Coastal seas constitute a link between land and the open ocean, and therefore play an important role in the global carbon cycle. Large amounts of carbon, of both terrestrial and marine origin, transit and are transformed in these waters, which belong to the more productive areas of the oceans. Despite much research has been done on the subject, there are still many unknown factors in the coastal sea carbon cycling. 

This doctoral thesis investigates the carbon dynamics in the Baltic Sea, with a focus on the production and fate of marine and terrestrial organic carbon and its influence on the air-sea CO₂ exchange in its northernmost part, the Gulf of Bothnia. The main approach is the use of a coupled 3D physical-biogeochemical model, in combination with a long series of measurements of physical and biogeochemical parameters. 

A new coupled 3D physical-biogeochemical model, which includes the stoichiometric flexibility of plankton and organic matter, is set up for the Gulf of Bothnia. It is found that phytoplankton stoichiometric flexibility in particular, with non-Redfieldian dynamics, is key to explaining seasonal pCO₂, dissolved organic carbon (DOC), and nutrient dynamics. If the Redfield ratio is instead used to predict organic carbon production, as done in most biogeochemical models currently in use, the uptake of atmospheric CO₂ is reduced by half. Furthermore, it is shown that the organic carbon production needed to reproduce the summer pCO₂ drawdown is larger than measured estimates of primary production. This discrepancy is attributed to a substantial production of extracellular DOC, which seems not to be captured by measurements. 

The dynamics of terrestrial dissolved organic carbon (tDOC) is studied by the use of a passive tracer released from rivers into the physical model of the Baltic Sea. It is found that 80% of the tDOC released in the Baltic Sea is removed, and the rest is exported to the North Sea. Two different parameterisations of tDOC removal are tested. In the first one a decay rate with a timescale of 1 year applied to 80% of the tDOC, and the remaining 20% is assumed to be refractory. In the second one a decay rate with a timescale of 10 years applied to 100% of the tDOC. Trying these parameterisations in a full biogeochemical model shows that only the one with the faster decay is able to reproduce observations of pCO₂ in the low-salinity region. A removal rate of one year agrees well with calculated removal rates from bacterial incubation experiments, indicating that bacteria have the potential to cause this remineralisation. It is not only remineralisation of tDOC that affects the pCO₂; it is also suggested that a strong tDOC induced light extinction is needed to prevent a too large pCO₂ drawdown by phytoplankton in the low salinity region.