This work continues the marine tradition of the Department of Zoology but attempts to open-up a new field by tackling for the first time the ecology of the shallow sea bed. False Bay is accessible and provides a large area that is virtually unaffected by rivers or the works of man. Its situation happens to be particularly interesting. It is the nearest part of the flat, submarine shelf called the Agulhas Bank, where there are valuable fishing grounds; and it lies at the boundary of two very different masses of water. Study of False Bay should, in addition, be of interest in throwing light on the contrast between the shallows of the sea and those of lagoons and estuaries. Field work started with dredging from hired fishing boats. It was soon obvious that one of the most important regions of the sea was escaping notice, that above the shallowest depth in which dredges could be worked and below low tide level. This region could only be studied by diving and so a programme of diving with frogman kit was started. Dredging and diving meant considerable training in non-academic techniques not only in the field but in designing new gear (eg. dredges, underwater camera and underwater electronic flash).

Only in 1954 the notes of late Prof. Moss on the genera Arthrocnemum and Salicornia were published, and although this was a big step forward in the classification of these genera, it proved to be unsatisfactory in many cases. This, however, should not throw the work on that subject done by him into shade, as shows his earlier works (Moss, 1910, 1912, 1914), and only his notes published by Prof. Adamson in 1954 were unfortunately fragmentary. The revision of the genus Arthrocnemum is taken up again, as so many salt marshes on the Cape Flats are drained and the rivers chanalized, so that the number of localities are increasing rapidly, and fresh material is essential for this work. The aim of this work is to cover as large a field as possible, but always with the eye on some further evidence for the taxonomy of the species. Unfortunately, only a few species which are growing on the Cape Peninsula and some fresh material of A. affine from Swakopmund was available for this study. This gives perhaps a bit one-sided view, but indications of similar problems in other species or of difficulties with explaining phenomena has been given for later research. Investigations were mainly done on fresh material, and dried herbarium specimen were usually only consulted for confirming certain characters found in fresh. To aid this it was made use of photos where ever possible, but it was found sometimes almost impossible to get clear pictures, as the flowers are minute and inconspicuous. For convenience, the whole treatise has been divided into three main parts: (1) The ecological part (autecology) in which the zonation phenomenon at Milnerton has been emphasized, but at the same time it gives an account of the particular habitat each species requires. (2) A brief account of the anatomy especially the abnormal secondary growth in the stem and root has been given, and a discussion of the origin of the fleshy segmented branches follows. (3) Lastly the history and the distribution of the whole genus Arthrocnemum, and a treatise of the classification and its difficulties of the species of the Cape Peninsula is produced. This work is purely preliminary, and it is tried to get access not only to the problematic taxonomy, but to an understanding of these plants as such. The author is well aware of some generalisations or speculative assumptions, but the main point of the treatise is to get a new approach to that difficult group of plants, of which each detail seems to be worth to be recorded for later evaluation.

