Investigating CO2 and CH4 Fluxes Across a Heterogeneous Restored Tidal Salt Marsh in the South San Francisco Bay, California, Using Eddy Covariance, Chamber, and Porewater Measurements

Interpretation of net ecosystem exchange (NEE) measurements of CO2 and CH4 in tidal wetlands is challenging due to effects from tidal activity and heterogeneous landscapes. In this study we are combining the use of eddy covariance (EC), soil chamber (SC), and porewater measurements to understand sources and sinks of CO2 and CH4 and improve NEE partitioning algorithms. This study takes place at a restored tidal salt marsh along the San Francisco Bay, California, consisting of three main cover types: Spartina foliosa, Salicornia pacifica, and bare mud flats. Three years of EC data show high net removal of CO2 (avg= -425 g C-CO2 m-2 yr-1, SD=45) and low CH4 emission (avg=0.5 g C-CH4 m-2 yr-1 SD=0.3). The high net carbon (C) removal at this marsh is not due to high photosynthetic rates, but extremely low respiration. By using SC measurements, we determined average daytime soil surface respiration from each cover type. Then, using generalized additive modeling (GAM), we identified predictor variables and modeled the soil surface fluxes over a one-year time period. The modeled soil respiration was tested as a predictor variable in an artificial neural network (ANN) partitioning approach. The best ANN model has an R2 value of 0.24 for about three years of data with an average R2 of 0.46 during the growing seasons. Traditional partitioning approaches such as the Reichstein and Lasslop approaches had R2 values of 0.10 and 0.01, respectively. Partitioning performance increased when water table depth variables were included, highlighting the significance of tidal activity on C exchange. The SC measurements also revealed that the average daytime soil surface emission of CH4 is greatest from soils in S. foliosa areas (7.30 nmol m-2 s-1, SE= 2.21) as hypothesized, however the highest CO2 emissions were from mud flat soils (0.861 mol m-2 s-1, SE=0.172), contrary to previous findings. After restoration, mud flat soils showed a coarsening of sediment particle size, potentially allowing for increased oxidation of organic matter and CH4. We plan to measure porewater concentrations and isotopic composition of CO2 and CH4 within the different covers to determine what processes are driving the emissions. The ultimate goal will be to improve the ability to model ecosystem exchange of CO2 and CH4 across restored heterogenous tidal wetland ecosystems.