Bibliography: pages 792-808. / Alien species known to be invading untransformed ecosystems in southern Africa, and, more particularly, those inside nature reserves, were identified. The extent and ecological impacts of these invasions were assessed. Their control within reserves was also evaluated. Research approaches used were; literature review (which included an international review), a detailed questionnaire survey of alien plant invasions in 307 reserves, rapid field surveys of 60 reserves, intensive case studies of four reserves (Cape of Good Hope Nature Reserve, Hluhluwe-Umfolozi Game Reserve, Kruger National Park, Pella Fynbos Research Site), international comparison with case-study reserves in other savanna and Mediterranean-type biomes, and field evaluation of control methods for alien plants in the two fynbos reserves. Ecological impacts of alien invasions throughout the subcontinent were determined from historical changes in vertebrate populations, including detailed studies of three native birds (Bostrychia hagedash, Lybius leucomelas, Ploceus velatus) expanding their ranges, partly in response to the spread of invasive alien trees. The results are presented in eight chapters, comprising 26 published (or submitted) papers, an introduction and a concluding summary. One chapter covers contributions to the theoretical understanding of invasion processes, including a prediction of their interaction with rapid global environmental change. The conclusion is reached that alien invasions pose a serious challenge to nature conservation in the region. Mostly this comes from alien woody plants but the importance of herbaceous plants has possibly been underestimated regionally. Introduced mammalian pathogens and predatory fishes have also had important effects. Alien invertebrates have been poorly studied (the ant lridomyrmex humilis poses a significant threat). Alien terrestrial vertebrates have generally had only localized effects. Alien plant invasions affect all biomes, with riparian ecosystems being regionally threatened. Mesic biomes and habitats are usually more invaded by alien plants than xeric equivalents. 281 alien vascular plant species were recorded invading vegetation within nature reserves (an average of 12 species per reserve) with an additional 200 species being possibly present but unrecorded (an average of 18 species per reserve). By 1984, the 54 plant taxa recorded invading reserves most frequently were estimated, on average, to be present in 30% of the 11cm x 11cm grid cells of the reserves they were invading. The average potential future extent of these invasions was estimated to be 51%. Control had, on average, been initiated for two plant species in each reserve and 18% of these operations had already resulted in complete eradication. Reported control costs were particularly high for the woody plants which pose a serious threat to the highly endemic flora of the fynbos biome, on average R48 284/reserve (R1,8/ha = US Dollar 1,2/ha) in 1983. These high costs were validated experimentally. A computerized optimization model, aimed at minimizing the costs of controlling the most intractable shrub invader of fynbos, Acacia saligna, was developed from the results of a field experiment at the Pella site. Practical field- scale control of these invasions was assessed to be feasible, using the results of repeated monitoring of permanent plots in the Cape of Good Hope Nature Reserve. Control strategies and regional priorities, based on the theoretical and practical insights gained from this study, are proposed. Even though the intensity of invasions is likely to increase in the foreseeable future, in part as a result of rapid man-induced changes in global climate, it is predicted that these invasions can be controlled if the correct approaches are adopted timeously. Failure to control them, will ensure that the extinction rate of native species will markedly increase and that ecosystem functioning will be altered significantly at a local scale and, conceivably also, at a regional scale.

Trees alter the environment by affecting microclimate, soil properties and adding structural complexity to an otherwise single stratum system. It is known that the herbaceous understory of Savanna ecosystems is most productive at intermediate tree density, where the shade and nutrients from the tree are more influential than the competition with the tree. Tree density varies in savannas due to disturbances, such as fire and herbivory. Yet, disturbances do not only alter density but can affect entire tree structure. It has been found that chronic ungulate browsing influences whole tree size, as well as canopy size and shape. Giraffe in south of Kruger National Park browse heavily on Acacia nigrescens found on basalt-derived soils. I studied the influence of trees above the giraffe browse trap (>6 m) as opposed to trees in the browse trap (< 6 m) on savanna ecosystem properties. I compared grass biomass, soil moisture, soil organic matter, soil carbon and soil nitrogen beneath tall trees vs. short trees, as well as in the open vs. underneath trees. Ecosystem properties were estimated and compared between a giraffe impacted to a giraffe absent area. Giraffe impacted area overall yielded lower grass biomass, soil moisture, soil nitrogen and soil carbon. The consequence is a trophic cascade of events from increased giraffe decreasing fires and in tum increasing tree density and hence carbon sequestration. As giraffes replace a single large tree by several small trees, shows that their population size is key determinant of ecosystem properties.