Non-trophic interactions are significant structuring agents of ecological communities. Knowledge of how this process drives ecosystem functioning and community structure either individually, or interactively with other processes, is however, limited, particularly in shallow soft- littoral ecosystems. At a local level, such systems are dominated by allogenic engineers such as the burrowing axiid sandprawn Callichirus kraussi Stebbing, which has important effects on macro- and meiobenthic assemblages. C. kraussi is distributed across the South African coastline, from the subtropical Mozambican border to the temperate west coast borders with Namibia. Bioturbation — the principal mechanism by which C. kraussi engineers influence associated biota in sedimentary systems, also has significant effects on sediment properties, biofilms, microalgal and microbial composition. However, theory suggests that ecosystem engineering effects are contextually dependent and contingent upon processes that are temporally and spatially variable. In South Africa, variations in background nutrient levels across the coastline is significant, with the West coast being dominated strong upwelling, which increases biological productivity. Within the west coast, upwelling is also seasonally variable, being strongest in summer. In view of this natural variability in nutrient levels across the coast, seasonal variability within the west coast and the fact that the sandprawn C. kraussi dominates across these conditions, the central focus of this PhD thesis was to investigate how ecosystem engineering by sandprawns C. kraussi and nutrient levels individually or interactively influence assemblages and ecological processes in coastal soft-sediment ecosystems. The study was carried out in Langebaan Lagoon, which is a marine lagoonal system on the west coast of South Africa that is dominated by sandprawns and subjected to seasonal upwelling that is a feature of the west coast. This thesis was based on the two principal approaches, viz. a field comparative study and in situ experiments. The field observational study investigated the responses of macroand meiofaunal communities to sandprawn bioturbation impacts between upwelling and non-upwelling seasons, with the aim of understanding how upwelling nutrient pulses modify these benthic assemblages. Benthic and water column chlorophyll-a (chl-a) levels were also measured, while meiofaunal communities were investigated within sandprawn burrows (burrow-walls) and at the sediment surface. It was hypothesized that chl-a levels, and community metrics would be lower in winter (non-upwelling) but increase in summer upwelling season due to nutrient pulses. It was also hypothesized that community metrics would be negatively correlated with sandprawn density due to bioturbatory effects (sediment turnover) in winter (non-upwelling) but this relationship would become neutral or positive increased in summer, due to increased productivity offsetting negative bioturbative effects. Clear, non-intuitive and ecologically interesting outcomes emerged from the field study. Firstly, benthic chl-a concentrations appeared to be lower in summer relative to winter in 2015, although, this pattern disappeared in 2016. In contrast, water column chl-a concentrations within the lagoon channel conformed to the posed hypothesis, being consistently greater in summer relative to winter. Even though both macro- and meiofaunal assemblages differed significantly between seasons, the hypothesis that community metrics would be greater in summer relative to winter season was not overwhelmingly supported by these findings. Pearson correlation analyses revealed that sandprawn effects were generally weaker within seasons but stronger when winter and summer data were combined and investigated per year. In terms of the latter, most community metrics and benthic chl-a levels were generally negatively correlated with sandprawn abundances, but this varied seasonally and spatially. Overall, results of the comparative study suggest that increases in water column chl-a levels do not necessarily translate into increases in benthic chl-a and community metric levels. One possible reason for this is that increasing bioturbation by C. kraussi overrides nutrient enrichment effects. This conclusion though is spatially dependent, given that results of correlation analyses were site-specific. The factorial field experiment employed in this dissertation manipulated nutrients (fertilizer capsules, Plantacote N: P: K) and sandprawns (C. kraussi densities) to investigate their individual and interactive effects on benthic assemblages. I hypothesized within the context of the grazer-reversal hypothesis of Proulx and Mazumder (1998) that, responses of diversity metrics should alter from a linear decrease at the ambient nutrient level to a unimodal hump-shaped response at the intermediate nutrient level and then to a linear increase at highest nutrients. Generally, emerging patterns for meiofaunal diversity metrics and individual morphotypes were inconsistent with the prediction of the grazer-reversal model. Instead, responses were of an increasing and decreasing nature. The outcomes of this investigation, however, revealed that meiofaunal community structure at the sediment surface was significantly affected by the main effect of sites and sandprawn densities, while within burrow-walls, the interaction between sites and nutrients, but also sandprawn densities alone, significantly affected meiofaunal community structure. Overall, meiofaunal results of this investigation showed limited support for the grazer-reversal hypothesis. Although the findings of this investigation also indicated limited support for the grazer-reversal hypothesis on the macrofauna community, interactive effects of nutrient enrichment and sandprawn densities emerged to influence the macrofaunal community structure and abundances of certain individual species/taxa. Generally, macrofaunal community metrics exhibited both increasing and decreasing patterns in response to experimental treatments, however, in some instances unimodal hump-shaped patterns emerged. There was evidence of macrofaunal functional groups (i.e. suspension feeders) conforming to the hypothesis posed, but this was site-specific. Overall, I conclude that an interplay between nutrients and sandprawns does not exert strong influences on the benthic communities of Langebaan Lagoon. However, sandprawn ecosystem engineering, mainly in the form of bioturbation overrides nutrient enrichment effects in regulating benthic assemblages.

Extralimital giraffe introductions in the south-western region of Kgalagadi Transfrontier Park, South Africa, may result in changes to the vegetation structure for the region through overutilization of two keystone tree species in the area. In the areas where giraffe density was highest, I determine the relative abundance of giraffe; their diet preference and the potential reasons for this preference. Using both spoor and dung counts as indicators of giraffe presence, transects were carried out over approximately 20 km of the Auob River. Percentage occurrence was plotted onto a map of the river, along with the location of boreholes that supply the only surface water. Little trend could be seen, although occurrence appeared slightly higher near boreholes. The two trees that most commonly occur in the Auob duneveld, Acacia erioloba and Acacia haematoxylon, were examined for nutritional content, and both chemical and physical defences - all of which are thought to affect herbivore selection. Stable isotope analysis and linear mixing model were used to determine the proportion of each species in the diet of giraffe using the leaves of both species and giraffe dung. The majority of the giraffe's diet consists of A. haematoxylon (mean = 79 %, s.d. = 20.5 %) despite having less crude protein and similar condensed tannin content. Relative lack of physical defences appears to encourage giraffes to exploit A. haematoxylon, which has shorter, weaker thorns. Without management intervention, A. haematoxylon could suffer selective mortality through the impacts of giraffe browsing, resulting in a loss of species and structural diversity of the landscape.

