Foraminiferal Calcification Response to Glacial-Interglacial Changes in Atmospheric CO2

See allHide authors and affiliations

Science  02 Aug 2002:
Vol. 297, Issue 5582, pp. 833-836
DOI: 10.1126/science.1072815


A record of foraminiferal shell weight across glacial-interglacial Termination I shows a response related to seawater carbonate ion concentration and allows reconstruction of a record of carbon dioxide in surface seawater that matches the atmospheric record. The results support suggestions that higher atmospheric carbon dioxide directly affects marine calcification, an effect that may be of global importance to past and future changes in atmospheric CO2. The process provides negative feedback to the influence of marine calcification on atmospheric carbon dioxide and is of practical importance to the application of paleoceanographic proxies.

Higher concentrations of carbon dioxide in the atmosphere cause surface seawater to become more acidic and lower the calcium carbonate saturation state through the consequent decrease in [CO3 2–], the carbonate ion concentration (1). Predictions suggest that the carbonate saturation state will be reduced by 30% relative to the preindustrial level by the middle of the 21st century (1, 2). This has raised concern because of evidence that carbonate saturation is correlated with the rate of production of marine calcium carbonate and because of studies showing that coral reefs and some species of coccolithophorids (major producers of marine carbonate) are sensitive to elevated CO2 pressure (P co 2) (3–7). The hypothesis that growth rate is a function of [CO3 2–] is also consistent with inorganic studies (8, 9).

If marine calcification is sensitive to the concentration of atmospheric carbon dioxide, its effect should be reflected in the paleoceanographic record as a response to glacial-interglacial fluctuations in P co 2. Foraminifera constitute an important fraction of marine plankton and play a significant role in the carbon cycle through the production of shell calcite. Increased carbonate ion concentrations in culturing experiments using the planktonic foraminifer Orbulina universa have been shown to produce higher shell weights in similarly sized organisms, interpreted as a consequence of thicker shell walls resulting from higher rates of calcification (10, 11). We have found that shell weights of several species of planktonic foraminifera from core top sediments vary systematically as a function of latitude in the North Atlantic. By combining these findings with a record of shell weight across glacial-interglacial Termination I, we demonstrate that the changes are as a result of ambient [CO3 2–] changes rather than calcification temperature and are consistent with known changes in atmospheric P co 2. The link between marine calcification and [CO3 2–] provides a negative feedback to changes in atmospheric P co 2. This observation is also of practical importance in paleoceanography because shell weight is used as an index of carbonate dissolution at the seafloor and, thus, of past changes in deep-sea [CO3 2–] (12).

Measured weights of several planktonic foraminiferal species (picked from narrow size fractions) from a North Atlantic latitudinal transect (13) increase by a factor of about 2 between 60° and 30°N (Fig. 1A). Due to the strong temperature dependence of CO2 solubility in seawater, and the subsequent dissociation of CO2(aq) into HCO3 and CO3 2–, modern open ocean surface water [CO3 2–] varies as a function of temperature. Thus, the observed trend in shell weight would fit with a [CO3 2–] control as well as a temperature control. An offset in the shell weights ofPulleniatina obliquiloculata and Neogloboquadrina dutertrei has been reported between the tropical Atlantic, Indian, and Pacific Ocean basins, but no such offset is found for the shallower dwelling Globigerinoides sacculifer (12). The similarity in surface ocean temperature profiles between these ocean basins seems to negate a temperature control on shell weight. However, [CO3 2–]-temperature relations are not constant from basin to basin and, as a result, varying offsets in [CO3 2–] occur at different depths between the oceans. These offsets may be sufficient to explain the anomalous offsets in shell weight.

Figure 1

(A) Shell weights of four species of planktonic foraminifera from the North Atlantic between 60° and 30°N latitude plotted against modern sea surface temperature (SST) at the core sites (28). Also shown are corresponding values of surface water [CO3 2–] (based on the linear relation between temperature and [CO3 2–]) corrected to preindustrial values (20). Solid circles,Globorotalia bulloides (300 to 355 μm); open circles,Globorotalia truncatulinoides (300 to 355 μm); solid squares, Globorotalia inflata (300 to 355 μm); open squares, Neogloboquadrina pachyderma (dextral) (250 to 300 μm). (B) Shell weight and (C) size-normalized shell weight for G. bulloides plotted against preindustrial [CO3 2–] (20). Heavy lines in (B) and (C) represent best-fit exponential curves to the data.

