High-frequency dynamics of dissolved carbon fluxes from machine learning and their contribution to carbon budgets in tidal marsh wetlands in the Yangtze River Estuary, China

Accurately quantifying the carbon budgets of coastal wetlands is crucial; however, methodological constraints have limited our understanding of lateral carbon exchange. In this study, we take advantage of machine learning algorithms to develop a high-frequency estimation method for the dissolved carbon flux from tidal marshes to offshore areas based on discrete sampling for dissolved carbon concentration and long-term high-frequency data for water environmental parameters and water fluxes. We quantified the lateral carbon fluxes in two tidal marshes with different tidal characteristics in the Yangtze River Estuary, China, and elucidated their seasonal and tidal-cycle dynamics. Accordingly, we simultaneously observed the lateral and vertical (between wetlands and atmosphere) carbon fluxes to improve our understanding of how lateral carbon flux contributes to the carbon budget of tidal marsh wetlands. The results showed that lateral carbon flux constituted a major component of the coastal carbon cycle and accounted for >85% of the total carbon budget. The dissolved inorganic carbon outflux is an order of magnitude greater than the dissolved organic carbon. The dissolved carbon export varied significantly with tidal current and seasonal changes. The majority of this export occurred in summer and autumn. The lateral carbon outputs at the Jiuduansha (JDS) and Chongming Xisha (CMXS) sites were 608.11 ± 32.87 and 717.15 ± 16.27 g C m⁻² yr⁻¹, respectively. Incorporating these values into the tidal marsh wetland carbon budget yielded net ecosystem carbon balances of 47.96 ± 35.96 and 85.85 ± 22.55 g C m⁻² yr⁻¹ for the JDS and CMXS sites, highlighting the role of tidal marshes as carbon sinks despite high outputs. The results of this study not only reduce the uncertainty in capturing the dynamics of lateral carbon exchange but also provide novel insights into the quantification of carbon budgets and a deeper understanding of carbon cycling processes within tidal marsh ecosystems.