Complex Interactions Between Weather, and Microbial and Physicochemical Water Quality Impact the Likelihood of Detecting Foodborne Pathogens in Agricultural Water

Prevalence of STEC virulence markers and Salmonella as a function of abiotic factors in agricultural water in the southeastern United States

Fresh produce can be contaminated by enteric pathogens throughout crop production, including through contact with contaminated agricultural water. The most common outbreaks and recalls in fresh produce are due to contamination by Salmonella enterica and Shiga toxin-producing E. coli (STEC). Thus, the objectives of this study were to investigate the prevalence of markers for STEC (wzy, hly, fliC, eaeA, rfbE, stx-I, stx-II) and Salmonella (invA) in surface water sources (n = 8) from produce farms in Southwest Georgia and to determine correlations among the prevalence of virulence markers for STEC, water nutrient profile, and environmental factors. Water samples (500 mL) from eight irrigation ponds were collected from February to December 2021 (n = 88). Polymerase chain reaction (PCR) was used to screen for Salmonella and STEC genes, and Salmonella samples were confirmed by culture-based methods. Positive samples for Salmonella were further serotyped. Particularly, Salmonella was detected in 6/88 (6.81%) water samples from all ponds, and the following 4 serotypes were detected: Saintpaul 3/6 (50%), Montevideo 1/6 (16.66%), Mississippi 1/6 (16.66%), and Bareilly 1/6 (16.66%). Salmonella isolates were only found in the summer months (May-Aug.). The most prevalent STEC genes were hly 77/88 (87.50%) and stx-I 75/88 (85.22%), followed by fliC 54/88 (61.63%), stx-II 41/88 (46.59%), rfbE 31/88 (35.22%), and eaeA 28/88 (31.81%). The wzy gene was not detected in any of the samples. Based on a logistic regression analysis, the odds of codetection for STEC virulence markers (stx-I, stx-II, and eaeA) were negatively correlated with calcium and relative humidity (p < 0.05). A conditional forest analysis was performed to assess predictive performance (AUC = 0.921), and the top predictors included humidity, nitrate, calcium, and solar radiation. Overall, information from this research adds to a growing body of knowledge regarding the risk that surface water sources pose to produce grown in subtropical environmental conditions and emphasizes the importance of understanding the use of abiotic factors as a holistic approach to understanding the microbial quality of water.

Methodological differences between studies confound one-size-fits-all approaches to managing surface waterways for food and water safety

Even though differences in methodology (e.g., sample volume and detection method) have been shown to affect observed microbial water quality, multiple sampling and laboratory protocols continue to be used for water quality monitoring. Research is needed to determine how these differences impact the comparability of findings to generate best management practices and the ability to perform meta-analyses. This study addresses this knowledge gap by compiling and analyzing a data set representing 2,429,990 unique data points on at least one microbial water quality target (e.g., Salmonella presence and Escherichia coli concentration). Variance partitioning analysis was used to quantify the variance in likelihood of detecting each pathogenic target that was uniquely and jointly attributable to non-methodological versus methodological factors. The strength of the association between microbial water quality and select methodological and non-methodological factors was quantified using conditional forest and regression analysis. Fecal indicator bacteria concentrations were more strongly associated with non-methodological factors than methodological factors based on conditional forest analysis. Variance partitioning analysis could not disentangle non-methodological and methodological signals for pathogenic Escherichia coli, Salmonella, and Listeria. This suggests our current perceptions of foodborne pathogen ecology in water systems are confounded by methodological differences between studies. For example, 31% of total variance in likelihood of Salmonella detection was explained by methodological and/or non-methodological factors, 18% was jointly attributable to both methodological and non-methodological factors. Only 13% of total variance was uniquely attributable to non-methodological factors for Salmonella, highlighting the need for standardization of methods for microbiological water quality testing for comparison across studies.IMPORTANCEThe microbial ecology of water is already complex, without the added complications of methodological differences between studies. This study highlights the difficulty in comparing water quality data from projects that used different sampling or laboratory methods. These findings have direct implications for end users as there is no clear way to generalize findings in order to characterize broad-scale ecological phenomenon and develop science-based guidance. To best support development of risk assessments and guidance for monitoring and managing waters, data collection and methods need to be standardized across studies. A minimum set of data attributes that all studies should collect and report in a standardized way is needed. Given the diversity of methods used within applied and environmental microbiology, similar studies are needed for other microbiology subfields to ensure that guidance and policy are based on a robust interpretation of the literature.

Scale and detection method impacted Salmonella prevalence and diversity in ponds

Site-specific approaches for managing food safety hazards in agricultural water require an understanding of foodborne pathogen ecology. This study identified factors associated with Salmonella contamination in Virginia ponds. Grab samples (250 mL, N = 600) were collected from 30 sites across nine ponds. Culture- and culture-independent (CIDT)-based methods were used to detect Salmonella in each sample. Salmonella isolated by culture-based methods were serotyped by Kauffman-White classification. Environmental data were collected for each sample. McNemar's χ2 was used to determine if Salmonella detection differed by testing method. Separate mixed effect models were used to identify environmental factors associated with culture and CIDT-based Salmonella detection. Separate models were built for each pond, and for all ponds combined. Salmonella detection differed significantly (p < 0.001) between CIDT (31 %; 183/600)- and culture (13 %; 77/600)-based methods. Culture-based methods yielded 11 different serovars. All cultured Salmonella samples were confirmed by CIDT; 42.1 % of CIDT Salmonella-positive samples could be cultured. Associations between environmental factors and Salmonella detection also varied substantially by pond and detection method. In the all-pond model, associations were observed for five factors (total coliforms, Escherichia coli, air temperature, UV, rain) for both culture- and CIDT-based Salmonella detection. Rain prior to sampling (24 h) increased odds of Salmonella detection for culture (OR = 5.09) and CIDT (OR = 3.62) in the all-pond model. When all the pond data were used, models masked associations at the individual pond level, as there were noticeable differences between ponds and the odds of isolating Salmonella by environmental factors. Ponds were within a 187-ha area in this study, emphasizing water management needs to be individualized (i.e., assess hazards/risks by pond). Results also highlight detection methods and scale strongly affect observed water quality and should be considered when developing monitoring programs to develop guidance for growers.

Oversampling methods for machine learning model data training to improve model capabilities to predict the presence of Escherichia coli MG1655 in spinach wash water

We assessed the efficacy of oversampling techniques to enhance machine learning model performance in predicting Escherichia coli MG1655 presence in spinach wash water. Three oversampling methods were applied to balance two datasets, forming the basis for training random forest (RF), support vector machines (SVMs), and binomial logistic regression (BLR) models. Data underwent method-specific centering and standardization, with outliers replaced by feature-specific means in training datasets. Testing occurred without these preprocessing steps. Model hyperparameters were optimized using a subset of testing data via 10-fold cross-validation. Models were trained on full datasets and tested on newly acquired spinach wash water samples. Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic Sampling approach (ADASYN) achieved strong results, with SMOTE RF reaching an accuracy of 90.0%, sensitivity of 93.8%, specificity of 87.5%, and an area under the curve (AUC) of 98.2% (without data preprocessing) and ADASYN achieving 86.55% accuracy, 87.5% sensitivity, 83.3% specificity, and a 92.4% AUC. SMOTE and ADASYN significantly improved (p < 0.05) SVM and RF models, compared to their non-oversampled counterparts without preprocessing. Data preprocessing had a mixed impact, improving (p < 0.05) the accuracy and specificity of the BLR model but decreasing the accuracy and specificity (p < 0.05) of the SVM and RF models. The most influential physiochemical feature for E. coli detection in wash water was water conductivity, ranging from 7.9 to 196.2 µS. Following closely was water turbidity, ranging from 2.97 to 72.35 NTU within this study.

