Scholes, Robert John2015-09-112015-09-112015-09-11http://hdl.handle.net/10539/18624A thesis submitted to the Faculty of Science University of the Witwatersrand, Johannesburg for the Degree or Doctor or Philosophy October 1987A t h r e e - y e a r study was un dertaken in the semi-arid (500mm p a . ) n o rth -eastern lowland area of South Afric a. All the woody plants were removed from one -hectare savanna plots on three d i f f e r e n t soil types , and key hydrological and biological changes were monitored relative to adjacent controls. Runoff increased initially, but decreased once the grass cover increased. Deep drainage and lateral subsurface flow increased on the sandiest site. Evaporation from the soil surface increased on the heavier t e x t u r e d soils. The duration of plant - available water in the soil increased on all cleared p l o t s . & The p re - c l e a r in g woody plant abo veground biomass was in the range of of which 0 . 6 6 to 0 . 8 0 t ha was 5.6 to 11.2 t ha -1 leaf biomass. The annual herbaceous production was stron gly rainfal' dep endent, averaging 1 to 1.5 t ha -1 , and increased by 0 . 4 - 0 6 t .ha -1 except on the most f erti le site ( 0 . 6 - 2 0 t ha ) . Total available forage increased with c l e a ring, but so did its variability The observed changes in herbaceous layer palatability could not be a t trib u t e d to clearing. Woody plants and grasses wore shown to have w a te r- u s e niche separation in both rooting depth and time of water use. Simulation over forty years of wetting patterns indicated 75 to 85 % niche overlap, w t h separation on the depth axis more important in sandy sites, and on the time axis in clayey sites. Competition between woody plants and grasses was strongly asymmetrical in favour of woody plants. w >enSavanna ecology--South AfricaDeforestationVegetation dynamicsSoils--AnalysisThesis