Effect of Weather on the Die-Off of Escherichia coli and Attenuated Salmonella enterica Serovar Typhimurium on Preharvest Leafy Greens following Irrigation with Contaminated Water

The log-linear die-off rate proposed by FSMA is not always appropriate, as the die-off rates of foodborne bacterial pathogens and specified agricultural water quality indicator organisms appear to commonly follow a biphasic pattern with an initial rapid decline followed by a period of tailing. While we observed substantial variation in the net culturable population levels of Salmonella and E. coli at each time point, die-off rate and FSMA compliance (i.e., at least a 2 log10 die-off over 4 days) appear to be impacted by produce type, bacteria, and weather; die-off on lettuce tended to be faster than that on spinach, die-off of E. coli tended to be faster than that of attenuated Salmonella, and die-off tended to become faster as relative humidity decreased. Thus, the use of a single die-off rate for estimating time-to-harvest intervals across different weather conditions, produce types, and bacteria should be revised.


Supplementary Information
. Summary statistics for weather factors across 11 trials over the entire 96h experiment.  Table 3. b Q1: 1 st quartile, or 25% of observations are below and 75% of observations are above this value; Q3: 3 rd quartile, or 75% of observations are below and 25% of observations are above this value. c SD: standard deviation in the weather variables across trials. a Descriptions of each weather variable can be found in Table 3. b Q1: 1 st quartile, or 25% of observations are below and 75% of observations are above this value; Q3: 3 rd quartile, or 75% of observations are below and 25% of observations are above this value. c SD: standard deviation in the weather variables across trials.      Table 3. c Coefficients were estimated using multivariable mixed effects linear regression via the lmer() function in R. Cohort was included as a random effect in all models. d 95% CI indicates a 95% confidence interval.  Table 3. c Coefficients were estimated using multivariable mixed effects linear regression via the lmer() function in R. Cohort was included as a random effect in all models. d 95% CI indicates a 95% confidence interval.        Table 3 (N=140). Each point represents the segment 1 die-off rate and relative humidity range values for an experimental plot. Pink points and lines represent Salmonella and blue points and lines represent E. coli. The regression model was fit using the lmer function in R; relative humidity range, produce type, and bacteria were included in the model as fixed effects, trial was included in the model as a random effect, and segment 1 die-off rate was the outcome.  Table 3 (N=140). Each point represents the segment 1 die-off rate standard error and max. temperature values for an experimental plot. Pink points and lines represent Salmonella and blue points and lines represent E. coli. The regression model was fit using the lmer function in R; maximum temperature, produce type and bacteria were included in the model as fixed effects, trial was included in the model as a random effect, and segment 1 die-off rate standard error was the outcome. Figure S4. Final mixed effects multivariable regression model for the relationship between relative humidity range (%) and the breakpoint (days) between segment 1 and segment 2 for the experimental plots; variables are defined in Table 3 (N=140). Each point represents the breakpoint and relative humidity range values for an experimental plot. Pink points and lines represent Salmonella and blue points and lines represent E. coli. The regression was model fit using the lmer function in R; relative humidity range and bacteria were included in the model as fixed effects, trial was included in the model as a random effect, and breakpoint was the outcome.
Appendix A. E. coli and Salmonella die-off in in soil and E. coli strain characterization following a simulated overhead irrigation event in Davis, California.
In two California trials (CA1 and CA2), die-off of Salmonella and E. coli were assessed in the soil of the baby spinach and lettuce plots following the simulated overhead irrigation event with PBS inoculated with rifampicin resistant E. coli and rifampicin resistant, attenuated Salmonella. The inoculum concentration of E. coli was 5.8 and 5.7 log10 CFU/mL in trials CA1 and CA2, respectively. The inoculum concentration of Salmonella was 5.8 log10 CFU/100mL in both trials. The soil samples were collected at 24, 48, 72, and 96h following inoculation. One sample was collected per plot per time point (i.e., 1 from each of 4 spinach plots and 1 from each of four lettuce plots per timepoint for a total of 64 samples per trial). To do so, the protocol described by Lopez-Velasco et al., (2012) was used. Briefly, soil was collected from the surface to a depth of 5-7 cm in 10 random locations per plot and composited. An amount of 100 g of soil was taken from the homogenized bulk-composite (approximately 500 g) and added to 200 mL of 0.01 M sodium phosphate supplemented with 0.05% Tween 20 (Fisher, Fair Lawn, NJ). The suspension was gently shaken and then allowed to settle for 20 min. Aliquots of 100 μL of the supernatant were spread onto an ECC+R plate and incubated at 37°C for 24h. Additionally, 25 mL of the soil-extraction supernatant was transferred into 75 mL of TSB+R and incubated for 18-24 h at 37°C for enrichment. Following incubation, 100 µL was streaked onto ECC+R and incubated at 37°C for 24h. The blue colonies and white colonies on ECC+R plates were recorded as E. coli and Salmonella, respectively. All data analysis was performed in R version 3.5.3 (R Core Team, Vienna, Austria). The Salmonella and E. coli counts per plate were converted to log10 CFU/ 100g of soil. Thirty-seven samples were below the limit of quantification; all were positive by enrichment. For these samples, 10 rounds of imputations were performed from a uniform distribution. A linear model was fit to the imputed dataset using the pool function from the mice package (van Buuren & Groothuis-Oudshoorn, 2011); the outcome was log10 CFU Salmonella or E. coli/ 100g soil. Backwards selection was performed to determine which on the following predictors were retained in the model: bacteria (Salmonella and E. coli), trial (CA1 and CA2), produce type (spinach and lettuce), and time (in hours). The final model can be found in Table SA1. Time was excluded from the final model; all other predictors were retained. In addition, the differential survival of the three E. coli inoculum strains (TVS 353, TVS 354, and TVS355) was assessed. Up to 16 isolates per sample were characterized using the PCR protocol developed to distinguish between the 3 inoculum strains (see main text for a description of the protocol). In total, 103 isolates were characterized in trial CA1 and 161 isolates were characterized in trial CA2. To determine the effect of trial, time, and produce type on the survival of the 3 E. coli inoculum strains, multinomial regression was performed using the multinom function in the "nnet" package (Venables & Ripley, 2002). The predictors tested for inclusion in the model were trial (CA1 and CA2), produce type (spinach and lettuce), and time. The results of the final model can be found in Table SA2.

