Environmental and biodiversity representativeness of global protected area network

Background/Question/Methods

Protected areas have long been the leading conservation instrument in response to continuing biodiversity loss worldwide. While about 13% of global land surface is currently covered by protected areas, the coverage has to be increased to at least 17% by 2020 to meet the Aichi Biodiversity Target. However, lack of good understanding of the environmental and biodiversity representativeness of a place with respect to its surrounding region limits our ability to maximize conservation outcomes by adjusting and expanding the current protected area network. To address this, we developed four indices of environmental representativeness as measures of the environmental (dis)similarity between a protected area and its surrounding region using remotely sensed global environmental variables, including land cover, topography, land surface temperature, solar radiation and cloud cover. We evaluated and compared the usefulness of the indices for estimating biodiversity representativeness of ~500 protected areas, measured by the percentage of regional vertebrate species protected. We then estimated and mapped both environmental and biodiversity representativeness for all IUCN category I-VI protected areas across the globe using quasi-binomial regression models built with the developed representativeness indices, regional species richness, protected area size and regional productivity.

Results/Conclusions

Among the four indices, the pixel-level environmental distance (ED) shows the highest correlation with the biodiversity representativeness, followed by the Sorensen similarity (SS) calculated based on the environmental classes. Whle the volume ratio of convex hulls (CH) in the environmental space and the environmental class-level distance (CD) have weaker correlations with biodiversity representativeness, they contain some complementary information. The quasi-binomial regression models built with combinations of the four indices effectively predicted the probability that a species in the regional pool occurs within a protected area and estimated its biodiversity representativeness. The developed indices and estimated environmental and biodiversity representativeness provide essential information to evaluate the performance of existing protected areas and to guide the expansion of the current network. Going beyond reserve evaluation and selection, the indices can also be applied to any place around the world to fill the knowledge gap on fine-grain local species richness using the information on regional richness, which can be reliably obtained from expert range maps at coarse resolutions.