Introduction
The savanna ecosystem is typically formed by a patchy mosaic of trees, shrubs and herbs, characterized by the dominance of grasses that create a more or less continuous ground cover (Walker 1987). The structure and abundance of savanna vegetation are driven by four main determinants: water availability, soil properties, fire, and herbivory (Huntley and Walker 2012, Bond 2019). Water and nutrient supply control primary production (Scholes 1990, Skarpe 1991), while fire and herbivory act as consumers of accumulated biomass (Archibald and Hempson 2016). However, although many experiments addressed complete or partial herbivore exclusion (Olff and Ritchie 1998, Staver and Bond 2014, Anderson et al. 2007), it has been pointed out that quantification of the herbivore impact under different nutrient and water availability in savanna ecosystems is still missing (Staver et al. 2021).
Large herbivores differ in their dietary niches, typically divided along the browser-grazer continuum, where three dietary guilds are defined: grazers, browsers, and mixed-feeders (Hofmann and Stewart 1972, Owen-Smith 1997, du Toit 2003). Grazers feed primarily on bulk/roughage that is mainly composed of grass, while browser diets consist of leaves and branches of woody species and forbs where the concentration of nutrients is higher. Mixed-feeders combine both food sources according to their changing availability and palatability during the dry and rainy seasons (Hofmann and Stewart 1972, du Toit et al. 2003, Codron et al. 2007). Grasses are therefore directly affected by grazers and mixed-feeders. Associated activities of animals such as trampling, wallowing, urinating, defecating, and digging also need to be considered when exploring the plant-herbivore relationships in savanna vegetation (Cumming 1982), especially in areas where animals aggregate and/or spend longer periods (Veldhuis et al. 2018). Large herbivores also reduce the woody layer that competes with understory species (Cumming 1982, Bond 2019). Large mammalian herbivores have been estimated to consume about half of the plant production in African savannas with intact faunas (Phillipson 1973), and herbivores have been shown to use ~57% of the grass biomass in exclosure experiments (Staver et al. 2021). The effect of herbivores on grasses is most evident under heavy grazing, which often concentrates at water sources (Thrash et al. 1993, Pringle 2004). Abundance and diversity of vegetation can be affected up to several hundreds of meters from the water source (Olivier and Laurie 1974, Thrash 1998a, Todd 2006, Smit and Grant 2009). Therefore, rivers in savanna systems can be hypothesized to affect vegetation in at least two ways: (i) higher soil moisture supports deep-rooting plants, and this effect interacts with (ii) an increased impact of animals that come to the water. On the other hand, herbivore abundance is driven by forage quality that is higher in nutrient-rich sites, such as fertile soils in termite mounds or grazing lawns (Mayengo et al. 2020).
Soil moisture and nutrients are essential drivers of grass biomass (Dye and Spear 1982, Zambatis 2003). Water supply controls both the growth rate and the length of the growing period, while soil nutrients, among which phosphorus and nitrogen are the most important, influence the growth rate (Scholes 1990). Soil nutrients are primarily determined by the character of bedrock – nutrient-rich soils weather from basic igneous bedrock, such as basalt, gabbro, or dolerite, while nutrient-poor soils from acidic bedrock, like granite, sandstone, or gneiss (Venter et al. 2003). Soil type also influences soil moisture by its water-holding capacity and water penetration during heavy rain; in clayey soils derived from basalts, the runoff is quicker and water does not reach the deeper soil horizons (Colgan et al. 2012). Coarse-textured sandy soils allow for greater infiltration and deeper percolation of rainwater than heavier-textured soils, consequently increasing moisture storage in the subsoil. Fine-textured soils that are high in clay content are much more xeric (Dye and Spear 1982). These differences influence plant species richness and diversity of all functional groups, shrubs, forbs, and grasses (Jacobs and Naiman 2008, Hejda et al. 2022). In the Kruger National Park, South Africa, plant communities on granites are richer in species than those on basalts (Hejda et al. 2022), where soils are more fertile but create less favourable conditions for most plant species due to rapid desiccation and mechanical stress caused by deep cracks in soil that damage fine roots (Muvengwi et al. 2018).