Safety:
a. While all strains used in the current study are non-pathogenic (generic E. coli) or attenuated (Salmonella), proper safety measures must still be taken to limit human exposure risks. During inoculation goggles, gloves, a face covering, and a Tyvek suit should be worn to prevent any unnecessary exposure to the inoculum. b. Naturally occurring rifampicin resistant E. coli or Salmonella that may be isolated over the course of this study has the potential to be pathogenic. Use proper aseptic technique and take proper safety measures to limit human exposure to all isolates. a. Field Set-up (see Figure SB1 for a diagram of sample field) i. Three experimental replicate fields (i.e. cohorts) will be conducted consecutively in each location (Davis, California; Freeville, New York; and Murcia, Spain); additionally, 3 or fewer experimental replicates may be conducted in Salinas, CA. ii. Each cohort will require eight or more 1.5x4 m plots. Lettuce will be sown in at least four plots and spinach will be sown in at least four plots, providing 4 "technical replicates" per produce commodity per experimental field (cohort). Additional plots may be planted to protect against poor stand germination. iii. There should be a ≥1 m buffer between all plots in the same cohort. iv. All fields should be ≥12 meters apart.

b. Plot preparation and seeding
i. Apply fertilizer and herbicide, as needed, to each field. The exact composition and type of fertilizer and herbicide is to be based on industry practice and field needs in the respective study location and cohort. Record type and amount applied. ii. Create a raised bed the length of the plot. Record height of beds. iii. Prior to seeding apply sufficient water to wet each bed using overhead irrigation. iv. Sow 6 rows of seed per bed using a seeding rate of 1.5 inches (~105 plants per row; ~630 plants per plot).

c. Irrigation
i. Overhead irrigation should be performed as needed, including post-inoculation.
ii. Measure the following water quality characteristics for the water being used for overhead irrigation. Obtain these measurements ≥ 3 times during the study period, ideally immediately preceding or following irrigation. 1. pH 2. Turbidity 3. Soluble Iron iii. Record when the fields are irrigated, and the approximate volume of water applied.

