Predictions of microsite requirements for seven common Afromontane forest tree species were made by testing above ground relative growth rate (RGR) and photosynthetic response of seedlings of the respective species under different light intensities. Virgilia oroboides and Halleria lucida, with either a very high photosynthetic rate or RGR, were regarded to be pioneer species. Podocarpus latifolius was regarded to be a deep shaded species due to its low photosynthetic and RGR's. Rapanea melanophloeos, Ilex mitis, Olea capensis sub Ssp. macrocarpa and Kiggelaria africana either had moderately high RGR or photosynthetic rate and were regarded to be light shaded species. Predictions were tested in Grootvadersbosch (34ᵒ05's,20ᵒ50'E) forest, where frequency of seedlings beneath the canopy and under a canopy gap were recorded and tested using a x²-goodness of fit test, against the null hypothesis that there are equal frequencies of seedlings beneath the canopy and canopy gaps. Results confirmed predictions made for latifolius, O. capensis sub sp. macrocarpa, and R. melanophloeos. Low seedling frequencies for the other species, either due to a restricted distribution of adult trees (e.g. V. oroboides) or high levels of herbivore attack, made it difficult to test predictions. Significantly higher frequencies of seedlings of P. latifolius and O. capensis sub sp. macrocarpa beneath the canopy than in the gap of dry forest suggest that gaps in drier forests are stressful sites for regeneration. This led to the hypothesis of expecting a greater dominance of shade tolerant species in the canopy of dry forests than in moist forests. McKenzie's (1978) findings support this hypothesis. It was concluded that Afromontane forests are largely made up of deep or light shaded tree species, with pioneer species confined to forest margins. The role of canopy gaps can be seen to play a minor role promoting regeneration.

The evolution of non-flying mammal pollination has given rise to a broad suite of adaptive plant traits including dull coloured and geoflorous flowers, copious sucrose rich nectar, nocturnal anthesis and nectar production and a musky odour. The Fynbos endemic, Liparia parva (Fabaceae) has been recognised to exhibit several of these traits. Based on this observation, field studies were carried out on the Cape Peninsula, South Africa, to investigate the prediction that L. parva is rodent pollinated. Several lines of evidence indicate that flowers of L. parva are visited by rodents. These include; the presence of Liparia pollen in the faeces of live trapped rodents; observations of captive rodents selectively foraging at flowers of L. parva; floral debris underneath L. parva plants; and video footage captured of a rodent visiting L. parva flowers in the field. However, a strong plant-pollinator relationship was apparent only in the Cape spiny mouse, Acomys subspinosis. Captive Acomys foraged non-destructively at flowers of L. parva, becoming visibly dusted with pollen. The exclusion of rodents from flowers resulted in a significant reduction in seedpod set, indicating rodents do contribute to pollination success in L. parva. Additional evidence that L. parva is adapted to pollination by rodents includes nocturnal floral anthesis and large amounts of total nectar in inflorescences. The findings of this study provide substantial evidence for rodent pollination in L. parva and thus represent the first report of rodent pollination in a legume.

The aim of this study was to gain a comprehensive understanding of the patterns of carbon partitioning, storage and remobilization in Acacia karroo during the juvenile life history stage. Tuber total nonstructural carbohydrate (TNC) concentrations and δ¹³C values were determined in plants from two different stages in the juvenile life history of A. karroo. These were one year after a fire when the plant consisted of numerous leafy shoots or coppices (coppicing stage) and three years after a fire when the plant consisted of one pole like stem (gulliver stage). Gullivers were found to have mean TNC pools of 150g and mean TNC concentrations of 33%. Coppices had mean TNC pools of97g and TNC concentrations of 24%. Both total TNC pools and TNC concentrations in gullivers were significantly higher (p < 0.05) than in coppices. Carbon isotopes were used to determine whether growth was based on carbon reserves as heterotrophic growth shows a distinct enrichment in δ¹³C. The water relations of plants can also influence the δ¹³C values of plant growth. However, there were no significant differences in root size and depth between the stages, indicating that all plants had access to similar water sources. Mean δ¹³C values from the stems of plants in the gulliver stage were significantly enriched(> 1‰) in ¹³C compared to both coppicing plants (p < 0.01) and adults (p < 0.05). The negative δ¹³C values in coppice stems suggest that their growth is not based on stored carbon. The enriched δ¹³C values found in the gulliver stems support the hypothesis that carbon reserves are utilized to achieve fast growth rates in an attempt to escape the fire trap. However, the small magnitude of the differences in δ¹³C between the two stages suggests post-burn regrowth is derived from both current photosynthate and stored carbon.

