Predicting the fate of coastal wetlands in the face of sea-level rise requires a comprehensive understanding of wetland biogeochemistry, as there are strong feedbacks between plant productivity, sea level, and wetland surface elevation. Notably, there is a parabolic relationship between flooding and plant biomass, and biomass contributes to building surface elevation of a wetland relative to sea level. A host of existing deterministic models capture these feedbacks and predict critical tipping points at which plants are flooded too much and no longer contribute to building surface elevation, leading to wetland collapse. There is a breadth of existing data important to parameterizing these models, including paired elevation and biomass estimates, that are now increasingly publicly available. Fusing these data and models in an ecological forecasting framework promotes a more holistic understanding of biogeochemical processes, allows for forecasts to be made with uncertainty, and provides insight to where uncertainty arises from. Forecast uncertainty is partly a function of how sensitive a model is to parameters and the uncertainty in estimates of those parameters. Here, we assess how uncertainties in predictions of coastal wetland stability are related to sensitivities to biomass-related parameters using an existing process model, and uncertainties in those parameters based on existing data collected across the Chesapeake Bay region.
Results/Conclusions
Coastal wetland stability is sensitive to parameters that explain relationships between flooding depth and plant biomass. For example, a 20% increase in maximum biomass production at optimal flooding depth can lead to a 24.7% increase in predicted surface elevation of the wetland over 100 years. Increases in the flooding tolerance range of plants by just 20cm can result in a 4.9% increase in predicted surface elevation. These sensitivities are large enough to change our predictions of the year of wetland collapse (i.e. no biomass production) by over a decade. Estimating parameters related to the relationship between flooding and plant biomass is tractable, but results in different uncertainties based on how data are collected. Qualitatively, fitting data collected from mesocosm flooding experiments results in reduced parameter uncertainty as compared to fitting data collected from observational field surveys. We suggest that uncertainty in forecasts of wetland stability are partly the result of the sensitivities and uncertainties of these biomass parameters and could be reduced by more directed data collection. Better understanding where and how this uncertainty arises will serve as important guidance to continued model development and field surveys.