The solubility of calcite (CaCO3) in sea water increases with pressure,
so that the ocean is typically supersaturated at shallow and intermediate
depths and undersaturated in the deepest waters. Only a fraction of the global
calcite production is buried, and this proportion depends on the area of the
sea floor that is shallower than the depth of calcite saturation. Any
imbalance in the sources (terrestrial weathering and alteration) and sinks
(deep sea and shallow water deposition) for CaCO3 will change the ocean
dissolved carbonate ion concentration ([CO3=]) in the direction of
restoring throughput balance. Through this mechanism, called "CaCO3
compensation" 2, 3, 4, the steady state deep
sea calcite burial rate is a slave to weathering less shallow water deposition,
which we refer to as the "deep sea carbonate influx". The pH equilibrium
reaction
maintains an inverse relationship between [CO3=] and pCO2.
We simulate the coupled ocean / sediment carbon cycle using the global ocean
circulation / carbon cycle model of Maier-Reimer7, 8, to
which we have coupled the sediment calcite dissolution model of
Archer6. Without the detailed sedimentary component the ocean model
is unable to drive the atmospheric pCO2 to lower glacial values without
violating paleoceanographic data7. Here we predict the ocean
alkalinity ([HCO3-] + 2[CO3=] + [B(OH)4-]) and
total CO2 ([CO2] + [HCO3-] + [CO3=]), rather than
specifying present-day values, based on the constraint that the deep sea
calcite burial rate equals the deep sea carbonate influx (which is imposed as a
source of alkalinity and total CO2 to the surface ocean, in the ratio of 2:1).
The deep sea alkalinity influx for the "standard case" (intended to mimic
present-day conditions) is 1.2 GT CaCO3 yr-1( Ref. 9, 10). The
model distributions of alkalinity and total CO2, and therefore the calcite
saturation state of deep water, compare well with oceanic observations (Figure
1a and Refs. 7 and 8). Although the glacial distributions of nutrients,
d13C, and sedimentary calcite were influenced by changes in
circulation, we use the present-day ocean circulation field throughout.
The calcite diagenesis model6 calculates the calcite dissolution
rate from the steady-state concentration profiles of the carbonate species CO2,
HCO3-, and CO3= in the diffusive sediment pore water.
Within the constraints of in situ microelectrode data11,
sediment trap data, pore water chemical data, and calcite accumulation rates,
the model predicts the observed patterns of calcite preservation in the
ocean6 ( Figure 1b). The steady state sedimentary calcite
distribution in the "standard case" model is compared with observations in
Figures 2a and 2b. The model calcite distribution is smoother than
observations because of the smoothness of the model topography.
One proposed explanation for lower glacial pCO2 is called the "coral reef
hypothesis"12, 13. In the present day ocean, up to half
of the total oceanic CaCO3 removal may be by growth of coral reef
complexes13. The rate of coral growth might be tied to
fluctuations in sea level relative to the level of the continental shelves,
with low coral growth rates during glacial times when the shelves are exposed
and surface temperatures are cooler. A shift of the "locus of CaCO3
deposition" to the deep sea would require a higher ocean [CO3=], and
hence a lower pCO212, 13. The steady state pCO2
predicted by our model appears to respond linearly to variations in the deep
sea carbonate influx (Figure 3a and 3b). In order for our model to achieve the
magnitude of the observed glacial / interglacial pCO2 changes, the glacial deep
sea calcite burial rate has to be 2-3 times higher than today's value. Under
these conditions, high-calcite sediments dominate the sea floor (Figure 2c).
This prediction is clearly at odds with the sedimentary record14,
15. Although a better test of the coral reef hypothesis would be a
time dependent calculation13, and we present a steady state, the
atmospheric pCO2 recorded by the Vostok ice core was depressed to the 180 - 220
uatm range for 50 kyr16. This time period is long enough for
sediments to have approached their steady state condition. Our results support
the conclusion that cycles in the global growth rate of corals would have a
significant impact on pCO2; however, we show that the coral-reef mechanism by
itself cannot account for the entire glacial / interglacial pCO2 signal.
We propose a new explanation for lower glacial pCO2 that relies in part on the
effect of sedimentary organic carbon degradation on calcite dissolution.
