Characterization of ecosystem's climate and geographical distribution of two pasture species using random forest modeling in Central Zagros region

Climatic conditions largely determine the status and characteristics of plants and plant communities in different regions. Therefore, in order to study range plant species, the tolerance of each species to these factors should be determined. In this study, climatic features and requirements of two important grass species with conservation and forage importance (Agropyron trichophorum Richt. & Bromus tomentellus Boiss.) were studied in the Central Zagros. For this purpose, 19 bioclimatic variables used as inputs to Random Forest (as a machine learning approach) in R statistacial environment. The results showed that total annual precipitation (BIO12), the average temperature of the wettest season (BIO8) and the mean diurnal range (BIO2) for both species had the greatest importance and the sum of these three variables described more than 68 and 64 percent of variance in the B. tomentellus and A. trichophorum, respectively. The estimated accuracy by TSS index was calculated for B. tomentellus as 0.95 and Sensitivity and Specificity values were estimated as 0.93 and 0.96, respectively. For A. trichophorum, these values (TSS, Sensitivity& Specificity) calculated as 0.92, 0.89 and 0.93; respectively. The results of this study explain the cliamtic niche of the studied species. Top priority places, where protective actions has to be taken, can be determined using the results of this study. Potentially climatic suitable habitats can be proposed for reintroduction of these valuable species as well.