Grasslands are a widespread and globally important biome providing key ecosystem services including C storage, regulation of the water cycle and diverse assemblages of grasses, forbs, and woody plant species. These plant species often have unique physiological and morphological mechanisms that facilitate persistence through time. Plant survival strategies often require trade-offs that balance competitive efficiency for resources, and an ability to tolerate (or avoid) frequent periods of resource limitation. In addition, alterations in bottom-up or top-down drivers (or their interactions) have the potential to modify coexistence dynamics among diverse species assemblages. Grasslands face multiple threats, including changes in drought intensity and bush encroachment - a process that results in increased woody plant abundance corresponding with decreased herbaceous plant abundance. The combination of reduced soil moisture and shifts in plant dominance from herbaceous to woody are likely to alter water pools in the soil profile. Altered ecohydrology may occur because woody plants tend to utilise soil moisture at deeper soil depths, while grasses are typically dependent on surface soil moisture. In order to predict changes in grassland vegetation structure and the hydrological cycle following bush encroachment, a greater understanding of changes in soil water pools at multiple soil depths is required, as well as the responses of these pools to changes in precipitation patterns. The Konza Prairie Biological Station in eastern Kansas, USA, supports ecological research focused on the dynamics and drivers of tallgrass prairie. In addition to biotic research, detailed landscape and hydrological characterisation of the site have provided insight about the roles of climate and land-use history on site ecohydrology. Fire frequency on Konza is prescribed at varying intervals for different watersheds, resulting in C4 grass dominance with frequent fire, mixed species assemblages with c. three-year fire frequencies, and bush encroachment in areas that were infrequently burned. Here, I will present a conceptual overview linking the physiological and morphological characteristics among C4 grasses, C3 forbs, and C3 woody plants. Fundamental ecophysiological differences among woody and herbaceous plants are manifest as distinct traits when competition for water is high, compared to coexistence when water is plentiful. Ultimately, these hydraulic traits and ecohydrological strategies among species influence landscape patterns, ecosystem processes, and susceptibility to drought. I will provide evidence that woody plants alter the ecohydrological dynamics of grasslands by changing infiltration pathways, altered groundwater recharge, and distinct patterns of plant water use. Identifying these unique traits of woody shrubs may prove useful for successful ecosystem management by utilising management strategies that increase stress among undesirable woody species and maximise the likelihood of mortality during prescribed fires.

Mopaneveld ecosystems cover approximately 555,000 km2 of rangelands and protected areas across southern African, and support extensive cattle and game farming, and eco-tourism. Grass production is critical for sustaining these industries, but is apparently lower in Mopaneveld than other savannas with similar rainfall, i.e. the grass layer in Mopaneveld has a low rain-use efficiency (RUE). Is this due to the abiotic environment (climate or soils), or biotic factors that could be managed (tree cover and grazing)? Four hypotheses that may explain the low productivity of Mopaneveld grass communities were tested by comparing Mopaneveld sites with nearby sites of non-Mopaneveld savanna in the central Lowveld region of South Africa: 1) The climate of Mopaneveld is less favourable for grass growth; 2) The soils of Mopaneveld are less favourable for grass growth; 3) A long history of over-grazing has degraded the grass productivity of Mopaneveld; 4) Mopane trees suppress grass productivity by reducing soil water content in the upper soil layers. Contrary to expectations, means for 5 to 10 years of annual biomass data from four Mopaneveld and four non-Mopaneveld sites in the central Lowveld had a similar range of RUE (0.06 to 0.39, and 0.05 to 0.22 g/m2/mm rain, respectively). Historical rainfall and temperature data for stations close to each group of sites revealed only minor differences in aridity (evaporation / rainfall) and intervals between rainfall events, between each group of sites. Soil types did not differ substantially between sites. The hypothesis that Mopaneveld is less productive because of historical overgrazing was supported, as sites with low RUE had long histories of heavy grazing. Plots within these sites that were recently protected from grazing had similar RUE to protected plots at the non-Mopaneveld sites. Support for the final hypothesis, that Mopane trees suppress grass productivity, already exists from the results of bush-clearing trials in Mopaneveld in other regions. Additional support for this was found in the form of increased grass production in response to thinning Mopaneveld at one of the sites. These results suggest that low grass productivity in Mopaneveld is caused by over-grazing and/or bush thickening rather than any abiotic factors. Better grazer management and bush-clearing should therefore form an integral part of any efforts to restore the extensive areas of degraded Mopaneveld in the communal rangelands and game farms of southern Africa.

