Detecting Potential Biodiversity Hotspots for Development of REDD+ Safeguards Based on Analyses of Land-Cover Complexity in East Java, Indonesia

We examined a new method to detect the biodiversity hotspots in terms of complex patch mosaics at a regional scale in East Java, Indonesia, in order to develop the safeguard against further expansion of monocultures by REDD+. A land-cover map consisting of five major land-cover types (forest, agricultural land, bare land, water, and residential) was generated with a 30 m x 30 m resolution by the unsupervised classification of a Landsat8-OLI image. Shannon's diversity index (H') was calculated for each of 10.98 ha (11 x 11 pixels) landscape throughout the study area based on the dominance of the land-cover types by five calculation methods with different combinations of land-cover types. Then, the landscapes of upper 5% in H' was selected as the potential hotspots in terms of highly complex patch mosaics. Among the five potential hotspots, the calculation of H' with four land-cover types (forest, agriculture, water and bare land) was thought to be most suitable to set conservation targets at a regional scale, because the potential hotspots by this method showed aggregated distribution patterns, and was less sensitive to the small residential patches. While, no clear distribution trend was observed along the environmental gradients stage.