Mitigating risks and maximizing sustainability of treated wastewater reuse for irrigation

Scarcity of freshwater for agriculture has led to increased utilization of treated wastewater (TWW), establishing it as a significant and reliable source of irrigation water. However, years of research indicate that if not managed adequately, TWW may deleteriously affect soil functioning and plant productivity, and pose a hazard to human and environmental health. This review leverages the experience of researchers, stakeholders, and policymakers from Israel, the United-States, and Europe to present a holistic, multidisciplinary perspective on maximizing the benefits from municipal TWW use for irrigation. We specifically draw on the extensive knowledge gained in Israel, a world leader in agricultural TWW implementation. The first two sections of the work set the foundation for understanding current challenges involved with the use of TWW, detailing known and emerging agronomic and environmental issues (such as salinity and phytotoxicity) and public health risks (such as contaminants of emerging concern and pathogens). The work then presents solutions to address these challenges, including technological and agronomic management-based solutions as well as source control policies. The concluding section presents suggestions for the path forward, emphasizing the importance of improving links between research and policy, and better outreach to the public and agricultural practitioners. We use this platform as a call for action, to form a global harmonized data system that will centralize scientific findings on agronomic, environmental and public health effects of TWW irrigation. Insights from such global collaboration will help to mitigate risks, and facilitate more sustainable use of TWW for food production in the future.

Population dynamics of Listeria spp., Salmonella spp., and Escherichia coli on fresh produce: A scoping review

Collation of the current scope of literature related to population dynamics (i.e., growth, die-off, survival) of foodborne pathogens on fresh produce can aid in informing future research directions and help stakeholders identify relevant research literature. A scoping review was conducted to gather and synthesize literature that investigates population dynamics of pathogenic and non-pathogenic Listeria spp., Salmonella spp., and Escherichia coli on whole unprocessed fresh produce (defined as produce not having undergone chopping, cutting, homogenization, irradiation, or pasteurization). Literature sources were identified using an exhaustive search of research and industry reports published prior to September 23, 2021, followed by screening for relevance based on strict, a priori eligibility criteria. A total of 277 studies that met all eligibility criteria were subjected to an in-depth qualitative review of various factors (e.g., produce commodities, study settings, inoculation methodologies) that affect population dynamics. Included studies represent investigations of population dynamics on produce before (i.e., pre-harvest; n = 143) and after (i.e., post-harvest; n = 144) harvest. Several knowledge gaps were identified, including the limited representation of (i) pre-harvest studies that investigated population dynamics of Listeria spp. on produce (n = 13, 9% of pre-harvest studies), (ii) pre-harvest studies that were carried out on non-sprouts produce types grown using hydroponic cultivation practices (n = 7, 5% of pre-harvest studies), and (iii) post-harvest studies that reported the relative humidity conditions under which experiments were carried out (n = 56, 39% of post-harvest studies). These and other knowledge gaps summarized in this scoping review represent areas of research that can be investigated in future studies.

Efficacy of Peracetic Acid and Chlorine on the Reduction of Shiga Toxin-producing Escherichia coli and a Nonpathogenic E. coli Strain in Preharvest Agricultural Water

Produce-borne outbreaks of Shiga toxin-producing Escherichia coli (STEC) linked to preharvest water emphasize the need for efficacious water treatment options. This study quantified reductions of STEC and generic E. coli in preharvest agricultural water using commercially available sanitizers. Water was collected from two sources in Virginia (pond, river) and inoculated with either a seven-strain STEC panel or environmental generic E. coli strain TVS 353 (∼9 log10 CFU/100 mL). Triplicate inoculated water samples were equilibrated to 12 or 32°C and treated with peracetic acid (PAA) or chlorine (Cl) [low (PAA:6ppm, Cl:2-4 ppm) or high (PAA:10 ppm, Cl:10-12 ppm) residual concentrations] for an allotted contact time (1, 5, or 10 min). Strains were enumerated, and a log-linear model was used to characterize how treatment combinations influenced reductions. All Cl treatment combinations achieved a ≥3 log10 CFU/100 mL reduction, regardless of strain (3.43 ± 0.25 to 7.05 ± 0.00 log10 CFU/100 mL). Approximately 80% (19/24) and 67% (16/24) of PAA treatment combinations achieved a ≥3 log10 CFU/100 mL for STEC and E. coli TVS 353, respectively. The log-linear model showed contact time (10 > 5 > 1 min) and sanitizer type (Cl > PAA) had the greatest impact on STEC and E. coli TVS 353 reductions (p < 0.001). E. coli TVS 353 in water samples was more resistant to sanitizer treatment (p < 0.001) indicating applicability as a good surrogate. Results demonstrated Cl and PAA can be effective agricultural water treatment strategies when sanitizer chemistry is managed. These data will assist with the development of in-field validation studies and may identify suitable candidates for the registration of antimicrobial pesticide products for use against foodborne pathogens in preharvest agricultural water treatment.

Impact of Acidulants on Salmonella and Escherichia coli O157:H7 in Water Microcosms Containing Organic Matter

As demands for fresh water become more competitive between the processing plant and other consumers of water such as municipalities, interest has grown in recycling or reusing water for food processing. However, recycling the processing water from a poultry plant, for example, represents challenges due to increased organic loads and the presence of bacterial contaminants including foodborne pathogens. The objective in the current study was to evaluate the inactivation of Salmonella and E. coli O157:H7 using combinations (0.5% and 1%) of sodium bisulfate (SBS) and 1% lactic acid (LA) in water and water with organic matter in the form of horse blood serum (0.3%) with exposure times of 1 min and 5 min. Pathogen reductions after a 5 min exposure time were greater than corresponding reductions after a 1 min exposure time for all acid solutions. The Salmonella counts were significantly reduced (i.e., ≥1 log-unit) in all acid solutions after a 5 min exposure time with the combination of LA + SBS acid solutions being more effective than the corresponding 2% LA solutions. None of the acid solutions were effective in reducing the E. coli O157:H7 after a 1 min exposure time. The 1% LA + 1% SBS solution was the most effective acid solution against both pathogens and was the only acid solution effective in reducing E. coli O157:H7 by at least one log unit after 5 min of exposure.

Scale of analysis drives the observed ratio of spatial to non-spatial variance in microbial water quality: insights from two decades of citizen science data

AIMS

While fecal indicator bacteria (FIB) testing is used to monitor surface water for potential health hazards, observed variation in FIB levels may depend on the scale of analysis (SOA). Two decades of citizen science data, coupled with random effects models, were used to quantify the variance in FIB levels attributable to spatial versus temporal factors.

METHODS AND RESULTS

Separately, Bayesian models were used to quantify the ratio of spatial to non-spatial variance in FIB levels and identify associations between environmental factors and FIB levels. Separate analyses were performed for three SOA: waterway, watershed, and statewide. As SOA increased (from waterway to watershed to statewide models), variance attributable to spatial sources generally increased and variance attributable to temporal sources generally decreased. While relationships between FIB levels and environmental factors, such as flow conditions (base versus stormflow), were constant across SOA, the effect of land cover was highly dependent on SOA and consistently smaller than the effect of stormwater infrastructure (e.g. outfalls).

CONCLUSIONS

This study demonstrates the importance of SOA when developing water quality monitoring programs or designing future studies to inform water management.

Developing a Decision-making Tool for Agricultural Surface Water Decontamination Using Ultraviolet-C Light

Ultraviolet-C (UV-C) light-assisted water treatment systems are an increasingly investigated alternative to chemical sanitizers for agricultural surface water decontamination. However, the relatively high concentration of particulate matter in surface water is a major challenge to expanding its application in the production of fresh produce. The objective of this project was to test the efficacy of two commercial UV-C devices to reduce the microbial risk of agricultural water in order to develop a web application to assist growers in decision-making related to the on-farm implementation of UV-C technologies for agricultural water treatment. An on-farm study using three agricultural water sources was performed to determine the microbial reduction efficacy of a low power, low flow (LP/LF; 1-9 gallons per minute (GPM), 1.34-gallon capacity) and a high powered, high flow (HP/HF; 1-110 GPM, 4.75-gallon capacity) device at flow rates of 6, 7, and 9 GPM. A threshold of 30% UVT for the HP/HF device was observed, wherein lower water transmissibility significantly impacted microbial inactivation. Although less effective at lower %UVT, the LP/LF device costs less to install, maintain, and operate. The observations were used to design an online tool for growers to calculate the predicted reduction of generic Escherichia coli using either device based on the %UVT of their water source. However, because this study utilized an exploratory and proof-of-concept approach, the experimental flow rates were limited to reflect the capacities of the smaller unit (9 GPM) for direct comparison to the larger unit. Thus, the preliminary model and tool are largely limited to the experimental conditions. Yet, these results of this study demonstrate the utility of UV-C light in reducing the microbial risk of agricultural water, and future studies using different UV-C devices and higher flow rates will expand the use of the decision-making tool.

