Research Article

The Southern Ocean’s Role in Carbon Exchange During the Last Deglaciation

See allHide authors and affiliations

Science  03 Feb 2012:
Vol. 335, Issue 6068, pp. 557-561
DOI: 10.1126/science.1208163


Changes in the upwelling and degassing of carbon from the Southern Ocean form one of the leading hypotheses for the cause of glacial-interglacial changes in atmospheric carbon dioxide. We present a 25,000-year-long Southern Ocean radiocarbon record reconstructed from deep-sea corals, which shows radiocarbon-depleted waters during the glacial period and through the early deglaciation. This depletion and associated deep stratification disappeared by ~14.6 ka (thousand years ago), consistent with the transfer of carbon from the deep ocean to the surface ocean and atmosphere via a Southern Ocean ventilation event. Given this evidence for carbon exchange in the Southern Ocean, we show that existing deep-ocean radiocarbon records from the glacial period are sufficiently depleted to explain the ~190 per mil drop in atmospheric radiocarbon between ~17 and 14.5 ka.

The transition between the Last Glacial Maximum [LGM; ~22 to 18 ka (thousand years ago)] and the warm climate of the Holocene was accompanied by an ~80 parts per million rise in atmospheric CO2 (1) and a substantial reduction in the ratio of radiocarbon (14C) to 12C in the atmosphere (2). Of particular interest is the large and rapid drop in atmospheric radiocarbon during the first pulse of CO2 rise at the last deglaciation (3). One hypothesis that has been proposed to explain the rise in atmospheric CO2 and the concurrent fall in atmospheric radiocarbon content is the “isolated reservoir” hypothesis, which calls for ventilation of a carbon-rich, radiocarbon-depleted water mass that was isolated from the atmosphere during the glacial period (4). Increased upwelling in the Southern Ocean has been suggested as a potential route for CO2 outgassing to the atmosphere during the last deglaciation (5). In this study, we generate reconstructions of Southern Ocean radiocarbon to provide new insights into the storage and release of carbon from the deep ocean.

Radiocarbon is a useful tracer of carbon cycle processes because it is produced in the atmosphere and then is transferred to other carbon reservoirs, where it decays with a half-life of 5730 years. Isolation of deep-water masses allows time for decay of radiocarbon and sequestration of carbon. However, evidence for the presence and subsequent release of a carbon-rich and radiocarbon-depleted glacial water mass remains controversial due to the challenges associated with producing reliable radiocarbon records. Radiocarbon reconstructions require accurate measurement of the 14C/12C ratio and, crucially, accurate and precise determinations of absolute (calendar) age. Deep-ocean radiocarbon records have been made using benthic foraminifera (6), but dissolution of carbonate, low abundances of benthic foraminifera, and the susceptibility of low sedimentation rate cores to the mixing effects of bioturbation limit the availability of appropriate samples in deep water. Absolute-age models in sediment cores often rely on radiocarbon analyses of planktic foraminifera, which can be strongly influenced by large variations in surface reservoir ages (7, 8). Additionally, variations in foraminiferal abundance or species composition through time can cause biases in benthic-planktic age offsets (9, 10).

In spite of these difficulties, much effort has been made to identify an isolated carbon reservoir in the glacial ocean, but direct evidence remains elusive. Broecker (3) has argued that the similarity between planktic-benthic radiocarbon age offsets from today and the LGM from a suite of cores in the Pacific has limited the potential volume for an abyssal isolated reservoir, requiring it to be thin, deep, and extremely radiocarbon-depleted. Skinner et al. (8) recently suggested the presence of a 4000-year-old water mass in the deep South Atlantic (3770 m) during the LGM. Although an intriguing result, the spatial extent of this water mass is not well constrained, and the associated age model requires large changes in the surface reservoir age.

A series of recent papers showing evidence for extreme radiocarbon depletions at intermediate depths during periods of deglacial CO2 rise (1113) have also sparked interest in this isolated reservoir hypothesis. These papers invoke the transfer of old carbon from a deep isolated reservoir to intermediate depths via upwelling in the Southern Ocean and transport by Antarctic Intermediate Water (AAIW) to locations in the Northern Hemisphere.

Here we provide a glacial and deglacial time series of radiocarbon from deep-sea corals collected from sites in the Southern Ocean at depths corresponding to modern-day Upper Circumpolar Deep Water (UCDW) and AAIW. Uranium-thorium (U-Th) dating is used to provide an independent calendar age that does not depend on reservoir ages or stratigraphic correlation. The location of our samples in the Drake Passage allows us to examine direct links between the deep ocean and the atmosphere because it is a region of mixing and upwelling of major deep-water masses.

