Mapping and predicting forest loss in a Sumatran tiger landscape from 2002 to 2050

Erin E Poor; Yang Shao; Marcella J Kelly

doi:10.1016/j.jenvman.2018.10.065

Mapping and predicting forest loss in a Sumatran tiger landscape from 2002 to 2050

J Environ Manage. 2019 Feb 1:231:397-404. doi: 10.1016/j.jenvman.2018.10.065. Epub 2018 Oct 24.

Authors

Erin E Poor¹, Yang Shao², Marcella J Kelly³

Affiliations

¹ Virginia Tech, Department of Fish and Wildlife Conservation, 100 Cheatham Hall, 310 W. Campus Drive, Blacksburg, VA 24060, USA. Electronic address: erinpoor@vt.edu.
² Virginia Tech, Department of Geography, 115 Major Williams Hall, 220 Stanger Street, Blacksburg, VA 24060, USA.
³ Virginia Tech, Department of Fish and Wildlife Conservation, 100 Cheatham Hall, 310 W. Campus Drive, Blacksburg, VA 24060, USA.

PMID: 30368149
DOI: 10.1016/j.jenvman.2018.10.065

Abstract

Riau Province in central Sumatra, with its peatland, lowland, and montane forest habitats, was once a stronghold for Sumatran tiger (Panthera tigris sumatrae) populations. Today, Riau may have one of the highest deforestation rates in the world and wildlife populations are dwindling, with natural forest now comprising approximately only 18% of the province, mostly contained within protected areas. Agriculture (acacia, rubber, and oil palm) makes up the majority of Riau's land cover and deforestation for the creation of new plantations is rampant. Natural forest and tigers still remain in Bukit Tigapuluh National Park and Rimbang Baling Wildlife Reserve, which remain connected to tiger populations in montane forest on the western edge of Sumatra. In this study, using freely available Landsat imagery and a maximum likelihood classification algorithm, we create land cover maps for central Sumatra from 2002 to 2016. We then use current land cover, elevation, and slope variables to predict changes from forest to plantation from 2016 to 2050 at five year intervals using a multilayer perceptron neural network. Finally, we compare connectivity based on a 100 km distance threshold (based on potential tiger dispersal) across the landscape and across years. Land cover maps had 80-90% accuracy, and we predict forest in Tesso Nilo and the western edge of the study area to be lost by 2050 given current rates of deforestation. Our connectivity analysis shows that Tesso Nilo and the area between Rimbang Baling and Bukit Tigapuluh are important components for maintaining connectivity throughout the study area. Focusing conservation and rehabilitation efforts on forests close to plantations in flat areas, including Tesso Nilo, is necessary to maintain forests and increase connectivity in Riau to ensure future habitat connectivity for survival of tigers and Sumatra's other diverse endemic species.

Keywords: Connectivity; Deforestation; Land cover map; Land cover prediction; Palm oil.

MeSH terms

Animals
Conservation of Natural Resources
Ecosystem
Forests
Indonesia
Tigers*