Outbreaks of human illness caused by enteric pathogens such as Salmonella are increasingly linked to the consumption of fruits and vegetables. Knowledge on the factors affecting Salmonella proliferation on fresh produce therefore becomes increasingly important to safeguard public health. Previous experiments showed a limited impact of pre-harvest production practices on Salmonella proliferation on tomatoes, but suggested a significant effect of harvest time. We explored the data from two previously published and one unpublished experiment using regression trees, which allowed overcoming the interpretational difficulties of classical statistical models with higher order interactions. We assessed the effect of harvest time by explicitly modeling the climatic conditions at harvest time and by performing confirmatory laboratory experiments. Across all datasets, regression trees confirmed the dominant effect of harvest time on Salmonella proliferation, with humidity-related factors emerging as the most important underlying climatic factors. High relative humidity the week prior to harvest was consistently associated with lower Salmonella proliferation. A controlled lab experiment confirmed that tomatoes containing their native epimicrobiota supported significantly lower Salmonella proliferation when incubated at higher humidity prior to inoculation. The complex interactions between environmental conditions and the native microbiota of the tomato crop remain to be fully understood.
