Predicting metapopulation responses of a tidal wetland annual to environmental stochasticity and water dispersal through an individual-based model

Freshwater tidal wetlands are a complex environment for annual plants. Seedling establishment and survival may be limited by a variety of factors, including competition with perennials and the twice-daily inundation of seeds and seedlings. Hence such species are often endemic and rare. Their observed population dynamics can be extraordinary, with individuals reappearing in certain patches where they had been absent for several seasons, and with total populations varying by orders of magnitude between years. Many interacting influences are thought to be at play here, including seed banks and water-based seed dispersal (hydrochory). So far it is not known (1) to what degree environmental stochasticity is likely to affect the population's survival in its natural habitat, (2) what role hydrochory plays in propagating and maintaining the species, and (3) how these two factors interact with one another. We therefore took the annual Aeschynomene virginica (Sensitive joint-vetch, SJV) as an example and developed an individual-based model in a geographically precise replica of its Holts Creek, Virginia, habitat. The model represents SJV's life cycle and is calibrated to data from a variety of empirical studies on the plant. Vital rates are partly calibrated from aerial imagery providing estimates of the biomass of specific patches. Simulated seeds enter the river network based on their proximity to the water's edge, and then travel upstream and downstream according to estimated flow rates, float times, and implantation probabilities. Additionally, random seasonal environmental conditions are imposed, depressing or inflating vital rates within prescribed ranges. We found that as environmental stochasticity increased to more than relatively modest levels, the long-term survival probability of the species precipitously declined. Hydrochory, though it may have played an important role in the past in allowing SJV to reach the regions in which it now thrives, had little impact on the plant's long-term likelihood of survival for our study population. Nevertheless, the model's performance indicates the existence of additional key factors at play in SJV's metapopulation dynamics that were not considered or quantified so far. These may include the varying elevation of habitat patches and the corresponding variability in submersion time, which should be taken into account in future modeling of annuals in freshwater tidal wetlands. We conclude that population models which include detailed representations of the spatial and temporal heterogeneity of environmental drivers can deliver important general insights even if they must be tied to specific study sites.