Habitat suitability modeling of cave dweller bats in Rudbar county to introduce the conservation-management unit of forest

Forest management involves resource management, natural disturbance, and wildlife habitat. To achieve a sustainable equilibrium, managers require explicit ecological information about the distribution of biodiversity hotspots in the matrix of the mountain-forest landscape. Cave-roosting bats are a keystone species in caves and an ecological indicator of forest quality that are sensitive to ecosystem changes caused by human activities. Spatial explicit models are appropriate tools to identify biologically sensitive areas in achieving key habitats of the forest. In this study, we used the geographical locations of caves in the study area for habitat suitability modeling of cave-dwelling bats. We used spatial explicit models with variables of agriculture, aspect, climate, forest, NDVI, precipitation, range, and water resources (a combination of rivers and streams) and six algorithms: Boosted Regression Tree, Multivariate Adaptive Regression Splines, Mixture Discriminant Analysis, Random Forest, Recursive Partitioning and Regression Trees, and Support Vector Machines in the sdm package. According to the AUC criterion, the best models were RF and BRT with a value of 0.87, and the RPART model with a value of 0.67 had the lowest value of this measure. In explaining the models, water resources and forest variables had the highest scores. We determined the suitable habitats predicted by the ensemble model as a unit of forest ecosystem management at the landscape level. An area of the management unit that has been disrupted by roads and other activities such as sand extraction and tourism development has been determined as a conservation unit.