant disease-causing microorganisms can enter. Human-pathogenic bacteria are believed to enter surface waters mainly through contamination from fecal material from wildlife and livestock directly or indirectly via contaminated water, soil, or debris. Plant pathogens can enter surface water sources through many routes, including infested soil, water, and debris; cull piles; and field drainage tiles (2). The types and frequency of disease-causing microorganisms can differ from one geographic location to the next, due to climate, available hosts, and other environmental characteristics.

There has been increasing scrutiny of agricultural water with respect to food safety, especially for fruit and vegetable growers, whose produce is likely to be consumed raw. Two groups of bacteria that are frequently associated with food-borne illnesses caused by contaminated produce are pathogenic Escherichia coli and Salmonella species. Both of these groups of bacteria have been reported from surface water sources and could be introduced into the fruit- and vegetable-growing environment through irrigation (3). Once introduced, these bacteria have the potential to cause food-borne illness due to the consumption of contaminated produce. Escherichia coli and Salmonella have been shown to be capable of surviving on plant surfaces and persisting in the soil for long periods of time (3–6). There is even evidence that when these bacteria, whose main hosts are mammals, are exposed to surface water before introduction to a plant surface, they are better able to survive and persist on plant surfaces due to a stress response that causes the bacteria to produce structures that mediate bacterial attachment and exopolysaccharides that protect them from desiccation and UV exposure (7, 8). Food-borne illness can lead to death, and great financial losses can result if an outbreak is associated with a farm or type of produce (9). The Food Safety Modernization Act (FSMA), which was signed into law in 2011, focuses on the prevention of contamination in the growing environment and requires many growers to monitor their irrigation water for generic E. coli, an indicator of fecal contamination, and discontinue use if levels exceed 235 CFU/100 ml or an average over 5 samples of 126 CFU/100 ml (10).

All major groups of plant pathogens, which include bacteria, viruses, fungi, nematodes, and oomycetes, have been found in irrigation water. In a 2005 review of plant pathogens in irrigation water, Hong and Moorman listed 8 species of bacteria, 43 species of oomycetes, 27 species of fungi, 10 viruses, and 13 species of nematodes that have been recovered from irrigation water sources, and many of these were from surface water (2). Among the plant pathogens, the oomycetes present the biggest waterborne threat for fruit and vegetable growers. In particular, members of the phylum Heterokontophyta in the genera Phytophthora and Pythium are pathogenic to plants, are well suited to be spread through surface water, and are commonly referred to as water molds. Many members of the genera Phytophthora and Pythium produce abundant numbers of biflagellate, asexual zoospores that are able to move through water toward potential plant hosts. Zoospores are believed to be primarily responsible for water infestations, but other structures, such as sporangia, oospores, and mycelia, may also infest water sources. Zoospores have been reported to remain viable in surface waters for hours to weeks, depending on the species and environmental factors, such as temperature and pH (11). Phytophthora capsici is the causal agent of phytophthora blight of many vegetables, including tomato, pepper, cucurbits, and snap and lima beans, and has many weed hosts (12, 13). The pathogen has been found in surface waters in several agricultural regions of the United States, but its presence in New York State surface waters was previously unknown.

This study was designed to develop a better understanding of the prevalence and concentrations of generic E. coli, Salmonella spp., and hymexazol-insensitive (HIS) oomycete plant pathogens in surface waters used for irrigation of fruit and vegetable crops in New York State. Water quality parameters, irrigation site properties, and precipitation data were collected and analyzed for correlations with the presence and levels of microorganisms in surface water irrigation sources. This study adds to the growing body of information regarding food-borne and plant pathogen risks associated with surface irrigation water.

MATERIALS AND METHODS

Water sampling and analysis.

