Global Ecology and Conservation (Nov 2021)
Using an integrative mapping approach to identify the distribution range and conservation needs of a large threatened mammal, the Asiatic black bear, in China
Abstract
Assessing a species’ threatened status and then developing specific conservation strategies accordingly rely heavily on knowing that species’ complete and accurate spatial distribution. In this study, we used the Asiatic black bear (Ursus thibetanus) in China to represent a large threatened species for which distribution information is limited and spatially biased. We grouped the two main sources of black bear occurrence data into two different resolutions: (1) coarse resolution data corresponding specific management units (e.g., nature reserves) that cover large areas, and (2) fine resolution data composed of longitude and latitude records that were bias in their geographic range. Our distribution mapping approach integrated those two data types to examine black bear spatial patterns across the country. We used both presence and absence data in the Random Forest algorithm to predict black bear distribution at coarse (30 km) and fine (3 km) resolutions, and then refined the coarse-scale prediction with the fine-scale prediction using a map fusion technique based on the Bayes theorem. We thus generated an integrated high-resolution range map that was both more accurate than the coarse-scale map and more representative of the black bear geographic range than was the fine-scale map. Our results showed that the total estimated range of Asiatic black bears in China was 462.3 × 103 km2, 77.50% less than the most recent IUCN range map and 70.90% less than the area of habitat (AOH) estimation. Using those results, we identified two island and six mainland management units in China and, based on the predicted habitat conditions, proposed specific conservation strategies for each unit. Our study results provide practical knowledge and pragmatic guidance for future conservation planning and action for this species, and our framework provides an example and template for range estimations of species with similar types of occurrence records.
