Predictors of Salmonella enterica contamination in agricultural and livestock-impacted natural watersheds.

Salmonella enterica (S. enterica), a ubiquitous zoonotic foodborne pathogen, remains a worldwide public health hazard and economic burden. In recent years, outbreaks associated with the consumption of plant-based foods probably contaminated by irrigation water highlights the importance of water sources. This study investigated anthropogenic and environmental factors influencing S. enterica occurrence in natural watersheds impacted by agricultural and livestock industries in a 10-month longitudinal study in Paraiba, Brazil. Water samples were obtained from multiple sites within the three major river basins by modified Moore Swabs (MMS) and processed by conventional S. enterica isolation methodologies. Physicochemical parameters, climate, and human activities near the water sources were recorded. A logistic regression model was fitted using Generalized Linear Model (GLM) and further adjusted according to the selected variables using the Least Absolute Shrinkage and Selection Operator (LASSO) method. A non-statistical decision tree model was also fitted using the rpart package in R. Season, rainfall regime, water physicochemical features, and anthropogenic activities were significantly associated with S. enterica contamination. According to the regression tree analysis, rainfall within the sampling month was the strongest predictor of S. enterica recovery, potentially due to leaching from soil or runoff from adjacent human and animal activities. The complexity of multivariate conditions driving S. enterica contamination in surface waters highlights the need for region-specific investigations.