During the growing seasons of 2010 and 2011, a pathogen survey was conducted in actively used surface water irrigation reservoirs throughout vegetable-growing regions of New York (Fig. 1). Water samples were collected monthly between May and October (Table 1) and were collected near the surface where water was drawn for irrigation. During each sampling event, 2 liters was collected in sterile 1-liter bottles (Nalgene, Penfield, NY) using a telescopic swing sampler (Nasco, Fort Atkinson, WI). Water temperature was measured for samples collected in 2011 using a pond thermometer (Lifeguard Aquatics, Cerritos, CA), and samples were stored in a cooler for transport back to the laboratory, where they were processed within 24 h. pH (HI 2211 pH/ORP meter; Hanna, Woonsocket, RI) and turbidity (2100P portable turbidimeter; Hach, Loveland, CO) measurements were taken for all samples collected in 2011. To test for the presence of generic E. coli and Salmonella spp., 100 ml of water was filtered using 47-mm, 0.45-μm-pore-size filters (Thermo Fisher Scientific, Waltham, MA). The filters were placed onto violet red bile agar (VRBA; Hardy Diagnostics, Santa Maria, CA) containing 4-methylumbelliferyl-fl-d-glucuronide (MUG; Hach, Loveland, CO) to test for E. coli and on bismuth sulfite agar (Criterion, Santa Maria, CA) to test for Salmonella spp. Bacterial samples were incubated at 37°C for 18 h. Samples from 18 sites were assayed for both E. coli and Salmonella spp. (Table 1). Primary identification of E. coli was determined by its characteristic appearance on VRBA-MUG and fluorescence under long-wave UV (365-nm) light. Primary identification of Salmonella spp. was based on diagnostic appearance (dark brown to black colonies with or without a metallic sheen) on bismuth sulfite agar. Representative colonies from E. coli- and Salmonella species-positive samples were cryopreserved at −80°C in Luria-Bertani broth with 15% glycerol for later PCR confirmation (14). The DNA from cryopreserved samples was extracted as previously described (15). Confirmation that an organism was E. coli was performed using primers UP and DOWN, as described by Chen and Griffiths (16). Confirmation that an organism was a Salmonella sp. was performed using primers 139 and 141, as previously described (17).

FIG 1.

Map of irrigation water sampling sites in New York. Each black circle represents an irrigation reservoir used in this study.

TABLE 1.

Surface water irrigation sites sampled and water quality parameters

Site	County	Source	Bait trap	Date (mo) of sampling		pHa	Turbiditya (NTU)	Tempb (°C)
				2010	2011
A	Genesee	Pond	Yes	May–September	NTc	8.20	NT	NT
Bd	Monroe	Canal	No	May–October	Aug–October	8.33	7.93	18.9–26.7
C	Orleans	Creek	No	May–September	NT	8.31	NT	NT
D	Orleans	Creek	Yes	July–September	NT	8.05	NT	NT
E	Monroe	Creek	Yes	May–September	NT	8.37	NT	NT
F	Columbia	Pond	Yes	June–September	NT	9.59	NT	NT
Gd	Columbia	Pond	Yes	June–September	June–October	9.15	58.27	18.9–26.7
Hd	Rensselaer	Pond	Yes	June–September	June–October	8.43	4.09	18.9–26.7
Id	Rensselaer	Creek	Yes	June–September	June–October	8.11	2.34	16.11–23.9
Jd	Rensselaer	Pond	Yes	June–September	June–October	8.92	3.75	16.7–26.7
K	Greene	Pond	Yes	June–September	NT	8.24	NT	NT
L	Greene	Pond	No	June–September	NT	8.10	NT	NT
Md	Erie	Pond	Yes	June–September	June–October	8.22	3.41	18.9–27.8
Nd	Erie	Pond	Yes	June–September	June–October	8.17	3.40	18.9–26.7
Od	Erie	Pond	Yes	June–September	June–October	8.03	0.75	18.9–26.7
Pd	Erie	Pond	Yes	June–September	June–October	8.17	7.23	17.8–26.7
Qd	Erie	Pond	Yes	June–September	June–October	8.97	5.44	18.9–26.7
Rd	Niagara	Pond	Yes	June–September	June–October	7.71	15.46	19.4–26.7
Sd	Niagara	Pond	No	June–October	June–October	8.20	2.08	18.9–27.8
T	Monroe	Pond	Yes	June–September	NT	8.29	NT	NT
Ud	Ontario	Pond	Yes	NT	May–October	8.74	16.52	13–27.8
Vd	Ontario	Pond	Yes	NT	May–October	9.32	9.70	18.9–27.8
Wd	Tompkins	Pond	No	NT	May–October	8.31	2.85	18.9–26.7
Xd	Tompkins	Creek	No	NT	May–October	8.38	3.20	16.7–24.4
Yd	Tompkins	Creek	No	NT	June–October	8.32	2.67	18.9–24.4
Zd	Niagara	Pond	Yes	NT	June–October	8.58	5.73	18.9–27.8
AA	Niagara	Pond	No	NT	June–October	8.18	3.91	20–25.6
BB	Orleans	Creek	No	NT	Aug–October	8.39	7.72	18.9–25.6
CC	Orleans	Creek	No	NT	Aug–October	8.42	3.39	18.9–25.6
DD	Monroe	Creek	No	NT	Aug–October	8.49	3.76	20.6–25.6
EE	Monroe	Creek	No	NT	Aug–October	8.32	2.44	18.9–25.6
FF	Wyoming	Creek	No	May	NT	8.77	NT	NT
GG	Ontario	Pond	No	NT	June	8.67	6.88	27.8
HH	Orleans	Creek	No	August	NS	8.61	NT	NT
JJ	Ontario	Creek	No	October	June	7.91	1.55	17.8
KK	Washington	Pond	Yes	NT	October	8.17	2.06	18.9
LL	Washington	Pond	No	NT	October	7.08	7.35	18.9
MM	Washington	Pond	No	NT	October	7.79	4.15	18.9