Aloe dichotoma populations differ in their rates of recruitment and mortality across the species distribution in South Western Africa. Little is known about how recruitment and mortality rates are affected by local climate and ecological factors. Using repeat photography and a statistical recruitment model, the recruitment history of an emerging population of over 200 A. dichotoma individuals near Kenhardt was simulated. Historical rainfall data was used in order to assess the possible link between rainfall and recruitment. Recruitment patterns were found to be a combination of episodic events of varying length and magnitude superimposed over a background of low-rate continuous recruitment. The largest, most extended episodic recruitment period peaked in the mid-1970s and led to the establishment of most of the individuals in the Kenhardt population. These trees were generally between 150 and 250 cm in height (2013). Newly discovered juveniles below 100 cm in height were predominantly the result of relatively small and brief episodic recruitment events that occurred in the mid and late 2000s. Between episodic recruitment events, evidence for low-rate continuous recruitment is present, particularly in more recent years. Episodic periods of high-rate recruitment were found to be linked to high rainfall. The largest, most extended recruitment period coincided with the exceptionally high rainfall of the 1970s. The more recent episodic recruitment events also coincided with high rainfall spikes during the 2000s. While correlations between rainfall variables and recruitment were significant, many other factors may affect the strength of this relationship.

Widdringtonia cedarbergensis is a critically endangered conifer restricted entirely to the Cederberg Mountain range in the Western Cape of South Africa. This study aimed to assess whether contemporary climate change could be a driver in the recently documented tenfold decline of W. cedarbergensis in the last 40 years. By building distribution models using Maxent bioclimatic modelling for pre- and post- 1970’s point occurrence data, the distribution of the species pre- and post-1970’s were compared and assessed for significant change. Models predicted a clear range contraction in W. cedarbergensis from pre- to post- 1970. The present model predictions of suitability occur in a tight altitudinal band on the mid to upper slopes of the middle Cederberg (minimum of 1048m to a maximum of 1530m). However, the climatic predictions associate the present model distribution with relatively warmer and drier areas compared to that of the pre- 1970 model, suggesting a shift down slope rather than upslope. This is possibly due to the exaggeration of the upper limit of distribution for the pre-1970’s model that predicts W. cedarbergensis to occur as high as 1850m. This is unlikely due to the high prevalence of frost at this elevation. Though any climatic signal was complicated due to historical data error, high confidence in the present model prediction adds a valuable contribution to a body of literature that documents the decline of W. cedarbergensis and has important implications for conservation management of the species.

A fundamental goal of evolutionary biology is to understand how natural selection, random genetic drift and gene flow interact to promote adaptive trait divergence within species. Environmental gradients are ideal opportunities for disentangling the relative effects of selection and gene flow in promoting local adaptation among populations experiencing different selective regimes. In this study a multidisciplinary approach combining the methods of sensory ecology, functional morphology, population genetics and functional genetics was used to explore the relative roles of neutral and adaptive processes in the evolution of sensory divergence in Cape horseshoe bats, Rhinolophus capensis. Geographic variation in echolocation resting frequencies (RFs) in this species is characterised by increasing frequency from west (75.7 kHz: xeric habitats) to east (86 kHz: mesic habitats) across their distribution in South Africa. Here the species is found across a wide range of habitats characterised by a gradient of increasing vegetation clutter from xeric habitats in the west, to mesic habitats in the east. To better understand how selection contributes to the evolution of RF variation in R. capensis, the relationships between RF and different ecological and morphological correlates of echolocation frequency were explored.