There is a paucity of information on plant dispersion in African savannas, in particular the grasses, herbaceous plants and dwarf shrubs of the herbaceous layer. This is partly due to the lack of data analyses on plant dispersion derived from density estimates, specifically adequately replicated data sets allowing for full statistical evaluation. Recent comprehensive replicated density surveys across controls and three burn treatments in four landscapes of the Kruger National Park have facilitated the calculation of density indices for all species encountered. Variance estimators were generated using randomisation with replacement which, together with the mean density, allowed computation of dispersion indices. These indices indicate random, uniform or contagious dispersion. These indices are compared and shown to differ across the landscapes under three different fire regimes illustrating the effect of fire on plant dispersion per growth form. Results show a range of influences on dispersion. This is a step in the understanding of the functioning of vegetation structural layers under differing management regimes.

Protected areas (PAs) adjacent to the Kruger National Park (KNP) are divided into fenced and unfenced properties. I discuss land use development to provide context of how these savannas have evolved over the last century. While the PAs have embraced the basic philosophies of the KNP management approach and have similar high level objectives, they function at different spatial, and thus temporal scales due largely to fencing and the subsequent provision of water in these areas.?I examine bottom up (environmental, rainfall) conditions leading into five drought periods dating back to 1982/83 and top down drivers focusing on animal number and type on fenced versus areas open to the larger KNP system. We then examine the response variables, grass species composition, cover and standing crop and the capacity of these to attain a variety of potential land use objectives.?Using a data set spanning some thirty years, we provide scenarios that offer an early warning system to managers irrespective of land use objectives and management approach (laisse faire or active). We present scenarios, one of which illustrates a case where despite early warnings to landowners over a two year time period, a lack of landowner action resulted in a failed grass layer and ultimately an 85% mortality of the buffalo population during a drought.

In the Eastern Cape (MAP ~ 600 mm), bush encroachment and transition of savanna to broadleaf thicket are widespread. We examined the effect of fire frequency on woody plant density, size structure, and composition at the University of Fort Hare long-term fire trials, where five experimental burn regimes (without any grazing) have been applied since 1980: one, two, three, four, and six-year burns plus unburned controls. We wanted to determine (1) the trend in woody density over time from 1980 to 2018, (2) whether high fire frequency acted as a filter excluding broadleaf species associated with thicket, and (3) whether fire reduced tree density compared to unburned controls. Tree densities and size class distribution had been monitored in fixed plots at approximately five-year intervals between 1980 and 2009. In 2018, we conducted a full census of woody individuals > 50 cm height in the experimental plots and determined the density of individuals < 50 cm height in smaller subplots. Woody density had increased substantially over the course of the experiment and was dominated by Vachellia karroo, but all tree species persisted even under frequent fires. The density of V. karroo and total tree density were greatest at fire frequencies of between 2–4 years, while the unburned control plots had the lowest tree densities. Vachellia karroo individuals in the larger height classes ( > 2 m) were found in all treatments but were most abundant under triennial burning. This suggests that fire alone does not maintain the savanna state and that herbivory is likely to play an important role in keeping trees from escaping into the larger size classes. The low density of trees in all size classes in the unburned controls suggests that either a lack of seed scarification reduces germination, or that grass competition reduces seedling recruitment. This will be experimentally tested.