PATHOGEN RESPONSES TO ENVIRONMENTAL STRESSORS IN PRE-HARVEST AND POST- HARVEST STAGES OF PRODUCE PRODUCTION By Dimple Sharma A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Food Science – Doctor of Philosophy 2025 ABSTRACT Fresh produce can serve as a vehicle for foodborne pathogens, posing serious public health risks. These pathogens, including Shiga toxin-producing Escherichia coli (STEC), Salmonella enterica, and Listeria monocytogenes, can contaminate agricultural environments through multiple pathways. The first part of the research includes the factors that affect the survival of STEC, S. enterica, and L. monocytogenes in soil extracts simulating flooded or stagnant water conditions. Chemical analysis revealed that high- nutrient extracts, characterized by elevated nitrogen, phosphorus, and carbon levels, promoted pathogen persistence, whereas native microbiomes played a critical role in reducing pathogen populations. Microbiome analysis indicated higher microbial diversity in low-nutrient extracts, suggesting complex interactions that influence pathogen survival. Additionally, a long short-term memory (LSTM) model accurately predicted pathogen survival based on soil parameters, demonstrating the potential for predictive tools in food safety risk assessments. The second phase of this research examined the impact of cold storage on the survival and transition to physiological state of STEC O157:H7 on romaine lettuce. Lettuce harvested at 9 °C and stored at 2 °C for five days exhibited increased dormancy, transitioning into persister and viable but non-culturable (VBNC) states. STEC cell populations including persister cells showed reduced virulence, acid tolerance, and chlorine tolerance over time, underscoring the need to consider storage- induced physiological changes in microbial risk assessments. Finally, we investigated how enrichment time, strain variability, and persister cell populations influenced STEC O157:H7 detection using a commercially available qPCR-based system. Detection rates improved significantly with 24-hour enrichment compared to 8-hour enrichment, and the Yuma, AZ 2018 outbreak strain demonstrated higher detection rates than other strains. Additionally, samples with persister cell populations exceeding 68% were more likely to be detected, highlighting the importance of optimizing enrichment protocols to enhance detection reliability. This study underscores the significance of soil ecology, post-harvest storage conditions, and detection methodologies in improving microbial risk assessments in the fresh produce supply chain. Fresh produce can sometimes be contaminated with pathogens. These pathogens can be deadly and can contaminate produce by various means. The first part of the research includes the factors that affect the survival of STEC, S. enterica, and L. monocytogenes in soil extracts simulating flooded or stagnant water conditions. Chemical analysis revealed higher total nitrogen, phosphorus, and carbon in high-nutrient extracts, which supported pathogen survival, whereas the presence of native microbiomes reduced pathogen levels. Microbiome analysis showed greater diversity in low-nutrient extracts, indicating distinct microbial interactions that influence pathogen persistence. A long short-term memory (LSTM) model effectively predicted pathogen survival based on soil parameters, underscoring the potential for predictive tools in food safety risk assessments. The second part of the research includes evaluation of the impact of cold storage on the survival and physiological state of Shiga toxin-producing E. coli (STEC) O157:H7 on romaine lettuce. Lettuce harvested at 9 °C followed by storage at 2 °C for five days exhibited increased transformation into dormant states, including persister and viable but non-culturable (VBNC) cells. These dormant cells showed reduced virulence, acid tolerance, and chlorine tolerance over time. These findings highlight the need to account for storage-induced physiological changes when assessing microbial risks in the food supply chain. Finally, we investigated how enrichment time, strain variability, and persister cell populations influence STEC O157:H7 detection using the commercially available qPCR-based detection system. Detection rates increased significantly with 24-hour enrichment compared to 8-hour enrichment and the Yuma, AZ 2018 outbreak strain exhibited higher detection rates than other strains. Samples with persister cell percentages exceeding 68% were more likely to be detected, emphasizing the importance of optimizing enrichment protocols to improve detection reliability. This study highlights the importance of soil ecology, storage conditions, and detection methods, for better risk assessment analysis. ACKNOWLEDGMENTS I would like to express my humble gratitude and service to HDG Sri Bhaktisiddhanta Saraswati Thakur, HDG Srila Prabhupada, HH Indradyumna Swami Maharaj, HG Yugal Kishore Das, HG Shyamananda Krishna Das, HG Deen Dayal Das and other people whose names I am unable to mention here, and who are very-very close to my heart, especially for helping even with questions and favors that were never asked. “Hare-Krishna Hare-Krishna, Krishna-Krishna Hare-Hare Hare-Rama Hare-Rama, Rama-Rama Hare-Hare" With this prayer, I would like to thank the lord for EVERYTHING. My deepest gratitude to my mom Suman, my dad Karan, my brother Hetanshu, and my more than a sister Ekta for their loving support. Papa, thank you for believing in me, more than anybody else. Mamma, and papa, thank you for opening the doors of education for me, thank you for being by my side through the entire process; this is YOURS more than it is mine. I would like to express my sincere gratitude to my primary advisor, Dr. Teresa Bergholz for molding me into a scientist with your knowledge, expertise, and your person. I would like to thank my committee members Jeffery Swada, Jade Mitchell, Leslie Bourquin, and Shannon Manning for their guidance, and encouragement. Special thanks to the department of Food Science and Human Nutrition for the opportunities to research and teach. Thanks to Graduate Student Association for giving the department a homely environment. Thanks to Cleary Catur, Shaney Rump, Alyssa Chechak, Corrine Kamphuis, Diane Luong, Xheni Alibashi, Nolan Schinderle, Jonas Ahonen, and Avery Evans for all the help in the lab. Thanks to my friends for standing by my side through the thicks and thins of this process. I would also especially like to express my gratitude for being a great support in and out of the lab towards Dr. Joshua Owade, Yawei Lin, Hazel Zeng, De’Anthony Morris, and Jun Nam. iv PREFACE Foodborne outbreaks associated with fresh produce are a persistent public health concern, with pathogens such as STEC, S. enterica, and L. monocytogenes posing serious risks to consumers. Contamination often occurs pre-harvest, influenced by environmental factors such as flooding, soil nutrient composition, and microbial dynamics. While past studies have investigated microbial risks in agriculture, there remains a gap in understanding how soil chemistry and microbiomes influence pathogen persistence. This study aims to bridge that gap by evaluating pathogen survival in soil extracts, focusing on nutrient availability and microbial interactions. The integration of predictive modeling tools, such as LSTM models, offers a promising approach to assessing pathogen risks in flooded agricultural environments if used along with quantitative risk assessment. By incorporating key soil parameters, these models can enhance our ability to forecast pathogen survival and inform agricultural practices. This research contributes to the development of more accurate food safety risk assessments, supporting efforts to minimize contamination risks and improve outbreak prevention strategies. This research improves food safety risk assessments, helping to reduce contamination risks and strengthen outbreak prevention strategies. Understanding how environmental stressors impact pathogen behavior is crucial for enhancing food safety. For instance, STEC O157:H7, a significant cause of leafy green related outbreaks, exhibits physiological adaptations during storage and transport, which may impact virulence and detectability. Cold storage conditions have been shown to induce dormancy in STEC O157:H7, potentially affecting stress tolerance and sanitization efficacy. Current risk assessment models often overlook such physiological changes, leading to potential underestimation of pathogen survival. This study also investigates how enrichment times, strain variability, and persister cell populations can influence the detection of STEC O157:H7, emphasizing the need for optimized detection protocols. This study is essential for strengthening food safety by uncovering factors that affect pathogen survival in agricultural soil and cold storage, improving risk assessments, and refining detection methods to reduce contamination and prevent foodborne outbreaks. v TABLE OF CONTENTS 1. LITERATURE REVIEW ...................................................................................................................... 1 1.1. Nutritional Benefits and Food Safety Risk .................................................................................... 1 1.2. Epidemiology and Pathogenesis of Listeria Monocytogenes ........................................................ 1 1.3. Epidemiology and Pathogenesis of Salmonella ............................................................................. 2 1.4. Epidemiology and Pathogenesis of STEC ..................................................................................... 3 1.5. Transmission of These Pathogens to Lettuce and Other Fresh Produce ........................................ 3 1.6. Pre-Harvest Produce Contamination and Regulations ................................................................... 4 1.7. Effect of Soil Composition on Pathogen Survival ......................................................................... 4 1.8. Impact of Irrigation, Rainfall, and Flood Events on Pathogen Survival in Agricultural Soils ...... 5 1.9. Harvesting Process and Contamination Controls ........................................................................... 5 1.10. Pre-Processing Time Delays .......................................................................................................... 6 1.11. Post-Harvest Controls for Fresh Produce ....................................................................................... 6 1.12. Bacteria Can Enter Dormant States to Escape Stress .................................................................... 7 1.13. Detection of STEC ......................................................................................................................... 7 1.14. Goals for Detection ........................................................................................................................ 8 1.15. Gaps in Knowledge Related to Health Risk and Bacterial Detection Ability ................................ 9 IMPACT OF BIOTIC AND ABIOTIC FACTORS ON LISTERIA MONOCYTOGENES, 2. SALMONELLA ENTERICA, AND ENTEROHEMORRHAGIC ESCHERICHIA COLI IN AGRICULTURAL SOIL EXTRACTS ....................................................................................................... 10 2.1. Abstract ........................................................................................................................................ 10 Introduction .................................................................................................................................. 11 2.2. 2.3. Materials and Methods ................................................................................................................. 12 2.4. Results .......................................................................................................................................... 18 2.5. Discussion .................................................................................................................................... 24 2.6. Conclusions .................................................................................................................................. 27 2.7. Funding ........................................................................................................................................ 27 3. DYNAMICS OF PHYSIOLOGICAL CHANGES OF STEC O157:H7 ON ROMAINE LETTUCE DURING COLD STORAGE, AND SUBSEQUENT EFFECT ON VIRULENCE AND STRESS TOLERANCE .............................................................................................................................................. 28 3.1. Abstract ........................................................................................................................................ 28 3.2. Introduction .................................................................................................................................. 29 3.3. Methods ........................................................................................................................................ 31 3.4. Results .......................................................................................................................................... 37 3.5. Discussion .................................................................................................................................... 52 3.6. Conclusions .................................................................................................................................. 55 3.7. Acknowledgment ......................................................................................................................... 56 3.8. Funding ........................................................................................................................................ 56 4. IMPACT OF DORMANT STATES AND ENRICHMENT PROTOCOLS ON DETECTION OF STEC O157:H7 ON ROMAINE LETTUCE .............................................................................................. 57 4.1. Abstract ........................................................................................................................................ 57 4.2. Significance and Impact of The Study ......................................................................................... 57 4.3. Introduction .................................................................................................................................. 58 4.4. Methods ........................................................................................................................................ 60 4.5. Results and Discussion ................................................................................................................ 63 4.6. Conclusions .................................................................................................................................. 69 vi 4.7. Funding ........................................................................................................................................ 70 5. CONCLUSIONS, FUTURE WORK, AND LIMITATIONS .............................................................. 71 5.1. Limitations ................................................................................................................................... 72 5.2. Future Work ................................................................................................................................. 73 BIBLIOGRAPHY ........................................................................................................................................ 76 APPENDIX A: SOIL EXTRACTS ............................................................................................................. 98 APPENDIX B: PHYSIOLOGICAL CHANGES IN STEC ...................................................................... 106 APPENDIX C: DETECTION OF STEC .................................................................................................. 112 vii 1. LITERATURE REVIEW 1.1. Nutritional Benefits and Food Safety Risk According to Dietary Guidelines for Americans, adults should consume fruits and vegetables. The recommended amount is 1.5-2 of fruits and 2-3 cup-equivalents of vegetables everyday [1,2]. Leafy greens are a rich source of folate, magnesium, iron, fiber, etc. They are also low in sodium, low in carbohydrates, low in cholesterol, and help in preventing cancer [3]. However, the consumption of fresh produce has also been linked to outbreaks of STEC, Salmonella, and L. monocytogenes [4–6]. The high incidence of illnesses caused by these pathogens, along with the severity of the diseases they trigger, poses a significant public health. According to the Centers for Disease Control, these pathogens are among the top five causes of hospitalizations, illnesses, and fatalities [7], and can cause various symptoms including diarrhea, vomiting, stomach cramps etc. [8–10]. The groups of people who are more susceptible to Listeriosis are immunocompromised people, pregnant women, and elderly people [11]. The most severe outcome of an E. coli infection is hemolytic uremic syndrome (HUS), which manifests in 5-10% of cases [8]. In cases of Salmonellosis, the diarrhea can sometimes become severe and even bloody, indicating a more advanced infection. The severity of the symptoms often requires medical intervention, and in extreme cases, hospitalization may be necessary for proper treatment and recovery [10]. 1.2. Epidemiology and Pathogenesis of Listeria Monocytogenes Listeria monocytogenes (L. monocytogenes) is a Gram-positive bacterium, that is a motile facultative anaerobe and can survive in environments such as soil, water, and produce [12–14]. L. monocytogenes can grow within a temperature range of -0.4°C to 45 °C, with its optimal growth temperature being 37 °C [15]. It can grow between a pH of 4.6 and 9.5 and can tolerate up to 20% salt conditions [16,17]. It is psychrophilic and has an advantage over other microbes that are not able to survive at colder temperatures. Study shows that L. monocytogenes was able to grow at 4 °C for 3 months [18]. Listeriosis symptoms can range from a few days to several weeks. Symptoms for mild infection include muscle pain, vomiting, fever, diarrhea, and nausea, and severe listeriosis include loss of balance, stiff neck, convulsions, headache and confusion [19,20]. The gastrointestinal tract is the primary site of infection [21]. Infection starts after 20 hours of ingestion [22], and the incubation period is 20-30 days [23,24]. Listeria can cause infection in the brain [25], liver [26], bloodstream [25], uterus [25], and gastrointestinal tract [27]. L. monocytogenes can be spread by either oral route, such as eating contaminated vegetables, dairy products, or meat products, or it can be transferred by mother to fetus during pregnancy or at the time of birth [28]. After getting into the host cells by the oral route, L. monocytogenes penetrates the intestinal 1 epithelial barrier itnto the lamina propia by entering into the intestinal lumen; where it later gets distributed through the lymph and blood to spleen, liver, and other organs [29,30]. A virulence component in L. monocytogenes is Listeriolysin (LLO), and this is what helps in making up of the membrane aperture for lysing the phagosome membrae [30]. After entering host cells by endocytosis, hosts’ phagosomes can combine with lysosomes to annihilate the pathogen, but LLO can help L. monocytogenes to escape the phagosome and survive in the host cells [31]. The bacterial infection can spread with the help of surface protein called Actin polymerase A (ActA), that aids in the bacteria to polymerize actin after entering the cytoplasm [32]. L. monocytogenes can cause programmed or unprogrammed cell death i.e., necrosis, and apoptosis respectively [33], and the host major histocompatibility complex (MHC) generates inflammatory factors in response [34]. Different T cell responses in host cells are activated so that the bacterial infection can be removed[33]. L. monocytogenes, according to CDC, has been linked to outbreaks relating to milk products, deli products, fresh produce such as fruits, vegetables, and leafy greens [35]. It has a significant mortality rate of 20-30% worldwide [20]. It is estimated that there are about 1600 listeriosis annual infections, with a 94% hospitalization rate [36]. 1.3. Epidemiology and Pathogenesis of Salmonella Salmonella is a Gram-negative, rod-shaped facultative anaerobe that belongs to the Enterobacteriaceae family. Salmonella enterica is divided into six categories viz., Salmonella enterica subsp. enterica, Salmonella enterica subsp. arizonae, Salmonella enterica subsp. houtenae, Salmonella enterica subsp. indica, Salmonella enterica subsp. salamae and Salmonella enterica subsp. diarizonae [37,38]. According to CDC, 1.35 million illnesses, 26500 hospitalizations, and 420 deaths are caused by Salmonella every year. Symptoms of Salmonella include fever, diarrhea, and stomach cramps [39]. Serious symptoms include sepsis, meningitis, endocarditis, and pneumonia [40,41]. Routes of exposure are eating contaminated food, drinking contaminated water, and touching infected animals, their feces, or their environment. The infectious dose of most strains of Salmonella is 106 to 109 cells [42], which is decreased if the stomach acidity is buffered. Salmonella doesn’t colonize the stomach but colonizes the lumen of the small intestine, and then the large intestine [43], where it passes by through specialized ileal cells called M cells. Salmonella is transferred to the intestinal mucosa, where it adheres to microfold (M) cells of Peyer’s patches, and invades epithelial cells, later reaching the lymph follicles and mostly doesn’t progress beyond that and can cause symptoms [42,44]. Inflammatory reaction is caused when epithelial cells are invaded, followed by diarrhea, which can also cause ulceration and destruction of mucosa [43]. For invading the host cells, Salmonella enterica serovar Typhimurium uses a type 3 secretion system (T3SS) that is encoded by Salmonella pathogenicity Island (SP-1) and SP1-T3SS effectors are known to contribute to the formation of ruffles that enclose the bacteria on the intestinal membrane [45]. Salmonella either encloses itself in the 2 membrane or divides rapidly and can also replicate within the cytosol of the cell. Because of high motility of the cytosolic cells, Salmonella can squeeze out of the membrane layer and potentially spread to other organs [46] . After intestinal infection, the liver, spleen, gall bladder, and bile become infected, which causes intestinal infection, as two weeks after initial ingestion, infected bile can cause secondary intestinal infection [42]. People who can become infected include infants who are not breastfed, children under 5 years, adults over 65 years, those with weakened immune system, and individuals taking certain medicines like antacids [39]. 1.4. Epidemiology and Pathogenesis of STEC There are many pathotypes of E. coli, based on the location of the infection, on surface antigens, and based on virulence genes [47–50]. E. coli strains are divided into different sequence types (STs) also based on housekeeping genes [49,50]. STEC can be detected by slide agglutination test, or PCR and DNA probes for genes stx (for Shiga-toxin) and eae (for intimin). EHEC infections are asymptomatic to severe [51]. EHEC including Shiga toxin-producing E. coli (STEC) O157:H7 infections include non-bloody or bloody diarrhea, hemorrhagic colitis, and hemolytic uremic syndrome (HUS) [52,53]. Once EHEC enters via the oral route, it makes its way through the intestine, and produces Shiga toxin (Stx), that plays a role in EHEC-associated disease [54]. EHEC is also called STEC, as the Shiga toxins produced by EHEC closely resemble the ones produced by Shigella dystenteriae 1 and 2, which are called Stx1 and Stx2 [55]. Out of the many serotypes being found in animals, and humas, only a few are associated with human diseases viz., O26:H11, O91:H21, O111:H8, O157:NM, and O157:H7 [56]. EHEC has been linked with outbreaks linked to a wide range of foods including carrots, greens, nuts, burgers, dairy products etc. [57]. A pathogenesis trait of EHEC is formation of attaching and effacing (A/E) lesions [58]. EHEC binds very tightly to intestinal epithelial cells, which causes absorptive disruption of the intestinal microvilli, and the cytoskeletal proteins of the hosts are accumulated under the attached bacteria [59]. This process leads to formation of pedestal-like structures. The LEE, a pathogenicity island of non-E. coli origin, primarily encodes the ability to form A/E lesions and was identified in 1995 [60,61]. The LEE harbors genes for a Type III Secretion System (T3SS). 1.5. Transmission of These Pathogens to Lettuce and Other Fresh Produce STEC, Salmonella, and L. monocytogenes have been repeatedly linked to fresh produce, as demonstrated by numerous outbreaks [62–64]. The contamination of fresh produce can occur pre-, at-, and post-harvest. Pre-harvest contamination can occur by application of raw manure, lapses in worker hygiene, contaminated irrigation water, or direct fecal deposition [65]. As runoff water from manure or from flooding events can be a potential source [66,67]. FDA final water rule requires farms to evaluate potential for 3 flooding [68]. Greens shouldn’t be harvested within 9 meters from the edge of flooded area and the same flooded area shouldn’t be used up to 60 days due to potential contamination, according to California’s Leafy Greens Marketing Agreement (LGMA) [69–72]. In 2018, the source of an outbreak was traced back to lettuce that came from Yuma, Arizona. The tests showed that the irrigation water was contaminated with STEC that was from animal sources and FDA suggests that the water might have been contaminated by manure from the cattle farm nearby [73,74]. In 2021, there was a voluntary recall issued on Dole salad products, as they detected the presence of L. monocytogenes on harvesting equipment of iceberg lettuce [75]. In July 2021, BrightFarms recalled their packaged salads sold in five northern U.S. states due to S. Typhimurium contamination; subsequent testing did not detect the pathogen in the facility but identified a direct genetic match to the outbreak strain in an outdoor pond [76]. 1.6. Pre-Harvest Produce Contamination and Regulations Different pathways, including contaminated irrigation water, the application of raw manure, direct fecal deposition by wildlife, and lapses in worker hygiene can introduce these pathogens onto fresh produce in the pre-harvest setting [77]. Moyne Et. Al. showed that if the contamination of leafy greens with STEC O157:H7 occurs close to harvest, it leads to an increase in survivability of STEC O157:H7 on leafy greens [78]. After heavy rains or flooding, nutrients from the soil might seep into standing water, influencing pathogen dynamics in agricultural settings. Moreover, the runoff of manure from animal farms into crop fields and water sources, compounded by flooding events, can contribute to the dissemination of pathogens in agricultural environments [79,80]. It's advised by LGMA to avoid utilizing the flooded soil for planting for a period of up to 60 days [81]. 1.7. Effect of Soil Composition on Pathogen Survival The composition of soils used for growing fresh produce varies, leading to differences in pathogen survival [82]. Variations in soil composition are linked to physicochemical properties like pH, porosity, aggregation, and cation exchange, as well as biological factors such as microbial abundance and diversity. Studies highlight that soil texture and chemical composition primarily dictate pathogen survival [77,83– 86]. Salmonella demonstrated enhanced survival rates in loamy soil compared to sandy soil [87]. Additionally, soil texture primarily influenced the long-term survival of L. monocytogenes, while soil chemical properties had a greater impact on its short-term survival, typically lasting less than two weeks [88]. Unfavorable conditions such as low pH and nutrient competition within agricultural soils have been shown to negatively affect the survival of EHEC [89]. The use of soil amendments is known to influence pathogen behavior in soils, particularly with biological soil amendments like dairy manure and poultry litter. These have been associated with increased survival of foodborne pathogens. Studies have 4 demonstrated that Salmonella introduced into amended soils displayed higher survival compared to those introduced into unamended soil [90]. It was observed that attenuated STEC O157:H7 exhibited enhanced survival in soils amended with poultry litter when compared to soils amended with horse manure or dairy manure [91]. Studies on amended soils have uncovered a correlation between higher levels of organic carbon, phosphorus, and nitrogen and the survival of organisms like STEC and Salmonella [90,92]. 1.8. Impact of Irrigation, Rainfall, and Flood Events on Pathogen Survival in Agricultural Soils Irrigation has been shown to impact pathogen survival in amended soils. Salmonella can persist for up to 129 days in soils amended with irrigation every day, which was 89 days longer than in non-amended soils. Irrigation or rainfall can induce pathogen growth in various soils [93]. This can be because these soils tend to have increased accessibility of water-soluble nutrients [80]. Flood events may cause soil saturation and the formation of standing water. This can create favorable conditions for the transmission and survival of foodborne pathogens. In a field study, flooding events caused significant increases in E. coli numbers in soils situated 0.5 and 1.5 meters from the flooded area, with E. coli persisting in the flooded soils for over 60 days [94]. In laboratory settings, soil or compost extracts have been used to replicate soil saturation and standing water conditions for evaluating pathogen survival. In liquid extracts of soil compost, STEC O157:H7, Salmonella, and L. monocytogenes initially showed regrowth only in the absence of indigenous microbiota after an initial decrease [95]. The behavior of Salmonella was notably influenced by nutrient composition in extracts from both amended and unamended soils, with a significantly shorter lag phase and higher maximum density observed in extracts from amended soil compared to unamended soils [90,96]. Although extensive data exist on soil and water chemistry in soil extracts containing foodborne pathogens, a comprehensive evaluation of both factors over a prolonged period is still lacking. The microbiome within these soils and extracts is dynamic, with fluctuations expected in available carbon, phosphorus, and nitrogen due to microbial metabolic activities. The indigenous microbes in soil and water, acting as direct competitors for nutrients, are likely to impact the behavior and survival of pathogens. Integrating both variables will assist in constructing models to predict the persistence and dissemination of pathogens in flooded agricultural systems. 1.9. Harvesting Process and Contamination Controls The harvesting time is very important as the produce can be contaminated while handling. During pre-harvest and harvest, factors that can influence survival of STEC O157:H7 are atmospheric and environmental conditions, types of irrigation, soil conditions, species of vegetables, and inherent properties of microbial cells [97,98]. Atmospheric and environmental conditions include temperature, relative 5 humidity, light intensity, and seasonal changes. LGMA recommends not harvesting leafy products at least up to 1.52 m radius of known scat [98]. LGMA provides numerous guidelines for handling produce during harvest, including the following: three 100 mL samples must test negative for generic E. coli, water quality must remain stable for at least 21 days before the scheduled harvest, and irrigation water must be kept free from contamination by nearby raw manure storage. Raw manure and soil amendments are to be avoided. Proper cleaning, sanitation, and maintenance should be routinely done for equipment used in hydration or hand-held equipment during harvesting. Single-use packing materials, liners for food-contact surfaces, and reusable containers should be used. These guidelines play a crucial role in maintaining produce safety and quality during harvest. 