Poisonous plants are one of the most important causes of economic losses in the livestock industry throughout the world, especially in those areas where rangeland and pasture grazing are practiced. In the livestock industry in South Africa, total annual costs of plant and fungal poisoning (mycotoxicosis) are conservatively estimated to be R104 506 077, 8% of which is due to D. cymosum poisoning. There is no antidote for D. cymosum poisoning of livestock and wide scale eradication by conventional control methods are labour intensive, expensive and often impractical. In South Africa, the communal or emerging farming sectors are the most affected. This is mainly due to the high costs associated with control measures such as fencing, supplementary feeding and veterinary expenses, and / or lack of grazing management with livestock numbers exceeding the biological carrying capacity of the rangeland due to lack of grazing lands. Proper rangeland management is the surest and most economical means of reducing plant poisoning of livestock. Focused research on the ecology of these poisonous plants in rangeland communities to improve rangeland management would assist in the development of these strategies. This study investigated D. cymosum infested savanna communities, focusing on understanding how negative (inter- and intraspecific competition) interactions influence community structure, dynamics and productivity and how plants in D. cymosum communities avoid these negative interactions by investigating their seasonal phenological patterns. Dichapetalum cymosum coexists with trees, such as Burkea africana, Ochna pulchra and Terminalia serecia, as well as shrubs species, such as Pygmaeothamnus zeyheri and Perinari capensis, in well drained, nutrient poor soils. Some plant ecologists contend that in an environment where water is limited, competition is inevitable amongst plants occupying the same above-ground stratum and the same soil horizon. Others contend that plants avoid competition with each other by sharing resources spatially and temporarily. Field experiments were conducted to investigate how the above species interact and coexist with each other in two South African savanna communities. Fourteen (100 m x 100 m) D. cymosum infested sites were identified in each community. Disperal analysis using nearest neighbour distance was used to investigate competition among species, and above ground flowering phenology along niche axes to determine temporal and spatial sharing of resources. The dispersal analysis revealed aggregated populations among species when intraspecific and combined (all individuals independent of species) analyses were conducted. However, in all instances, aggregation among species was not significant. No interspecific competition was observed among species when correlation analysis was performed between nearest neighbour distance and combined canopy cover of the nearest neighbour pair. Intraspecific competition was, however, observed for tree species T. sericea (n = 128; r = 0.3952; P < 0.0001) and B. africana (n = 166; r = 0.49926; P <0.0001) and a shrub species, D. cymosum (n = 391; r = 0.39788; P <0.0001). Segregation was found between O. pulchra and both B. africana (S = 0.999, χ² = 102.7588, P <0.0001) and T. sericea (S = 0.999, χ² = 57.8571, P <0.0001). Shrub species were also segregated, all with interspecific nearest neighbour pairs occurring less often than expected. The vegetative phenology of all experimental plant species followed the rainfall gradient. Differences in reproductive phenologies were observed between O. pulchra and both B. africana and T. sericea. Dichapetalum cymosum also differed from P. capensis and P. zeyheri in their reproductive phenologies. The differences in the reproductive strategies of at least one of the species in each growth form account for the observed spatial distribution amongst species in these communities. The observed growth patterns shown by the vegetative phenologies, however, suggest that lengthy retention of nutrients is a strategy to avoid competition for nutrient uptake with other species in these communities. Segregation between species and positive correlation iii between nearest neighbour distance and combined canopy cover of the nearest neighbour pairs suggest that intraspecific competition and interspecific facilitation determine D. cymosum woody plant community structure. This study had limited application to rangeland management. However, it can be concluded that grazing of D. cymosum communities should take place during mid-summer, when enough grazing material is available to allow animals to vary their diet. The introduction of animals in poor condition or naïve animals into these lands should be avoided in winter and spring as they will graze non-selectively resulting in D. cymosum poisoning. To utilize these areas as grazing lands, supplements need to be provided to assist in the detoxifications of toxins once ingested. / Dissertation (MSc Agric)--University of Pretoria, 2013. / gm2014 / Plant Production and Soil Science / unrestricted