Diagenetic models for calcite dissolution predict that a significant fraction
of the calcite rain to the sediments dissolves in response to the addition of
CO2 to the pore water by oxic organic carbon degradation5,
6 (a process we will refer to as "respiratory calcite dissolution",
see Figure 1b and 1c). If sedimentary calcite dissolution is represented as
the sum of respiratory dissolution and dissolution driven by the overlying
water [CO3=], an increase in one, globally, will require a decrease
in the other to maintain steady state. Therefore an increase in the organic
carbon to calcite rain rate ratio to the sediments will drive an increase in
[CO3=], and a corresponding decrease in pCO2, through the mechanism
of CaCO3 compensation.
We test this hypothesis by altering the fraction of organic carbon and calcite
production that reaches the sea floor (Figure 3b, c). Since only a small
fraction of production reaches the ocean bottom, altering this fraction has
only a minor impact on the tracer source/sink functions in the water column.
In order to isolate the effect of respiratory calcite dissolution, we ran model
simulations both with and without the effect of respiratory calcite
dissolution. When respiratory calcite dissolution is neglected in the
simulations, we find that pCO2 is somewhat sensitive to calcite rain to the sea
floor, and insensitive to organic carbon rain, consistent with previous
studies17. In contrast, when respiratory calcite dissolution is
included in the simulation, pCO2 becomes much more sensitive to organic carbon
and calcite rain rates (Figure 3b-d, solid lines). We also simulate an
ecological shift from calcitic to silicious organisms by varying the production
ratio of calcite to organic carbon. For example, a decrease in calcite
production by 40% globally drives the pCO2 of the atmosphere near glacial
values.
The required shifts in organic carbon and calcite dissolution are qualitatively
supported by observations from glacial sediments. An increase in organic
carbon production is suggested by an increased benthic-planktonic gradient in
d13C18, increased sedimentary organic carbon
concentration19 and burial rates20, and a change in
foraminiferal assemblages21. In general, calcitic organisms are
replaced by siliceous diatoms in regions of high productivity and lower
temperatures22. Both conditions are thought to be generally more
prevalent under the glacial climate. In high latitudes, the front between
silica and calcite production moved equatorward23 during the
glacial. Thus we see a general increase in organic production, and a potential
shift from calcite toward siliceous production during the last glacial. We
conclude that, in contrast to the coral reef hypothesis, the requirements of a
"change in production" hypothesis seem to fit more comfortably within the
constraints of the available data.
The model predicts that the alkalinity and pH of the whole ocean were higher
during glacial time than at present. One test of this hypothesis will be
paleo-pH estimates derived from isotopic signature of boron24 in
sedimentary calcite (A. Sanyal and G. Hemming, personal communication).
Another test of the hypothesis can be made by comparing model time-dependent
behavior to atmospheric pCO2 recorded in ice cores. The ~10 kyr time scale of
data from Byrd1 is clearly longer than would be expected from a
glacial / interglacial change in surface ocean chemistry 25,
26 or ocean circulation17. The model experiment in
Figure 4 started at 18 kyr BP with a low value for calcite production relative
to organic carbon, and simulated the removal of carbon from the ocean by forest
re-growth (as inferred from the d13C values of benthic
foraminifera27) by removing 0.055 GT C yr-1 from the
atmosphere, from 17.5-7.5 kyr BP, for a total of 550 GT C. The model captures
the long term behavior of the observed CO2 record, but misses short-term
excursions (e.g. at 12 Kyr BP) that are probably associated with fast changes
in ocean circulation, sea surface temperature, or terrestrial biomass, that are
not simulated by the model. The model also neglects changes in terrestrial
weathering intensity and coral growth that may be significant to the global
alkalinity budget. However, we conclude that the similarity between the model
pCO2 and data from Byrd supports the possibility that lower glacial pCO2 values
were caused by a higher glacial whole ocean alkalinity. While the "coral reef"
hypothesis (a higher deep sea calcite burial rate during glacials) appears by
itself to be inadequate as an explanation for high glacial alkalinity, we look
instead to a change in the relative production rates of calcite and organic
carbon.
Acknowledgements. This work supported by the Lamont Doherty
Earth Observatory, and benefitted from discussion with Wally Broecker, Steve
Emerson, Lloyd Burkle, Robin Keir, Mitch Lyle, Doug MacAyeal, and several other
anonymous reviewers. 