Improving the viability of large-mammal populations by using habitat and landscape models to focus conservation planning

Abstract

Context. Assessing the viability of animal populations in the wild is difficult or impossible, primarily because of limited data. However, there is an urgent need to develop methods for estimating population sizes and improving the viability of target species. Aims. To define suitable habitat for sambar (Cervus unicolor), banteng (Bos javanicus), gaur (Bos gaurus), Asian elephant (Elephas maximus) and tiger (Panthera tigris) in the Western Forest Complex, Thailand, and to assess their current status as well as estimate how the landscape needs to be managed to maintain viable populations. Methods. The present paper demonstrates a method for combining a rapid ecological assessment, landscape indices, GIS-based wildlife-habitat models, and knowledge of minimum viable population sizes to guide landscape-management decisions and improve conservation outcomes through habitat restoration. Key results. The current viabilities for gaur and elephant are fair, whereas they are poor for tiger and banteng. However, landscape quality outside the current distributions was relatively intact for all species, ranging from moderate to high levels of connectivity. In addition, the population viability for sambar is very good under the current and desired conditions. Conclusions. If managers in this complex wish to upgrade the viabilities of gaur, elephant, tiger and banteng within the next 10 years, park rangers and stakeholders should aim to increase the amount of usable habitat by ~2170 km2 or 17% of existing suitable habitats. The key strategies are to reduce human pressures, enhance ungulate habitats and increase connectivity of suitable habitats outside the current distributions. Implications. The present paper provides a particularly useful method for managers and forest-policy planners for assessing and managing habitat suitability for target wildlife and their population viability in protected-area networks where knowledge of the demographic attributes (e.g. birth and death rates) of wildlife populations are too limited to perform population viability analysis.