Comparative genomics unveils extensive genomic variation between populations of Listeria species in natural and food-associated environments

Comprehending bacterial genomic variation linked to distinct environments can yield novel insights into mechanisms underlying differential adaptation and transmission of microbes across environments. Gaining such insights is particularly crucial for pathogens as it benefits public health surveillance. However, the understanding of bacterial genomic variation is limited by a scarcity of investigations in genomic variation coupled with different ecological contexts. To address this limitation, we focused on Listeria, an important bacterial genus for food safety that includes the human pathogen L. monocytogenes, and analyzed a large-scale genomic dataset collected by us from natural and food-associated environments across the United States. Through comparative genomics analyses on 449 isolates from the soil and 390 isolates from agricultural water and produce processing facilities representing L. monocytogenes, L. seeligeri, L. innocua, and L. welshimeri, we find that the genomic profiles strongly differ by environments within each species. This is supported by the environment-associated subclades and differential presence of plasmids, stress islands, and accessory genes involved in cell envelope biogenesis and carbohydrate transport and metabolism. Core genomes of Listeria species are also strongly associated with environments and can accurately predict isolation sources at the lineage level in L. monocytogenes using machine learning. We find that the large environment-associated genomic variation in Listeria appears to be jointly driven by soil property, climate, land use, and accompanying bacterial species, chiefly representing Actinobacteria and Proteobacteria. Collectively, our data suggest that populations of Listeria species have genetically adapted to different environments, which may limit their transmission from natural to food-associated environments.

AI-enabled biosensing for rapid pathogen detection: From liquid food to agricultural water

Rapid pathogen detection in food and agricultural water is essential for ensuring food safety and public health. However, complex and noisy environmental background matrices delay the identification of pathogens and require highly trained personnel. Here, we present an AI-biosensing framework for accelerated and automated pathogen detection in various water samples, from liquid food to agricultural water. A deep learning model was used to identify and quantify target bacteria based on their microscopic patterns generated by specific interactions with bacteriophages. The model was trained on augmented datasets to maximize data efficiency, using input images of selected bacterial species, and then fine-tuned on a mixed culture. Model inference was performed on real-world water samples containing environmental noises unseen during model training. Overall, our AI model trained solely on lab-cultured bacteria achieved rapid (< 5.5 h) prediction with 80-100% accuracy on the real-world water samples, demonstrating its ability to generalize to unseen data. Our study highlights the potential applications in microbial water quality monitoring during food and agricultural processes.

Sanitizer Type and Contact Time Influence Salmonella Reductions in Preharvest Agricultural Water Used on Virginia Farms

No Environmental Protection Agency (EPA) chemical treatments for preharvest agricultural water are currently labeled to reduce human health pathogens. The goal of this study was to examine the efficacy of peracetic acid- (PAA) and chlorine (Cl)-based sanitizers against Salmonella in Virginia irrigation water. Water samples (100 mL) were collected at three time points during the growing season (May, July, September) and inoculated with either the 7-strain EPA/FDA-prescribed cocktail or a 5-strain Salmonella produce-borne outbreak cocktail. Experiments were conducted in triplicate for 288 unique combinations of time point, residual sanitizer concentration (low: PAA, 6 ppm; Cl, 2-4 ppm or high: PAA, 10 ppm; Cl, 10-12 ppm), water type (pond, river), water temperature (12°C, 32°C), and contact time (1, 5, 10 min). Salmonella were enumerated after each treatment combination and reductions were calculated. A log-linear model was used to characterize how treatment combinations influenced Salmonella reductions. Salmonella reductions by PAA and Cl ranged from 0.0 ± 0.1 to 5.6 ± 1.3 log10 CFU/100 mL and 2.1 ± 0.2 to 7.1 ± 0.2 log10 CFU/100 mL, respectively. Physicochemical parameters significantly varied by untreated water type; however, Salmonella reductions did not (p = 0.14), likely due to adjusting the sanitizer amounts needed to achieve the target residual concentrations regardless of source water quality. Significant differences (p < 0.05) in Salmonella reductions were observed for treatment combinations, with sanitizer (Cl > PAA) and contact time (10 > 5 > 1 min) having the greatest effects. The log-linear model also revealed that outbreak strains were more treatment-resistant. Results demonstrate that certain treatment combinations with PAA- and Cl-based sanitizers were effective at reducing Salmonella populations in preharvest agricultural water. Awareness and monitoring of water quality parameters are essential for ensuring adequate dosing for the effective treatment of preharvest agricultural water.

Waterborne Infectious Diseases Associated with Exposure to Tropical Cyclonic Storms, United States, 1996-2018

In the United States, tropical cyclones cause destructive flooding that can lead to adverse health outcomes. Storm-driven flooding contaminates environmental, recreational, and drinking water sources, but few studies have examined effects on specific infections over time. We used 23 years of exposure and case data to assess the effects of tropical cyclones on 6 waterborne diseases in a conditional quasi-Poisson model. We separately defined storm exposure for windspeed, rainfall, and proximity to the storm track. Exposure to storm-related rainfall was associated with a 48% (95% CI 27%-69%) increase in Shiga toxin-producing Escherichia coli infections 1 week after storms and a 42% (95% CI 22%-62%) in increase Legionnaires' disease 2 weeks after storms. Cryptosporidiosis cases increased 52% (95% CI 42%-62%) during storm weeks but declined over ensuing weeks. Cyclones are a risk to public health that will likely become more serious with climate change and aging water infrastructure systems.

Conditional Forest Models Built Using Metagenomic Data Accurately Predicted Salmonella Contamination in Northeastern Streams

The use of water contaminated with Salmonella for produce production contributes to foodborne disease burden. To reduce human health risks, there is a need for novel, targeted approaches for assessing the pathogen status of agricultural water. We investigated the utility of water microbiome data for predicting Salmonella contamination of streams used to source water for produce production. Grab samples were collected from 60 New York streams in 2018 and tested for Salmonella. Separately, DNA was extracted from the samples and used for Illumina shotgun metagenomic sequencing. Reads were trimmed and used to assign taxonomy with Kraken2. Conditional forest (CF), regularized random forest (RRF), and support vector machine (SVM) models were implemented to predict Salmonella contamination. Model performance was assessed using 10-fold cross-validation repeated 10 times to quantify area under the curve (AUC) and Kappa score. CF models outperformed the other two algorithms based on AUC (0.86, CF; 0.81, RRF; 0.65, SVM) and Kappa score (0.53, CF; 0.41, RRF; 0.12, SVM). The taxa that were most informative for accurately predicting Salmonella contamination based on CF were compared to taxa identified by ALDEx2 as being differentially abundant between Salmonella-positive and -negative samples. CF and differential abundance tests both identified Aeromonas salmonicida (variable importance [VI] = 0.012) and Aeromonas sp. strain CA23 (VI = 0.025) as the two most informative taxa for predicting Salmonella contamination. Our findings suggest that microbiome-based models may provide an alternative to or complement existing water monitoring strategies. Similarly, the informative taxa identified in this study warrant further investigation as potential indicators of Salmonella contamination of agricultural water. IMPORTANCE Understanding the associations between surface water microbiome composition and the presence of foodborne pathogens, such as Salmonella, can facilitate the identification of novel indicators of Salmonella contamination. This study assessed the utility of microbiome data and three machine learning algorithms for predicting Salmonella contamination of Northeastern streams. The research reported here both expanded the knowledge on the microbiome composition of surface waters and identified putative novel indicators (i.e., Aeromonas species) for Salmonella in Northeastern streams. These putative indicators warrant further research to assess whether they are consistent indicators of Salmonella contamination across regions, waterways, and years not represented in the data set used in this study. Validated indicators identified using microbiome data may be used as targets in the development of rapid (e.g., PCR-based) detection assays for the assessment of microbial safety of agricultural surface waters.

