Search

Savanna plant communities change considerably across time and space. The processes driving savanna plant species diversity, coexistence and turnover along environmental gradients are still unclear. Understanding how species respond differently to varying environmental conditions during the seedling stage, a critical stage for plant population dynamics, is needed to explain the current composition of plant communities and to enable us to predict their responses to future environmental changes. Here we investigate whether seedling response to changes in resource availability, and to competition with grass, varied between two functional groups of African savanna trees: species with small leaves, spines and N-fixing associations (fine-leaved species), and species with broad leaves, no spines, and lacking N-fixing associations (broad-leaved species). We show that while tree species were strongly suppressed by grass, the effect of resource availability on seedling performance varied considerably between the two functional groups. Nutrient inputs increased stem length only of broad-leaved species and only under an even watering treatment. Low light conditions benefited mostly broad-leaved species’ growth. Savannas are susceptible to ongoing global environment changes. Our results suggest that an increase in woody cover is only likely to occur in savannas if grass cover is strongly suppressed (e.g. by fire or overgrazing). However, if woody cover does increase, broad-leaved species will benefit most from the resulting shaded environments, potentially leading to an expansion of the distribution of these species. Eutrophication and changes in rainfall patterns may also affect the balance between fine- and broad-leaved species.

Large herbivore grazing is a widespread disturbance in mesic savanna grasslands which increases herbaceous plant community richness and diversity. However, humans are modifying the impacts of grazing on these ecosystems by removing grazers. A more general understanding of how grazer loss will impact these ecosystems is hampered by differences in the diversity of large herbivore assemblages among savanna grasslands, which can affect the way that grazing influences plant communities. To avoid this we used two unique enclosures each containing a single, functionally similar large herbivore species. Specifically, we studied a bison (Bos bison) enclosure at Konza Prairie Biological Station, USA and an African buffalo (Syncerus caffer) enclosure in Kruger National Park, South Africa. Within these enclosures we erected exclosures in annually burned and unburned sites to determine how grazer loss would impact herbaceous plant communities, while controlling for potential fire-grazing interactions. At both sites, removal of the only grazer decreased grass and forb richness, evenness and diversity, over time. However, in Kruger these changes only occurred with burning. At both sites, changes in plant communities were driven by increased dominance with herbivore exclusion. At Konza, this was caused by increased abundance of one grass species, Andropogon gerardii, while at Kruger, three grasses, Themeda triandra, Panicum coloratum, and Digitaria eriantha increased in abundance.

The African elephant (Loxodonta africana) is a large-bodied, generalist herbivore that eats both browse and grass. The proportions of browse and grass consumed are largely expected to reflect the relative availability of these resources. We investigated variations in browse (C(3) biomass) and grass (C(4)) intake of the African elephant across seasons and habitats by stable carbon isotope analysis of elephant feces collected from Kruger National Park, South Africa. The results reflect a shift in diet from higher C(4) grass intake during wet season months to more C(3) browse-dominated diets in the dry season. Seasonal trends were correlated with changes in rainfall and with nitrogen (%N) content of available grasses, supporting predictions that grass is favored when its availability and nutritional value increase. However, switches to dry season browsing were significantly smaller in woodland and grassland habitats where tree communities are dominated by mopane (Colophospermum mopane), suggesting that grasses were favored here even in the dry season. Regional differences in diet did not reflect differences in grass biomass, tree density, or canopy cover. There was a consistent relationship between %C(4) intake and tree species diversity, implying that extensive browsing is avoided in habitats characterized by low tree species diversity and strong dominance patterns, i.e., mopane-dominated habitats. Although mopane is known to be a preferred species, maintaining dietary diversity appears to be a constraint to elephants, which they can overcome by supplementing their diets with less abundant resources (dry season grass). Such variations in feeding behavior likely influence the degree of impact on plant communities and can therefore provide key information for managing elephants over large, spatially diverse, areas.

Mesic grasslands in North America and South Africa share many structural attributes, but less is known of their functional similarities. We assessed the control of a key ecosystem process, aboveground net primary production (ANPP), by interannual variation in precipitation amount and pattern via analysis of data sets (15- and 24-year periods) from long-term research programs on each continent. Both sites were dominated by C(4) grasses and had similar growing season climates; thus, we expected convergence in precipitation-ANPP relationships. Lack of convergence, however, would support an alternative hypothesis-that differences in evolutionary history and purportedly greater climatic variability in South Africa fundamentally alter the functioning of southern versus northern hemisphere grasslands. Neither mean annual precipitation nor mean ANPP differed between the South African and North American sites (838 vs. 857 mm/year, 423.5 vs. 461.4 g/m(2) respectively) and growing season precipitation-ANPP relationships were similar. Despite overall convergence, there were differences between sites in how the seasonal timing of precipitation affected ANPP. In particular, interannual variability in precipitation that fell during the first half of the growing season strongly affected annual ANPP in South Africa (P0.01), but was not related to ANPP in North America (P = 0.098). Both sites were affected similarly by late season precipitation. Divergence in the seasonal course of available soil moisture (chronically low in the winter and early spring in the South African site vs. high in the North American site) is proposed as a key contingent factor explaining differential sensitivity in ANPP to early season precipitation in these two grasslands. These long-term data sets provided no support for greater rainfall, temperature or ANPP variability in the South African versus the North American site. However, greater sensitivity of ANPP to early season precipitation in the South African grassland suggests that future patterns of productivity may be more responsive to seasonal changes in climate compared with the North American site.