Niche modelling the distributions of large Acacia nigrescens and Sclerocarya birrea trees.

Thesis (M.Sc.)-University of KwaZulu-Natal, Westville, 2011.

MaxEnt modelling uses only the known locations of a species to predict the overall distribution of a species. Large trees are important for the functioning of savanna ecosystems, bringing nutrients to the surface, providing shelter to animals and providing a number of ecological functions. Large trees have been identified as declining in density in many southern African reserves, making the conservation of large trees within reserves an issue in park management, such as in Kruger National Park (KNP) and Hluhluwe iMfolzi Parks (HiP). Two species of primary concern are Acacia nigrescens and Sclerocarya birrea, which have similar distributions in Southern Africa. Effective management of large trees requires understanding their distribution within reserves and any potential distribution changes. By determining the current locations of a species, and using GIS layers of environmental variables to predict the extent of habitats that could support the species, niche models can predict species distribution. Maximum Entropy techniques evaluate the probability of finding the species in raster squares, with values for environmental factors controlling distribution. For this study, the locations of A. nigrescens and S. birrea trees higher than 5 m were recorded in KNP and HiP, and were used in conjunction with MaxEnt to produce distribution probability maps for both species in each reserve. In HiP, the distribution map was compared with an independent existing data set to determine if the predicted distributions were accurate. The factors effecting their distribution were compared between HiP and KNP to determine why the species were found together in KNP but not in HiP. MaxEnt could predict the locations of the species within HiP, but predictions were better for A. nigrescens than S. birrea. In both Reserves, rainfall was the best predictor of tree location, along with elevation. The niche overlap was higher in KNP, where both species are well within their total species range, than in HiP where A. nigrescens was at the edge of its distribution. These variables that are limiting distribution at a reserve scale will have an influence on the overall distribution of the species. Niche models can be used to inform the establishment of botanical reserves or other management strategies that can help preserve large trees within reserves.