Species distribution potential of striped hyaena (Hyaena hyaena) in response to climate change in Iran

Carnivores have always been exposed to threat-inducing processes because of their placement at the top of the food pyramid. Nowadays, approximately 65% of carnivores are listed as Critically Endangered (CR) or Vulnerable (V) on the IUCN Red List of threatened species. On a global scale, in the order Carnivora the Hyaenidae family is the smallest with only four species. Hyaena hyaena is the only member of the Hyaenidae family found in Iran, and is listed as Near Threatened (NT) in the IUCN Red List. The current study was aimed at modelling the species distribution of the striped hyaena (Hyaena hyaena), which is wide in Iran. Using nineteen layers of so-called Bioclimatic variables, the correlations between those variables were analyzed and then highly correlated variables were excluded from the modelling process. Finally, eight species distribution models from the sdm package (GLM, GAM, BRT, SVM, RF, MARS, CART, and FDA) in R software were used. Given that the output of species distribution models is, unavoidably, often uncertain, one possible solution is to estimate the difference between projections and, to reduce the uncertainty, the use of an ensemble prediction system instead of using a single modelling method. Therefore, after determining the potential habitats of the Hyaena hyaena using those eight mentioned models and by using the ensemble prediction system, the best regions for the distribution of this species in Iran were estimated. The results of this study showed that annual mean temperature, seasonal precipitation and precipitation during the warmest season have the greatest influence on the distribution of Hyaena hyaena. In general, the FDA, GAM, BRT, CART and GLM models are fairly reliable, the MARS model is very reliable, and SVM and RF models are completely reliable. The results showed that the GLM, GAM, BRT, MARS, CART, RF models demonstrated that the suitable areas for Hyaena hyaena are generally the central regions of Iran, while the SVM and FDA models predicted the southern margin of the Caspian sea as containing the best regions for the distribution of this species. The difference in the predictions of each model makes confirms the uncertainty between the models which confirms the necessity of using the ensemble method. The results of the ensemble model showed that the most suitable regions for the Hyaena hyaena species distribution are the semi-arid and central steppe regions of Iran.