Grass cover
The abundance of herbivores in our plots strongly depended on water
availability/proximity to water; almost two-thirds of the individuals
were recorded at perennial rivers, less than one-third by seasonal
rivers, and only ~7% at crests. The concentration of
herbivores at perennial rivers increases the impact on grasses, leading
to a lower grass cover. This is in accordance with previous studies
(Thrash et al. 1993, Thrash 1998b, Todd 2006, Smit and Grant 2009) that
found the lowest cover of herbs and shrubs at artificial watering points
and its increase with increasing distance from the water source. Olivier
and Laurie (1974) report a similar pattern from the Mara River in
Tanzania, where grass cover increased from 34% to 71% over one
kilometer from the river, and grazing intensity declined from 86% at 20
m from the river to 35% at 880 m apart. Soil moisture could be higher
near perennial and seasonal rivers, influencing the growth and abundance
of deeper-rooting grasses. In general, grass biomass and cover increase
with water availability (Zambatis 2003, Staver et al. 2019), and
experiments show that grass abundance is higher in moist than in arid
sites after herbivore exclusion (Staver and Bond 2014). This indicates
that herbivore pressure, especially heavy grazing and other effects of
animal presence typical of areas along perennial rivers (Olivier and
Laurie 1974), likely outweighs the positive effect of enhanced water
availability. This notion is supported by the intermediate grass cover
and species richness we recorded at seasonal rivers, where herbivore
abundances were also moderate. It needs to be noted, however, that the
low grass cover by rivers, which is a proxy of actual biomass, does not
necessarily mean low biomass production. Grazing can be compensated by
increased growth of grasses (McNaughton 1979, Ritchie and Penner 2020),
which is stimulated by increased nutrient input from dung and urine and
by lower self-shading (Augustine et al. 2003, Zimmermann et al. 2010).
Grazing lawns that often occur nearby rivers are an example of a highly
productive grass ecosystem maintained by intensive grazing (McNaughton
1984, Archibald 2008, Hempson et al. 2015).
Finally, grass cover significantly differed between bedrock types, and
this difference was most pronounced on crests where it was on average
almost twice as high (53.4% cover) on clayey soils derived from basalts
than on sandy soils on granites (29.4%). This corresponds with the
results of Dye and Spear (1982), who found greater grass biomass on
clayey soils in Zimbabwean savanna systems. In other habitats, such as
seasonal and perennial rivers, the effect of bedrock was not as
pronounced, probably over-ridden by grazing. Grasses on crests sometimes
created a continuous cover, often dominated namely by Bothriochloa
radicans and abundant Themeda triandra or Cenchrus
ciliaris (Table 3).