1.10. Pre-Processing Time Delays Contamination of produce can occur not only before harvest but also during post-harvest handling. In the processing stage, bacteria are exposed to various stressors [99], and pathogens can infiltrate produce through scars or damaged skin [100]. In numerous recent outbreaks of STEC O157:H7 associated with leafy greens, most initial cases in the outbreaks originated from states along the East Coast or in the Midwest, and from eastern Canada [101–103]. Even when the product was distributed nationwide, illnesses were region-based. When harvested produce is processed in close proximity to the harvesting facility, and then transported to retail stores, it’s called source processing; and when it’s transported to processing centers, so that it can be processed close to retail stores, it’s called forward processing. In the latter case, raw produce is being held at cold transportation for longer time, if the produce is sent far away compared to close-by from the farm. The difference between both these scenarios is that when the product had to be transported far away, it was transported at a colder temperature. It is suspected that leafy greens that went through pre-processing delays in relation to forward processing are more likely to be linked to illnesses. Low temperatures before processing can affect pathogen behavior. Bacteria may address this problem by entering into dormant states 1.11. Post-Harvest Controls for Fresh Produce Post-harvest controls for fresh produce, such as washing cut or shredded lettuce with sanitizers and applying antimicrobial treatments [104], play a crucial role in reducing contamination. Studies have shown that STEC O157:H7 survives better when lettuce is cut with blunt knives compared to sharp razors [105], highlighting the impact of cutting methods on pathogen survival. USDA provides guidelines for certain sanitizers including chlorine, hydrogen peroxide, ozone, peroxyacetic acid/peracetic acid, phosphoric acid, potassium hydroxide, and sodium hydroxide [106]. Although numerous studies showed some cells STEC O157:H7 declines during sanitizer treatment, but some cells are still able to survive [70–72,107]. Other 6 non- conventional techniques include electrolyzed water, plasma light treatment, cold nitrogen plasma, and irradiation [70–72,107]. FDA suggests considering the temperature of the wash water [108]. Alternative treatments like ionizing radiation, or dry cleaning are considered for water sensitive products. FDA also asks to have cooler temperature for products to maintain produce quality [108]. 1.12. Bacteria Can Enter Dormant States to Escape Stress When bacterial cells go through stressful conditions, they can transition to dormant states like persister cells and VBNC cells. This state is temporary during which the cells have a reduced metabolic activity and no growth [109]. Once the stressful condition is removed and the nutrients are provided, dormant cells can resuscitate and can start growing again [110,111]. These cells are more tolerant to antibiotics and other stressful conditions [110,112]. The process of dormancy involves a spectrum of physiological transformations, spanning from actively proliferating cells to stationary phase cells, and culminating in the formation of persister cells or viable but nonculturable (VBNC) cells [113,114]. Persister cells are a subpopulation that is non growing or slow growing and is able to resist stressful conditions [115]. VBNC cells do not grow and have been previously referred to as conditionally viable environmental cells (CVEC) and active but non culturable (ABNC) cells [116,117]. In dormant states, bacteria doesn’t have metabolic activity or growth [118]. Studies have shown that STEC O157:H7 can transform into persister and VBNC cells under pre-harvest conditions on leafy greens and in water sources [119–121]. During pre-harvest conditions, when STEC on lettuce was exposed to cold temperature of 8 °C, the culturable cells of STEC transition into VBNC cells [122]. STEC can transform into dormant states, not only during pre-harvest, but also during post-harvest conditions. STEC was seen to transform into VBNC cells when the process wash water was treated with chlorine [123]. STEC at 15 °C could also transform into persister cells during post-harvest conditions, of spinach wash water [124]. The number of studies on ability of STEC O157:H7 to enter dormant states are limited, especially during pre- processing stage of lettuce production. 1.13. Detection of STEC STEC infection is linked to many food commodities including meat, milk, and vegetables [125– 127]. Fresh produce is consumed raw, and hence it is even more important to have STEC detection strategies in fresh produce including leafy greens. Different methods of detection include phenotypic and molecular diagnostic methods. Phenotypic methods include biochemical testing, the use of chromogenic media, and matrix assisted laser desorption/ionization-time of flight mass spectroscopy (MALDI-TOF MS). Molecular diagnostic methods include hybridization-based methods, amplification methods, DNA microarrays, whole genome sequencing, and multi-omics approach [128]. A standard method for laboratory procedures for 7 microbiological analysis of food and cosmetics is Bacteriological Analytical Manual (BAM) by U.S. Food and Drug Administration (FDA) [129], where test sample goes through enrichment, then the cells from presumptive positive samples are plated on selective media for STEC O157:H7 [130]. There are many commercially available detection methods including (LAMP) [131], lateral flow immunoassay (LFIA) [132], verotoxin-enzyme linked immunosorbent assay (VT-ELISA) [132] , and NeoSeekTM STEC [133], culture and PCR based methods such as BAX® system [134,135], ANSR system [136]. Limitations of detection methods include distinguishing virulent strains [137], detecting active toxins [138], not identifying emerging or non-viable variants , and producing false positives [139]. STEC is likely present in very low quantities to be detected, so to bring it to a threshold to be detected usually it is enriched. The enrichment time is usually between 8-24 h [122,137,138]. Since the enrichment steps involve the growing of bacteria that are culturable on regular media, most detection methods do not give consideration to the proportion of cells that are in persister cells. 1.14. Goals for Detection Detecting the pathogen helps in preventing and treating infections, to safeguard food safety, and to understand the spread of the disease. Detection is done for the following reasons- a) spread of infection can be prevented if the presence of pathogen is known, b) detecting pathogens in food products can help prevent foodborne illnesses, c) detection in samples such as water and soil can help identify potential source of contamination, so that preventative measures can be taken by this environmental sampling[122–125]. Since lettuce has been contaminated with STEC multiple times, and has caused outbreaks, there is an urgent need to know how to rapidly and effectively detect STEC on leafy greens, and lettuce specifically. FDA partners with private and public sectors to enhance safety of leafy greens in relation to STEC by implementation of Leafy Green STEC Action Plan (LGAP), which includes focused sampling, prioritized inspections, root cause investigations, and advancements in science of detection and prevention [140]. Inspections for detection are conducted at majorly three priority areas viz., prevention, response, and addressing knowledge gaps. Detection of STEC O157:H7 is part of prevention that also includes testing agricultural water, enhancing inspections, audits, and certification programs for leafy greens during growing season etc. From the farm, detection tests have been directly conducted on lettuce itself. Harvest equipment can also be tested for STEC O157:H7 [140]. FDA is also working on enhancing sample collection methods for detecting STEC from wind and dust and also on high- risk virulence factors in STEC associated with leafy greens. The overarching goal of these efforts is to strengthen early detection capabilities, reduce contamination risks, and ultimately protect public health by ensuring the safety of leafy greens throughout the supply chain. 8 1.15. Gaps in Knowledge Related to Health Risk and Bacterial Detection Ability Soil chemistry, microbiome dynamics, and nutrient availability influence pathogen persistence in agricultural fields, particularly after flooding events like those in Salinas Valley, California. Despite pathogen decline over pathogens such as L. monocytogenes, Salmonella, and STEC can persist in flooding environments, highlighting food safety risks. There needs to be the presence of soil modelling tools to make predictions about the survival of pathogens. A simple soil modelling tool can help farmers understand the risk of survival of pathogen in flooded waters and improve risk mitigation and agricultural decision-making. As observed in outbreaks, the health risks associated with microbial detection in physiology and virulence may increase. Previous studies in quantitative microbial risk assessment (QMRA) have mainly concentrated on STEC O157:H7 in leafy greens and lettuce, motivated by past outbreaks and aimed at evaluating intervention measures such as irrigation water quality, temperature control, UV radiation, chlorine sanitizers, and others [86,141–144]. But these QMRA assessments have typically focused on detectable and quantifiable culturable bacteria. The health risks associated with post-harvest processing and handling, including timing and temperature effects on pathogen behavior, are not well-documented, underscoring a significant knowledge gap. Research needs to be done on how the physiological state of STEC O157:H7 can impact its tolerance to stresses, virulence, and detectability. 9 2. IMPACT OF BIOTIC AND ABIOTIC FACTORS ON LISTERIA MONOCYTOGENES, SALMONELLA ENTERICA, AND ENTEROHEMORRHAGIC ESCHERICHIA COLI IN AGRICULTURAL SOIL EXTRACTS Microorganisms, 2024, 12, 1498. MDPI (https://doi.org/10.3390/ microorganisms12071498) Dimple Sharma1, Autumn L. Kraft2, Joshua O. Owade1,3, Mateja Milicevic3, Jiyoon Yi3, and Teresa M. Bergholz1* 1 Department of Food Science and Human Nutrition, Michigan State University 2 Department of Microbiological Sciences, North Dakota State University (Present- U.S. Food and Drug Administration) 3 Department of Biosystems and Agricultural Engineering, Michigan State University * Corresponding author 2.1. Abstract Outbreaks of Enterohemorrhagic Escherichia coli (EHEC), Salmonella enterica, and Listeria monocytogenes linked to fresh produce consumption pose significant food safety concerns. These pathogens can contaminate pre-harvest produce through various routes, including contaminated water. Soil physicochemical properties and flooding can influence pathogen survival in soils. We investigated survival of EHEC, S. enterica, and L. monocytogenes in soil extracts designed to represent soils with stagnant water. We hypothesized pathogen survival is influenced by soil extract nutrient levels and presence of native microbes. Chemical analysis revealed higher levels of total nitrogen, phosphorus, and carbon in high- nutrient soil extracts compared to low-nutrient extracts. Pathogen survival was enhanced in high-nutrient, sterile soil extracts, while the presence of native microbes reduced pathogen numbers. Microbiome analysis showed greater diversity in low-nutrient soil extracts, with distinct microbial compositions between extract types. Our findings highlight the importance of soil nutrient composition and microbial dynamics in influencing pathogen behavior. Given key soil parameters, long short-term memory model (LSTM) effectively predicted pathogen survival. Integrating these factors can aid in developing predictive models for pathogen persistence in agricultural systems. Overall, our study contributes to understanding the complex interplay in agricultural ecosystems, facilitating informed decision-making for crop production and food safety enhancement. 10 Keywords: Soil extracts, Listeria monocytogenes, Salmonella enterica, Enterohemorrhagic Escherichia coli, native microbiome. 2.2. Introduction Outbreaks of Enterohemorrhagic Escherichia coli (EHEC), Salmonella, and Listeria monocytogenes have been linked to the consumption of fresh produce [62–64]. These pathogens are a significant food safety concern due to the high number of illnesses and severity of disease [145]. These pathogens can be introduced onto fresh produce in the pre-harvest environment through various routes including contaminated irrigation water, application of raw manure, direct fecal deposition by wildlife, and lapses in worker hygiene [65]. Additionally, the runoff of manure from animal farms to crop fields and water sources, as well as flooding events, can contribute to pathogen spread in agricultural settings [66,146]. The Food and Drug Administration (FDA) final water rule recognizes this risk, as farmers are required to evaluate potential for flooding as sediments in these environments can serve as a harbor site for foodborne pathogens [68]. If fields are flooded, the California Leafy Greens Marketing Agreement (LGMA) [69] requires that leafy greens are not harvested within 9 meters from the edge of the flooded area due to potential contamination. It also has been suggested to not use the same flooded soil for planting for up to 60 days [98]. Soils used for production of fresh produce vary in composition and pathogen survival can differ as well [147]. The differences in physiochemical properties like pH, porosity, aggregation, and cation exchange, and biological properties including abundance and diversity of microbes of the soils affect pathogen survival [65,147–151]. Loamy soil supported greater Salmonella survival compared to sandy soil[152]. Soil texture mainly affected long-term survival of L. monocytogenes, while short term (< 2 weeks) survival was influenced more by soil chemical properties [153]. Adverse conditions like low pH and competition for nutrients within agricultural soils have been shown to negatively affect the survival of EHEC [154]. Other than the intrinsic variation, soil management practices also alter physicochemical properties and thus influence foodborne pathogen survival in soils. Application of soil amendments is known to influence pathogen behavior in soils, where the presence of biological soil amendments, like dairy manure and poultry litter, has been associated with increased survival of foodborne pathogens. Salmonella inoculated into amended soils had greater survival compared to when inoculated into unamended soil [155]. Attenuated E. coli O157:H7 survived better in soils amended with poultry litter compared to soils amended with horse manure or dairy manure [156]. In establishing that amended soils had elevated levels of organic carbon, phosphorus, and nitrogen, there was a positive correlation with the survival of organisms such as E. coli and Salmonella [155,157]. 11 Irrigation or rainfall can cause pathogens to grow in certain soils [158], likely due to increased availability of water-soluble nutrients [66]. Salmonella survived for 129 days in daily irrigated amended soils, 89 days longer than non-amended soils[158]. Other than irrigation, flooding events may lead to soil saturation and standing water, which may provide more favorable conditions for foodborne pathogen transmission and survival. In a field study, flooding events led to significant increases in E. coli levels in soils that were 0.5 and 1.5m from the flooded area, and E. coli persisted in the flooded soils for over 60 days [159]. Soil or compost extracts have been utilized in the laboratory to mimic soil saturation and standing water to assess pathogen survival. In liquid extracts of soil compost, EHEC O157:H7, Salmonella, and L. monocytogenes exhibited an initial probable decline in numbers but showed regrowth only in the absence of indigenous microbiota [160]. Salmonella behavior was significantly impacted by nutrient composition in extracts of amended and unamended soils, where the length of lag phase was significantly shorter and maximum density was significantly higher in extracts from amended soil [155,156]. While there has been a collection of soil and water chemistry data for soil extracts containing foodborne pathogens, there isn’t a thorough assessment that considers both factors over an extended period. The microbiome present in these soils and extracts is dynamic, with the levels of available carbon, phosphorus, and nitrogen likely to fluctuate due to the metabolic activities of the microbiome. The native microbes in the soil and water, serving as direct competitors for nutrients, are expected to play a role in influencing the behavior and survival of pathogens. Integrating both variables will help develop models to predict the persistence and dissemination of pathogens in flooded agricultural systems. In our research, we investigated the survival of EHEC, Salmonella enterica, and Listeria monocytogenes in two different extracts from soils commonly found in agricultural settings for production of fresh produce. We hypothesized that pathogen survival would be enhanced with high quantity of nutrients and inhibited by presence of native microbes. By unraveling the dynamics of soil chemistry, microbial composition, and pathogen survival, our study establishes groundwork for informed decision-making in agriculture. The conditions provided in this study simulate scenarios of stagnant water after intense rainfall or flooding, as well as overflow from nearby drainage that could be harboring pathogens. In detailing best practices for the management of agricultural fields post-flooding events for production of fresh produce, it is necessary to have a comprehensive understanding of the factors influencing pathogen behavior. 2.3. Materials and Methods 2.3.1. Bacterial strain preparation Two strains each of Salmonella, L. monocytogenes, and EHEC (Table 2.1) were used in the first phase of the study. For the second phase of the study, one representative strain was used from each species 12 (bolded in Table 2.1). Both L. monocytogenes strains and S. enterica strain FSL-S10-1646 were sourced from the Food Safety Lab at Cornell University. EHEC strains were sourced from the Thomas S. Whittam STEC collection at MSU. S. enterica strain Mdd314 was sourced from the USDA ARS in Beltsville, MD and was previously used in a soil extract study [156]. The Salmonella Newport strain, Mdd314, which was previously selected for resistance to 80 µg/mL rifampicin was used [161]. The L. monocytogenes strain, 10403S, was already resistant to 80 µg/mL streptomycin. The EHEC strain, MI-0041B, was selected for resistance to 80 µg/mL rifampicin (Thermo Scientific, Waltham, MA) using the process of random mutant selection [18]. This EHEC strain was streaked onto a Luria Bertani (LB) plate (Invitrogen, Carlsbad, California) and incubated at 37 °C for 24h. One colony was transferred to 50 mL LB broth with 4 µg/mL rifampicin. The broth was incubated at 37 °C for 24h with shaking. A 100 µL aliquot of this overnight culture was transferred to a new 50 mL LB broth with 40 µg/mL rifampicin and incubated at 37 °C for 24h with shaking. A 100 µL aliquot of this overnight culture was added to 50 mL LB broth with 80 µg/mL rifampicin and incubated for 24h with shaking. Confirmation of growth was done by taking a loopful of the overnight culture and streaking onto LB media plated supplemented with 80 µg/mL rifampicin. Of the overnight culture LB broth supplemented with 80 µg/mL rifampicin, 800 µL was added to 200 µL of 75% glycerol and stored in a -80 °C freezer. All the strains were stored in Brain-heart infusion broth (BHIB) (Neogen, Lansing, MI) with glycerol at -80 °C for use throughout the experiments. Table 2.1. List of strains used. Strain 10403S H7858 Species Serogroup (serovar) Source Listeria monocytogenes 1/2 a Skin Lesion Listeria monocytogenes 4b MI-0041B Escherichia coli DA-5 Escherichia coli O157 O121 Hot Dog Human Human FSL-S10-1646 Salmonella enterica Enteritidis Environmental, Produce Mdd314 Salmonella enterica Newport Tomato Bold letters indicate the representative strains (10403S, MI-0041B, and Mdd314) used in the second phase of study. For each inoculation, L. monocytogenes strains were streaked onto BHI plates and incubated at 37 °C. The Salmonella and EHEC strains were streaked onto LB plates and incubated at 37 °C. Isolated single colonies were picked and transferred into BHI broth + 80 µg/mL streptomycin for L. monocytogenes strains, and LB broth + 80 µg/mL rifampicin for EHEC and Salmonella strains. These were incubated at 37 °C without shaking for 16 h. The resultant plate count data revealed the concentration of each strain to be approximately 109 CFU/mL [156]. 13 2.3.2. Soil sample collection and extract preparation Two different types of soil were collected in bulk from the North Dakota State University (NDSU) Beef Cattle barn and from a cornfield close to NDSU campus, which were then stored in 6-7 kg batches in large Ziploc bags at –80 °C to be used throughout the experiment, as described previously [156]. 2.3.2.1. Phase-1 The first set of experiments was conducted to evaluate the impact of sterilization of the soil extracts on pathogen behavior. Soil extracts were prepared following the methodology developed [156]. To prepare soil extracts, 25g soil was taken from the frozen sample and added to 50 mL Milli-Q ultrapure water. These suspensions were held at 4 °C for 24h with slow agitation at 30 rpm. After incubation, the suspensions were centrifuged at 5000 rpm; the liquid portion was kept as the soil extract and was used for the soil extract assays. Preliminary chemical analysis of the soil extracts based on Total Organic Carbon (TOC), nitrogen, and phosphorous revealed differences in the cornfield and beef barn samples, which led them to be categorized as low-nutrient and high-nutrient respectively (Table 2.2). To assess the impact of the native soil microbes on pathogen behavior, the soil extracts were divided in half and those portions were sterilized via filtration of the extract through a 0.22 µm filter. The extracts were classified as follows, high-nutrient sterile (HS), high-nutrient non-sterile (HNS), low-nutrient sterile (LS), and low-nutrient non-sterile (LNS). 2.3.2.2. Phase-2 The second set of experiments was designed to quantify pathogen behavior in the soil extracts and assess the microbiome composition over time. Large scale preparation of the soil extracts was done to provide sufficient volume for pathogen inoculation, chemical analysis, and DNA isolation for microbiome analysis. For each high-nutrient and low-nutrient soil, 625 g was added to 1250 mL of Milli-Q ultrapure water. Soil extract was collected the same way as described in section 2.2.1. Aliquots of 49.5 mL were made for each soil type for pathogen experiments, and 50 mL aliquots were made for chemical analysis for each time point. 2.3.3. Inoculation of soil extracts In the first phase of study, high- and low-nutrient soil extracts, which were either sterile or non- sterile were inoculated individually with the two strains of three pathogens each. For the assay, 0.5 mL of bacterial inoculum diluted in Butterfield’s buffer (pH 7.2) was added to 4.5 mL soil extract, for a final concentration of 104 CFU/mL. The inoculated soil extracts were incubated at 15°C for 14 d. Plate count data for each of the six strains was taken at days- d0, d1, d2, d3, d4, d6, d8, d10, d12, and d14, respectively, after the time of inoculation. The samples were collected at the same time for each day. In the second phase 14 of study, one strain from each of the three pathogen types was chosen (bolded in Table 2.1). Extracts were inoculated the same way as for the first phase, but only non-sterile samples were used in this phase. 2.3.4. Enumeration of pathogens from soil extracts For enumeration of pathogens, media supplemented with antibiotics was used, as described in section 2.1. Salmonella and EHEC were enumerated on xylose lysine deoxycholate (XLD) supplemented with rifampicin (XLD, BD Diagnostics, Berkshire, UK) and MacConkey agar (BD Diagnostics, Berkshire, UK) supplemented with rifampicin, respectively. L. monocytogenes was plated on Rapid LM (Bio-Rad Laboratories, Inc., Hercules, CA) supplemented with streptomycin. XLD and MacConkey agar plates were incubated at 37 °C and Rapid LM plates were incubated at 30 °C for 24 h. Q-Count (Color Q-Count, Spiral Biotech Inc., Norwood, MA) was used for enumeration of colonies. Data was collected at d0, d1, d4, d6, d8, d10, and d14 at the same time each day. 2.3.5. Enumeration of mesophilic aerobic microbes Plate count data for mesophilic aerobic heterotrophic microbes were collected at the same data points as described in section 2.4. Soil extracts aliquots were diluted in PBS and plated onto LB agar and incubated at 30 °C for 24 h, similar to the process used by [156]. Resulting colonies were enumerated and are referred to as mesophilic aerobic microbes throughout the manuscript. To observe changes in the mesophilic aerobic microbe count in the absence of pathogens, a set of soil extracts was left uninoculated and mesophilic aerobic microbes were sampled and quantified at the same time points. 2.3.6. Chemical analysis of soil extracts Samples of soil extracts were collected from the uninoculated non-sterile extracts for chemical analysis at the same data points in phase 2. These soil extract samples were sent to the NDSU soil testing lab for chemical analysis. Soil extract samples were stored at -20 °C after collection and prior to the analysis. Chemical analysis was performed for total organic carbon (TOC), nitrogen (N) and phosphorus (P), as well as alkalinity, pH, and determining the concentration of sodium (Na), calcium (Ca), magnesium (Mg), potassium (K), copper (Cu), iron (Fe), manganese (Mn), zinc (Zn), and chloride ions (Cl-). For total N, and P, Kjeldahl digestion was performed. In this method, using selenium catalyst, organic bound element is digested with concentrated sulfuric acid and potassium sulfate [162]. Alkalinity was measured by quantifying carbonates and bicarbonates, which can indicate the local geochemical environment. To accurately measure the quantity of carbonates and bicarbonates in a sample, a pH electrode is combined with CO2 electrode [163]. For phase 1, samples of the soil extracts were taken at the start of the experiment, for phase 2, samples for chemical analysis were taken at every sampling point as described in 2.3. 15 2.3.7. DNA extraction Total genomic DNA was extracted from each soil extract sample at each sample time point using the Power Soil DNA extraction kit (Qiagen, Ann Arbor, MI). DNA was isolated from 3 independent replicates of each pathogen in each soil extract (HNS and LNS) as well as 3 independent replicates of each soil extract that remained uninoculated. Isolated DNA was stored at –20 °C. 2.3.8. PCR and Sequencing The 465 bp (base pair) 16S rDNA region was amplified using PCR with V3 forward and V4 reverse primers 5’-CCTACGGGAGGCAGCAG-3'– and 5’-GGACTACHVGGGTWTCTAAT-3’respectively [164] (Invitrogen). ThermoScientificTM DreamTaqTM Hot start DNA polymerase was used for PCR. The PCR was performed using the following conditions: 2 min initial denaturation at 95 °C, denaturation at 95 for 40s, annealing at 50 °C for 30s, extension at 72 °C for 1 min, final extension at 72°C for 7 min. The cycle of denaturation, annealing, and extension was repeated 30 times. After the final extension, the samples were held at 4 °C until they were taken out of the thermocycler, and later stored at -20 °C. The genomic region that was amplified from the samples were sent to Genomics core facility in MSU (Michigan State University, East Lansing, MI) for amplicon sequencing with the Illumina MiSeq. 2.3.9. Statistical analysis and predictive modelling 2.3.9.1. Pathogen survival Pathogen count data from both phases of the experiment were collected for three biological and two technical replicates. CFU/mL values were log-transformed for each time point and were used for subsequent analyses. One-way analysis of variance (ANOVA) in R programming language was used to test for differences in the microbial counts and chemical composition of different soil extracts. For means that were statistically (p<0.05) different, Tukey’s Honest Significant Difference (HSD) was used to separate the individual differences among the soil extracts. Generalized linear model with random effects were used to test for the effect of different nutrient levels and pathogen on the survival of the cells over different days. Estimated marginal means was used to compare the survival of different pathogens in different soil extracts (p<0.05). 2.3.9.2. 16S sequence analysis The sequences that were obtained after amplicon sequencing using Illumina were analyzed using QIIME2 [165]. QIIME2 was used to determine OTUs (operational taxonomical units) and their abundance. It was also used to plot a relative abundance of different taxonomical units, alpha diversity plots, and Bray- Curtis dissimilarity plots. Two types of alpha diversity plots were created; one to plot the microbial diversity in low- and high-nutrient soil extracts, and another for microbial diversity in uninoculated soil extracts, and 16 soil extracts each inoculated with three pathogens. The plots were viewed on https://view.qiime2.org/. The files used for viewing on this website can be accessed here (https://osf.io/2wsy8/). Similarity percentages for abundance of different taxa from day 0 to 14 was analyzed using Vegan package in R programming language. This was done using a modified version of the simper command [166,167]. 2.3.9.3. Dimensionality reduction for predictive modeling Several methodologies were explored to create a predictive model for pathogen survival rates. Initially, input data were scaled to a range between 0 and 1 by subtracting the mean from the values and subsequently dividing by standard deviation. Non-numeric values, which represented unmeasured pathogen counts for non-inoculated control soil extract samples, were replaced with zeros. This approach was selected to retain essential data within the limited dataset, thus facilitating the inclusion of complete rows containing other measurements from these control samples in further analyses. Principal component analysis (PCA) was then utilized for dimensionality reduction using the scikit-learn PCA transformer in Python [168], retaining 95% of the variance. This approach selected the most important variables in soil chemistry and microbial composition data. 2.3.9.4. Development and evaluation of predictive models Following data cleaning and dimensionality reduction, various modeling techniques including deep learning, machine learning, and ensemble methods were employed to develop robust predictive models. The models were trained to learn the correlation between soil extract dataset (i.e., soil chemistry and microbial composition) and pathogen counts for day 0. Then, the models were tested using data from day 14 to assess their ability to predict pathogen survival based on the soil chemistry and microbial composition from that day. A deep learning approach involved constructing a deep neural network (DNN) with two layers and ReLU activation functions using PyTorch, a Python-based deep learning library [169]. To prevent training data overfitting, k-fold cross-validation with three splits was used alongside an early stopping mechanism based on loss. Machine learning algorithms such as the random forest (RF) regressor and the support vector machine (SVM) regressor were also employed, using the scikit-learn library. These models were fitted using grid search cross-validation for hyperparameter tuning to prevent overfitting. Furthermore, an ensemble model using a bagging strategy was developed to combine predictions from RF and SVM to ascertain the optimal model. Additionally, a sequential forecasting approach using long short-term memory (LSTM) was developed to improve pathogen survival prediction. To ensure class balance for input data, data were manually split into training and testing sets so that each sequence represented a time-series data for respective soil extract samples over the data collection period. For the training of LSTM, initial sequential data from days 0 and 1 were used to forecast pathogen survival for subsequent days 4, 6, 8, 10, and 14. 17 Training was done in batches of 10 epochs until no further improvement in model validation loss was observed. This model was developed and trained using TensorFlow [170], with mean squared error (MSE) as the loss function and Adam optimizer for optimization of the model. Overall, the effectiveness of all models in pathogen survival prediction was assessed and compared using MSE on the test dataset. 