Sample description and results. Forty solitary deep-sea corals (37 Desmophyllum dianthus, two Balanophyllia malouinensis, and one Flabellum curvatum) were selected for this study from a suite of dredge-collected corals from Burdwood Bank and seamounts in the Drake Passage (Fig. 1) (14, 15). Coral samples were screened to identify glacial and deglacial specimens by radiocarbon reconnaissance dating (14), and these samples were then U-Th dated. Precise radiocarbon measurements (16) were made on the corals with ages within the time period of 0 to 35 ka to calculate past Δ14C values (17) of dissolved inorganic carbon (DIC) at depths bathed today by AAIW and UCDW (18). Here we refer to the radiocarbon records by the names of the water masses at those depths today, while recognizing that different water masses could have occupied those depths in the past.

Fig. 1

(A) Locations of dredging sites (white circles) and sediment cores. Core SO161-SL22 [36.2°S, 73.7°W, 1000 m; (28)] is indicated by a blue triangle, core MD07-3076 CQ [44.1°S, 14.2°W, 3770 m; (8)] is indicated by a brown diamond, and core TN057-21 [41.1°S, 7.8°E, 4981 m; (27)] is indicated by a pink square. (B) Map of the dredging sites (white circles). The Polar Front is indicated by a blue line (38). SFZ, Shackleton Fracture Zone. Depth-latitude sections of salinity (C) and oxygen (D) from the GLODAP data set for the latitudinal transect indicated by the red box from (B) (19). Corals (white circles) come from depths bathed today by Antarctic Intermediate Water (AAIW) and Upper Circumpolar Deep Water (UCDW). LCDW, Lower Circumpolar Deep Water.

The reconstructed Δ14C values for corals at depths corresponding today to AAIW follow the atmospheric Δ14C curve (2) at an offset [ΔΔ14C, supporting online material (SOM)] ranging from –80 to –190‰ (Fig. 2), with modern corals reflecting the Δ14C of ambient seawater DIC (19). The largest radiocarbon offsets in AAIW occurred at 13.3 ± 0.1 and 13.7 ± 0.1 ka during the Bølling-Allerød/Antarctic Cold Reversal (BA/ACR; ~14.6 to 12.8 ka).

Fig. 2

(A) Δ14C reconstructed from Drake Passage deep-sea corals bathed today by AAIW (dark blue stars) and UCDW (red stars). Uncertainties are represented by 2σ error ellipses. Two previously dated corals from Sars Seamount (39, 40) with ages of ~16.5 and 16.6 ka are included in the UCDW group. Also shown are reconstructed Δ14C records from benthic foraminifera from cores in the South Atlantic (brown diamonds) (8), North Pacific (gold circles) (6), Cape Basin (pink squares) (27), and Chilean margin (light blue triangles) (28). Atmospheric radiocarbon (IntCal09) (2) is shown by the gray lines (±2σ). Arrow 1 marks the drop in UCDW radiocarbon over the HS1/Mystery Interval that parallels the drop in the atmospheric record. Arrow 2 marks the increase in UCDW radiocarbon after the HS1/Mystery Interval. (B) Offset of reconstructed deep-sea coral Δ14C from contemporaneous atmosphere (ΔΔ14C) for AAIW (dark blue stars) and UCDW (red stars) corals (this study) (39, 40). Also included are the offsets of the reconstructed benthic foraminiferal Δ14C from the same records as shown in (A). All ellipses are 2σ. Contours of constant radiocarbon age offset from the contemporaneous atmosphere are shown as gray lines to illustrate the effect of the changing atmospheric radiocarbon inventory on ΔΔ14C values (SOM). Dashed line at 14.6 ka divides the plot between the relatively radiocarbon-depleted waters of the glacial period and the relatively radiocarbon-enriched waters post–Mystery Interval.

