Seed banks buffer plant populations against environmental variation and population stochasticity. Specifically, delayed germination should evolve to maximize the long-term geometric population growth rate and seed banks should be more likely to be selected for in populations which experience higher levels of interannual variation in fitness. Research with Clarkia xantiana ssp. xantiana suggests the presence and importance of a soil seed bank in this species. In this study, we leverage intraspecific variation in the seed bank to assess whether seed germination and survival probabilities are consistent with expectations from models for bet hedging. We ask three questions: (1) Is there intraspecific variation in germination probability? (2) Is germination probability negatively correlated with temporal variance in fitness? (3) Is germination probability negatively correlated with seed survival? Finally, we use stochastic population models to calculate the optimal germination strategy and compare these predictions to the observed germination probability.
Results/Conclusions
We use experimental and observational data from a published study on C. xantiana demography to fit Bayesian models for population- and year-specific vital rates (seed germination, survival, seedling survival to fruiting, fecundity). We find that germination probability varies among populations of C. xantiana, roughly doubling across the range of the species from ~0.10 in western range edge populations and ~0.20-0.25 in eastern range edge populations. Germination probability is not correlated with temporal variance in fitness (Pearson’s r = -0.16, 95% credible interval overlaps zero). Germination probability is negatively correlated with seed survival (Pearson’s r = -0.16, 95% credible interval does not overlap zero). The optimal germination strategy predicted by population models in which fitness is density-independent do not match the observed germination strategies. Our results provide insight into the selective pressures shaping the evolution of seed banks; our next steps will evaluate whether the observed germination strategies are better predicted by population models with density-dependent fitness or models with predictive germination.