2.4. Results 2.4.1. Pathogen behavior is influenced by soil extract chemistry and presence of native microbiome For Phase 1 of the study, the initial inoculum in all four soil extracts (high nutrient sterile (HS), high nutrient non-sterile (HNS), low nutrient sterile (LS) and low nutrient non-sterile (LNS)) for all strains ranged between 3.39 to 3.78 log CFU/mL (Figure A. 1). Over the 14-day period, strains inoculated into HS had a significant increase (p<0.05) ranging from 0.65 to 2.32 log CFU/mL (Figure 2.1). When the native microbiome remained in the HNS soil extract, the overall pathogen reductions ranged from 0.37 to 0.62 log CFU/mL after 14 d. The low nutrient extract did not support pathogen growth, even when sterilized. Some strains had a higher reduction, such as EHEC DA-5 (1.57 ± 0.36 log CFU/mL) and S. enterica mdd314 (1.18 ± 0.14 log CFU/mL), while others such as L. monocytogenes H7858 had an average log reduction of 0.55 ± 1.09 in the LS extract. Pathogen reductions were greatest (p <0.05) in the LNS extract ranging from 1.66 (L. monocytogenes 10403S) to 3.24 (S. enterica mdd314) log CFU/mL. Figure 2.1. Bacterial population over 14 days. 18 Figure 2.1. (cont’d) Bacterial population growth or reduction (log CFU/mL) after 14 d in high- (dark red and dark blue) and low- (light red and light blue) sterile (right) and non-sterile (left) soil extracts. Bars represent the average difference in log CFU/mL from 0 to 14 d. Initial chemical assessment of the soil extracts showed higher levels of total nitrogen, phosphorus and carbon in the high-nutrient than the low-nutrient soil extracts (Table 2.2). Relative to the low-nutrient soil extract, the high-nutrient soil extract had 5x greater total nitrogen, 8x greater total phosphorus and 9x greater total carbon. Table 2.2. Preliminary chemical analysis results of soil extracts from two different soil sources. Soil extract type Total N (ppm) Total P (ppm) Total C (ppm) High nutrient (beef 134.5 ± 25.6 22.4 ± 4.4 248.7 ± 58.8 barn) Low nutrient (corn 30.1 ± 8.5 2.8 ± 0.1 27.5 ± 2.7 field) The data is reported in parts per million (ppm) and is collected from three replicates and average and standard deviation are reported for each compound. During phase 2 of the study, there were no significant changes in chemical composition of the soil extracts over time (p>0.05). We then tested for significant differences between LNS and HNS soil extracts and identified several chemical parameters that were significantly different between the extract types. Alkalinity, chloride ions, K, Na, Fe, total P, total N and TOC were significantly (p<0.05) higher in HNS compared to LNS (Table S 2.1). 2.4.2. Levels of mesophilic aerobic microbes and pathogens are affected by soil extract nutrient composition Since survival was similar between strains in non-sterile extracts in the first phase, only one of each pathogen was chosen for the second phase. Pathogens had an initial inoculum of approximately 4 log CFU/mL, in HNS soil extract and declined by an average of 0.90 ± 0.73 log CFU/mL for EHEC, 1.15 ± 0.81 log CFU/mL for L. monocytogenes and 0.61 ± 0.81 for Salmonella over 14 d (Figure 2.2a). Pathogen reductions were significantly greater (p<0.05) in LNS, with an average reduction of average of 1.82 ± 0.66 log CFU/mL for EHEC, 2.19 ± 0.72 log CFU/mL for L. monocytogenes and 1.93 ± 0.88 for Salmonella over 14 d (Figure 2.2a). All pathogen counts were significantly different at 14 d compared to 0 d (p>0.05). The average initial concentrations for native mesophilic aerobic microbes were 6.84 ± 0.l39 log CFU/mL 19 in HNS and 4.76 ± 0.28 log CFU/mL in LNS (Figure 2.2b). Mesophilic aerobic microbes were measured for soil extracts inoculated with pathogens and those that remained uninoculated. There were no significant differences (p<0.05) in mesophilic aerobic bacteria, among LNS and HNS samples that were inoculated compared to the ones that were not inoculated. Significant (p<0.05) differences in the overall levels of mesophilic aerobic microbes were observed between HNS and LNS, with significantly higher levels in HNS (Figure 2.2b). At 14 d, mesophilic aerobic microbes were 7.26 ± 0.21 log CFU/mL in HNS and 4.94 ± 0.27 log CFU/mL in LNS. Figure 2.2. Pathogen (a) and mesophilic aerobic microbe (b) counts over 14 d. Pathogen (a) and mesophilic aerobic microbe (b) counts over 14 d in high-nutrient (HNS) and low- (LNS) nutrient soil extracts. Four colors represent the presence of different pathogens in soil extracts and an uninoculated soil extract. Each data point is an average of four biological and two technical replicates. Error bars represent standard deviation of the counts within replicates. 2.4.3. Influence of soil extract composition on the microbiome Microbiome diversity was assessed for each soil extract type over time using 16S rDNA sequencing. Overall, a greater number of taxa were identified in LNS (415) compared to in HNS (370). The top 5 taxa in HNS included Pseudomonadaceae, Moraxellaceae, Weeksellaceae, Flavobacteriaceae, and Xanthomonadaceae, while the top 5 taxa in LNS included Burkholderiaceae, Sphingobacteriaceae, Pseudomonadaceae, Hymenobacteraceae, and Chitinophageceae (Figure A. 2). Significant changes in relative abundance of taxa were determined between HNS and LNS soil extracts over the 14-d incubation. Taxa that were present at significantly greater levels in HNS included Pseudomonadaceae, Moraxellaceae, Weeksellaceae, Dysgonomonadaceae, while Flavobacteriaceae, Paludibacteraceae, Crocinitomicaceae,and Alteromonadaceae, were present in HNS but absent in LNS (Figure A. 3). Taxa that 20 were present at significantly greater levels in LNS included Sphingobacteriaceae and Burkholderiaceae, while Verrucamicrobiaceae, Sphingomonadaceae, Pedosphaeraceae, Enterobacteriaceae, Diplorickettsiaceae, and Caulobacteraceae (Figure A. 3). For HNS soil extracts, significant changes in taxa abundance over time included a decrease in Moraxellaceae and Dysgonomonadaceae from day 0 to day 14, while Pseudomonadaceae and Weeksellaceae significantly increased from day 0 to day 14. For LNS soil extracts, Sphingobacteriaceae and Enterobacteriaceae significantly increased in abundance from day 0 to day 14, while Burkholderiaceae and Pseudomonadaceae significantly decreased in abundance from day 0 to day 14. Figure 2.3. Alpha diversity plot for microbiome samples. Alpha diversity (Shannon entropy) for microbiome samples based on soil extract type., viz. high- and low- nutrient non-sterile samples. Each box plot represents Shannon index values for 84 samples for each soil extract type. Boxes represent the 25th and 75th percentiles and the horizontal bar represents the median value for each distribution. Whiskers show the 10th and 90th percentiles and outliers are indicated with solid circles. For HNS soil extracts inoculated with EHEC, Burkholderiaceae and Sphingobacteriaceae were more abundant compared to HNS soil extracts inoculated with Salmonella or L. monocytogenes (Figure A. 4). Pseudomonadaceae, Moraxellaceae, Weeksellaceae were more abundant in HNS soil extracts inoculated with Salmonella or L. monocytogenes compared to HNS soil extracts inoculated with EHEC. Moraxellaceae decreased significantly from day 0 to day 14 for HNS soil extracts inoculated with Salmonella or L. monocytogenes, while increasing in HNS soil extracts inoculated with EHEC. For LNS soil extracts inoculated with Salmonella, Weeksellaceae and Moraxellaceae were present, while absent in LNS soil extracts inoculated with L. monocytogenes (Figure A. 5). LNS soil extracts inoculated with L. 21 monocytogenes had Pedosphaeraceae and Bacteriovoracaceae, present, which were absent in LNS soil extracts inoculated with Salmonella. 2.4.4. Soil extract microbiome diversity was not affected by pathogen inoculation, but was influenced by soil extract type Alpha diversity was determined for soil extracts inoculated with each pathogen, in comparison to soil extracts that remained uninoculated. The average Shannon Entropy of the soil extract microbiomes were not significantly different (p>0.05) by inoculum type (Figure A. 6). As there were no differences in alpha diversity by inoculum type, we also examined alpha diversity by soil extract nutrient level. The average Shannon Entropy for microbiomes in high nutrient soil extracts was significantly lower (p<0.05) compared to microbiomes in low nutrient soil extracts (Figure 2.3), indicating greater diversity in taxa for LNS. Beta diversity, as determined by the Bray-Curtis dissimilarity also indicated a distinct difference in soil extract microbiome composition by nutrient levels of the extracts (Figure 2.4). Figure 2.4. Bray-Curtis dissimilarity plot. Bray-Curtis dissimilarity plot for samples from HNS (red) and LNS (blue) soil extracts. 2.4.5. Identification of key variables by PCA PCA was conducted to reduce the dimensionality of the soil extract dataset, which encompasses a large number of variables related to soil chemistry and microbial compositions. The results in Figure A. 7 indicate that 75 principal components explain 95% of the total variance in the soil extract dataset. These components were therefore selected as the most important variables for further predictive modeling. 22 Additionally, the PCA results showed that the observations from day 0 were more effectively represented by the major principal components compared to subsequent time points (Figure A. 8). It was also found that soil chemistry data exhibited a strong positive relationship with the first principal component (Figure A. 9). 2.4.6. Prediction of pathogen survival based on soil chemistry and microbial composition Using the key variables identified through PCA, various models were trained to predict pathogen survival after 14 days based on the soil characteristics. The DNN model demonstrated the least effectiveness, with an MSE of 1.545, due to the small data size that typically hinders the generalization capabilities of complex models like DNNs. Machine learning models such as RF and SVM exhibited better performance, achieving MSE values of 0.693 and 0.808, respectively. An ensemble model that integrated predictions from both RF and SVM through a bagging method slightly improved the model performance, achieving an MSE of 0.685. The LSTM model outperformed the others with an MSE of 0.09 for sequential prediction of pathogen survival after day 1. The results in Figure 2.5 further illustrate the variance in survival rates of each pathogen in soil extracts with different nutrient levels. Overall, the LSTM model accurately forecasted the pathogen survival trends based on the initial sequences. However, the predicted EHEC count on day 14 was lower than observed in low nutrient soil extract, whereas the predicted L. monocytogenes count was higher than observed. This performance was slightly improved when additional datasets up to day 4 were included in model training, resulting in an MSE of 0.078 (Figure A. 10). 23 Figure 2.5. Time-series prediction of pathogen survival. Time-series prediction of pathogen survival by the LSTM model trained on sequential datasets from days 0–1. Top row: high nutrient soil extract. Bottom row: low nutrient soil extract. 2.5. Discussion 2.5.1. Significant differences in soil extract chemistry were associated with differences in pathogen survival and native microbiome diversity Nutrient rich soil extracts promoted pathogen growth, as shown by the greater increase in pathogen numbers in the HNS soil extracts compared to the LNS soil extracts. This is likely due to the HNS soil extracts having significantly higher amounts of TOC, P, N, Fe, Na, K, Cl-, and alkalinity. These findings are similar to those from other studies utilizing sterilized soil or compost extracts. In sterilized soil extracts generated from poultry litter-amended and unamended soils, Salmonella Newport grew over the 96h incubation period [156]. Significantly more TOC, P, and N were found in poultry litter-amended soil 24 extracts compared to unamended soil extracts, which was associated with shorter lag phase and higher maximum growth levels for Salmonella in the amended vs. unamended soil extracts [156]. In autoclaved compost extracts with 40-50% moisture, inoculated Salmonella increased up to 4-log CFU/g over 3d at approximately 24ºC [171]. Taken together, these studies indicate that the best conditions for pathogen growth are those associated with high nutrient levels in soil extracts. In soils, foodborne pathogen growth is not typically observed, though physicochemical parameters are associated with differences in survival. EHEC O157:H7 survival was positively correlated with available organic carbon and total nitrogen in soils collected from two fresh produce growing regions, Salinas, California and Yuma, Arizona [172]. Soil chemistry was the principal factor explaining the variation in L. monocytogenes survival in soils over 14 days, where the composition of Ca, Mg, and K explained 55.4% of variability in survival profiles [153]. Other than nutrient quantity, pathogen behavior was also influenced by the presence or absence of the native microbes in the soil extract. Here we utilized soil that was collected at a single timepoint and held at -80°C until needed for the experiments. We do recognize that this storage could impact the native microbes in the soil, though any differences would be consistent throughout the experiment. Even in the presence of high nutrient levels, when native microbes were present, the pathogens were unable to increase in number. Under low nutrient conditions, the presence of native microbes led to a greater reduction in pathogen numbers compared to when they were absent. A similar result was seen in the study utilizing poultry litter amended and unamended soil extracts. When native microbes were present in the extracts, Salmonella Newport grew in the high-nutrient poultry litter amended extract, though less than when it was sterilized. Salmonella was unable to grow in the unamended soil extracts in the presence of the native microbes [156]. Similar findings have been reported for soils treated to purposely remove native microbes. In soil microcosms, E. coli O157:H7 had greater survival in soils with reduced levels of native microbes than those containing the full repertoire of native microbes [173]. E. coli O157:H7 survival was significantly longer in autoclaved soils compared to natural soils [174]. These data all highlight the contribution of both nutrient composition and microbiomes to foodborne pathogen survival in soils and soil extracts. The native soil microbiome is known to have antagonistic effects on survival of foodborne pathogens including Salmonella, L. monocytogenes and EHEC. Survival of E. coli O157:H7 was inversely related to microbial diversity [173,175]. Once invasion happens by a pathogen in a new environment it can use various ways to establish in the environment including competition, antagonism, and predation. Once the pathogen invader is established in the new environment, it grows, spreads, and can displace or shift the resident taxa [176], but microbial diversity in soil can control the extent to which bacterial invaders can establish in the soil [175]. In comparison of different soil management practices, L. monocytogenes was 25 found to have steeper reductions in soils with more diversity but was not impacted by community composition [177] similar to our findings, where pathogen reductions were greater in LNS, which had a higher level of diversity compared to HNS. While specific interactions between native microbiome taxa and foodborne pathogens have not been examined extensively in soils or soil extracts, some antagonistic interactions have been reported. Pseudomonas species, members of the family Pseudomonadaceae, have been shown to significantly reduce the growth of E. coli O157:H7 under varying environmental conditions [178]. Pseudomonas species have also been shown to have antagonistic effects against Salmonella, and L. monocytogenes in solutions and on surface of fruit [179,180]. Burkholderia species, members of the family Burkholderiaceae, are known for their diversity in metabolic capabilities, including the production of numerous secondary metabolites with antimicrobial activity [181]. We observed a high abundance of Burkholderiaceae and Pseudomonadaceae in soil extracts, which could be associated with antagonistic effects toward the inoculated foodborne pathogens. We observed that Burkholderiaceae and Sphingobacteriaceae had a higher relative abundance in soil extracts inoculated with Salmonella or L. monocytogenes compared to EHEC in HNS. Nutrient quantity is likely to play a role in relative abundance of certain families in the soil extract microbiome, and some families of native microbiome can compete with certain pathogens. It is difficult to directly compare the impacts of soil and soil microbiome variables on different foodborne pathogens, as few studies have examined pathogens in the same soils or soil extracts. The data reported here is one of the first studies to assess multiple foodborne pathogens in the same soil extract system. More studies are needed that compare multiple pathogens in the same system. 2.5.2. AI models can predict pathogen survival given key soil variables Despite the challenges posed by a relatively small size of soil extract samples, effective model training results were achieved by employing PCA for dimensionality reduction (Figure 2.5). This strategy was crucial in preventing overfitting on the training data, given the exponential growth in sample size requirements relative to the number of input features. A commonly referenced ‘factor 10’ rule of thumb suggests approximately 10 data points per variable, although this may be considered conservative [182]. Therefore, follow-on studies with larger sample sizes would further enhance the robustness and generalizability of our predictive modeling approach. Among the various models tested, the LSTM demonstrated the most effective prediction, particularly due to its suitability for forecasting applications that require analysis of sequential data. Unlike other models, LSTM leverages memory gates to selectively retain past information and its sequential order, assisting in future predictions [183]. This attribute makes LSTM particularly useful for pattern recognition and forecasting tasks. Previous studies have also shown that LSTMs are well-suited for modeling time- 26 series changes in biological data, including bacterial population behavior in food [184] and dynamic changes in spore concentration [185]. However, the prediction performance varied across different pathogen types and nutrient conditions (Figure 2.5). LSTM predictions were less accurate for EHEC in low nutrient soil extract and L. monocytogenes in high nutrient soil, but this discrepancy reflects distinct patterns in the actual survival data. Notably, our model training did not differentiate between pathogen types, thus learning a generalized trend across all pathogens. This study highlights that the interaction between soil characteristics and pathogen survival can differ among pathogen types. Future research should focus on collecting more data per pathogen type and developing pathogen-specific models, tailored to enhance accuracy and applicability in diverse environmental conditions. 2.6. Conclusions Understanding the microbial dynamics resulting from the interplay of factors such as soil chemical composition and agricultural management practices can be instrumental in developing predictive models for pathogen persistence in agricultural systems [177]. Here we focused on mimicking situations that could occur in agricultural fields following heavy rains and flooding events, like in Salinas Valley California in 2022 and January 2023 [186]. Even though pathogen numbers declined over the 14-d in both high- and low-nutrient soil extracts, L. monocytogenes, Salmonella, and EHEC persisted over this period, emphasizing the food safety concerns posed by events such as flooding and runoffs. Available macro- and micro-nutrients influenced pathogen behavior in the soil extracts, though the effects of nutrient levels were minimized by the presence of the native microbiome. These findings highlight the importance of understanding the complex interplay between soil microbiota and pathogens for effective risk mitigation. While this study explores chemistry, microbiome dynamics, and pathogen survival, future research avenues may delve deeper into the mechanisms underlying these interactions. Additionally, the implications of these findings on crop health, yield, and the broader ecosystem merit further investigation. Incorporating temporal dynamics and seasonal variations could enhance the robustness of future studies in this domain. In conclusion, this manuscript contributes significantly to our understanding of the intricate web of interactions in agricultural ecosystems. By unraveling the dynamics of soil chemistry, microbiome composition, and pathogen survival, the study paves the way for informed decision-making in agriculture, ultimately benefiting both crop production and food safety. 2.7. Funding This research was supported in part by the intramural research program of the U.S. Department of Agriculture, National Institute of Food and Agriculture, Hatch Project#MICL02681 and by NDSU EPSCoR award FAR0032092. 27 3. DYNAMICS OF PHYSIOLOGICAL CHANGES OF STEC O157:H7 ON ROMAINE LETTUCE DURING COLD STORAGE, AND SUBSEQUENT EFFECT ON VIRULENCE AND STRESS TOLERANCE Authors: Dimple Sharma1, Joshua O. Owade2, Corrine J. Kamphius1,3, Avery Evans1,4, E. Shaney Rump1, Cleary Catur1,5, Jade Mitchell2, Teresa M. Bergholz1* 1 Affiliation 1; Department of Food Science and Human Nutrition, Michigan State University, East Lansing, MI sharmad6@msu.edu 2 Affiliation 2; Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI owadejo1@msu.edu 1,3 Affiliation 3; Department of Food Science and Human Nutrition, Michigan State University, East Lansing, MI, Current location- Ascension St. John Hospital Health Center: LabCorp Division, Grosse Pointe, MI kamphu11@msu.edu 1,4 Affiliation 4; Department of Food Science and Human Nutrition, Michigan State University, East Lansing, MI, Current location- Detroit Medical Center University Laboratories, Detroit, MI evansave@msu.edu 1 Affiliation 5; Department of Food Science and Human Nutrition, Michigan State University, East Lansing, MI rumpeile@msu.edu 1,5 Affiliation 6; Department of Food Science and Human Nutrition, Michigan State University, East Lansing, MI, Current location- Roskam Foods, Kentwood, MI caturcle@msu.edu 2 Affiliation 7; Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI jade@msu.edu * Correspondence: tmb@msu.edu 3.1. Abstract If lettuce is contaminated in the field, Shiga-toxin producing E. coli (STEC) O157:H7 can survive through the distribution chain. Prolonged cold storage during transportation may impact pathogen physiology, affecting subsequent stress survival and virulence. Greenhouse-grown Romaine lettuce, inoculated with three STEC O157:H7 strains, was harvested after 24h and stored at 2°C for 5d following 28 4h at harvest temperature (9°C/17°C). Culturable, persister, and viable but non-culturable (VBNC) cells were quantified. Virulence was evaluated using Galleria mellonella, and acid tolerance at pH 2.5 and tolerance to 20-25ppm chlorine were quantified. Colder harvest temperature (9°C) before cold storage led to higher transformation of STEC O157:H7 into dormant states and decreased virulence in most cases. Increasing length of cold storage led to decreased virulence and acid tolerance of STEC O157:H7 on lettuce, while having no significant effect on chlorine tolerance. These findings highlight that entry of STEC O157:H7 into dormant states during harvest and transportation at cold temperatures leads to decreased stress tolerance and virulence with increasing cold storage. Changes in STEC O157:H7 physiology on lettuce during cold storage can be integrated into risk assessment tools for producers, which can assist in identifying practices that minimize risk of STEC O157:H7 from consumption of lettuce. Keywords: Persister, VBNC, chlorine, low pH, low temperature, Galleria mellonella, harvest temperature 3.2. Introduction Fresh produce is consumed widely and has been associated with human illnesses due to contamination by pathogens. Of all foodborne illnesses linked to identified pathogens that occurred in the U.S. between 1998 and 2020, 9.18% are linked to leafy greens. Romaine lettuce, iceberg lettuce, and other lettuces account for 60.8% of leafy green associated outbreaks, and 75.7% of leafy green associated illnesses. Pathogens associated with the majority of illnesses attributed to leafy greens are Shiga toxin- producing Escherichia coli, Salmonella [187,188], Norovirus, and Campylobacter [189]. According to the Centers for Disease Control and Prevention (CDC), STEC ranks among the top five foodborne pathogens for causing hospitalizations [7], and can cause various symptoms including diarrhea, vomiting, and stomach cramps [190]. The most severe symptom of STEC infection is hemolytic uremic syndrome (HUS), which manifests in 5-10% of cases [8]. It is estimated that 19.8% of STEC outbreaks from 1998 to 2020 have been associated with Romaine lettuce resulting in more than 12,000 illnesses annually in the United States [189]. In recent years, multiple STEC outbreaks attributed to Romaine lettuce have been reported [6,12–16], and some strains associated with these outbreaks have recently been classified as recurring, emerging, or persistent by the CDC [197]. Lactuca sativa L. (lettuce) is the third most consumed crop in the U.S. [198]. The production of leafy greens in the U.S. largely occurs in Arizona (~30%) and California (~70%) [199,200]. Lettuce is either field- packed or placed in bins for further processing after cooling [201]. After harvesting, lettuce is transported to cooling centers and is chilled using vacuum cooling, air cooling or forced air cooling [202,203]. Once cooled, field-packed lettuce is transported for distribution and sale. Lettuce designated for 29 further processing is transported to a processing facility. Differences in transportation time can occur prior to processing, depending on where processing facilities are located. For lettuce processing facilities located outside the West Coast, long haul transportation is used where truck temperature is maintained at <5°C [204,205]. When lettuce is stored at <8°C, with low relative humidity (<55%), STEC is unable to grow, though it can survive [206–208]. STEC can tolerate cold stress when present on lettuce, surviving at 5-6 ℃ for 10-14 days [209,210]. In addition to cold stress, STEC present on lettuce experiences multiple stresses during post-harvest handling. Once lettuce arrives at a processing facility, it is trimmed, cored, cut, washed, centrifuged, and then packaged for further distribution and sale [211]. Sanitizers are included in wash water to reduce cross contamination. Commonly used sanitizers for fresh produce are chlorine, peroxyacetic acid (PAA), and ozone for wash water treatments [212–214]. All three of these sanitizers pose oxidative stress to microbes, and PAA also poses a low pH stress [213]. STEC is capable of surviving on lettuce under conditions including cold temperature, sanitizer treatment and acid treatment [209]. STEC levels on lettuce can be reduced by approximately 1.5 to 5 log CFU/g due to sanitizer treatment; this variability is likely due to differences in organic load in the wash water, potential for bacterial adherence and/or internalization [215]. However, some STEC cells on lettuce can survive despite the application of various sanitizers in lettuce processing water [216,217]. STEC that survived on lettuce after washing with sanitizers can persist during cold storage [218]. While studies have shown that STEC on lettuce can survive at refrigeration temperatures and exposure to sanitizers during washing, it is not known if the length of time in cold storage prior to washing influences STEC tolerance to wash water sanitizers. One of the ways bacteria can adapt to unfavorable environments is by entering dormant states, where metabolic activity is reduced [219]. The process of dormancy involves a spectrum of physiological transformations, spanning from actively proliferating cells to stationary phase cells, and culminating in the formation of persister cells or VBNC cells. Studies have shown that STEC can transform into dormant states such as persister and VBNC cells under pre-harvest conditions on leafy greens [220–222]. When STEC was exposed to cold stress of 8°C in the phyllosphere of lettuce pre-harvest, transformation into VBNC cells occurred 8-9 days post-inoculation [220]. In post-harvest conditions, STEC in spinach wash water transformed into persister cells at 15°C [223]. STEC inoculated into the process water transformed into VBNC cells when process water was treated with chlorine [123]. These studies suggest that STEC can alter its physiological state by entering dormant states when presented with stress, but if the length of cold storage impacts the entry into dormant states on lettuce is still unknown. These studies suggest that STEC can alter its physiological state by entering dormant states when presented with stress, but if the length of cold storage impacts the entry into dormant states on lettuce is still unknown. 30 Stresses similar to those imposed on lettuce, as well as entry into dormant states are known to impact the virulence potential of STEC. STEC on lettuce stored at 4°C for 9-10 days shows higher expression levels of virulence-related genes [224,225]. For STEC which entered the VBNC state on lettuce at 8°C, Shiga toxin was still detected even when culturable cells were not present, indicating the STEC retained the potential for virulence. [220]. Although cold temperatures can impact STEC VBNC formation and virulence, it is unclear how the length of time in cold storage on lettuce may influence virulence potential. Lettuce transported from the West coast to the East coast may be held for days at refrigeration temperature during transport prior to processing, likely impacting pathogen physiology. Physiological changes such as entering a dormant state can protect cells from environmental stress. While STEC O157:H7 can enter the persister or VBNC state on lettuce, the extent to which this occurs during harvest, cooling, and refrigerated transport is unknown. Changes in the physiological state of STEC O157:H7 have the potential to impact risk of illness associated with contaminated lettuce due to alterations in tolerance to sanitizers in wash water as well as virulence properties. The objectives of this study were to investigate the effect of the length of time in cold storage on a) the physiological state of STEC O157:H7 on Romaine lettuce; b) subsequent acid tolerance and chlorine tolerance; and c) virulence of STEC O157:H7. 3.3. Methods 3.3.1. Experimental Design 3.3.1.1. Greenhouse propagation of Romaine lettuce Romaine lettuce plants, Lactuca sativa var. Solid King (Seedway, NY) and Parris Island (Gardner’s Basics, UT) were grown in the greenhouse complex at Michigan State University. The temperature was maintained between 18-24 °C with 14/10h light/dark periods. Miracle Gro potting mixture (Scotts Miracle- Gro, Marysville, OH) was used, along with Peter’s Excel pH low 15-7-25 fertilizer (George, UT). About 45–55-day old plants were used for the studies. 3.3.1.2. STEC strains Three STEC O157:H7 strains used were obtained from Michigan Department of Health and Human Services (MDHHS) and were associated with the following outbreaks: Yuma, Arizona in 2018 (PNUSAE013458) [194], Central Coast, California in 2018 (PNUSAE019890) [196], and Salinas, California in 2019 (PNUSAE044369) [195]. 31 3.3.1.3. Selection of rifampicin resistant isolates To distinguish inoculated pathogen from native microbes of lettuce, the STEC strains were selected for resistance to rifampicin (Thermo SCIENTIFIC, Waltham, MA) by sequentially exposing them to increasing concentrations of rifampicin as described previously [226]. The strain stocks stored in 15% glycerol at -80 °C were streaked on to Luria Bertani (LB) plates (Invitrogen, Carlsbad, California) and incubated for 24h at 37 °C. A single colony was used for incubating 5 mL LB broth with 4 µg/mL rifampicin, which was incubated at 37 °C with shaking for 24h. A 100 µL aliquot was then transferred to 50 mL LB broth supplemented with 40 µg/mL rifampicin and incubated at 37 °C for 24h with shaking. Next day, a 100 µL from this culture was transferred to a 50 mL LB broth supplemented with 80 µg/mL rifampicin and grown at 37 °C with shaking for 24h. Next day, a loopful culture was streaked onto LB plates supplemented with 80 µg/mL rifampicin for confirmation of growth. From the overnight culture, 800 µL was added to 200 µL of 75% glycerol and stored at -80 °C for use throughout the experiments. 