Various approaches could be made to test the relation between [CO3 2–] and shell weight. It has suggested that a comparison of modern foraminiferal tests taken from the water column with those from core tops would discriminate between a temperature control and a carbonate ion control (14). If the control is carbonate ion, foraminifera from the water column should weigh less than those from core tops, reflecting the decrease in surface ocean [CO3 2–] resulting from atmosphericP co 2 increase since preindustrial times, whereas a temperature control would produce a change in the opposite direction. An alternative yet analogous approach, which we have adopted, is to chart the evolution of shell weight through time. We analyzed the shell weights of Globigerina bulloidesthrough Termination I in a shallow sediment core from the North Atlantic (NEAP 8K, collected as part of the North East Atlantic Palaeoceanography and Climate Change (NEAPACC) project, 59°48'N 23°54'W; water depth, 2360 m). Results were combined with paleotemperature (via Mg/Ca thermometry) and salinity records from the same core in order to constrain the major parameters associated with the ocean carbonate system. AtmosphericP co 2 at the last glacial maximum (LGM) was ∼90 ppmV lower than preindustrial (Holocene) values (thus, surface ocean [CO3 2–] was higher) and temperature was lower as well. Therefore, a temperature control would be indicated by lower shell weights before Termination I, whereas a carbonate ion control would be indicated by higher shell weights.

Down-core records of shell weight and size-normalized shell weight (15) for G. bulloides are plotted with the δ18O record of G. bulloides from the same core (Fig. 2). Shell weights are highest during the LGM and show a large decrease into the early Holocene. This trend is also recognizable in other, lower resolution, records used to study carbonate dissolution in the Atlantic Ocean (16,17). The fact that shell weights are heavier during glacial times suggests that the control on shell weight is carbonate ion and not calcification temperature. A potential complication is that, alternatively, the control is higher carbonate preservation in the glacial samples.

Figure 2

(A) Measured shell weight and (B) size-normalized shell for G. bulloidesplotted with δ18O [data from (32)] for the same species in NEAP 8K [age model from (33)].

NEAP 8K is situated well above the modern-day lysocline in the NE Atlantic Ocean and is bathed in high-δ13C, low-nutrient North Atlantic deep water (NADW). During the LGM, the site lay at the transition between similarly high-δ13C glacial North Atlantic intermediate water (GNAIW) and low-δ13C Antarctic bottom water (AABW) (18, 19). The increased influence of low-δ13C, nutrient-rich waters at the site of NEAP 8K would probably have caused a lowering of bottom water carbonate saturation and a decrease in preservation potential of carbonate sediments. Therefore, we should expect to see lighter glacial shell weights in NEAP 8K if dissolution was a major control on shell weight at this site. The observed increase in glacial shell weight suggests that this is not a preservation signal.

To evaluate quantitatively whether the record through Termination I can be ascribed to a carbonate ion effect, we first used the data from the core-top transect to establish the relation between [CO3 2–] and foraminiferal shell weight and then go on to estimate surface-ocean paleo-P co 2. Shell weight (Fig. 1B) and size-normalized weight data (Fig. 1C) for G. bulloidesare plotted against preindustrial [CO3 2–]surface(50m)(20). Adjustment to preindustrial values is necessary for the calibration as the average age of core-top samples is >200 years, since when surface-ocean [CO3 2–] has decreased as atmospheric P co 2increased. Comparison of the plots highlights the benefit of correcting weight data for size variations. Samples with considerably lower weights than would be predicted from the overall trend (Fig. 1B) are those with a mean shell size that is lower than average. Recognizing the uncertainties implicit from the scatter in the correlation (of course, the ages of the core tops vary and, hence, our assumption of a constant offset between modern and preindustrial [CO3 2–] is not strictly correct), this calibration is used to interpret the down-core size-normalized weight record from NEAP 8K in terms of [CO3 2–] (Fig. 3A). The antiphase relation between the records of calculated [CO3 2–] and atmospheric CO2 from the Vostok ice core (21) supports the hypothesis that foraminiferal shell weight is related to a component of the ocean-atmosphere carbonate system.