Donkeys are used extensively in Namaqualand and other rural areas in South Africa for pulling carts, carrying loads, ploughing and threshing grain. There is, however, an ongoing debate among farmers, agriculturalists and the wider population about the implications of keeping donkeys, particularly on communal lands. Donkeys are reported to cause more damage to the veld than other animals, and it is commonly said that one donkey eats as much as seven goats. This project was initiated as a case study to investigate the impacts of donkeys in a communal area in the succulent shrublands of Namaqualand, and to generate suggestions for the management of donkey populations. The degree of competition between goats and donkeys was investigated by predicting daily food intake using a model based on ruminant and non-ruminant digestion, which takes into account the animal's body weight and digestibility characteristics of its diet. The ratio predicted food intakes is discussed in the light of the degree of dietary and habitat overlap, which were obtained from field observations. Habitat use was scored using dung frequency transects, and feeding was directly observed. It was found that in the wet season, one donkey eats as much as 5 / 7.6 goats (donkey weight set at 200 and 250 kg respectively). In the dry summer, this ratio increases to 7.6 / 8.6 because in ruminants, the passage of food through the gut slows down as the digestibility of the diet decreases. Habitat overlap is concentrated in the riverine areas and sandy pediments, which make up 15 % of the landscape and are strongly favoured by donkeys. No significant differences were found to exist between the intensities of goat use of the different habitats. The botanical composition of the diets of donkeys and goats were found to be similar, but there is a great difference in the quality of the diet consumed by each species as goats are able to select plant parts of a higher quality. The implications of this are that whereas a donkey takes in as much dry matter as 5 - 8.6 goats, the diet of this food is not high enough to satisfy the dietary requirements of goats, and hence fewer goats than predicted can be supported for every donkey that is eliminated. Donkeys are commonly reported to have destructive feeding habits; little of this was directly observed, but donkeys were found to bite deeper into the wood of shrubs, which may damage the plants in the long term. There are two main management measures that should be applied. The feral donkey population in Paulshoek, estimated at 50-100 animals, must be eliminated - even at a conservative estimate, these animals could be replaced with 250 goats. However, one must look beyond the number to the processes taking place to make management effective. Even if one donkey can be replaced with fewer than 5 goats, the vegetation is relieved of the indirect effects donkeys have through their feeding habits. Control of working donkeys should focus on the riverine and flat, sandy areas, as they are potentially of high nutritional importance (especially the riverine areas which support grass throughout the year), and which are also the most susceptible to erosion due to their sparse vegetation cover. These areas are small and valuable, and plans should be made to protect some areas through the growing season.

Protea witzenbergiana possesses some features common to that of therophilous Proteas in the Cape Floristic Region, such as a cup-shaped, downwards-hanging inflorescence surrounded by red involucral bracts, which emits a yeasty, musky odour. The flowering season occurs in the winter and the plants occur in small, localized and isolated stands in the Witzenberg fynbos. However, the species has never been investigated as to whether it is pollinated by small mammals. This study investigated whether the plant is rodent-pollinated, and if so to what extent by various species. Fieldwork was carried out over three days of live trapping, during and after the flowering season. Exclosure experiments were set up in order to assess seed set. Mammal droppings were collected, processed and pollen grains were counted. All small mammal species excepting one (Otomys irroratus) contained pollen grains in their faeces. Oendromus melanotis, Mus minutoides and Aethomys namaquensis were found to be the most important pollinators. The small mammal community showed a distinct change in size and composition after flowering season had ended, suggesting that the availability of nectar resources may be an important food supply to small mammals in the area. Implications of these findings could be relevant for conservation and co-evolutionary studies.

The subtidal seaweed communities of the KwaZulu-Natal coast are poorly understood. A recent collaborative research project between Belgian and South African phycologists, however, is currently investigating the diversity and biogeographical affinities of the marine benthic algal flora of this region. The newest leg of the project is focusing on correlating biogeographic patterns with abiotic factors in the area between the St. Lucia estuary and Sodwana Bay, where it has become evident that there is a huge turnover in species composition. The present study examines seaweed community change along a depth gradient (1-30 m depth) at Sodwana Bay on the KwaZulu-Natal (KZN) coast, South Africa. All seaweed cover within 25 x 25 cm quadrats was collected, identified and weighed. The environmental variables, depth and % sand cover, were also measured. Both biomass and presence/absence data were ordinated to show community differentiation related to the environmental variables. The seaweed communities are largely red turfs, with a total of 82 Rhodophyte, 14 Chlorophyte, and 8 Phaeophyte taxa recorded. These turf communities showed exceptional diversity with 105 taxa (>25 % of recorded KZN flora) occurring in the total sample area of 1.56m². Seaweed communities clearly change with depth, but the biggest change occurs between the shallow subtidal zone (approx 1 m depth) and deeper waters (>5 m depth), mostly likely due to the effects of wave action; thereafter communities do change with depth, but the effects are less pronounced. Biomass (p=0.049, significant) and seaweed diversity (p=0.078, not significant) were both shown to decrease, and % bare (seaweed-free) cover (p=0.002, significant) increased, with increasing depth. It is likely that these changes with increased depth can be attributed to the effects of increased light attenuation on ecological functioning in the subtidal communities. Sand inundation patterns are shown to be important in determining seaweed community structure, but do not appear to be predictable along a depth gradient (p=0.544, not significant).