3.3.1.4. Spray inoculation of lettuce For every experiment, rifampicin-resistant bacterial stock cultures from -80 °C were streaked on to tryptic soy agar plates (Neogen, MI) supplemented with 80 µg/mL rifampicin (TSA+ rif) and incubated for 24h at 37 °C. A single colony from the plates was used to inoculate 5 mL LB+ rifampicin broth, which was then incubated at 37 °C for 18h with shaking at 150 rpm. For subsequent inoculation, a 100 µL culture was inoculated into 100 mL LB+ rifampicin, and incubated for 18h at 37 °C. The culture was centrifuged at 8000 rpm for 15 min at 4 °C, and the pellet was suspended in 50 mL PBS. A hand-held sprayer (Miles Scientific, Newark, DE) was used to spray inoculate the Romaine lettuce plants individually in pots in a biosafety cabinet, with a target of 5.5-6 log CFU/g of STEC O157:H7. It was left to dry for 2 min, and once all the plants were inoculated, they were transferred to a Caron plant growth chamber model 7303-22-1 (Caron, OH) at 17 °C for 24h. Twenty-one plants were inoculated for each strain, each biological replicate, and each harvest temperature. 3.3.1.5. Lettuce harvesting After 24h at 17 °C in the growth chamber, lettuce plants in pots were removed and the head of lettuce was harvested by cutting at the base with a sterile knife. After harvesting, lettuce heads were placed into 81.2 cm ´ 50.8 cm ´ 25.4 cm grated plastic bins (Grainger, Lake Forest, Illinois) lined with a plastic bag, and these grated bins were wrapped loosely with another plastic bag. They were then returned to the Caron growth chamber at the harvest temperature (9 °C or 17 °C) for 4h. Harvest temperatures and time were determined from industry supplied data (see section 2.3). 32 3.3.1.6. Cold storage The bins with harvested lettuce were transferred to cold storage for 5d at 2.2 °C. Temperature data was collected from the time of harvest to the last day of cold storage using an external stainless steel temperature probe (Thermo SCIENTIFIC, Waltham, MA). 3.3.2. Sample collection Lettuce samples were collected at the time of inoculation to enumerate culturable cells. For every sampling point, one head of lettuce was used as a unit. A sterile knife was used to chop lettuce on a sterile cutting board. It was first sliced horizontally, then diagonally twice. Samples were collected before harvest, after harvest, and every day at the same time during each day of 5 days of cold storage. The parameters measured at these time points were culturable cells, injured cells, persister cells, VBNC cells, acid tolerance, chlorine tolerance, and virulence. 3.3.2.1. Culturable cells For measuring culturable cells, 10 g of inoculated lettuce was homogenized in Seward STOMACHER 400 (Seward, Worthing,UK) for 1 min in a 7.5”X12” filter Whirl-Pak stomacher bag containing 90 mL of phosphate buffer saline (PBS) solution at pH 7.2. A 100 µL was used to make appropriate dilutions and plated onto TSA+ rif plates. Plates were incubated at 37°C for 18-20 hours. 3.3.2.2. Injured cells Injured cells were measured at every sampling point. A 100 µL sample was taken from the PBS and inoculated lettuce slurry from 2.2.1, and appropriate dilutions were made. Dilutions were plated onto selective media of MacConkey Sorbitol + rif on plates. Plates were incubated at 37°C for 18-20 hours. The cells that were counted using MacConkey Sorbitol + rif plates were the cells that were not injured. The cells that didn’t survive out of a total of culturable cells were injured cells. Injured cells were calculated by subtracting the number of surviving cells from the number of culturable cells and expressed as a percentage. 3.3.2.3. Persister cells Minimum inhibitory concentration (MIC) of antibiotic ciprofloxacin for the STEC O157:H7 strains was determined based on the concentration of ciprofloxacin where cell growth did not occur. This was measured by measuring turbidity during 18h incubation at 37 °C using plate reader (Molecular devices, San Jose, CA). MIC was measured for each strain individually. MIC for PNUSAE013458 is 0.064ng/µL, for PNUSAE019890 and PNUSAE044369 is 0.128ng/µL. 10X MIC for the identification of persister cells is calculated to be 0.64ng/µl for PNUSAE013458, and 1.28 ng/µl for PNUSAE019890 and PNUSAE044369. The methodology for measuring persister cells was adapted from the method described by Thao et al. [42]. An aliquot of 1 mL in an Eppendorf tube was taken from the homogenate of 10 g of inoculated 33 lettuce and PBS from 2.1 and was exposed to 10X MIC of ciprofloxacin. This was done by adding 5 µL of 1000X concentrated stock solution of ciprofloxacin, i.e., 0.64 mg/mL or 1.28 mg/mL depending upon the strain. This mixture was incubated at 37 °C for 3h with shaking at 150 rpm. Later, a 3 mL syringe with a removable filter head (Sterilitech, Aurburn, WA) with a 0.22 µm filter (Isopore, Millipore Sigma, Burlington, MA) was used capture the bacterial cells. The filter was then added to 1 mL of PBS solution and vortexed (Thermofisher Scientific, Waltham, MA) at 2000 rpm for 3 min to suspend the cells in PBS solution. Appropriate dilutions were made and plated on TSA+ rif plates. 3.3.2.4. VBNC cells A 10 mL sample was taken from inoculated lettuce and PBS homogenate from 2.1 and centrifuged at 10,000 rpm at 8 °C for 10 min. The supernatant was discarded, and 1 mL of PBS solution was added to suspend the cells by mixing with a pipette and shaking. To make the final concentration of 50 µM, 3.125 µL of propidium monoazide (PMA) (Biotium, US), and 250 µL of PMA enhancer (Biotium, US), was added to the cell suspension. The samples were stored in the dark before exposure to UV light for 30 min in a PMA-LiteTM device (Biotium, Fremont, CA). This is a LED light box that photoactivates the samples treated with photoreactive devices such as PMA, that crosslinks with dead cell DNA, and will only let live cell DNA amplify. Every 10 min, the tubes that were kept in the PMA-LiteTM device were vortexed in minimum lighting at 2000 rpm for 5s. Powerlyzer kit (Qiagen, MD) was used for DNA extraction. For the samples to be sent to Michigan State University (MSU) core genomics facility for quantitative PCR (qpCR), the samples were prepared by adding SYBR green (Qiagen,MD), qPCR grade water (ThermoSCIENTIFIC), primers targeting ORF Z3276 gene [227] viz., forward- 5′-GCACTAAAAGCTTGGAGCAGTTC and reverse- 5′- AACAATGGGTCAGCGGTAAGGCTA in the ratio of 10:6:2. A 2 µL sample was added to this working solution of 18 µL, which was read in QuantStudio 7 Flex qPCR (Thermo Fisher SCIENTIFIC, Waltham, MA) in MSU core genome facility. This data is for calculating the total number of viable cells using standard curves (R2>90%) prepared before with known cell concentrations. 3.3.2.5. Acid tolerance The method utilizes synthetic gastric fluid (SGF) [228] at pH 2.5. The SGF is freshly prepared one hour prior to the experiment and serves as a representative simulation of the acidic environment found in the human stomach. Ten grams of lettuce was added to a stomacher bag and 90 mL SGF was added to that bag. It was homogenized for 1 min at normal speed in filter bag in a Seward STOMACHER 400 (Weber SCIENTIFIC, NJ). Serial dilutions were made and plated on TSA+ rif plates, to enumerate the initial level of cells. These inoculated lettuce containing bags were incubated at 37 °C for 2h. After incubation, 34 appropriate dilutions were made from the bag and plated on TSA+ rif plates which were incubated at 37°C for 24 h before enumeration. 3.3.2.6. Chlorine tolerance For test samples, 40 g of inoculated lettuce was added in a mesh bag and the bag was dropped in a 1 L of 0.05M KH2PO4 solution and exposed to 20-25 ppm of chlorine solution (XY-12) for 30s. Free chlorine was measured using an ORION AQUAfast AQ3700 instrument (Thermo SCIENTIFIC, Waltham, MA) with the test kit-AC2071. Chlorine was measured before and after the lettuce bag was dropped in 1 L solution. After chlorine exposure, the lettuce in mesh bag was placed in a stomacher bag, and 90.4 µL solution of sodium thiosulfate was added to neutralize the chlorine. For negative control, forty grams of inoculated lettuce was added to 160 mL of 0.05M KH2PO4 solution. Appropriate dilutions were made and plated on TSA+ rif plates which were incubated at 37°C for 24 h before enumeration. 3.3.2.7. Virulence assays Virulence of STEC O157:H7 strains were determined using killing assays of Galleria mellonella larvae (Speedway, WI). These methods were adapted from another research paper [229]. G. mellonella larvae that arrived within 5-7 days were used for the assay and were stored at 15 °C until the time of experiments. To do the baseline assays for determining LD50 (lethal dose at which 50% population is killed), G. mellonella larvae were injected with each bacterial strain. In this case, a total of ten larvae were used for each strain, each harvest temperature, and each of the three biological replicates. The inoculum was prepared by streaking the bacterial freezer stocks to a TSA+ rif plate, overnight incubation at 37 °C, inoculating 5 mL TSB with one single colony from the overnight grown culture, incubating the inoculated broth at 37 °C for 18h at 150 rpm shaking, centrifuging at 8000 rpm for 10 mins, and resuspending in PBS solution. Bacterial dilutions were made ranging from 102 to 108 cells/mL for larvae inoculations. Larvae were inoculated at the 3rd posterior pro-leg with insulin syringes with a TRIDAK STEPPER. Ten larvae each were stabbed with an empty needle, and with PBS solution were used as negative controls. After inoculation, the larvae were incubated at 37 °C for 5d in the dark. They were scored every 24h ± 0.5h for taking the count of live larvae that were pale yellow. The dead black larvae were taken out every time while counting. These larvae did not respond to any stimulation [230]. This baseline data was collected for each strain in three biological replicates, which was later used as an input for Probit regression model analysis to calculate the LD50 [231]. LD50 values were used to choose the optimum inoculum during the experiment for three strains, and two harvest temperatures, which was 20mL sample and a 1/10th dilution of sample. The LD50 values for Yuma 2018, Central Coast 2018 and Salinas 2019 strains were 4.5, 6, and 6.7 log CFU/g respectively. During the experiments, from the inoculated lettuce and PBS homogenate from 2.2.1, 1 mL and 1/10th dilution of the same was used to inoculate larvae with 20 µL solutions. Larvae stabbed with 35 an empty insulin needle, and with PBS solution were used as negative controls. The virulence assays were conducted in three biological replicates for each strain, and each harvest temperature. 3.3.3. Industry supplied data Time and temperature profile data was sourced from an industry partner data detailing the harvesting, transport and cooling conditions of Romaine lettuce. The time and temperature profile data were collected for 3 years and 2 months (February 2016 to April 2019). Of the collected datapoints, 5615 were for Romaine lettuce grown and harvested in the Salinas, California growing region, and 4623 for the Yuma, Arizona growing region. The data were analyzed in the R programming language [232] for summaries and probability distributions of the data. The temperature and times for harvesting, transportation and cooling were first checked for outliers and distributions using boxplots and histograms, respectively. Using the group function, the median, upper and lower quartile of these data was generated for the different locations. Differences in the time and temperature data for the two locations were evaluated using pairwise tests, and where the data was not normal, Mood’s median test was used. Time taken to cool the lettuce at the cooling center was generated from the difference between the time at the start of cooling and at the end of cooling. Time taken from harvest to cooling was computed by calculating the difference between the time at the end of harvesting and the start of cooling. 3.3.4. Data analysis 3.3.4.1. Generating standard curve for the VBNC cells To generate standard curves for qPCR, different bacterial concentrations (108, 107, 106, 105 and 104) were prepared. Culture was prepared as described in 1.4 and then diluted to make appropriate dilutions. Above mentioned dilutions of the bacteria were added to stomacher bags containing 10 g of lettuce. PBS was added to make the liquid portion up to 90 mL including the inoculations. Samples were homogenized at normal speed for 1 min in Seward STOMACHER 400 (Weber SCIENTIFIC, NJ), and 10 mL sample was collected, which was then centrifuged at 10000 rpm for 10 mins at 8°C. The pellet was suspended in 1 mL of PBS, 250 µL of PMA enhancer (Biotium, US), and 3.125 µL and 4.688 µL of PMAxx dye (Biotium, US) to make the final concentration of the dye to be 50 µM and 75 µM respectively out of 20 mM stock solution. Samples were held in the dark for 10 min, exposed to UV light for 30 min with vortexing after every 10 min at 2000 rpm for 5s. DNA extraction and qPCR master mix preparation was done as described in 2.3, and sent to MSU core Genomics for QuantStudio 7 Flex qPCR (Thermo Fisher Scientific). Samples at concentrations of 108, 106, and 104 were plated to collect plate count data. A standard curve (R2>90%) with plate count data against qPCR data was prepared for each strain comprising 3 biological replicates. 36 3.3.4.2. Statistical analyses All the statistical analyses of the microbial data were done in R programing language [232]. Log Nt/No or counts at (log) at time (t) days from the baseline (time 0 days) were calculated using the culturable counts to describe changes due to chlorine and acid. variables, the fraction of VBNC and percentages of persister states were also calculated. The length of cold storage, harvest temperatures, and each strain were analyzed as independent variables. Package dplyr was used to group the response variables by independent variables [233]. Normality of the data was tested using Wilk’s Shapiro test and outliers were visualized using box plots. Correlation between microbial counts and time, VBNC fraction and time, and persister percentage and time was analyzed. ANOVA was used to test if harvesting conditions at different temperatures and strains influenced the culturable counts, injured cell percentages, persister cell percentages, VBNC counts, acid tolerance, and chlorine tolerance. Two-way analysis of covariance (ANCOVA) was used to test the main effects and the interaction of harvest temperature and strain type on the microbial behavior. Estimated marginal means (emmeans) of the Package emmeans [234] was utilized to separate statistically different (p<0.05) means. Regression tree analysis was done to check the effect of independent variables on microbial behavior [235]. Virulence was analyzed using binomial logistic regression, with survival as count for pathogen that were alive after 4 days and all three independent variables were evaluated. 3.4. Results 3.4.1. Harvest temperatures and cut-to-cool times of Romaine lettuce from Salinas and Yuma growing regions The median temperature of Romaine lettuce in bins at the time of harvest was 13.55 °C (min=- 10.00, max=31.77 ℃). Lettuce harvested from the Salinas growing region had statistically significantly higher median bin temperatures, 14.4 ℃, than those from the Yuma growing region, 12.00 ℃ (p<0.001, Mood’s median test, Figure 3.1A). The median temperature at the end of cooling of lettuce from the Salinas region was 1.38 °C, which was statistically different from that of the Yuma region, which was 1.50°C (p<0.001, Mood’s Median test, Figure 3.1B). The time from the end of harvest to the start of cooling is called the cut-to-cool time. The cut-to-cool time for Romaine lettuce harvested from the Salinas growing region (163 min) was significantly lower (p<0.001, Mood’s median test) than that for lettuce from the Yuma growing region (232 min), see Figure 3.2A. While statistically different (p<0.001, Mood’s median test, Figure. 3.2B), the median time taken to cool Romaine lettuce from the Yuma growing region was 33 min similar to that from the Salinas growing region at 32 min. 37 Figure 3.1. Average temperature of Romaine lettuce in the bins at harvest (A) and after cooling (B). Boxplots represent 5615 and 4623 datapoints collected from the Salinas and Yuma growing regions, respectively. Figure 3.2. Harvest-to-cool time (A) and time taken to cool Romaine lettuce at the cooling plants (B). 38 Figure 3.2. (cont’d) This data is for Romaine lettuce harvested from two different growing regions. Boxplots represent 5615 and 4623 datapoint collected from Salinas and Yuma growing regions, respectively. 3.4.2. Behavior of STEC O157:H7 on inoculated Romaine lettuce at the two harvest temperatures from cut to cool 3.4.2.1. Transitions in physiological states of STEC O157:H7 on inoculated lettuce To assess the effects of harvest temperature on STEC O157:H7 physiology, we selected the harvest temperatures representing the 25th and 75th percentiles from the Yuma and Salinas growing regions, respectively. For the experiment, the cut-to-cool time of 240 minutes was selected, which was similar to the median of cut-to-cool time for Romaine lettuce from the Yuma growing region. The greenhouse grown lettuce was inoculated with the different STEC O157:H7 strains and lettuce was incubated at the selected harvest temperatures for 240 min before cooling. Harvested Romaine lettuce had geometric mean of culturable counts of STEC O157:H7 of 4.25- 6.57 and 4.16-5.02 log CFU/g for the three strains at 9 and 17℃, respectively. The average percentages of injured cells were 26.5-81.1% and 58.6-72.7%, the average percentage of cells in the persister state was 2.4-20.0 and 6.7-18.9, and the average VBNC cell counts were -0.47-0.54 and -0.37-0.99 log CFU/g after harvest at 9 and 17 ℃, respectively. The effect of harvest temperature and strain on the physiological state was statistically evaluated using factorial ANOVA tests, accounting for the interactions. For changes in culturable and injured cells, significant differences (p<0.05, ANOVA) were observed among the strains (Figure 3.3A and 3B). The Yuma outbreak strain had greater changes in culturable, and injured cells compared to the Central Coast outbreak strain (Figure B. 3). The change in persister percentages and VBNC counts were dependent on strain and harvest temperature, as evidenced by a significant interaction effect (p<0.05, ANOVA; Figures B. 1 and B. 2). The Salinas 2019 outbreak strain had a significantly higher persister percentage and amount VBNC cells compared to that of the Central Coast 2018 outbreak strain on lettuce harvested at 9 ℃ (Figure 3.3C and D). For lettuce harvested at 17 ℃, the Yuma outbreak strain had a higher amount of VBNC cells compared to the Central Coast 2019 outbreak strain. 39 Figure 3.3. Physiological state of STEC O157:H7 during harvest. Boxplots representing (A) change in culturable cells, (B) injured cell percentage, (C) persister cell percentage, and (D) VBNC cells at 4h after harvest. X- axis shows harvest temperature. Culturable cells and VBNC cells are shown as change in log CFU/g, whereas persister cells, and injured cells are shown as change in percentages on Y-axis. Solid boxplots represent 25th and 75th percentile of the data. Outliers are shown by solid dots. Each boxplot represents one strain with at least three biological replicates and two technical replicates. 40 3.4.2.2. Effect of harvest temperature on acid and chlorine tolerance of STEC O157:H7 on inoculated lettuce To quantify changes in acid and chlorine tolerance, the change in log values as Nt/ N0 was over the 4h period was calculated. The change in log reduction was computed as the difference of the LRV at harvest 0h and after 4h of incubation at harvest temperatures (9 and 17 °C). Higher median values for changes in log reduction were reported for the Central Coast outbreak strain for both chlorine and acid tolerance compared to the other 2 strains (Figure 3.4). Despite the median differences, strain and harvest temperature did not significantly (p>0.05, ANOVA) effect the change in log reduction due to acid and chlorine treatments. The interaction of harvest temperature and strain did not significantly (p>0.05, ANOVA) affect the change in log reduction due to acid nor chlorine treatments. Figure 3.4. STEC O157:H7 log reduction due to acid treatment. Boxplots representing (from top) change in log reduction of culturable cells due to (A) exposure to pH 2.5 for 2h, and (B) exposure to chlorine for 30s. X- axis shows harvest temperature. Y- axis shows the change in reduction in log CFU/g. Solid boxplots represents 25th and 75th percentile of the data. Outliers are shown 41 Figure 3.4. (cont’d) by solid dots. Each boxplot represents one strain with at least three biological replicates and two technical replicates. 3.4.3. Transitions in physiological states of STEC O157:H7 on Romaine lettuce during cold storage 3.4.3.1. Changes in culturable cells during cold storage In computing the changes in STEC O157:H7 culturable cells on Romaine lettuce during cold storage, the log reduction over the period was calculated with day 0 as the baseline. Strains from the Central Coast outbreak and Yuma outbreak consistently showed higher median log reduction values over the cold storage period for Romaine lettuce harvested at 9 and 17 ℃ compared to day 0 (Figure 3.5A). Correlation analysis of the log reduction in culturable cells over the period of storage showed a weak relationship between reduction and harvest temperature at 9 (r=-0.1) and 17 ℃ (r=-0.1). Analysis of covariance (ANCOVA) with days as a covariate was used to assess the effect of the interaction of harvest temperature and strain on the culturable cells during cold storage. The interaction of harvest temperature and strain did not significantly (p>0.05) effect the levels of the culturable cells on stored lettuce. However, the main effects of strain and harvest temperatures significantly (p<0.05, ANCOVA) influenced the change in culturable cells. The Central Coast outbreak strain had significantly (p<0.05, ANCOVA) higher log reduction values compared to other strains over the period of cold storage for lettuce harvested at both temperatures (Figure C. 4). For the other strains, harvest temperature had no significant (p>0.05, ANCOVA) effect on the log reduction of culturable cells during cold storage. 42 Figure 3.5. Physiological state of STEC O157:H7 during cold storage. 43 Figure 3.5. (cont’d) Boxplots representing culturable cells (A), injured cell percentage (B), persister cell percentage (C), and VBNC cells (D) during 5 days of cold storage. Culturable cells and VBNC cells are shown as reduction (log CFU/g), and persister cells, and injured cells are shown as percentages (%) on Y-axis. Solid boxplots represent the 25th and 75th percentile of the data. Outliers are shown by solid dots. Each boxplot represents one strain with at least three biological replicates and two technical replicates. To assess the effect of harvest temperature, cold storage days, strains and their interactions on log reduction of culturable cells, a regression tree analysis was used. Optimal pruning of the tree was done using a defined complexity parameter (CP), known to determine how deep a tree will grow, and cross- validation error (xerror): a CP that minimized the xerror was selected to optimally account for explained variance. The tree had 3 splits optimally explaining the variance of culturable cells during cold storage (CP=0.01, xerror= 0.856). While regression trees have the disadvantage of variation in the splits, bootstrapping (n=100) was used to minimize variation in establishing the most important factors explaining variance in the culturable cells. Cold storage days accounted for 56.1% of the variance and was the most important factor explaining the reduction in culturable cells (Figure B. 5). Regardless of the harvest temperature, lettuce inoculated with the Central Coast 2018 outbreak strain and stored for >1 day averaged the highest log reduction (0.72 log CFU/g, probability= 0.24), followed by the Yuma 2018 outbreak strain in cold storage for same time (0.34 log CFU/g, probability=0.27), see Figure 3.6. The lowest log reduction was reported for the Salinas 2019 outbreak strain regardless of storage time (-0.11 log CFU/g, probability=0.38). Figure 3.6. Regression tree analysis for culturable cells over cold storage. 44 Figure 3.6. (cont’d) Regression tree analysis of the combined effect of harvest temperature, strains and cold storage period on the log reduction of culturable cells over cold storage period. The criteria used in splitting the trees are indicated for each decision node and the relative probability of each decision indicated as a %. 3.4.3.2. Transition of STEC O157:H7 cells to the injured state during cold storage The median values of injured cells increased from 66.4% (95% CI= 31.6%,91.1%) at 0 days to 72.7 % (95% CI= 41.9%, 96.9%) after 5 days in cold storage. Whereas the median value of injured cells on lettuce harvested at 9 ℃ was 62.5% (95% CI= 26.8%, 94.6%), those on lettuce harvested at 17 ℃ was 71.6% (95% CI= 27.9%, 95.8%). The median values of injured cells for strains from the Central Coast, Salinas and Yuma outbreaks were 52.3% (95% CI= 21.0%, 78.7%), 78.6% (95% CI= 32.7%, 97.1%) and 65.1 % (95% CI= 28.4%, 86.9%), respectively. The strain from the Salinas outbreak had relatively higher median values for the percentage of injured cells than those from the Central Coast and Yuma outbreak strains for both harvest temperatures (Figure 3.5B). There were weak correlations (r≤0.1) for injured cells and cold storage times at both 9 and 17 ℃ temperatures. Regression methods were used to assess the combined effect of harvest temperature, strain and cold storage days on the percentage of injured cells. Different combinations of factors were incorporated into regression models, and the best fitting combinations were selected using the corrected Akaike Information Criterion (AICc). The selected model inputs were evaluated for their effect on the injured cells. Strain significantly (p<0.001) affected the transition to injured cells during cold storage (Figure 3.7). The Salinas outbreak strain had significantly (p<0.05) more cells transition into injured cells than that from the Central Coast strain. Figure 3.7. Beta coefficients from beta regression model evaluating the effect of strain, harvest temperature and cold storage period on injured cells. 45 Figure 3.7. (cont’d) The response variable, injured cells was transformed into the interval (0,1) and the factors of strain, harvest temperature and cold storage days evaluated using beta regression. Confidence interval of beta coefficient >0 or <0 denote significant (p<0.05) effect, whereas a confidence interval including 0 is not significant (p>0.05). Strain was transformed into a dummy variable with Central Coast 2018 outbreak strain as a baseline category 3.4.3.3. Transition of microbial cells into the persister state during cold storage The median persister percentage increased from 8.78% (95% CI= 0.7%, 42.0%) on day 0 to 19.6% (95% CI= 1.5%, 74.4%) after 5 days of cold storage. The median values of persister cells on lettuce that was harvested at 9 and 17 ℃ was 24.4% (95% CI= 2.03%, 78.2%) and 13.3% (95% CI= 0.8%, 64.9%), respectively. The outbreak strains from Central Coast 2018, Salinas 2019 and Yuma 2018 had median persister percentage of 18.6% (95% CI= 1.7%, 68.3%), 24.6% (95% CI= 2.2%, 73.7%) and 6.57% (95% CI= 0.8%, 63.6%), respectively over the cold storage period. Lettuce harvested at 9 ℃ had relatively higher persister percentage for all strains over the cold storage period compared to that harvested at 17 ℃ (Figure 3.5C). Spearman rank correlation tests showed weak correlation between persister percentage and cold storage days for lettuce that was harvested either at 9 ℃ (r=0.12) or 17 ℃ (r=0.14). Regression analysis was conducted to evaluate the combined effect of strain, harvest temperature and cold storage period on transition to persister state. With increasing cold storage days, the persister formation also increased (beta=0.082, p=0.009), see Figure 3.8. On the other hand, increasing harvest temperature had a negative impact (beta=-0.046, p=0.002) on persister formation. Strain differences also affected the formation of persisters. Fewer persister cells were formed by the Yuma 2018 outbreak strain than Central Coast 2018 outbreak strain (beta=-0.046, p<0.001). The interaction of the factors did not significantly (p>0.05) affect the formation of persisters during cold storage. 46 Figure 3.8. Effect of harvest temperature, strain and cold storage period on the formation of persister cells in lettuce. Persister fraction was presented as a proportion (0, 1) and the analysis was conducted using beta regression function. Significant (*) values <0 indicate decreasing persister formation with increasing values, whereas significant values >0 indicate increasing persister formations with the values. Strain was transformed into a dummy variable with Central Coast 2018 as a baseline category. 3.4.3.4. Transition of microbial cells to the VBNC state during cold storage Log increase for the VBNC cells during the cold storage period ranged from -3.62 to 3.90. Whereas on lettuce harvested at 9 ℃, there was an increase of 0.002 log CFU/g in VBNC over the cold storage period, lettuce harvested at 17 ℃ had -0.01 log. Among the strains, Central Coast 2018 outbreak strain had the highest increase of 0.284 logs across the period of storage, Yuma 2018 and Salinas 2019 outbreak strains had -0.269 and -0.003 logs. Over cold storage, the Central Coast 2018 outbreak strain had a relatively higher increase in transition to VBNC cells than other strains (Figure 3.5D). Correlation tests for trends of transition to VBNC showed that period of storage was not correlated to formation of the dormant cells at either harvest temperatures, 9 ℃ (r=0.03) and 17 ℃ (r=-0.11). ANCOVA tests showed that the interaction of harvest temperature and strain significantly (p<0.001) affected the formation of VBNC cells. The Central Coast 2018 outbreak strain had increased VBNC formation during cold storage compared to the Yuma 2018 and Salinas 2019 outbreak strains (Figure B. 6). The Yuma 2018 outbreak strain showed higher formation of VBNC cells at 9 ℃ than at 17 ℃. Regression tree analysis showed that the variance in VBNC formation was optimally explained by 8 splits (CP=0.01, xerror=0.78; Figure 3.9). The highest formation of VBNC cells during cold storage occurred for Central Coast 2018 outbreak strain, stored for ≥1 day on lettuce harvested at 17 ℃ (1 log, probability 0.12). The least formation of VBNC cells (-1.1 logs, probability=0.19) was reported for Yuma 2018 outbreak on produce that was harvested at 17 ℃ and stored for ≥5 days. Cold storage days accounted for 47.6% of the variance in the formation of VBNC cells (Figure B. 5). 47 Figure 3.9. Regression tree analysis for VBNC cells over cold storage. Regression tree analysis of the combined effect of harvest temperature, strains and cold storage period on the formation of VBNC cells over cold storage period. The criteria used in splitting the trees are indicated for each decision node and the relative probability of each decision indicated as a %. 3.4.4. Acid and chlorine tolerance of STEC O157:H7 on inoculated lettuce during cold storage The median log reduction due to acid treatment for all the strains increased from 0.18 (95% CI=- 0.48,0.97) on day 0 to 0.32 (95% CI=-0.35, 1.37) at 5 days of cold storage. The Salinas 2019 outbreak strain had higher median log reduction values on lettuce harvested at 9 ℃ over the cold storage days than either the Yuma 2018 or Central Coast 2018 outbreak strains on lettuce at same harvest temperature (Figure 3.10). There was weak to no correlations between cold storage days and log reduction due to acid treatment, for lettuce harvested at either 9 ℃ (r= 0.28) or 17 ℃ (r=0.02). The interaction of harvest temperature and strain significantly (p<0.001, ANCOVA) affected the log reduction due to acid treatment during cold storage. The Salinas 2019 outbreak strain had significantly higher log reduction at 9 ℃ (0.81) than at 17 ℃ (0.07). 