Patterns of segregation among Protea species were investigated. Interspecific competition was studied as a cause of these patterns. Two different methods were used for measuring competition in a mixed stand of P. lepidocarpodendron and P. corona/a growing on the Cape Peninsula. These were the plot-based averaging method, which considers stand density, and the nearest neighbour approach, which considers competitive impacts as a function of the immediate neighbouring plant. Both methods demonstrated negligible interspecific, and strong intraspecific competitive effects on fecundity. However strong interspecific competition appears to be occurring at earlier stages of the life cycle and may account for habitat segregation of P. coronata and P. lepidocarpodendron. P. lepidocarpodendron juveniles appear to outgrow and suppress P. coronata plants. Survivors of this interaction grow to full maturity, giving an apparent lack of species interactions when measured in terms of fecundity. The same results were demonstrated in a mixed stand of P. lepidocarpodendron and H. suaveolens, where H. suaveolens suppresses P. lepidocarpodendron. The study indicates the importance of shrub architecture in reducing species interactions, and juvenile phase properties in determining competitive displacement. Soil preferences and variable fire responses between the species were studied in an attempt to explain the restricted distribution of the stronger competitor, P. lepidocarpodendron at the study site. No conclusions were reached, but the restriction of this species from certain sites appears to be a fire response.

I aim to determine whether or not remote sensing, through multispectral, satellite and digital photography, is a feasible and accurate method for determining drought stress in Fynbos vegetation. I hypothesize that (1) water stress in fynbos is detectable with the use of a remote sensing index, namely NDVI and (2) that the remotely sensed trends will correlate with ground truth measures of water stress.

Acacia karoo arises following disturbance in the Dune Forests of Northern KwaZulu-Natal. Patches of A. karoo which had arisen as a result of the abandonment of fields by subsistence farmers in a dune forest matrix were studied. Previous authors have considered A. karoo to be part of a successional mechanism which leads towards a climax forest canopy (Mentis and Ellery, 1994; von Maltitz et al, 1996). This succession is questioned as a number of the conclusions of these authors were not verified by my own work. The present paradigm holds that the A. karoo is even aged and will senesce after 25 years, while recruitment does not occur in the patches. I find no evidence of canopy senescence and microhabitats within which recruitment of A. karoo occurred were identified. Further the introduction of other species in the patches is facilitated not through the A. karoo canopy but was shown to be facilitated by the presence of multi-stemmed canopy-trees of other species which had not been removed by the farmers. However, this introduction of other species into the patches was still limited, with certain species not found in even the very oldest patches (37 years old). No discernible chronosequence could be identified when comparing patches of different age, which casts doubt on the belief that these patches will eventually revert to climax dune forest.

Fifty-two sites throughout the Cape Peninsula area burnt in the January 2000 fires were sampled to investigate the effects of fire severity on the recruitment of invasive and fynbos plants with particular emphasis on seed size. Four hundred fynbos species were divided into big- (>l0mg) and small-seeded (<10mg) species. Different measurements of burnt plant skeletons were examined to find the most appropriate indicator of the biological effects of a fire. It was found that the measurement of thinnest branch diameter of plant species with similar growth forms can be used as estimates of fire severity. Seedling density of fynbos species decline with an increase in fire severity, whereas recruitment of alien seedlings is not affected by fire severity. Big-seeded species and seedlings showed no significant correlation with fire severity, whereas the density of small-seeded species and seedlings correlated significantly with fire severity. It is likely that variations in fire severity that influences the depth to which seeds are killed will affect the ability of small seeds to emerge. High severity burns may thus have a different community composition by favouring the survival of big-seeded species.

