The objective of this study was to describe the impacts of the density of Chromolaena odorata (chromolaena) on species composition in coastal grasslands and to investigate serial changes in the vegetation following the implementation of a burning programme. The thesis deals with key ecological concepts and issues, so a comprehensive literature review is included. Chromolaena invades coastal grasslands that are not burnt regularly (i.e. biennially). Grasslands that were not burnt for 30 years were seral to secondary forest. The successional pathway from open grassland to closed canopy forest varied according to soil type. Coastal grasslands on Glenrosa soils were characterised by savanna at an intermediate stage between the grassland and forest states. Shading ended the persistence of savanna species (e.g. Combretum molle, Dichrostachys cinerea and Heteropyxis natalensis) in forest, whereas forest precursors (e.g. Canthium inerme, Maytenus undata and Protorhus longifolia) only established where fire was absent. Chromolaena infestations were characterised by multi-stemmed adult plants of variable height (i.e. 1-3 m), depending on soil type. Regic sands did not support stratified woody vegetation and chromolaena infestations were self-supporting, reaching a maximum height of 1.5 m. Glenrosa soils supported tree communities and chromolaena reached more than 3 m in places. The density of chromolaena affected species composition in grasslands with moderate to dense stands (> 5 adult plants m ¯² or >50000 shrubs ha ¯¹). Chromolaena stands became monospecific when the number of adult plants exceeded 7 m ¯². Succession to forest also ceased once chromolaena became thicket-forming. Fire-induced mortality of the chromolaena depended on grass fuel loads. Grass cover of 30% (c. 1 000 kg ha ¯¹) was required to achieve 80% mortality of the parent infestation after the initial burn. Dense infestations could only be killed by running head-fires from adjacent grasslands into thickets. Under conditions where head-fires could not be used, infestations were slashed and burnt at the height of the dry season (July to August) to achieve an 80% kill rate. Seedlings were killed (99%) by annual burning in sparse (≤ 10000 shrubs ha ¯¹) to moderate < 50 000 shrubs ha ¯¹) infestations. The suppression of chromolaena and other alien species, establishing on bare ground after clearing dense infestations, required chemical control until grass cover was sufficient (i.e. 1 000 kg ha ¯¹) to effect uniform burning. Certain secondary alien invaders (e.g. Lantana camara, Psidium guajava and Solanum mauritianum) persisted by coppicing profusely after fire and herbicides need to be integrated into burning programmes when these species occur. Grasslands on regic sands (e.g. Ischaemum fasciculatum, Panicum dregeanum and Themeda triandra) were more resilient to the modifying effects of woody vegetation, than grasslands on Glenrosa soils. Grasslands on Glenrosa soils did not revert to an open state but persisted as ruderal savanna grassland (e.g. Eragrostis curvula, Hyparrhenia tamba and Cymbopogon validus) once fire-resistant tree species (e.g. Combretum molle and Heteropyxis natalensis) had established. Depending the objectives for land management and the vegetation's condition, coastal grasslands can be rehabilitated and managed in multiple states, i.e. grassland, savanna or forest communities. A state-and-transition model based on the empirical data recorded in the study is presented and shows chromolaena altering vegetation states from open grassland to chromolaena dominated thicket. The model illustrates chromolaena thickets as the dominant phase of a moist coastal forest/savanna succession, irrespective of soil type, in absence of appropriate land management practices (e.g. control burning and integrated control of alien vegetation). This model should aid in planning strategies for the control of chromolaena in subtropical grasslands in South Africa. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 2000.

A series of trials and investigations were implemented to address concerns surrounding the dynamics of the fire-climax wooded/edaphic grasslands within the iSimangaliso Wetland Park, northern KwaZulu-Natal, South Africa. The research problem surrounded inadequate historical evaluations of changes in vegetation structure, grasslands progressing to a woody dominated composition, and increases in Helichrysum kraussii (Curry bush). These were addressed as follows: Firstly, the recovery of vegetation in response to different periods of fire exclusion in different communities along a topographical gradient of a coastal dune area, was assessed over a two year period. Secondly, the regeneration after wildfire of the persistent, stress tolerant shrub H. kraussii, was studied on different catenal positions with differing fire exclusion periods and with and without defoliation of surrounding plant biomass in the coastal edaphic grasslands north of Manzengwenya, South Africa. Thirdly, aerial photography from 1937, 1975, and 2000 was georectified, digitised and analysed using a Geographic Information System to examine broad vegetation changes in response to different management regimes for a site on the Eastern Shores of Lake St Lucia and a site within the Tewate Wilderness Area. In the absence of fire, the coastal edaphic grasslands progressed to a closed canopied scrub forest within six years. An increase in fire exclusion period resulted in a decrease in species abundance, an increase in woody height, and a decrease in plant density. Richness increased initially but declined marginally with increased fire exclusion period. Higher lying east and west facing sites had a better veld condition index compared with bottom sites and had an increased response (vigour) to defoliation but were far more likely to succeed through to woody scrub forest. Woody plant biomass vigour was greater for west facing sites. Ordination of species composition across sites in response to fire exclusion and catenal position revealed greater similarities within exclusion periods than between. Bottom sites were more similar with similarity decreasing for east and west facing sites. Fire exclusion resulted in an initial increase in woody species and a subsequent increase in herbaceous species. iii Growth response of H. kraussii was unaffected by catenal position and fire exclusion period, whereas defoliation of surrounding grass tended to increase in size (P<0.05). Density and height for this species however increased with increasing fire exclusion. An increase in soil moisture negatively affected H. kraussii growth indicating susceptibility to high water tables. The number of other woody species establishing beneath H. kraussii may be due to changes in the transmission of light through the canopy where an increase in canopy diameter resulted in an increase of photosynthetically active radiation at the soil surface. The effects of fire on landscape change were investigated for the Eastern Shores and Tewate Wilderness Area, iSimangaliso Wetland Park, South Africa using aerial photography. Changes to historical disturbance regimes largely through active exclusion of fire resulted in the majority of the higher lying coastal grasslands changing to savanna scrub or closed canopied forest within 63 years on the Eastern Shores. The degree of fragmentation of these grasslands was greatly reduced within the Tewate Wilderness Area where disturbance regimes included greater frequencies of fire. Hygrophilous grasslands remained largely unaffected by woody encroachment but did not preclude woody species establishment indicating possible susceptibility during long drier periods. Frequent fires result in the maintained distribution of the higher grasslands. This vegetation type is a system which becomes resilient in response to fire, whereas in the absence of fire readily progresses to Dune Forest. The coastal grasslands above the high water table are therefore highly unstable and transformed easily in the absence of regular disturbance. It would appear that a threshold of approximately six years exists, after which substantial management intervention may be required to reverse the succession back to grassland. The growth of H. kraussii was unaffected by fire and remained persistent irrespective of fire exclusion period. An ability to attain size (height and canopy diameter) was limited with increased soil moisture but density was reduced through regular burning. Frequent fires are necessary to reduce density of H. kraussii and reduce the competitive advantage gained with age. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2007.