48 Regression tree analysis optimally explained the variance in the increase in log reduction in 6 splits (CP=0.017, xerror=0.950). The Salinas 2019 outbreak strain on lettuce harvested at 9 ℃ had the highest increase (1.1, probability=0.07) under cold storage (≥2 days), see Figure 3.11. The Central Coast 2018 outbreak strain had the lowest log reduction due to acid treatment during cold storage (0.11, probability=0.43). Cold storage day explained the highest variance (49.3%) of change in log reduction due to acid treatment (Figure B. 5). Figure 3.10. STEC O157:H7 reduction due to acid (A) and Chlorine (B). Boxplots of the changes in acid (A) and Chlorine (B) tolerance of the STEC O157:H7 cells in leafy greens stored under cold temperature over 5 days. Acid and chlorine tolerance were measured as difference in log reduction due to acid and chlorine treatment, respectively, from 0 days of cold storage. Solid boxplots represent 25th and 75th percentile of the data. Outliers are shown by solid dots. Each boxplot represents one strain with at least three biological replicates and two technical replicates. 49 Figure 3.11. Regression tree analysis for log reduction due to acid during cold storage. Regression tree analysis of the combined effect of harvest temperature, strains and cold storage period on the change in log reduction values due to acid treatment. The criteria used in splitting the trees are indicated for each decision node and the relative probability of each decision indicated as a %. The median log reduction due to chlorine treatment increased from 0.93 (95% CI=-0.34, 1.91) on day 0 to 1.01 (95% CI=-0.00, 1.80) at 5 days of cold storage. Partial correlations accounting for the chlorine concentrations showed weak relationship between cold storage days and log reduction due to chlorine treatment at 9℃ (0.08) and 17℃ (0.19). ANCOVA analysis with chlorine concentration as a covariate showed that harvest temperature and strain did not significantly (p>0.05) affect the log reduction due to chlorine treatment. The effect of harvest temperature, strain and cold storage with chlorine concentration as a covariate were assessed using regression trees. Produce that was inoculated with outbreak strain Salinas 2019 and Yuma 2018 and stored for >2 days in cold storage was attained in chlorine washing (>20 ppm) had the highest log reduction (1.8, probability=0.07). Produce that was stored for ≥2 days in the cold had higher log reduction than those stored in the cold of <2 days. Concentration of chlorine (41.9%) and cold 50 storage days (31.3%) accounted for the highest variance in the change in log reduction due to chlorine treatment (Figure B. 5). Figure 3.12. Regression tree analysis for log reduction values due to chlorine treatment. Regression tree analysis of the combined effect of harvest temperature, strains and cold storage period on the change in log reduction values due to chlorine treatment. The criteria used in splitting the trees are indicated for each decision node and the relative probability of each decision indicated as a % 3.4.5. Impact of cold storage and harvest temperature on virulence of STEC O157:H7 Binomial logistic regression with count for survival of the G. mellonella larvae as the response variable was used to evaluate the effect of strain, harvest temperature and cold storage time on virulence. Strain specific differences from lettuce harvested at different harvesting temperatures over the cold storage period are as shown in Figure 3.13. The Central Coast 2018 outbreak strain on lettuce harvested at 9 ℃ and stored for 0 days was the baseline group. Larvae from the baseline group (day 0) waere more likely (odds=3.76, p<0.001) to survive compared to larvae inoculated with STEC from lettuce held in cold storage over extended periods of time. Lower survival of the larvae (Odds=0.276, p=0.015) occurred when the STEC O157:H7 inoculum was from lettuce harvested at 17 ℃ rather than at 9 ℃ (Figure C. 9). The time in cold storage and strain differences did not significantly (p>0.05) affect the larvae survival rates. 51 Interaction was reported between the strains and harvesting temperatures. Better survival (odds=16.8, p=0.001), of larvae was seen, when they were inoculated with the Salinas 2019 outbreak strain from lettuce harvested at 17 ℃ compared to the baseline group. Figure 3.13. Survival of STEC O157:H7 inoculated G. mellonella larvae. Survival of G. mellonella larvae inoculated with STEC O157:H7 from produce harvested at two different temperatures and stored in the cold over five days. The points are the observed values of the proportion of the surviving G. mellonella larvae that whereas the line are the fitted predicted values from the binomial logistic regression model. 3.5. Discussion 3.5.1. Changes in physiology during cut-to-cool were minimal, while significant differences in physiology occurred during cold storage Cold storage is used for storage and transportation of perishable foods including fruits and vegetables [236–238]. Refrigeration poses cold stress [239], so any pathogens present on those foods need to adapt to cold stress in order to survive. The optimal growth temperature of E. coli is 37°C, and its growth is impaired under 21°C and stops at 7.5°C [240]. When the temperature shifts from optimal to low, bacterial 52 cell growth is arrested [241]. During that temperature shift, phenotypic changes in bacteria happen such as alterations in cell membrane, change in translational and transcriptional machinery, and production of cold shock acclimation proteins [242]. In STEC O157:H7, activation of genes related to the barrier function of the outer membrane, polysaccharide synthesis, and curli production increases [243]. In addition to these responses, STEC can enter dormant states on exposure to lower temperatures [222,244]. In inoculated field water at 15°C, approximately 10-23% of the overall population of four STEC strains were seen to transform into persister cells [222]. When inoculated onto lettuce leaves, STEC O157:H7 maintained culturability at 16°C, but entered the VBNC state when lettuce was stored at 8°C [244]. Here, we utilized 2°C to represent refrigerated storage of lettuce. The impacts of temperature shifts were evaluated by the two different harvest temperatures, which did have an effect on the entry of different STEC O157:H7 strains into dormant states. When inoculated lettuce was harvested at 9°C compared to at 17°C, a greater proportion of the Salinas 2019 outbreak strain entered the persister and VBNC states during cold storage, a trend similarly observed for VBNC cell formation in the Yuma 2018 outbreak strain. It may be that exposure to a lower temperature during harvest could prime the pathogen to better adapt to cold stress, by enabling entry into dormant states. Taken together, our data and previous research demonstrate that STEC O157:H7 enters dormant states as part of the adaptation to cold stress. When bacteria experiences cold stress, translation efficiency is reduced due to RNA folding, with other changes such as reduced membrane permeability, lower substrate diffusion rates, and disrupted ion equilibrium [245–247]. Activation of the general stress response occurs when temperatures are shifted from 37-35°C to 25-14°C [248–250]. Various cold shock proteins (CSPs) are involved with adaptation to cold temperature. A class of CSPs, called the RNA chaperones, keep the RNA single-stranded during initial stages of cold stress [247,251]. At temperatures of 4°C and 18°C, cspA and cspG respectively are reported to be upregulated in STEC O157:H7 in lettuce lysates [252,253]. Expression of cspH and survival of persisters is known to be associated [70]. CspD is regulated by (p)ppGpp independently of RpoS [255] and is also known to play a role in persister cell formation [256]. This could be one of the pathways that STEC O157:H7 utilizes to enter into the persister state in response to cold stress. 3.5.2. Increasing cold storage time led to reduced acid tolerance STEC O157:H7 is very tolerant to low pH [257–259], with minimal decrease in cell numbers after 5h at pH 2.5 [239]. Cold stress negatively impacts acid tolerance in STEC O157:H7, as 4 weeks of cold stress of 4°C and pH 5.5 prior to exposure to pH 1.5-2 in tryptic soy broth, led to decreased tolerance [260]. In this study, STEC O157:H7 was stressed through comparatively low pH and low temperature for 4 weeks prior to the actual treatment of ph1.5-2.0, whereas no additional stress was given in our studies prior to the acid treatment. We found that during cold storage, lettuce harvested at 9°C had higher acid tolerance 53 compared to that harvested at 17°C. With our research, we also found that cold storage led to decreased tolerance to pH 2.5, where cold storage of 2 days or more led to higher log reduction values for the Salinas 2019 outbreak strain, showing that acid tolerance decreases as duration of cold storage increases. This is in line with another study that shows that STEC O157:H7 on cut lettuce survived the treatment of SGF at pH 2.0 [261]. 3.5.3. Increasing cold storage time did not influence chlorine tolerance STEC O157:H7 can survive on leafy greens even after washing with chlorine-based sanitizers [104,216,262–264]. The duration that the pathogen remains on pre-harvest lettuce leaves has been shown to increase tolerance to post-harvest chlorine washing. A 90s chlorine exposure had 3.79-5.37 log reduction of STEC O157:H7 in wash water of inoculated fresh-cut leafy greens [216]. In a study where lettuce was inoculated with STEC O157:H7, significantly lower reductions in a chlorine-based wash water were seen at 3- and 5-days post-inoculation compared to 1-day post-inoculation [262]. In our study, we found that STEC O157:H7 on lettuce stored in the cold for 5 days was able to survive chlorine treatment, with change in log reductions ranging from 0.58-1.8 log CFU/g, and that chlorine tolerance varied over the length of cold storage. 3.5.4. Increasing cold storage time led to decreased virulence in G. mellonella Even as sudden temperature fluctuations lead to production of heat and cold-shock proteins to protect bacterial functions, shifts in temperature can also lead to the production of bacterial virulence factors [265]. Stress response and virulence genes such as rpoS, stx1, and osmY have been shown to be highly expressed during cold shock transition from 37°C to 4°C or 7.5°C [266,267]. At longer cold storage temperatures, for cells inoculated on leafy greens, virulence genes such as iha, stx2A, and eae have increased expression [268,269]. In lettuce lysate, a cold shock of 10 min or 1h led to increased expression of stx1A [267]. Shiga toxins have a cytotoxic effect, and cytotoxicity is an aspect of virulence that can be quantified in laboratory studies. A study in tissue culture cells showed that stressed STEC O157 cells had higher cytotoxicity than non-stressed controls [270]. This study also showed that chlorine, starvation, acid, or oxidative stressed STEC O157 cells have higher survival during cold storage of 4 weeks at 8, 12, and 16°C compared to non-stressed cells [270]. We assessed relative virulence in our study with the insect model Galleria mellonella. Mammalian models of infection can raise ethical issues, generally are limited to small samples sizes, and are costly, and insect models provide a valuable alternative. G. mellonella is a common insect model that is used to investigate bacterial pathogenesis [270,271]. Our data shows that Central Coast 2018 outbreak strain had a positive relationship between survival of larvae and cold storage, meaning that the virulence decreased with 54 cold storage. This means that cold transportation of lettuce decreases virulence of STEC O157:H7 if it happened to contaminate lettuce. Our results are different than what the previous study reported on cytotoxicity for stressed STEC O157 cells, though it is important to note that the previous research was conducted in cell culture rather than a more complex model host [270]. This might be due to the reason that we held inoculated lettuce at 2.2°C for five days, and STEC didn’t go through the stressors before cold storage that the study described. Previous studies have shown that cold temperature leads to increased virulence gene expression, but only the effect of short-term shock has been evaluated on virulence phenotypes. In our study, we used long-term exposure of STEC to cold stress ranging from 1-5 days and virulence phenotypes were measured in an invertebrate model of infection. 3.5.5. Strains affected the variability in the physiological, tolerance, and virulence response of the strains The strains that we used belong to different phylogenetic categories. Central Coast 2018 and Salinas 2019 outbreak strains are more closely related to each other than to the Yuma outbreak strain, which belongs to the REPEXH01 category [272]. Central Coast 2018 outbreak strain belongs to Clade 1, and Salinas, 2019 strain belongs to Clade 2 of REPEXH02 [273]. A recent study showed that REPEXH01 strain led to lower persister cell formation and a higher VBNC cell formation compared to REPEXH02 strain [274]. A similar trend was seen with our data, where Yuma strain had lower persister cell formation, and a higher VBNC cell formation compared to Salinas strain. Central Coast 2018 outbreak strain had the least increase in LRV during chlorine treatment meaning that it was able to tolerate chlorine stress. Strain differences were also seen in a study related to STEC O157:H7 inoculated lettuce has shown that one strain inoculated on lettuce lost culturability in 7 days while the other lost culturability in 15 days with all the variables provided to them were same [244]. Differences in data due to strain have been seen previously. Previous data and our data suggest that one strain is not representative of a serotype and can behave differently under same conditions. 3.6. Conclusions Our study aimed to understand the microbial dynamics that occur when pathogenic cells such as STEC O157:H7 are exposed to temperature shifts during harvesting and storage. These findings can inform the assessment of risks posed by STEC O157:H7 on lettuce. We found that both low harvest temperatures and prolonged cold storage increased the transition of STEC O157:H7 into dormant states. This transition poses a risk of underestimating pathogen survival when microbial concentrations in lettuce are evaluated, which is important for more accurate risk modeling. Our study highlights the need to account for conditions that promote dormancy when using such data in QMRA. Furthermore, we demonstrated that the efficacy 55 of sanitization processes, such as chlorine washing to minimize post-harvest cross-contamination, is influenced by the exposure of pathogens to temperatures at harvest or during storage. We also established that increasing the duration of cold storage did not result in evidence of increased strain virulence. Future risk assessments should therefore not only consider the effects of harvesting and cold storage temperatures on STEC O157:H7 die-off or growth but also account for physiological changes that could otherwise lead to the underestimation of risks. We concluded that colder harvest temperature followed by cold storage led to a higher transformation of STEC O157:H7 into dormant states. The length of cold storage led to greater reduction of STEC O157:H7 due to acid treatment, but not chlorine treatment. These results demonstrate that forward processing of lettuce does not lead to increased risk of STEC O157:H7 survival or stress tolerance. 3.7. Acknowledgment We would like to thank Church Brothers Farms in Salinas, CA, for their Romaine harvest temperature data. 3.8. Funding The funding of this study was made possible by The Center for Produce Safety and the U.S. Department of Agriculture’s (USDA) Agricultural Marketing Service through grant AM21SCBPCA1002. 56 4. IMPACT OF DORMANT STATES AND ENRICHMENT PROTOCOLS ON DETECTION OF STEC O157:H7 ON ROMAINE LETTUCE Authors: Dimple Sharma1, Joshua O. Owade 2, Teresa M. Bergholz1* 1 Department of Food Science and Human Nutrition, Michigan State University, East Lansing, MI  2 Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI  * Corresponding author  4.1. Abstract Physiological changes such as dormancy can protect STEC O157:H7 cells from environmental stress, potentially reducing detectability. STEC O157:H7 has been associated with outbreaks linked to leafy greens, particularly romaine lettuce, accounting for 60.8% of all leafy green-related STEC outbreaks. This study evaluates the impact of enrichment time, the persister dormancy state, and strains of STEC O157:H7 on detection using a commercially available qPCR-based detection system. STEC O157:H7 cells were inoculated into simulated agricultural water and subsequently onto romaine lettuce, held at 4°C, and then incubated with varying enrichment times of 8 and 24 hours. Persister percentages, ranging from <1% to 85%, were calculated based on survival after antibiotic exposure. Results revealed that detection rates increased with enrichment time: 8-hour enrichment yielded positive detection rates between 4-36%, while 24-hour enrichment ranged from 66-86%, for all strains (p<0.001). We used three strains associated with Romaine lettuce outbreaks: the Central Coast, CA 2018, Salinas, CA 2019, and Yuma, AZ 2018 outbreak strains. The Yuma 2018 outbreak strain was more likely to be detected compared to other two strains (p=0.01, OR=2.2). These findings established that longer enrichment time of 24h provided a better detection of the dormant states than shorter enrichment time of 8h regardless of the strain variability. The findings also show that detection after 8 h of enrichment time varied based on the strain. 4.2. Significance and Impact of The Study Detection of pathogens is a common procedure to see check for produce contaminated. In detection protocols, an enrichment time of 8-24h in a culture media is commonly recommended, but this wide range impact recovery efficiency. Pathogens such as STEC O157:H7 (Shiga-toxin-producing Escherichia coli O157:H7) are known to exist in different physiological states, including persister state. There is potential 57 for variation among strains in recovery from different physiological states during enrichment. This study contributes to efforts to minimize outbreaks and enhance foodborne pathogen surveillance by detection methods, by taking into consideration the length of enrichment time, dormant state of pathogen, and strain variability. 4.3. Introduction Physiological changes such as entering a dormant state can protect cells from environmental stress, which may influence detectability. Recent studies have demonstrated that Shiga toxin-producing Escherichia coli (STEC) O157:H7 cells can enter a persister state, a type of dormancy characterized by resistance to environmental stresses [111,122,124,275]. According to the latest European Union surveillance report, STEC infection is the fourth most important reported zoonosis [276]. STEC infection has been linked to food commodities including vegetables, meat, and milk [125,277,278]. Among these, ready-to-eat vegetables, including leafy greens like romaine lettuce, undergo minimal processing steps to effectively reduce contamination. This makes it even more important to have methods that detect STEC in raw foods such as leafy greens. STEC has been linked to outbreaks related to leafy greens [127,193,279– 281]. Out of all leafy green outbreaks, 60.8% occur in lettuce and 19.8% of the total STEC outbreaks are associated with romaine lettuce [126]. For detection of STEC O157:H7 in food samples, culture-based methods with or without polymerase chain reaction (PCR) have been used. Standard methods for detection can be found in the Bacteriological Analytical Manual (BAM) by U.S. Food and Drug Administration (FDA) [129]. According to FDA-BAM, first samples are enriched in media, then quantitative PCR (qPCR) is used. In the process of qPCR, only genes specific to STEC O157:H7, such as for Shiga-toxin (stx1 and stx2), and beta glucuronidase (uidA) are amplified. The sample results are shown as negatives, or presumptive positives. Later, the presumptive positives are plated on specific media that allows the growth of only STEC O157:H7. FDA-BAM protocol recommends incubating samples during enrichment step for 24h [129], whereas other methods use shorter enrichment times. There are numerous commercially available, Association of Official Analytical Chemists- (AOAC) approved, methods to qualitatively test for STEC in terms of positive or negative in different food matrices including leafy greens viz. loop-mediated isothermal amplification (LAMP) [131], lateral flow immunoassay (LFIA) [132], verotoxin-enzyme linked immunosorbent assay (VT-ELISA) [132], and NeoSeekTM STEC [133]. Detection methods for STEC have limitations in detecting active toxins [138], distinguishing virulent strains [137], producing false positives [139], and not identifying emerging or non-viable variants [135]. DuPont Qualicon BAX® System [134] (now under Hygiena) has 58 been used for detection of STEC O157:H7 in lettuce. We used Hygiena BAX® System for our studies, as it has been used for detection of STEC O157:H7 in leafy greens. STEC O157:H7 can be present in lower quantities [282] than what’s required for it to be detected using any culture-based detection methods. In detection methodology, the enrichment step brings the STEC quantity to a threshold where the DNA can be amplified in PCR machines, and can be read using software. The enrichment step involves incubating test samples in growth media for a specified period to detect STEC. This duration can be less than 8h [283] or overnight [133,284], depending on the method. Some protocols specify a range of 8-24h or 12-24h [134,136]. For immunomagnetic separation (IMS), the length of enrichment time between 8-24h did not impact the detection of STEC from inoculated cattle feces [129]. IMS uses magnetic beads to separate out the target cells from the sample after enriching the sample [285], whereas culture and PCR-based methods use enrichment of bacterial cells and then detect the cells by using amplification through polymerase chain reaction (PCR) [134]. The widely used culture-based methods are meant to detect the culturable fraction of STEC cells. This technique can also culture persister cells as a subset of culturable cells without distinguishing their dormant state. Bacterial population can consist of cells that can be in dormant state like persister cells and viable but non-culturable (VBNC) cells [111]. STEC can enter into these dormant states on exposure to stress. Studies have shown that STEC can transform into persister cells or VBNC cells on leafy greens during pre-harvest conditions [122,124,275]. During post-harvest conditions, STEC can transform into persister cells or VBNC cells on exposure to cold or chlorine stress [122,286]. While we know that STEC can enter into dormant states on exposure to stress, the impact this transition has on the detectability using standard methods remains less understood. In this study, we tested detection of STEC O157:H7 on romaine lettuce using BAX® System. Changes in the physiological state of STEC O157:H7 have the potential to impact the risk of illness associated with contaminated lettuce due to alterations in detectability. Previous studies have used detectable, quantifiable vegetative bacteria for microbial populations in foods [286–297]. There is less literature present on the impact of persister state, and effect of enrichment time on detectability of STEC O157:H7 using culture- and PCR-based methods. Our goal was to study a) the effect of persister dormant state on detection of STEC O157:H7 on romaine lettuce using BAX® System, and b) effect of enrichment time of 8h and 24h in detection of STEC O157:H7 on romaine lettuce using BAX® System. 59 4.4. Methods 4.4.1. Agricultural water preparation The composition of the simulated agricultural water was developed based on average concentrations of various compounds from rivers, lakes, troughs, and fecal contaminated puddle water as reported by Avery Et. Al. [298]. The compounds were phosphate (2.6X10-5 Molar Potassium phosphate monobasic), nitrate (3.7X10-5 M potassium nitrate), ammonium (2.67X10-4 M Ammonium sulfate), calcium (9.2X10-4M Calcium chloride), magnesium (3.65X10-4 M Magnesium chloride), and sodium (2.29X10-3 M Sodium chloride). A 100 ml of agricultural water was prepared for each sample. The prepared agricultural water was sterilized by filtering through a 0.22 µm pore size filter (Isopore, Millipore Sigma, Burlington, MA) with a removable head (Sterilitech, Aurburn, WA). The method diagram is provided as Figure 4.1. Figure 4.1. Methodological approach for evaluating the effect of dormant persister state on microbial detectability. 4.4.2. Inoculation of agricultural water with STEC O157:H7 strains Three STEC strains were individually grown on Luria Bertani (LB) (Invitrogen, Carlsbad, California) media supplemented with 80 mg/mL rifampicin and incubated at 37 °C for 24h. One colony was taken using a loop and added to 5ml LB+rif broth and incubated at 37 °C for 18h. One mL was added 60 to an Eppendorf tube and centrifuged at 10,000 rpm for 15 mins. After discarding the supernatant, the pellet was suspended in 1 mL of phosphate buffered saline (PBS). Appropriate dilutions were made, and 1 mL of 108 CFU/mL was added to 99 mL of simulated agricultural water to make a total 100 mL at 106 CFU/mL with each of the three strains individually. The STEC O157:H7 strains used in this study were obtained from Michigan Department of Health and Human Services (MDHHS) and were associated with the following outbreaks: Yuma, AZ in 2018 (PNUSAE013458) [299], Central Coast, CA in 2018 (PNUSAE019890) [300], and Salinas, CA in 2019 (PNUSAE044369) [301]. The inoculated water was stored at 15 °C throughout the experiments. Enumeration of STEC O157:H7 was done by making appropriate dilutions in PBS, plating onto LB plates, and incubating for 18-20 hours at 37 °C. 4.4.3. Lettuce grown in greenhouse Romaine lettuce plants, Lactuca sativa var. Parris Island (Gardner’s Basics, UT) were grown in the greenhouse complex at Michigan State University. The growth conditions were a temperature of 18-24 °C with 14/10h light/dark periods. Miracle Gro potting mixture (Scotts Miracle-Gro, Marysville, OH) was used, along with Peter’s Excel pH low 15-7-25 fertilizer (George, UT). Approximately 50–60-day old plants were used for experiments. 4.4.4. Measurement of culturable cells A 100 µL sample from all the inoculated water samples was taken, diluted in PBS and plated onto LB media. The inoculated LB media plates were incubated at 37 °C for 18 to 20 hours. The next day, the number of colonies were counted, and cell concentrations were calculated. 4.4.5. Determination of persister percentages First, persister cells were measured by following the protocol mentioned in this section, along with measuring the culturable cells. Persister percentages were calculated by dividing the number of persister cells with culturable cells and then multiplying by 100. Persister percentages were calculated from an overall culturable cells [124]. Minimum inhibitory concentration (MIC) of antibiotic ciprofloxacin for the STEC O157:H7 strains was calculated by the concentration of ciprofloxacin at which bacterial cells were not able to grow. It was measured by turbidity during 18h incubation at 37 °C using plate reader (Molecular Devices, San Jose, CA). MIC for PNUSAE013458 strain is 0.064 ng/µL, for PNUSAE019890 and PNUSAE044369 is 0.128 ng/µL. 10X MIC for the identification of persister cells is calculated to be 0.64 ng/µl for PNUSAE013458, and 1.28 ng/µl for PNUSAE019890 and PNUSAE044369. Persister cells were calculated by exposing to 10X the concentration of MIC of ciprofloxacin for 3h at 37 °C at 150 rpm. A 3 mL syringe with a removable filter head (Sterilitech, Aurburn, WA) with a 0.22 µm filter (Isopore, Millipore Sigma, Burlington, MA) was used to capture the bacterial cells that survived the 61 antibiotic exposure. The filter was added to 1 mL of PBS solution and vortexed (ThermoFisher Scientific, Waltham, MA) at 2000 rpm for 3 min to suspend the cells in PBS. Appropriate dilutions were made and plated on LB+rif plates. Culturable cells were enumerated by making appropriate dilutions of the agricultural water and plating on LB+rif. The total number of persister cells were divided by the overall culturable cells and then multiplied by 100, giving the percentage that represents the persister population in the culturable cells. 4.4.6. Inoculation of chopped lettuce Twenty-five grams of lettuce was chopped into 2X2 inches wide squares and added to Whirlpack bags. Appropriate dilutions of the inoculated agricultural water were made to have a target of 2 and 15 cells for each lettuce sample by using the concentrations of culturable cells from a day before. After dilutions of the original agricultural water, bacterial inoculum ranging from 0.2-1 mL estimating 2 or 15 bacterial cells was added on 25 g of lettuce in each bag. These bags were kept at 2.2 °C overnight. The next day the actual MPN of STEC O157:H7 on lettuce was measured. 4.4.7. Measurement of minimum probable number (MPN) of STEC O157:H7 The inoculated lettuce in mesh stomacher bags (Thermofisher Scientific, Waltham, MA) were filled with 75 mL of LB broth for a 1:4 dilution of lettuce with LB growth media. This was homogenized using a Seward STOMACHER 400 (Weber SCIENTIFIC, NJ) for 60s at medium speed. The 1 mL 96-deep-well plates were used for this experiment. Five replicates of 1:100, 1:1000, and 1:10000 dilutions of the lettuce homogenate were grown in LB + rif media in 96 well plates at 37 °C. The growth was observed after 16- 18h for MPN estimation [302]. 4.4.8. Sample preparation for BAX testing The inoculated lettuce bags were kept at 2.2 °C for 16-18 h and sent for testing via courier to a third-party testing company. The detection experiments were done with 8 and 24-h enrichment time as suggested in BAX testing protocol [134]. 4.4.9. Preparation of controls Twenty-five grams of lettuce sample were added to bags, without any pathogen being inoculated on to them. These were also sent for detection and acted as negative controls. This was done to make sure that the lettuce was not already contaminated and that it didn’t interfere with our results. Twenty-five grams of lettuce sample were added to bags and inoculated with bacterial cells from each strain that came directly from LB culture. These samples were sent for detection and acted as positive controls. This was done to ensure if STEC O157:H7 was not detected, it was because of strain differences and not because they went 62 through different lengths of optimal conditions of 8 or 24h during enrichment after overnight cold stressful conditions. 4.4.10. Analysis methods The microbial detection data was collected in binary (positive and negative) and computed as proportions. Both descriptive and statistical analysis of the detection data was done in R programming language [303,304]. To test the association between both incubation time and strain and detection, chi- square tests and odds ratios were used; and wherever the assumptions of the chi-square tests were not met, the Fisher’s exact test was used. For tests for the relationship between persister percentage and detection, a point-biserial correlation test was used. Regression techniques including logistic and classification trees were evaluated for fitting of the effects of incubation time, strain and persister percentage on the microbial detection, and the Bayesian Information Criteria (BIC) used to select the best fitting model. Since classification trees best fitted the combined effects of the independent variables, while it cannot provide the statistical significance of the same, we used the conditional inference trees (CIT) to fit the effects [303,304]. Statistical significance was tested at p<0.05. 4.5. Results and Discussion 4.5.1. Persister cell percentages varied with strains STEC O157:H7 can enter into dormant states on exposure to various stresses and can become resistant to stresses [111,122,124,275]. Persister cells were calculated as a percentage of the total overall culturable cells. It was observed that the percentage of persister cells increased with prolonged exposure to stress, as demonstrated by our data in nutrient-deprived agricultural water at 15°C, aligning with findings from previous research [275,305]. Inoculation of nutrient-limited agricultural water with the three strains was done on different days ranging from 2- 450 days, in order to have a wide range of persister percentages for the detection experiments, with the lower percentages being the newest set inoculated (1-2 days) agricultural water samples (e.g., 0.19, 0.26%), whereas the highest (e.g., 68, 85%) being mostly the oldest set inoculated agricultural water samples. The range of the persister percentage used in this study for all the strains was 0.12-85% (Figure 4.2). 63 Figure 4.2. Box plots showing distribution of persister cell percentages of the different strains. The X-axis shows three different outbreak strains of Central Coast, CA 2018 (n=72); Salinas, CA 2019 (n=48); and Yuma, AZ 2018 (n=88), whereas Y-axis shows persister cells in percentage. 