Ecological and carcinogenic risk assessment of potentially toxic elements in rangelands and croplands around Lake Junin (Peru): Integrating remote sensing, machine learning, and land cover segmentation

• Autores: Pizarro Carcausto, Samuel Edwin, Requena Rojas, Edilson Jimmy, Barboza, Elgar, Peña Elme, Eunice Dorcas, Arias Arredondo, Alberto Gilmer, Ccopi Trucios, Dennis
• Formato: artículo
• Fecha de Publicación: 2025
• Institución: Instituto Nacional de Innovación Agraria
• Repositorio: INIA-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.inia.gob.pe:20.500.12955/2854
• Enlace del recurso: http://hdl.handle.net/20.500.12955/2854; https://doi.org/10.1016/j.scitotenv.2025.180327
• Nivel de acceso: acceso abierto
• Materia: Heavy metals; Ecological risk assessment; Human health risk; Remote sensing; Machine learning; Soil contamination; Andean wetlands; Metales pesados; Evaluación de riesgos ecológicos; Riesgo para la salud humana; Teledetección; Aprendizaje automático; Contaminación del suelo; Humedales andinos; https://purl.org/pe-repo/ocde/ford#4.01.04; Human health; Salud humana; Rangelands; Pastizales; Andean region; Región andina

The Junín Lake basin, a critical high-altitude ecosystem in the central Peruvian Andes, faces severe contamination from potentially toxic elements (PTEs) driven by mining activities, agriculture, and urbanization. This study evaluates the spatial distribution, ecological risk, and human health implications of 14 heavy metals, metalloids, and trace elements in surface soils surrounding the lake. Using 211 soil samples, we integrated remote sensing, land cover classification, and Random Forest machine learning models with spectral, edaphic, topographic, and proximity-based environmental covariates to predict contamination patterns and assess risk. Results reveal extreme contamination, with arsenic (As), lead (Pb), cadmium (Cd), and zinc (Zn) concentrations exceeding ecological thresholds by over 100-fold in agricultural zones. Ecological risk assessments using contamination degree (mCD), pollution load index (PLI), and risk index (RI) indicated that over 99 % of the study area exhibits very high to ultra-high contamination levels. Human health risk analysis identified unacceptable carcinogenic risks from As, Pb, and Cr across adult and pediatric populations, with arsenic presenting the greatest concern. The integration of geospatial tools and machine learning enabled precise identification of contamination hotspots and vulnerable land cover types, demonstrating the value of AI approaches for monitoring contaminated territories. These findings underscore the urgent need for coordinated environmental management, targeted remediation strategies, and community-based monitoring to protect public health and preserve Andean ecosystem integrity.