Ecological and Statistical Evaluation of Genetic Algorithm (GARP), Maximum Entropy Method, and Logistic Regression in Predicting Spatial Distribution of <i>Astragalus</i> sp.

This study aims to evaluate the potential habitat of <i>Astragalus</i> sp. using three different species distribution modeling methods: the maximum entropy (MaxEnt) model, the Genetic Algorithm for Rule-Set Production (GARP), and logistic regression. The primary objective was to identify key environmental factors that influence the spatial distribution of <i>Astragalus</i> sp. in the Savar-Abad basin's rangelands. Vegetation sampling was carried out across diverse vegetation types within the study area, using 2-10 square meter plots to capture a representative sample of plant species distribution. Soil sampling was conducted at varying depths to capture essential soil properties, including physical (clay, gravel, silt, and sand) and chemical factors (organic matter, electrical conductivity, pH, and lime). Soil maps were generated using interpolation techniques to visualize soil variation across the area. The sampling strategy was designed to ensure comprehensive data collection, allowing for robust model training and validation. MaxEnt, which is a presence-only model, outperformed both the GARP and logistic regression models in predicting suitable habitats for <i>Astragalus</i> sp. Results revealed that soil salinity, elevation, and soil acidity significantly influenced species distribution. The findings also suggest that elevation and salinity have the most substantial effects on habitat suitability, while soil texture (clay, silt, and sand) plays a secondary role. These results are valuable for rangeland management, offering insights into areas where <i>Astragalus</i> sp. could thrive or where interventions might be necessary to improve habitat conditions. In terms of management, this study highlights the importance of considering both ecological and environmental factors when planning conservation and restoration activities for rangelands. The ability to predict species distribution can help optimize resource allocation for habitat restoration and enhance biodiversity conservation efforts.