Spatial Versus Nonspatial Variance in Fecal Indicator Bacteria Differs Within and Between Ponds

Surface water environments are inherently heterogenous, and little is known about variation in microbial water quality between locations. This study sought to understand how microbial water quality differs within and between Virginia ponds. Grab samples were collected twice per week from 30 sampling sites across nine Virginia ponds (n = 600). Samples (100 mL) were enumerated for total coliform (TC) and Escherichia coli (EC) levels, and physicochemical, weather, and environmental data were collected. Bayesian models of coregionalization were used to quantify the variance in TC and EC levels attributable to spatial (e.g., site, pond) versus nonspatial (e.g., date, pH) sources. Mixed-effects Bayesian regressions and conditional inference trees were used to characterize relationships between data and TC or EC levels. Analyses were performed separately for each pond with ≥3 sampling sites (5 intrapond) while one interpond model was developed using data from all sampling sites and all ponds. More variance in TC levels were attributable to spatial opposed to nonspatial sources for the interpond model (variance ratio [VR] = 1.55) while intrapond models were pond dependent (VR: 0.65-18.89). For EC levels, more variance was attributable to spatial sources in the interpond model (VR = 1.62), compared to all intrapond models (VR < 1.0) suggesting that more variance is attributable to nonspatial factors within individual ponds and spatial factors when multiple ponds are considered. Within each pond, TC and EC levels were spatially independent for sites 56-87 m apart, indicating that different sites within the same pond represent different water quality for risk management. Rainfall was positively and pH negatively associated with TC and EC levels in both inter- and intrapond models. For all other factors, the direction and strength of associations varied. Factors driving microbial dynamics in ponds appear to be pond-specific and differ depending on the spatial scale considered.

Salmonella Prevalence Is Strongly Associated with Spatial Factors while Listeria monocytogenes Prevalence Is Strongly Associated with Temporal Factors on Virginia Produce Farms

The heterogeneity of produce production environments complicates the development of universal strategies for managing preharvest produce safety risks. Understanding pathogen ecology in different produce-growing regions is important for developing targeted mitigation strategies. This study aimed to identify environmental and spatiotemporal factors associated with isolating Salmonella and Listeria from environmental samples collected from 10 Virginia produce farms. Soil (n = 400), drag swab (n = 400), and irrigation water (n = 120) samples were tested for Salmonella and Listeria, and results were confirmed by PCR. Salmonella serovar and Listeria species were identified by the Kauffmann-White-Le Minor scheme and partial sigB sequencing, respectively. Conditional forest analysis and Bayesian mixed models were used to characterize associations between environmental factors and the likelihood of isolating Salmonella, Listeria monocytogenes (LM), and other targets (e.g., Listeria spp. and Salmonella enterica serovar Newport). Surrogate trees were used to visualize hierarchical associations identified by the forest analyses. Salmonella and LM prevalence was 5.3% (49/920) and 2.3% (21/920), respectively. The likelihood of isolating Salmonella was highest in water samples collected from the Eastern Shore of Virginia with a dew point of >9.4°C. The likelihood of isolating LM was highest in water samples collected in winter from sites where <36% of the land use within 122 m was forest wetland cover. Conditional forest results were consistent with the mixed models, which also found that the likelihood of detecting Salmonella and LM differed between sample type, region, and season. These findings identified factors that increased the likelihood of isolating Salmonella- and LM-positive samples in produce production environments and support preharvest mitigation strategies on a regional scale. IMPORTANCE This study sought to examine different growing regions across the state of Virginia and to determine how factors associated with pathogen prevalence may differ between regions. Spatial and temporal data were modeled to identify factors associated with an increased pathogen likelihood in various on-farm sources. The findings of the study show that prevalence of Salmonella and L. monocytogenes is low overall in the produce preharvest environment but does vary by space (e.g., region in Virginia) and time (e.g., season), and the likelihood of pathogen-positive samples is influenced by different spatial and temporal factors. Therefore, the results support regional or scale-dependent food safety standards and guidance documents for controlling hazards to minimize risk. This study also suggests that water source assessments are important tools for developing monitoring programs and mitigation measures, as spatiotemporal factors differ on a regional scale.

Soil Collected from a Single Great Smoky Mountains Trail Contains a Diversity of Listeria monocytogenes and Listeria spp

Listeria monocytogenes, a foodborne pathogen, and other Listeria spp. are present in natural environments. Isolating and characterizing strains from natural reservoirs can provide insight into the prevalence and diversity of Listeria spp. in these environments, elucidate their contribution to contamination of agricultural and food processing environments and food products, and lead to the discovery of novel species. In this study, we evaluated the diversity of Listeria spp. isolated from soil in a small region of the Great Smoky Mountains National Park, the most biodiverse national park in the U.S. National Park system. Of the 17 Listeria isolates recovered, whole-genome sequencing revealed that 14 were distinct strains. The strains represented a diversity of Listeria species (L. monocytogenes [n = 9], L. cossartiae subsp. cossartiae [n = 1], L. marthii [n = 1], L. booriae [n = 1], and a potentially novel Listeria sp. [n = 2]), as well as a diversity of sequence types based on multilocus sequence typing (MLST) and core genome MLST, including many novel designations. The isolates were not closely related (≥99.99% average nucleotide identity) to any isolates in public databases (NCBI, PATRIC), which also indicated novelty. The Listeria samples isolated in this study were collected from high-elevation sites near a creek that ultimately leads to the Mississippi River; thus, Listeria present in this natural environment could potentially travel downstream to a large region that includes portions of nine southeastern and midwestern U.S. states. This study provides insight into the diversity of Listeria spp. in the Great Smoky Mountains and indicates that this environment is a reservoir of novel Listeria spp. IMPORTANCE Listeria monocytogenes is a foodborne pathogen that can cause serious systemic illness that, although rare, usually results in hospitalization and has a relatively high mortality rate compared to other foodborne pathogens. Identification of novel and diverse Listeria spp. can provide insights into the genomic evolution, ecology, and evolution and variance of pathogenicity of this genus, especially in natural environments. Comparing L. monocytogenes and Listeria spp. isolates from natural environments, such as those recovered in this study, to contamination and/or outbreak strains may provide more information about the original natural sources of these strains and the pathways and mechanisms that lead to contamination of food products and agricultural or food processing environments.

Structural Equation Models Suggest That On-Farm Noncrop Vegetation Removal Is Not Associated with Improved Food Safety Outcomes but Is Linked to Impaired Water Quality

While growers have reported pressures to minimize wildlife intrusion into produce fields through noncrop vegetation (NCV) removal, NCV provides key ecosystem services. To model food safety and environmental tradeoffs associated with NCV removal, published and publicly available food safety and water quality data from the Northeastern United States were obtained. Because data on NCV removal are not widely available, forest-wetland cover was used as a proxy, consistent with previous studies. Structural equation models (SEMs) were used to quantify the effect of forest-wetland cover on (i) food safety outcomes (e.g., detecting pathogens in soil) and (ii) water quality (e.g., nutrient levels). Based on the SEMs, NCV was not associated with or had a protective effect on food safety outcomes (more NCV was associated with a reduced likelihood of pathogen detection). The probabilities of detecting Listeria spp. in soil (effect estimate [EE] = -0.17; P = 0.005) and enterohemorrhagic Escherichia coli in stream samples (EE = -0.27; P < 0.001) were negatively associated with the amount of NCV surrounding the sampling site. Larger amounts of NCV were also associated with lower nutrient, salinity, and sediment levels, and higher dissolved oxygen levels. Total phosphorous levels were negatively associated with the amount of NCV in the upstream watershed (EE = -0.27; P < 0.001). Similar negative associations (P < 0.05) were observed for other physicochemical parameters, such as nitrate (EE = -0.38). Our findings suggest that NCV should not be considered an inherent produce safety risk or result in farm audit demerits. This study also provides a framework for evaluating environmental tradeoffs associated with using specific preharvest food safety strategies. IMPORTANCE Currently, on-farm food safety decisions are typically made independently of conservation considerations, often with detrimental impacts on agroecosystems. Comanaging agricultural environments to simultaneously meet conservation and food safety aims is complicated because farms are closely linked to surrounding environments, and management decisions can have unexpected environmental, economic, and food safety consequences. Thus, there is a need for research on the conservation and food safety tradeoffs associated with implementing specific preharvest food safety practices. Understanding these tradeoffs is critical for developing adaptive comanagement strategies and ensuring the short- and long-term safety, sustainability, and profitability of agricultural systems. This study quantifies tradeoffs and synergies between food safety and environmental aims, and outlines a framework for modeling tradeoffs and synergies between management aims that can be used to support future comanagement research.

