Geospatial modelling of Prosopis cineraria (L.) Druce dominance using interpolation and machine learning techniques...

Prosopis cineraria (L.) Druce, a keystone species in hot arid and semi-arid ecosystems of India, contributes significantly to ecological stabilization, carbon sequestration, and traditional agroforestry systems. This study employs an integrated geospatial and statistical framework to assess the spatial dominance and habitat suitability of P. cineraria across diverse ecological gradients. Using field-based relative importance value (RIV) data from 322 sites, inverse distance weighting (IDW) was applied to interpolate species dominance patterns. Ensemble species distribution modelling (ESDM) was implemented with seven machine learning algorithms (e.g., random forest, MaxEnt, artificial neural network) and environmental predictors-bioclimatic, edaphic, topographic, and anthropogenic-to delineate habitat suitability. Results indicated that bioclimatic variables, particularly precipitation seasonality (Bio15) and temperature extremes (Bio5, Bio6), were the most influential, with random forest achieving the highest predictive accuracy (AUC = 0.98; TSS = 0.89). IDW interpolation identified strong P. cineraria dominance in western-central districts (Jodhpur, Pali, Nagaur), while ESDM projected ~ 428,407 km² of suitable habitat, largely overlapping with field-derived hotspots. Niche hypervolume analysis revealed a broad core niche but a more restricted realized distribution, constrained by human pressures and environmental factors. These findings provide evidence that could guide zone-specific conservation strategies, including the prioritization of high-RIV areas for in situ protection and ecological restoration in low-dominance regions. While the results highlight the ecological importance of P. cineraria, further field validation and socio-economic assessments are recommended before direct policy application.