Probabilistic forecasting of coastal wetland elevation and soil carbon response to sea-level rise
Tidal wetlands build soil as a dynamic response to inundation, resulting in carbon sequestration and resilience to sea-level rise. Process models can be used to make site-specific forecasts, anticipate future changes, and plan management responses. But to be actionable, forecasts from these models need to be accompanied by descriptions of their confidence. In this study, we demonstrate the application of ecological forecasting principles to the unique case of tidal wetland vulnerability and carbon storage, given accelerating and uncertain sea-level rise. We present a case study using the Cohort Marsh Equilibrium Model at the Global Change Research Wetland in Maryland, USA. We used meta-analyses to quantify lab- and sample-level measurement uncertainties for common soil measurements and to develop Bayesian model priors. Reanalysis of previous literature supported predictive relationships between belowground turnover and mean annual temperature, as well as between soil erodibility and suspended sediment concentration that could be transferable to other locations. The model forecasted a likely conversion of low-elevation marsh to open water by the 2060s, accompanied by a decrease in marsh carbon storage capacity. Forecasted uncertainty in elevation was dominated by structural uncertainty in the model, while carbon accumulation rates were more influenced by parameter uncertainty. After 2050, and increasingly as the forecast approached 2100, uncertainty in sea-level rise projections contributed to uncertainty in both forecasts. Forecasts were most sensitive to processes related to organic matter preservation, plant productivity (maximum aboveground biomass and belowground turnover), and organic soil packing density. The soil formed was an organic-rich peat (>70% organic matter according to the data), and model predictions were insensitive to parameters associated with mineral sedimentation. We identify additional field data collection, such as on decomposition trajectories and plant productivity, and data-model coupling of near-term elevation records as the most impactful strategies for reducing uncertainties in future forecasting efforts. Future work can build on this forecasting framework in other locations, using it for adaptive management, identifying appropriate interventions to mitigate wetland loss, and projecting future carbon removal capacity in response to accelerated sea-level rise.