POSTDISPERSAL SEED PREDATION IN GRASSLAND - ITS MAGNITUDE AND SOURCES OF VARIATION

1 Post-dispersal seed predation by rodents and invertebrates, was studied in two grassland sites in Berkshire, UK. The influence of season, seed density, seed burial, seed species, as well as within- and between-habitat variation on the removal of nineteen species of herbaceous seeds was investigated using exclosure experiments. 2 Analysis examined two components of seed predation: seed encounter (the probability of at least one seed being removed) and seed exploitation (the proportion of seeds removed once encountered). Seed removal was greatest from treatments to which rodents had access, losses attributable to invertebrates were negligible. 3 Burial reduced rodent encounter at both seed densities used and explained almost 21% of the variation, while having only a limited effect on the proportion of seeds exploited. Burial increased the variation in encounter attributable to the other sources examined and altered the direction of habitat, season and species effects. Interpretation of surface seed predation data to details of plant recruitment for seeds which form buried seed banks should be undertaken with caution. 4 Examination of rodent seed removal patterns revealed that on average 14% of the variation in both seed encounter and exploitation was attributable to small-scale spatial differences within habitats related to the distribution of small mammals. 5 No significant between site variation was found for encounter (< 2%) while seed density effects accounted for approximately 11% of the variation. Single seeds were encountered less than half as frequently as seeds in groups of 10. Density-dependence varied with seed size, with predation of large seeds less influenced by a reduction in seed density. 6 No significant species variation was found for seed encounter (species means ranged from 50 to 60%). In contrast, species effects for seed exploitation were significant and species means ranged from 20 to 95%.