Thesis presented for the Degree of Doctor of Philosophy at the University of the Witwatersrand / Successional changes in the vegetation after fire were studied in several fynbos communities of the south-western Cape Province of South Africa. The study sites were located in the mountains, at altitudes between 300 and 1000 m a.s.l., in areas with winter rainfall regimes -1 and annual precipitation of about 900 to 1000 mm. yr Soils are highly leached, derived principally from quartzites. The two main sites were Zachariashoek near Paarl, where summers tend to be rather dry, and Jakkalsrivier east of Grabouw, where summer drought is ameliorated by fog precipitation and cloudiness. Successional changes were followed for intervals of up to 10 yr between fires, as well as for similar periods in vegetation that had been unburnt for 25 yr. Vegetational changes were analysed by means of repeated floristic assessments on permanent quadrats and point-quadrat sampling of canopy cover composition on these and on larger plots. At Jakkalsrivier, recently burnt and long unburnt vegetation were also compared by paired samples. Demographic trends in populations of prominent shrub species were followed by repeated censuses of tagged samples in unburnt and recently burnt vegetation. Also at Jakkalsrivier, the effects of fire on resources available to plants were examined by sampling soil moisture and soil mineral nutrients, as well as by following trends in xylem pressure potentials in selected species of plants and analysing their foliar nutrient concentrations. Effects of fire on microclimate were tested by comparative studies on burnt and unburnt sites. All fynbos communities sampled proved to be highly stable in the face of fire. Essentially, the pre-fire species composition was regained in 2-3 yr in every case. Species were added after fire, partly because of the appearance of ephemerals with life histories tied to fire, but also because of the reappearance of longer-lived plants as well as through the readier detection of species in vigorous vegetative form. The species richness of the regenerating corrununities tended to be quadratically related to pre-fire biomass, as predicted from current succession theory. Most species in any corrununity (about 70% on average) regenerated vegetatively by sprouting after fire. The relative numbers of species that regenerated germinatively, i.e. the seeders, did not vary in a manner predictively related to corrununity biomass. There were relatively few species with specialised life histories based on reseeding, such as those with canopy-stored seed and ephemerals with presumably specialised requirements for germination. Virtually no recruitment could be found among plants in the older (about 25 yr) vegetation, in contrast with lowland fynbos sites, where recruitment of herbaceous species occurs, and some mountain fynbos sites on more fertile soils, where forest precursors may sometimes colonise. Canopy redevelopment after fire indicated similar resilience among the different corrununities, despite variation in regrowth rates. Pre-fire growth-form composition was restored within around 10 yr. Maximum leaf-area indices ranged from about 1,5 to 2,5, although corrununi ties on phreatic sites had leaf-area indices exceeding 3,0. There was no evidence for a suppression of the understoreys by overstorey layers, mainly because the latter were sparse despite the abundance of tall broad-sclerophyllous shrubs in certain habitats. This was because the taller shrubs had sparse or slender crowns, or both, and because mortality tended to thin the populations before dense canopies developed. Trends in the composition of the canopy varied among corrununities. corrununities in productive habitats, i.e. in this case on phreatic sites, were dominated in the early stages by a relatively luxurious growth of ephemeral herbs and soft shrubs which declined within around 3-4 yr. Other sites had very sparse ephemeral cover, the early stages being dominated mainly by Restionaceae, Cyperaceae, and other sprouting herbs, and sprouting and seeding shrubs, which were constituents of the pre-fire canopies. In this respect, the fynbos is clearly distinguished from the California chaparral, for example, where ephemerals tend to dominate the post-fire stages on most sites. There was no evidence that fire had any effect on the water relations of regenerating vegetation, although stream discharge is known to be increased by fire in these environments. There was tentative evidence, in enhanced foliar concentrations of some mineral nutrients, that regenerating species of climax plants exploited nutrients released in fire. However, any such responses were small, especially in comparsion with responses observed in chaparral, for example. Ephemeral shrubs had much higher concentrations of foliar nutrients overall than climax species, tending to confirm the correlations found in Australian heathlands between plant life-history and nutrient economy. The effects of fire on microclimate were pronounced, especially on the thermal and water vapour regimes experienced by seedlings and sprouts. These extremes did not, however, appear as water stress in regenerating plants. Despite relatively sparse canopies, mature vegetation did reduce light at the ground to levels likely to affect seedling recruitment and survival. Preliminary experiments with a local dominant shrub, Leucadendron xanthoconus, showed a pronounced intolerance of shading and hence that light attenuation by canopies must be implicated in successional processes. The demographic studies indicated that density-dependent effects were not important in survival of plants. Two species of fire ephemeral shrubs effectively died out within four years, being characterised by markedly higher growth rates than climax species and brief and early fecundities. Climax shrubs had more or less constant rates of mortality over time, though populations in unburnt vegetation tended to have slightly higher rates of mortality than young populations. Densities of seedling populations were very high, but mortality rates were extremely low. In summary, it may be said that the fynbos communi ties studied here are very stable under a given fire regime. Recovery is rather rapid, being apparently achieved within 10 yr. Not much change occurs in older vegetation, but there was a gradual attrition of populations of dominant shrubs, without recruitment, with rare exceptions. Summer droughts in these montane environments are evidently not sufficiently marked for water deficits to play a primary role in succession, so that fire has no effect on plant water relations. Nutrient responses are relatively weak, and masked in the plants by the low rates of metabolism in climax species. Succession after fire is distinguished by the recovery of pre-fire communities, and subsequent inhibition of recruitment. This inhibition is probably through the effects of canopies on microclimate, although the interactions between especially plants and animals have been implicated in succession in other studies. / AC2017