It has been reported that plant species diversity in savannas, including that of grasses, is often positively affected by free-ranging large herbivores (Olff and Ritchie 1998, Anderson et al. 2007, Díaz et al. 2007, Jacobs and Naiman 2008, Staver and Bond 2014). Grass diversity is expected to be highest under intermediate herbivore pressure because under high pressure, only species adapted to disturbance survive, as demonstrated by papers studying the effect of artificial water points or conducted in smaller game reserves (Thrash et al. 1993, Todd 2006, Smit and Grant 2009, Fenetahun et al. 2021), while at low densities or when herbivores are excluded, a few dominant grass species tend to prevail (Olff and Ritchie 1998, Anderson et al. 2007). By this mechanism, grazing affects competitive relationships among grass species. There are two possible scenarios: (i) the preferentially grazed species decrease because their biomass is being removed (Walker 1987), or (ii) they increase due to the suppression of other species in the plant community that are intolerant of grazing (McNaughton 1979, Hempson et al. 2015). The former mechanism is typical for heavy grazing by livestock or in game reserves, where it can lead to the replacement of palatable perennial grasses with unpalatable annuals (Werger 1977, Skarpe 1991). The latter option is likely to manifest in grazing lawns formed by short, palatable grasses with a high content of nutrients that provide a high-quality forage (McNaughton 1984, Cromsigt and Te Beest 2014, Hempson et al. 2015). Therefore, the intensity of grazing and herbivore species composition shape traits of species forming the grass community, such as small stature, high shoot density, and presence of belowground meristems to avoid herbivory pressure (Augustine and McNaughton 1998, Díaz et al. 2007), or high nitrogen leaf content in fast growing species in grazing lawns adapted to grazing (Arnold et al. 2014). Many plant species in African savannas are tolerant of grazing due to a long evolutionary history with herbivores (Cumming 1982). However, hypotheses diverge regarding what is expected to stimulate grass productivity (Staver et al. 2021). The grazing optimization hypothesis suggests that intermediate levels of herbivory should have the largest stimulating effect on productivity (McNaughton 1979, Hilbert et al. 1981), whereas alternative models indicate that productivity is most stimulated by intense but episodic grazing (Oesterheld and McNaughton 1991, Ritchie and Penner 2020). Studies have shown that the impact of free-ranging herbivores, unlike those in game reserves where animals do not migrate, is affected by migrations, which may change the abundance of herbivores and their species composition (Bakker et al. 2006).
Finally, the herbivore effect on grass abundance and species richness may interact with environmental settings such as nutrients and water availability (Milchunas at al. 1988, Veldhuis et al. 2014). The effect of herbivory on species richness is usually more pronounced in more nutrient-rich and productive ecosystems (Olff and Ritchie 1998, Bakker et al. 2006) or in burnt sites (Eby et al. 2014). However, some comprehensive large-scale studies only found an effect of herbivory, regardless of nutrient conditions. Borer et al. (2014) and Koerner et al. (2018) found that herbivore-induced change in plant dominance was the best predictor of plant diversity regardless of site productivity worldwide. Grazing pressure can lead to severe overgrazing and decreased grass diversity under arid and nutrient-poor conditions (Milchunas et al. 1988, Bakker et al. 2006). In contrast, under high water and nutrient supply, a similar level of grazing can cause a suppression of dominant species and increase grass species richness (Belsky 1992, Proulx and Mazumder 1998, Ritchie and Olff 1999).
To provide an insight into the relationship between plants and herbivores in an African savanna, we explored the interaction of grazing, varying water availability, and soil properties influencing plant communities dominated by grasses in the Kruger National Park, South Africa. Specifically, we asked (i) what is the relationship between either grass species abundance or grass species richness and the abundance and species richness of large herbivores, and (ii) how these patterns change with regard to water availability manifest by location of study sites at perennial rivers, seasonal rivers and dry crests on granite vs. basaltic bedrock.