Patterns in longevity of soil seedbanks in fire-prone communities of south-eastern Australia
Tony D. Auld, David A. Keith and Ross A. Bradstock
Australian Journal of Botany
48(4) 539 - 548
Seed burial in nylon mesh bags over a 2-year period was used to examine seed longevity patterns in 12 shrub and two graminoid species in fire-prone habitats around Sydney, south-eastern Australia. Most species released a large fraction of their annual seed-crop in a dormant state and all species showed evidence for some form of persistent seedbank. However, regressions of seed persistence over time were in most cases poor predictors of seed decay (9 of 14 study species).
Considerable variation in the degree and pattern of seed longevity was apparent in the study species. Three functional groupings of species are suggested. (1) Seed half-lives in the soil predicted to be greater than 2 years and evidence of imposed secondary dormancy (continuous, Kunzea spp. or seasonal, Grevillea caleyi). Only Kunzea capitata and G. caleyi showed significant seed decay in this group. (2) Seed half-lives in the soil predicted to be greater than 2 years and no evidence of secondary dormancy (nine species). Six species had high seed dormancy at release (only two of which showed significant seed decay). Three species had initial seed dormancy of 40–57%—two (Asterolasia elegans and Zieria involucrata) with significant decay only for the non-dormant seed fraction, and one (Comesperma ericinum) with significant decay of both the dormant and non-dormant seed fractions. (3) Two species (Darwinia biflora and Persoonia pinifolia) showed evidence of very short mean half-lives of seeds in the soil (0.4–1.0 years). The threatened species, D. biflora, had a rapid initial seed decay over 6 months followed by little decay for 18 months, and the half-life of seeds is likely to be a poor predictor of seed longevity. For P. pinifolia, maintenance of a soil seedbank is predicted to be dependent on continual inputs of seeds locally or dispersal of seeds from other sites.
Full text doi:10.1071/BT99046
© CSIRO 2000