| Abstract: | Grassland dynamics in a degraded disclimax grassland dominated by Danthonia caespitosa Gaudich. are examined using both demographic and multivariate approaches in an experiment designed to determine the effect of grazing intensity and exclosure on pasture dynamics. The experiment ran for 20 years from 1949 to 1968, using permanent quadrats at 3 grazing intensities and within exclosures. Demographic studies of some perennial grass species demonstrated markedly different responses to grazing; Danthonia caespitosa was unaffected by grazing but responsive to seasonal rainfall differences. Enteropogon acicularis survived only on protected sites. Numerical classification of total species set (121 species) for six observation periods demonstrated that community types were sensitive to differences in winter rainfall, and time since the start of experiment. Principal component analysis of permanent quadrat observations for individual years demonstrates quadrat trajectories which confirm this and indicate progressive divergence of the successional trends of the grazed and ungrazed quadrats. Repeated analysis on grazed quadrats only, shows that three components of pasture dynamics can be recognized; these are trend (succession?) and seasonal differences, each of which account for about 20% of the variance, and differences due to soil heterogeneity in the experimental paddock (8% of variance accounted for). No effect of grazing intensity was detected. Multivariate techniques can provide a clear partitioning of types of dynamic behaviour present in grassland communities. It is concluded that partitioning of environmental heterogeneity prior to demographic studies would increase their sensitivity. |
