An operational-age-associated transition in groundwater PFAS fingerprints within a complex industrial park in China

Complex industrial parks host heterogeneous per- and polyfluoroalkyl substance (PFAS) emission sources from multiple industries. However, whether groundwater PFAS fingerprints are governed primarily by industry-specific emission patterns or by the regulatory context prevailing at factory establishment remains unresolved. Using a self-organizing map (SOM), permutational multivariate analysis of variance (PERMANOVA), and an extreme gradient boosting model with SHapley Additive exPlanations (XGB-SHAP), this study characterized groundwater PFAS fingerprints and their drivers across 20 factories spanning five industrial sectors in a complex industrial park in China. A total of 48 PFAS were detected, with concentrations from 10.9 to 926 ng/L (mean: 136 ng/L). Within the park, factory operational age was more strongly associated with PFAS fingerprints than industry type. PERMANOVA attributed 73.0% of fingerprint variance to operational age (p = 0.04), versus a non-significant 7.10% for industry type (p = 0.16). Operational age also ranked first in the XGB-SHAP model. The age-SHAP pattern changed sign near 10.5 years, while linear regression estimated a compositional crossover at 10.88 years. Together, these results supported a candidate transition interval of approximately 10-11 years within a historical context charactertized by regulatory, technological, and raw-material shifts. Five emission clusters were consolidated into three patterns with stratified risk profiles. A descriptive synthesis of Chinese industrial-park studies of groundwater (2009-2025) showed a similar transition from perfluorooctanoic acid (PFOA) to short-chain PFAS around 2014-2015, providing contextual support for the site-level pattern. These findings necessitate tiered regulation prioritizing legacy long-chain residues at older factories and proactive short-chain- or novel-PFAS surveillance at newer ones.