4.5.2. Enrichment time impacts detectability From our study, we established that 30% of all the samples tested positive after an 8h enrichment, whereas 76.9% samples tested positive after a 24h enrichment period. All the strains had increasing detection with increasing incubation time, from 8 to 24 h (Table 1). Positive samples after 8h enrichment for different strains ranged from 4.1-36.4%, whereas positives after 24h enrichment for different strains ranged from 64 66.7- 86.4%. Statistical analysis showed that STEC O157:H7 on lettuce when enriched for 24h is more likely (odds ratio=11.55, p-value= 1.54e-15, Fisher’s Exact test) to be detected than at 8h. Table 4.1. Detection of positive samples after enrichment times of 8h, and 24h. Statistical tests Enrichment time 8h 24h Frequency of positives 23/104 (30.7%) 80/104 (76.9%) Odds ratio for positives 1 11.56 4.5.3. Detectability varied by strain of STEC O157:H7 on lettuce We used three strains linked to Romaine lettuce outbreaks for our studies viz., Yuma, AZ 2018 outbreak strain, Central Coast, CA 2018 outbreak strain, and Salinas, CA 2019 outbreak strain. Positive samples detected at 8h of enrichment from outbreak strains Central Coast 2018, and Salinas 2019, and Yuma 2018 were 16.6% (6/36), 4.1% (1/24), and 36.4% (16/44) respectively (Figure 4.3A). Strain was significantly (p=0.014, ꭓ2 = 8.61, chi-square) associated with the detectability of the pathogen. The Yuma 2018 outbreak strain was more likely (OR=2.21, CI= 1.17-4.21, p=0.014) to be detected than the outbreak strain Central Coast 2018 (Figure 4.3B). The Salinas 2019 outbreak strain was not significantly different from the Central Coast, CA 2018 outbreak strain (odd=0.91, CI= 0.43-1.94). 65 Figure 4.3. Bar graphs showing detection of the STEC O157:H7 on Romaine lettuce based on strains and enrichment time. (A) is the proportion of positive samples by strains, while (B) is the odds ratio for detection of positive samples for different strains. 66 4.5.4. Interaction of persister fraction, incubation time and strains affected the detectability of the pathogens The fitting of the interactive effect of incubation time, strains and persister fraction was first evaluated by including different combinations of factors in logistic model and classification trees. With a BIC value of 116.3, the classification tree had the best fit of the interaction. The BIC values for the logistic model without interactions and those with interactions were 234.1 and 257.1, respectively. Classification trees utilize gini impurity to select how random selection of response can result in misclassification. In order to test for statistical significance of the classifications and the factors, we used the conditional inference trees . Analysis of the interactive effect of all the factors revealed that at 24 h of incubation time, the detectability had the highest accuracy, p<0.001 (Figure 4.3). Incubation time was the most important factor that most influenced pathogen detection (Figure 4.4). Persister percentages by themselves didn’t impact detectability as shown by point biserial correlation test (p=0.49). Correlation tests showed a weak relationship (r=0.093) between persister % and detection. For samples that were incubated for 8 h, the Yuma 2018 strain was more likely to be detected, and the outbreak strains Salinas 2019 and Central Coast 2018 had better detectability (0.56/1 compared to 0.04/1) when the persister fraction was >68%, p=0.004. This is opposite of our hypothesis that cells won’t be able to be detected when more cells enter into persister state. This might be because of the reason that there were two persister percentages in the category of >68%, i.e., 68%, 80%, and 85%. But all of these samples consisted of only one strain, i.e., the Central Coast 2018 outbreak strain. We know that these strains differ by genetic variations. So, these results of better detection at >68% persister percentage, are highly unlikely to be representative of the whole STEC O157:H7 population and should be taken with careful consideration. So far, there has been no study that’s been done to observe the effect of persister state on the pathogen’s ability to be detected. CIT for the data, shows that incubation time is the most significant factor out of all in determining the number of samples that are detected positive (Figure 4.5). It shows that samples that are enriched for 24h have 0.76/1 chance of being detected. There have been other studies proving the same point that with a longer enrichment time compared to a shorter enrichment time, a pathogen is more likely to be detected, like STEC O157:H7 in ground beef and sprouts [306]. A 6h enrichment time in various protocols is not enough for detection of STEC O157:H7, when inoculated in low quantities in ground beef [307]. Our results and previous research show that enrichment time plays an important role in STEC O157:H7 being detected. Confidence inference tree shows that when it comes to being detected at 8h, the Yuma, AZ 2018 outbreak strain behaved differently compared to Central Coast, CA 2018, and Salinas, CA, 2019 outbreak strains (Figure 4.2), and is not dependent upon the persister percentage. According to Fisher’s Exact test, strains had a significant effect on being detected (p=0.01). The Yuma, AZ 2018 outbreak strain was more 67 likely to be detected compared to the Central Coast, CA 2018 outbreak strain at 24h of enrichment (odds ratio=2.2). This might be due to genetic variations. These strains are classified amongst the REP strains, which mean recurring, emergent, and persistent strains. This classification is determined by the Center for Disease Control (CDC) [308]. The Yuma, AZ 2018 outbreak strain belongs to clade 8, while the other two strains belong to clade 2 based on their genetic differences [309]. Central Coast, CA 2018 and Salinas, CA 2019 outbreak strains, belonging to REPEXH02 category are more closely related to each other compared to the Yuma, AZ 2018 outbreak strain, which belongs to REPEXH01 category [310,311]. More studies with larger number of strains would need to be done to clearly see if there is a significant difference in detectability among these REP strains. Figure 4.4. Analysis of the interactive effect of incubation time, strain and persister percentage on the detectability of the pathogens on lettuce. 68 Figure 4.4. (cont’d) Confidence Inference tree showing incubation time (p<0.001), strain (p=0.012), and persister percentage (p=0.004) to be important factors in bringing variability to detection results. The bottom most part of the plot represents the probability of positives in each case. The variability that the factors bring to detectability of STEC O157:H7 are shown in Figure 4.5. Figure 4.5. Variable importance of factors affecting detectability of pathogens on lettuce. 4.6. Conclusions The findings of this study highlight the critical impact of enrichment time on the detectability of STEC O157:H7 in romaine lettuce using commercially available qPCR-based test system. Detection rates significantly increased with longer enrichment, with only 30.7% of samples testing positive after 8 hours compared to 76.9% after 24 hours. Additionally, strain differences influenced detectability, with the Yuma 2018 outbreak strain more likely to be detected than others. The study also found that samples with persister cell percentages above 68% were significantly more detectable, although it only represented one strain. So, more research needs to be done where the data for higher percentages of persisters is taken from different strains. The strains behave differently already based on their genetic differences. These results emphasize 69 the need to optimize enrichment protocols, considering pathogen dormancy and strain variability, to improve foodborne pathogen surveillance and minimize contamination risks in the food supply chain. 4.7. Funding The funding for this study was made possible by The Center for Produce Safety and the U.S. Department of Agriculture’s (USDA) Agricultural Marketing Service through grant AM21SCBPCA1002. 70 5. CONCLUSIONS, FUTURE WORK, AND LIMITATIONS Pathogen persistence in agricultural environments is influenced by a combination of soil chemistry, microbiome interactions, and environmental stressors. This study integrates findings from multiple investigations to examine how factors such as flooding impact survival, how cold storage can impact survival, tolerance, and virulence of pathogens and how enrichment time in detection methodologies can impact detectability of foodborne pathogens, highlighting critical gaps in current risk assessment and food safety practices. This study underscores the pivotal role of soil chemistry, microbiome interactions, and nutrient availability in determining pathogen persistence in agricultural environments, particularly in the aftermath of extreme weather events such as flooding. Despite an overall decline in pathogen populations over time, L. monocytogenes, Salmonella, and STEC exhibited substantial persistence, underscoring the food safety concerns. These advancements will have significant implications. The native microbiome, along with factors such as soil chemistry, flooding conditions, and microbial diversity, plays a crucial role in limiting pathogen growth. These findings highlight the need for predictive soil modeling tools that can effectively assess the risks of pathogen survival in flooded agricultural soils, using LSTM models. The development of such models, incorporating simple yet insightful matrices, would enable farmers and policymakers to make informed decisions regarding risk mitigation strategies, ultimately enhancing food safety and agricultural sustainability. Current microbial risk assessments have predominantly focused on culturable pathogens such as STEC O157:H7 in leafy greens, often overlooking critical physiological changes that influence pathogen survival, virulence, and detectability. Our research revealed that cold storage conditions increase the transition of STEC O157:H7 into dormant states, raising the risk of underestimating pathogen persistence using conventional microbial detection methods. Additionally, exposure to temperature fluctuations during harvesting and storage has been shown to impact pathogen physiology, potentially altering survival rates and detection accuracy. This highlights the necessity of refining risk assessments to account for physiological adaptations that may render pathogens more resilient under specific environmental conditions. The food industry can enhance quality control by considering how cold storage, stress tolerance, strain variability, harvest temperature, and physiological states impact pathogen survival and virulence. These insights will help regulatory agencies refine safety standards and improve microbial monitoring techniques. Another significant finding of this study is that the detectability of STEC O157:H7 in romaine lettuce is strongly influenced by enrichment time, and strain variability. Our results demonstrate that 71 extending the enrichment period significantly improves detection rates, highlighting the importance of optimizing detection protocols for more accurate pathogen surveillance. Additionally, genetic differences among pathogen strains play a crucial role in detectability, as shown by the higher detection rates of the Yuma 2018 outbreak strain compared to others. This underscores the necessity of conducting strain-specific studies to refine foodborne pathogen monitoring approaches. Moreover, our data suggests that samples with high persister cell percentages are more readily detectable, though further research across diverse strains is needed to establish broader applicability. Our findings highlight the risk of underestimating contamination when relying solely on culture-based methods. Future investigations should prioritize enhancing enrichment methodologies to improve the reliability of foodborne pathogen detection, minimizing contamination risks in the food supply chain. 5.1. Limitations While this study provides valuable insights, some limitations must be acknowledged. 5.1.1. Temporal Data Limitations in Modeling The LSTM model's improved performance when incorporating data from day 4 suggests that early time points alone may not capture key survival trends. The exclusion of additional time points could limit the predictive power of the model, as pathogen behavior may exhibit non-linear patterns that emerge beyond the initial days of observation. This constraint affects the ability to make accurate early-stage predictions based solely on limited temporal data. 5.1.2. Climate Variability and Experimental Scale Constraints Environmental factors influencing pathogen survival over multiple lettuce growth cycles were not fully controlled. Seasonal fluctuations, temperature shifts, humidity levels, and other climatic variables could introduce inconsistencies in pathogen behavior, contributing to variability in persistence outcomes. The inability to isolate these factors from the study results may limit the precision of the findings, and might have caused variability with the replicates. The chlorine treatment experiments were conducted in a 5L container rather than a pilot-scale plant, which may not fully replicate real-world conditions. Differences in factors such as solution dynamics, pathogen exposure, and chlorine effectiveness at larger scales could result in deviations from the outcomes observed in this study. The smaller experimental scale might affect the accuracy of persistence and treatment efficacy assessments. 72 5.1.3. Strain-Specific Findings The observed higher persister percentage was limited to a single strain, restricting the generalizability of the findings across different strains. Variability among strains in formation of persister cells could lead to different outcomes under similar conditions, making it difficult to understand the results to a broader pathogen population of different strains. 5.2. Future Work Building on these limitations, in future many approaches can be considered to better understand pathogen persistence, refine predictive models, and improve food safety interventions. 5.2.1. In relation to impact of factors on survival of pathogens in soil extracts 5.2.1.1. Enhancing AI-Based Predictive Modeling The findings indicate that the LSTM model performed better when incorporating data from day 4, suggesting that early prediction is currently limited. To make AI models more useful for real-world applications, research should focus on refining existing models or exploring alternative modeling approaches. The goal would be to develop a robust predictive model that allows farmers to collect early data (e.g., up to day 2) while still providing reliable pathogen survival predictions for later days. Incorporating additional environmental and microbial community data into AI models could further enhance predictive accuracy and risk assessment capabilities. 5.2.1.2. Investigating the Role of the Microbiome in Pathogen Suppression Evidence suggests that some families of native microbiota are more effective than others in limiting pathogen survival. Future research should explore the microbiome’s role in suppressing pathogen persistence and identify specific microbial interactions that contribute to pathogen control. Metagenomic studies could provide insights into microbial competition, antagonistic interactions, and the production of antimicrobial compounds. Identifying naturally occurring microbial solutions could pave the way for biocontrol strategies that reduce pathogen survival without relying on chemical treatments. 5.2.2. In relation to impact of cold storage on physiological state, virulence, and tolerance of STEC O157:H7 on romaine lettuce 5.2.2.1. Addressing Climate Variability in Pathogen Persistence Environmental factors such as temperature fluctuations, humidity levels, and seasonal weather variations introduce variability that can affect pathogen survival. Long-term, multi-location studies should be conducted to understand the influence of different climatic conditions on pathogen survival across 73 diverse agricultural landscapes. Alternatively, controlled experiments can be designed to eliminate weather- related variability. Such studies could involve greenhouse or growth chamber experiments where environmental parameters are tightly regulated. 5.2.2.2. Scaling Up Decontamination Studies The effectiveness of decontamination strategies, such as chlorine treatment, may vary significantly depending on scale. While this study used a 5L container, scaling up these experiments to pilot-scale or industrial-scale food processing plants will provide more applicable and reliable data. Larger-scale studies will help identify operational challenges, optimize treatment parameters, and assess real-world efficacy, ensuring that chlorine sanitation is practical and implementable in commercial settings. 5.2.2.3. Understanding Molecular Pathways Governing Bacterial Dormancy Under Cold Stress Cold storage is a common practice in food supply chains, yet the mechanisms governing bacterial dormancy and survival under cold stress remain poorly understood. Research should focus on identifying molecular pathways that regulate dormancy and resuscitation in foodborne pathogens. Investigating gene expression changes, stress response mechanisms, and metabolic adaptations during cold storage and transportation will provide critical insights into how pathogens evade decontamination efforts and persist in food products. 5.2.3. In relation to impact of physiological state, and enrichment time on detectability of STEC O157:H7 on romaine lettuce 5.2.3.1. Expanding Strain Variability Studies Given that the observed higher persister percentage was noted in only one strain, it is essential to investigate whether this phenomenon is strain-specific or a broader microbial trait for all the strains of STEC O157:H7. Comparative studies incorporating a wide range of pathogen strains from different geographic locations and environmental backgrounds will help determine if detectability is linked to specific genetic or physiological traits. Genomic and transcriptomic analyses of these strains may also reveal underlying molecular mechanisms contributing to detectability. Addressing the challenges posed by pathogen survival requires an approach that integrates predictive soil modeling, comprehensive microbial risk assessments, and advanced detection techniques. Future research should incorporate seasonal variations, environmental stressors, and pathogen physiological adaptations to develop more accurate predictive models. Strengthening food safety protocols 74 and contamination prevention strategies will not only be better for public health but also support sustainable agricultural practices. 75 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. BIBLIOGRAPHY DGA Dietary Guidelines for Americans, 2020-2025 and Online Materials | Dietary Guidelines for Americans Available online: https://www.dietaryguidelines.gov/resources/2020-2025- dietary-guidelines-online-materials (accessed on 8 May 2024). Lee, S.H.; Latetia, ; Moore, V.; Park, S.; Harris, D.M.; Blanck, H.M. Morbidity and Mortality Weekly Report Adults Meeting Fruit and Vegetable Intake Recommendations-United States, 2019. 2022. Dark Green Leafy Vegetables : USDA ARS Dark Green Leafy Vegetables : USDA ARS Available online: https://www.ars.usda.gov/plains-area/gfnd/gfhnrc/docs/news-articles/2013/dark-green- leafy-vegetables/ (accessed on 9 May 2024). Turner, K.; Nou Moua, C.; Hajmeer, M.; Barnes, A.; Needham, M. Overview of Leafy Greens- Related Food Safety Incidents with a California Link: 1996 to 2016. 2018, doi:10.4315/0362- 028X.JFP-18-316. Self, J.; Conrad, A.; Stroika, S.; Jackson, A. Emerging Infectious Diseases • Www. 2019, doi:10.3201/eid2507.180761. Krishnasamy, V.P.; Marshall, K.; Dewey-Mattia, D.; Wise, M. Outbreak Characteristics and Epidemic Curves for Multistate Outbreaks of Salmonella Infections Associated with Produce: United States, 2009-2015 HHS Public Access. Foodborne Pathog Dis 2020, 17, 15–22, doi:10.1089/fpd.2019.2711. Foodborne germs and Illnesses CDC Foodborne Germs and Illnesses | CDC Available online: https://www.cdc.gov/foodsafety/foodborne-germs.html (accessed on 8 May 2024). Symptoms E. https://www.cdc.gov/ecoli/ecoli-symptoms.html (accessed on 8 May 2024). coli CDC Symptoms | E. Coli | CDC Available Symptoms Listeria CDC https://www.cdc.gov/listeria/symptoms.html (accessed on 8 May 2024). | Listeria Symptoms | CDC Available online: online: Symptoms Salmonella CDC Symptoms | CDC Available online: https://www.cdc.gov/salmonella/general/salmonella-symptoms.html (accessed on 8 May 2024). | Salmonella People at Risk | Listeria | CDC People at Risk | Listeria | CDC Available online: https://www.cdc.gov/listeria/risk.html (accessed on 8 May 2024). 12. Weis, J.; Seeliger, H.P.R. Incidence of Listeria Monocytogenes in Nature. Appl Microbiol 1975, 30, 29–32. 13. Linke, K.; Rückerl, I.; Brugger, K.; Karpiskova, R.; Walland, J.; Muri-Klinger, S.; Tichy, A.; Wagner, M.; Stessl, B. Reservoirs of Listeria Species in Three Environmental Ecosystems. Appl Environ Microbiol 2014, 80, 5583–5592. 14. Welshimer, H.J.; Donker-Voet, J. Listeria Monocytogenes in Nature. Appl Microbiol 1971, 21, 516–519. 15. 16. Allerberger, F. Listeria: Growth, Phenotypic Differentiation and Molecular Microbiology. FEMS Immunol Med Microbiol 2003, 35, 183–189. Bucur, F.I.; Grigore-Gurgu, L.; Crauwels, P.; Riedel, C.U.; Nicolau, A.I. Resistance of Listeria Monocytogenes to Stress Conditions Encountered in Food and Food Processing Environments. Front Microbiol 2018, 9, 2700. 76 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Buchanan, R.L.; Stahl, H.G.; Whiting, R.C. Effects and Interactions of Temperature, PH, Atmosphere, Sodium Chloride, and Sodium Nitrite on the Growth of Listeria Monocytogenes. J Food Prot 1989, 52, 1–844. Gray, M.L.; Stafseth, H.J.; Thorp Jr, F.; Sholl, L.B.; Riley Jr, W.F. A New Technique for Isolating Listerellae from the Bovine Brain. J Bacteriol 1948, 55, 471–476. Osek, J.; Lachtara, B.; Wieczorek, K. Listeria Monocytogenes – How This Pathogen Survives in Food-Production Environments? Front Microbiol 2022, 13. Buchanan, R.L.; Gorris, L.G.M.; Hayman, M.M.; Jackson, T.C.; Whiting, R.C. A Review of Listeria Monocytogenes: An Update on Outbreaks, Virulence, Dose-Response, Ecology, and Risk Assessments. Food Control 2017, 75, 1–13. Farber, J.M.; Peterkin, P. Listeria Monocytogenes, a Food-Borne Pathogen. Microbiol Rev 1991, 55, 476–511. Dalton, C.B.; Austin, C.C.; Sobel, J.; Hayes, P.S.; Bibb, W.F.; Graves, L.M.; Swaminathan, B.; Proctor, M.E.; Griffin, P.M. An Outbreak of Gastroenteritis and Fever Due to Listeria Monocytogenes in Milk. New England Journal of Medicine 1997, 336, 100–106. Linnan, M.J.; Mascola, L.; Lou, X.D.; Goulet, V.; May, S.; Salminen, C.; Hird, D.W.; Yonekura, M.L.; Hayes, P.; Weaver, R. Epidemic Listeriosis Associated with Mexican-Style Cheese. New England Journal of Medicine 1988, 319, 823–828. Schuchat, A.; Deaver, K.A.; Wenger, J.D.; Plikaytis, B.D.; Mascola, L.; Pinner, R.W.; Reingold, A.L.; Broome, C. V; Swaminathan, B.; Hayes, P.S. Role of Foods in Sporadic Listeriosis: I. Case- Control Study of Dietary Risk Factors. JAMA 1992, 267, 2041–2045. Osek, J.; Wieczorek, K. Listeria Monocytogenes—How This Pathogen Uses Its Virulence Mechanisms to Infect the Hosts. Pathogens 2022, 11, 1491. Vázquez-Boland, J.A.; Kuhn, M.; Berche, P.; Chakraborty, T.; Domı́nguez-Bernal, G.; Goebel, W.; González-Zorn, B.; Wehland, J.; Kreft, J. Listeria Pathogenesis and Molecular Virulence Determinants. Clin Microbiol Rev 2001, 14, 584–640. 27. Magalhães, R.; Mena, C.; Ferreira, V.; Silva, J.; Almeida, G.; Gibbs, P.; Teixeira, P. Listeria Monocytogenes. Encyclopedia of Food Safety, Second Edition, Volume 1-4 2023, 1–4, V2-164- V2-178, doi:10.1016/B978-0-12-822521-9.00045-9. 28. 29. 30. 31. Listeriosis online: https://www.health.ny.gov/diseases/communicable/listeriosis/fact_sheet.htm (accessed on 23 March 2025). Infection) Available (Listeria Zhang, T.; Abel, S.; Abel zur Wiesch, P.; Sasabe, J.; Davis, B.M.; Higgins, D.E.; Waldor, M.K. Deciphering the Landscape of Host Barriers to Listeria Monocytogenes Infection. Proceedings of the National Academy of Sciences 2017, 114, 6334–6339. Portnoy, D.A.; Jacks, P.S.; Hinrichs, D.J. Role of Hemolysin for the Intracellular Growth of Listeria Monocytogenes. J Exp Med 1988, 167, 1459–1471. Beauregard, K.E.; Lee, K.-D.; Collier, R.J.; Swanson, J.A. PH-Dependent Perforation of Macrophage Phagosomes by Listeriolysin O from Listeria Monocytogenes. J Exp Med 1997, 186, 1159–1163. 77 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. Kocks, C.; Gouin, E.; Tabouret, M.; Berche, P.; Ohayon, H.; Cossart, P. L. Monocytogenes- Induced Actin Assembly Requires the ActA Gene Product, a Surface Protein. Cell 1992, 68, 521– 531. Cong, Z.; Xiong, Y.; Lyu, L.; Fu, B.; Guo, D.; Sha, Z.; Yang, B.; Wu, H. The Relationship between Listeria Infections and Host Immune Responses: Listeriolysin O as a Potential Target. Biomedicine & Pharmacotherapy 2024, 171, 116129. Trowsdale, J. Genomic Structure and Function in the MHC. Trends in Genetics 1993, 9, 117– 122. CDC https://www.cdc.gov/listeria/outbreaks/index.html (accessed on 5 February 2025). Listeria Outbreaks | CDC Available Infection Listeria | online: Scallan, E.; Hoekstra, R.M.; Angulo, F.J.; Tauxe, R. V; Widdowson, M.-A.; Roy, S.L.; Jones, J.L.; Griffin, P.M. Foodborne Illness Acquired in the United States—Major Pathogens. Emerg Infect Dis 2011, 17, 7. Le Minor, L.; Popoff, M.Y. Request for an Opinion: Designation of Salmonella Enterica Sp. Nov., Nom. Rev., as the Type and Only Species of the Genus Salmonella. Int J Syst Evol Microbiol 1987, 37, 465–468. Penner, J.L. International Committee on Systematic Bacteriology Taxonomic Subcommittee on Enterobacteriaceae: Minutes of the Meeting, 8 September 1986, Manchester, England. Int J Syst Evol Microbiol 1988, 38, 223–224. Carter, J.D.; Hudson, A.P. Reactive Arthritis: Clinical Aspects and Medical Management. Rheumatic Disease Clinics of North America 2023, 35, 21–44, doi:10.1016/j.rdc.2009.03.010. Bastin, M.L.T.; Neville, N.R.; Parsons, R.E.; Flannery, A.H.; Tennant, S.J.; Johnson, C.A. An Unusual Case of Salmonella Enteritidis Causing Pneumonia, Septic Shock and Multiple Organ Failure in an Immunocompetent Patient. IDCases 2016, 6, 85–89. Cheng, W.-L.; Li, C.-W.; Li, M.-C.; Lee, N.-Y.; Lee, C.-C.; Ko, W.-C. Salmonella Infective Endocarditis. Journal of Microbiology, Immunology and Infection 2016, 49, 313–320. Finlay, B.B. Molecular and Cellular Mechanisms of Salmonella Pathogenesis; 1994; Giannella, R.A. Salmonella. Medical Microbiology 1996, Chapter 21. Hurley, D.; McCusker, M.P.; Fanning, S.; Martins, M. Salmonella–Host Interactions–Modulation of the Host Innate Immune System. Front Immunol 2014, 5, 481. Zhu, H.; Sydor, A.M.; Boddy, K.C.; Coyaud, E.; Laurent, E.M.N.; Au, A.; Tan, J.M.J.; Yan, B.- R.; Moffat, J.; Muise, A.M. Salmonella Exploits Membrane Reservoirs for Invasion of Host Cells. Nat Commun 2024, 15, 3120. 46. Westerman, T.L.; Bogomolnaya, L.; Andrews-Polymenis, H.L.; Katherine Sheats, M.; Elfenbein, J.R. The Salmonella Type-3 Secretion System-1 and Flagellar Motility Influence the Neutrophil Respiratory Burst. PLoS One 2018, 13, doi:10.1371/JOURNAL.PONE.0203698. 47. 48. Kaper, J.B.; Nataro, J.P.; Mobley, H.L.T. Pathogenic Escherichia Coli. Nat Rev Microbiol 2004, 2, 123–140. Clermont, O.; Christenson, J.K.; Denamur, E.; Gordon, D.M. The C Lermont E Scherichia Coli Phylo‐typing Method Revisited: Improvement of Specificity and Detection of New Phylo‐ groups. Environ Microbiol Rep 2013, 5, 58–65. 78 49. Wirth, T.; Falush, D.; Lan, R.; Colles, F.; Mensa, P.; Wieler, L.H.; Karch, H.; Reeves, P.R.; Maiden, M.C.J.; Ochman, H. Sex and Virulence in Escherichia Coli: An Evolutionary Perspective. Mol Microbiol 2006, 60, 1136–1151. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. Jaureguy, F.; Landraud, L.; Passet, V.; Diancourt, L.; Frapy, E.; Guigon, G.; Carbonnelle, E.; Lortholary, O.; Clermont, O.; Denamur, E. Phylogenetic and Genomic Diversity of Human Bacteremic Escherichia Coli Strains. BMC Genomics 2008, 9, 1–14. Fatima, R.; Aziz, M. Enterohemorrhagic Escherichia Coli (EHEC). StatPearls. Treasure Island (FL) 2019. Penzel, A.; Schützler, K.; Dröge, J.; Mellmann, A.; Ehricht, R.; Engelmann, I.; Braun, S.D.; Schleenvoigt, B.T.; Löffler, B.; Rödel, J. Rapid Culture-Based Identification of Shiga Toxin- Producing Escherichia Coli and Shigella Spp./Enteroinvasive E. Coli Using the Eazyplex® EHEC Complete Assay. European Journal of Clinical Microbiology & Infectious Diseases 2020, 39, 151–158. K Rahjerdi, A.; Amani, J.; Rad, I.; Jafari, M.; H Salmanian, A. Designing and Structure Evaluation of Multi-Epitope Vaccine against ETEC and EHEC, an in Silico Approach. Protein Pept Lett 2016, 23, 33–42. Yara, D.; Greig, D.; Dallman, T.; Jenkins, C. Shiga Toxin Prophage Analysis of Clinically Relevant Enterohaemorrhagic E. Coli Isolates Using Nanopore Sequencing 2019. Nguyen, Y.; Sperandio, V. Enterohemorrhagic E. Coli (EHEC) Pathogenesis. Front Cell Infect Microbiol 2012, 2, 90. Gonzalez, G.M.A.; Cerqueira, M.F.A. Shiga Toxin‐producing Escherichia Coli in the Animal Reservoir and Food in Brazil. J Appl Microbiol 2020, 128, 1568–1582. Outbreaks of E. Coli https://www.cdc.gov/ecoli/outbreaks/index.html (accessed on 11 February 2025). | E. Coli Infections Infection | CDC Available online: Croxen, M.A.; Finlay, B.B. Molecular Mechanisms of Escherichia Coli Pathogenicity. Nat Rev Microbiol 2010, 8, 26–38. Schmidt, M.A. LEEways: Tales of EPEC, ATEC and EHEC. Cell Microbiol 2010, 12, 1544– 1552. 60. McDaniel, T.K.; Jarvis, K.G.; Donnenberg, M.S.; Kaper, J.B. A Genetic Locus of Enterocyte Effacement Conserved among Diverse Enterobacterial Pathogens. Proceedings of the National Academy of Sciences 1995, 92, 1664–1668. 61. 62. 63. 64. Elliott, S.J.; Wainwright, L.A.; McDaniel, T.K.; Jarvis, K.G.; Deng, Y.; Lai, L.-C.; McNamara, B.P.; Donnenberg, M.S.; Kaper, J.B. The Complete Sequence of the Locus of Enterocyte Effacement (LEE) from Enteropathogenic Escherichia Coli E2348/69. Mol Microbiol 1998, 28. Krishnasamy, V.P.; Marshall, K.; Dewey-Mattia, D.; Wise, M. Outbreak Characteristics and Epidemic Curves for Multistate Outbreaks of Salmonella Infections Associated with Produce: United States, 2009–2015. Foodborne Pathog Dis 2020, 17, 15, doi:10.1089/FPD.2019.2711. Self, J.; Conrad, A.; Stroika, S.; Jackson, A. Emerging Infectious Diseases • Www. 2019, doi:10.3201/eid2507.180761. Turner, K.; Nou Moua, C.; Hajmeer, M.; Barnes, A.; Needham, M. Overview of Leafy Greens- Related Food Safety Incidents with a California Link: 1996 to 2016. 2019, doi:10.4315/0362- 028X.JFP-18-316. 79 65. 66. 67. 68. 69. 70. Brandl, M.T. Fitness of Human Enteric Pathogens on Plants and Implications for Food Safety 1. 2006, doi:10.1146/annurev.phyto.44.070505.143359. Cho, K.H.; Wolny, J.; Kase, J.A.; Unno, T.; Pachepsky, Y. Interactions of E. Coli with Algae and Aquatic Vegetation in Natural Waters. 2021, doi:10.1016/j.watres.2021.117952. Bergholz, P.W.; Strawn, L.K.; Ryan, G.T.; Warchocki, S.; Wiedmann, M. Spatiotemporal Analysis of Microbiological Contamination in New York State Produce Fields Following Extensive Flooding from Hurricane Irene, August 2011. J Food Prot 2016, 79, 384–391, doi:10.4315/0362-028X.JFP-15-334. FSMA Final Rule on Pre-Harvest Agricultural Water FSMA Final Rule on Pre-Harvest Agricultural Water | FDA Available online: https://www.fda.gov/food/food-safety- modernization-act-fsma/fsma-final-rule-pre-harvest-agricultural-water (accessed on 16 May 2024). California Leafy Greens Marketing Agreement California Leafy Greens Marketing Agreement | Home Page Available online: https://lgma.ca.gov/ (accessed on 28 April 2024). Neal, J.A.; Cabrera-Diaz, E.; Márquez-González, M.; Maxim, J.E.; Castillo, A. Reduction of Escherichia Coli O157: H7 and Salmonella on Baby Spinach, Using Electron Beam Radiation. J Food Prot 2008, 71, 2415–2420. 71. Mukhopadhyay, S.; Sokorai, K.; Ukuku, D.O.; Jin, T.; Fan, X.; Olanya, O.M.; Leng, J.; Juneja, V. Effects of Direct and In-Package Pulsed Light Treatment on Inactivation of E. Coli O157: H7 and Reduction of Microbial Loads in Romaine Lettuce. LWT 2021, 139, 110710. 72. 73. 74. 75. 76. 