Toward agent-based models for pre-harvest food safety

The safety of fresh produce is a concern because many types of produce are consumed raw, and the incidence of produce-borne diseases appears to be increasing. Good agricultural practices (GAPs) can limit the chances for contamination of produce, but farms are open systems where interactions with surrounding lands can override benefits from GAPs. The extent to which interactions between farms and the surrounding landscape contribute to sporadic disease is not yet clear, but outbreaks have arisen from contamination of produce by wildlife, manure, or irrigation water. We hypothesize that the ability to control transmission of pathogens to produce, before harvest, partially depends on understanding the dynamics of pathogen populations, with some landscapes exhibiting combinations of characteristics that increase pathogen dispersal to croplands. The landscape processes leading to contamination are difficult to predict with current techniques. In this perspectives paper, we advance a foundation for modeling food contamination processes in agricultural landscapes using a combination of remote-sensing technologies and ecological modeling of pathogen dispersal, enabled by geographic information systems. With the development of predictive models, cultivation practices can be adapted to local conditions to reduce food contamination by controlling the connectivity between foodborne pathogen reservoirs and food production blocks on farms.