COS 86-3
Response of savanna tree cover to the near-extirpation of large mammalian herbivores in Mozambique's Gorongosa National Park, 1977-2012

Wednesday, August 12, 2015: 2:10 PM
325, Baltimore Convention Center
Joshua Daskin, Ecology and Evolutionary Biology, Princeton University, Princeton, NJ
Marc Stalmans, Department of Scientific Services, Gorongosa National Park, Mozambique
Robert M. Pringle, Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ

Large mammalian herbivores (LMH) exert strong effects on vegetation in many ecosystems, but their population densities are decreasing (native ungulates) over much of Africa and Asia due to anthropogenic influences. Experiments at tractable spatial and temporal scales suggest that LMH declines in Africa may have strong community and ecosystem level impacts, in part by influencing the extent of tree cover. However, the difficulty of manipulating LMH populations has largely limited the study of LMH’s role in regulating savanna tree cover to these relatively short-term, small-scale experiments, which may not accurately simulate the effects of real-world species loss. We used the war-driven near-extirpation of LMH from Gorongosa National Park, Mozambique, as a natural experiment to test how an abrupt decline in all large-mammal species affected savanna tree cover over 35 years. To measure changes in tree cover, we classified woody vs. herbaceous vegetation across the entire 4000 km2park from satellite images taken before (1977) and after (2012) LMH declines.


We found a 28% increase in park-wide tree cover (from 29 to 37%); most regions of the park showed far greater increases (51-134%), while only on the large Cheringoma plateau (1050 km2) in the east of the park had nearly steady tree cover. Cheringoma also had historically lower LMH population densities. The observed increase in tree cover suggests that a release from browsing pressure following the war-driven mammal declines increased tree recruitment and/or growth. Potential alternative drivers of tree cover, such as rainfall and fire regimes, showed no directional trend during this period and thus cannot account for increased tree cover. Our analysis provides a multi-decadal, whole-ecosystem example of LMH impacts on savanna vegetation structure. Ongoing changes to LMH population densities present the strong potential shifts between woodland and grassland savanna habitats, and for cascading effects on other species.