^aThese values represent the averages for all samples collected from each site in 2011.

^bThese values represent the ranges for all samples collected from each site in 2011.

^cNT, not tested.

^dSites assayed for E. coli and Salmonella spp. (all sites were assayed for oomycetes).

To assay for oomycetes, 1 liter of water was filtered using a sterile magnetic filter funnel (Pall, Port Washington, NY) and 47-mm, 5.0-μm-pore-size filters (EMD Millipore, Billerica, MA). One or more filters were used per 1-liter sample to prevent clogging. The filters were inverted onto PARPH agar and incubated at 25°C for up to 5 days. PARPH agar (cornmeal agar containing [per liter] 5 mg pimaricin, 250 mg ampicillin [Na salt], 10 mg rifampin, 50 mg pentachloronitrobenzene, and 50 mg hymexazol) is selective for HIS oomycetes, which include most Phytophthora spp. and some Pythium spp. (18). Single isolates were transferred to new PARPH agar as growth occurred. All sites were assayed for oomycetes.

Oomycete baiting in water.

Oomycetes were assayed directly from water samples and also with the use of pear, cucumber, and lemon leaf baits. Three different baits were used to attract a wide variety of oomycete plant pathogens. Green pears (Bartlett) were collected and placed in cold storage (4°C) until needed. Cucumbers were purchased from a local grocery store as needed. Lemon ( Citrus limon) trees were maintained in a greenhouse, and leaves were collected as needed. Before use, all baits were surfaced sterilized in 10% bleach for 1 min and rinsed with distilled water. Baits (2 cucumbers, 2 pears, and 2 lemon leaves) were placed in single-door rigid live traps (Edmbg, Fort Wayne, IN) from which the trap mechanisms had been removed, and cylindrical pieces of polyethylene foam (Gladon, Oak Creek, WI) were attached to the sides to keep the traps upright and floating. A landscaping brick attached to wire was used as an anchor to keep the traps from washing ashore. Baited traps were floated in the irrigation sources for 7 days, and the baits were then collected, transported back to the laboratory in a cooler, and processed within 24 h. The baits were rinsed with sterile distilled water. Tissue was excised from single lesions, placed onto PARPH agar, and incubated at 25°C for up to 5 days. Baits still in good condition (few or no lesions) were placed in a moist chamber for up to 5 days at 25°C for further lesion development; the baits were checked daily, and tissue was isolated from new lesions.

DNA extraction, PCR, and identification of oomycetes.

Oomycete isolates were grown in 15% clarified V8 broth for DNA extraction. Mycelium was rinsed with distilled water, and DNA was extracted according to a cetyltrimethylammonium bromide (CTAB)-based method developed as previously described, with modifications (polyvinylpyrrolidone and β-mercaptoethanol were omitted from extraction buffer A) (19). The 5.8S rRNA gene, internal transcribed spacer (ITS) region 1 (ITS1), and ITS2 were amplified using 50-μl PCR mixtures with primers ITS4 and ITS5, as previously described (20), under the following conditions: initial denaturation at 95°C for 5 min, followed by 35 cycles at 95°C for 1 min, 56°C for 1 min, and 72°C for 1 min and a final extension step of 72°C for 10 min. Amplification was performed using an Eppendorf Mastercycler gradient thermocycler (Eppendorf, Hauppauge, NY.) Amplicons were sequenced at Cornell University's Biotechnology Resource Center (Ithaca, NY). Identification was based on comparison of the sequences to the sequences of previously identified oomycetes in GenBank (http://www.ncbi.nlm.nih.gov/GenBank/) and the Phytophthora Database (www.phytophthoradb.org).

Calculations and statistics.

The total prevalence of E. coli, Salmonella, or oomycetes was calculated as the number of positive samples (≥1 CFU/100 ml for E. coli or Salmonella, ≥1 CFU/liter for oomycetes) divided by the total number of samples assayed for the particular organism. For each organism, prevalence was also calculated for samples within each water source type (pond, creek, or canal).

The parameters recorded from water samples for use as possible predictor variables in statistical analyses were site, pH, turbidity, water temperature, presence of livestock near the irrigation site, presence of fowl in or near the irrigation site, precipitation amounts for the day of sampling, and precipitation amounts 3 days prior to sampling. The presence or absence and concentration of each study organism were used as response variables. Precipitation data were collected using the National Climate Data Center Climate Data Online mapping tool, with data from the nearest weather station with complete precipitation data available being selected (all stations used were within 15 km of a site) (http://www.ncdc.noaa.gov/cdo-web/).

All regression trees were built using random effect-estimation methods (RE-EMs) (21). Regression trees were computed using the REEMtree package (version 2.14.2) in R, where irrigation site was specified as the random effect (R Core Development Team, Vienna, Austria). This method was chosen because it was designed for analysis of longitudinal data where data collected from repeated sampling of the same subjects or sites are not independent. Each response variable for qualitative and quantitative measures of study organisms was analyzed in a correlation matrix, using JMP Pro (version 11) statistical software (SAS Institute, Cary, NC), with possible predictor variables; those variables with the greatest correlation coefficients were used for building regression trees (see Tables S1 to S3 in the supplemental material). A correlation matrix was used because the units of measure between variables differed. Each regression tree was run with a 10-fold cross validation.

Cluster analyses were used to study the relationship between oomycete counts and turbidity because turbidity was the variable with the greatest correlation to oomycete counts but was not a branch in the RE-EM regression tree analysis. The cluster analyses were computed for each irrigation source separately because source and oomycete counts were correlated. All cluster analyses (k-mean method) were performed using JMP Pro (version 11) software.

Nucleotide sequence accession numbers.

One representative DNA sequence from each named species found in this study has been deposited in the GenBank database, and these can be found under accession numbers KJ855323, KJ855327, KJ865226, and KJ865241, respectively.

RESULTS

Prevalence of E. coli, Salmonella spp., and oomycetes.

Across all E. coli-positive samples (n = 123), 33% were positive. Samples from 18 sites were assayed for E. coli, and those from 16 of these sites were positive at least once during the survey (Table 2). The two sites where E. coli was not detected during the survey were the canal site (site B) and one pond in Tompkins County (site W). For each water source, the prevalences of E. coli-positive samples were 34%, 32%, and 0%, in ponds, creeks, and the canal, respectively (Table 2). The counts in Escherichia coli-positive samples ranged from 1 to >300 CFU/100 ml. Three samples from two different pond sites (sites H and P) contained E. coli at levels of >300 CFU/100 ml.

TABLE 2.

