Potential distribution models from two highly endemic species of subterranean rodents of Argentina: which environmental variables have better performance in highly specialized species?

South American rodents of the genus Ctenomys (tuco-tucos) occupy the underground environment, present high specificity to loose and friable soils and have restricted mobility, with a generally fragmented distribution. We use species distribution models (SDMs) in two Ctenomys species from the Atlantic coast and continental areas of Argentina. We develop SDMs using Maxent software for Ctenomys australis and Ctenomys talarum, which coexist in a narrow coastal landscape with restricted distributions. We model the potential distributions of both species using, first, bioclimatic variables (Group 1), and second, Landsat 8 bands and granulometric layers (Group 2). According to the known distributions of the species, the Group 2 variables showed the greatest accuracy for inferring their potential distributions. The most important variables for predicting habitat suitability were, primarily, the majority of granulometric variables and some Landsat 8 bands such as the bands 4 and 5, related to the vegetation cover. We also analyze the level of overlapping niches between these two species, and we found that there is a certain degree of geographical overlap between them, and also present ecologically similar niches, despite the fact that the characteristics of their habitats differ in certain aspects. We conclude that in tuco-tucos species, their potential distributions are better predicted by variables that consider the particular characteristics of soils and cover vegetation, since they are specialized species of substrates. Also, a higher spatial resolution allows a better performance of the Ctenomys species models, which was expected for species with restricted distributions.