The reconstructed Δ14C values for corals from depths bathed today by UCDW follow atmospheric Δ14C throughout the Holocene, with offsets from the contemporaneous atmosphere ranging from –110 to –155‰, consistent with nearby water column Δ14C measurements (19). During the LGM, radiocarbon in UCDW was 230‰ depleted from the contemporaneous atmosphere, ~80‰ lower than values observed at these sites today. UCDW was even more depleted from the contemporaneous atmosphere before the LGM at 25.6 ± 0.3 ka during Heinrich Stadial 2 (HS2; ~27 to 24 ka), with an offset of –370‰, and also during Heinrich Stadial 1 (HS1; ~18 to 14.6 ka), with offsets of –250 to –300‰. A marked increase in UCDW Δ14C occurred at the start of the BA/ACR, giving ΔΔ14C values of only –140‰. The offsets of UCDW Δ14C from the contemporaneous atmosphere then remained relatively constant until the modern day, with the exception of slightly enriched radiocarbon values (ΔΔ14C = –87 to –105‰) immediately after the end of the Younger Dryas (YD; 12.8 to 11.5 ka) (Fig. 2).

Discussion. Large radiocarbon depletions in the glacial Southern Ocean could be caused by a reduction in air-sea gas exchange (2022) [for instance, by an increase in sea-ice cover (23) at deep-water formation sites] or by increased stratification within the Southern Ocean (24). Additionally, global ocean changes in stratification (25) could contribute to lower Δ14C values in the glacial deep ocean by reducing mixing between the abyssal ocean and better-ventilated water masses above. A reduction in the flux of well-ventilated (radiocarbon-enriched) northern-source waters to the Southern Ocean could also lead to more depleted radiocarbon in UCDW. We explore the potential of these mechanisms to explain the features that we observe in the Southern Ocean radiocarbon records, with reference to additional published records from the global ocean.

Polar stratification has been suggested to play a fundamental role in lowering atmospheric CO2 during glacial times by reducing the ventilation of CO2 from the deep ocean (26). The depleted radiocarbon values that we detect in glacial UCDW are indeed consistent with reduced ocean ventilation during the last glacial period and imply that carbon at these depths was more isolated from the atmosphere than it is today. To examine the extent of this isolation and depletion, we compare our Drake Passage data to two deep radiocarbon records from benthic foraminifera from the South Atlantic (Figs. 1 and 2) (8, 27). The deeper record [4981 m in modern-day Antarctic Bottom Water (AABW) (27)] is broadly consistent with our UCDW data during the deglaciation, suggesting common influences on the radiocarbon values in the Southern Ocean regions of UCDW and AABW formation. The second record [3770 m in modern-day Lower Circumpolar Deep Water (LCDW) (8)] is quite different during the glacial, with Δ14C values up to 500‰ lower than those in the atmosphere, or 290‰ lower than those in UCDW and AABW (Fig. 2). This 500‰ radiocarbon depletion from the atmosphere may have been caused by an increase in the proportion of recirculated deep water from the Pacific. Some of the offset between the two South Atlantic foraminiferal radiocarbon records may arise from the large differences in surface reservoir ages used for their respective age models (SOM). However, even allowing for changes to the sediment core age models, the records from UCDW, LCDW, and AABW suggest that there was a geochemical gradient between deep Southern Ocean water masses at the LGM, consistent with reduced vertical mixing across water-mass boundaries (25).

At the end of the glacial period and during HS1, the Δ14C of UCDW, LCDW, and AABW became more similar, as the Δ14C of UCDW and AABW decreased and that of LCDW increased (Fig. 2). These changes suggest increased deep mixing in the Southern Ocean and a breakdown of the deep vertical stratification that appears to characterize the LGM. This breakdown of stratification is consistent with evidence for increased nutrient upwelling from opal accumulation records (Fig. 3), hypothesized to result from a southern shift in the westerly winds (5). The timing of this event is concurrent with the initial deglacial rise in atmospheric CO2 (Fig. 3) (1) and lends support to the idea that upwelling of deep carbon-rich waters to shallow depths in the Southern Ocean was a major cause of deglacial atmospheric CO2 rise. The AAIW radiocarbon records from Burdwood Bank and off the coast of Chile (28) were more enriched than the other Southern Ocean water masses during HS1, indicating that there was a substantial influence of atmospheric and surface-ocean radiocarbon on AAIW Δ14C and that AAIW was well ventilated over this time interval.

Fig. 3

(A) Atmospheric CO2 (parts per million by volume) from EPICA Dome C (1) for data <22 ka on the time scale of Lemieux-Dudon et al. (41) and from Vostok ice core (42) for the data points >22 ka. (B) δ18O record from the EDML ice core (43), also on the time scale of Lemieux-Dudon et al. (41). (C) Atmospheric Δ14C (‰) from IntCal09 (2). (D) Offset (‰) of reconstructed Δ14C from that in the contemporaneous atmosphere (ΔΔ14C) for AAIW (blue), UCDW (red), and LCDW (brown). Error ellipses represent 2σ. (E) Opal flux (g cm–1 ky–1) south of the Polar Front in the Indian Ocean sector of the Southern Ocean (core TN057-13-4PC; 53.2°S, 5.1°E) as a proxy for Southern Ocean nutrient upwelling (5).

