Assessing presence of per- and polyfluoroalkyl substances (PFAS) in the Indian River Lagoon: A Bayesian approach to understanding the impact of environmental stressors

Abstract

Per- and polyfluoroalkyl substances (PFAS) are persistent environmental pollutants, and their presence in aquatic environments, especially coastal waters, poses significant ecological and human health risks. This study investigates the occurrence and behavior of four PFAS compounds in the Indian River Lagoon, a biodiverse estuarine ecosystem located in Florida USA, by evaluating how ecological and hydroclimatic factors influence PFAS occurrence. A Bayesian Logistic Regression Model (BLRM) was employed to quantify the relationships between environmental stressors such as salinity, precipitation, river discharge, water temperature, and pH, and the presence of these PFAS compounds. The BLRM approach not only estimated the log odds of PFAS presence but also provided posterior estimates and odd ratios, making it a transparent and interpretable model compared to other machine learning techniques. The results indicate that salinity is a significant negative predictor for all PFAS compounds, showing a decrease in PFAS presence with increasing salinity. Precipitation exhibited a statistically significant positive association with PFBS, PFOA, and PFHxS, whereas river discharge negatively affected PFNA and PFOA. Model diagnostics confirmed BLRM's robustness, with posterior predictive checks showing strong alignment between observed PFAS presence and the model's predictions, validating its accuracy. The study highlights BLRM's advantages in environmental modeling, identifying key stressors and the direction of their effects on PFAS occurrence. It emphasizes the importance of ecological and hydroclimatic factors, such as salinity, precipitation, and river discharge, in understanding PFAS behavior in coastal ecosystems. These insights aid future risk assessments and management strategies to mitigate PFAS contamination in aquatic environments.