1. Two species of seed feeding bruchid, Algarobius prosopis and A. bottimeri, were introduced to South Africa in 1987 and 1990 respectively, to curb the spread of the invasive weed Prosopis. Failure of A. bottimeri to establish and reduced effectiveness of A. prosopis due to interference by livestock resulted in the release of a third species, Neltumius arizonensis in 1992. 2. Algarobius prosopis populations have remained high but N. arizonensis has not been as successful. 3. Populations ofboth bruchids showed similar emergence patterns through the sampling period hence there is no temporal partitioning of resources. 4. Percentage emergence of N. arizonensis was low at the start of the season but increased as the season progressed with higher proportions of N. arizonensis emerging from 'tree' pods than 'ground' pods. Regardless of the spatial partitioning of resources by the two bruchids, N. arizonensis emergence numbers are still far lower than those of A. prosopis (approximately 1:4 respectively). 5. Oviposition strategies of N. arizonensis results in high levels of egg parasitism by Uscana sp., Trichogrammatidae, that A. prosopis escapes by ovipositing in concealed places avoiding detection by parasitoids. 6. Egg parasitism is extremely high on N. arizonensis eggs (67%) and has resulted in failure by N. arizonensis to establish large populations and hence its effectiveness as a biocontrol agent is negligible.

The habitat of an animal is extremely important as it provides that animal with the necessary resources for fulfilling its life-history requirements (Brennan & Block, 1993; Beyer et al. 2010). A habitat is defined as a region in environmental space which comprises of multiple abiotic and biotic variables influencing an animal's location (Krausman, 1999; Beyer et al. 2010). Animals tend to utilise discrete areas within a habitat, constituting part of their home range. Home range analysis helps to delineate the area used by an animal habitually and areas of concentrated use (Samuel et al. 1985; Seaman & Powell, 1996; Moorcroft et al. 1999; Mitchell, 2007; Rodgers & Kie, 2011). Patterns of differential use of space within an animal's home range are the result of competing demands and trade-offs. In this study, a kernel technique was used to determine the home range of four satellitetracked groups of Southern Ground-Hornbills Bucorvus leadbeateri in the Associated Private Nature Reserves (APNR) in the South African lowveld. Satellite data were analysed in ArcGIS® 9.3 to quantify habitat selectivity by groups of ground-hornbills at different times of the year to determine a) favoured habitat types, and b) the resolution with which they perceive their environment. Each of the I groups showed variation in the utilisation and extent of their home ranges on a seasonal basis. Home range sizes contracted towards the nest during the summer breeding season (December to March) and expanded during the dry season (April to September). Within the home range of one of the groups the physical characteristics of habitat types (i.e. vegetation types) were sampled at 250 random co-ordinates in order to assess whether habitat preference at the meso-scale can be explained by the physical attributes of that vegetation type. By profiling and quantifying the vegetation of areas in the home range that are used by Southern Ground-Hornbills to differing degrees, this information could be used as a proxy to facilitate re-introduction efforts, by providing a tool to identify optimal landscape configurations.

