Research Article

Southern Hemisphere Water Mass Conversion Linked with North Atlantic Climate Variability

See allHide authors and affiliations

Science  18 Mar 2005:
Vol. 307, Issue 5716, pp. 1741-1746
DOI: 10.1126/science.1102163


Intermediate water variability at multicentennial scales is documented by 340,000-year-long isotope time series from bottom-dwelling foraminifers at a mid-depth core site in the southwest Pacific. Periods of sudden increases in intermediate water production are linked with transient Southern Hemisphere warm episodes, which implies direct control of climate warming on intermediate water conversion at high southern latitudes. Coincidence with episodes of climate cooling and minimum or halted deepwater convection in the North Atlantic provides striking evidence for interdependence of water mass conversion in both hemispheres, with implications for interhemispheric forcing of ocean thermohaline circulation and climate instability.

Water mass formation in the Southern Hemisphere oceans is involved in the global thermohaline circulation (THC) through the convection and interocean exchanges of surface, intermediate, and bottom waters. These water masses transport heat, salt, and freshwater through the world ocean and are relevant for a range of climatic processes. Yet, water mass production in the Southern Hemisphere has long been perceived as a passive player in ocean and climate dynamics, and research into the causes of rapid climate change has focused on the northern North Atlantic as the presumed primary driver of global ocean THC (1, 2). While evidence is accumulating that climatic upheaval occurred within the recent geologic past with substantial forcing from ocean-climate processes (1), numerical and conceptual models are increasingly highlighting an active involvement of water mass conversion in the Southern Hemisphere (37). The role of Antarctic Intermediate Water (AAIW) in redistributing heat and freshwater within the upper ocean is particularly recognized as pivotal in these changes (4, 7, 8). However, detailed insight into the linking between AAIW and past ocean-climate change has long been hampered by a lack of observational time series of sufficient length and temporal resolution. Fine-scale paleoceanographic records are now available from the southwest Pacific that document the history of AAIW variability at multicentennial scales and enable a detailed assessment of Southern Hemisphere water mass conversion and its linking with climate variability in both hemispheres.

Benthic isotope records. We have generated benthic foraminiferal oxygen (δ18Ob) and carbon isotope (δ13Cb) records along 36-m-long IMAGES (International Marine Past Global Changes Studies) core MD97-2120 from Chatham Rise off New Zealand (45°32.06′S, 174°55.85′E) (Fig. 1A). δ18Ob and δ13Cb were measured on monospecific samples of the benthic foraminifers Cibicidoides wuellerstorfi, C. cicatricosus, and C. kullenbergi. In intervals lacking Cibicidoides spp., the records were complemented by analyses on Melonis barleeanum and Bulimina aculeata to maintain fine-scale resolution along the entire core (9). The age scale for core MD97-2120 is derived from accelerator mass spectrometry (AMS) 14C dating, a tephrochronological marker event, and tuning of the benthic and planktonic records to paleoclimatic profiles from the North Atlantic and Antarctica, respectively (9). The core spans the last three climatic cycles and at its base reaches glacial termination IV at 340,000 years (340 ky) before the present (B.P.). High sedimentation rates at the core site permit a mean temporal resolution along the records of 199 ± 94 years (at 2-cm sample spacing). At a water depth of 1210 m, core MD97-2120 lies close to the AAIW core layer (Fig. 1B). The stable isotope records thus provide the first detailed documentation of the history of mid-depth circulation and AAIW renewal at high southern latitudes over the past three glacial-interglacial cycles.

Fig. 1.

Subtropical gyre circulation in the South Pacific and the Southern Ocean meridional overturning cell. (A) Schematic surface circulation in the South Pacific subtropical gyre with areas of SAMW and AAIW formation (8, 27). PF, polar front; SAF, subantarctic front; STC, subtropical convergence. (B) World Ocean Circulation Experiment (WOCE) meridional salinity transect in the southwest Pacific (44). Superimposed, circulation in the SOMOC (45). AABW, Antarctic Bottom Water; (U/L)CDW, (Upper/Lower) Circumpolar Deep Water. Oceanic fronts as in (A).

