A Multi-Factor Analysis Linking Environmental Stressors with Presence of Per-and Polyfluoroalkyl Substances (PFAS) in a Coastal Lagoon
By Sanneri E. Santiago Borrés, Sunil Kumar, Jean C. Bonzongo, Katherine Y. Deliz Quiñones, and Antarpreet Jutla
ACS ES&T Water
July 12, 2024
DOI: 10.1021/acsestwater.4c00286
Per- and polyfluoroalkyl substances (PFAS) pose significant environmental and health risks due to their unique physicochemical properties, which influence their environmental fate and transport. This study investigates presence of PFAS in biweekly surface water samples from 17 sites in the Indian River Lagoon (IRL) in Brevard County, FL, using Classification and Regression Trees (CART), a machine-learning method, and incorporates seven environmental stressors (salinity, dissolved oxygen, pH, water temperature, precipitation, wind speed, and river discharge). CART analysis revealed salinity as a crucial factor associated with the occurrence of PFAS throughout the IRL, which can be attributed to PFAS sorption behaviors at the water–solid interface and the water residence time at the lagoon. In addition, it highlights better predictive accuracy for long-chain carboxylic PFAS compounds. The performance of the CART analysis yielded average sensitivity and specificity values of ∼87 and ∼57%, respectively. These findings offer valuable insights for future studies leveraging machine-learning approaches to swiftly and systematically identify areas with increased PFAS occurrence in water systems. This approach represents a crucial first step in the implementation of machine-learning to assess PFAS exposure risks in aquatic environments, suggesting the need for targeted monitoring and interventions based on identified environmental stressors.
Topics: