Plants are the major source of food or energy required to sustain life on the planet, but humans are grappling with the deteriorating conditions of natural ecosystems such as compositional change, desertification, invasive plants and soil erosion. In the face of global climate change and growing demands for agricultural productivity, future pressures on grassland ecosystems will intensify, therefore sustainable utilization of all plant resources is of vital importance to enhance food security within the limits of good conservation. The semi-arid grasslands of southern Africa represent major grassland resources for grazing. Herbage production in these areas is determined not only by water and nutrient availability, but also by controlled and uncontrolled fires. Since fire is regarded as a natural factor in savannas, it is essential to develop a deeper understanding of the role of fire in community structure and function for the development of appropriate burning regimes. A study was conducted in the Eastern Cape of South Africa where the rural communities are faced with the challenges of rangeland degradation in the form of encroachment by unacceptable bush, karroid, macchia and less desirable grass species, as well as soil erosion. The main objective of this thesis was to investigate the roles of competition and disturbance regimes (fire and simulated non-selective grazing) on species composition, habitat productivity and the performances of selected species from this semi-arid savanna. Long-term effects of burning frequency on herbaceous species composition, Leaf Area Index (LAI), Photosynthetically Active Radiation (PAR) within the herbaceous canopy, biomass production and soil chemical properties were investigated. These studies were conducted on a fire trial set up in 1980 at the University of Fort Hare research farm in the Eastern Cape, South Africa. The treatments comprise an annual, biennial, triennial, quadrennial, sexennial and no burn control, all replicated twice in a Complete Randomized Design. The data from the trial collected between 1980 and 2008 were used to determine compositional variation for herbaceous species using the Non-metric Multidimensional Scaling and Bray-Curtis Dissimilarity tests. The PAR ceptometer was used to determine LAI and intercepted PAR, while random samples were harvested from 1m² quadrats from each plot. Soil samples were taken at four depths (0-2 cm; 2-4 cm; 4-6 cm and 6-8 cm) from each plot and analyzed for pH, Ca, K, P, total C and total N. The Resin-Bag technique was used to determine nitrogen mineralization. Burning frequency caused significant variation in herbaceous species composition over time. The species were distributed along gradients of increasing burning frequency, and these responses were in three categories: Those that increased with burning frequency such as Themeda triandra; those that decreased with burning frequency such as Melica decumbens, and those that showed little response such as Panicum maximum. The three-year burn resulted in the highest compositional variation, light interception, Leaf Area Index, aboveground biomass production, while the annual, biennial and no burn treatments resulted in the lowest. The fact that infrequent burning resulted in higher species variation, improved habitat productivity due to increased leaf area for light interception shows that appropriate use of fire can maintain a more diverse and productive savanna system. Burning frequency had significant effects on the soil properties, while soil depth did not show any significance. Frequent burning increased soil pH, K, Ca, and Na, but reduced C, N, P and N mineralization. There was a negative correlation between burning frequency and N mineralization, but no correlation existed between N mineralization and total N, total C or the C:N ratio. These results imply that frequent burning can cause nutrient losses and a greater nutrient limitation to plants in the long-term, especially soil C and N loss from combustion of organic material in the soil top layer. The ability of shade-tolerant plants to persist under shade and regular defoliation such as in burnt and grazed systems may be of greater importance for long-term productivity and sustainability of forage crops. It is therefore imperative to explore the mechanisms by which some species were favoured by frequent burning which created low shade conditions, while others were favoured by high shade conditions where burning is infrequent or absent. A pot experiment was conducted to investigate the shade tolerances of seven grass species that were abundant in the long-term fire trial. The test species were Cymbopogon plurinodis, Digitaria eriantha, Eragrostis curvula, Melica decumbens, Panicum maximum, Sporobolus fimbriatus and Themeda triandra. Individual grass tillers of each species were collected from the natural vegetation, propagated in separate seedling trays and transplanted into individual pots, and were grown under five shading treatments: full sun (0 % shading), 55 %; 70 %; 85 % and 93 % shading respectively. Shading significantly reduced the dry matter production of all the species. Biomass production of all the species decreased linearly to varying degrees with an increase in shade intensity. Digitaria eriantha and Eragrostis curvula were most adversely affected by shading, hence are classified as shade intolerant, while Melica decumbens was the least affected by shading, and is hence classified as shade tolerant. Cymbopogon plurinodis, Panicum maximum, Sporobolus fimbriatus and Themeda triandra are classified as moderately shade-tolerant. From the results it was apparent that some species could perform optimally in partial shade than in full sunlight, and these results lead to a conclusion that for satisfactory natural regeneration and seedling growth of this savanna vegetation would require a gap large enough to provide at least 30 % of ambient light. Investigating patterns in competitive effects and responses of species in these communities may not only explain the abundance of each species, but may also provide insight into the nature of forces that affect the structure and function of that community. Since fire, herbivory and soil nutrients are natural drivers of savanna community structure and function, their influence on competitive interactions of selected species were investigated. Two experiments were conducted to investigate the competitive effects and responses of eight selected common species in the area. The test species (phytometers) included one woody shrub, Acacia karroo and seven grass species namely: Cymbopogon plurinodis, Digitaria eriantha, Eragrostis curvula, Melica decumbens, Panicum maximum, Sporobolus fimbriatus and Themeda triandra. In an outdoor plot experiment the responses of the phytometers to competition from neighbours (0; 2 and eight neighbours respectively), fertility (fertilized, unfertilized) and clipping (clipping, no clipping) were investigated. The second comprised a pot experiment where the competitive effects of the species were investigated. Each species was grown under 3 levels of fertility (0 %; 50 % and 100 % Hoagland‘s solution) and clipping (clipping, no clipping) in pots filled with fine river sand and 4 neighbours. Competition intensity, soil fertility and clipping had significant effects on the biomass production of the phytometer species. Acacia karroo and Melica decumbens, exhibited the weakest competitive effects and responses, and incurred the highest mortalities after clipping and with 8 neighbours. Digitaria eriantha and Panicum maximum exhibited the strongest competitive effects and responses, especially in high fertility, and experienced the lowest mortalities. T.triandra exhibited stronger competitive effect after clipping in low fertility, while A. karroo and C. plurinodis exhibited stronger competitive effects in moderate (50 %) fertility. Cymbopogon plurinodis, Eragrostis curvula and Sporobolus fimbriatus ranked between these two extreme groups in terms of competitive effects and responses. Relative Competitive Interaction increased with soil fertility and number of neighbours in the absence of clipping. These results indicate that in general, taller or broad-leaved grass species outgrow the shorter ones, and this gives them a competitive advantage over light and soil resources. One of the range management objectives in the False Thornveld of the Eastern Cape is to promote the abundance of Themeda triandra, which is of high forage value and an indicator of rangeland that is in good condition. The general situation under livestock farming conditions in this area is that if the grass sward is optimally grazed and rested then there is a great potential for Themeda triandra to dominate.The results of the competition experiments indicated that the species exhibits strong competitive interaction, and also exhibited stronger competitive effect after clipping in low fertility. These results imply that it has a low response and a high effect, an attribute that would enhance its performance if it is moderately grazed or the area is burnt. The species is also moderately shade tolerant, and this may explain why it thrives in burnt and moderately grazed areas. These studies have demonstrated the important role that competition and disturbance in the form of fire and herbivory play in the maintenance of this savanna grassland. Through natural selection species are able to occupy different niches in the same area and coexist in a heterogeneous environment and minimize their chances of extinction. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2012.