Effects of elephant densities on landscape heterogeneity in relation to surface water availability: A GIS and remote sensing-based approach

Abstract

Predictions have been made on the effect of increased elephant (Loxodonta africana) densities on landscape heterogeneity change as a result of artificial waterholes in semi-arid savanna landscapes of southern Africa. However, limited effort has been put forth to test these predictions spatially. This thesis sought to test the utility of a Geographic Information System (GIS) and remote sensing-based approach to understand the effects of elephant density on landscape heterogeneity change around artificial waterholes in Hwange National Park (HNP). Firstly, an objective method based on satellite remotely-sensed data was tested for efficiency in detecting and mapping waterholes at an optimal threshold. In this regard, the performance of the Modified Normalised Difference Water Index (MNDWI) and newly developed Superfine Water Index (SWI) were evaluated at selected optimal thresholds, using Landsat image data. Kappa coefficient results indicated that the MNDWI detects waterholes better than the SWI and that the optimal threshold for detecting waterholes using MNDWI is -0.29. The study further validated accuracy of MNDWI-detected waterholes using spatial and temporal rainfall data and tested for aridity, using the remotely-sensed waterholes. Secondly, the study related the spatial distribution of waterholes detected using MNDWI to elephant density and assessed short-term (seasonal scale) elephant density effects on vegetation heterogeneity change using the coefficient of variation of the Normalised Difference Vegetation Index (NDVI CoV). It also assessed long-term elephant density effects as a result of artificial waterholes on landscape heterogeneity change using landscape metrics. The study also tested the potential influence of an existing spatial rainfall gradient on landscape heterogeneity change. Results show that density of elephants is significantly influenced by spatial distribution of waterholes in the late dry season. Results show that in the short-term there is no relationship between NDVI CoV change and elephant density. In the long term, results show that woodland mean patch size, woodland class proportion and bushland patch density decrease while bushland mean patch size and bushland class proportion increase with increasing density of artificial waterholes. Results suggest that consistently high elephant density in the area with high artificial waterhole density leads to a decrease in woody vegetation structural diversity as woodland patches are converted to bushland as a result of coppicing due to intensive browsing by elephants. However, homogenisation of vegetation in the landscape is being inhibited by other mediating factors which are converting bushland to grassland. Results provide evidence that landscape heterogeneity dynamics in HNP are not significantly influenced by spatial rainfall variability. In conclusion, this thesis demonstrated that GIS techniques and remotely-sensed data can effectively be used to detect and map waterholes, assess waterhole-elephant density dynamics and quantify elephant density effects on landscape heterogeneity change over short and broad temporal intervals for sustainable management of semi-arid savanna ecosystems.