Understanding life history traits and how they influence population-level processes under different conditions have been important areas of study in evolution and ecology. One application in ecology has been to determine whether there are general responses that apply to species with similar life history traits, which could provide a predictive understanding of species with shared traits rather than having to adopt a speciesspecific approach to management. In this study, an analysis of life history traits was applied to cycads, the oldest living seeds plants which are at risk of extinction due to vulnerability to illegal wild harvesting and habitat loss. The study focused on South African cycad species within the genus Encephalartos, as well as two species from Swaziland. The first objective was to identify groups of species with similar life histories and to assess whether these groups are aligned with phylogenetic relationships or an association with particular environments. The second objective was to determine whether species with shared suites of life history traits exhibited a similar population stage structure. Hierarchical cluster analysis and Principal Component Analysis (PCA) were used to identify groups based on key life history traits, namely, cone number and coning interval, number of seeds produced, number of stems, stem height and plant habit. Four life history groups emerged, these were: Group 1 with single-stemmed, subterranean species such as E. villosus; Group 2 generally characterised by medium-height species with many stems such as E. lehmannii; Group 3 with the tallest species which generally produce the most cones at the most frequent interval such as E. natalensis; and Group 4 with dwarf-arborescent, multi-stemmed species such as E. horridus. The groups were not found to correspond with the most recent phylogeny for Encephalartos and thus phylogenetic constraints were excluded. An analysis of environmental variables for 193 populations of species in all groups, including maximum and minimum temperatures, mean annual precipitation, moisture index, rainfall co-efficient of variation and fire return frequency, showed that the means per group were not significantly different from each other for most of the variables. However, it was evident that the variable ranges showed clear trends with biological significance. These tended towards more stable, less variable environments with more predictable rainfall for Group 1 species, and less stable, more variable environments with less predictable rainfall for species within Group 4. Groups 2 and 3 showed a large range and overlap in environmental associations with no consistent patterns. Population structure is valuable for determining whether recruitment in a population is adequate for species survival. Static life tables and Chi-Squared analyses were used to test differences in the population structure and survivorship patterns of species between and within the life history groups. If particular demographic patterns were shown to be emergent properties of certain suites of life history traits, autecological studies may be avoided for predictive-ecological conservation frameworks. The results showed that species in Group 1 had populations with a high proportion of juveniles and fewer reproductive adults, indicating high recruitment as well as high juvenile mortality. In contrast, species in Group 4 had populations with a high proportion of adults and fewer juveniles indicating low recruitment and high juvenile mortality with primary investment into adult plant persistence through vegetative suckering. Group 2 consisted mostly of rare species with very small populations and high proportions of adult plants showing evidence of episodic recruitment events. Species in Group 3 showed a similar trend towards intermittent recruitment with adult persistence, however, no explicit population structure emerged. This led to the conclusion that Groups 1 and 4 are relatively consistent and predictable, however, Groups 2 and 3 would better benefit from an autecological approach to management. This study was helpful to assemble life history data for South African Encephalartos, as well as environmental parameters for each species. Ultimately, it has shown that primary investment into seeds versus persistence, or a combination thereof, can influence population structure.

The handful of studies that have investigated chimpanzee and gorilla seed-dispersal identify these primates as important dispersal agents. These studies do not, however, make any measure of the 'quality' of the dispersal service offered by chimpanzees and gorillas. Determining 'quality' requires a measure of the dispersal distance and the microsite to which the seeds are dispersed. In this study, I report the first estimate of seed dispersal curves for chimpanzees and gorillas. Seed dispersal curves were produced by combining ape movement data with gut passage curves from literature. The derived dispersal distances for chimpanzees and gorillas are similar c. 7.7 km; this is surprisingly large when compared with other seed dispersal agents. This is likely due to a combination of foraging behaviour and gut physiology. At a species level, chimpanzees (Pan troglodytes) were shown to direct dispersal of Uapaca palidosa to favourable microsites even though gorillas (Gorilla gorilla) were responsible for moving a greater number of seeds. This study presents a novel method for the rapid derivation of dispersal curves and highlights the importance of incorporating species. level as well as community level studies to assess the quality of seed dispersal agents. It is my hope that the methods presented here be applied elsewhere so that the role of extant megaherbivores as seed dispersal agents be incorporated into future models that investigate forest dynamics.

