The greenhouse gas concentrations have varied nearly synchronously with temperature and ice volume over the last 600 ka at least (Fig. 5.20) with the difference between interglacial and glacial periods reaching about 80 ppm for carbon dioxide and 300 ppb for methane. This corresponds to a radiative forcing of nearly 3 W m-2 and thus to a strong amplifying mechanism for the cooling during glacial periods. However, as mentioned in the latest IPCC report, "the qualitative and mechanistic explanation of these CO2 variations remains one of the major unsolved question in climate research" (Jansen et al. 2007).
The land biosphere can not be responsible for this decrease in the CO2 concentration during glacial periods. Because of the advance of ice sheets, the land area available for vegetation growth declines significantly. Furthermore, the lower temperatures induce less evaporation over the oceans and less precipitation over land. The fraction of dry areas and desert, which only store a small amount of carbon compared to, for instance, forest was thus larger. All these factors lead to a decrease in the carbon storage over land which was not compensated for by the growth of terrestrial vegetation on the new land area associated with the lower sea level. As a consequence, changes in the land biosphere during glacial periods tend to increase the atmospheric CO2 concentration by an amount which is estimated to be around 20 ppm.
The cause of the decline must thus lie in the ocean, the geological processes being too slow to account for the observed changes. Because of the accumulation of freshwater with nearly zero dissolved organic carbon and alkalinity in the ice sheets, the salinity, DIC and Alk of the ocean increases. This leads to an increase of pCO2 in the ocean. However, it can be shown that this is overoutweighed whelmed by the higher greater solubility of CO2 in the ocean due to cooling. The net effect is a small decrease of the atmospheric concentration of CO2, but it is insufficient to explain all the 80 ppm decrease between interglacial and glacial periods.
This decrease must therefore be related to changes in the ocean circulation and/or the soft tissue and carbonate pumps. All these factors have a large influence on the distribution of DIC and Alk in the ocean and thus on the ocean-atmosphere CO2 exchanges. Most hypotheses emphasises the role of the Southern Ocean. A strong argument in favour of this is the very similar evolution of atmospheric CO2 concentration and Antarctic temperatures (Fig. 5.20). At present, there is a strong upwelling of deep water, rich in nutrients and DIC, in that area. Biological activity is insufficient to fix the excess carbon and some the carbon coming from the deep ocean is transferred to the atmosphere. If in glacial periods, this upwelling (and more generally the connection between surface and deep water) or the biological production had changed, this would have a considerable influence on the concentration of CO2.
The upwelling might have been reduced at the LGM by a northward shift of the westerlies in the Southern Ocean and thus by the divergence associated with the wind-induced Ekman transport, but this still needs to be confirmed. The weaker hydrological cycle during cold periods, and the associated increase in the Earth's surface covered by dry areas, had probably lead to a greater dust transport towards the Southern Ocean. This had brought a large amount of iron to the Southern Ocean. As a consequence, biological production might have been higher during glacial times as this micro-nutriment strongly limits the primary production for the present-day condition in the area. Both of these effects could thus have played a role in the observed atmospheric CO2 concentration decrease.
It has also been suggested that the supply of iron to the Southern Ocean by dust have induced a large-scale shift in the ecosystem from phytoplankton producing calcium carbonate towards species which do not form CaCO3. This would have decreased the intensity of the carbonate pump, so inducing a decrease in the CO2 concentration.
Many other explanations have been suggested, but it seems that, on its own, none of them can explain the 80 ppm change. It is very likely that some of them play an important role, while others have a negligible influence. However the relative importance of the various explanatory factors is presently unknown.