The reduced atmospheric CO₂ and land biota (e.g., Crawley, 1995) clearly indicates that the oceanic storage of carbon was greater during the last glacial period than today. For the cause, several mechanisms to increase the efficiency of "biological pump" have been proposed including intense supply to the surface ocean of either iron through the atmosphere (Martin, 1990) or nutrients due to change in high-latitude deep water formation (e.g., Knox and McElroy, 1984; Sarmiento and Toggweiler, 1984). However,  the increased organic carbon production, by itself, is insufficient to explain the lowered atmospheric CO₂.  This is because, in order to maintain the atmospheric CO₂ concentration at 190-200 ppm, alkalinity and pH in the surface ocean must be higher by ~85 µeq/L and ~0.14 units, respectively, than those in the pre-industrial period according to the thermodynamic equilibrium (see Figure 1; Nozaki and Oba, 1995). This pH shift seems to be supported by recent measurements on boron isotopes in foraminifera (Sanyal et al., 1995). Thus, some change in carbonate chemistry of the ocean must be involved (Boyle, 1988; Broecker and Peng, 1989; Archer , 1996).

          Diatoms dominate phytoplankton blooms in the ocean, since they utilize nitrate readily and often preferentially (Dugdale et al., 1995). Coccoliths must also play a significant role in carbon cycling in the ocean, because of their importance in abundance as indicated by the average production ratio of biogenic opal : carbonate = 2 : 1 (Broecker and Peng, 1982). Furthermore, they are involved into the reaction of carbonate chemistry given by,

Thus, growth of coccolith shells tends to increase not only dissolved inorganic carbon in the deep sea through their sinking and dissolution but also atmospheric CO₂ through gas exchange across the sea surface.
            Phytoplankton grow by utilizing nitrate supplied from the sub-surface waters underneath the seasonal thermocline either by diffusion and upwelling or winter convection. In the presence of dissolved Si (at > 2 µM; Egge and Aksenes, 1992; Dunne et al., 1999), diatoms bloom first. When dissolved Si is depleted and, coccoliths and dinofragelates can dominate in utilizing remaining nitrate and iron. Thus, in the regions of high-nitrate and Si-depleted regimes like the sub-Antarctic and the North Atlantic where the vertical concentration gradient ratio of DSi/DNO₃ is much lower than 1.0 mol mol^-1 (Brezezinski and Nelson, 1995), diatoms can readily be replaced by coccoliths. This is consistent with the observation of the particulate flux by using a sediment trap (Honjo, 1997), revealing that the North Pacific is dominated by opal production whilst the North Atlantic is dominated by carbonate production.
           The increased supply of atmospheric dust to the surface oceans during the glacial periods was marked as deduced from the ice core records (Royer et al., 1983; DeAngelis et al., 1987). Dissolution of Si and iron from eolian dust into the surface waters in the high-nitrate and presently low-Si regime would enhance diatom production. Then, coccoliths must compete with diatoms for available nitrate and would therefore be inhibited for their bloom. This change would lead to an increased organic C/CaCO₃ rain ratio to the deep sea, resulting in increases in pH and alkalinity of surface waters and thereby reducing atmospheric CO₂. Archer and Maier-Reimer (1994) have shown that about 40% decrease in calcite production is sufficient to derive atmospheric CO₂ down close to glacial values. If this reduction of carbonate production in the surface ocean is compensated thoroughly by diatom production, it would lead to an increase in the opaline flux by ~20 % during the glacial period. This "alkalinity pump" hypothesis was first described in Nozaki and Oba (1995) and recently followed by Harrison (2000).
           Charles et al. (1991) studied the opal and carbonate contents in the deep-sea sediment cores collected from the Southern Ocean, and indicated that the opal contents mainly reflect the productivity change in the surface ocean, despite of possible changes of opal contents by other factors such as dissolution in the deep sea and sediment pore waters, bulk sediment accumulation rate, etc. They also concluded that the biological productivity in the Southern Ocean during the glacial period did not increase as a whole to the extent that the dynamic biogeochemical models for glacial-interglacial pCO₂ change (e.g., Sarmiento and Toggweiler, 1984) predict. The sub-Antarctic cores, V22-108 and RC11-120 collected in the Atlantic and Indian sectors, respectively, north of the Polar Front, beautifully shows opal/CaCO₃ change suggested by the silicon-induced "alkalinity pump" scenario (Figure 2a). The carbonate contents in the cores are high during the inter-glacial and low during the glacial period, in agreement with a consequence of carbonate chemistry that high production of carbonate must accompany an increase of atmospheric CO₂ through gas exchange across the air-sea interface (see, equation 1). There is a strong inverse relationship between opal and carbonate content (Figure 2b), suggesting that plankton species (diatoms versus coccoliths) change depending upon the availability of dissolved Si in the surface ocean may explain the variation of glacial and inter-glacial pCO₂ without significant change in productivity as a whole. We note that sedimentary carbonates contains not only coccoliths but also significant amount of foraminifera. However, preferential dissolution of small-sized coccoliths in the water column and sediment pore waters tends to protect the large-sized foraminifera against further dissolution. This would lead to a consequence that high coccolith production in the surface water is reflected in the high carbonate burial rate, just like the case of opal in the cores as discussed by Charles et al. (1991). Otherwise, it seems difficult to explain the tight inverse correlation between carbonate and opal contents (Figure 2b).

