Improving soil carbon estimates of mudflats in Araca Bay using spatial models that consider riverine input, wave exposure and biogeochemistry

A persistent challenge to managing coastal wetlands in a way that will mitigate climate impacts is understanding how soil organic carbon (Corg) stocks change over spatial and temporal scales. This study tested how seasonal change in riverine flows and soil biogeochemical attributes (grain-size, degree of wave disturbance and microphytobenthic productivity) can influence the Corg stock estimates of an unvegetated tidal flat within a Brazilian subtropical bay. We used a Generalized Additive Modelling (GAM) approach to accurately predict and map subtle changes in the Corg pools of these expansive, but often overlooked coastal ecosystems. The most important predictors across seasons were sediment grain-size and wave exposure. Seasonally important covariates were riverine flow and microphytobenthic productivity (during the warm and wet summer season). We then compared the high-resolution map outputs derived from models with more traditional approaches (interpolation and extrapolation) to see how these different methods might influence estimates of the carbon stocks for the entire bay (i.e., an area covering 115 ha). Modelled estimates for the bay (1945 +/- 45 Mg Corg) where more conservative and had lower associated error than interpolated (2196 +/- 185 Mg) and extrapolated estimates (2535 +/- 117 Mg). Given the current urgency to understand the carbon sequestration potential of coastal ecosystems, the present findings are important and draw attention to the often overlooked, but important carbon pools of intertidal mudflats.