Evaluating logistic regression model capability to determine spatial distribution map of plant species (Case study: Taleghan Miany rangelands)

The purpose of this study was to assess the logistic regression model capability to map the spatial distribution of plant species in Taleghan Miany rangelands. Systematic-random sampling method was used to collect vegetation and environmental data (such as topography and soil factors). Three transects with 150 m lengths were established in each sampling unit and the vegetation cover percentage were determined in 15 plots along each transect. Soil samples were collected from the two sampling plots in each transect randomly. Geo-statistical methods were used to produce the environmental maps. Topographic maps were also produced using digital elevation model. Species distribution models were prepared using logistic regressions equations in Geographic Information System (GIS). Kappa index (k) was used to evaluate the agreement between prediction models and field data. According to the results, kappa index obtained for Agropyron intermedium habitat model was 0.83. The calculated Kappa for Stipa barbata and Astragalus gossypinus habitat models were 0.79 and 0.80 respectively. This index for Thymus kotschyanus habitat was 0.68. This study indicated that logistic regression model could predict the habitat suitability of the four studied species appropriately. The produced logistic regression model for predicting Agropyron intermedium habitat was more efficient because this species has more specific habitat requirements. The results of this study can be used in rehabilitation, conservation and management programs in rangeland ecosystems.