             Another aspect of the core data is that between 19.4 and 24.1 kyr, the carbonate plus opal content is only 20-25%, but reaches ~90% for the last ~10 kyr. Assuming 2.7 g/m³ as mineral density and 0.75 for the sediment porosity, these correspond to the non-carbonate and non-opal fluxes of 4.3 mg cm^-2yr^-1 during the last glacial maximum and 0.37 mg cm^-2yr^-1 during the Holocene. An order of magnitude higher detrital material flux in the former is not surprizing, considering that, during the glacial period, the atmospheric dust flux markedly inceased as recorded in the ice cores and also errosion of coastal sediments exposed by lowered sea level might be enhanced. Both of these sources would increase iron and dissolved Si in surface waters and lend added support to the Si-induced alkalinity pump. It should be remembered that the mechanism takes effect only in the Si-depleted oceanic regions like the sub-Antarctic and the North Atlantic, and therefore, the global calculation done by Harrison (2000) is difficult to prove the effect.
             This alkalinity pump mechanism described above does not suffer from the critical issues raised for the previous models. First, it does not cause any large-scale dissolved oxygen depletion in the deep sea, a problem in most current biological pump models of atmospheric CO₂ variability (e.g., Boyle, 1988; Broecker and Peng, 1994). Secondly, such plankton species change in the surface ocean can occur rapidly enough to explain the sharp rise and fall of atmospheric CO₂ within a few decades as recorded in the ice cores, which would have been obscured if the variation of carbonate dissolution in the bottom of the ocean was a primary cause (Boyle, 1988; Archer and Maier-Reimer, 1995).
 

Reference
Archer, D., An atlas of the distribution of calcium carbonate in sediments of the deep sea, Global Biogeochem. Cycles, 10, 159-174, 1996.
Archer D. and E. Maier-Reimer, Effect of deep-sea sedimentary calcite preservation on atmpspheric CO₂ concnentration. Nature, 367, 260-263, 1994.
Boyle, E. A., The role of vertical fractionation in controlling Late Quaternary atmospheric carbon dioxide, J. Geophys. Res., 93, 15701-15714, 1988.
Broecker, W. S., and T. H. Peng, Tracers in the Sea, 690pp., Eldigio Press, Palisades, New York, 1982.
Broecker, W. S., and T. H. Peng, Greenhouse puzzles. Part II. Archer's world: CO₂'s glacial hideout? 47pp., Eldigio Press, Palisades, New York, 1994.
Broecker, W. S. and T. H. Peng, The cause of the glacial and interglacial atmospheric CO₂ change: A polar alkalinity hypothsis, Global Biogeochem. Cycles, 3, 215-240, 1989.
Brzezinski, M. A., and D. M. Nelson, The annual silica cycle in the Sargasso Sea near Bermuda, Deep-Sea Res. I, 42, 1215-1237, 1995.
Charles, C. D., P. N. Froelich, M, A. Zibello, R. A. Mortlock and J. Morley, Biogenic opal in Southern Ocean sediments over the last 450,000 years: Implication surface water chemistry and circulation, Paleoceanogr., 6, 697-728, 1991.
Crawley, T. J., Ice age terrestrial carbon changes revisited, Global Biogeochem. Cycles, 9, 377-389, 1995.
DeAngelis, M., N. I. Barkov, and V. N. Petrov, Aerosol concentrations over the last climatic cycle (160 kyr) from an Antarctic ice core, Nature, 325, 318-321, 1987.
Dunne, J. P., J. W. Murray, and A. K. Aufdenkampe, Silicon-nitrogen coupling in the equatorial Pacific upwelling zone, Global Biogeochem. Cycles, 13, 715-726, 1999.
Dugdale, R. C.,  F. P. Wilkerson, and H. J. Minas, The role of the silicate pump in deriving new production, Deep-Sea Res. I, 42, 697-719, 1995.
Egge, J. K. and D. L. Aksenes, Silicate as regulating nutrient in phytoplankton competition.  Mar. Ecol. Prog. Ser.  83, 281-289, 1992.
Harrison, K. G., Role of increased marine silica input on paleo-pCO₂ levels. Paleocanography, 15, 292-298, 2000.
Honjo, S., The rain of ocean particles and Earth's carbon cycle, Oceanus, 40, 4-7, 1997.
Jenkins, W. J., Nitrate flux into the euphotic zone near Bermuda, Nature, 331, 521-523, 1988.
Knox, F. E., and M. McElroy, Changes in atmospheric CO₂; influence of marine biota at high latitudes. J. Geophys. Res., 89, 4629-4637, 1984.
Martin, J. H., Glacial-interglacial CO2 change: The iron hypothesis, Paleoceanogr., 5, 1-13, 1990.
Nozaki, Y. and T. Oba, Dissolution of calcareous tests in the ocean and atmospheric carbon dioxide. In: Biogeochemical Processes and Ocean Flux in the Western Pacific,  H. Sakai and Y. Nozaki, editors., pp.83-92, TerraPub., Tokyo, 1995.
Sarmiento, J. L., and R. Toggweiler, A new model for the role of the oceans in determining atmospheric pCO₂. Nature, 308, 621-624, 1984.
Sanyal, A., G. Hemming, and W. S. Broecker, The pH of the glacial ocean as reconstructed from boron isotope measurements on foraminifera, Nature, 308, 621-624, 1995.