Species distribution and forest type mapping in Mexico

The study analyzed the potential of multi-temporal satellite remote sensing data (Terra-MODIS) for phenological studies in Mexico. Within two conceptually and methodologically independent approaches, we analyzed vegetation index data to continuously predict species' occurrence and linked these predictions to a satellite data based land cover map of Mexico. Phenological metrics derived from interpolated MODIS EVI time series together with topographic data of the SRTM mission and bioclimatic variables (WorldClim) were used as environmental predictors for Maxent distribution modeling of six Quercus species. We evaluated model performance based on 20% independent occurrence data using model sensitivity, model specificity, and AUC statistics. Forest classes were extracted from a recently accomplished national land cover map of Mexico (NALCMS) with approximate accuracy of 82%. We found a remarkable spatial agreement of modeled Quercus distribution and forest classes. The value-added combination of both approaches allows for improved vegetation maps and the methodology has a great potential of transferability.