Determining the change of sea surface CO2 fugacity ( fCO2) is important as the fCO2 gradient between the atmosphere and the ocean dictates the direction of CO 2 flux and the fate of this greenhouse gas. While substantial efforts have been dedicated to the study of fCO2 trends in the open ocean, little is known regarding how fCO 2 levels change in ocean margins. Meanwhile, hypoxia (i.e., dissolved oxygen concentration, or DO, less than 2 mg L--1) is becoming an increasing global threat in coastal areas. Elucidating the carbon sources that consume DO is important because it helps to make proper mitigation plans.

In Chapter II, I used a newly available, community-based global CO 2 database (Surface Ocean CO2 Atlas version 3) to develop a new statistical approach based on Generalized Additive Mixed Modeling (GAMM) to interpret oceanic fCO2 changes in ocean margins. This method utilized Julian day of year, sea surface salinity, sea surface temperature, and sampling date as predictors. Using the GAMM method, I was able to derive multi-decadal fCO2 trends with both improved precision and greater robustness to data gaps compared to the existing method. In Chapter III, I used the GAMM method on global ocean margins (within 400 km from the shore and 30 °S--70 °N) and found that fCO2 trends closely followed the atmospheric fCO2 increase rate. Further analysis suggested that fCO2 trends in Western Boundary Current- and Eastern Boundary Current-influenced areas differed in response to thermal (temperature) and nonthermal (chemical and biological) effects. These differences were due to heterogonous physical, chemical, and biological responses under climate change forcing, leading to divergent trends in CO2 sinks and sources among different ocean margins.

To address the hypoxia formation mechanism question, I adopted the stable carbon isotope (delta13C) of dissolved inorganic carbon (or DIC, the end product of organic carbon degradation) as a proxy to trace back the delta13C of remineralized organic carbon that was responsible for DO consumption in the northern Gulf of Mexico (Chapter IV) and two semi-arid coastal bays in south Texas (Baffin Bay and Oso Bay) (Chapter V), the two areas that both experience seasonal bottom water hypoxia. My findings suggested that terrestrial carbon contributed to oxygen consumption in limited extent and mostly focused in areas where river water influence was significant in the northern Gulf of Mexico, while for the vast shelf areas marine-produced organic carbon was the dominant contributor to hypoxia formation. In Baffin Bay and Oso Bay, however, phytoplankton, seagrass/marsh organic carbon, and refractory terrestrial organic carbon all contributed to the DO loss under different hydrological conditions.

This study provided a comprehensive data-driven analysis on ocean margin fCO2 changes on a multi-decadal timescale and revealed different behaviors of the two types of boundary current-dominated systems. Regarding the hypoxia formation mechanism in the different coastal and estuarine environments, my study suggested that eutrophication remained the top stressor that could lead to hypoxia formation. Therefore, sustained efforts that focus on reducing nutrient pollution should still be carried out to mitigate the hypoxia stress for the both ecologically and economically important coastal and estuarine systems.