Frequency of samples and sites positive for E. coli, Salmonella, and oomycetes in irrigation water sources

Organism and site	Frequencya
	Overall	Pond	Creek	Canal
E. coli	40 (123)	34 (99)	6 (19)	0 (5)
Sites positive for E. coli	16 (18)	13 (14)	3 (3)	0 (1)
Salmonella	53 (123)	46 (99)	5 (19)	2 (5)
Sites positive for Salmonella	17 (18)	13 (14)	3 (3)	1 (1)
Oomycetes	184 (210)	131 (155)	47 (47)	6 (8)
Sites positive for oomycetes	38 (38)	24 (24)	13 (13)	1 (1)

^aFrequency data are shown as the number of samples or sites positive (total number of samples or sites).

Across all Salmonella-positive samples (n = 123), 43% were positive (≥1 CFU/100 ml). Samples from 18 sites were assayed for Salmonella, and those from 17 of the sites were positive at least once during the survey (Table 2). The single site where Salmonella was not detected during the survey was a pond site (site W) in Tompkins County. For each water source, the prevalences of Salmonella-positive samples were 46%, 26%, and 40% in ponds, creeks, and the canal, respectively (Table 2). The prevalence for Salmonella was equal to or higher than that observed for E. coli in each water source, with Salmonella counts more frequently exceeding 300 CFU/100 ml. The counts in Salmonella-positive samples ranged from 1 to >300 CFU/100 ml, with 23 samples from 14 different sites (sites B, H, G, J, M to S, U, V, and Z) producing Salmonella counts of >300 CFU/100 ml.

Across all oomycete samples (n = 210), 88% were positive (≥1 CFU/liter), and one or more samples from all sites (n = 38) assayed for HIS Phytophthora and Pythium during the survey were positive for oomycetes (Table 2). A total of 1,093 oomycetes were isolated, including 10 species of Phytophthora and 11 species of Pythium that were identified during the survey (see Table S4 in the supplemental material). The most prevalent oomycetes (present in ≥10% of samples) were Phytophthora lacustris, Phytophthora hydropathica, Phytophthora irrigata, and Pythium litorale. Many isolates (n = 150) of both Phytophthora and Pythium could not be identified to the species level because their ITS DNA sequences did not clearly align with sequences of known species available in GenBank or the Phytophthora Database. The frequency of oomycetes ranged from 0 to 200 CFU/liter of irrigation water. One hundred percent of samples from creeks were positive, while 85% of samples from ponds and 75% of samples from the canal were positive.

Longitudinal study of E. coli.

Regression trees were computed for E. coli-positive samples using the variables with the highest correlation for E. coli-positive samples, including the presence of Salmonella, irrigation source, and precipitation 3 days prior to sampling. The tree for E. coli-positive samples resulted in a root node only, suggesting that the variables used did not have a strong predictive power for distinguishing between positive and negative samples. A regression tree was also computed for E. coli counts (numbers of CFU/100 ml) using the variables for Salmonella counts and precipitation 3 days prior to sampling. The tree for E. coli counts resulted in a root node only, suggesting that the variables used did not have strong predictive power for characterizing the number of E. coli CFU/100 ml. No parameters in the study were strongly correlated to the presence or levels of E. coli in the surface water irrigation sources surveyed.

Longitudinal study of Salmonella spp.

A regression tree for distinguishing Salmonella-positive samples using variables for the presence of E. coli, irrigation source, and all precipitation variables resulted in a single-node tree with precipitation of ≥0.64 cm (cumulative total 3 days prior to sampling) as the split. Fifty-one percent (n = 88) of samples from sites with less than 0.64 cm precipitation were positive, while only 23% (n = 35) of samples from sites with greater than 0.64 cm precipitation were positive. A regression tree for Salmonella counts using the variables for precipitation (3 days prior to sampling) and E. coli counts showed samples from sites with 0.38 to 0.64 cm precipitation were associated with the highest Salmonella counts (average, 175 CFU/100 ml). Samples from sites receiving 0 to 0.38 cm of precipitation had average counts of 68 CFU/100 ml, while samples from sites receiving more than 0.64 cm of precipitation had average Salmonella counts of 12 CFU/100 ml. No splits occurred with E. coli counts as a predictor variable, suggesting that the levels of the two bacteria were not strongly correlated.

