With a spate of recent outbreaks of such foodborne pathogens as Salmonella, Shiga toxin-producing E. coli, and Listeria monocytogenes, the ability to predict where and how these deadly microbes enter the food supply chain could save lives and prevent disease. Cornell University researchers have created a method that uses geospatial algorithms, foodborne pathogen ecology, and Geographic Information System (GIS) tools to predict hot spots where these pathogens may be present and spread on farms prior to harvest. Many of the recent outbreaks of foodborne pathogens have been linked to contamination on the farm.
The method, which can be applied to any farm, uses classification tree tools with remotely sensed data, such as topography, soil type, weather trends, and proximity to various sources (water, forests) to predict areas where pathogens are likely to be present.
“We wanted to see if we could identify factors that gave us a higher or lower prevalence of finding these pathogens,” said Laura Strawn, a graduate student in the field of food science and lead author of a study published in Applied and Environmental Microbiology. “We can look at a farm and use this data analysis tool to tell the farmer where these hotspots may be for foodborne pathogens.” By knowing where the hot spots are, farmers may then implement such preventive practices as draining standing water, adjusting where livestock graze, or planting crops that should be consumed cooked rather than raw, Strawn added.
The researchers collected 588 samples soil, water, feces, and drag swabs (gauze attached to a string and dragged over a field) from four produce fields on five farms each. Samples were collected four times a year, during each season, from 2009 to 2011. The prevalence of L. monocytogenes, Salmonella, and E. coli were 15%, 4.6%, and 2.7%, respectively, across all the samples. Listeria monocytogenes and Salmonella were detected more frequently in water samples from irrigation sources or nearby streams, while E. coli was found in equal distributions across all the sample types.
Listeria monocytogenes and Salmonella were found in higher frequencies in areas with moist soils; well-drained fields had lower Salmonella prevalence. Knowledge of such factors would help predict whether an area of a farm may be at higher risk. For Listeria, proximity to water, pastures, livestock, and grazing cattle, wildlife habitation, and nearby impervious surfaces, roads, and ditches all predicted a higher prevalence of the pathogen.
Once such factors have been identified, the GIS platform may be used to filter out specific areas based upon those factors (such as filtering areas that have moist soils and close proximity to water) to create a color-coded map of any farm area with predicted prevalence for a pathogen.