Unveiling the fate of heavy metals along the soil-rice-human pathway: Source-sink quantification, rhizospheric processes, and health implications.

Rapid industrial and agricultural intensification is increasing multi-source heavy metals (HMs) inputs into farmland, escalating risks to soil-crop system safety and human health. Limited source-sink flux data and inaccurate exposure estimates hinder precise assessment of HMs pollution in the soil-crop-human system. This study examines the impact of rapid industrial and agricultural intensification on HMs contamination in farmland, focusing on cadmium (Cd) risk to human health. Combining HMs flux monitoring, Cd bioaccessibility, socioeconomic and geographic data, and health risk modeling, we assessed dietary Cd exposure in a typical heavily populated, anthropogenically impacted watershed. Soils exhibited weak alkalinity (mean pH 7.75) and Cd contamination, with 15.5 % of soil samples exceeding China's risk thresholds. Cd isotope analysis and flux data identified atmospheric deposition and crop harvesting as dominant Cd sources and sinks, contributing 81 % and 64 %, respectively, with a net annual Cd soil accumulation of -0.44 μg/kg. Machine learning (XGBoost) identified the rice Cd bioconcentration factor (CdB), soil pH, organic matter (SOM), and cation exchange capacity (CEC), as key predictors of non-carcinogenic risk and carcinogenic risk. Population density, distance to the main river, the enhanced vegetation index, and atmospheric deposition are non-negligible contributors to the carcinogenic risk. Two-stage regression confirmed that indicated soil pH, CEC, and SOM regulate CdB, while iron, phosphorus, and potassium contents in root-surface iron plaque control Cd uptake via redox-sensitive pathways. This study advances regional-scale risk assessment by linking source apportionment, rhizosphere processes, and human exposure, providing a novel insight for targeted pollution control and sustainable agricultural management.