Vegetation stability was investigated within the Cape of Good Hope Nature Reserve by incorporating three vegetation surveys spanning a 44 year time period. The goals of this study were to investigate changes in overall species diversity within the reserve, stability of community composition within and between sites, and which factors are influencing stability. A survey was conducted on 26 vegetation plots across the reserve. Data from this survey and two previous surveys was used to compare species diversity between time periods. Vegetation compositional similarity was also compared between sites. I tested for the effect of difference in vegetation age and soil type as predictors of vegetation groups. Results showed an overall decrease in species diversity. Analysis ofvegetation groups showed that vegetation types are mostly stable as current vegetation can be predicted by groups based on 1966 data. Vegetation is influenced significantly by fire history and soil types.

Cooking shelters are found at all the 28 stocking posts in the Paulshoek village. They are round circular structures with walls made from plant materials to provide insulation. Food is prepared and cooked inside the shelters, providing protection from the elements mainly cold and wind. These shelters also serve as social meeting and entertainment venues. Due to the poverty of the area alternatives to constructing cooking shelters are too expensive. Thus the sustainable use of plants used to construct cooking shelters will prevent both species loss and economic expense. There are three types of cooking shelters: full canopy, half canopy and open shelters. 18 shelters including all three types were measured. The volume of plant material calculated and the species composition estimated. The full canopy shelters are generally the largest and well built. These shelters are used in winter and need to be sturdy to withstand the elements and last for a number of years [ 15 years]. Open shelters are less permanent and built more often. As they are not required to be as sturdy a higher proportion of less desirable but more available plant species were used. However in all types the major species used was Polymita albiflora. This species is desirable as a construction material as it packs tightly and the dense foliage supply insulation. Although most shelters were made up of more than one species, only Galenia africana and Euphorbia mauritanica are used consistently in small quantities. Both these species are common and widespread and the amounts being harvested are insufficient to cause any negative effects on the species population dynamics. However P. albiflora is restricted to quartzite slopes or shallow pockets in granite in Northern Namaqualand. Thus harvesting could be destructive to the natural populations. A cooking shelter was dismantled and a size class distribution of P. albiflora used for construction calculated. Transects were run in four separate P. albiflora populations and a size class distribution for all sites calculated. The majority of plants used to build cooking shelters are in the size class 2 0.005-0.015 m³. The size class distribution of natural populations of P. albiflora showed the smallest size class [ <0.005 m³] to have the highest frequency. There was a decrease in the frequency for the remaining size classes. Harvesting from those size classes could cause this reduction. However it is more likely that high mortality during the establishment phase causes this decrease. A regression between volume and number of capsules showed that the plants being removed are not the only reproductively active plants. Small plants [0.001m³] are capable of producing a few capsules each containing more than 100 seeds. Also the few large plants produce up to 300 capsules per plant. P. albiflora is also used as a poor quality fuelwood for baking. Baking occurs approximately once a week. However only populations within the immediate vicinity are threatened by overexploitation due to the efforts required for fetching these plants far outweighs their use as a fuelwoood. This study indicates that P. albiflora is being harvested well within the sustainable limit. P. albiflora grow in dense stand and removal of plants within these will most likely be replaced by small plants or seeds of this species.

Pinus resinosa stands in Minnesota, USA were surveyed and fine and coarse roots from stands were analysed for nitrogen isotopes in an attempt to determine if fire effects nitrogen cycling. Fire increased the % nitrogen of coarse roots and changed the ratio of C: N. It may also decrease the depth of the A-horizon. It did not however affect the nitrogen isotope ratio of stands. Isotope values of coarse roots were affected by the depth of the A-horizon, which is consistent with other studies. A correlation between the ¹⁵N values of coarse and fine roots was found however fine roots were enriched proportionately to coarse roots. This suggests that they may be using a more enriched source of nitrogen such as ammonia although this is not conclusive. A PCA performed on site variables and isotope values determined the two major axes for sites were bum history and nitrogen cycle openness but that burning did not affect the second axis. A stepwise backwards regression showed the best predictors of a roots isotope value were, in decreasing order, the isotope value of coarse roots for fine roots and vice versa, the % nitrogen of roots and the age of stands. Coarse and fine roots showed opposite trends with % nitrogen. As the % nitrogen increased, fine roots were more ¹⁵N enriched, again pointing to different nitrogen sources for coarse and fine roots. Older stands do appear to have a more open nitrogen cycle, as roots were ¹⁵N enriched in these stands. The availability of different forms of nitrogen may influence the ¹⁵N signal of roots especially if fine roots are preferentially using ammonia. Therefore, it is not possible to say if isotope values are due to stand openness, or due to different resources use by roots.