Longitudinal study of HIS oomycetes.

A regression tree was computed for oomycete-positive samples using irrigation source as the predictive variable. The tree resulted in a root node only, suggesting that irrigation source did not have a strong predictive power for distinguishing positive and negative samples. A regression tree was also computed for oomycete counts using irrigation source and turbidity as predictive variables. The tree for oomycete counts had one split for irrigation source, showing that the canal and creek samples (n = 54) had, on average, higher oomycete counts, 19 CFU/liter, than pond samples (n = 156), which had an average oomycete count of 12 CFU/liter.

Cluster analyses were performed for oomycete counts and turbidity for creek and pond samples separately. A cluster analysis was not done for the canal site because of the small sample size. For creek samples, the majority (18/25) belonged to a cluster with an average turbidity and an average oomycete count of 2.4 nephelometric turbidity units (NTU) and 29 CFU/liter, respectively. The remaining samples (7/25) were in a cluster characterized by samples with a higher average turbidity and a higher average oomycete count of 4.0 NTU and 73 CFU/liter, respectively. For pond samples, the majority (85/90) were in a cluster characterized by an average turbidity and an average oomycete count of 5.7 NTU and 18 CFU/liter, respectively. The remaining samples (5/85) were characterized by a higher average turbidity level and a higher average oomycete count of 63.9 NTU and 78 CFU/liter, respectively.

DISCUSSION

Over the course of this study, E. coli, Salmonella spp., and HIS oomycetes were identified in surface water irrigation sources in New York State. The presence and levels of E. coli in surface water irrigation sources were not closely associated with any of the parameters tested in this study. Contaminated runoff from precipitation events and livestock in the proximity are considered likely sources for E. coli contamination of surface waters, but this study did not find a strong link between those or any other parameters tested and E. coli prevalence. Approximately 1/3 of all samples were positive for generic E. coli, showing that the risk of introducing E. coli into the produce preharvest environment exists when surface water is used for irrigation. Fruit and vegetable growers are concerned with high levels of E. coli in their irrigation water because it is associated with fecal contamination and a greater risk of introducing pathogenic bacteria, which could lead to food-borne illness in association with their produce. Three samples from this study had E. coli levels above the regulatory threshold of >235 CFU/100 ml, but no sites had a mean of greater than 126 CFU/100 ml over 5 samples. Further investigation is necessary to understand why and when a surface water sample could have E. coli levels that exceed regulatory thresholds, as this information would be very valuable to any grower using surface water in the produce production environment.

Salmonella-positive samples and high population levels were most correlated with periods of rainfall of less than 0.64 cm (3 days before sampling), when growers are more likely to use water for irrigation. The lowest levels of Salmonella were associated with heavy rainfall amounts (>0.64 cm). Inconsistencies with the correlation of Salmonella levels and precipitation have been reported previously (3, 22). Other factors, such as soil type, could influence the association of precipitation and Salmonella levels in irrigation sources (3). The presence of Salmonella was not strongly correlated with the presence of E. coli, suggesting that testing for E. coli would not provide information about Salmonella levels in irrigation sources. Currently, irrigation water is not routinely tested for Salmonella. In this study, Salmonella was more prevalent than E. coli in surface irrigation water, with more samples having counts exceeding >300 CFU/100 ml. Other studies have also found Salmonella to be highly prevalent in surface water (3). Major sources of water contamination by Salmonella and E. coli are similar: wildlife, livestock, and humans (23–25). The greater prevalence and concentration of Salmonella compared to those of E. coli may be due to the level of persistence of Salmonella in surface water sources compared with that of other potential pathogenic enteric bacteria, as has been previously shown (26). The medium used for the selective isolation of E. coli or Salmonella spp. has the potential to grow nontarget species that can be morphologically similar to the target species. The results for a representative sample of positive colonies from the selective medium used were confirmed by PCR, but the possibility of false-positive results does exist for E. coli and Salmonella counts.

