The influence of grazing on seed production and seed banks of some African savanna grasslands.
Species composition, population structure and the seed banks of Aristida bipartita, Bothriochloa insculpta, Heteropogon contortus, Themeda triandra and Urochloa mosambicensis, were described for 9 sites in mesic savanna grasslands, South Africa (seed banks for 5) on a similar soil type ranging in grazing history from sustained heavy grazing (>20 years) to light grazing. The influence of 2 seasons of grazing on the seed production and the seed bank of T. triandra, Setaria incrassata, H. contortus, A. bipartita, B. insculpta and Digitaria eriantha, was estimated at 2 other sites. Lightly grazed sites were characterized by the longer-lived palatable perennials T. triandra, B. insculpta, H. contortus and D. eriantha, and heavily grazed sites by the shorter-lived perennials U. mosambicensis, Sporobolus nitens, Chloris virgata, the unpalatable A. bipartita, and some forb species. Grass tuft size declined with increasing grazing pressure. Sustained heavy grazing reduced or eliminated T. triandra, B. insculpta and H. contortus from the seed bank, but increased the abundance of A. bipartita and U. mosambicensis. Rainfall influenced the seasonal pattern of seed production, which decreased with intra-seasonal drought. The number of seeds produced/m2 of tuft basal area varied from 21 000 for S. incrassata to 1.9 million for A. bipartita. Annual seed production/m2 with light grazing was ≤400 for T. triandra, 500-600 for H. contortus, 1300-4000 for S. incrassata, >10 000 for A. bipartita, and 300 for B. insculpta. Seed viability varied from 0 for D. eriantha to 68% for S. incrassata, and varied within a season for T. triandra (24-49%) and S. incrassata (26-68%). The seeds of all species tested had a period of innate dormancy of <1 year. Seven to ten grass and 6-7 forb species were found in the seed banks at 2 sites. Perennial grasses constituted 85-95% of the seed bank, with >90% of all seeds occurring in the top 2 cm of soil. Seed density differed markedly between sites and grazing treatments. A simple model of the seed bank dynamics of the perennial grasses showed that because their seeds germinate readily and have a limited survival when in secondary dormancy, the seed bank size was determined primarily by the species abundance in the vegetation. Thus, the seed banks of T. triandra and H. contortus, which produce small numbers of seeds, could easily be eliminated by sustained grazing. This was also indicated by the distribution of these species along the gradient of grazing history.