What on earth? The impact of digestates and composts from farm effluent management on fluxes of foodborne pathogens in agricultural lands

The recycling of biomass is the cornerstone of sustainable development in the bioeconomy. In this context, digestates and composts from processed agricultural residues and biomasses are returned to the soil. Whether or not the presence of pathogenic microorganisms in these processed biomasses is a threat to the sustainability of the current on-farm practices is still the subject of debate. In this review, we describe the microbial pathogens that may be present in digestates and composts. We then provide an overview of the current European regulation designed to mitigate health hazards linked to the use of organic fertilisers and soil improvers produced from farm biomasses and residues. Finally, we discuss the many factors that underlie the fate of microbial pathogens in the field. We argue that incorporating land characteristics in the management of safety issues connected with the spreading of organic fertilisers and soil improvers can improve the sustainability of biomass recycling.

Environmental and anthropogenic factors associated with the likelihood of detecting Salmonella in agricultural watersheds

Surface water is one of the primary sources of irrigation water for produce production; therefore, its contamination by foodborne pathogens, such as Salmonella, may substantially impact public health. In this study, we determined the presence of Salmonella in surface water and characterized the relationship between Salmonella detection and environmental and anthropogenic factors. From April 2019 to February 2020, 120 samples from 30 sites were collected monthly in four watersheds located in two different central Chile agricultural regions (N = 1080). Water samples from rivers, canals, streams, and ponds linked to each watershed were obtained. Surface water (10 L) was filtrated in situ, and samples were analyzed for the presence of Salmonella. Salmonella was detected every month in all watersheds, with a mean detection percentage of 28% (0%-90%) across sampling sites, regardless of the season. Overall, similar detection percentages were observed for both regions: 29.1% for Metropolitan and 27.0% for Maule. Salmonella was most often detected in summer (39.8% of all summer samples tested positive) and least often in winter (14.4% of winter samples). Random forest analysis showed that season, water source, and month, followed by latitude and river, were the most influential factors associated with Salmonella detection. The influences of water pH and temperature (categorized as environmental factors) and factors associated with human activity (categorized as anthropogenic factors) registered at the sampling site were weakly or not associated with Salmonella detection. In conclusion, Salmonella was detected in surface water potentially used for irrigation, and its presence was linked to season and water source factors. Interventions are necessary to prevent contamination of produce, such as water treatment before irrigation.

Humans and Hoofed Livestock Are the Main Sources of Fecal Contamination of Rivers Used for Crop Irrigation: A Microbial Source Tracking Approach

Freshwater bodies receive waste, feces, and fecal microorganisms from agricultural, urban, and natural activities. In this study, the probable sources of fecal contamination were determined. Also, antibiotic resistant bacteria (ARB) were detected in the two main rivers of central Chile. Surface water samples were collected from 12 sampling sites in the Maipo (n = 8) and Maule Rivers (n = 4) every 3 months, from August 2017 until April 2019. To determine the fecal contamination level, fecal coliforms were quantified using the most probable number (MPN) method and the source of fecal contamination was determined by Microbial Source Tracking (MST) using the Cryptosporidium and Giardia genotyping method. Separately, to determine if antimicrobial resistance bacteria (AMB) were present in the rivers, Escherichia coli and environmental bacteria were isolated, and the antibiotic susceptibility profile was determined. Fecal coliform levels in the Maule and Maipo Rivers ranged between 1 and 130 MPN/100-ml, and 2 and 30,000 MPN/100-ml, respectively. Based on the MST results using Cryptosporidium and Giardia host-specific species, human, cattle, birds, and/or dogs hosts were the probable sources of fecal contamination in both rivers, with human and cattle host-specific species being more frequently detected. Conditional tree analysis indicated that coliform levels were significantly associated with the river system (Maipo versus Maule), land use, and season. Fecal coliform levels were significantly (p < 0.006) higher at urban and agricultural sites than at sites immediately downstream of treatment centers, livestock areas, or natural areas. Three out of eight (37.5%) E. coli isolates presented a multidrug-resistance (MDR) phenotype. Similarly, 6.6% (117/1768) and 5.1% (44/863) of environmental isolates, in Maipo and Maule River showed and MDR phenotype. Efforts to reduce fecal discharge into these rivers should thus focus on agriculture and urban land uses as these areas were contributing the most and more frequently to fecal contamination into the rivers, while human and cattle fecal discharges were identified as the most likely source of this fecal contamination by the MST approach. This information can be used to design better mitigation strategies, thereby reducing the burden of waterborne diseases and AMR in Central Chile.

Correlation value determined to increase Salmonella prediction success of deep neural network for agricultural waters

The use of computer-based tools has been becoming popular in the field of produce safety. Various algorithms have been applied to predict the population and presence of indicator microorganisms and pathogens in agricultural water sources. The purpose of this study is to improve the Salmonella prediction success of deep feed-forward neural network (DFNN) in agricultural surface waters with a determined correlation value based on selected features. Datasets were collected from six agricultural ponds in Central Florida. The most successful physicochemical and environmental features were selected by the gain ratio for the prediction of generic Escherichia coli population with machine learning algorithms (decision tree, random forest, support vector machine). Salmonella prediction success of DFNN was evaluated with dataset including selected environmental and physicochemical features combined with predicted E. coli populations with and without correlation value. The performance of correlation value was evaluated with all possible mathematical dataset combinations (nCr) of six ponds. The higher accuracy performances (%) were achieved through DFNN analyses with correlation value between 88.89 and 98.41 compared to values with no correlation value from 83.68 to 96.99 for all dataset combinations. The findings emphasize the success of determined correlation value for the prediction of Salmonella presence in agricultural surface waters.

Prevalence and Clonal Diversity of over 1,200 Listeria monocytogenes Isolates Collected from Public Access Waters near Produce Production Areas on the Central California Coast during 2011 to 2016

A 5-year survey of public access surface waters in an agricultural region of the Central California Coast was done to assess the prevalence of the foodborne pathogen Listeria monocytogenes. In nature, L. monocytogenes lives as a saprophyte in soil and water, which are reservoirs for contamination of preharvest produce. Moore swabs were deployed biweekly in lakes, ponds, streams, and rivers during 2011 to 2016. L. monocytogenes was recovered in 1,224 of 2,922 samples, resulting in 41.9% prevalence. Multiple subtypes were isolated from 97 samples, resulting in 1,323 L. monocytogenes isolates. Prevalence was higher in winter and spring and after rain events in some waterways. Over 84% of the isolates were serotype 4b. Whole-genome sequencing was done on 1,248 isolates, and in silico multilocus sequence typing revealed 74 different sequence types (STs) and 39 clonal complexes (CCs). The clones most isolated, CC639, CC183, and CC1, made up 27%, 19%, and 13%, respectively, of the sequenced isolates. Other types were CC663, CC6, CC842, CC4, CC2, CC5, and CC217. All sequenced isolates contained intact copies of core L. monocytogenes virulence genes, and pathogenicity islands LIPI-3 and LIPI-4 were identified in 73% and 63%, respectively, of the sequenced isolates. The virulence factor internalin A was predicted to be intact in all but four isolates, while genes important for sanitizer and heavy metal resistance were found in <5% of the isolates. These waters are not used for crop irrigation directly, but they are available to wildlife and can flood fields during heavy rains. IMPORTANCE Listeria monocytogenes serotype 4b and 1/2a strains are implicated in most listeriosis, and hypervirulent listeriosis stems from strains containing pathogenicity islands LIPI-3 and LIPI-4. The waters and sediments in the Central California Coast agricultural region contain widespread and diverse L. monocytogenes populations, and all the isolates contain intact virulence genes. Emerging clones CC183 and CC639 were the most abundant clones, and major clones CC1, CC4, and CC6 were well represented. CC183 was responsible for three produce-related outbreaks in the last 7 years. Most of the isolates in the survey differ from those of lesser virulence that are often isolated from foods and food processing plants because they contain genes encoding an intact virulence factor, internalin A, and most did not contain genes for sanitizer and heavy metal resistance. This isolate collection is important for understanding L. monocytogenes populations in agricultural and natural regions.