The δ18Ob record (Fig. 2B) displays prominent glacial-interglacial modulation that reflects the growth and decay of global ice volumes and associated sea-level fluctuations during the last three climatic cycles (10). Glacial-to-interglacial δ18Ob amplitudes of 1.6 to 1.98‰ exceed the coeval mean-ocean δ18O change of 0.8 to 1.1‰ (11, 12), documenting changes of temperature and salinity (T-S) at intermediate water depth. Additional variability is seen in fine structure that closely follows the suborbital changes in the Antarctic Vostok ice-core deuterium (δD) record (13) that reflects atmospheric temperature changes over Antarctica. Correlation with Vostok δD has been found for a northeast Atlantic deepwater δ18Ob record and used to infer sea-level variations during Antarctic climatic events (14) with considerable meltwater contributions from Northern and Southern Hemisphere ice sheets (15). Sea-level oscillations on millennial time scales have been confirmed recently by using high-resolution isotope data from a Red Sea sediment core (16) and are estimated on the order of 30 ± 5 m. Close fine-structural similarity of our δ18Ob record to these millennial-scale sea-level changes indicates predominant control of global ice-volume fluctuations on the δ18Ob signature in our core. The residual amplitude, up to 0.35‰, not explained by global sea level, reflects changes in mid-depth T-S hydrography that occurred in conjunction with Southern Hemisphere climatic shifts.

Fig. 2.

AAIW proxy records along core MD97-2120 compared with surface hydrographic profiles of the same core and the Vostok climate record over the past 340 ky. (A) Glacial terminations of the planktonic foraminiferal (Globigerina bulloides) δ18O record of core MD97-2120 (25). (B) δ18Ob record of core MD97-2120 measured on Cibicidoides spp. and supplemented by analyses on M. barleeanum and B. aculeata (9). (C) Antarctic Vostok deuterium (δD) record (13). (D) G. bulloides SSTMg/Ca (25) and (E) δ13Cb of core MD97-2120. (F) Glacial terminations of the δ13Cb record from South Atlantic Site 1089 (35). Insets during MIS 5 are a δ13Cb section and ice-rafted debris (IRD) record from the deep North Atlantic (NEAP 18K) (46). The NEAP-18K age scale is adjusted to that of core MD97-2120 using benthic δ8O. Vertical gray bars mark times of reduced North Atlantic deepwater formation associated with increased meltwater influx into the northern North Atlantic [H-events: H1 to H6 (40), h8 to h13 (47), H12 and H13 (48)] and North Atlantic cold events C20 to C24 between 75 and 110 ky B.P. (46). Gray arrows mark negative δ13Cb anomalies at the end of glacial terminations in core MD97-2120 and at Site 1089. Brackets indicate positive δ13Cb excursions during glacial terminations. Age control points are shown along the bottom axis: 14C AMS dates (green), Kawakawa tephra (red), δ18Ob tie-points to North Atlantic core MD95-2042 (blue), and SSTMg/Ca tie-points to the Vostok deuterium record (black) (9, 25). Alternate“SFCP” age scale at bottom is from (49) [see (9)]. Glacial terminations I to IV are labeled TI to TIV.

Benthic δ13Cb levels along core MD97-2120 are reduced during glacial periods of the past 340 ky (Fig. 2E). δ13Cb is derived from δ13C of total dissolved carbon (δ13CTCO2) in the ocean that is controlled by biological nutrient cycling and water mass chemical “aging” (17). This makes foraminiferal δ13Cb a valuable proxy to reconstruct past water mass variability. Several lines of evidence (9) strongly suggest that δ13Cb variability along our record has not been substantially driven by secondary overprints, for example, from organic debris fluff layers at the sediment-water interface (18), temperature-dependent fractionation during air-sea gas exchange in the convection region (19), and source-water (“preformed”) δ13CTCO2 changes. Rather, it is suggested that δ13Cb fluctuations at the mid-depth core site are linked directly with changes in the rate of AAIW production.

The substantial glacial δ13Cb depletion in core MD97-2120 is contrasted by δ13Cb levels in glacial sections of North Atlantic and Pacific mid-depth cores that remain similar to or are elevated over interglacial levels, indicating that mid-depth ventilation in these basins remained high over long time scales (20, 21). Full-glacial δ13Cb depletion by 0.8‰ (Fig. 2) is substantially in excess of the mean-ocean δ13CTCO2 depletion of 0.32‰ (21). It is also in excess of the change of only 0.2‰ recorded at intermediate water sites in the subantarctic East Atlantic [ODP Site 1088 (22), core TN057-20 (23)] but is similar to the δ13Cb found at mid-depth in the Tasman Sea (24). We note that the South Atlantic records do not display typical Holocene levels (Site 1088) or do not reach late Holocene times (TN057-20) at their core tops, whereas amplitudes during previous glacial-interglacial transitions are similar to those in our record (Site 1088) (fig. S5). Although genuine differences may exist between ocean basins, we suggest that the South Atlantic sites do not capture the full amplitude of the last glacial-interglacial change.

