Using Remote Sensing to Uncover the Drivers of Land-cover Change that Elephant Habitat Use in the Serengeti Ecosystem

Estes, Anna Bond, Department of Environmental Sciences, University of Virginia
Shugart, Hank, Department of Environmental Sciences, University of Virginia
Wilbur, Henry, Department of Biology, University of Virginia
Epstein, Howie, Department of Environmental Sciences, University of Virginia
D’Odorico, Paolo, Department of Environmental Sciences, University of Virginia

Increasing human populations and land conversion around protected areas can exacerbate human - wildlife conflict, increase fire prevalence, and restrict the movement of large mammals, including elephants. Human - wildlife conflict erodes local support of protected areas and has substantial economic consequences for subsistence livelihoods. Increases in fire frequency and restriction of elephants inside protected areas can impact woody vegetation. The interactions between humans, fire and elephants, and the resultant influence on savanna ecosystems is however not well understood, and has seldom been investigated at larger spatial scales. In this dissertation, I used remote sensing to investigate these interactions at the scale of a 25,000 kmz ecosystem. I used time series of Landsat satellite images to identify patterns of land - cover change in and around the protected areas in the Greater Serengeti Ecosystem in Tanzania and Kenya, and compared these changes with human population trends, fire histories, and models of elephant habitat use constructed with data from GPS tracking collars. I found that the rate of human population growth and agricultural conversion was greatest near the western boundary of the protected area, likely because of lack of arable land farther away. At each of ~10 - year time intervals between 1984 and 2011, there were numerous, patchy changes between woodland and grassland (only about half the woodland and grassland remained stable), yet averaged across the ecosystem, there was only a 4% reduction in woodland. Higher fire frequency was correlated with woodland declines, and the only period in which woodlands increased coincided with low elephant numbers. NDVI, distance to rivers, and land cover were major determinants of elephant habitat suitability, and some of the most highly suitable habitat was found adjacent to increasing human populations in western Serengeti. The presence of highly suitable habitat along the park boundary accurately predicted hotspots of human - elephant conflict, indicating that knowledge of underlying habitat preference is an important aspect of creating successful mitigation strategies. The results of this dissertation show that remote sensing and species distribution models can be powerful tools to better understand the interactions of disturbances in savanna ecosystems, and can yield information essential to designing appropriate ecosystem management plans.

Note: Abstract extracted from PDF file via OCR

PHD (Doctor of Philosophy)
All rights reserved (no additional license for public reuse)
Issued Date: