A Bayesian Network Approach to Probabilistic Risk Assessment of Pesticides in Rice Fields

Summary

Today’s aquatic environment is constantly exposed to various pollutants resulting from anthropogenic activities, such as the use of plant protection products in agricultural practices. The traditional environmental risk assessment is usually based on a calculated risk characterisation such as a quotient representing a ratio of exposure to effects. This deterministic approach uses single values of predicted environmental concentration and predicted no effect concentration to calculate a risk quotient, and accounts for uncertainty by using an assessment factor. Contrary to this deterministic approach, we try to account for variability and uncertainty by integrating probability distributions for exposure and effects that are propagated throughout a Bayesian network model. We focus on the risk assessment of various pesticides in a representative study area and their potential risk to the aquatic ecosystem. This study is carried out for a rice field in Spain (Albufera region) and offers a transparent way of estimating the risk inside a rice paddy and drainage water. While facilitating the communication of estimates and uncertainties, the developed model can also predict the probability of several levels of the risk quotient.