Habitat distribution modeling of some plant species using logistic regression in the semi-arid rangelands. (Case study: Eshtehard rangelands)

The aim of this study is habitat species distribution modeling using logistic regression in Eshtehard rangelands. Homogeneous units was determined based on the slope and elevation maps and satellite images. Per unit, three transects with a length of 750 meters was established according to systematic-random method. At the beginning and end of each transect soil profile was dug, Sampling was done from the depths 0-20 and 20-80. Soil variables including gravel, clay,sand, silt, lime, organic matter, pH and electrical conductivity was measured. Also for recording the presence or absence of plant species, 15 plots was established along each transect. Size of plot sampling was determined rectangular area 2 square meters according to the type and distribution of plant species with minimum area. In the next step regression relationship between environmental factors and presence of habitat using SPSS software was obtained.The map of variables was produced by Kriging interpolation method in GIS software. The regression equation related to the presence of any type were applied on maps of effective environmental factors, and prediction map produced for each habitat. The correspondence between the predicted and actual maps was assessed using the kappa coefficient. Correspondence between the predicted and actual maps was assessed for habitat Halocnemum strobilaceum in good level (k=0/66), for habitat of Artemisia sieberi in intermediate level (k=0/46), for Artemisia sieber- Stipa barbataþ in intermediate level too (K=0/54) and for habitat of Artemisia sieberi- Salsola richteriþ in poor level (k=0/29).