High-resolution species-distribution model based on systematic sampling and indirect observations

Abstract

Species distribution models (SDMs) are often limited by the use of coarse-resolution environmental variables and by the number of observations required for their calibration. This is particularly true in the case of elusive animals. Here, we developed a SDM by combining three elements: a database of explanatory variables, mapped at a fine resolution; a systematic sampling scheme; and an intensive survey of indirect observations. Using MaxEnt, we developed the SDM for the population of the Asiatic wild ass (Equus hemionus), a rare and elusive species, at three spatial scales: 10, 100, and 1000 m per pixel. We used indirect observations of feces mounds. We constructed 14 layers of explanatory variables, in five categories: water, topography, biotic conditions, climatic variables and anthropogenic variables. Woody vegetation cover and slopes were found to have the strongest effect on the distribution of wild ass and were included as the main predictors in the SDM. Model validation revealed that an intensive survey of feces mounds and high-resolution predictor layers resulted in a highly accurate and informative SDM. Fine-grain (10 and 100 m) SDMs can be utilized to: (1) characterize the variables influencing species distribution at high resolution and local scale, including anthropogenic effects and geomorphologic features; (2) detect potential population activity centers; (3) locate potential corridors of movement and possible isolated habitat patches. Such information may be useful for the conservation efforts of the Asiatic wild ass. This approach could be applied to other elusive species, particularly large mammals.