Outstanding features of our δ13Cb record are frequent positive excursions in glacial intervals with amplitudes of up to 0.8‰, that is, δ13Cb values during some of these events reach full-interglacial levels (Figs. 2 and 3). The events coincide with Southern Hemisphere warm anomalies seen in Mg/Ca-derived sea surface temperatures (SSTMg/Ca) (25) and in Antarctic ice-core records (13) (Figs. 2 and 3). Consistent correlation of intermediate water δ13Cb and Southern Hemisphere temperature variability on orbital to millennial time scales suggests direct linking of water mass conversion at high southern latitudes with climatic changes.

Fig. 3.

Paleoceanographic records along core MD97-2120 for the past 80 ky compared with benthic and planktonic stable isotope records from deepwater North Atlantic core MD95-2042 (14, 37) and South Atlantic Site 1089 (35). (A) Planktonic δ18O record from core MD95-2042. (B) SSTMg/Ca and (C) δ13Cb records of core MD97-2120. (D and E) δ13Cb records of North Atlantic core MD95-2042 and South Atlantic Site 1089, respectively. (F) δ18Ob records from MD97-2120 (blue) and MD95-2042 (green) for stratigraphic control. Gray bars, age control points, and secondary age axis as in Fig. 2. Yellow bar marks the ACR in SSTMg/Ca and the concomitant negative δ13Cb anomaly in core MD97-2120. Note correlation of positive δ13Cb excursions during H-events in core MD97-2120 with negative δ13Cb anomalies in core MD95-2042 and at Site 1089. Bracket marks the abrupt increase and positive δ13Cb plateau during the deglaciation; arrow indicates negative post-termination δ13C anomalyb in core MD97-2120. Alternate “SFCP” age scale at bottom is from (49) [see (9)].

The Southern Ocean meridional overturning cell. Primary forcing of intermediate water formation at southern latitudes is exerted by air-sea interaction in the Subantarctic Zone (SAZ); cross-frontal northward Ekman transport of cold, fresh Antarctic surface waters; and the southward advection of heat and salt within the subtropical gyre circulation (8, 26). These processes result in surface buoyancy loss in the SAZ and determine the rate and variability of Subantarctic Mode Water (SAMW) formation (Fig. 1A). Wintertime cooling causes SAMW density to further increase until finally the densest classes of SAMW convect and feed AAIW (8, 26, 27). Primary sources for AAIW renewal are in the Southeast Pacific and Southwest Atlantic, with additional sources in the South Pacific and Indian Oceans where Antarctic surface waters are directly transformed into AAIW through air-sea buoyancy fluxes and the southward movement of the circum-Antarctic fronts (8, 28). Increased westerly wind stress has been suggested to contribute to and invigorate shallow overturning in the Southern Hemisphere by enhancing northward Ekman transports and the subduction and convection of AAIW (29).

Environmental conditions were different during glacial periods for an array of ocean and atmospheric parameters with implications for the operation of the Southern Ocean Meridional Overturning Circulation (SOMOC). Substantial atmosphere and surface ocean cooling (13, 25), in conjunction with intensified winds, resulted in a northward displacement of circumpolar ocean fronts and the westerly wind belt by some 5° in latitude (30). Expansion of the sea-ice cover by as much as 70% (30) led to enhanced seasonal meltwater supply to the Southern Ocean, thereby increasing the stratification of the upper water column (31, 32).

The longer term changes of δ13Cb along our record reflect the response of the SOMOC to glacial boundary conditions. Glacial intensification of the westerlies plausibly intensified northward Ekman freshwater transports, which freshened the surface layer in the SAZ. Whereas today the effect of these low-salinity waters on surface buoyancy in the SAZ is partly offset by their low temperatures (26, 29), increased freshwater transports under glacial conditions plausibly enhanced the salinity anomaly across the SAZ and reduced production of AAIW. This process would then contribute to the large glacial δ13Cb depletion in core MD97-2120. Increased upwelling of UCDW in response to enhanced glacial wind stress and stronger bottom-water formation (e.g., 33) with upward displacement of the AAIW-UCDW interface are also consistent with low glacial intermediate water δ13Cb values. However, while the operation of such processes cannot be excluded, the direct coupling of high-amplitude δ13Cb changes with Southern Hemisphere warming strongly suggests changes in AAIW production as the dominant factor in causing δ13Cb variability along our record. As we will discuss below, the positive mid-depth δ13Cb anomalies in our record are coeval with pronounced deepwater δ13Cb minima that are linked with reductions in the southward advection of North Atlantic deep water (NADW) (22, 23, 34, 35). Apparent decoupling of Southwest Pacific mid-depth δ13Cb from deepwater δ13Cb variability requires changes in AAIW production so as to maximize δ13Cb in core MD97-2120 and steepen vertical gradients between the mid-depth and deeper water column.