The classical grazing lawn model is an intensely-grazed patch composed of short-statured, grazing-tolerant grass species. The formation and maintenance of these communities is controlled by positive feedbacks between grazers and the high-quality resource forage that the component grass species provide. Different nutrient cycling dynamics among the lawns identifies two discrete lawn types in the savanna: bioticallydriven lawns on nutrient-rich gabbroic soils and abiotically-driven lawns at sodic sites. We were interested in identifying whether the biotic and sodic lawns represented two distinct systems in terms of the feedback responses among herbivores, decomposers and grass and decomposer community assemblages in a mesic savanna. We sampled these components of the abiotic and biotic template of five sodic and five biotic lawns in the Kruger National Park. We used β diversity in grass and dung beetle community assemblages among the lawns to identify whether sodic and biotic lawns were distinct for grass percent cover and dung beetle species abundance. Four and three categories of lawns were identified for these traits, respectively, and placed the lawns on a gradient from biotic-like to sodic-like with a range of intermediates. Soil Na content was higher among sodic lawns but these levels did not manifest themselves in the grass foliar Na content, as for biotic lawns. Herbivore utilization of the sodic lawns was higher than the biotic lawns. Biotic lawns showed no difference in herbivore metabolic biomass between the late-wet and early-mid dry season. We concluded that the systems of nutrient cycling and lawn maintenance are distinct between the biotic and sodic lawns, but that the lawns exist along a gradient in terms of their community characteristics and abiotic features. Efforts to classify grazing lawns will present benefits in improving our understanding of their dynamics and, resultantly, the management and conservation approaches that use them to control herbivore populations in African savanna ecosystems.

The floral characteristics of small-mammal pollinated (SMP) Protea species have been assessed in a number of previous studies. This study aimed to determine whether the inflorescences of Protea canaliculata, Protea sulphurea and Protea humiflora possess these traits and are pollinated by small mammal species. An additional aim of this study was to determine whether there is a variation in pollinator efficiency of different animal species. Floral characteristics that may influence plantpollinator interactions were measured, including floral dimensions, nectar production and spectral reflectance. Live-trapping was conducted using Sherman traps and mean facial and faecal pollen load was determined for the different species caught. Furthermore pollinators were observed through footage from motion sensor cameras placed facing the inflorescences of SMP proteas. The results of this study confirmed that Protea canaliculata, Protea sulphurea and Protea humiflora are pollinated by small mammal pollinators. The evidence supporting this is that the afore-mentioned species have traits that correspond to those possessed by known small-mammal pollinated proteas including: bowlshaped inflorescences, high nectar concentrations (ranging between 24.1-42.9%), sucrose-rich nectar composition, a "yeasty" scent, floral colours that are visible to small mammals, and a winter flowering season. These proteas were found to have separated peak flowering times, providing a nectar source throughout winter for small mammals at this site. Fifty-eight small mammals of seven different species, were trapped in P. canaliculata and P. sulphurea stands over 98 hours. The average nighttrapping success was 22.7% and day-trapping success was 5.7%, indicating a relatively abundant nocturnal small-mammal population. A separation in pollinator efficiency was observed for different small mammal species, with Elephantulus edwardii identified as the most effective pollinator as it showed the greatest pollen removal (highest faecal pollen load) and spent the longest time foraging on inflorescences (±28 seconds per inflorescence). Another important pollinator was Aethomys namaquensis because it visited flowers 75% more frequently than any of the other pollinators. Camera trapping was shown to be a superior method than conventional trapping for assessing pollination by providing insight into pollinator behaviour, identifying new pollinators of 'trap-shy' species and also due to its more animal-friendly disposition.

The challenge in answering the question of whether or not plants evolve different mating systems to accommodate their respective pollinators lies in finding a pair of closely related species differing only in pollination syndromes. Furthermore, it has been hypothesized that Non-Flying Mammal Pollination (NFMP) arises from a bird pollinated ancestor as a result of their limited distributions. Liparia splendens subsp. splendens and L. parva are two genetically indistinguishable species that are thought to differ in pollination syndromes and co-occur with similar distributions, densities and have sympatric pollinators. Speculations that closely related sister species L. splendens subsp. splendens and L. parva are bird and nonflying mammal pollination (NFMP) respectively have been confirmed here. Furthermore, mating system divergence in terms of nectar volume and sugar concentration, pollen ovule rations and selfcompatibility has been investigated. There was no difference in nectar volume between the two species investigated, however, nectar concentrations have been found to be significantly higher in L. parva, which is thought to have evolved through selective pressures of pollinator preference. Similarly, pollen ovule ratios in L. parva (22663) are statistically higher than that of L. splendens subsp. splendens (17360), which is predicted to facilitate gene-flow between populations. Both species have been shown to have early-acting self-incompatible (ESI). Similar genetic variation and gene-flow of the two species in question suggest that NFMP is similar to that of bird pollination in its ability to maintain high levels of genetic diversity. The case of Liparia provides a basis to reject the hypothesis of NFMP evolution from a matrix of bird pollinated ancestors due to similar pollinator efficiencies.