Spatial risk zonation of phosphorus loss in agricultural catchments shaped by phosphorus forms and transport pathways

Identifying spatial risk zones of phosphorus (P) loss remains difficult in agricultural catchments because P forms differ in source availability, leaching potential, and transport pathways. Existing risk assessment frameworks mainly emphasize surface erosion and runoff, making it difficult to capture form-specific risks associated with subsurface leaching. Here, we developed a spatial risk assessment framework for total phosphorus (TP), particulate phosphorus (PP), colloidal phosphorus (CP), and dissolved phosphorus (DP) by integrating soil-column leaching experiments, random forest modelling, field observations of hydrological pathway partitioning, and Monte Carlo-Sobol sensitivity analysis of weighting parameters. The RF models showed moderate predictive performance for leaching loads (R 2 = 0.612–0.661), with pH, redox potential (Eh), electrical conductivity, total dissolved solids, and water-dispersible colloid mass showing relatively high model-based importance. Field observations of TP export showed that surface flow was dominant, while interflow and baseflow made measurable contributions, indicating that non-surface P transfer occurred at the catchment scale. Integrated risk analysis revealed contrasting source-transport sensitivities among P forms: TP and PP risk outputs were more sensitive to transport-related weighting parameters, whereas CP and DP outputs were more sensitive to source-related parameters. For TP, the relative contribution of the leaching-related indicator decreased from 33.1% to 17.5% across risk levels, whereas that of soil P status increased from 23.2% to 31.5%. This framework clarifies how P forms and transport pathways jointly shape spatial risk zonation and supports form-specific, zone-targeted management of agricultural P loss.