Modeling plant species distribution is based on the relationships of plant species presence with environmental variables. Using these models enables managers to predict the potential habitats from some environmental variables. Nowadays there is more attention on presenting the habitat potential mapping for various species by fast growing geo-statistical methods along with improvements in geographic information systems. This study aimed to identify environmental variables affecting on Ferula ovina (Boiss) distribution and presenting its potential distribution map using Generalized Additive Model (GAM) and Bayesian Belief Network () in a local scale. The presence and absence of the species in 278 sites (140 presence sites 138 absence sites) were collected using a stratified- random sampling approach in Feridoonshar rangeland located in Isfahan province. Some environmental maps were produced for the entire studied area using geo-statistical methods including krigging and inverse distance weight method. The relationships of presence and absence of the species with the environmental variables were modeled using GAM and methods. According to the GAM model, the presence of Ferula ovina was significantly correlated with some environmental variables including the average annual temperature, range of annual temperature, soil caco 3 content, soil saturation percentage, organic matter content and silt and clay contents. The sensitivity analysis of model revealed that the grazing pressure has more influence than environmental variables on the species distribution in the studied area. Also organic matter was identified as the most important determinant environmental variables with the presence of the species. According to the response curves of the species, the chance of the species occurrence were more in the habitats with average annual temperature 9 to 11 centigrade , range of annual temperature 62- 65 centigrade, 10 to 30 percent of soil caco 3 , 4 to 6 percent of organic matter, 45 to 60 percent of soil saturation, 10 to 30 percent of silt and 45 to 60 percent of clay in its soil texture. GAM model accuracy was then evaluated using independent data. The produced species distribution model had high accuracy in a local scale indicated by calculated Kappa coefficient and ROC area under curve plots being 0.52 and 0.75 respectively. According to GAM results, 43299 hectare or 43% of the study area was recognized as suitable area for the species occurrence. The model accuracy was also assessed by the obtained data from 31 independent sites. The prediction of habitat suitability of Ferula ovina using model was compared with the measured percentage cover of the species. The model enables to predict the habitat suitability of the species appropriately. According to the results, Ferula ovina has relatively broad ecological amplitude and the capability to regenerate naturally if its habitat properly managed. The studied modeling approaches can be used broadly to achieve appropriate management plans for allocating appropriate areas for rehabilitation and protection of other valuable native plant species in rangeland ecosystems. Keywords Species distribution model, Ferula ovina , General additive model, Bayesian belief network, Geographic information system, Feridonshahr, Isfahan province