The vegetation of the Mgeni Estuary, Mhlanga Estuary and Mdloti Estuary was analysed according to the Braun-Blanquet phytosociological method using quadrats of 4m2 and 25m2 in area placed subjectively and the Point-Centred Quarter method with points chosen whilst traversing. A classification of plant communities is given. These are described floristically and related to habitat variables. Indicator species of the climatic climax of coast forest occur within the study area. The absence of coast forest is ascribed to low altitude, tidal inundation and the consequent effect of salinity, basal inundation resulting in a high water-table and the influence of man. The floodplain of the Mgeni Estuary is dominated by mangrove vegetation comprising mainly Bruguiera gymnorrhiza and Avicennia marina whereas vegetation established on Athlone Island is dominated by mesophytic thicket comprising mainly Schinus terebinthifolius, Lantana camara, Chromolaena odorata and Cardiospermum grandiflorum. The floodplain of the Mhlanga Estuary is dominated by Phragmites australis. The shores of the Mdloti Estuary are dominated by Barringtonia racemosa, Phragmites australis and Echinochloa pyramidalis. Major differences in vegetation patterns of the three systems are related primarily to the differences in the open nature of the river mouths. This is controlled mainly by river flow and longshore drift. Differences in vegetation patterns within an estuary are dependent on differences in tolerances to salinity, basal inundation and shade, together with variations in altitude and edaphic factors and competition between species. Generally soils of Mgeni Estuary had higher contents of small sized fractions, bulk densities, reserve acidities, organic matter, salts and exchangeable bases and lower pH than soils at Mhlanga Estuary and Mdloti Estuary. Differences exist between mangrove and non-mangrove soils at Mgeni Estuary and differences between the non- mangrove soils at the three study sites. Major differences in soil characteristics are as a result of differences in tidal inundation, geogenetic parameters and biotic factors. Information on topography, hydrology, geology, climatic factors, biotic factors and historical background of the area is given. A check-list of vascular plants is included. The work is illustrated by 44 figures. / Thesis (M.Sc.)-University of Durban-Westville, 1986.

The overall aim of this study was to develop an understanding of the vegetation ecology of the lower Mkuze River floodplain from a landscape ecology perspective. The lower Mkuze River floodplain and its associated wetlands are located east of the Lebombo Mountains and north of Lake St. Lucia on the Maputaland Coastal Plain in northern KwaZulu-Natal. This system is defined as a storage floodplain wetland and comprises a mosaic of different wetland types. In addition it has a complex history of resource use and management. Landscape ecology proved to be an ideal theoretical framework for this study because it enables the examination of complex ecological processes and phenomena in an integrated and holistic manner. It achieves this by explicitly recognizing the spatial heterogeneity, dynamics and hierarchical organization of the landscape; concepts that proved useful in developing an understanding of the ecological patterns and processes operating within the lower Mkuze River floodplain. The vegetation of the study area was classified, using multivariate techniques, into six plant communities. The distribution of these plant communities was correlated with underlying environmental gradients that summarized the interactions between hydrology, substrate properties and topography within the floodplain system. Within the study area the Phragmites mauritianus reed swamp community was found where there was slow moving water, in semi- to permanently saturated soil. This was usually around the edges of pans or in extensive stands in low-lying areas in the distal reaches of the floodplain. The Imperata cylindrica hygrophilous grassland community was uncommon and was found in isolated stands towards the edge of the region of seasonal flooding. The Echinochloa pyramidalis backswamp community was the most extensive of all the plant communities identified. This community was tolerant of flooding and was found in damp places such as seasonal pans, backswamps and riverbanks as well as in standing water. The distribution of the Ficus sycomorus riparian forest community was restricted to elevated levees adjacent to the river channel that experienced inundation when floods were large enough to overtop channel banks. The Cynodon dactylon floodplain community was generally found towards the floodplain-terrestrial upland boundary in elevated areas with sandy well-drained soils. The Acacia xanthophloea woodland community was distributed on the floodplain margin in elevated areas on sandy soils, primarily fringing the linear pans draining towards the Mkuze River from the north. The description of the plant community types and the underlying environmental determinants of their distribution provided a useful foundation for the examination of ecological processes and phenomena operating at spatially coarser levels within the landscape hierarchy. Plant communities were aggregated into functional types based on criteria such as exposure to similar flooding and sedimentation regimes. The identification and mapping of these functional types, using a Geographical Information System (GIS), enabled one to identify a hydrogeomorphic continuum that described the interaction between floodplain processes and vegetation distribution. Within the study area the proximal-seasonally inundated functional type comprised plant community types found on channel levees and within backswamp areas. These areas were functionally connected