Developing a probabilistic approach to a PFAS food web-based model for ecological risk assessment

In site-specific investigations, food web risk assessment models are a complementary approach to conventional sampling and analysis of biota tissue, which in some cases can be impractical or impossible. One such model has recently been developed, allowing risk assessors to predict concentrations of per- and polyfluoroalkyl substances (PFAS) across various trophic levels in aquatic habitats. However, a key limitation lies in the use of point-valued bioaccumulation metrics , which hinders the understanding and communication of the inherent uncertainty associated with model predictions for PFAS concentration in biota. The objective of this study is to describe how this model was advanced into a probabilistic framework using Monte Carlo simulations. This enhancement involved quantifying variability for the bioaccumulation factors (BAFs) associated with each of the 18 PFAS compounds within the model, and these were integrated into the model as probability distributions . The upgraded model generates distributions for PFAS concentrations in biota and facilitates sensitivity analyses to pinpoint the critical inputs exerting the most influence on model predictions . In addition to allowing risk assessors to communicate uncertainty, the enhanced model can serve as a guidepost for directing research efforts toward obtaining high-quality data for pivotal inputs and optimizing site-specific investigations considering biota sampling.