Recent studies on the interception and utilization of occult precipitation (fog, cloud-borne mist and dew) have revealed that the direct wetting of foliage provides a water subsidy to plants of various ecosystem types. In this study, we investigate the presence of foliar uptake, and the effects of misting on the plant water potential of species representing diverse functional types, namely ericoids, proteoids and restioids in Fynbos species occurring within the Cape Fold mist belt. In this study, foliar uptake after 180-min submergence in distilled water was demonstrated by five of the seven species investigated. These species included all the restioids and ericoids investigated in this study. By contrast, the proteoids L. conocarpodendron and L. laureolum were found to show no significant amount of foliar uptake or increased leaf water content (%). There was an increase in the average, normalized leaf water content in individuals subjected to misting treatments in both proteoids, L. laureolum and L. conocarpodendron. Similarly, there was also an overall increase in plant water status, as shown by the increased water potential in individuals that were subjected to the misting treatment. It was found that control individuals showed a decrease in plant water potential (i.e. lost water) during the day, as can be expected when soil water is not replenished. All species showed significant stomatal conductance, during both night and day. Results indicate that misting events have a significant effect on the overall plant water status in all functional types and the presence of foliar uptake in both ericoids and restioids; thus indicating that cloud events may have an important effect on the vulnerability of these species to drought, under the precepts of global climate change.

The use of phylogeography in plant systems has been on the increase in recent years with the use of chloroplast DNA to detect sufficient intraspecific variation to reach significant conclusions about plant species histories, both temporally and spatially. In this study, the geographic structure and possible origin of the morphologically variable Themeda triandra is explored. The trnF - trnC and psbD - trnS gene regions of the cpDNA were used to find 12 haplotypes found in 11 populations of T. triandra that encompass the species large distributional range. A haplotype tree was constructed that showed the relationship of the 11 haplotypes (haplotype_H12 was excluded as it fell outside of the 95% confidence limit), with haplotype H6 inferred to be the ancestral haplotype. A nested clade analysis was performed with the results used to infer the geographic structure of T. triandra within South Africa. Significant results showed that there was restricted gene flow with nested clades involving the three Free State populations, indicating that there are barriers to gene flow with other haplotypes. The ancestral haplotype showed long distance colonisation, with a probable root of this colonisation being the Kruger National Park. This is the proposed point of introduction of T. triandra into South Africa, with results from this study supporting this proposal. A substantial amount of gene flow (25.49%; AMOV A) between populations is observed, with this probably being due to the widespread distribution of haplotypes H6 and H10. It is thought that T. triandra followed two migration routes within South Africa: one along the coast, with the other inland above the escarpment where populations became genetically isolated from populations below the escarpment. Further studies may look for a correlation between morphological variants of T. triandra and the cpDNA haplotypes found within the species.

Bibliography: p. 129-143. / The broad aim of this thesis is to investigate the formulation and usefulness of a very general model for plant population dynamics. In chapter 1, the goal of generality is discussed, particularly in the light of the lack of interaction between field and experimental population studies on the one hand and theoretical population dynamics on the other hand. A distinction is ma.de between descriptive and axiomatic theories, and it is suggested that they serve different purposes. The advantages of a. rigorous framework are pointed out and the basic elements of the continuum approach are introduced. In chapter 2, the model is proposed, the existence and uniqueness of solutions to its equations is proved, and an algorithm for numerically -approximating transient solutions is discussed. The question of generality is addressed in two places, and it is argued that the basic framework presented here is in principle adequate to model the processes of plant population dynamics in full detail, though the existence proof cannot to accommodate all possible models. In particular, models with time lags are excluded. Further limitations of the existence proof ill terms of constitutive relations are pointed out. In consequence, the theory here presented does not fully exploit the possibilities for generality inherent in the basic equations. In chapter 3, the question of what data would allow identification of factors determining somatic growth and mortality is investigated computationally. It is shown that using only the average size is insufficient. A class of models which includes all possible combinations of three types of size dependence in somatic growth and mortality is formulated. Qualitative parameter estimation for the various models yields size distributions that can be classified into the following biologically meaningful groups: group (i) has no models that use dependence on relative size; group (ii) has all the models in which somatic growth depends on relative size group (iii) has the models where only mortality depends on relative size. Thus it appears that size distribution may be used to distinguish various forms of size dependence in somatic growth and mortality. In chapter 4, a lottery model criterion for coexistence of plants with disjoint generations is developed, which is shown to require relative density dependence. Computer simulations aiming to initiate the use of exploratory calculations in studies of coexisting serotinous proteoids in fynbos indicate that the aspect of plant population dynamics most sensitive to density dependence is seed production, then somatic growth, while mortality is least sensitive to density dependence.