to the Mkuze River in that they were exposed to seasonal flood events and associated sedimentation. The distal-permanently inundated functional type was typically found in the lower reaches of the floodplain that were rarely exposed to hydrological and sedimentological inputs from the Mkuze River. This functional type was permanently inundated and characterized by standing water and/or permanently saturated soils that were generally associated with the large floodplain pans. The distal-infrequently inundated functional type was typically located in sandy areas along the southern distal reaches of the floodplain. These areas were infrequently inundated by overbank floodwaters from the Mkuze River and were not characterized by substantial clastic sedimentation. The distribution and interaction between these functional types made it possible to develop process-based understanding of the ecosystem patterns and processes operating within the lower Mkuze River floodplain. Landscape ecology theory emphasises the importance of a temporal analysis of spatial heterogeneity and the role of disturbance in ecosystem patterns and processes. Therefore a temporal analysis of the landscape mosaic from 1937 to 1996 was undertaken, using a GIS, in order to quantify landscape change over time. The landscape characteristics utilised to examine this change were total category area, percentage contribution to the total landscape area, number of patches, mean patch size, median patch size, patch size standard deviation and the mean perimeter-area ratio. These spatial statistics were calculated for each year using PATCH ANALYST, an ArcView GIS extension and they were used to illustrate the role of anthropogenic disturbance on the landscape mosaic at a variety of levels within the landscape hierarchy. Anthropogenic disturbance was found to affect landscape content and configuration and therefore had the potential to undermine the underlying environmental determinants of landscape patterns and processes. Once the underlying functional processes are undermined, irreversible ecosystem degradation is a possible outcome. The examination of the different levels within the landscape hierarchy and the dynamics of ecosystem patterns and processes operating within the Mkuze River floodplain made it possible to develop deeper insights into ecosystem patterns and processes than a conventional vegetation ecology study that typically focuses primarily on plant community classification. The use of landscape ecology as an overarching theory that guided the research process and aided the interpretation of findings by explicitly recognising the importance of examining spatial heterogeneity, hierarchical organisation and dynamics, proved invaluable in developing process-based understanding of the lower Mkuze River floodplain. / Thesis (M.Sc.)-University of Natal, Durban, 2001.

The impounding of the waters of the Pongolo river, upstream of its floodplain on the Mocambique coastal plain, may adversely affect the functioning of the floodplain system. A multidisciplinary study of the functioning of the floodplain was initiated to provide a basis for the development of a management strategy for the floodplain. The study reported in this dissertation considered the flood dependence and functioning of the vegetation of the seasonally inundated area. The vegetation was mapped and the communities ordinated, according to the Braun-Blanquet technique, in relation to their positions relative to high flood level (HFL) and the level of the water after flood subsidence (i.e. maximum retention level, MRL). Community distribution was shown to be strictly determined by both the height of the floods and by the MRL. It was concluded that periodic floods are essential for the maintenance of the communities. The Cynodon dactylon (L.) Pers. Community, which forms extensive meadows in the zone of periodic inundation, was studied in detail. As the water level receded during winter, productivity was high (up to 23 kg ha¯¹ d¯¹1 dry mass) and a palatable sward was produced. This is heavily grazed by domestic stock, but as the soil dries out and water stress becomes significant, production decreases, C. dactylon becomes less palatable, and grazing shifts to newly exposed areas. The shift in grazing allows the build up of a large standing crop of both grazeable and ungrazeable (below ground and stolons) material. At the time of inundation by the next floods c. 910 kg ha¯¹ of dry mass, c. 17 kg ha¯¹ nitrogen and c. 2 kg ha¯¹ phosphorus have been removed by grazers. It is concluded that this production, which is flood dependent, forms an important supplement to stock grazing during winter. It is suggested that this source of grazing could be stimulated by irrigation during winter. Cynodon dactylon is shown to decompose rapidly during inundation, losing half of its mass and nutrients in c . 28 days. It therefore represents a major energy and nutrient input during the aquatic phase. The extent to which it is grazed during submergence is unknown. The nutrient input is derived ultimately from the soils of the inundated areas and, since nutrients are being removed by both terrestrial grazers and flushing, continued production is reliant upon the annual sediment load reaching the floodplain. Most of the sediment load will now be deposited in the impoundment, and fertilization may be necessary to maintain productivity. The response of C. dactylon to the seasonal fluctuations in water level are used to formulate proposals for water release from the dam. These include proposals for the short-term, i.e. until the demand for irrigation water conflicts with the requirements of the floodplain, and for the long-term, when less water will be available for the floodplain / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1981.