The structuring of ant assemblages in a Southern African savanna was investigated using data from the only long-term, large-scale savanna fire experiment in Africa. A comprehensive survey of three habitats in the Kruger National Park (KNP) revealed a total of 169 ant species from 41 genera. The sampling efficiency and consistency of pitfall traps and Winkler samples for inventory, bioindicator and ecological studies in savanna habitats was compared using ants. Pitfall traps were more efficient and productive than Winkler sampling for epigaiec ants, with a greater total species richness and higher abundance of ants recorded. Suggestions were made to improve Winkler sampling output, and to allow quantitative data to be collected. With the structuring of local assemblages, competition was the most significant local factor tested. The relationship between ant dominance and ant species richness was consistent across three continents. A model developed to test mechanisms that could be responsible for the form of this relationship supported the hypothesis that competitive exclusion by dominant ants at least partially reduces species richness. Stress was only partially responsible for low dominance and low species richness, while scatter in the data points is related to patchiness of ants at baits. These findings contrast strongly with previous claims regarding the relationship between richness and dominance. Habitat complexity was not found to play an important role in determining ant assemblage body size in this savanna system. The size-grain hypothesis (Kaspari&Weiser 1999) which predicts that environmental rugosity results in positive allometric scaling of leg length on body length because of changes in locomotion costs, was tested by comparing the body sizes of ants from areas of contrasting habitat complexity. No support for the hypothesis was found. Phylogenetic independent contrast methods did however support the allometric relationship found by Kaspari and Weiser (1999). Ant assemblages in KNP exhibited a remarkable degree of resistance, and in some cases resilience, to burning. Species richness or abundance did not vary with different burning treatments, although ant assemblage composition was sensitive to burning treatment. This difference, however, was only pronounced between burnt and unburnt plots, not between burning treatments. The degree of response of ant assemblages is likely to be related to two main contributory factors: mean annual rainfall and changes in vegetation structure with burning, and the assemblage's history of association with fire. An overview of published research on the effects of fire on fauna in Southern Africa was undertaken. Few studies have examined the effects of fire on amphibians or reptiles and few experimental studies have been undertaken using an experimental fire regime applied over appropriately long time intervals. Most studies provided no information on the scale of the study. Replication was often not reported, and was generally inadequate. Information on the effects of fire on fauna in Southern Africa is fragmentary, and consequently informed management decisions regarding the consequences of burning policies on the conservation of biodiversity both within and outside protected areas are problematic. Recommendations and suggestions for improving fire research are given.
Thesis (DPhil (Zoology))--University of Pretoria, 2005.