Seed bank bias: Differential tracking of functional traits in the seed bank and vegetation across a gradient.

A goal in trait-based ecology is to understand and predict plant community responses to environmental change; however, diversity stored within seed banks that may expand or limit these responses is typically overlooked. If seed banks store attributes that are more advantageous or vulnerable under future conditions, they could impact community adaptability to change and disturbance. We explored compositional differences between seed banks and vegetation (i.e. seed bank bias) across a twelve-site gradient of increasingly higher and older soil terraces, asking: How do seed banks contribute to taxonomic and functional composition, and what do shifts in seed bank biases along the gradient (i.e. tracking) reveal about the processes driving seed bank variation and its implications for community adaptability? Across the gradient, seed banks stored distinct pools of species that added to species richness but not functional dispersion. Seed banks were generally biased towards short-life histories and 'fast' species with small seeds, thinner and more acquisitive roots, and lower root biomass allocation; however, trait means in the seed bank and vegetation sometimes shifted along the gradient, amplifying or reversing these biases. For example, species with higher specific leaf area (tied to rapid resource acquisition) tended to dominate vegetation on lower soil terraces but were more common in the seed bank on higher terraces - at least when patterns were weighted by species' relative abundances. Although seed banks were generally characterized by 'fast' attributes, observed shifts in seed bank biases across the gradient - particularly in leaf traits - demonstrate that environment can impact stored diversity, and consequently, our expectations for future vegetative turnover. The seed bank bias patterns that we characterized could be the result of many potential processes, including environment- or trait-driven variation in seed bank inputs (seed production, dispersal) or losses (seed desiccation, germination), and may have important implications for a system's adaptive capacity. Only by integrating seed banks into the functional ecology agenda will we be able to unpack these processes and use seed banks more effectively in both prediction and ecosystem management.