'n Plantekologiese verwysingsperseelnetwerk vir die Hoëveldlandboustreek

Abstract

A PLANT ECOLOGICAL REFERENCE AREA NETWORK FOR THE HIGHVELD AGRICULTURE REGION As a necessity for determining field condition, a reference standard by which pasture evaluation in South Africa can be applied, has been created. To be of any importance such a standard has to be compared with a known standard or benchmarksites (Tainton, 1984). Truter (1988) has determined a transec of benchmarck sites from Harrismith in the east to Potchefstroom in the west. In order to extrapolate the results of these selected benchmarks to the adjacent Agriculture Regions a extrapolation basis had to be found. Species composition data were collected on different toporaphical units in the study area, whereby special attention was given to the vegetation on the crest, middleslopes as well as the pediments in each landtype. Nearest plant recordings were made at 200 points in each plot, using the wheelpoint apparatus of Tidmarsh and Havenga (1955). During data collection sample plots were selected to represent various degrees of degradation. Special attention was given to sampling fence line effects, vegetation at different distances from animal concentration areas and the ungrazed vegetation eg. next to unfenced cultivated areas. The plots were separately ordinated for each topographical unit in the different landtypes by means of the DECORANA ordination technique and in particular the Detrended Correspondence Analysis (Gauch, 1982). The ordination has been repeated for each rainfall region and topographical unit. The (x)-axis representing the grazing gradient was in each case identified by the ordinated positions of sampling plots that had been surveyed at increasing distances from watering points, and those with known histories. The data matrices of the land type ordination resulted in a diverse dissimilar pattern of sampling plots, while the ordination in which the rainfall regions had been used, resulted in a orderly distribution pattern of the sampling plots. For this reason it was decided to use a rainfall gradient as a basis for extrapolation. Mathematical fittings of species present on the grazing gradients, i.e. on the (x) - axis, of the identified habitat types were executed by means of the polynomial regression techniques. The value on the (x) - axis where the potential matrix from which the reference areas can be chosen, occured, was to determine the theoretical compositions for the reference areas of each topographical unit in each different rainfall zone. The mentioned (x) - axis value is described by Bosch, et al., (1987) as the position where an equilibrium is reached between the negative effects of grazing and the natural tendency of the vegetation to reach the climax composition. This position is identified in the lightly to moderately grazing zone on the grazing gradient. With the help of these regressions the values on the (x)-axis at these positions, have been used to determine the percentage species composition for each theoretical reference points on the different topographical units in the different rainfall zones. These theoretical reference points were ordinated together with the original data matrices. The sampling plots nearest to the theoretical reference points and those having the highest percentage preferred species present, were taken as the multiple reference areas. By comparing the species composition of these multiple reference areas to that of the teoretical reference points, the multiple areas were reduced to only one reference area, with a similar species composition as the teoretical reference point. This has been done for each topographical unit in each different rainfall zone. Theoretical reference points were also calculated for the degraded vegetation that occurs in the study region, by employing the computer model of Species Composition Change, developed by Janse van Rensburg (1987). The approach to provide a network of reference areas (benchmarck sites) on an extrapolation basis for the Highveld Agricultural Region with the aid of ordination and regression techniques, was to a large extent succesful. The identified network of reference areas for the High veld Region can thus be of significant value for use in the range monitoring programmes in this region.