For the first time in human history, more people live in urban areas than in rural areas, and the patterns of suburbanization and urban sprawl once characteristic of North America are now present globally (Obaid 2007). As conservation biologists seek to prioritize conservation efforts worldwide, urbanization and agricultural development emerge as two of the most extensive processes that threaten biodiversity. Suburban and rural sprawl are significant drivers of forest fragmentation and biodiversity loss (e.g., Murphy 1988; Radeloff et al. 2005). Data on human impacts is often averaged across political boundaries rather than biogeographic boundaries, making it challenging to use existing data sets on human demography in ecological studies and relate human population change to the changes in populations of other species. Remotely sensed data can make major contributions to mapping human impacts in ecologically relevant ways. For example, Ricketts and Imhoff (2003) assigned conservation priorities (based on species richness and endemism) for the United States and Canada using several different types of remotely sensed data. For mapping urban cover, they used the map of “city lights at night” from the Defense Meteorological Satellite Program (Imhoff et al. 1997) to classify land as urbanized or not urbanized. For mapping agricultural cover, they used the USGS North America Seasonal Land Cover map (Loveland et al. 2000), derived from the Advanced Very High Resolution Radiometer (AVHRR), lumping five categories to create an agricultural land class. For ecological data, they used a compilation of ecoregion boundaries combined with range maps for over 20,000 species in eight taxa (birds, mammals, butterflies, amphibians, reptiles, land snails, tiger beetles, and vascular plants; Ricketts et al. 1999). Analyzing these data, Ricketts and Imhoff (2003) identified a strong correlation between species richness and urbanization. Of the 110 ecoregions studied, 18 ranked in the top third for both urbanization and biodiversity (species richness, endemism, or both); some of the ecoregions identified as priorities were not identified by a previous biodiversity assessment that did not include the remotely sensed mapping of urbanization (Ricketts et al. 1999).
Mapping the species richness and composition of tropical forests from remotely sensed data with neural networks
