I describe the spider biodiversity for a savanna ecosystem, assess sampling techniques, investigate surrogate measures of species richness and measure the biotic and abiotic processes affecting spider diversity. Spiders were sampled at Makalali Game Reserve, Northern Province, South Africa from February to December 1999 using pitfall traps, sweep netting, beating and active searching. A total of 4832 individuals from 268 species (14 potentially new), 147 genera (8 endemic and 2 new records for South Africa) and 37 families (1 new record for South Africa) were recorded. There was no overall significant difference in spider diversity among different physiognomic habitat types. However, analysing the results at a functional group level revealed that the web builders were significantly affected by the habitat type. Mopane woodland habitat type had the greatest number of web builders and general bushveld the least. Sweeping and active searching sampled the greatest number of individuals and species respectively. I recommend a combination of at least beating and active searching, which together sampled the highest number of unique species, for efficient and cost effective surveys. There was a significant relationship between the spider species richness and other invertebrate richness. However, the relationship is not significant when functional groups are considered separately. There was also a significant relationship between the number of species and families and species and genera. However, species level identifications remain ideal for conservation purposes. Inexperienced participants significantly overestimate the number of species. The use of surrogates is not supported by the work conducted in this study. It is still unclear what biotic and abiotic processes or combination of processes influence spider diversity patterns at the local scale. Different spider functional groups are significantly influenced by different factors. However, habitat diversity (branches and vegetation density) was the most common factor influencing spider diversity . Predicted diversity (modelled using GIS and beta-coefficients from multiple regression analyses) was higher than measured diversity values. While further research into the role of other environmental variables is clearly required, current reserve management should aim to maximise microhabitat structural diversity.

Thesis (M.Sc.)-University of Natal, Durban, 2000.