higher total reflectance from wavelengths between 450-900 nm than glabrous leaves (t = 4.0; d.f. = 18; p b 0.05). Hairy-leaved individuals tended to occur on granitic clay soils, whilst glab- rous individuals occurred on sandstone derived soils. However, there was no correlation between leaf traits and soil character- istics including pH, electro-conductivity and the abundance of major cations. There was also no clear correlation between mean annual rainfall and leaf traits. Thus, despite clear dif- ferences in leaf traits, these were apparently not associated with the edaphic or climatic factors under consideration in this study. A possible explanation for the leaf trait variation is a difference in phenology associated with water-permeability of the various soils. Preliminary studies show that hairy-leaved individuals flower later in the year than do glabrous individuals. doi:10.1016/j.sajb.2009.02.103 Determinants of community assembly and phylogenetic structure of Tetraria communities J.A. Slingsby, M.D. Cramer, G.A. Verboom Department of Botany, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa Previous work has demonstrated that co-occurrence of species in the fynbos sedge genus Tetraria at fine spatial scales is non- random with respect to phylogeny. This pattern may arise because the functional traits that determine habitat preference and interspecific interactions are phylogenetically structured. Whilst previous work has demonstrated that the evolution of a number of traits within the genus Tetraria are phylogenetically conserved, neither the utility of the traits, nor their association with specific environmental variables, were established. No direct link between ecological process and community structure could thus be inferred. I report the results of a study in which we investigated, within a phylogenetic framework, the utility and relationships with environmental variables of a range of morphological and physiological traits in the genus Tetraria. We use this information, along with species co-occurrence data and a phylogeny for the group, to infer which functional traits and ecological processes determine the structure and assembly of Tetraria communities. doi:10.1016/j.sajb.2009.02.104 Whole-ecosystem functioning on Marion Island - past, present and future V. Smith Department of Botany and Zoology, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa A long-term goal of the research program on Sub-Antarctic Marion Island (47°S, 38°E) is to construct whole-island budgets for plant biomass, primary production and nutrient cycling. Primary production and nutrient cycling have so far been measured for 8 of the island's 41 plant communities. This involved analyses of the seasonal changes in plant biomass and chemical composition, changes the sizes of soil nutrient pools, and studies of plant litter decomposition, soil carbon efflux and nutrient inputs to the communities by abiotic and biotic processes. The biomass, production and decomposition studies were carried out on a per-species basis and took nearly 20 years. At the end of that period we still seemed to be a long way from being able to quantify primary production and nutrient cycling, or to construct biomass and nutrient budgets, for the whole island. A recent classification of the island's terrestrial habitats relied on ordinations of plant guilds and soil chemistry. The habitats separate out in the ordination space in a pattern that can be readily interpreted in terms of the environmental and ecol- ogical forcing variables at the island (soil moisture and hy- drology, exposure to wind, soil type (mineral volcanic ash to highly organic peat), wind-blown salt spray and manuring and trampling by seals and seabirds). These same factors are also cardinal determinants of ecological function processes, such as photosynthesis rate, decomposition, soil microbial activity and nutrient cycling, so it was predicted that the habitat classification would emphasize the between-habitat variation in ecosystem function. The production, decomposition, nutrient cycling and soil heterotrophic activity values for the eight communities studied (each represents a different habitat) are highly correlated with the scores of the communities on the ordination axes. This suggests that the plant guilds of the habitat classification are analogues of plant functional types that can be used to economically estimate biomass, primary production and nu- trient cycling for the remaining habitats, and thus attain the goal of these parameters on a whole-island basis. The viability of this approach is discussed and some early results of its implementa- tion presented. doi:10.1016/j.sajb.2009.02.105 Fire frequency in the fynbos: Quantifying variation and dependency on fuel age D. Southey a,c , L.G. Underhill b , G.F. Midgley c , W.J. Bond a a Botany Department, National Botanical Institute, Private Bag X7, Claremont 7735, South Africa b Animal Demography Unit, University of Cape Town, Private Bag, Rondebosch 7701, South Africa c Global Change and BioAdaptation Division, National Botanical Institute, Private Bag X7, Claremont 7735, South Africa One of the most important aspects of fire, that influences both the natural and human environment, is its frequency. The frequency and variability of fire occurrence in fynbos is one of the major components responsible for plant community di- versity, and also indicates the frequency at which lives and 420 SAAB Annual Meeting Abstracts