Before the LGM, at around the time of HS2 (25.6 ka), Drake Passage UCDW had a radiocarbon-depleted signature similar to that of the LCDW site (ΔΔ14C = –370‰; Fig. 2). This similarity between Δ14C in UCDW and LCDW during HS2 is also observed during HS1, suggesting a systematic link between Heinrich stadials and changes in circulation and carbon ventilation in the Southern Ocean (29).

By the start of the BA/ACR, the offset of UCDW Δ14C from the atmosphere was similar to that observed through the Holocene (Fig. 2). This decrease in the magnitude of ΔΔ14C is caused by the combined effect of the large drop in atmospheric Δ14C coupled with a rise in the Δ14C of UCDW (Figs. 2 and 3). An increase in marine Δ14C is also observed in records from the deep North Atlantic (30) and deep South Atlantic (8, 27), suggesting widespread deep ventilation of these regions (Figs. 2 and 3). In contrast to the Atlantic, the Δ14C in the deep North Pacific remains relatively constant (6). However, the drop in atmospheric Δ14C causes these deep North Pacific waters to appear relatively less radiocarbon depleted from the contemporaneous atmosphere by the BA/ACR (Fig. 2). The increase in the Δ14C (and decrease in the magnitude of ΔΔ14C) of UCDW at the start of the BA/ACR is likely due to higher Δ14C values in the deep waters that are upwelled in the Southern Ocean to form UCDW (8). This change may be caused by a deepening of North Atlantic Deep Water supplying a greater flux of well-ventilated (radiocarbon-enriched) waters to the deep Atlantic and Southern Oceans (27).

In contrast, the ΔΔ14C of AAIW is more depleted during the BA/ACR, with values similar to those in UCDW (Figs. 2 and 3). This decrease in the radiocarbon offset between AAIW and UCDW indicates a reduced influence of surface waters and the atmosphere on AAIW Δ14C relative to the deep ocean. This change may be achieved by increased stratification between surface and intermediate waters. Increased surface stratification is also supported by lower opal accumulation fluxes south of the Polar Front, which are interpreted as a reduction in the supply of nutrients to the surface at this time (Fig. 3) (5). An increase in Southern Ocean surface stratification could be associated with an increase in sea-ice cover (23) and may contribute to the deglacial pause in atmospheric CO2 rise. AAIW then returns to a more ventilated state during the YD, further supporting the link between enhanced exchange between the upper Southern Ocean and the atmosphere during periods of atmospheric CO2 rise (Fig. 3).

Lack of extreme 14C depletion in Antarctic Intermediate Water. Our Southern Ocean AAIW data follow the atmospheric radiocarbon record at relatively constant offsets, with absolute values similar to those of the sediment core record from the Chile Margin (Figs. 1 and 2) (28). Additionally, there is no evidence for large (>300‰) radiocarbon depletions in any of the deep South Atlantic (8, 27) or Southern Ocean water masses after 14.6 ka. These results are in apparent contradiction with several studies that have invoked an extremely radiocarbon-depleted AAIW to explain the observations at intermediate depths in the North Pacific, South Atlantic, Arabian Sea, and high-latitude North Atlantic (1113, 31). Our new data from the Atlantic sector of the Southern Ocean make an extremely radiocarbon-depleted AAIW increasingly unlikely. Although our UCDW data do support a Southern Ocean mechanism for the release of carbon during the deglaciation, the radiocarbon-depleted signature does not appear to be transported to the Northern Hemisphere by AAIW. Recent modeling work (32) suggests that propagating the signature of an extremely radiocarbon-depleted water mass to the North Pacific via AAIW is improbable, because the signal would be rapidly dissipated and diluted by the carbon in the rest of the ocean and the atmosphere. Thus, it appears that although there is a widespread distribution of extreme radiocarbon depletions at intermediate depths, they are unlikely to be caused by a common radiocarbon-depleted water mass that was transported over large distances.

