Integrating agroecological suitability of cacao (Theobroma cacao L.) with biodiversity and land-use constraints in Peru

• Autores: Cotrina Sanchez, Alexander, Guzman Valque, Betty Karina, Barboza, Elgar, Oliva, Manuel, Huaman Pilco, Angel Fernando, Rojas Briceño, Nilton B.
• Formato: artículo
• Fecha de Publicación: 2026
• Institución: Instituto Nacional de Innovación Agraria
• Repositorio: INIA-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.inia.gob.pe:20.500.12955/3022
• Enlace del recurso: http://hdl.handle.net/20.500.12955/3022; https://doi.org/10.1016/j.agsy.2026.104637; https://www.sciencedirect.com/science/article/abs/pii/S0308521X26000053
• Nivel de acceso: acceso abierto
• Materia: Agroforestry systems; Cacao cultivation; Ensemble model; EUDR; Biomod2; Sistemas agroforestales; Cultivo de cacao; Modelo de conjunto; https://purl.org/pe-repo/ocde/ford#4.01.06; Theobroma cacao; Biodiversidad; Biodiversity; Almacenamiento; Storage; Deforestación; Deforestation; Utilización de la tierra; Land use; Restauración; Restoration; Modelización; Modelling

CONTEXT: Cacao cultivation is vital for rural economies in Peru, but its expansion often overlaps with sensitive ecosystems, raising concerns for biodiversity conservation. Despite international commitments to deforestation-free supply chains, integrated analyses combining agroecological suitability with land-use constraints remain scarce in Peru. OBJECTIVES: This study aims to identify suitable areas for cacao cultivation under multiple exclusion scenarios, evaluate conflicts with biodiversity and conservation areas, and quantify degraded lands that could provide opportunities for agroforestry-based restoration. METHODS: Cacao suitability was modelled with an ensemble of nine machine-learning algorithms using bioclimatic, edaphic, and topographic predictors. Outputs were filtered to exclude biophysical barriers and overlaid with national-scale layers of species richness, protected areas, forest cover, and degraded lands through GIS-based spatial analysis to evaluate exclusion scenarios and trade-offs. RESULTS AND CONCLUSIONS: The ensemble achieved high predictive power, with Random Forest (AUC = 0.997) and XGBoost (AUC = 0.972) performing best. Highly suitable areas were concentrated in the Andean-Amazon transition, especially in San Martín, Cusco, Huánuco, and Junín departments, where they overlapped with biodiversity hotspots and legally protected areas. Degraded yet suitable lands highlighted opportunities to expand cacao through agroforestry systems, reducing forest pressure and enhancing ecological restoration. SIGNIFICANCE: By integrating suitability modelling with national-scale geospatial layers, this study delivers a framework linking crop suitability with land-use constraints. The findings support national-scale planning while remaining adaptable to local contexts. They also align with international policy frameworks such as the European Deforestation Regulation (EUDR), promoting sustainable cacao production, biodiversity conservation, and long-term rural development in Peru.