Modeling the Ecological Preferences and Adaptive Capacities of Kentucky Bluegrass Based on Water Availability Using Various Machine Learning Algorithms

This study examined the habitat suitability of Kentucky bluegrass (Poa pratensis L.) in Iran’s Fars province, a region characterized by diverse climatic conditions and significant ecological challenges. Utilizing a multi-technique approach that included species distribution models (SDMs) based on machine learning algorithms, geographic information systems (GIS), and remote sensing, we analyzed environmental factors such as climate variables, soil properties, and water availability to understand their influence on habitat suitability. The results indicated that Kentucky bluegrass shows a strong preference for areas near water sources, and its distribution is significantly affected by soil salinity and texture. Among the models tested, Random Forest (RF) and Support Vector Machines (SVMs) demonstrated the highest predictive accuracy. Based on the RF model, the most suitable habitats were identified in the counties of Sepidan, Beyza, Bavanat, Pasargad, and Abadeh. At the same time, areas with lower suitability included Eqlid, Marvdasht, Zarghan, and Arsanjan. Although this study primarily focused on current distribution patterns, the findings provide important insights into the ecological preferences and adaptive capacities of Kentucky bluegrass. These insights are essential for the development of targeted conservation strategies in transitional climate zones. Future studies are recommended to explore the species’ response to future climate scenarios, enhancing its resilience against global climate change.