Figure 3

(A) Calculated surface seawater [CO3 2–] estimated from NEAP 8K together with atmospheric [CO2] from the Vostok ice core. (B) Calculated surface seawaterP co 2(aq) with Vostok atmospheric [CO2].

The relation between atmosphericP co 2 and the carbonate species within surface waters can be described to a first approximation by the thermodynamic relationsEmbedded Imagewhich lead toEmbedded Imagewhere K 0 is the solubility coefficient of CO2 in seawater,K 1 and K 2 are the first and second dissociation constants of carbonic acid [functions of temperature (T) and salinity (S)], K′ isK 0 K 1 /K 2, and TA is the total alkalinity (22). Accordingly, knowledge of [CO3 2–], T, S, and TA, will allow approximate calculation of P co 2. Temperature and salinity were estimated by combining Mg/Ca thermometry with δ18O data on the same foraminiferal samples (Fig. 4) (23). Although TA is a linear function of salinity in the modern surface ocean, it is not clear whether this relation is constant through time. Therefore, estimates were made for different values of TA (23).

Figure 4

(A) Mg/Ca and δ18O forG.bulloides from NEAP 8K. (B) Mg/Ca from NEAP 8K with Faunal (SIMAX) and alkenone (Uk37) temperature estimates from nearby BOFS 16K (collected as part of the Biogeochemical Ocean Flux Study) (36). (C) Calculated calcification temperatures from Mg/Ca [Mg/Ca = 0.72 × e0.1 × T, where T is temperature in °C] and corresponding values of δ18Oseawater for the same samples.

The record of surface-ocean P co 2(P co 2(aq)) for NEAP 8K obtained in this manner displays a trend similar to that observed for atmospheric CO2 as recorded in Antarctic ice (Fig. 3B). The glacial to interglacial increase in P co 2(aq) of around 80 to 90 ppm agrees with estimates from ice records. In general,P co 2(aq) values are about 40 ppm lower than atmospheric CO2 concentrations. This is in line with observations made in the modern ocean at this latitude (24) and is probably related to the strong temperature control on K 0,1,2 and the relatively long (in the order of 1 year) equilibration time between CO2(atmos) and P co 2(aq)(25). Errors in estimating T and S result in uncertainties in calculated P co 2(aq) of about 10 ppm (fig. S1A) (23). In this calculation (Fig. 3B), TA was kept constant at 2320 μeq kg−1, the present value at this site. By letting TA vary as a function of salinity, the maximum glacial-interglacial shift in P co 2(aq)is 70 to 80 ppm, although the amplitude of variability during glacial times is increased (fig. S1B) (23).

The amplitude of variability inP co 2(aq) on shorter time scales than glacial-interglacial changes is relatively high compared with that observed in ice records. Although the Vostok data shown here are not of particularly high resolution (more variability may be expected on shorter time scales), we may well expect atmospheric records to show less variability than their marine counterparts. This is due to the near homogeneous state of atmospheric gases compared with the highly variable nature of P co 2 in the surface ocean.

The study suggests that glacial-interglacial changes in foraminiferal shell weight are related to changes in ambient carbonate ion concentration through time in response to changing atmosphericP co 2. If this proves to be systematic of marine carbonate production in general, the process provides negative feedback to past changes in carbon dioxide as well as to anthropogenic carbon dioxide increases in the future. Though it is as yet unclear how precise this approach might be (secondary controls such as nutrient availability may also be important), measurements of shell weight offer the potential to determine which areas of the global ocean have the greatest control over shifts in atmospheric CO2. A further application would be to investigate longer-term trends in [CO3 2–] andP co 2. The observations made here also provide potential to refine the estimates of deep-ocean [CO3 2–] based on the assumption that foraminiferal shell weight is an index of dissolution (12). Such estimates clearly require prior knowledge of initial weight, which might be attained from a nearby shallow-core calibration or possibly by using an atmospheric P co 2 record. It is interesting to note that the shell weights of undissolved foraminiferal calcite appear to trace surface-ocean [CO3 2–], whereas those of dissolved foraminiferal calcite appear to trace [CO3 2–] of the deep ocean.

Supporting Online Material


Fig. S1


  • * To whom correspondence should be addressed. E-mail: sbar98{at}


View Abstract

Stay Connected to Science

Navigate This Article