Seed bank dynamics in restored grassland following the sowing of high- and low-diversity seed mixtures

Soil seed banks on ex-arable land are dominated by undesirable ruderal species that compete with “desirable” target species during grassland restoration. At the same time, for continued regeneration, the latter often functionally depend on gap colonization from the seed bank, which serves as a buffer against local extinction. Nonetheless, few studies have so far investigated the effects of restoration practices on seed bank dynamics. Using a multisite experiment investigating techniques for restoring lowland mesotrophic grassland, we studied the effects of seedbed preparation (shallow cultivation using harrows or discs vs. deep cultivation using a plow) and of seed mixtures (species-rich grass–forb mixes vs. species-poor grass-only mixes vs. unseeded natural regeneration) on 7 years of post-restoration seed bank dynamics. We assessed how these practices affected density and diversity of sown and unsown species in the seed bank. Seed bank dynamics were much more strongly affected by seed sowing than by cultivation. Grass sowing resulted in stronger seed bank decline of unsown grasses, and additional forb sowing in stronger decline of unsown forbs. Higher seed densities and species richness of sown forbs colonizing from neighboring plots sown with the grass–forb mix were observed under natural regeneration than in the grass-only sown treatment, reflecting grass priority effects on sown forb colonization in the latter. Sowing of diverse target species mixtures was associated with the greatest shift in seed bank composition away from extant ruderal species towards sown target species. Our results illustrate the usefulness of seed bank monitoring for assessing restoration progress.