We expected to find the answer in existing land cover products A

We expected to find the answer in existing land cover products. As we shall now explain, these products are not sufficient for our needs. While GlobCover (ESA and UCLouvain 2010) maps croplands and urban areas, mosaics of croplands and natural areas and a variety of other ecosystems, it incorrectly evaluated

the extent of land conversion and subsequent availability of lion habitat. For example, an immense area, nearly 500 km from north to south and stretching over 4,000 km west to east across the entire map (and to areas further east of it), indicates no land use conversion (Fig. 1). Such an area would be of obvious conservation value if intact; however our mapping, using Google Earth imagery at an elevation of ~10 km, shows that people have converted virtually the entire area to cropland (Fig. 1). Fig. 1 In West Africa, there is a large overlap (purple) between Erismodegib in vivo GlobCover’s (ESA and UCLouvain 2010) mapping of anthropogenic land uses (i.e. croplands, cropland mosaics and urban

areas) with areas of user-identified land conversion. GlobCover, however, misses NSC23766 molecular weight large areas (shown in red) that it classifies as unmodified savannahs, but which show fine-grained, extensive conversion to crops when viewed in high-resolution imagery. At the bottom left is Google Earth imagery of a roughly 9 by 5 km area viewed at ~10 km above the surface. It shows an extensive mosaic of fields, even more apparent at lower elevation (bottom right). (Color figure online) Calibration of land use conversion with human population density Since GlobCover (ESA and UCLouvain 2010) is unsuitable for our purposes, we explored whether models of human population provided a better correlation with land conversion. The aim was to find an estimate of human population density that best matched extensive land conversion. We used four focus areas distributed throughout the African lion’s range to compare human population at various FAK inhibitor densities with a high-resolution satellite-based land conversion layer (Supplemental materials, Fig. S1). Figure 2 shows the proportion of overlap in areas between the

user-identified land conversion and people at varying densities across the four focus areas. We define overlap as being when the layers indicate both conversion and the Ribonucleotide reductase threshold for human population density is met, and also where there is no conversion and the threshold is not met. For all four areas, overlap peaks between 10 and 25 people per km2. (Details are in Supplemental materials, Table S2). This permitted us to use human population density as a proxy for land-use conversion for areas where we did not define the latter directly. When the user-identified land conversion layer was not available, we used a density of 25 people per km2 to constrain LCUs, a threshold we consider further in the “Discussion” section. Fig.

Comments are closed.