Predicting the distribution of plant species habitats using maximum entropy model (A case study in rangelands of western Taftan, Southeastern Iran)

摘要

This study was aimed to identify the most important effective variables affecting the distribution of plants habitat and mapping of those habitats using Maximum Entropy model. Vegetation sampling was carried out using systematic randomiz ed method. Soil samples were taken from 0 - 30 and 30 - 80 cm depths in each habitat by digging eight soil profiles. Soil samples were analyzed for physico - chemical parameters. For modeling by Maximum Entropy method, layers of physical variables were prepared with GIS and Geostatistics. Habitats distribution was modeled using MaxEnt software. Accuracy of classification by models and agreement of predicted and observed maps were evaluated through calculation of Kappa coefficient and curve area statistic. Results showed that altitude, degree of slope, available moisture and percentage of sand in superficial soil, gypsum in subsoil, gravel in topsoil, lime in both depths, and geological structure were most effective variables in habitat distribution. The agreement of predicted and observed maps was very good for the habitats of Amygdalus scoparia , Artemisia aucheri, Haloxylon persicum with Kappa coefficient of 0.82, 0.76 and 0.75, respectively. In addition, the Kappa values were 0.69 (good) and 0.55 (fair) for the h abitat of Zygophyllum eurypterum and Artemisia sieberi , respectively. This study concludes that the outcomes of Maximum Entropy Model can make possible the selection of suitable plant species for rehabilitation of degraded rangeland in addition to the iden tification of effective environmental factors in plants habitat distribution.