Analyzing spatial variability of drivers of coastal wetland loss in the northern Gulf of Mexico using Bayesian multi-level models

A modeling framework that can quantify the impact of multiple potential drivers on coastal wetland loss while estimating its uncertainties is needed to better conserve and restore these valuable ecosystems. We developed a Bayesian multi-level modeling framework to predict the areal wetland loss in the northern Gulf of Mexico, driven by relative sea-level rise (RSLR), vegetation productivity, tidal range, wave height, and coastal slope. We further investigated how the impact of these site or local-scale biogeophysical factors on wetland loss differed by watershed unit. The results indicate the importance of accounting for watershed boundaries where the coastal wetlands are located when evaluating their loss. The effects of RSLR, wave height, and tidal range on wetland loss differed by watershed. Inclusion of streamflow and land use/land cover at the watershed scale did not improve model predictions while other relevant watershed-scale covariates, such as sediment availability, are not available. Based on the 95% credible intervals of the posteriors of the model coefficients, there were only five watersheds among the 14 studied where RSLR, wave height, or tidal range, showed positive or negative effects on coastal wetland loss. The results demonstrate the complexity of modeling wetland loss processes as no single dominant driver was identified for most of the watersheds. Furthermore, the need to account for multiple drivers simultaneously to ensure the effectiveness of costly wetland restoration projects clearly appears warranted when evaluating wetland loss processes over a broad spatial scale.

Automated summary

A Bayesian multi-level modeling framework was developed to predict wetland loss in the northern Gulf of Mexico, highlighting the importance of considering both biogeophysical factors and watershed boundaries in evaluating coastal wetland loss processes.