Salmonella enterica Serovar Diversity, Distribution, and Prevalence in Public Access Waters from a Central California Coastal Leafy Green Growing Region during 2011 - 2016

Prevalence and serovar diversity of Salmonella enterica was measured during a five-year survey of surface waters in a 500 mi^2 agricultural region of the Central California Coast. Rivers, streams, lakes, and ponds were sampled bimonthly resulting in 2,979 samples. Overall prevalence was 56.4% with higher levels detected in Spring than in Fall. Small, but significant, differences in prevalence were detected based on sample locations. Detection of Salmonella was correlated positively with both significant rain events and, in some environments, levels of generic Escherichia coli. Analysis of 1,936 isolates revealed significant serovar diversity, with 91 different serovars detected. The most common isolated serovars were S. enterica subsp. enterica serovars I 6,8:d:- (406 isolates, 21.0%, and potentially monophasic Salmonella Muenchen), Give (334 isolates, 17.3%), Muenchen (158 isolates, 8.2%), Typhimurium (227 isolates, 11.7%), Oranienburg (106 isolates, 5.5%), and Montevideo (78 isolates, 4%). Sixteen of the 24 most common serovars detected in the region are among the serovars reported to cause the most human salmonellosis in the United States. Some of the serovars were associated with location and seasonal bias. Analysis of XbaI Pulsed Field Gel Electrophoresis (PFGE) patterns of strains of serovars Typhimurium, Oranienburg, and Montevideo showed significant intra-serovar diversity. PFGE pulsotypes were identified in the region for multiple years of the survey, indicating persistence or regular re-introduction to the region. Importance Non-typhoidal Salmonella is the among the leading causes of bacterial foodborne illness and increasing numbers of outbreaks and recalls are due to contaminated produce. High prevalence and 91 different serovars were detected in this leafy green growing region. Seventeen serovars that cause most of the human salmonellosis in the United States were detected, with 16 of those serovars detected in multiple locations and multiple years of the 5-year survey. Understanding the widespread prevalence and diversity of Salmonella in the region will assist in promoting food safety practices and intervention methods for growers and regulators.

Evaluation of E. coli in sediment for assessing irrigation water quality using machine learning

Fresh produce irrigated with contaminated water poses a substantial risk to human health. This study evaluated the impact of incorporating sediment information on improving the performance of machine learning models to quantify E. coli level in irrigation water. Field samples were collected from irrigation canals in the Southwest U.S., for which meteorological, chemical, and physical water quality variables as well as three additional flow and sediment properties: the concentration of E. coli in sediment, sediment median size, and bed shear stress. Water quality was classified based on E. coli concentration exceeding two standard levels: 1 E. coli and 126 E. coli colony forming units (CFU) per 100 ml of irrigation water. Two series of features, including (FIS) and excluding (FES) sediment features, were selected using multi-variant filter feature selection. The correlation analysis revealed the inclusion of sediment features improves the correlation with the target standards for E. coli compared to the models excluding these features. Support vector machine, logistic regression, and ridge classifier were tested in this study. The support vector machine model performed the best for both targeted standards. Besides, incorporating sediment features improved all models' performance. Therefore, the concentration of E. coli in sediment and bed shear stress are major factors influencing E. coli concentration in irrigation water.

Levels of Salmonella enterica and Listeria monocytogenes in Alternative Irrigation Water Vary Based on Water Source on the Eastern Shore of Maryland

Irrigation water sources have been shown to harbor foodborne pathogens and could contribute to the outbreak of foodborne illness related to consumption of contaminated produce. Determining the probability of and the degree to which these irrigation water sources contain these pathogens is paramount. The purpose of this study was to determine the prevalence of Salmonella enterica and Listeria monocytogenes in alternative irrigation water sources. Water samples (n = 188) were collected over 2 years (2016 to 2018) from 2 reclaimed water plants, 3 nontidal freshwater rivers, and 1 tidal brackish river on Maryland's Eastern Shore (ESM). Samples were collected by filtration using modified Moore swabs (MMS) and analyzed by culture methods. Pathogen levels were quantified using a modified most probable number (MPN) procedure with three different volumes (10 liters, 1 liter, and 0.1 liter). Overall, 65% (122/188) and 40% (76/188) of water samples were positive for S. enterica and L. monocytogenes, respectively. For both pathogens, MPN values ranged from 0.015 to 11 MPN/liter. Pathogen levels (MPN/liter) were significantly (P < 0.05) greater for the nontidal freshwater river sites and the tidal brackish river site than the reclaimed water sites. L. monocytogenes levels in water varied based on season. Detection of S. enterica was more likely with 10-liter filtration compared to 0.1-liter filtration. The physicochemical factors measured attributed only 6.4% of the constrained variance to the levels of both pathogens. This study shows clear variations in S. enterica and L. monocytogenes levels in irrigation water sources on ESM. IMPORTANCE In the last several decades, Maryland's Eastern Shore has seen significant declines in groundwater levels. While this area is not currently experiencing drought conditions or water scarcity, this research represents a proactive approach. Efforts, to investigate the levels of pathogenic bacteria and the microbial quality of alternative irrigation water are important for sustainable irrigation practices into the future. This research will be used to determine the suitability of alternative irrigation water sources for use in fresh produce irrigation to conserve groundwater.

Integrative Survey of 68 Non-overlapping Upstate New York Watersheds Reveals Stream Features Associated With Aquatic Fecal Contamination

Aquatic fecal contamination poses human health risks by introducing pathogens in water that may be used for recreation, consumption, or agriculture. Identifying fecal contaminant sources, as well as the factors that affect their transport, storage, and decay, is essential for protecting human health. However, identifying these factors is often difficult when using fecal indicator bacteria (FIB) because FIB levels in surface water are often the product of multiple contaminant sources. In contrast, microbial source-tracking (MST) techniques allow not only the identification of predominant contaminant sources but also the quantification of factors affecting the transport, storage, and decay of fecal contaminants from specific hosts. We visited 68 streams in the Finger Lakes region of Upstate New York, United States, between April and October 2018 and collected water quality data (i.e., Escherichia coli, MST markers, and physical–chemical parameters) and weather and land-use data, as well as data on other stream features (e.g., stream bed composition), to identify factors that were associated with fecal contamination at a regional scale. We then applied both generalized linear mixed models and conditional inference trees to identify factors and combinations of factors that were significantly associated with human and ruminant fecal contamination. We found that human contaminants were more likely to be identified when the developed area within the 60 m stream buffer exceeded 3.4%, the total developed area in the watershed exceeded 41%, or if stormwater outfalls were present immediately upstream of the sampling site. When these features were not present, human MST markers were more likely to be found when rainfall during the preceding day exceeded 1.5 cm. The presence of upstream campgrounds was also significantly associated with human MST marker detection. In addition to rainfall and water quality parameters associated with rainfall (e.g., turbidity), the minimum distance to upstream cattle operations, the proportion of the 60 m buffer used for cropland, and the presence of submerged aquatic vegetation at the sampling site were all associated based on univariable regression with elevated levels of ruminant markers. The identification of specific features associated with host-specific fecal contaminants may support the development of broader recommendations or policies aimed at reducing levels of aquatic fecal contamination.

New standards at European Union level on water reuse for agricultural irrigation: Are the Spanish wastewater treatment plants ready to produce and distribute reclaimed water within the minimum quality requirements?

The new European regulation on minimum quality requirements (MQR) for water reuse (EU, 2020/741) was launched in May 2020 and describes the directives for the use of reclaimed water for agricultural irrigation. This Regulation will be directly applicable in all Member States from 26 June 2023. Since its publication in 2020, concerns have raised about potential non-compliance situations in water reuse systems. The present study represents a case study where three different water reuse systems have been monitored to establish their compliance with the MQR. Each water reuse system includes a wastewater treatment plant (WWTP), a distribution/storage system and an end-user point, where water is used for irrigation of leafy greens. The selected water reuse systems allowed us to compare the efficacy of water treatments implemented in two WWTPs as well as the impact of three different irrigation systems (drip, furrow and overhead irrigation). The presence and concentration of indicator microorganisms (Escherichia coli and C. perfringens spores) as well as pathogenic bacteria (Shiga toxin-producing, E. coli (STEC), E. coli O157:H7, and Salmonella spp.) were monitored in different sampling points (influent and effluent of the WWTPs, water reservoirs located at the distribution system and the end-user point at the irrigation system as well as in the leafy greens during their growing cycle. Average levels of E. coli (0.73 ± 1.20 log cfu E. coli/100 mL) obtained at the point where the WWTP operator delivers reclaimed water to the next actor in the chain, defined in the European regulation as the 'point of compliance', were within the established MQR (<1 log cfu/100 mL) (EU, 2020/741). On the other hand, average levels of E. coli at the end-user point (1.0 ± 1.2 log cfu/100 mL) were below the recommended threshold (2 log cfu E. coli/100 mL) for irrigation water based on the guidance document on microbiological risks in fresh fruits and vegetables at primary production (EC, 2017/C_163/01). However, several outlier points were observed among the samples taken at the irrigation point, which were linked to a specific cross-contamination event within the distribution/storage system. Regarding pathogenic bacteria, water samples from the influent of the WWTPs showed a 100% prevalence, while only 5% of the effluent samples were positive for any of the monitored pathogenic bacteria. Obtained results indicate that reclaimed water produced in the selected water reuse system is suitable to be used as irrigation water. However, efforts are necessary not only in the establishment of advance disinfection treatments but also in the maintenance of the distribution/storage systems.

