Holistic Estuarine Monitoring: Data-Driven and Process-Based Coupling of Biogeochemical Cycles of Per- and Polyfluoroalkyl Substances

Better understanding the fate and transport of estuarine per- and polyfluoroalkyl substances (PFASs) requires coupling multiple matrix-specific biogeochemical roles, rather than relying on a single-matrix approach. We therefore evaluated sediment and biological matrices (blue mussels (BMs), Mytilus edulis; and hardshell clams (HSCs), Mercenaria mercenaria) as complementary indicators of PFAS contamination across three locations over a 240-day period following a spill event. A three-tiered analytical approach was applied: Tier 1 used non-parametric statistics to assess the broad-spectrum detection patterns for a total of 40 PFASs (n = 47 samples); Tier 2 employed generalized regression (adaptive Elastic Net), random forest, and artificial neural networks to model the concentrations of the most frequently detected PFASs (PFOS, PFOA, PFHxA, and PFOSA) (n = 188 observations); and Tier 3 implemented a system dynamics model to mechanistically couple the PFOS and 5:3 FTCA fate. The results suggest that the sediment acted as a long-term sink for legacy long-chain compounds (99.3%, primarily PFOS), while the biota, particularly BMs, acted as sensitive recorders of acute pulses and hydrophilic precursors, uniquely accumulating 5:3 FTCA during spring pulses (p < 0.001). All the models identified the matrix type as the dominant driver of the most prevalent PFAS concentrations. A reliance on sediment monitoring alone may fail to capture the majority of the active contamination burden sequestered in the biota, suggesting that effective risk assessment necessitates an integrated view.