Southern Hemisphere and Deep-Sea Warming Led Deglacial Atmospheric CO2 Rise and Tropical Warming

See allHide authors and affiliations

Science  19 Oct 2007:
Vol. 318, Issue 5849, pp. 435-438
DOI: 10.1126/science.1143791


Establishing what caused Earth's largest climatic changes in the past requires a precise knowledge of both the forcing and the regional responses. We determined the chronology of high- and low-latitude climate change at the last glacial termination by radiocarbon dating benthic and planktonic foraminiferal stable isotope and magnesium/calcium records from a marine core collected in the western tropical Pacific. Deep-sea temperatures warmed by ∼2°C between 19 and 17 thousand years before the present (ky B.P.), leading the rise in atmospheric CO2 and tropical–surface-ocean warming by ∼1000 years. The cause of this deglacial deep-water warming does not lie within the tropics, nor can its early onset between 19 and 17 ky B.P. be attributed to CO2 forcing. Increasing austral-spring insolation combined with sea-ice albedo feedbacks appear to be the key factors responsible for this warming.

The data obtained from high-latitude ice cores establish a close temporal relation between varying concentrations of atmospheric CO2 and atmospheric temperatures during glacial terminations (1). However, uncertainty in the gas-age chronologies and inadequate temporal resolution in many proxy climate reconstructions have hampered efforts to establish the exact phasing of events during glacial terminations, a necessary step in understanding the physical relation between CO2 forcing and climate change (2). Arguably, the most robust estimate of changes in mean global temperatures that accompany glacial terminations is the amount of heat stored in the oceans. The large size of the oceanic reservoir and the long residence time of deep waters mean that deep-water temperatures reflect a globally averaged record of Earth's radiative-heat balance (3). At present, we cannot quantify changes in ocean heat content at the last glacial termination. However, because deep waters formed at high latitudes carry with them the nearly conservative properties of temperature and salinity, it is possible to establish the relative timing of high-latitude versus low-latitude climatic change at glacial terminations by dating co-occurring benthic (bottom-dwelling) and planktonic (surface-dwelling) foraminiferal paleotemperature records in a marine core from a tropical location that has sufficient temporal resolution. We used radiocarbon (14C) dating to establish the timing of the deep-sea and tropical–surface-ocean temperature changes during the last glacial termination and compared this history with the timing of CO2 change and deglacial warming in the southern high latitudes during the last glacial termination.

Marine sediment core MD98-2181 from the Morotai Basin (Fig. 1) is ideally suited for documenting the timing of both deep-sea and tropical–surface-water temperature change during the last glacial termination because the sediments at this site are an admixture of planktonic and benthic foraminiferal carbonate that accumulated at rates of 50 to 80 cm/thousand years (ky) over the past glacial-to-interglacial transition (4). Sampling this core at the centimeter scale provides a temporal resolution for each sample of ∼25 to 50 years (5). This core was recovered from a depth of 2114 m, where it was bathed in the upper Pacific Deep Water, a water mass whose temperature and salinity are acquired within the Southern Ocean (6). The site is also located within the Pacific Warm Pool (7), where the δ18O and Mg/Ca of planktonic foraminifera record the temperature and salinity of western tropical Pacific surface water (8, 9).

Fig. 1.

Location of tropical sites (MD98-2181, MD98-2176 and MD98-2170), high–southern latitude Ocean Drilling Program site 1233, and the EPICA Dome C ice core. At 2100 m, the MD98-2181 site is bathed by the Pacific Deep Water (supporting online material). The depth of the isopycnal surface σt = 27.6 kg/m3 (relative to the 0 m) is shown with color shading (39), pre-bomb 14C ages from the Global Data Analysis Project data set (40) are interpolated onto σt = 27.6 kg/m3 (contour lines), geostrophic velocity on σt = 27.6 kg/m3 is shown (blue and magenta vectors, the scale of the magenta vectors is reduced by 60% relative to the blue vectors for better visibility), and the locations of marine cores are indicated.

We compiled 42 planktonic (4) and 8 mixed-benthic (Table 1) 14C ages from late glacial and Holocene sections of this site (Fig. 2). Broecker et al. (10) noted previously that the benthic/planktonic 14C age difference from the glacial termination horizons of this core averages ∼1300 years, similar to the 14C offset between surface and deep waters today (fig. S2). Of the eight benthic/planktonic pairs we have from the deglacial section, the average age difference is 1493 (±367) years, similar to Broecker et al.'s estimate. The reservoir age of surface waters in the source region of the Pacific Deep Water may have been slightly older during the glacial period, which would explain the somewhat higher surface/deep-water age difference in the glacial samples (11). However, given the small range in Δ14C, we see no evidence for a substantial shift in the surface/deep-water age difference and therefore estimate that the transit time of deep water from the Southern Ocean to the north equatorial Pacific was ∼1000 (±300) years during the glacial period, as in the modern ocean (12). The conversion of 14C ages to a calendar-age model was performed with the CALIB 5.0.2 routine using the Marine04 calibration data set (13). A reservoir-age correction of 560 years was applied for the planktonic samples younger than 13,000 years; a correction of 630 years was applied for older samples. At each stratigraphic horizon, the benthic foraminiferal δ18O value reflects the deep-water temperature and oxygen isotope composition acquired from the Southern Ocean ∼1000 years earlier (fig. S2). Therefore to compare the temporal phasing of deep-water– and tropical–surface-water changes at the deglaciation, an independent chronology was established for the benthic δ18O record by adding 1000 years to the planktonic calendar age at each core horizon.

Fig. 2.

14C ages of planktonic (solid circles) and benthic foraminifers (open circles) from MD98-2181. The average benthic/planktonic 14C age difference is 1493 (±367) years through the late glacial and early deglacial portion of the core. 14C dates for terrestrial wood from the glacial section indicates that the surface reservoir age in the Western tropical Pacific was not much higher during the glacial period (10). 14C data from 941 to 961 cm were excluded because of a sediment disturbance at this depth (4).

Table 1.

Benthic foraminiferal 14C ages from the Morotai Basin core MD98-2181 (6°N, 126°E, 2.1 km). All 14C measurements were conducted at the Woods Hole Accelerator Laboratory. NOSAMS, National Ocean Sciences Accelerator Mass Spectrometry.

View this table:

Previous studies (8, 9, 1418) have shown that during the last glacial termination, tropical Pacific sea surface temperatures (SSTs) began to warm at ∼17.5 thousand years before the present (ky B.P.) and that the warming preceded the first major shift in seawater δ18O associated with the disintegrating glacial ice sheets (17). The timing of tropical SST warming is best documented in the MD98-2181 core, which has the highest sediment-accumulation rate and the most detailed 14C age control of the available cores (fig. S1). However, two other closely associated cores, MD98-2176 and MD98-2170, compliment the MD98-2181 SST record and provide a composite representation of the Western Pacific Warm Pool during the last deglaciation (8). During the Last Glacial Maximum, the SSTs at these sites averaged 26°C (Fig. 3) (8, 9). SSTs increased by ∼1°C between 17 and 16 ky B.P., and by 14.6 ky B.P. they had reached ∼28°C across the Warm Pool. The deglacial warming within the tropical Pacific Warm Pool occurred in close association with increasing concentrations of atmospheric CO2 (14); this finding has provided support for hypotheses that call on the CO2 increase itself to explain the deglacial warming (1921). However, the exact phasing between the CO2 increase and the tropical SST warming is not well established because of the uncertainty in assigning gas ages to CO2 records (1, 2224). In the DomeCicecore, for example, the concentrations of both CO2 and CH4 begin to increase above their glacial values at 525 m. In the European Project for Ice Coring in Antarctica (EPICA) Dome C 2 (1) time scale, this corresponds to a gas age of 17.3 ky B.P., which would mean that the initial rise in CO2 occurred essentially simultaneously with warming in the tropical Pacific. However, by synchronizing CH4 records from the Antarctic Dome C core and the Greenland Ice Sheet Project (GISP) core (23) and adopting the GISPII gas-age model, the initial CO2 rise could be shifted to as early as 18 ky B.P. and therefore slightly earlier than the tropical SST warming (Fig. 3). In either case, the concentration of atmospheric CO2 did not begin to rise until after 19 ky B.P.

Fig. 3.

Temporal phasing of the Pacific deep- and tropical–surface-water deglacial stratigraphy compared with the EPICA Dome C CO2 record (1). The averaged mean longitude spring insolation (21 August to 20 November) at 65°S insolation and the Dome Fuji ice core temperature deviation (ΔTsite) relative to the mean of the last 10 thousand years is shown (34). The yellow shading spans the time period between the initial warming of deep-sea temperatures and the onset of tropical Pacific SST warming. The Dome C CO2 record was plotted on the GISPII gas-age model (15) by synchronizing the GISP II and Dome C CH4 stratigraphies. The deep Pacific δ18O record is based on multiple species of benthic foraminifera from MD98-2181 (Cibicidoides mundula, blue circles; Uvigerina costata, red circles; Uvigerina hispida, black squares; Cibicidoides wuellerstorfii, green diamonds; red squares, Uvigerina sp.) and is depicted on an age scale that represents the surface conditions at the source region of the Pacific Deep Water. The gray shading indicates a ∼200-year uncertainty in the benthic ages. A disturbed interval between the 941- and 961-cm sections of MD98-2181 encompassing the interval associated with the Antarctic Cold Reversal is excluded. ppmv, parts per million by volume; VPDB, Vienna Pee Dee belemnite.

If atmospheric CO2 was the sole driver of deglacial temperature changes, the deep-sea temperatures should reflect this with an appropriate lagged relation that would account for the response time between CO2 forcing and the turnover time of deep waters. The Pacific is the best candidate for testing the temporal relation because it is volumetrically the largest deep-water reservoir. Shackleton (19) recognized the phasing similarity between δ18O of O2 gas trapped in the Vostok ice core (25) and the orbital-insolation record (June insolation at 65°N) and argued that the O2 record reflects the influence of solar-insolation forcing vis-à-vis ice-volume variations. He tuned the deep-sea (benthic) δ18O record from Pacific core VM19-30 to the Vostok O2 δ18O stratigraphy and applied the orbital-insolation chronology to it. The advantage of this approach would be to provide an estimate of the deep-sea temperature change, which is the residual benthic δ18O signal that is not accounted for by the Vostok O2 δ18O record. However, this method also imposed a phase relation between the deep-sea temperature change and the ice-volume components of the benthic δ18O record. As a result, the “tuned” VM19-30 benthic δ18O record does not exhibit any deglacial δ18O change until ∼15 ky B.P., nearly 3000 years after atmospheric CO2 began to rise (fig. S2). In a subsequent study of deep Pacific temperature changes during the last glacial termination, Martin et al. (20) used Mg/Ca–based estimates of benthic temperature change to argue that deep Pacific temperatures warmed by 2°C during the glacial termination and that the magnitude of these deep-sea temperature changes was consistent with a forcing by CO2 during the deglaciation. Here again, the issue is timing. Did deep-sea temperature change lag the CO2 forcing? The records used by Martin et al. (20) to argue for the CO2 forcing of deep-sea temperatures were also tuned to the orbital-insolation record through a cross correlation of benthic δ18O records to Shackleton's VM19-30 record. Consequently, these cores do not provide an independent constraint on the temporal relation between temperature change and CO2 concentration.

We circumvented the assumptions associated with tuning benthic records by independently dating benthic foraminifera from MD98-2181 (Table 1). In Fig. 3, the independently dated MD98-2181 benthic δ18O stratigraphy (representing Southern Ocean surface conditions 1000 years earlier) is compared with the western tropical Pacific SST reconstructions. The maximum benthic δ18O values occurred between 20 and 19 ky B.P. Three species of benthic foraminifera exhibited a δ18O value of 4.8 per mil (‰) during this time period, a value that is 1.8‰ higher than modern benthic values at this site. Taking the mean ocean glacial-to-interglacial δ18O shift to be 1‰ (19, 2630), the observed 1.8‰ difference between Holocene and glacial δ18O values at MD98-2181 implies that temperatures at 2100 m in the Pacific were ∼2.5°C colder and/or the 18O/16O of the deep water was substantially higher at the glacial maximum. Pore-water Cl profiles from deep-sea sediment cores suggest that deep waters were more salty during the Last Glacial Maxmium, perhaps by as much 1‰ (30), but there is no evidence to suggest that the 18O/16O of the Pacific deep water was substantially higher (after correcting for the 1‰ ice-volume effect) (31). Considering these results together with the Mg/Ca data from the eastern Pacific (32) indicates that the primary influence on the early deglacial benthic δ18O evolution was temperature change. The benthic δ18O values from MD98-2181 began to decrease from the cold glacial maximum values at 19 to 18.5 ky B.P. and decreased progressively by ∼0.5‰ over the next 1000 years. If this 0.5‰ decrease was entirely due to temperature, it means that nearly all of the glacial/interglacial deep-water warming occurred before 17.5 ky B.P. and therefore before both the onset of deglacial warming in tropical Pacific surface waters and the increase in atmospheric CO2 concentration.

A comparable warming has also been observed in surface-water records from the midlatitudes in the south Pacific (33). At this latitude the surface waters also warmed by 2°C between 19 and 17 ky B.P., leading the tropical warming by ∼1000 years (Fig. 2). The onset and temporal evolution of deglacial warming over the Antarctic continent (34) also matches the deep-ocean record of warming (Fig. 3). These records, together with the synchronous retreat of mid–southern-latitude glaciers (35), confirm that the onset of deglacial warming throughout the Southern Hemisphere occurred long before deglacial warming began in the tropical surface ocean. An early warming in the Antarctic has important implications for understanding what may have caused an asynchronous deglacial warming pattern between the low and high southern latitudes. In particular, it means that the mechanism responsible for initiating the deglacial events does not lie directly within the tropics itself, nor can these events be explained by CO2 forcing alone. Both CO2 and the tropical SSTs did not begin to change until after 18 ky B.P. (22), approximately 1000 years after the benthic δ18O record indicates that the Southern Ocean was warming.

The rise in Southern Ocean temperatures coincided with the retreat of Antarctic sea ice and high-elevation glaciers in the Southern Hemisphere (36, 37). The explanation for the early warming in the Southern Hemisphere could involve increasing springtime solar insolation, which is well correlated with the retreat of sea ice and with the history of sea-salt accumulation in the Antarctic Dome C ice core (fig. S5). We suggest that the trigger for the initial deglacial warming around Antarctica was the change in solar insolation over the Southern Ocean during the austral spring that influenced the retreat of the sea ice (38). Retreating sea ice would have led to enhanced Ekman transport in the Southern Ocean and decreased stratification due to stronger air-sea fluxes. We hypothesize that these forcings promoted enhanced ventilation of the deep sea and the subsequent rise in atmospheric CO2.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S5

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article