77. Posada-Izquierdo, G.D.; Pérez-Rodríguez, F.; López-Gálvez, F.; Allende, A.; Gil, M.I.; Zurera, G. Modeling Growth of Escherichia Coli O157: H7 in Fresh-Cut Lettuce Treated with Neutral Electrolyzed Water and under Modified Atmosphere Packaging. Int J Food Microbiol 2014, 177, 1–8. The Counter How a 3.5-Mile-Long Irrigation Canal Used by 23 Farms Caused the Yuma E. Coli Outbreak Available online: https://thecounter.org/yuma-lettuce-ecoli-outbreak-fda/ (accessed on 26 February 2025). NPR What Sparked An E. Coli Outbreak In Lettuce? Scientists Trace A Surprising Source : The Salt : online: https://www.npr.org/sections/thesalt/2018/08/29/642646707/investigators-track-contaminated- lettuce-outbreak-to-a-cattle-feedlot (accessed on 26 February 2025). Available NPR FDA, D. Outbreak Investigation of Listeria Monocytogenes: Dole Packaged Salad (December 2021) | FDA Available online: https://www.fda.gov/food/outbreaks-foodborne-illness/outbreak- investigation-listeria-monocytogenes-dole-packaged-salad-december-2021 (accessed on 26 February 2025). FDA, B. Outbreak Investigation of Salmonella Typhimurium: BrightFarms Packaged Salad Greens (July 2021) | FDA Available online: https://www.fda.gov/food/outbreaks-foodborne- illness/outbreak-investigation-salmonella-typhimurium-brightfarms-packaged-salad-greens- july-2021 (accessed on 26 February 2025). Brandl, M.T. Fitness of Human Enteric Pathogens on Plants and Implications for Food Safety. Annu 367–392, Phytopathol doi:10.1146/ANNUREV.PHYTO.44.070505.143359/CITE/REFWORKS. 2006, Rev 44, 78. Moyne, A.; Sudarshana, M.R.; Blessington, T.; Koike, S.T.; Cahn, M.D.; Harris, L.J. Fate of Escherichia Coli O157: H7 in Field-Inoculated Lettuce. Food Microbiol 2011, 28, 1417–1425. 80 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. Bergholz, P.; Strawn, L.K.; Ryan, G.; Warchocki, S.; Wiedmann, M. Spatiotemporal Analysis of Microbiological Contamination in New York State Produce Fields Following Extensive Flooding from Hurricane Irene, August 2011. J Food Prot 2016, 79, 384–391, doi:10.4315/0362- 028X.JFP-15-334. Cho, K.H.; Wolny, J.; Kase, J.A.; Unno, T.; Pachepsky, Y. Interactions of E. Coli with Algae and Aquatic Vegetation in Natural Waters. 2021, doi:10.1016/j.watres.2021.117952. California Leafy Greens Marketing Agreement California Leafy Greens Marketing Agreement | Home Page Available online: https://lgma.ca.gov/ (accessed on 28 April 2024). Bach, S.J.; Stanford, K.; Mcallister, T.A. Survival of Escherichia Coli O157:H7 in Feces from Corn- and Barley-Fed Steers. 2005, doi:10.1016/j.femsle.2005.08.030. Artz, R.R.E.; Townend, J.; Brown, K.; Towers, W.; Killham, K. Soil Macropores and Compaction Control the Leaching Potential of Escherichia Coli O157:H7. Environ Microbiol 2005, 7, 241– 248, doi:10.1111/J.1462-2920.2004.00690.X. Ihssen, J.; Grasselli, E.; Bassin, C.; Franç ois, P.; Piffaretti, J.-C.; Kö ster, W.; Schrenzel, J.; Egli, T.; Egli egli, T. Comparative Genomic Hybridization and Physiological Characterization of Environmental Isolates Indicate That Significant (Eco-)Physiological Properties Are Highly Conserved in the Species Escherichia Coli. 2007, doi:10.1099/mic.0.2006/002006-0. Ihssen, J.; Egli, T. Global Physiological Analysis of Carbon- and Energy-Limited Growing Escherichia Coli Confirms a High Degree of Catabolic Flexibility and Preparedness for Mixed Substrate Utilization. Environ Microbiol 2005, 7, 1568–1581, doi:10.1111/J.1462- 2920.2005.00846.X. Franz, E.; Van Bruggen, A.H.C. Ecology of E. Coli O157:H7 and Salmonella Enterica in the Primary Vegetable Production Chain. Crit Rev Microbiol 2008, 34, 143–161, doi:10.1080/10408410802357432. Alegbeleye, O.; Sant, A.S. Survival of Salmonella Spp. under Varying Temperature and Soil Conditions. 2023, doi:10.1016/j.scitotenv.2023.163744. Locatelli, A.; Spor, A.; Jolivet, C.; Piveteau, P.; Hartmann, A. Biotic and Abiotic Soil Properties Influence Survival of Listeria Monocytogenes in Soil. PLoS One 2013, 8, 75969, doi:10.1371/journal.pone.0075969. Xing, J.; Wang, H.; Brookes, P.C.; Salles, J.F.; Xu, J. Soil PH and Microbial Diversity Constrain the Survival of E. Coli in Soil. 2018, doi:10.1016/j.soilbio.2018.10.013. Shah, M.K.; Bradshaw, R.; Nyarko, E.; Handy, E.T.; East, C.; Millner, P.D.; Bergholz, T.M.; Sharma, M. Salmonella Enterica in Soils Amended with Heat-Treated Poultry Pellets Survived Longer than Bacteria in Unamended Soils and More Readily Transferred to and Persisted on Spinach. Appl Environ Microbiol 2019, 85, doi:10.1128/AEM.00334-19/FORMAT/EPUB. Sharma, M.; Millner, P.D.; Hashem, F.; Vinyard, B.T.; East, C.L.; Handy, E.T.; White, K.; Stonebraker, R.; Cotton, C.P. Survival of Escherichia Coli in Manure-Amended Soils Is Affected by Spatiotemporal, Agricultural, and Weather Factors in the Mid-Atlantic United States. Appl Environ Microbiol 2018, 85, doi:10.1128/AEM.02392-18/FORMAT/EPUB. Neher, D.; Cutler, A.; Weicht, T.; Sharma, M.; Millner, P.; Deborah Neher, C.A. Composts of Poultry Litter or Dairy Manure Differentially Affect Survival of Enteric Bacteria in Fields with Spinach. 2019, doi:10.1111/jam.14268. 81 93. 94. 95. 96. 97. 98. 99. Hammami, R.; Crippen, T.L.; Strawn, L.K.; Bardsley, C.A.; Weller, D.L.; Ingram, D.T.; Chen, Y.; Oryang, D.; Rideout, S.L. Strain, Soil-Type, Irrigation Regimen, and Poultry Litter Influence Salmonella Survival and Die-off | www.frontiersin.org 2021, 1, 590303, doi:10.3389/fmicb.2021.590303. in Agricultural Soils. Frontiers in Microbiology Callahan, M.T.; Micallef, S.A.; Sharma, M.; Millner, P.D.; Buchanan, R.L. Metrics Proposed to Prevent the Harvest of Leafy Green Crops Exposed to Flood Water Contaminated with Escherichia Coli. Appl Environ Microbiol 2016, 82, 3746–3753, doi:10.1128/AEM.00052- 16/FORMAT/EPUB. Kim, J.; Jiang, X. The Growth Potential of Escherichia Coli O157:H7, Salmonella Spp. and Listeria Monocytogenes in Dairy Manure-Based Compost in a Greenhouse Setting under Different Seasons. 2010, doi:10.1111/j.1365-2672.2010.04841.x. Shah, M.K.; Bradshaw, † Rhodel; Nyarko, E.; Millner, P.D.; Neher, D.; Weicht, T.; Bergholz, T.M.; And, †; Sharma, M. Survival and Growth of Wild-Type and RpoS-Deficient Salmonella Newport Strains in Soil Extracts Prepared with Heat-Treated Poultry Pellets. 2019, doi:10.4315/0362-028X.JFP-18-465. Owade, J.O.; Bergholz, T.M.; Mitchell, J. A Review of Conditions Influencing Fate of Shiga Toxin‐producing Escherichia Coli O157: H7 in Leafy Greens. Compr Rev Food Sci Food Saf 2024, 23, e70013. Food Safety Program, 2024 California Leafy Greens Marketing Agreement | LGMA Connect Available online: https://lgma.ca.gov/lgma-connect (accessed on 28 April 2024). Declan Iwu, C.; Okoh, A.I. Preharvest Transmission Routes of Fresh Produce Associated Bacterial Pathogens with Outbreak Potentials: A Review. 2019, doi:10.3390/ijerph16224407. 100. Guo, X.; Chen, J.; Brackett, R.E.; Beuchat, L.R. Survival of Salmonellae on and in Tomato Plants from the Time of Inoculation at Flowering and Early Stages of Fruit Development through Fruit Ripening. Appl Environ Microbiol 2001, 67, 4760–4764, doi:10.1128/AEM.67.10.4760- 4764.2001. 101. Bottichio, L.; Keaton, A.; Thomas, D.; Fulton, T.; Tiffany, A.; Frick, A.; Mattioli, M.; Kahler, A.; Murphy, J.; Otto, M.; et al. Clinical Infectious Diseases Shiga Toxin-Producing Escherichia Coli Infections Associated With Romaine Lettuce-United States, 2018. Clinical Infectious Diseases ® 2020, 71, 323–353, doi:10.1093/cid/ciz1182. 102. Coulombe, G.; Catford, A.; Martinez-Perez, A.; Buenaventura, E. General Interest Outbreaks of Escherichia Coli O157:H7 Infections Linked to Romaine Lettuce in Canada from 2008 to 2018: An Analysis of Food Safety Context. 2020, doi:10.4315/JFP-20-029. 103. Marshall, K.E.; Hexemer, A.; Seelman, S.L.; Fatica, M.K.; Blessington, T.; Hajmeer, M.; Kisselburgh, H.; Atkinson, R.; Hill, K.; Sharma, D.; et al. Emerging Infectious Diseases • Www Lessons Learned from a Decade of Investigations of Shiga Toxin-Producing Escherichia Coli Outbreaks Linked and Canada. 2020, 26, doi:10.3201/eid2610.191418. to Leafy Greens, United States 104. Owade, J.O.; Bergholz, T.M.; Mitchell, J. A Review of Conditions Influencing Fate of Shiga Toxin‐producing Escherichia Coli O157: H7 in Leafy Greens. Compr Rev Food Sci Food Saf 2024, 23, e70013. 105. Gleeson, E.; O’beirne, D. Effects of Process Severity on Survival and Growth of Escherichia Coli and Listeria Innocua on Minimally Processed Vegetables. Food Control 2005, 16, 677–685. 82 106. USDA Allowed Detergents and Sanitizers for Food Contact Surfaces and Equipment in Organic Operations; 2025; 107. Cui, H.; Bai, M.; Yuan, L.; Surendhiran, D.; Lin, L. Sequential Effect of Phages and Cold Nitrogen Plasma against Escherichia Coli O157: H7 Biofilms on Different Vegetables. Int J Food Microbiol 2018, 268, 1–9. 108. 109. 110. FDA Guidance for Industry: Guide to Minimize Microbial Food Safety Hazards for Fresh Fruits and Vegetables | FDA Available online: https://www.fda.gov/regulatory-information/search-fda- guidance-documents/guidance-industry-guide-minimize-microbial-food-safety-hazards-fresh- fruits-and-vegetables (accessed on 14 February 2025). Lennon, J.T.; Jones, S.E. Microbial Seed Banks: The Ecological and Evolutionary Implications of Dormancy. Nat Rev Microbiol 2011, 9, 119–130. Li, L.; Mendis, N.; Trigui, H.; Oliver, J.D.; Faucher, S.P. The Importance of the Viable but Non- Culturable State in Human Bacterial Pathogens. Front Microbiol 2014, 5, 258. 111. Ayrapetyan, M.; Williams, T.; Oliver, J.D. Relationship between the Viable but Nonculturable State and Antibiotic Persister Cells. 2018, doi:10.1128/JB.00249-18. 112. Maisonneuve, E.; Gerdes, K. Molecular Mechanisms Underlying Bacterial Persisters. Cell 2014, 157, 539–548. 113. Helaine, S.; Kugelberg, E. Bacterial Persisters: Formation, Eradication, and Experimental Systems. Trends Microbiol 2014, 22, 417–424. 114. Nowakowska, J.; Oliver, J.D. Resistance to Environmental Stresses by Vibrio Vulnificus in the Viable but Nonculturable State. FEMS Microbiol Ecol 2013, 84, 213–222. 115. Balaban, N.Q.; Merrin, J.; Chait, R.; Kowalik, L.; Leibler, S. Bacterial Persistence as a Phenotypic Switch. Science (1979) 2004, 305, 1622–1625. 116. Faruque, S.M.; Biswas, K.; Udden, S.M.N.; Ahmad, Q.S.; Sack, D.A.; Nair, G.B.; Mekalanos, J.J. Transmissibility of Cholera: In Vivo-Formed Biofilms and Their Relationship to Infectivity and Persistence in the Environment. Proceedings of the National Academy of Sciences 2006, 103, 6350–6355. 117. Nelson, E.J.; Chowdhury, A.; Flynn, J.; Schild, S.; Bourassa, L.; Shao, Y.; LaRocque, R.C.; Calderwood, S.B.; Qadri, F.; Camilli, A. Transmission of Vibrio Cholerae Is Antagonized by Lytic Phage and Entry into the Aquatic Environment. PLoS Pathog 2008, 4, e1000187. 118. Ayrapetyan, M.; Williams, T.; Oliver, J.D. Relationship between the Viable but Nonculturable State and Antibiotic Persister Cells. 2018, doi:10.1128/JB.00249-18. 119. Dinu, L.-D.; Bach, S. Induction of Viable but Nonculturable Escherichia Coli O157:H7 in the Phyllosphere of Lettuce: A Food Safety Risk Factor. Appl Environ Microbiol 2011, 77, 8295– 8302, doi:10.1128/AEM.05020-11. 120. Munther, D.S.; Carter, M.Q.; Aldric, C. V.; Ivanek, R.; Brandl, M.T. Formation of Escherichia Coli O157:H7 Persister Cells in the Lettuce Phyllosphere and Application of Differential Equation Models To Predict Their Prevalence on Lettuce Plants in the Field. Appl Environ Microbiol 2020, 86, doi:10.1128/AEM.01602-19. 121. Thao, S.; Brandl, M.T.; Carter, M.Q. Enhanced Formation of Shiga Toxin-Producing Escherichia Coli Persister Variants in Environments Relevant to Leafy Greens Production 2 3. 2019. 83 122. Dinu, L.D.; Bach, S. Induction of Viable but Nonculturable Escherichia Coli O157:H7 in the Phyllosphere of Lettuce: A Food Safety Risk Factors. Appl Environ Microbiol 2011, 77, 8295– 8302, doi:10.1128/AEM.05020-11. 123. Truchado, P.; Gómez-Galindo, M.; Gil, M.I.; Allende, A. Cross-Contamination of Escherichia Coli O157: H7 and Listeria Monocytogenes in the Viable but Non-Culturable (VBNC) State during Washing of Leafy Greens and the Revival during Shelf-Life. Food Microbiol 2023, 109, 104155. 124. Thao, S.; Brandl, M.T.; Carter, M.Q. Enhanced Formation of Shiga Toxin-Producing Escherichia Coli Persister Variants in Environments Relevant to Leafy Greens Production 2 3. 2019. 125. Bottichio, L.; Keaton, A.; Thomas, D.; Fulton, T.; Tiffany, A.; Frick, A.; Mattioli, M.; Kahler, A.; Murphy, J.; Otto, M.; et al. Shiga Toxin-Producing Escherichia Coli Infections Associated with Romaine Lettuce-United States, 2018. Clinical Infectious Diseases 2020, 71, E323–E330, doi:10.1093/cid/ciz1182. 126. Yang, X.; Scharff, R. Foodborne Illnesses from Leafy Greens in the United States: Attribution, Burden, and Cost. J Food Prot 2024, 87, doi:10.1016/j.jfp.2024.100275. 127. Irvin, K.; Viazis, S.; Fields, A.; Seelman, S.; Blickenstaff, K.; Gee, E.; Wise, M.E.; Marshall, K.E.; Gieraltowski, L.; Harris, S. An Overview of Traceback Investigations and Three Case Studies of Recent Outbreaks of Escherichia Coli O157: H7 Infections Linked to Romaine Lettuce. J Food Prot 2021, 84, 1340–1356. 128. Capps, K.L.; Mclaughlin, E.M.; Murray, A.W.A.; Aldus, C.F.; Wyatt, G.M.; Peck, M.W.; Van Amerongen, A.; Ariëns, R.M.C.; Wichers, J.H.; Baylis, C.L.; et al. Validation of Three Rapid Screening Methods for Detection of Verotoxin-Producing Escherichia Coli in Foods: Interlaboratory Study; 2004; 129. Ritter, V.; Kircher, S.; Dick, N.; Edward Richter, R. USDA FSIS, FDA BAM, and ISO Culture Methods BD BBLTM CHROMagarTM O157 Media; 130. 131. Feng, P.; Weagant, S.D.; Jinneman, K. Bacteriological Analytical Manual Chapter 4A: Diarrheagenic Escherichia Coli (July 2020 Edition); 2020; Stratakos, A.C.; Linton, M.; Millington, S.; Grant, I.R. A Loop-Mediated Isothermal Amplification Method for Rapid Direct Detection and Differentiation of Nonpathogenic and Verocytotoxigenic Escherichia Coli in Beef and Bovine Faeces. J Appl Microbiol 2017, 122, 817–828, doi:10.1111/jam.13381. 132. Capps, K.L.; Mclaughlin, E.M.; Murray, A.W.A.; Aldus, C.F.; Wyatt, G.M.; Peck, M.W.; Van Amerongen, A.; Ariëns, R.M.C.; Wichers, J.H.; Baylis, C.L.; et al. Validation of Three Rapid Screening Methods for Detection of Verotoxin-Producing Escherichia Coli in Foods: Interlaboratory Study; 2004; 133. Hosking, E.; Roman, B.; Alles, S.; Mozola, M.; Hinkley, S.; Cooper, K.; Keys, D.; Bastin, B.; Thompson, W.; Donofrio, R. NeoSeekTM STEC: A Multiplex Molecular Method for Detection and Identification of Select Shiga Toxin–Producing Escherichia Coli in Beef. J AOAC Int 2020, 103, 523–532, doi:10.5740/JAOACINT.19-0300. 134. 135. Tice, G.; Andaloro, B.; Fallon, D. EMERGENCY RESPONSE VALIDATION: Salmonella in Peanut Butter DuPont Qualicon BAX ® System Polymerase Chain Reaction Assay; 2009; Parsons, B.D.; Zelyas, N.; Berenger, B.M.; Chui, L. Detection, Characterization, and Typing of Shiga Toxin-Producing Escherichia Coli. Front Microbiol 2016, 7, 478. 84 136. Viator, R.; Alles, S.; Le, Q.N.; Hosking, E.; Meister, E.; Pinkava, L.; Tovar, E.; Mozola, M.; Rice, J. Validation of the ANSR® E. Coli O157:H7 Method for Detection of E. Coli O157:H7. J AOAC Int 2016, 99, 705–716, doi:10.5740/jaoacint.15-0309. 137. Madic, J.; Vingadassalon, N.; de Garam, C.P.; Marault, M.; Scheutz, F.; Brugere, H.; Jamet, E.; Auvray, F. Detection of Shiga Toxin-Producing Escherichia Coli Serotypes O26: H11, O103: H2, O111: H8, O145: H28, and O157: H7 in Raw-Milk Cheeses by Using Multiplex Real-Time PCR. Appl Environ Microbiol 2011, 77, 2035–2041. 138. Kim, J.; Lee, J.B.; Park, J.; Koo, C.; Lee, M.-S. Recent Advancements in Technologies to Detect Enterohaemorrhagic Escherichia Coli Shiga Toxins. J Microbiol Biotechnol 2023, 33, 559. 139. Grys, T.E.; Sloan, L.M.; Rosenblatt, J.E.; Patel, R. Rapid and Sensitive Detection of Shiga Toxin- Producing Escherichia Coli from Nonenriched Stool Specimens by Real-Time PCR in Comparison to Enzyme Immunoassay and Culture. J Clin Microbiol 2009, 47, 2008–2012. 140. Leonard, S.R.; Simko, I.; Mammel, M.K.; Richter, T.K.S.; Brandl, M.T. Seasonality, Shelf Life and Storage Atmosphere Are Main Drivers of the Microbiome and E. Coli O157:H7 Colonization of Post-Harvest Lettuce Cultivated in a Major Production Area in California. Environmental Microbiomes 2021, 16, doi:10.1186/S40793-021-00393-Y. 141. Danyluk, M.D.; Schaffner, D.W. Quantitative Assessment of the Microbial Risk of Leafy Greens from Farm to Consumption: Preliminary Framework, Data, and Risk Estimates. 2011, doi:10.4315/0362-028X.JFP-10-373. 142. Ottoson, J.R.; Nyberg, K.; Lindqvist, R.; Albihn, A. Quantitative Microbial Risk Assessment for Escherichia Coli O157 on Lettuce, Based on Survival Data from Controlled Studies in a Climate Chamber. 2011, doi:10.4315/0362-028X.JFP-10-563. 143. 144. Pang, H.; Lambertini, E.; Buchanan, R.L.; Schaffner, D.W.; Pradhan, A.K. Quantitative Microbial Risk Assessment for Escherichia Coli O157:H7 in Fresh-Cut Lettuce. 2017, doi:10.4315/0362-028X.JFP-16-246. Tromp, S.O.; Rijgersberg, H.; Franz, E. Quantitative Microbial Risk Assessment for Escherichia Coli O157:H7, Salmonella Enterica, and Listeria Monocytogenes in Leafy Green Vegetables Consumed at Salad Bars, Based on Modeling Supply Chain Logistics. 2010. 145. CDC Foodborne Germs and Illnesses Available online: https://www.cdc.gov/foodsafety/foodborne-germs.html (accessed on 5 January 2024). 146. Bergholz, P.W.; Strawn, L.K.; Ryan, G.T.; Warchocki, S.; Wiedmann, M. Spatiotemporal Analysis of Microbiological Contamination in New York State Produce Fields Following Extensive Flooding from Hurricane Irene, August 2011. J Food Prot 2016, 79, 384–391, doi:10.4315/0362-028X.JFP-15-334. 147. Bach, S.J.; Stanford, K.; Mcallister, T.A. Survival of Escherichia Coli O157:H7 in Feces from Corn- and Barley-Fed Steers. 2005, doi:10.1016/j.femsle.2005.08.030. 148. Artz, R.R.E.; Townend, J.; Brown, K.; Towers, W.; Killham, K. Soil Macropores and Compaction Control the Leaching Potential of Escherichia Coli O157:H7. Environ Microbiol 2005, 7, 241– 248, doi:10.1111/J.1462-2920.2004.00690.X. 149. Ihssen, J.; Grasselli, E.; Bassin, C.; Franç ois, P.; Piffaretti, J.-C.; Kö ster, W.; Schrenzel, J.; Egli, T.; Egli egli, T. Comparative Genomic Hybridization and Physiological Characterization of Environmental Isolates Indicate That Significant (Eco-)Physiological Properties Are Highly Conserved in the Species Escherichia Coli. 2007, doi:10.1099/mic.0.2006/002006-0. 85 150. 151. Ihssen, J.; Egli, T. Global Physiological Analysis of Carbon- and Energy-Limited Growing Escherichia Coli Confirms a High Degree of Catabolic Flexibility and Preparedness for Mixed Substrate Utilization. Environ Microbiol 2005, 7, 1568–1581, doi:10.1111/J.1462- 2920.2005.00846.X. Franz, E.; Van Bruggen, A.H.C. Ecology of E. Coli O157:H7 and Salmonella Enterica in the Primary Vegetable Production Chain. Crit Rev Microbiol 2008, 34, 143–161, doi:10.1080/10408410802357432. 152. Alegbeleye, O.; Sant, A.S. Survival of Salmonella Spp. under Varying Temperature and Soil Conditions. 2023, doi:10.1016/j.scitotenv.2023.163744. 153. Locatelli, A.; Spor, A.; Jolivet, C.; Piveteau, P.; Hartmann, A. Biotic and Abiotic Soil Properties Influence Survival of Listeria Monocytogenes in Soil. PLoS One 2013, 8, 75969, doi:10.1371/journal.pone.0075969. 154. Xing, J.; Wang, H.; Brookes, P.C.; Salles, J.F.; Xu, J. Soil PH and Microbial Diversity Constrain the Survival of E. Coli in Soil. 2018, doi:10.1016/j.soilbio.2018.10.013. 155. 156. Shah, M.K.; Bradshaw, R.; Nyarko, E.; Handy, E.T.; East, C.; Millner, P.D.; Bergholz, T.M.; Sharma, M. Salmonella Enterica in Soils Amended with Heat-Treated Poultry Pellets Survived Longer than Bacteria in Unamended Soils and More Readily Transferred to and Persisted on Spinach. Appl Environ Microbiol 2019, 85, doi:10.1128/AEM.00334-19/FORMAT/EPUB. Shah, M.K.; Bradshaw, R.; Nyarko, E.; Millner, P.D.; Neher, D.; Weicht, T.; Bergholz, T.M.; Sharma, M. Survival and Growth of Wild-Type and RpoS-Deficient Salmonella Newport Strains in Soil Extracts Prepared with Heat-Treated Poultry Pellets. J Food Prot 2019, 82, 501–506, doi:10.4315/0362-028X.JFP-18-465. 157. Neher, D.; Cutler, A.; Weicht, T.; Sharma, M.; Millner, P.; Deborah Neher, C.A. Composts of Poultry Litter or Dairy Manure Differentially Affect Survival of Enteric Bacteria in Fields with Spinach. 2019, doi:10.1111/jam.14268. 158. Hammami, R.; Crippen, T.L.; Strawn, L.K.; Bardsley, C.A.; Weller, D.L.; Ingram, D.T.; Chen, Y.; Oryang, D.; Rideout, S.L. Strain, Soil-Type, Irrigation Regimen, and Poultry Litter Influence | Salmonella Survival and Die-off www.frontiersin.org 2021, 1, 590303, doi:10.3389/fmicb.2021.590303. in Agricultural Soils. Frontiers in Microbiology 159. Callahan, M.T.; Micallef, S.A.; Sharma, M.; Millner, P.D.; Buchanan, R.L. Metrics Proposed to Prevent the Harvest of Leafy Green Crops Exposed to Flood Water Contaminated with Escherichia Coli. Appl Environ Microbiol 2016, 82, 3746–3753, doi:10.1128/AEM.00052- 16/FORMAT/EPUB. 160. Kim, J.; Shepherd, M.W.; Jiang, X. Evaluating the Effect of Environmental Factors on Pathogen Regrowth in Compost Extract. Microb Ecol 2009, 58, 498–508, doi:10.1007/S00248-009-9524- X/FIGURES/2. 161. Lin, Y.; Simsek, S.; Bergholz, T.M. Fate of Salmonella and Shiga-Toxin Producing Escherichia Coli on Wheat Grain during Tempering. Food Microbiol 2023, 111, 104194, doi:10.1016/j.fm.2022.104194. 162. Kroon, H. Determination of Nitrogen in Water: Comparison of a Continuous-Flow Method with on-Line UV Digestion with the Original Kjeldahl Method. Anal Chim Acta 1993, 276, 287–293, doi:https://doi.org/10.1016/0003-2670(93)80396-3. 86 163. 164. Zhan, N.; Huang, Y.; Rao, Z.; Zhao, X.-L. Fast Detection of Carbonate and Bicarbonate in Groundwater and Lake Water by Coupled Ion Selective Electrode. Chinese Journal of Analytical Chemistry 2016, 44, 355–360, doi:https://doi.org/10.1016/S1872-2040(16)60913-1. Schriefer, A.E.; Cliften, P.F.; Hibberd, M.C.; Sawyer, C.; Brown-Kennerly, V.; Burcea, L.; Klotz, E.; Crosby, S.D.; Gordon, J.I.; Head, R.D. A Multi-Amplicon 16S RRNA Sequencing and Analysis Method for Improved Taxonomic Profiling of Bacterial Communities. 2018, doi:10.1016/j.mimet.2018.09.019. 165. Bolyen, E.; Ram Rideout, J.; Dillon, M.R.; Bokulich, N.A.; Abnet, C.C.; Al-Ghalith, G.A.; Alexander, H.; Alm, E.J.; Arumugam, M.; Asnicar, F.; et al. Reproducible, Interactive, Scalable and Extensible Microbiome Data Science Using QIIME 2. 2019, doi:10.1038/s41587-019-0190- 3. 166. Team, R.C., Title Community Ecology Package. 2022. 167. 168. Package “vegan” Title Community Ecology Package Package “vegan” Title Community Ecology Package. 2024. Pedregosa FABIANPEDREGOSA, F.; Michel, V.; Grisel OLIVIERGRISEL, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Vanderplas, J.; Cournapeau, D.; Pedregosa, F.; Varoquaux, G.; et al. Scikit-Learn: Machine Learning in Python Gaël Varoquaux Bertrand Thirion Vincent Dubourg Alexandre Passos PEDREGOSA, VAROQUAUX, GRAMFORT ET AL. Matthieu Perrot. Journal of Machine Learning Research 2011, 12, 2825–2830. 169. Paszke, A.; Gross, S.; Massa, F.; Lerer, A.; Bradbury Google, J.; Chanan, G.; Killeen, T.; Lin, Z.; Gimelshein, N.; Antiga, L.; et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library. 2019. 170. Abadi, M.; Barham, P.; Chen, J.; Chen, Z.; Davis, A.; Dean, J.; Devin, M.; Ghemawat, S.; Irving, G.; Isard, M.; et al. This Paper Is Included in the Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’16). Open Access to the Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation Is Sponsored by USENIX. TensorFlow: A System for Large-Scale Machine Learning TensorFlow: A System for Large-Scale Machine Learning. 171. Kim, J.; Jiang, X. The Growth Potential of Escherichia Coli O157:H7, Salmonella Spp. and Listeria Monocytogenes in Dairy Manure-Based Compost in a Greenhouse Setting under Different Seasons., doi:10.1111/j.1365-2672.2010.04841.x. 172. Ma, J.; Ibekwe, A.M.; Crowley, D.E.; Yang, C.H. Persistence of Escherichia Coli O157:H7 in Major Leafy Green Producing Soils. Environ Sci Technol 2012, 46, 12154–12161, doi:10.1021/ES302738Z. 173. Dirk Van Elsas, J.; Semenov, A. V; Costa, R.; Trevors, J.T. MINI-REVIEW Survival of Escherichia Coli in the Environment: Fundamental and Public Health Aspects. ISME J 2011, 5, 173–183, doi:10.1038/ismej.2010.80. 174. Baker, C.A.; Lee, S.; Deid, J.; Jeong, K.C.; Schneiderid, K.R. Survival of Escherichia Coli O157 in Autoclaved and Natural Sandy Soil Mesocosms. 2020, doi:10.1371/journal.pone.0234562. 175. Mallon, C.A.; Dirk Van Elsas, J.; Salles, J.F. Microbial Invasions: The Process, Patterns, and Mechanisms. 2015, doi:10.1016/j.tim.2015.07.013. 176. Wang, H.; Jiang, X. Isolation and Characterization of Competitive Exclusion Microorganisms from Animal Wastes–Based Composts against Listeria Monocytogenes. J Appl Microbiol 2022, 132, 4531–4543, doi:10.1111/JAM.15528. 87 177. Devarajan, N.; Mcgarvey, J.A.; Scow, K.; Jones, M.S.; Lee, S.; Samaddar, S.; Schmidt, R.; Tran, T.D.; Karp, D.S. Cascading Effects of Composts and Cover Crops on Soil Chemistry, Bacterial Communities and the Survival of Foodborne Pathogens. J Appl Microbiol 2021, 131, 1564–1577, doi:10.1111/jam.15054. 178. Du¡y, G.; Whiting, R.C.; Sheridan, J.J. The Effect of a Competitive Microflora, PH and Temperature on the Growth Kinetics of Escherichia Coli O157: H7. 1999. 179. Messi, P.; Guerrieri, E.; Bondi, M. Antibiotic Resistance and Antibacterial Activity in Heterotrophic Bacteria of Mineral Water Origin. 2005, doi:10.1016/j.scitotenv.2004.12.005. 180. Iglesias, M.B.; Abadias, M.; Anguera, M.; Vi, I.; Nas, ~ Efficacy of Pseudomonas Graminis CPA- 7 against Salmonella Spp. and Listeria Monocytogenes on Fresh-Cut Pear and Setting up of the Conditions for Its Commercial Application. 2017, doi:10.1016/j.fm.2017.09.010. 181. Bach, E.; Pereira Passaglia, M.; Jiao, J.; Gross, H. Burkholderia in the Genomic Era: From the Discovery of New Antimicrobial Secondary Metabolites. 2021, Taxonomy doi:10.1080/1040841X.2021.1946009. to 182. Alwosheel, A.; Van Cranenburgh, S.; Chorus, C.G. Is Your Dataset Big Enough? Sample Size Requirements When Using Artificial Neural Networks for Discrete Choice Analysis. 2018, doi:10.1016/j.jocm.2018.07.002. 183. Sagheer, A.; Kotb, M. Time Series Forecasting of Petroleum Production Using Deep LSTM Recurrent Networks. Neurocomputing 2019, 323, 203–213, doi:10.1016/j.neucom.2018.09.082. 184. Hiura, S.; Koseki, S.; Koyama, K. Prediction of Population Behavior of Listeria Monocytogenes in Food Using Machine Learning and a Microbial Growth and Survival Database. Scientific Reports | 123AD, 11, 10613, doi:10.1038/s41598-021-90164-z. 185. Hui, W.; Shuyi, Y.; Wei, Z.; Junbo, P.; Haiyun, T.; Chunhao, L.; Jiye, Y. Modeling the Dynamic Changes in Plasmopara Viticola Sporangia Concentration Based on LSTM and Understanding the Impact of Relative Factor Variability. Int J Biometeorol 2023, 67, 993–1002, doi:10.1007/s00484-022-02419-7. 186. 187. Salinas Valley Flooding and the Testing Trap Salinas Valley Flooding and the Testing Trap | Food Safety Available online: https://www.food-safety.com/articles/8380-salinas-valley-flooding-and- the-testing-trap (accessed on 8 May 2024). Sant’Ana, A.S.; Franco, B.D.G.M.; Schaffner, D.W. Risk of Infection with Salmonella and Listeria Monocytogenes Due to Consumption of Ready-to-Eat Leafy Vegetables in Brazil. Food Control 2014, 42, 1–8. 188. Mishra, A.; Guo, M.; Buchanan, R.L.; Schaffner, D.W.; Pradhan, A.K. Development of Growth and Survival Models for Salmonella and Listeria Monocytogenes during Non-Isothermal Time- Temperature Profiles in Leafy Greens. Food Control 2017, 71, 32–41. 189. Yang, X.; Scharff, R. Foodborne Illnesses from Leafy Greens in the United States: Attribution, Burden, and Cost. J Food Prot 2024, 87, 100275. 190. Symptoms E. https://www.cdc.gov/ecoli/ecoli-symptoms.html (accessed on 8 May 2024). coli CDC Symptoms | E. Coli | CDC Available online: 191. Koutsoumanis, K.; Allende, A.; Alvarez-Ordóñez, A.; Bover-Cid, S.; Chemaly, M.; Davies, R.; De Cesare, A.; Herman, L.; Hilbert, F.; Lindqvist, R.; et al. Pathogenicity Assessment of Shiga Toxin-Producing Escherichia Coli (STEC) and the Public Health Risk Posed by Contamination of Food with STEC. EFSA Journal 2020, 18, doi:10.2903/J.EFSA.2020.5967. 88 192. FDA Factors Potentially Contributing to the Contamination of Romaine Lettuce Implicated in the Three Outbreaks of E. Coli O157:H7 During the Fall of 2019 | FDA Available online: https://www.fda.gov/food/outbreaks-foodborne-illness/factors-potentially-contributing- contamination-romaine-lettuce-implicated-three-outbreaks-e-coli (accessed on 12 August 2024). 193. Bottichio, L.; Keaton, A.; Thomas, D.; Fulton, T.; Tiffany, A.; Frick, A.; Mattioli, M.; Kahler, A.; Murphy, J.; Otto, M. Shiga Toxin–Producing Escherichia Coli Infections Associated with Romaine Lettuce—United States, 2018. Clinical Infectious Diseases 2020, 71, e323–e330. 194. CDC, Y. 2018 E. Coli Outbreak Linked to Romaine Lettuce A - Advice | Investigation Notice: Multistate Outbreak of E. Coli O157:H7 Infections April 2018 | E. Coli | CDC Available online: https://archive.cdc.gov/www_cdc_gov/ecoli/2018/o157h7-04-18/advice-consumers.html#print (accessed on 23 July 2024). 195. CDC, S. 2019 E. Coli Outbreak Linked to Romaine Lettuce - Investigation Updates | E.Coli Infections online: | E. https://archive.cdc.gov/www_cdc_gov/ecoli/2019/o157h7-11-19/updates.html (accessed on 23 July 2024). CDC Available | November 2019 Coli | 196. CDC, C.C. 2018 E. Coli Outbreak Linked to Romaine Lettuce B - Investigation Updates | E. Coli Infections Linked to Romaine Lettuce | November 2018 | E. Coli | CDC Available online: https://archive.cdc.gov/www_cdc_gov/ecoli/2018/o157h7-11-18/updates.html (accessed on 23 July 2024). 197. Chen, J.C.; Patel, K.; Smith, P.A.; Vidyaprakash, E.; Snyder, C.; Tagg, K.A.; Webb, H.E.; Schroeder, M.N.; Katz, L.S.; Rowe, L.A.; et al. Reoccurring Escherichia Coli O157:H7 Strain Linked to Leafy Greens–Associated Outbreaks, 2016–2019 - Volume 29, Number 9—September 2023 - Emerging Infectious Diseases Journal - CDC. Emerg Infect Dis 2023, 29, 1895–1899, doi:10.3201/EID2909.230069. 198. Sinkel; Daniel Good Agriculture Practices for Lettuce; 199. Rural Migration blog Lettuce Production and Trade - Rural Migration Blog | Migration Dialogue Available online: https://migration.ucdavis.edu/rmn/blog/post/?id=2699 (accessed on 25 July 2024). 200. USDA ERS USDA ERS - Chart Detail Available online: https://www.ers.usda.gov/data- products/chart-gallery/gallery/chart-detail/?chartId=89157 (accessed on 25 July 2024). 201. Lettuce AMRC Lettuce | Agricultural Marketing Resource Center Available online: https://www.agmrc.org/commodities-products/vegetables/lettuce (accessed on 26 August 2024). 202. Smith, R.; Cruz, S.; Benito, S. LEAF LETTUCE PRODUCTION IN CALIFORNIA; 203. Gombas, D.; Gorny, J. Commodity Specific Food Safety Guidelines for the Lettuce and Leafy Greens Supply Chain 2 5 A P R I L 2 0 0 6 TM COMMODITY SPECIFIC FOOD SAFETY GUIDELINES FOR THE LETTUCE AND LEAFY GREENS SUPPLY CHAIN • 1 ST Edition International Fresh-Cut Produce Association; 204. Brown, W.; Ryser, E.; Gorman, L.; Steinmaus, S.; Vorst, K. Transit Temperatures Experienced by Fresh-Cut Leafy Greens during Cross-Country Shipment. Food Control 2016, 61, 146–155, doi:10.1016/j.foodcont.2015.09.014. 205. Gu, G.; Ding, Q.; Redding, M.; Yang, Y.; O’Brien, R.; Gu, T.; Zhang, B.; Zhou, B.; Micallef, S.A.; Luo, Y. Differential Microbiota Shift on Whole Romaine Lettuce Subjected to Source or Forward Processing and on Fresh-Cut Products during Cold Storage. Int J Food Microbiol 2024, 416, 110665. 89 206. Khalil, R.K.S. Effect of Abusive Storage Temperatures on Growth and Survival of Escherichia Coli O157: H7 on Leafy Salad Vegetables in Egypt. LWT-Food Science and Technology 2016, 65, 954–962. 207. Choi, S.; Bang, J.; Kim, H.; Beuchat, L.R.; Ryu, J. Survival and Colonization of Escherichia Coli O157: H7 on Spinach Leaves as Affected by Inoculum Level and Carrier, Temperature and Relative Humidity. J Appl Microbiol 2011, 111, 1465–1472. 208. Owade, J.O.; Bergholz, T.M.; Mitchell, J. A Review of Conditions Influencing Fate of Shiga Toxin‐producing Escherichia Coli O157:H7 in Leafy Greens. Compr Rev Food Sci Food Saf 2024, 23, doi:10.1111/1541-4337.70013. 209. 210. Luo, Y.; He, Q.; McEvoy, J.L. Effect of Storage Temperature and Duration on the Behavior of Escherichia Coli O157: H7 on Packaged Fresh‐cut Salad Containing Romaine and Iceberg Lettuce. J Food Sci 2010, 75, M390–M397. Leonard, S.R.; Simko, I.; Mammel, M.K.; Richter, T.K.S.; Brandl, M.T. Seasonality, Shelf Life and Storage Atmosphere Are Main Drivers of the Microbiome and E. Coli O157: H7 Colonization of Post-Harvest Lettuce Cultivated in a Major Production Area in California. Environ Microbiome 2021, 16, 1–23. 211. KING, A.D.; MAGNUSON, J.A.; TÖRÖK, T.; GOODMAN, N. Microbial Flora and Storage Quality of Partially Processed Lettuce. J Food Sci 1991, 56, 459–461, doi:10.1111/j.1365- 2621.1991.tb05303.x. 212. Baur, S.; Klaiber, R.; Hammes, W.P.; Carle, R. Sensory and Microbiological Quality of Shredded, Packaged Iceberg Lettuce as Affected by Pre-Washing Procedures with Chlorinated and Ozonated Water. Innovative Food Science and Emerging Technologies 2004, 5, 45–55, doi:10.1016/j.ifset.2003.10.002. 