Tree mortality, following surfaces fires, is much greater in forests than in savannas, despite similarities in the bark thickness of tree trunks. This difference is thought to influence the distributions of forest and savanna biomes. There is no consensus and very little research on how fires actually kill trees. Though tree canopies are too high above the flames to be charred, leaves are often scorched over the full tree height; hence the focus on differences in tree canopy recovery. I explored two possible mechanisms for differences in crown resprouting following canopy scorch: (1) differences in bud banks and their insulation, (2) differences in the vulnerability to disruption of the hydraulic system by heat scorch. I simulated canopy scorch using nine congeneric savanna-forest pairs in three Reserves in KwaZulu Natal, South Africa. Contrary to predictions, there was no difference in the presence of a viable bud bank between congeneric species. The resprouting response varied between congeneric pairs with only the Combretum, Rhus and Acacia pairs showing savanna species with a significantly greater ability to resprout. The vulnerability to hydraulic disruption was measured using wood density as a proxy for conduit reinforcement. The results are inconclusive and possibly affected by the growth rates of certain shade-tolerant forest trees. The differences in crown resprouting following canopy scorch of forest and savanna trees is not the result of differences in bud banks and resprouting ability. The differences in insulation of vulnerable tissue by bark may be related to the protection of the hydraulic system rather than the buds and/or cambium.

The main aim of this study was to determine if Ischyrolepis sieberi, a common restiod species found at high altitudes in the Cederberg captures fog to supplement its water needs. This was done by comparing the isotopic ratios of δD and δ¹⁸O in captured fog, rainwater, and stream water and comparing these to the δD and δ¹⁸O values found in the xylem water of Ischyrolepis sieberi plants growing at the study site. The δD values of the collected fog samples were enriched relative to rainwater, stream water, and xylem water from I. sieberi, however these differences were not significant. The δ¹⁸O values of the I. sieberi xylem water were significantly (p < 0.01) more enriched than rain, stream, and fog water. The xylem water of I. sieberi was depleted in δD and enriched in δ¹⁸O relative to all other water sources. The plants therefore did not appear to be utilizing fog during the three months of this study. The δ¹⁸O values of I. sieberi were consistently 4-6‰ enriched compared to rainwater, suggesting that they are utilizing rainwater, which has undergone fractionation due to evaporation in the soil before being taken up by the plants. This study was run during the wet winter months. The δD and δ¹⁸O values from previous data suggest that the I. sieberi plants do utilize fog during the dry summer months.

Forest and thicket communities are distributed throughout fynbos. The species constituents exhibit life history characteristics significantly different from fynbos species, in that reproductive biology and recruitment are not coupled to fire. The limited occurrence of forests and thickets suggest that there may be specific abiotic factors which limit their distribution. In an attempt to determine which factors are significant in prediction, data from 600 plots covering a substantial part of the Cape Peninsula was analyzed using linear regression models. Although trends in predictive variables were detected for both forest and thickets, the models were largely unsuccessful. This was due both to the lack of inclusion or poor transformation of certain factors and the fact that not all potential forest and thicket sites are filled. The role of fire as the stochastic element in the system as well as the significance of measured environmental variables in determining forest and thicket location is discussed.

I investigated the infraspecific diversity among individuals of Aspalathus linearis from five localities in the Cederberg Fynbos region using morphological traits and two DNA- based methods: inter-simple sequence repeats (ISSR) and a plastid marker. A principle component analysis (PCA) based on seven morphological variables grouped the samples into three different growth forms, specifically an erect-form, a prostrate-form and a shrub-form. The genetic variability of four plastid markers and one nuclear marker was determined for seven individuals selected from the populations. The trnL'UAAF-trnFGAA plastid marker was used to amplify the remaining samples and revealed four haplotypes, with a basal haplotype fixed in both the prostrate form and a single population of the erect form, a unique haplotype fixed in the shrub form, and a mix of two other haplotypes in the remaining erect forms. The congruence between haplotypes and regeneration strategy suggests a longstanding pattern of restricted gene flow. However, a small sample size associated with the plastid marker data limits or reduces the certainty of these findings.

Clonal growth has been studied little in Africa, and models of savanna tree-grass interactions have not included vegetatively reproducing trees or shrubs. Root suckering has been observed in both subspecies of Dichrostachys cinerea that occur in Southern Africa. This study, performed in Hluhluwe-iMfolozi Game Reserve, KwaZulu Natal South Africa, examines variability in subspecies distribution and the extent of root suckering under different fire regimes, and levels of herbivory. Variations in subspecies distribution are clear. Subspecies africana, with greater structural defenses, survives in areas of high herbivory but is rarely seen under high fire disturbance. The opposite is true for subspecies nyassana, which is poorly defended against herbivory. There was no difference in vegetative propogation by root suckering in areas of different fire regime, rainfall, soil type or degree of herbivory. The implication is that the management of the species will be a formidable task. A combination of burning during summer, when below ground reserves are depleted, and heavy browsing may aid in the control of this species.