Monitoring Environmental Conditions.
a. The following weather data should be collected from the closest weather station to the field; all measurements should be collected for the shortest possible time interval. ii. Within 24 h. of removing the plates from the incubator, flood 1 plate per strain with 3 mL PBS. iii. Resuspend cell using a spreader and transfer resuspended cells using a pipette into a beaker holding 97 mL PBS (1 lawn plate resuspended in 100 mL). iv. Homogenize each suspension using a 1 mL pipette. c. Washing step and inoculum preparation i. NOTE: See Figure SB3. for a diagram of the washing step.
ii. Transfer 10 mL of each bacterial suspension into separate 15 mL Falcon tubes (i.e., there should be 5 falcon tubes total, one per strain). Centrifuge at 2,500 x g for 5 minutes and pipette off the culture supernatant. iii. Wash the pellet twice with 10 mL PBS and re-suspend in 5 mL of PBS using the same centrifugation conditions as above. iv. Measure the optical density (OD600) of each bacterial suspension. v. Compare the OD600 measurement to the standard curve provided. Adjust the concentration of the bacterial suspension of each strain to ~log 9 CFU/mL using the standard curve. vi. Transfer each strain's bacterial suspension to the 4°C cold room.

d. Dilution of E. coli
i. NOTE: See Figure SB4. for a diagram of the preparation of the E. coli cocktail. ii. Retrieve each bacterial suspension from the 4°C cold room. iii. Combine 4 mL of each of the E. coli suspensions into a 15 mL Falcon tube (i.e., 12 mL total, 4 mL per strain). Vortex. iv. Perform serial dilutions of the E. coli cocktail using sterile PBS to achieve a concentration of log 6 CFU/mL using the scheme from the f. After converting the colony count to CFU per mL, the concentration of each organism should be between 10 4 and 10 5 CFU/ mL. 3.6. Inoculation NOTE: Practice using the sprayer with water prior to performing inoculation to ensure a consistent volume inoculum is applied over all plots.

a. Timing of inoculation
i. Inoculation should be performed once the lettuce plants for a given cohort have 6 true leave (approx. 30 to 40 days after planting). The spinach plants will have 10 to 12 true leaves. ii. Try to target inoculation for a day without rain and with minimal wind. b. Inoculation i. NOTE: See Figure SB5 for a picture of an example inoculation.
ii. Set the sprayer at approx. 27-30 psi and apply at pre-calibrated walking speed to deliver spray at 2L/100 feet (30.5 m). iii. During inoculation, hold the sprayer so the nozzle is approx. 1 m above the lettuce/ spinach. 1. The primer sequences to detect each of the E. coli inoculum strains are as follows: Primer Name Target Strain  Sequence  353F  TVS353  TGACGGACAGGGACTCTATCTG  353R  TVS353  CAGCGTTCGCTCACTGAGAG  354F  TVS354  TAGGTTTGTTCACATTAGGTGATGTCG  354R  TVS354  AAATGTGGGTATGGCATATGGCAG  355F  TVS355  GTGACACCAATGACATCTGATGTTATCC  355R  TVS355  CGTCCTTATCCTGTTGGCTTGTG  35XF  All 3  4. Gel electrophoresis should be performed using a 1.8% agarose gel at 80 volts.

Sample
The "BenchTop pGEM DNA Marker" should be used as the ladder. 5. The gel will take approx. 2 hours to run. 6. Ethidium bromide should be used for staining. i. Randomly select a single spinach plant.

Detection of Viable But Not Culturable
ii. Wipe the scissors with a 20% bleach wipe followed by two 70% ethanol wipes.
iii. Cut the selected plant 2 cm above the soil line and place the plant in a pre-labeled, Whirl-Pak bag.
3. The reaction components should be combined according to the set-up below. The reaction components should be combined according to the set-up below.
One no-template control should be included per run (i.e., replace DNA template with sterile dH2O).

Difficulty re-suspending cells.
a. Colonies on older plates (> 4 days) become dried out and stick to agar. b. ONLY use plates between 1 and 2 days old to facilitate easier re-suspension.

Plate is out of countable region.
a. For 0-8h samples, if plate is out of countable range and less than 48h has passed since sample collection, adjust the dilution accordingly and re-plate.

Uneven inoculation across and within plots.
a. Practice using the sprayer with uninoculated water prior to performing the experiment.

Splashing of inoculum when the 3 L bottles containing the inoculum are removed
from the sprayer. a. De-pressurize the 2 L bottles that contain the inoculum prior to removing them from the sprayer by following the steps outlined in Figure SB6