Arrested recovery in a sandy woodland correlates with a lack of heavy and long seeds in the seed bank

Abstract

The spontaneous recovery of habitats to pre-disturbance species' assemblages rarely occurs following major disturbances such as land clearing. The mechanisms filtering the recruitment of plant species into post-disturbance habitats are poorly understood. We set out to determine why some species do not return after large scale disturbances. We compared the trajectory of spontaneous recovery of an endangered woodland community in Eastern Australia every 6 months over 3.5 yr using 29 x 1 ha plots and three habitat conditions (cleared, degraded, intact woodlands, 'n' = 348 plant species). We used non-metric multidimensional scaling to ordinate the floristics and we compared changes over time using PERMANOVA and Procrustes rotation. We also compared recruitment for 38 woody species. Seed bank composition was determined from (1) germination trials conducted over 75 weeks and (2) sieve and search retrievals. Seed morphologies (shape, mass, dormancy mechanisms) found in the above- and belowground vegetation were compared. We found large differences in species composition between the cleared habitats and both the degraded and intact habitats due to a lack of native species in the cleared habitats. Initially cleared areas had similar rates of recruitment as degraded and intact habitats but in the last four monitoring visits recruitment in cleared habitats stalled. Overall few woody species recruited into cleared areas. For the seed bank, 23% of species were only found above ground and were not present in the seed bank, including many woody species. The majority of these species have a dormancy mechanism-so we excluded transiency as an explanation for their absence in the seed bank. However, these species were found to have significantly larger and longer seeds than species found below-ground only. Furthermore of the woody species that failed to recruit into habitats, 57% of these were seldom found in the seed bank, even though they were present in the above-ground vegetation, and all but one of these species had seeds in the top 20% of heavy-seeded species. We propose that large-seeded species experience higher levels of predation than smaller-seeded species and this filtering process is skewing the recovery of this endangered ecological community.