Modeling of Artemisia sieberi Besser Habitat Distribution Using Maximum Entropy Method in Desert Rangelands

Abstract

Predictive modeling of habitat distribution of range plant species and identification of their potential habitats play important roles in the restoration of disturbed rangelands. This study aimed to predict the geographical distribution of <em>Artemisia sieberi</em> and find the influential variables in the distribution of <em>A. sieberi</em> in the desert rangelands of central Iran. Maps of environmental variables were generated by GIS software (version 9.3)<em>.</em> Predictive distribution maps of <em>A. sieberi </em>were produced with Maximum Entropy Method (MaxEnt) and the existing data regarding this species. The agreement of predictive map with the actual map was checked by calculating Kappa coefficient value. Accuracy of predictive models was evaluated using the Area Under the Curve (AUC). Results showed that soil pH and lime content in the surface layer (0-30 cm) and silt percent in both surface and sub-surface soil depths (0-30 and 30-60) had the greatest impacts on the distribution of <em>A. sieberi</em> in the study area. Correspondence of actual map with the predictive one was assessed at a satisfactory level (Kappa coefficient = 0.70). MaxEnt is widely used as compared with the other standard methods since it only requires the presence data of a specific plant species to draw the distribution map of its habitat. Additionally, MaxEnt is a generative method and its output can be easily understood by the field practitioners.