Source Apportionment and Driving Factors of Heavy Metal Pollution in Paddy Soils Around a Mining Area Using PMF and XGBoost-SHAP Models
Identifying the sources and driving factors of heavy metals (HMs) in paddy soils around mining areas is crucial for safeguarding regional agricultural production and food security. This study focused on paddy soils near a mining area in southern Jiangxi Province, integrating Spearman correlation analysis, positive matrix factorization (PMF), and extreme gradient boosting (XGBoost) coupled with SHapley Additive exPlanations (SHAP) to quantitatively apportion potential ecological risks, pollution sources, and driving factors. Results showed that mean concentrations of Pb, Ni, and Cu exceeded the soil background values of Jiangxi Province, except for Cr. The mean comprehensive potential ecological risk index (RI) was 17.4, with 25.7% of sampling sites exhibiting moderate risk, and Ni being the primary ecological risk factor. Source apportionment identified three main sources: natural background (51.1%), industrial and mining activities (26.6%), and a mixed source of agricultural activities and traffic emissions (22.3%). Spatial variation of Cr, Cu, and Ni was predominantly governed by soil physicochemical properties (Fe and P contents), whereas Pb distribution was significantly influenced by industrial and mining activities. The combination of PMF and XGBoost-SHAP effectively delineated sources and driving factors, providing a scientific basis for targeted soil management and pollution control in mining-affected agricultural regions.