Potential and Limitations of Multi-Temporal Earth Observation Data to Improve Modeled Results of Tree Species Distribution in Mexico

Abstract

The study explored the potential of multi-temporal remote sensing data for distribution modeling of selected tree species belonging to the genera Pinus spp. (pine) and Quercus spp. (white oak) in Mexico. Several environmental predictor data sets at 1 km2 spatial resolution were used in combination with the Maxent algorithm (Phillips et al., 2004), namely (1) phenological metrics derived from the Terra-MODIS 16-day vegetation indices product MOD13A2 averaged over the seven years of the study period from 2001 to 2007, (2) topographic data (elevation, slope, and aspect) of the SRTM mission, and (3) a series of bioclimatic variables (WorldClim, Hijmans et al., 2005) derived from monthly temperature and rainfall values. Different model scenarios were compared and showed that remote sensing data contributed significantly to discover habitat characteristics even within similar climatic conditions. Moreover, a sharper delineation of the predicted areas and better exclusion of regions that had suffered land cover change was possible. The improved distribution maps can contribute to long-term and sustainable conservation planning and management of biodiversity hotspots.