The Mystery Interval. One of the more intriguing features of the atmospheric radiocarbon record is the ~190‰ drop in radiocarbon beginning at ~17 ka and ending at the beginning of the BA/ACR at ~14.5 ka (2). Production records from paleomagnetic and 10Be reconstructions can explain, at maximum, 40‰ of this drop (3335). Broecker and Barker (4) have suggested that this drop in atmospheric Δ14C requires a very large, or extremely depleted, oceanic radiocarbon reservoir. In the absence of evidence for such a widespread, old radiocarbon reservoir, this event remains unexplained (the “Mystery Interval”) (36). However, because the atmospheric carbon reservoir is so small compared to that of the ocean, we suggest that a direct outgassing of CO2 from the ocean to the atmosphere would not require an extremely depleted radiocarbon reservoir to force the atmospheric Δ14C drop, allowing it to be explained with available marine radiocarbon records (fig. S5). We use existing records to constrain the deep-water Δ14C values, and we use Southern Ocean radiocarbon records as evidence for the direct transfer of old carbon to the atmosphere during this time period. During the interval between 16.6 and 15.6 ka, the radiocarbon in the atmosphere decreased at a rate similar to the rate of change in the UCDW Δ14C record, whereas the radiocarbon content of the deep South Atlantic (LCDW) (8) increased (Fig. 2). This convergence of the UCDW and LCDW radiocarbon records is consistent with the ventilation of the deep ocean via the Southern Ocean.

If we assume that the glacial ocean deeper than 2500 m has a ΔΔ14C of at least –300‰ (~2000 radiocarbon years offset from atmosphere) as observed in the western equatorial Pacific (37), or that the glacial ocean deeper than 3700 m has a ΔΔ14C of at least –430‰ (~3000 radiocarbon years offset from atmosphere) as observed in the deep South Atlantic (8) and deep North Pacific (6), then only 3% of the carbon from those depths is needed to outgas directly to the atmosphere to decrease atmospheric Δ14C by 190‰. The process of driving down atmospheric Δ14C through outgassing of old carbon in the Southern Ocean would result in an ocean that would be apparently “younger” or less radiocarbon-depleted from the coeval atmosphere than before the Mystery Interval, as the data suggest (Fig. 2). In this case, “old” carbon would be added directly to the atmosphere rather than mixing through the entire ocean, forcing atmospheric Δ14C to fall faster than the Δ14C of most of the ocean.

A major caveat to this calculation is that it is not a transient experiment; hence, we have not taken into account the time scales over which the carbon could be released and subsequently taken up by the ocean. Nonetheless, it provides a possible explanation for the drop in atmospheric radiocarbon over the Mystery Interval that is consistent with available radiocarbon data from the intermediate and deep oceans. This calculation does not require the deep ocean to be as radiocarbon-depleted (old) as a previous calculation done by Broecker and Barker (4), which required the ocean deeper than 2500 m to have a radiocarbon age offset of ~6000 years, or the ocean deeper than 3700 m to have a radiocarbon age offset of ~14,000 years. The reason for this large difference is that Broecker and Barker assume that the nonisolated ocean maintains a constant radiocarbon age offset from the atmosphere after the Mystery Interval. This requirement forces oceanic radiocarbon to drop by ~150‰, which is a result that is not supported by available marine radiocarbon data. Thus, the large radiocarbon age offsets required by the calculation of Broecker and Barker are likely an overestimate of the deep-ocean isolation required to explain the atmospheric drop over the Mystery Interval. Our radiocarbon results from UCDW support the idea of a radiocarbon-depleted glacial ocean that is ventilated directly through the Southern Ocean. By the end of the Mystery Interval and the onset of the BA/ACR, the ocean as a whole is better ventilated, and there is no evidence for the large radiocarbon depletions found in glacial times.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Tables S1 to S3

References (44–56)

References and Notes

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16. Materials and methods are available as supporting material on Science Online.
  17. 17. 14C is a measure of the relative difference between the radiocarbon activity of an absolute international standard and that of a sample, after correction for both age and 13C.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 45.
  45. 46.
  46. 47.
  47. 48.
  48. 49.
  49. 50.
  50. 51.
  51. 52.
  52. 53.
  53. 54.
  54. 55.
  55. 56.
  56. 57.
  57. Acknowledgments: This research was funded by NSF grants 0636787, 0944474, 0902957, 0338087, and 0819714. The data reported in this study are tabulated in the supporting online material. We acknowledge R. Waller and the science parties of cruises LMG06-05 and NBP08-05 for supplying samples used in this study. We also thank the crews of the R/V Nathaniel B. Palmer and R/V Laurence M. Gould, NOSAMS staff, the WHOI plasma facility, J. Blusztajn, M. Auro, and two anonymous reviewers for help with this project.
View Abstract

Navigate This Article