Habitat suitability modeling for medicinal and weedy plant species using machine learning: A spatial AI approach for sustainable agroecosystems

The integration of artificial intelligence (AI), particularly machine learning algorithms (MLAs), into agriculture presents transformative opportunities for sustainable land and crop management. Among these, habitat suitability modeling plays a crucial role in understanding the ecological preferences and potential spread of medicinal plants, as well as the management of weeds, which are of significant environmental and agronomic interest. This study employs a spatially explicit modeling framework that incorporates topographic and bioclimatic variables and a suite of ۳۰ individual and ensemble MLAs to predict habitat suitability across diverse agroecosystems. By focusing on a broad set of species rather than a single taxon, we emphasize methodological generalizability and scalable application. Ensemble algorithms, particularly Random Forest and Gradient Boosting Machines, consistently outperformed individual models, achieving high accuracy (AUC > ۰.۹۰) in most scenarios. The outcomes support informed decision-making for the management of weedy species (for control or containment) and conservation or cultivation of medicinal plants, contributing to sustainable agriculture and biodiversity preservation. This AI-driven approach offers a powerful decision-support tool for land managers, researchers, and policymakers aiming to optimize agroecological planning under current and future environmental conditions.

کلیدواژه ها:

Artificial intelligence ، Machine learning ، Medicinal plants ، Weeds ، Habitat modeling