Random Forest-based spatial distribution from heavy metals and associated with ecological and health risk assessment in agricultural soils of the Mantaro Valley, Peru

Abstract

Heavy metal (HM) contamination of agricultural soils in Andean valleys poses a critical threat to ecological integrity, food safety, and human health. This study assessed the spatial distribution, contamination levels, ecological and health risk (non-carcinogenic and carcinogenic) associated with Zn, Pb, Cu, Cd, and As in agricultural soils of the Mantaro Valley in the central Andes of Peru. A total of 306 topsoil samples were analyzed. Random Forest (RF) was used to model the spatial distribution of HM concentrations, integrating spectral, climatic, topographic, edaphic, and distance-related covariates. The RF models showed consistent but element-specific predictive performance. The values of R² ranged from 0.76 to 0.89, with the highest values for Pb and As and the lowest for Cu. Spatial predictions should be interpreted with consideration of the irregular sampling design and inherent uncertainties in spatial extrapolation. Mean concentrations of Zn, Pb, Cu, As, and Cd reached 1347.87, 439.91, 201.71, 90.69, and 4.27 mg kg⁻¹, respectively, exceeding both regulatory thresholds and local geochemical background values. Contamination indices revealed marked enrichment, particularly for As and Cd. Ecological risk was primarily driven by Cd and As, with 70.10% and 61.41% of samples classified as very high ecological risk for these elements, respectively. The comprehensive potential ecological risk index indicated a critical scenario with 71.7% of samples exceeding the very high-risk threshold. Human health risk assessment identified As as the main contributor to both non-carcinogenic and carcinogenic risks, with children showing higher vulnerability than adults. These findings delineate priority agricultural areas where ecological risk and arsenic-related carcinogenic risk overlap, underscoring the urgent need for targeted soil monitoring, exposure reduction, and locally adapted risk management strategies in the Mantaro Valley.