Abstract
The environmental data used to calculate ecological niche models (ENM) are obtained mainly from ground-based maps (e.g., climatic interpolated surfaces). These data are often not available for less developed areas, or may be at an inappropriate scale, and thus to obtain this information requires fieldwork. An alternative source of eco-geographical data comes from satellite imagery. Three sets of ENM were calculated exclusively with variables obtained (1) from optical and radar images only and (2) from climatic and altitude maps obtained by ground-based methods. These models were compared to evaluate whether satellite imagery can accurately generate ENM. These comparisons must be made in areas with well-known species distribution and with available satellite imagery and ground-based data. Thus, the study area was the south-western part of Salamanca (Spain), using amphibian and reptiles as species models. Models’ discrimination capacity was measured with ROC plots. Models’ covariation was measured with a Spatial Spearman correlation. Four modelling techniques were used (Bioclim, Mahalanobis distance, GARP and Maxent). The results of this comparison showed that there were no significant differences between models generated using remotely sensed imagery or ground-based data. However, the models built with satellite imagery data exhibited a larger diversity of values, probably related to the higher spatial resolution of the satellite imagery. Satellite imagery can produce accurate ENM, independently of the modelling technique or the dataset used. Therefore, biogeographical analysis of species distribution in remote areas can be accurately developed only with variables from satellite imagery.