Pathogen and Surrogate Survival in Relation to Fecal Indicator Bacteria in Freshwater Mesocosms

The microbial quality of agricultural water for fresh produce production is determined by the presence of the fecal indicator bacterium (FIB) Escherichia coli, despite poor correlations with pathogen presence. Additional FIB, such as enterococci, have been utilized for assessing water quality. The study objective was to determine the survival times (first time to detect zero or censored) of FIB (E. coli and enterococci), surrogates (Listeria innocua, Listeria seeligeri, Salmonella enterica serovar Typhimurium, and PRD1), and pathogens (four strains each of pathogenic E. coli and Listeria monocytogenes and five Salmonella serovars) simultaneously inoculated in freshwater mesocosms exposed to diel and seasonal variations. Six separate mesocosm experiments were conducted for ≤28 days each season, with samples (sediment/water) collected each day for the first 7 days and weekly thereafter. Microorganisms survived significantly longer in sediment than in water (hazard ratio [HR] for water/sediment is 2.2; 95% confidence interval [CI], 1.79 to 2.71). Also, FIB E. coli survived significantly longer than FIB enterococcus (HR for enterococci/E. coli is 12.9 [95% CI, 8.18 to 20.37]) after adjusting for the sediment/water and lake/river effects. Differences in the area under the curve (calculated from log CFU or PFU over time) were used to assess pathogen and surrogate survival in relation to FIB. Despite sample type (sediment/water) and seasonal influences, survival rates of pathogenic Salmonella serovars were similar to those of FIB E. coli, and survival rates of L. monocytogenes and pathogenic E. coli were similar to those of FIB enterococci. Further investigation of microbial survival in water and sediment is needed to determine which surrogates are best suited to assess pathogen survival in agricultural water used in irrigation water for fresh produce. IMPORTANCE Contamination of fresh produce via agricultural water is well established. This research demonstrates that survival of fecal indicator bacteria, pathogenic microorganisms, and other bacterial and viral surrogates in freshwater differs by sample type (sediment/water) and season. Our work highlights potential risks associated with pathogen accumulation and survival in sediment and the possibility for resuspension and contamination of agricultural water used in fresh produce production. Specifically, a greater microbial persistence in sediments than in water over time was observed, along with differences in survival among microorganisms in relation to the fecal indicator bacteria E. coli and enterococci. Previous studies compared data among microbial groups in different environments. Conversely, fecal indicator bacteria, surrogates, and pathogenic microorganisms were assessed within the same water and sediment mesocosms in the present study during four seasons, better representing the agricultural aquatic environment. These data should be considered when agricultural microbial water quality criteria in fresh produce operations are being determined.

A Systematic Review of Listeria Species and Listeria monocytogenes Prevalence, Persistence, and Diversity throughout the Fresh Produce Supply Chain

Listeria monocytogenes is an increasing food safety concern throughout the produce supply chain as it has been linked to produce associated outbreaks and recalls. To our knowledge, this is the first systematic literature review to investigate Listeria species and L. monocytogenes prevalence, persistence, and diversity at each stage along the supply chain. This review identified 64 articles of 4863 candidate articles obtained from four Boolean search queries in six databases. Included studies examined naturally detected/isolated Listeria species and L. monocytogenes in fresh produce-related environments, and/or from past fresh produce associated outbreaks or from produce directly. Listeria species and L. monocytogenes were detected in each stage of the fresh produce supply chain. The greatest prevalence of Listeria species was observed in natural environments and outdoor production, with prevalence generally decreasing with each progression of the supply chain (e.g., packinghouse to distribution to retail). L. monocytogenes prevalence ranged from 61.1% to not detected (0.00%) across the entire supply chain for included studies. Listeria persistence and diversity were also investigated more in natural, production, and processing environments, compared to other supply chain environments (e.g., retail). Data gaps were identified for future produce safety research, for example, in the transportation and distribution center environment.

Interpretability Versus Accuracy: A Comparison of Machine Learning Models Built Using Different Algorithms, Performance Measures, and Features to Predict E. coli Levels in Agricultural Water

Since E. coli is considered a fecal indicator in surface water, government water quality standards and industry guidance often rely on E. coli monitoring to identify when there is an increased risk of pathogen contamination of water used for produce production (e.g., for irrigation). However, studies have indicated that E. coli testing can present an economic burden to growers and that time lags between sampling and obtaining results may reduce the utility of these data. Models that predict E. coli levels in agricultural water may provide a mechanism for overcoming these obstacles. Thus, this proof-of-concept study uses previously published datasets to train, test, and compare E. coli predictive models using multiple algorithms and performance measures. Since the collection of different feature data carries specific costs for growers, predictive performance was compared for models built using different feature types [geospatial, water quality, stream traits, and/or weather features]. Model performance was assessed against baseline regression models. Model performance varied considerably with root-mean-squared errors and Kendall's Tau ranging between 0.37 and 1.03, and 0.07 and 0.55, respectively. Overall, models that included turbidity, rain, and temperature outperformed all other models regardless of the algorithm used. Turbidity and weather factors were also found to drive model accuracy even when other feature types were included in the model. These findings confirm previous conclusions that machine learning models may be useful for predicting when, where, and at what level E. coli (and associated hazards) are likely to be present in preharvest agricultural water sources. This study also identifies specific algorithm-predictor combinations that should be the foci of future efforts to develop deployable models (i.e., models that can be used to guide on-farm decision-making and risk mitigation). When deploying E. coli predictive models in the field, it is important to note that past research indicates an inconsistent relationship between E. coli levels and foodborne pathogen presence. Thus, models that predict E. coli levels in agricultural water may be useful for assessing fecal contamination status and ensuring compliance with regulations but should not be used to assess the risk that specific pathogens of concern (e.g., Salmonella, Listeria) are present.

Listeria cossartiae sp. nov., Listeria farberi sp. nov., Listeria immobilis sp. nov., Listeria portnoyi sp. nov. and Listeria rustica sp. nov., isolated from agricultural water and natural environments

A total of 27 Listeria isolates that could not be classified to the species level were obtained from soil samples from different locations in the contiguous United States and an agricultural water sample from New York. Whole-genome sequence-based average nucleotide identity blast (ANIb) showed that the 27 isolates form five distinct clusters; for each cluster, all draft genomes showed ANI values of <95 % similarity to each other and any currently described Listeria species, indicating that each cluster represents a novel species. Of the five novel species, three cluster with the Listeria sensu stricto clade and two cluster with sensu lato. One of the novel sensu stricto species, designated L. cossartiae sp. nov., contains two subclusters with an average ANI similarity of 94.9%, which were designated as subspecies. The proposed three novel sensu stricto species (including two subspecies) are Listeria farberi sp. nov. (type strain FSL L7-0091T=CCUG 74668T=LMG 31917T; maximum ANI 91.9 % to L. innocua), Listeria immobilis sp. nov. (type strain FSL L7-1519T=CCUG 74666T=LMG 31920T; maximum ANI 87.4 % to L. ivanovii subsp. londoniensis) and Listeria cossartiae sp. nov. [subsp. cossartiae (type strain FSL L7-1447T=CCUG 74667T=LMG 31919T; maximum ANI 93.4 % to L. marthii) and subsp. cayugensis (type strain FSL L7-0993T=CCUG 74670T=LMG 31918T; maximum ANI 94.7 % to L. marthii). The two proposed novel sensu lato species are Listeria portnoyi sp. nov. (type strain FSL L7-1582T=CCUG 74671T=LMG 31921T; maximum ANI value of 88.9 % to L. cornellensis and 89.2 % to L. newyorkensis) and Listeria rustica sp. nov. (type strain FSL W9-0585T=CCUG 74665T=LMG 31922T; maximum ANI value of 88.7 % to L. cornellensis and 88.9 % to L. newyorkensis). L. immobilis is the first sensu stricto species isolated to date that is non-motile. All five of the novel species are non-haemolytic and negative for phosphatidylinositol-specific phospholipase C activity; the draft genomes lack the virulence genes found in Listeria pathogenicity island 1 (LIPI-1), and the internalin genes inlA and inlB, indicating that they are non-pathogenic.

