Consequences of neglecting cryptic life stages from demographic models

Information on individuals from all stages of life is crucial to explore their ecology, evolution and conservation biology. However, the life cycles of many species contain cryptic life stages that are difficult to detect and track over time and are therefore omitted from demographic models. One example is the dormant seed bank, an evolutionary bet-hedging mechanism that buffers plant populations in variable environments. To evaluate this methodological oversight, we conduct simulations to explore the effect of seed bank parameter uncertainties on demographic outputs such as the deterministic (lambda(1)) and stochastic population growth rate (lambda(s)), and extinction probabilities of 12 plant species. We have used uninformed and informed priors for seed bank parameters based on literature estimates, and reconstructed published models in which the seed bank was excluded without justification. Trials on removing the seed stage from models (6 species) explored the worst-case scenario for ignoring the seed bank. Inclusion of a seed bank and demographic uncertainty in seed bank parameters have little impact on stable populations (lambda(1 )similar or equal to 1) with high post-seedling survival. When populations deviate from stability or demonstrate temporal demographic variation, greater changes in lambda(1) and the range of possible growth rates caused by demographic uncertainty are observed. As expected, decreasing populations (lambda(2) < 1) benefit from the inclusion of a seed bank through increases in the growth rate and extinction times, whereas increasing populations (lambda(1) > 1) are slowed down. While germination estimates from the literature cannot accurately reflect those obtained in the field, they provide a starting point to assess the relative importance of a seed bank. The exclusion of the seed bank must be justified by confirming that dormancy is either non-existent or not important. Accounting for cryptic stages in demographic models will produce better informed management decisions for threatened or invasive species.