All samples (n = 6) from one pond (site W) in Tompkins County were found to be free of E. coli and Salmonella spp. This result could be due to sampling frequency (once a month), and additional sampling could determine if this site is consistently free of E. coli and Salmonella. A unique feature of the pond at site W is that it is the only embankment-style pond tested in this study, and it is located at a higher elevation than the surrounding cultivated land. All of the other ponds sampled in this study were excavation-style ponds. Embankment ponds are built on sloped terrain, where a high embankment is constructed at the end of the pond with the lowest elevation to trap water. Excavation ponds are built on level ground, usually at the lowest elevation on the farm, where groundwater is likely to be close to the surface. Pond style could play a role in the presence of organisms such as E. coli and Salmonella and should be further investigated.

Oomycetes were highly prevalent in all water sources, indicating that using any surface water source for irrigation or other agricultural application runs a high risk for introducing plant-pathogenic oomycetes into a produce-growing environment. Several other studies have found high oomycete prevalences in surface waters (27). We found higher levels of oomycetes in irrigation water sources from creeks than in those from ponds. The canal site had levels of oomycetes similar to those observed in creeks, but since only one canal site was sampled in this study, further investigation is necessary to describe oomycete prevalence in canals used for irrigation purposes. Irrigation ponds are primarily fed from groundwater and drainage tiles, while creeks are primarily fed from surface runoff; this could play a role in the difference in oomycete counts between ponds and creeks. Surface runoff is likely to contain oomycete-infested water and debris that can enter creeks anywhere along the banks. Water entering a pond through groundwater or a drainage tile passes through layers of soil, where some of the oomycetes may be trapped. Turbidity levels were found to be positively associated with oomycete levels in the surface waters sampled in this study. Greater turbidity may be associated with runoff water or debris containing larger amounts of oomycetes that enter a surface water source and could explain the positive correlation. In addition, turbidity could play a role in the persistence of oomycetes in surface water. Components of turbidity, such as clay, silt, organic matter, and microorganisms, could play a role in survival for some oomycetes and protect them from UV light exposure as well.

The most prevalent HIS oomycetes (present in ≥10% of samples) isolated in this survey were Ph. lacustris, Ph. hydropathica, Ph. irrigata, and Py. litorale. Each of these plant pathogens has previously been found in surface water sources and is not considered a major threat to fruit and vegetable growers (28–30). While Ph. irrigata and Py. litorale have been show to produce disease on some vegetable crops in greenhouse studies, no disease in plants grown in the field has been attributed to these species (28, 29). Other Phytophthora spp. isolated during this survey included Phytophthora citricola, Phytophthora cryptogea, Phytophthora gonapodyides, Phytophthora sansomeana, Phytophthora capsici, and Phytophthora nicotianae. These species have previously been isolated from surface water sources and are known to cause disease on some fruit and vegetables crops (13, 31, 32). Phytophthora gallica has not been previously reported from surface water and is pathogenic on some alder and beech trees but is not known to infect cultivated fruit and vegetable crops (33). Many of the Pythium spp. isolated during this survey, including Pythium helicoides, Pythium catenulatum, Pythium marisipium, Pythium myriotylum, Pythium irregulare, Pythium vexans, and Pythium adhaerens, have previously been reported from surface water sources (34–37). Four Pythium spp. isolated in this water survey, Pythium amasculinum, Pythium mercuriale, Pythium adhaerens, and Pythium oedochilum, have not been previously reported from surface water sources. Each Pythium species found in this survey has previously been isolated from a diseased crop plant (38–40).

The results from this study show that the prevalence of generic E. coli, Salmonella, and HIS plant-pathogenic oomycetes in surface water in New York State is high. Using surface water for irrigation puts a grower at risk for introducing potential human and plant pathogens into the growing environment. Water quality parameters, irrigation site properties, and precipitation data may be useful in helping predict the prevalence of potential pathogen contamination of surface water used for irrigation. Further studies are necessary to better understand the specific factors that influence pathogen contamination and persistence in surface water, as this information could be used to reduce the occurrences of food-borne illness and plant disease.