Purpose: A study was performed to test a set of hierarchical rules associated with a previously developed model to predict the prevalence of Listeria monocytogenes in New York State produce farms.
Methods: Predictive risk maps were developed for four enrolled produce farms. The expected prevalence of L. monocytogenes was determined for each field based on the hierarchical rules: proximity to water, roads, and pastures, and available soil moisture (AWS). Drag swab samples were collected from a subset of plots assigned to each risk category. Chi-square tests were used to evaluate whether each rule accurately predicted L. monocytogenes prevalence. Multivariable analyses were performed to test the effect of factors on the likelihood of L. monocytogenes isolation from each risk category.
Results: The overall prevalence of L. monocytogenes was 12% (128/1,056). Specifically, L. monocytogenes was detected in 24% (43/176), 8% (21/264), 17% (23/132), 5% (7/132), and 10% (34/352) of samples collected from risk categories near water, near roads, low AWS, high AWS, and near pastures, respectively. The expected prevalence of L. monocytogenes (previous model) was not significantly different from the observed prevalence of L. monocytogenes for risk categories: near water (P = 0.72), high AWS (P = 0.28), and near pasture (P = 0.54). Additionally, multivariable analyses showed that proximity to water and pasture were significantly associated with isolation of L. monocytogenes (OR = 3.8 and 2.4, respectively) from drag swab samples.
Significance: These findings suggest the risk of L. monocytogenes contamination on-farms is not uniform; instead risk of contamination is driven by environmental factors specific to each on-farm location. Thus, GIS offers growers’ science-based-data on pathogen risk to tailor GAPs.