A modelling approach to disentangle DOC release processes from riparian wetlands

Release of dissolved organic carbon (DOC) from soils into pore waters can occur via three distinct processes: (i) microbial decomposition of soil organic matter, producing soluble organic molecules; (ii) desorption of DOC from soil mineral surfaces, such as oxyhydroxides and clays, driven by increasing pH in pore waters and (iii) reduction of ferric iron in waterlogged soils, leading to the dissolution of amorphous Fe(III) oxyhydroxides that were previously coprecipitated with dissolved organic matter. Once released into pore waters, DOC can either be mineralised or exported to surface waters.

Riparian wetlands are important sources of organic matter and sites where all three DOC release mechanisms may occur. DOC release via microbial decomposition (process i) is typically associated with a rapidly cycling DOC pool that fuels microbial mineralisation, whereas desorption and reductive dissolution processes (ii and iii) are often linked to DOC mobilisation and export of a potentially more stable, mineral-associated pool of soil organic matter. Despite their importance, the relative contributions of these processes and their controlling factors remain incompletely understood. Improved understanding of these mechanisms may provide insight into the extent to which these processes are likely to influence the long-term carbon storage capacity of wetland soils.

Against this background, we propose and demonstrate a modelling approach to disentangle the processes and drivers of seasonal DOC mobilisation in riparian pore waters of the Kervidy-Naizin Critical Zone Observatory in western France. First, a principal component analysis was applied to weekly to biweekly measurements (period from November 2022 until May 2023 and 17 sites) of pH, nitrate, DOC, soluble reactive phosphorus, ferrous iron, five variables describing dissolved organic matter composition based on fluorescence properties, and hydrological variables. Scores of the first two principal components - interpreted as proxies for DOC desorption and reductive dissolution of coprecipitates - were extracted. Second, these component scores, together with two additional variables assumed to represent lateral DOC leaching and microbial decomposition of soil organic matter, respectively, were used as predictors in a generalized additive model (GAM) of DOC concentrations. Third, the GAM was used to quantify the relative contributions of the four processes to DOC increases.

Our analysis suggests that desorption was the dominant process responsible for DOC release during winter and spring in the studied riparian zones of the Kervidy-Naizin catchment. These results demonstrate that disentangling the processes contributing to seasonal DOC mobilisation in riparian soils - such as pH driven desorption and reductive dissolution of coprecipitates - is feasible using a combined multivariate and additive modelling approach. To further improve the quantification of individual process contributions, additional measures of DOC quality, for example derived from FT-ICR MS, are likely to be beneficial.