|dc.description.abstract||Long-term trends in rangeland sward dynamics (species composition, structure, productivity)
were examined on three trials established between 1989 and 1996 at Kokstad Research
Station in the Highland Sourveld, while animal performance (average daily gain and gain per
hectare) was examined on two of the trials. The region enjoys moderate rainfall of 782mm per
annum, with hilly topography, and soil depths ranging from >1m to <20cm. The first trial was
labelled the simulation trial, as it simulated a four-paddock rotational grazing system, in
which animals spent two weeks in each of three paddocks while the fourth was rested for the
entire season. The rested paddock was rotated each year. The trial tested two stocking rates
(0.5 and 1.0 AU.ha-1) at five ratios of cattle to sheep, ranging from cattle only to sheep only.
The trial was unreplicated, and was established in 1989 on flat topography with deep soils.
The second trial (labelled the flat two-paddock trial) was established in 1992 adjacent to the
simulation trial. The trial examined two stocking rates of sheep weaners (0.5 and 1.0 AU.ha-1
seasonally) in a continuous grazing two-paddock system, in which one paddock of each
treatment was burned and grazed continuously while the second paddock was rested, to be
burned and grazed in the following season. The trial was replicated twice. The third trial
(labelled the steep two-paddock trial) mimicked the grazing system of the flat trial, but was
located on a steep (c. 20%) West-facing slope with shallow soils. The trial incorporated two
additional treatments: an intermediate stocking rate of 0.7 AU.ha-1 and an ungrazed treatment.
Species composition of the sward was recorded biennially on all trials using the nearest plant-point
technique with between 200 and 800 points per paddock. Sward standing crop was
measured in the rested seasons of the simulation trial and at the beginning, middle and end of
each season in one paddock of each two-paddock treatment of the two-paddock trials. In the
two-paddock trials, sward standing crop was measured within and outside permanently placed
exclosure cages. Animals were weighed fortnightly.
The response of species to grazing pressure or animal type was mediated by soil depth and
slope, as well as the grazing system. Tristachya leucothrix declined on all grazed treatments.
The ungrazed treatments remained relatively stable over ten years. On the low stocking rate
treatments of the steep trial, unpalatable species increased, but so did Themeda triandra. The
heavily grazed treatment of the steep trial was surprisingly stable, with little significant
change in relative abundance of key species other than an increase in the unpalatable
Alloteropsis semialata and decline in T. leucothrix. The medium stocking rate treatment on
the steep trial showed significant shifts in relative abundance of key species, with declines in
T. triandra and T. leucothrix and increases in A. semialata and the unpalatable wiregrass D.
filifolius. These trends were not repeated on the flat trial, however, with T. triandra and A.
semialata increasing and all other key species declining or remaining stable. On the
simulation trial, species responded largely unpredictably with species abundances often
fluctuating considerably over time. Microchloa caffra and A. semialata increased
substantially in both the low and high stocking rate sheep-only treatments, with a concurrent
decline in T. triandra in the high stocking rate but not the low.
Changes in composition over time, as measured by Euclidean distance, showed that shallow
soils, high stocking rates and a high proportion of sheep caused greater shifts in species
composition over time than deep soils, low stocking rates or more cattle. Three treatments, the
sheep-only treatments on the simulation trial and the high stocking rate on the steep trial,
showed an initial rapid shift in composition over about 6 years, before stabilising in
subsequent seasons. The flat trial showed no substantial shift in composition over time. This
general pattern of change was confirmed by Non-Metric Multidimensional Scaling.
On the simulation trial, total standing crop was influenced by stocking rate and by the
proportion of sheep in most seasons. On the two-paddock trials, increasing stocking rate
significantly reduced sward vigour, and vigour declined over time.
Stocking rate reduced total standing crop on both trials at the end of the 2004/05 seasons and
the crop of unpalatable species on the steep trial. Total palatable plants were unaffected by
stocking rate on both trials.
The classic Jones-Sandland model of animal performance as influenced solely by stocking
rate was not supported. Sheep performance was influenced by stocking rate and the
interaction of stocking rate and seasonal rainfall. There was no difference in average daily
gain between treatments over time, and hence cumulative animal production per hectare
increased with increasing stocking rate. Animal performance was possibly influenced by
many factors beyond the scope of this study, including the effect of predator attacks on
surviving animals, and resource availability such as shade and shelter and high-production
patches in some paddocks and not others. Scale effects on ecology are being increasingly
investigated and a meta-analysis of this type shows that, even in one research farm, slight
differences in management and environment can have significant effects on plant and animal
responses to grazing.||en