Little is currently known about the spatial ecology of the endangered African penguin during the non-breeding season. As foraging success during this period is critical to adult survival, this project explores their dispersal patterns, as well as the degree of overlap with commercial purse-seine fisheries which target the same prey. African penguins from two colonies (Bird Island, Algoa Bay, south coast and Dassen Island, west coast) were tracked over 3 non-breeding seasons using Platform Terminal Transmitters and Global Positioning System devices to explore their pre- and post-moult dispersal patterns. Dispersal trips varied greatly between individuals, but there were still significant differences in the dispersal patterns found between islands and between life history-stages at Dassen Island. Bird Island penguins tended to make frequent, short trips, returning to their breeding colony in between, whereas Dassen Island penguins followed one of two strategies: either making few (up to 3) very long trips (up to 4000km), or adopting a central-place foraging pattern based at a point along the coast, distant from their breeding colony. At Dassen Island a clear spatial divide is evident with most pre-moulters dispersing south whereas most post-moult birds dispersed north, off the west coast. Bird Island penguins all moulted at their breeding colony and seldom visited either the adjacent coastline or other islands. During pre-moult, Dassen Island birds often spent nights close to shore along the coast or at other colonies and about a quarter of them moulted at another colony, most often at Stony Point, on the mainland east of Cape Point. Five dispersal patterns (defined in terms of the direction and distance of the furthest point reached) were identified for birds from Dassen Island and there is some indication that different dispersal patterns have distinct survival implications. Over 60% of PTT loss (a proxy for penguin mortality) occurred in a small area in the northern part of St Helena Bay, offshore from Lambert's Bay and Strandfontein and could be due to high levels of seal predation, which has been previously documented in the area. Commercial small pelagic fishing data were analysed over three years (2012-2014). Average fishing effort decreased each month from September to December (when hardly any fishing occurred in any of the three years) and there was much greater fishing effort along the west coast than on the south coast. Commercial fishing effort was compared to penguin track data to assess the degree of potential overlap between these two pelagic predators. Firstly, the spatial overlap between core penguin non-breeding dispersal areas and core fishing areas was assessed on a broad temporal scale (three years). Bird Island penguins were found to be more spatially removed from fishing activity with no overlap in core fishing and dispersal areas on the south coast. Compared to the size of their dispersal area, Dassen Island penguins which stayed along the lower west coast showed the highest overlap with core fishing areas. The two most common dispersal patterns exhibited by Dassen Island penguins both involve journeying to areas outside the range of purse-seine fishing, indicating the possibility of avoidance behaviour once the pressures of breeding are lifted. Direct (daily) overlap between commercial fishing effort and penguin tracks was studied although the low frequency of the penguin position data restricted the analysis to a 20 km scale. The results were similar to those described above, with most direct overlap occurring between Dassen Island and Gansbaai.

The increasing population and urbanization occurring in South Africa is threatening the survival of many environments, particularly those in close proximity to urban areas. Present legislation does not adequately protect the environment, and methods are needed to collect data to guide the development of conservation-orientated management and developmental policies. These data should provide a general description of the environment, Indicating sensitive areas, while taking physical, socio-economic and natural features into account. To be most meaningful, these data should be stored in a survey information system which is accessible to any person or institution involved in management or developmental policy formulation. Chapter 2 describes some of the components of such a system. Chapter 3 discusses the data collection component of the survey Information system. The use of vegetation features to describe the natural environment, and the use of land-use features for the socio-economic environment, are emphasized. Numerous methods are available to collect vegetation and land-use data, but for management purposes the methods should be elementary enough to be used by non-specialists while still supplying meaningful data. Aerial photography can be applied to vegetation and land-use studies. However, numerous inaccuracies can occur when using aerial photography as a source of data and Chapter 3 also discusses some of these limitations. The study area used to test proposed methods of collecting vegetation and land-use data for a management information system are described in Chapter 4. The study area consists of eleven study sites, each at a river mouth on the False Bay coast.

Includes bibliography. / This study aimed to describe patterns of distribution in terrestrial Afrotropical birds, to investigate the causes of these patterns, and examine how aspects of distributional patterns may be used to prioritize local regions for conservation attention. Presence-only data were gathered and digitized at one-degree square scale for 1686 terrestrial bird species that breed on or regularly visit sub-Saharan Africa as non-breeding migrants. Biogeographical analysis of the 1437 species that are globally restricted to sub-Saharan Africa (Afrotropical endemics) revealed a suite of geographical areas that have a homogenous and characteristic avifaunal composition, termed avifaunal zones. The approach used in this study ensured representativeness in the resultant biogeographical classification scheme, which was not biased towards avifaunas that are species rich or that contain many narrow endemics, and further included avifaunas that consisted of few, but taxonomically and ecologically distinct species (e.g. the Namib Province). Analysis of zonal boundaries exhibiting high levels of turnover, defined specifically as species replacement, were distinguished from zonal boundaries that are characterised by species richness gradients. For instance, the northern forest-savanna boundary between the Guineo-Congolian and Northern Savanna Subregions was shown to consist of a sharp ecotone between forest and savanna, whereas the boundary between the Northern Savanna and Northern Arid Subregions was shown to be dominated by species drop-outs. This shows that whereas the Northern Savanna Subregion represents a unique avifauna that is distinct from that of the Guineo-Congolian Subregion, the Northern Arid Subregion is merely a depauparate subset of the Northern Savanna avifauna. Patterns of species richness and narrow endemism where shown to differ between species groups that exhibit different life history characteristics (e.g. residents vs. migrants) and distributional characteristics Atrotropical endemics vs. nonendemics). Differences can probably be attributed to island biogeography and aerography theory.