Ensemble modelling the distribution and habitat suitability of wild goat Capra aegagrus in southwestern Iran

ABSTRACT Increasing anthropogenic pressures have caused drastic population decline of wild goats Capra aegagrus. The study aim was to determine habitat suitability and explore the factors influencing habitat selection and dispersal of wild goats in the Khaeiz and Sorkh protected area, southwestern Iran. We utilized ensemble modelling based on 25 replications of 10 species distribution models GLM (generalized linear model), GAM (generalized additive model), GBM (gradient boosting model), CTA (classification tree analysis), FDA (flexible discriminant analysis), ANN (artificial neural network), MARS (multivariate adaptive regression spline), RF (random forest), MaxEnt (maximum entropy), SRE (surface range envelope). Wild goat occurrence data were collected from 2019 to 2020 and related to five variables: predator distribution, slope, distance from road, distance from water resources, and normalized difference vegetation index (NDVI). We found that distribution of predators, slope, distance from water resources and distance from roads were the most important predictors of potential habitats for wild goats. The least important variable was NDVI. Probabilistic predictions suggest that only 5.89% of the area was highly suitable habitat, whereas 46.24% was unsuitable. Ensembling individually classified presence/absence maps further indicates that 79.72% of the study area was certainly unsuitable while only 8.06% was certainly suitable for wild goats. The results can help wildlife managers and policymakers work towards conservation and management goals and decreasing the conflicts between wildlife and humans.