The presented work examines in detail the carbon cycle of the North Sea and the main drivers of the interannual variability of the carbon fluxes by using the ecosystem model ECOHAM. This study consists of three main parts: In the first part, several long-term simulations with different scenarios conducted with ECOHAM were analysed. The main goal of these simulations was to examine the driving mechanisms of the interannual variability of the carbon shelf pump and especially of the air-sea flux of CO2 between ocean and atmosphere. It was found that the export of carbon into the North Atlantic via shelf pumping was mainly correlated with the amount of water that was exported into the North Atlantic. During the whole simulation period from 1970 to 2006, the North Sea was found to be a sink for atmospheric CO2 and absorbed in the mean about 1.31 mol C m-2 yr-1. The northern North Sea was responsible for most of the uptake, whereas the southern North Sea oscillated between being a sink or a source. During the last ten years of the simulation period, the uptake of atmospheric CO2 declined visibly. This trend in the simulation was mainly due to the increasing surface temperature of the North Sea and a decline in pH. During a long-term simulation without biological processes the North Sea was an increasing source for atmospheric CO2 due to river input of inorganic carbon, indicating that biology is the most important factor for the North Sea being a sink for atmospheric CO2. Within the simulation period, the North Sea was subject to a strong phosphate eutrophication. During this eutrophication, the primary production in the model increased, while the net ecosystem production, thus the primary production minus respiration, did not react. A biologically induced air-sea flux was calculated and it was shown that the biological influence of the air-sea flux can be attributed to the net ecosystem production. This explained, why the eutrophication phase had no visible impact on the air-sea flux. Furthermore, it was found that the biological and physical effects on the air-sea flux counteracted in some years, e.g in 1996, resulting in an intermediate net effect for the air-sea flux: In 1996, low surface temperatures favoured the dissolution of CO2, whereas the low temperatures and low river loads of nitrogen induced a low net ecosystem production. In the sum, both extreme values were balanced against each other. All in all, this part of the study gives insight into the most important processes driving the variability of the carbon fluxes of the North Sea. In the second part, the method to differentiate between the physical and biological impact on the air-sea flux, which was already used in the first study, was compared with another method to find the most appropriate approach to calculate the influence of biology on the air-sea flux. For the first method, the assumption was made that biological and physical effects on the air-sea flux add up. Therefore, long-term simulations with and without biology were conducted and the difference between the two simulated pCO2 values was taken as the biologically influenced pCO2. The second method was introduced by Takahashi et al. (1993), who linearised the temperature effect on pCO2 by using experiments and statistical analysis and thus calculated a linearisation coefficient to estimate the changes in pCO2 that are not due to temperature. The methods were compared by applying them at two stations in the North Sea, one in the northern and one in the southern area. The weak points in the superposition assumption were due to the fact that changes in temperature induce different changes in pCO2 at different pCO2 levels, while the Takahashi-approach does not account for external sinks or sources, as e.g. river inputs, in the area where the calculation is made. To account for different regional properties, horizontally different linearisation factors were calculated and displayed. Despite their respective weaknesses, results of both methods agree quite well. In the third part, a special aspect of the marine carbon cycle was implemented in the model and its impact on the carbon fluxes was analysed. The concept of the carbonate counterpump describes the release of CO2 into the water column due to biogenic calcification. The released CO2 may affect the air-sea flux and the carbon shelf pump. The reaction of calcification due to climate change and ocean acidification is currently under debate. In the North Sea, the calcification mainly occurs in the northern part and is attributed to coccolithophores. The coccolithophores were implemented in ECOHAM together with a prognostic representation of calcification and Total Alkalinity. The simulation results were compared with data from a ship expedition. The main goal of this study was to analyse the impact of the coccolithophores on the carbon cycle of the North Sea and the carbon shelf pump. It was found that the calcification reduced the uptake of atmospheric CO2 by the North Sea. But this reduction was lower than would be expected when regarding the amount of calcification because of the increased export of particulate inorganic carbon, which was stored in the deep water and in the sediment. In some scenario simulations with doubled atmospheric pCO2, a reduced calcification and a reduced export of particulate inorganic carbon was found. The decrease in calcification is a negative feedback on rising atmospheric pCO2, while the decrease in export is a positive feedback on rising atmospheric pCO2. In connection to the sinking of coccoliths, the ballast hypothesis was developed, which assumes that the sinking of organic matter is enhanced by coagulation of ballast minerals and organic material. The influence of this ballasting material on zooplankton fecal pellets was found to be very small in a sensitivity simulation. The export of carbon into the North Atlantic decreased by 10 % in a simulation with coccolithophores in comparison to a simulation without coccolithophores. In summary the model results point into the direction of a decreased efficiency of the carbon shelf pump due to the presence of coccolithophores. After these three chapters, some additional questions were discussed, for which there was no space in the three manuscripts. The influence of salinity on the gas exchange was investigated and the results of a long-term simulation with constant temperature and a long-term simulation with constant atmospheric pCO2 were discussed to further investigate the different impact factors on the air-sea flux. It was shown that the influence of the variability in salinity does play a minor role for the interannual variability of the air-sea flux. Further, it was found that at constant atmospheric pCO2, the decreasing trend in the air-sea flux remains, but the magnitude of the air-sea flux was lower. The temperature thus determined the trend, whereas the atmospheric pCO2 determined the strength of the air-sea flux. In summary, it was found that the most important factors for the variability of the air-sea flux were temperature, pH and net ecosystem production. The export into the North Atlantic was determined by the export of water. The biological contribution to the air-sea flux could be calculated by either assuming a superposition or by a linearisation of the temperature effect, depending on the local characteristics. Further, it was found that the carbonate counterpump slightly weakened the carbon shelf pump.