213. USDA Peracetic Acid Processing; 2000; 214. Use of Ozone to Enhance the Safety of Produce | National Agricultural Library Available online: https://www.nal.usda.gov/research-tools/food-safety-research-projects/use-ozone-enhance- safety-produce (accessed on 2 September 2024). 215. Davidson, G.R.; Kaminski-Davidson, C.N.; Ryser, E.T. Persistence of Escherichia Coli O157: H7 during Pilot-Scale Processing of Iceberg Lettuce Using Flume Water Containing Peroxyacetic Acid-Based Sanitizers and Various Organic Loads. Int J Food Microbiol 2017, 248, 22–31. 216. Davidson, G.R.; Kaminski-Davidson, C.N.; Ryser, E.T. Persistence of Escherichia Coli O157: H7 during Pilot-Scale Processing of Iceberg Lettuce Using Flume Water Containing Peroxyacetic Acid-Based Sanitizers and Various Organic Loads. Int J Food Microbiol 2017, 248, 22–31. 217. Davidson, G.R.; Buchholz, A.L.; Ryser, E.T. Efficacy of Commercial Produce Sanitizers against Nontoxigenic Escherichia Coli O157: H7 during Processing of Iceberg Lettuce in a Pilot-Scale Leafy Green Processing Line. J Food Prot 2013, 76, 1838–1845. 218. Wang, J.; Wang, S.; Sun, Y.; Li, C.; Li, Y.; Zhang, Q.; Wu, Z. Reduction of Escherichia Coli O157:H7 and Naturally Present Microbes on Fresh-Cut Lettuce Using Lactic Acid and Aqueous Ozone. RSC Adv 2019, 9, 22636–22643, doi:10.1039/c9ra03544c. 219. Ayrapetyan, M.; Williams, T.; Oliver, J.D. Relationship between the Viable but Nonculturable State and Antibiotic Persister Cells. 2018, doi:10.1128/JB.00249-18. 90 220. Dinu, L.D.; Bach, S. Induction of Viable but Nonculturable Escherichia Coli O157:H7 in the Phyllosphere of Lettuce: A Food Safety Risk Factors. Appl Environ Microbiol 2011, 77, 8295– 8302, doi:10.1128/AEM.05020-11. 221. Munther, D.S.; Carter, M.Q.; Aldric, C. V.; Ivanek, R.; Brandl, M.T. Formation of Escherichia Coli O157:H7 Persister Cells in the Lettuce Phyllosphere and Application of Differential Equation Models To Predict Their Prevalence on Lettuce Plants in the Field. Appl Environ Microbiol 2020, 86, doi:10.1128/AEM.01602-19. 222. 223. Thao, S.; Brandl, M.T.; Carter, M.Q. Enhanced Formation of Shiga Toxin-Producing Escherichia Coli Persister Variants in Environments Relevant to Leafy Greens Production 2 3. 2019. Thao, S.; Brandl, M.T.; Carter, M.Q. Enhanced Formation of Shiga Toxin-Producing Escherichia Coli Persister Variants in Environments Relevant to Leafy Greens Production. Food Microbiol 2019, 84, 103241. 224. Carey, C.M.; Kostrzynska, M.; Thompson, S. Escherichia Coli O157: H7 Stress and Virulence Gene Expression on Romaine Lettuce Using Comparative Real-Time PCR. J Microbiol Methods 2009, 77, 235–242. 225. 226. 227. Sharma, M.; Lakshman, S.; Ferguson, S.; Ingram, D.T.; Luo, Y.; Patel, J. Effect of Modified Atmosphere Packaging on the Persistence and Expression of Virulence Factors of Escherichia Coli O157: H7 on Shredded Iceberg Lettuce. J Food Prot 2011, 74, 718–726. Sharma, D.; Kraft, A.L.; Owade, J.O.; Milicevic, M.; Yi, J.; Bergholz, T.M. Impact of Biotic and Abiotic Factors on Listeria Monocytogenes, Salmonella Enterica, and Enterohemorrhagic in Agricultural Soil Extracts. Microorganisms 2024, 12, 1498, Escherichia Coli doi:10.3390/microorganisms12071498. Li, B.; Chen, J.Q. Real-Time PCR Methodology for Selective Detection of Viable Escherichia Coli O157: H7CELLS by Targeting Z3276 as a Genetic Marker. Appl Environ Microbiol 2012, 78, 5297–5304, doi:10.1128/AEM.00794-12/FORMAT/EPUB. 228. Beumer, R.R.; De Vries, J.; Rombouts, F.M. Campylobacter Jejuni Non-Culturable Coccoid Cells; 1992; Vol. 15;. 229. Cardenas-Alvarez, M.X.; Townsend Ramsett, M.K.; Malekmohammadi, S.; Bergholz, T.M. Evidence of Hypervirulence in Listeria Monocytogenes Clonal Complex 14. J Med Microbiol 2019, 68, 1677–1685, doi:10.1099/JMM.0.001076. 230. Scalfaro, C.; Iacobino, A.; Nardis, C.; Franciosa, G. Galleria Mellonella as an in Vivo Model for Assessing the Protective Activity of Probiotics against Gastrointestinal Bacterial Pathogens. FEMS Microbiol Lett 2017, 364, doi:10.1093/femsle/fnx064. 231. Finney, D.J.; Stevens, W.L. A Table for the Calculation of Working Probits and Weights in Probit Analysis; 1948; Vol. 35;. 232. Team, R.C., Title Community Ecology Package. 2022. 233. Wickham, H.; François, R.; Henry, L.; Müller, K.; Vaughan, D. Dplyr: A Grammar of Data Manipulation. R Package Version 1.1. 2. Computer software 2023. 234. Searle, S.R.; Speed, F.M.; Milliken, G.A. Population Marginal Means in the Linear Model: An Alternative to Least Squares Means. American Statistician 1980, 34, 216–221, doi:10.1080/00031305.1980.10483031. 91 235. 236. 237. Therneau, T.; Atkinson, B.; Ripley, B. Rpart: Recursive Partitioning and Regression Trees. R Package Version 4.1-15. Retrieved 2015, 13, 2015. James, S.J.; James, C.; Evans, J.A. Modelling of Food Transportation Systems–a Review. International Journal of Refrigeration 2006, 29, 947–957. Zhao, Y.; Zhang, X.; Xu, X. Application and Research Progress of Cold Storage Technology in Cold Chain Transportation and Distribution. J Therm Anal Calorim 2020, 139, 1419–1434. 238. Qi, T.; Ji, J.; Zhang, X.; Liu, L.; Xu, X.; Ma, K.; Gao, Y. Research Progress of Cold Chain Transport Technology for Storage Fruits and Vegetables. J Energy Storage 2022, 56, 105958. 239. Russell, N.J. Bacterial Membranes: The Effects of Chill Storage and Food Processing. An Overview; 2022; 240. Strocchi, M.; Ferrer, M.; Timmis, K.N.; Golyshin, P.N. Low Temperature-Induced Systems Failure in Escherichia Coli: Insights from Rescue by Cold-Adapted Chaperones. Proteomics 2006, 6, 193–206, doi:10.1002/PMIC.200500031. 241. Barria, C.; Malecki, M.; Arraiano, C.M. Bacterial Adaptation to Cold. Microbiology (United Kingdom) 2013, 159, 2437–2443. 242. Yamanaka, K. Cold Shock Response in Escherichia Coli. J Mol Microbiol Biotechnol 1999, 1, 193–202. 243. King, T.; Vockler, C.J.; Allnutt, T.R.; Fegan, N. Transcriptomic Response of Escherichia Coli O157 Isolates on Meat: Comparison between a Typical Australian Isolate from Cattle and a Pathogenic Clinical Isolate. Food Microbiol 2019, 82, 378–387, doi:10.1016/j.fm.2019.03.008. 244. Dinu, L.D.; Bach, S. Induction of Viable but Nonculturable Escherichia Coli O157:H7 in the Phyllosphere of Lettuce: A Food Safety Risk Factors. Appl Environ Microbiol 2011, 77, 8295– 8302, doi:10.1128/AEM.05020-11/ASSET/E8B601D4-47E0-4CF4-9E51- 69150FB498DF/ASSETS/GRAPHIC/ZAM9991027240004.JPEG. 245. Zhang, Y.; Gross, C.A. Cold Shock Response in Bacteria. Annu Rev Genet 2021, 55, 377–400. 246. Phadtare, S.; Inouye, M. Genome-Wide Transcriptional Analysis of the Cold Shock Response in Wild-Type and Cold-Sensitive, Quadruple-Csp-Deletion Strains of Escherichia Coli. J Bacteriol 2004, 186, 7007–7014. 247. Keto-Timonen, R.; Hietala, N.; Palonen, E.; Hakakorpi, A.; Lindström, M.; Korkeala, H. Cold Shock Proteins: A Minireview with Special Emphasis on Csp-Family of Enteropathogenic Yersinia. Front Microbiol 2016, 7, 1151. 248. White-Ziegler, C.A.; Um, S.; Pérez, N.M.; Berns, A.L.; Malhowski, A.J.; Young, S. Low Temperature (23 °C) Increases Expression of Biofilm-, Cold-Shock- and RpoS-Dependent Genes in 148–166, doi:10.1099/mic.0.2007/012021-0. Escherichia Coli K-12. Microbiology (N Y) 2008, 154, 249. King, T.; Kocharunchitt, C.; Gobius, K.; Bowman, J.P.; Ross, T. Global Genome Response of Escherichia Coli O157:H7 Sakai during Dynamic Changes in Growth Kinetics Induced by an Abrupt Temperature Downshift. PLoS One 2014, 9, doi:10.1371/journal.pone.0099627. 250. Kocharunchitt, C.; King, T.; Gobius, K.; Bowman, J.P.; Ross, T. Integrated Transcriptomic and Proteomic Analysis of the Physiological Response of Escherichia Coli O157:H7 Sakai to Steady- State Conditions of Cold and Water Activity Stress. Molecular and Cellular Proteomics 2012, 11, doi:10.1074/mcp.M111.009019. 92 251. Jiang, W.; Hou, Y.; Inouye, M. CspA, the Major Cold-Shock Protein of Escherichia Coli, Is an RNA Chaperone. Journal of Biological Chemistry 1997, 272, 196–202. 252. Crozier, L.; Hedley, P.E.; Morris, J.; Wagstaff, C.; Andrews, S.C.; Toth, I.; Jackson, R.W.; Holden, N.J. Whole-Transcriptome Analysis of Verocytotoxigenic Escherichia Coli O157: H7 (Sakai) Suggests Plant-Species-Specific Metabolic Responses on Exposure to Spinach and Lettuce Extracts. Front Microbiol 2016, 7, 1088. 253. Kennedy, N.M.; Mukherjee, N.; Banerjee, P. Escherichia Coli O157: H7 Cells Exposed to Lettuce Leaf Lysate in Refrigerated Conditions Exhibit Differential Expression of Selected Virulence and Adhesion-Related Genes with Altered Mammalian Cell Adherence. J Food Prot 2016, 79, 1259– 1265. 254. Keren, I.; Shah, D.; Spoering, A.; Kaldalu, N.; Lewis, K. Specialized Persister Cells and the Mechanism of Multidrug Tolerance in Escherichia Coli. J Bacteriol 2004, 186, 8172–8180. 255. Yamanaka, K.; Fang, L.; Inouye, M. The CspA Family in Escherichia Coli: Multiple Gene Duplication for Stress Adaptation. Mol Microbiol 1998, 27, 247–255. 256. Kim, Y.; Wang, X.; Zhang, X.; Grigoriu, S.; Page, R.; Peti, W.; Wood, T.K. Escherichia Coli Toxin/Antitoxin Pair MqsR/MqsA Regulate Toxin CspD. Environ Microbiol 2010, 12, 1105– 1121. 257. Jordan, K.N.; Oxford, L.; O’Byrne, C.P. Survival of Low-PH Stress by Escherichia Coli O157: H7: Correlation between Alterations in the Cell Envelope and Increased Acid Tolerance. Appl Environ Microbiol 1999, 65, 3048–3055. 258. Tosun, H.; GÖNÜL, Ş.A. The Effect of Acid Adaptation Conditions on Acid Tolerance Response of Escherichia Coli O157: H7. Turkish Journal of Biology 2005, 29, 197–202. 259. Choi, S.H.; Baumler, D.J.; Kaspar, C.W. Contribution of Dps to Acid Stress Tolerance and Oxidative Stress Tolerance in Escherichia Coli O157: H7. Appl Environ Microbiol 2000, 66, 3911–3916. 260. Elhanafi, D.; Leenanon, B.; Bang, W.; Drake, M.A. Impact of Cold and Cold-Acid Stress on Poststress Tolerance and Virulence Factor Expression OfEscherichia Coli O157: H7. J Food Prot 2004, 67, 19–26. 261. Chua, D.; Goh, K.; Saftner, R.A.; Bhagwat, A.A. Fresh‐cut Lettuce in Modified Atmosphere Packages Stored at Improper Temperatures Supports Enterohemorrhagic E. Coli Isolates to Survive Gastric Acid Challenge. J Food Sci 2008, 73, M148–M153. 262. Tyagi, D.; Kraft, A.L.; Smith, S.L.; Roof, S.E.; Sherwood, J.S.; Wiedmann, M.; Bergholz, T.M. Pre-Harvest Survival and Post-Harvest Chlorine Tolerance of Enterohemorrhagic Escherichia Coli on Lettuce. Toxins (Basel) 2019, 11, doi:10.3390/toxins11110675. 263. Abnavi, M.D. Chlorine Decay and Pathogen Cross Contamination Dynamics in Fresh Produce Washing Process; Cleveland State University, 2021; ISBN 9798535577586. 264. Garcia, A.; Mount, J.R.; Davidson, P.M. Ozone and Chlorine Treatment of Minimally Processed Lettuce. J Food Sci 2003, 68, 2747–2751. 265. Ramarao, N.; Nielsen-Leroux, C.; Lereclus, D. The Insect <Em>Galleria Mellonella</Em> as a Powerful Infection Model to Investigate Bacterial Pathogenesis. Journal of Visualized Experiments 2012, doi:10.3791/4392. 93 266. Kennedy, N.M.; Mukherjee, N.; Banerjee, P. Escherichia Coli O157: H7 Cells Exposed to Lettuce Leaf Lysate in Refrigerated Conditions Exhibit Differential Expression of Selected Virulence and Adhesion-Related Genes with Altered Mammalian Cell Adherence. J Food Prot 2016, 79, 1259– 1265. 267. Allen, K.J.; Lepp, D.; McKellar, R.C.; Griffiths, M.W. Examination of Stress and Virulence Gene Expression in Escherichia Coli O157: H7 Using Targeted Microarray Analysis. Foodborne Pathog Dis 2008, 5, 437–447. 268. Carey, C.M.; Kostrzynska, M.; Thompson, S. Escherichia Coli O157: H7 Stress and Virulence Gene Expression on Romaine Lettuce Using Comparative Real-Time PCR. J Microbiol Methods 2009, 77, 235–242. 269. Sharma, M.; Lakshman, S.; Ferguson, S.; Ingram, D.T.; Luo, Y.; Patel, J. Effect of Modified Atmosphere Packaging on the Persistence and Expression of Virulence Factors of Escherichia Coli O157: H7 on Shredded Iceberg Lettuce. J Food Prot 2011, 74, 718–726. 270. Yoo, B.K.; Liu, Y.; Juneja, V.; Huang, L.; Hwang, C.-A.; Desk, S. Effects of Stresses on the Growth and Cytotoxicity of Shiga-Toxin Producing Escherichia Coli in Ground Beef and Spinach. SDRP Journal Of Food Science & Technology 2016, 1. 271. Habets, A.; Antoine, C.; Wagemans, J.; Vermeersch, M.; Laforêt, F.; Diderich, J.; Lavigne, R.; Mainil, J.; Thiry, D. Impact of Shiga-Toxin Encoding Gene Transduction from O80: H2 Shiga Toxigenic Escherichia Coli (STEC) on Non-STEC Strains. Sci Rep 2022, 12, 21587. 272. Koutsoumanis, K.; Allende, A.; Alvarez-Ordóñez, A.; Bover-Cid, S.; Chemaly, M.; Davies, R.; De Cesare, A.; Herman, L.; Hilbert, F.; Lindqvist, R.; et al. Pathogenicity Assessment of Shiga Toxin-Producing Escherichia Coli (STEC) and the Public Health Risk Posed by Contamination of Food with STEC. EFSA Journal 2020, 18, doi:10.2903/J.EFSA.2020.5967. 273. Chen, J.C.; Patel, K.; Smith, P.A.; Vidyaprakash, E.; Snyder, C.; Tagg, K.A.; Webb, H.E.; Schroeder, M.N.; Katz, L.S.; Rowe, L.A.; et al. Reoccurring Escherichia Coli O157:H7 Strain Linked to Leafy Greens–Associated Outbreaks, 2016–2019 - Volume 29, Number 9—September 2023 - Emerging Infectious Diseases Journal - CDC. Emerg Infect Dis 2023, 29, 1895–1899, doi:10.3201/EID2909.230069. 274. Carter, M.Q.; Carychao, D.; Bono, J.L. Comparative Analyses of Persistence Traits in Escherichia Coli O157:H7 Strains Belonging to Different Clades Including REPEXH01 and REPEXH02 Strains. Front Microbiol 2024, 15, doi:10.3389/fmicb.2024.1501956. 275. Munther, D.S.; Carter, M.Q.; Aldric, C. V.; Ivanek, R.; Brandl, M.T. Formation of Escherichia Coli O157:H7 Persister Cells in the Lettuce Phyllosphere and Application of Differential Equation Models To Predict Their Prevalence on Lettuce Plants in the Field. Appl Environ Microbiol 2020, 86, doi:10.1128/AEM.01602-19. 276. 277. EFSA; ECDC The European Union One Health 2021 Zoonoses Report. EFSA Journal 2022, 20, doi:10.2903/j.efsa.2022.7666. Ethelberg, S.; Smith, B.; Torpdahl, M.; Lisby, M.; Boel, J.; Jensen, T.; Nielsen, E.M.; Molbak, K. Outbreak of Non-0157 Shiga Toxin Producing Escherichia Coli Infection from Consumption of Beef Sausage. Clinical Infectious Diseases 2009, 48, doi:10.1086/597502. 278. Martin, A.; Beutin, L. Characteristics of Shiga Toxin-Producing Escherichia Coli from Meat and Milk Products of Different Origins and Association with Food Producing Animals as Main Contamination 99–104, J doi:10.1016/j.ijfoodmicro.2011.01.041. Food Microbiol Sources. 2011, 146, Int 94 279. Turner, K.; Nou Moua, C.; Hajmeer, M.; Barnes, A.; Needham, M. Overview of Leafy Greens- Related Food Safety Incidents with a California Link: 1996 to 2016. 2018, doi:10.4315/0362- 028X.JFP-18-316. 280. Kintz, E.; Byrne, L.; Jenkins, C.; McCarthy, N.; Vivancos, R.; Hunter, P. Outbreaks of Shiga Toxin–Producing Escherichia Coli Linked to Sprouted Seeds, Salad, and Leafy Greens: A Systematic Review. J Food Prot 2019, 82, 1950–1958. 281. Marshall, K.E.; Hexemer, A.; Seelman, S.L.; Fatica, M.K.; Blessington, T.; Hajmeer, M.; Kisselburgh, H.; Atkinson, R.; Hill, K.; Sharma, D. Lessons Learned from a Decade of Investigations of Shiga Toxin–Producing Escherichia Coli Outbreaks Linked to Leafy Greens, United States and Canada. Emerg Infect Dis 2020, 26, 2319. 282. Michino, H.; Araki, K.; Minami, S.; Takaya, S.; Sakai, N.; Miyazaki, M.; Ono, A.; Yanagawa, H. Massive Outbreak of Escherichia Coli0157:H7 Infection in Schoolchildren in Sakai City, Japan, Associated with Consumption of White Radish Sprouts; 1999; Vol. 150;. 283. Baranzoni, G.M.; Fratamico, P.M.; Rubio, F.; Glaze, T.; Bagi, L.K.; Albonetti, S. Detection and Isolation of Shiga Toxin-Producing Escherichia Coli (STEC) O104 from Sprouts. Int J Food Microbiol 2014, 173, 99–104, doi:10.1016/j.ijfoodmicro.2013.12.020. 284. Rosauer, M.L.; Barnes, C.; Silbernagel, K.M.; Koch, K.M.; Thompson-Strehlow, L. Validation of the 3MTM Molecular Detection Assay 2 - STEC Gene Screen (Stx) for the Detection of Shiga Toxin Gene (Stx1 and/or Stx2) in Additional Matrixes: AOAC Performance Tested MethodSM071903. J AOAC Int 2022, 105, 1126–1135, doi:10.1093/jaoacint/qsac014. 285. Kanki, M.; Seto, K.; Kumeda, Y. Simultaneous Immunomagnetic Separation Method for the Detection of Escherichia Coli O26, O111, and O157 from Food Samples. J Food Prot 2014, 77, 15–22, doi:10.4315/0362-028X.JFP-13-225. 286. Truchado, P.; Gómez-Galindo, M.; Gil, M.I.; Allende, A. Cross-Contamination of Escherichia Coli O157: H7 and Listeria Monocytogenes in the Viable but Non-Culturable (VBNC) State during Washing of Leafy Greens and the Revival during Shelf-Life. Food Microbiol 2023, 109, 104155. 287. Heijnen, L.; Medema, G. Quantitative Detection of E. Coli, E. Coli O157 and Other Shiga Toxin Producing E. Coli in Water Samples Using a Culture Method Combined with Real-Time PCR. J Water Health 2006, 4, 487–498. 288. De Boer, E.; Heuvelink, A.E. Methods for the Detection and Isolation of Shiga Toxin‐producing Escherichia Coli. J Appl Microbiol 2000, 88, 133S-143S. 289. Khalil, R.K.S.; Gomaa, M.A.E.; Khalil, M.I.M. Detection of Shiga-Toxin Producing E. Coli (STEC) in Leafy Greens Sold at Local Retail Markets in Alexandria, Egypt. Int J Food Microbiol 2015, 197, 58–64. 290. Delbeke, S.; Ceuppens, S.; Holvoet, K.; Samuels, E.; Sampers, I.; Uyttendaele, M. Multiplex Real-Time PCR and Culture Methods for Detection of Shiga Toxin-Producing Escherichia Coli and Salmonella Thompson in Strawberries, a Lettuce Mix and Basil. Int J Food Microbiol 2015, 193, 1–7. 291. Costa-Ribeiro, A.; Lamas, A.; Mora, A.; Prado, M.; Garrido-Maestu, A. Moving towards On-Site Detection of Shiga Toxin-Producing Escherichia Coli in Ready-to-Eat Leafy Greens. Curr Res Food Sci 2024, 8, 100716. 95 292. Garrido-Maestu, A.; Azinheiro, S.; Carvalho, J.; Fuciños, P.; Prado, M. Optimized Sample Treatment, Combined with Real-Time PCR, for Same-Day Detection of E. Coli O157 in Ground Beef and Leafy Greens. Food Control 2020, 108, 106790. 293. Owade, J.O.; Bergholz, T.M.; Mitchell, J. A Review of Conditions Influencing Fate of Shiga Toxin‐producing Escherichia Coli O157: H7 in Leafy Greens. Compr Rev Food Sci Food Saf 2024, 23, e70013. 294. Danyluk, M.D.; Schaffner, D.W. Quantitative Assessment of the Microbial Risk of Leafy Greens from Farm to Consumption: Preliminary Framework, Data, and Risk Estimates. J Food Prot 2011, 74, 700–708. 295. Ottoson, J.R.; Nyberg, K.; Lindqvist, R.; Albihn, A. Quantitative Microbial Risk Assessment for Escherichia Coli O157 on Lettuce, Based on Survival Data from Controlled Studies in a Climate Chamber. J Food Prot 2011, 74, 2000–2007. 296. 297. Pang, H.; Lambertini, E.; Buchanan, R.L.; Schaffner, D.W.; Pradhan, A.K. Quantitative Microbial Risk Assessment for Escherichia Coli O157: H7 in Fresh-Cut Lettuce. J Food Prot 2017, 80, 302–311. Franz, E.; Tromp, S.O.; Rijgersberg, H.; Van Der Fels-Klerx, H.J. Quantitative Microbial Risk Assessment for Escherichia Coli O157: H7, Salmonella, and Listeria Monocytogenes in Leafy Green Vegetables Consumed at Salad Bars. J Food Prot 2010, 73, 274–285. 298. Avery, L.M.; Williams, A.P.; Killham, K.; Jones, D.L. Survival of Escherichia Coli O157: H7 in Waters from Lakes, Rivers, Puddles and Animal-Drinking Troughs. Science of the total environment 2008, 389, 378–385. 299. CDC, Y. 2018 E. Coli Outbreak Linked to Romaine Lettuce A - Advice | Investigation Notice: Multistate Outbreak of E. Coli O157:H7 Infections April 2018 | E. Coli | CDC Available online: https://archive.cdc.gov/www_cdc_gov/ecoli/2018/o157h7-04-18/advice-consumers.html#print (accessed on 23 July 2024). 300. CDC, C.C. 2018 E. Coli Outbreak Linked to Romaine Lettuce B - Investigation Updates | E. Coli Infections Linked to Romaine Lettuce | November 2018 | E. Coli | CDC Available online: https://archive.cdc.gov/www_cdc_gov/ecoli/2018/o157h7-11-18/updates.html (accessed on 23 July 2024). 301. CDC, S. 2019 E. Coli Outbreak Linked to Romaine Lettuce - Investigation Updates | E.Coli Infections online: | E. https://archive.cdc.gov/www_cdc_gov/ecoli/2019/o157h7-11-19/updates.html (accessed on 23 July 2024). CDC Available | November 2019 Coli | 302. Most online: connect.foodsafetyrisk.org/microbial/mpncalc/ (accessed on 27 January 2025). Calculator Available Probable Number https://pub- 303. Hothorn, T.; Hornik, K.; Zeileis, A. Unbiased Recursive Partitioning: A Conditional Inference Framework. Journal of Computational and Graphical statistics 2006, 15, 651–674. 304. Hothorn, T.; Hornik, K.; Zeileis, A. Ctree: Conditional Inference Trees. The comprehensive R archive network 2015, 8, 1–34. 305. Carter, M.Q.; Carychao, D.; Bono, J.L. Comparative Analyses of Persistence Traits in Escherichia Coli O157:H7 Strains Belonging to Different Clades Including REPEXH01 and REPEXH02 Strains. Front Microbiol 2024, 15, doi:10.3389/fmicb.2024.1501956. 96 306. Hara-Kudo, Y.; Onoue, Y.; Konuma, H.; Nakagawa, H.; Kumagai, S. Comparison of Enrichment Procedures for Isolation of Escherichia Coli O157:H7 from Ground Beef and Radish Sprouts; 1999; Vol. 50;. 307. Vimont, A.; Vernozy-Rozand, C.; Montet, M.P.; Lazizzera, C.; Bavai, C.; Delignette-Muller, M.L. Modeling and Predicting the Simultaneous Growth of Escherichia Coli O157:H7 and Ground Beef Background Microflora for Various Enrichment Protocols. Appl Environ Microbiol 2006, 72, 261–268, doi:10.1128/AEM.72.1.261-268.2006. 308. CDC, R. E. Coli REP Strain Notices | E. Coli Infection | CDC Available online: https://www.cdc.gov/ecoli/php/data-research/rep-strains/index.html (accessed on 27 February 2025). 309. Manning, S.D.; Motiwala, A.S.; Springman, A.C.; Qi, W.; Lacher, D.W.; Ouellette, L.M.; Mladonicky, J.M.; Somsel, P.; Rudrik, J.T.; Dietrich, S.E. Variation in Virulence among Clades of Escherichia Coli O157: H7 Associated with Disease Outbreaks. Proceedings of the National Academy of Sciences 2008, 105, 4868–4873. 310. Koutsoumanis, K.; Allende, A.; Alvarez-Ordóñez, A.; Bover-Cid, S.; Chemaly, M.; Davies, R.; De Cesare, A.; Herman, L.; Hilbert, F.; Lindqvist, R.; et al. Pathogenicity Assessment of Shiga Toxin-Producing Escherichia Coli (STEC) and the Public Health Risk Posed by Contamination of Food with STEC. EFSA Journal 2020, 18, doi:10.2903/J.EFSA.2020.5967. 311. Chen, J.C.; Patel, K.; Smith, P.A.; Vidyaprakash, E.; Snyder, C.; Tagg, K.A.; Webb, H.E.; Schroeder, M.N.; Katz, L.S.; Rowe, L.A.; et al. Reoccurring Escherichia Coli O157:H7 Strain Linked to Leafy Greens–Associated Outbreaks, 2016–2019 - Volume 29, Number 9—September 2023 - Emerging Infectious Diseases Journal - CDC. Emerg Infect Dis 2023, 29, 1895–1899, doi:10.3201/EID2909.230069. 97 APPENDIX A: SOIL EXTRACTS Figure A. 1. Survival of L. monocytogenes, EHEC, and Salmonella in 4 kinds of soil extracts. Survival of L. monocytogenes (1040S, H7858), EHEC (MI-0041B, DA-5), and Salmonella strains (S10 1646, Mdd314) in (a) LS-low-nutrient sterile, (b) LNS-low-nutrient non-sterile, (c) HS-high-nutrient sterile, and (d) HNS-high-nutrient non-sterile soil extracts over a two-week period. Each point represents four replicates of each strain and error bars denote SD. 98 Figure A. 2. Relative abundance of top 25 taxa from low- (top) and high- (bottom) nutrient soil extracts. The Y-axis shows the relative abundance of bacterial families, and X-axis shows soil extracts from different days. EC, LM, Sal, and Un refer to E. coli, L. monocytogenes, S. enterica, and uninoculated soil extracts respectively. Days are represented by d0, d1, d4, d6, d8, d10, and d14. The black vertical bars represent separation based on soil extract type. The black horizontal arrows represent the progression of the days from 0 to 14. 99 Figure A. 3. Taxa whose relative abundance differed on d0 and d14 in High and Low nutrient soil extracts. The X- axis shows nutrient soil extracts. HNS and LNS represent high- and low- nutrient soil extracts respectively. The Y-axis represents taxa at Family level. Figure A. 4. Taxa whose relative abundance differed from d0 to d14. 100 Figure A. 4. (cont’d) The X- axis shows bacteria in High-nutrient soil extracts; EHEC, Lm, and Sal represent soil extracts inoculated with E. coli, L. monocytogenes, Salmonella. and uninoculated soil extracts. The Y-axis represents taxa at Family level. Figure A. 5. Taxa whose relative abundance differed from d0 to d14. The X- axis shows bacteria in Low-nutrient soil extracts; Lm, and Sal represent soil extracts inoculated with L. monocytogenes, Salmonella, and uninoculated soil extracts. The Y-axis represents taxa at Family level. Figure A. 6. Alpha diversity plots of microbiome samples based on three pathogens. 101 Figure A. 6. (cont’d) Alpha diversity (Shannon entropy) or microbiome samples based on inoculation with a foodborne pathogen compared to soil extract samples that remained uninoculated. Each boxplot represents the Shannon index values for 42 samples for each inoculation type. Boxes represent the 25thand 75th percentiles and the horizontal bar represents the median value for each distribution. Whiskers indicate the 10th and 90th percentiles and outliers are indicated with circles. Figure A. 7. Eigenvalues and explained variances of principal components (PCs) generated by PCA. The dotted and solid blue lines represent the individual and cumulative proportions of the variance explained by PCs, respectively. 102 Figure A. 8. A PCA plot of variables and observations across sampling time points. Different clusters, identified by k-means clustering, are shown in different symbols. Their clear separation suggests that a model can effectively learn and capture patterns in the data for better predictions. PC1: the first principal component that captures the most variation in the dataset. PC2: the second principal component.  103 Figure A. 9. PCA loadings of top variables explaining variation in soil chemistry and microbial composition. PCA loadings of top variables on selected PCs that explain significant variation in the soil chemistry and microbial composition dataset. 104 Figure A. 10. Time-series prediction of pathogen survival. Time-series prediction of pathogen survival by the LSTM model trained on sequential datasets from days 0–4. Top row: high nutrient soil extract. Bottom row: low nutrient soil extract. 105 APPENDIX B: PHYSIOLOGICAL CHANGES IN STEC Figure B. 1. Effect of strain and harvest temperature on VBNC cells of STEC O157:H7 on Romaine lettuce. Post-hoc tests for pairwise comparison of the interactive effect of strain and harvesting temperature on VBNC cells of STEC O157:H7 on Romaine lettuce. The pairwise comparison for the significant effect of interaction of strain and harvesting temperature on VBNC cells was conducted using Tukey’s HSD test. For confidence interval values that are either higher or lower than 0 are significantly (p<0.05) different whereas one that includes zero in the range are not significantly (p>0.05) different.  106   Figure B. 2. Post-hoc tests for pairwise comparison of the interactive effect of strain and harvesting temperature on persister percentage of STEC O157:H7 on Romaine lettuce.   The pairwise comparison for the significant effect of interaction of strain and harvesting temperature on persister percentage was conducted using Tukey’s HSD test. For confidence interval values that are either higher or lower than 0 are significantly (p<0.05) different whereas one that includes zero in the range are not significantly (p>0.05) different.  Figure B. 3. Effect of strains on injured cells, and culturable cells of STEC O157:H7 on Romaine lettuce. Pairwise comparison of the effect of strain on the injured and culturable cells of STEC O157:H7 during incubation of Romaine lettuce at harvesting temperature. For confidence interval values that are either higher or lower than 0 are significantly (p<0.05) different whereas one that includes zero in the range are not significantly (p>0.05) different.  107     Figure B. 4. Pairwise comparison of the predicted log reduction values of the culturable cells for different strains in lettuce harvested at different temperatures. Following analysis of covariance of the effect of the strain and harvest temperature on the log reduction of culturable cells over cold storage period, marginal means function was used to estimate the average values.  Figure B. 5. Factors explain variance in physiological states, acid, and chlorine tolerance. 108     Figure B. 5. (cont’d) Most important factors explaining the variance in (A) culturable and (B) VBNC cells and (C) acid and (D) chlorine tolerance. The measurements were generated from 100 iterations of bootstrapping and the variance explained expressed as a proportion.  Figure B. 6. Pairwise comparison of the predicted increase in VBNC cells for different strains in lettuce harvested at different temperatures. Following analysis of covariance of the effect of the strain and harvest temperature on the formation of VBNC cells over cold storage period, marginal means function was used to estimate the average values.  Figure B. 7. Pairwise comparison of the predicted increase in log reduction due to acid treatment. 109     Figure B. 7. (cont’d) Following analysis of covariance of the effect of the strain and harvest temperature on log reduction due to acid treatment over cold storage period, marginal means function was used to estimate the average values.  Figure B. 8. Pairwise comparison of the predicted increase in log reduction due to chlorine treatment. Marginal means function has been used to predict the change in log reduction due to chlorine treatment for all the strains.  110   Figure B. 9. Odd ratios for the factors affecting the survival of G. mellonella after inoculation with STEC cells from cold stored lettuce. Baseline categories are outbreak strain Central Coast 2018, 0 day and 9 ℃ harvest temperature. Odd ratio >1 denote higher survival, <1 denote less survival whereas 1 denote equal survival to baseline.    111   APPENDIX C: DETECTION OF STEC Table C. 1. Table showing details about samples corresponding to persister percentage. Strain Replicate Enrichment Time Persister Percentage Detection Results Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Central Coast, CA 2018 Central Coast, CA 2018 R1 R2 R3 R1 R2 R3 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 R1 R2 8h 8h 8h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 0.12 0.12 0.12 0.12 0.12 0.12 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.19 0.26 0.26 0.26 0.26 0.26 0.26 0.26 0.26 0.35 0.35 0.35 0.35 0.35 0.35 0.35 0.35 0.50 0.50 0.50 0.50 0.50 0.50 1.09 1.09 N N N P P P P P P P P P P P N N N N N P P P N N P P P P P P N N N P P P N N 112 Table C. 1. (cont’d) Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 24h 24h 24h 1.09 1.09 1.09 1.09 1.09 1.09 2.47 2.47 2.47 2.47 2.47 2.47 2.47 2.47 2.50 2.50 2.50 2.50 2.50 2.50 2.70 2.70 2.70 2.70 2.70 2.70 2.70 2.70 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 3.30 N N N N N N N N N N P P P N N N N P P P N N N N N N N N N N N N P P P P N N N P P N 113 Table C. 1. (cont’d) Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 R1 R2 R1 R2 R1 R2 R1 R2 8h 8h 8h 8h 8h 8h 8h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 8h 8h 8h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 3.50 3.50 3.50 3.50 3.72 3.72 3.72 3.72 3.72 3.72 5.00 5.00 5.00 5.00 5.00 5.00 6.54 6.54 6.54 6.54 6.54 6.54 6.54 6.54 6.90 6.90 6.90 6.90 9.24 9.24 9.24 9.24 9.24 9.24 16.67 16.67 16.67 16.67 16.67 16.67 16.67 16.67 N N N N N N N P P P N N N N N P P P P P P P P P N N P N N N N P P P P P P P P P P P 114 Table C. 1. (cont’d) Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Yuma, AZ 2018 Yuma, AZ 2018 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 24h 24h 24h 24h 8h 8h 16.70 16.70 16.70 16.70 19.40 19.40 19.40 19.40 19.40 19.40 19.40 19.40 19.40 19.40 19.40 19.40 26.60 26.60 26.60 26.60 26.60 26.60 26.60 26.60 26.60 26.60 26.60 26.60 32.62 32.62 32.62 32.62 32.62 32.62 32.62 32.62 35.50 35.50 35.50 35.50 35.50 35.50 N N N N P N P P N N N N P N P P P P P P N N N N P P P P P P P P P P P P N N N N N N 115 Table C. 1. (cont’d) Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Yuma, AZ 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Central Coast, CA 2018 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 Salinas, CA 2019 R1 R2 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 R1 R2 R1 R2 R1 R2 R1 R2 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 R2 R3 R1 8h 8h 24h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 8h 8h 24h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 8h 24h 24h 24h 8h 8h 8h 24h 35.50 35.50 35.50 35.50 35.50 35.50 68.00 68.00 68.00 68.00 68.00 68.00 80.00 80.00 80.00 80.00 80.00 80.00 80.00 80.00 85.00 85.00 85.00 85.00 85.00 85.00 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 <1 N N P N P P N N N N N N P P P P P P P P N N N P P P N N N P P P N N N P P P N N N P 116 Table C. 1. (cont’d) Salinas, CA 2019 Salinas, CA 2019 R2 R3 24h 24h <1 <1 N P 117