Rangeland ecosystems of semi-arid and arid areas can stressed out by their specific physical, environmental and managemental conditions. Better understanding of the relations between a variety of factors can help us improve our management plans, for which we have to study the relations between different plant species and their distribution patterns. Classification and Regression Tree (CART) was used to model potential habitats for four halophyte pastoral plants at Hoz Soltan area in Qom province. Occurance data of species have been acquired from field trips and using GPS points in the area which included 36 present and absent points. 12 environmental layers used for modeling process included (12 soil factors). ArcGIS geostatistics (Kriging) method was used to zone all maps of the study area. After establishing the relations in R software, ecological space model was transmuted to geographical spaces as maps and also a Tree diagram. Performing complementary RDA analysis different reactions of the studied species to the soil factors and two-plot diagrams of species-environment have been generated. Clustering analysis was used to generate dendrogram of the studied species. Results of CART model suggested that the distribution of Halocnemum strobilaceum is related to calcium, calcium sulphates, Sand and pH percentages, Seidlitzia rosmarinus species to calcium, calcium carbonates and Sand percentages, Tamarix ramossisima species to Sand percentage and Artemisia sieberi species to calcium, calium carbonate and Sand; this variables affect the distribution patterns of the studied species strongly. Results of model evaluation showed that CART model’s performance for AUC, Kappa’s coefficient and CCR accuracy is as follows respectively for each species: Halocnemum strobilaceum 0.91 – 0.68 – 0.86, Seidlitzia rosmarinus 0.91 – 0.69 – 0.86, Tamarix ramossisima 0.93- 0.62 – 0.96 and Artemisia sieberi 0.91 – 0.65 – 0.86. Performance of AUC coefficient according to Swets Categories (1988) was excellent and according to landis and Koch categories (1997) Kappa coefficient’s performance was good. The results of CART modeling and complementary RDA analysis have also been coherent. Key words: species distribution pattern models, potential habitat of four pastoral species, CART model, Geographical Information System, R software, Hoz Soltan area, Qom province