Environmental dissemination of pathogenic Listeria monocytogenes in flowing surface waters in Switzerland

Listeria monocytogenes is an opportunistic pathogen that is widely distributed in the environment. The aquatic environment may represent a potential source for the transmission of L. monocytogenes to animals and the food chain. The present study assessed the occurrence of L. monocytogenes in 191 surface water samples from rivers, streams and inland canals throughout Switzerland. Twenty-five (13%) of the surface water samples contained L. monocytogenes. Whole genome sequence (WGS) data were used to characterize the 25 isolates. The isolates belonged to major lineages I and II, with the majority assigned to either serotype 1/2a (48%), or 4b (44%). The predominant CCs identified were the hypervirulent serotype 4b clones CC1 and CC4, and the serotype CC412; all three have been implicated in listeriosis outbreaks and sporadic cases of human and animal infection worldwide. Two (8%) of the isolates belonged to CC6 which is an emerging hypervirulent clone. All isolates contained intact genes associated with invasion and infection, including inlA/B and prfA. The four CC4 isolates all harbored Listeria pathogenicity island 4 (LIPI-4), which confers hypervirulence. The occurrence of L. monocytogenes in river ecosystems may contribute to the dissemination and introduction of clinically highly relevant strains to the food chain.

Predictive Models May Complement or Provide an Alternative to Existing Strategies for Assessing the Enteric Pathogen Contamination Status of Northeastern Streams Used to Provide Water for Produce Production

While the Food Safety Modernization Act established standards for the use of surface water for produce production, water quality is known to vary over space and time. Targeted approaches for identifying hazards in water that account for this variation may improve growers’ ability to address pre-harvest food safety risks. Models that utilize publicly-available data (e.g., land-use, real-time weather) may be useful for developing these approaches. The objective of this study was to use pre-existing datasets collected in 2017 (N = 181 samples) and 2018 (N = 191 samples) to train and test models that predict the likelihood of detecting Salmonella and pathogenic E. coli markers (eaeA, stx) in agricultural water. Four types of features were used to train the models: microbial, physicochemical, spatial and weather. “Full models” were built using all four features types, while “nested models” were built using between one and three types. Twenty learners were used to develop separate full models for each pathogen. Separately, to assess information gain associated with using different feature types, six learners were randomly selected and used to develop nine, nested models each. Performance measures for each model were then calculated and compared against baseline models where E. coli concentration was the sole covariate. In the methods, we outline the advantages and disadvantages of each learner. Overall, full models built using ensemble (e.g., Node Harvest) and “black-box” (e.g., SVMs) learners out-performed full models built using more interpretable learners (e.g., tree- and rule-based learners) for both outcomes. However, nested eaeA-stx models built using interpretable learners and microbial data performed almost as well as these full models. While none of the nested Salmonella models performed as well as the full models, nested models built using spatial data consistently out-performed models that excluded spatial data. These findings demonstrate that machine learning approaches can be used to predict when and where pathogens are likely to be present in agricultural water. This study serves as a proof-of-concept that can be built upon once larger datasets become available and provides guidance on the learner-data combinations that should be the foci of future efforts (e.g., tree-based microbial models for pathogenic E. coli).

Total Coliform and Generic E. coli Levels, and Salmonella Presence in Eight Experimental Aquaponics and Hydroponics Systems: A Brief Report Highlighting Exploratory Data

Although many studies have investigated foodborne pathogen prevalence in conventional produce production environments, relatively few have investigated prevalence in aquaponics and hydroponics systems. This study sought to address this knowledge gap by enumerating total coliform and generic E. coli levels, and testing for Salmonella presence in circulating water samples collected from five hydroponic systems and three aquaponic systems (No. of samples = 79). While total coliform levels ranged between 6.3 Most Probable Number (MPN)/100-mL and the upper limit of detection (2496 MPN/100-mL), only three samples had detectable levels of E. coli and no samples had detectable levels of Salmonella. Of the three E. coli positive samples, two samples had just one MPN of E. coli/100-mL while the third had 53.9 MPN of E. coli/100-mL. While the sample size reported here was small and site selection was not randomized, this study adds key data on the microbial quality of aquaponics and hydroponics systems to the literature. Moreover, these data suggest that contamination in these systems occurs at relatively low-levels, and that future studies are needed to more fully explore when and how microbial contamination of aquaponics and hydroponic systems is likely to occur.

The Composition of Microbial Communities in Six Streams, and Its Association With Environmental Conditions, and Foodborne Pathogen Isolation

Surface water used for produce production is a potential source of pre-harvest contamination with foodborne pathogens. Decisions on how to mitigate food safety risks associated with pre-harvest water use currently rely on generic Escherichia coli-based water quality tests, although multiple studies have suggested that E. coli levels are not a suitable indicator of the food safety risks under all relevant environmental conditions. Hence, improved understanding of spatiotemporal variability in surface water microbiota composition is needed to facilitate identification of alternative or supplementary indicators that co-occur with pathogens. To this end, we aimed to characterize the composition of bacterial and fungal communities in the sediment and water fractions of 68 agricultural water samples collected from six New York streams. We investigated potential associations between the composition of microbial communities, environmental factors and Salmonella and/or Listeria monocytogenes isolation. We found significantly different composition of fungal and bacterial communities among sampled streams and among water fractions of collected samples. This indicates that geography and the amount of sediment in a collected water sample may affect its microbial composition, which was further supported by identified associations between the flow rate, turbidity, pH and conductivity, and microbial community composition. Lastly, we identified specific microbial families that were weakly associated with the presence of Salmonella or Listeria monocytogenes, however, further studies on samples from additional streams are needed to assess whether identified families may be used as indicators of pathogen presence.

Landscape, Water Quality, and Weather Factors Associated With an Increased Likelihood of Foodborne Pathogen Contamination of New York Streams Used to Source Water for Produce Production

There is a need for science-based tools to (i) help manage microbial produce safety hazards associated with preharvest surface water use, and (ii) facilitate comanagement of agroecosystems for competing stakeholder aims. To develop these tools an improved understanding of foodborne pathogen ecology in freshwater systems is needed. The purpose of this study was to identify (i) sources of potential food safety hazards, and (ii) combinations of factors associated with an increased likelihood of pathogen contamination of agricultural water Sixty-eight streams were sampled between April and October 2018 (196 samples). At each sampling event separate 10-L grab samples (GS) were collected and tested for Listeria, Salmonella, and the stx and eaeA genes. A 1-L GS was also collected and used for Escherichia coli enumeration and detection of four host-associated fecal source-tracking markers (FST). Regression analysis was used to identify individual factors that were significantly associated with pathogen detection. We found that eaeA-stx codetection [Odds Ratio (OR) = 4.2; 95% Confidence Interval (CI) = 1.3, 13.4] and Salmonella isolation (OR = 1.8; CI = 0.9, 3.5) were strongly associated with detection of ruminant and human FST markers, respectively, while Listeria spp. (excluding Listeria monocytogenes) was negatively associated with log10 E. coli levels (OR = 0.50; CI = 0.26, 0.96). L. monocytogenes isolation was not associated with the detection of any fecal indicators. This observation supports the current understanding that, unlike enteric pathogens, Listeria is not fecally-associated and instead originates from other environmental sources. Separately, conditional inference trees were used to identify scenarios associated with an elevated or reduced risk of pathogen contamination. Interestingly, while the likelihood of isolating L. monocytogenes appears to be driven by complex interactions between environmental factors, the likelihood of Salmonella isolation and eaeA-stx codetection were driven by physicochemical water quality (e.g., dissolved oxygen) and temperature, respectively. Overall, these models identify environmental conditions associated with an enhanced risk of pathogen presence in agricultural water (e.g., rain events were associated with L. monocytogenes isolation from samples collected downstream of dairy farms; P = 0.002). The information presented here will enable growers to comanage their operations to mitigate the produce safety risks associated with preharvest surface water use.