Rapid departures from low glacial δ13Cb background levels are closely linked with Southern Hemisphere warming, providing striking evidence that the production of newly ventilated AAIW was tightly coupled with abrupt changes in Southern Hemisphere climate. Relaxation and southward shifting of circumpolar wind trajectories, consistent with hemisphere-wide warming, presumably resulted in a decrease of northward Ekman freshwater transports and thus weakened the freshwater anomaly in the SAZ. Increased surface salinity in the Southwest Pacific (25) further suggests enhanced southward salt transports in the western branch of the subtropical gyre, causing rapid buoyancy loss and preconditioning of the SAZ toward intensified AAIW formation. Weakened easterly winds in the tropics at these times (36) could have aided the process through intermittent releases of saline waters from the western tropical Pacific into the subtropical gyre, thereby enhancing southward salt transports and providing the impetus for AAIW production.

Interhemispheric linking. Comparison with a high-resolution deepwater δ13Cb record from the North Atlantic (37) shows that some δ13Cb anomalies in our core are in antiphase with Northern Hemisphere δ13Cb. This is particularly evident during Heinrich (H) meltwater events H4 and H5 in the northern North Atlantic, when freshwater surges from the Northern Hemisphere ice sheets slowed down NADW convection (9, 14) (Figs. 2 and 3). The ventilation collapses associated with these events are evident in discrete δ13Cb minima along the North Atlantic deepwater record but are directly contrasted by δ13Cb maxima in our core. Positive δ13Cb excursions are also developed in our record during the episodes of the North Atlantic H1, H3, H5a, and H6 events. However, these events are not obviously antiphased with corresponding changes in the North Atlantic, perhaps as a result of insufficient temporal resolution of the northern record. We note, however, that the millennial-scale mid-depth δ13Cb variability along our records is in pace with SST changes in the same core, which in turn are synchronous with the Antarctic temperature history and in antiphase with the Greenland ice-core record (38). This linking therefore supports an out-of-phase correlation of the mid-depth δ13Cb ventilation events with Northern Hemisphere cooling and NADW reduction (38).

The incursion of warm episodes in planktonic records from our core during periods of convection slowdown and cooling in the Northern Hemisphere is consistent with the concept of a bipolar antiphase response of climatology (39). While the climatic “seesaw” (39) links the interhemispheric temperature history with changes in marine northward heat transport, it does not predict the observed antiphased pattern between water mass conversion in both hemispheres. The apparent NADW-AAIW anticorrelation suggests that anomalous climate and ocean perturbations in one hemisphere force an immediate response of water mass conversion in the other hemisphere. Numerical and conceptual models postulate the existence of such a mechanism (3, 4, 6, 7) and predict interhemispheric THC changes as a result of varying freshwater budgets at high latitudes and concomitant changes of interhemispheric density gradients. Such a scenario particularly holds for periods of North Atlantic meltwater surges during H-events in that the increased freshwater input to the northern North Atlantic reversed the meridional density gradient in favor of increased AAIW formation (6, 7). Intensified production and northward flow of AAIW then forced a compensating southward flow of saline surface and thermocline waters (6, 7), thereby further diminishing Atlantic salt inventories and, ultimately, stabilizing the NADW “off” mode.

Glacial terminations. Abrupt changes also mark the end of glacial periods as climate shifted to its warm mode. Abrupt increases in δ13Cb occur early in the deglacial phase, when values increase by up to 0.8‰ within as little as 200 to 700 years (Fig. 4). These sudden shifts span most of the total glacial-interglacial δ13Cb amplitude, but they lag the likewise abrupt warming seen in the planktonic SSTMg/Ca profile by some 1 to 3 ky. The delayed response of initial mid-depth ventilation increase to Southern Hemisphere warming points to transient inertia of buoyancy budgets across the SAZ, plausibly caused by enhanced northward meltwater fluxes and reduced southward salt transports in the subtropical gyre during large-scale deglacial warming.

Fig. 4.

Mid-depth ventilation changes during the past four glacial terminations compared with surface ocean records from the Southwest Pacific and deep North Atlantic ventilation records. (A to D) Terminations I to IV. Records in each panel are from top to bottom: North Atlantic planktonic δ18O record of core MD95-2042 (14) (TI and TII) and Site 980 (41) (TIII and TIV); southwest Pacific planktonic δ18O (lilac) and SSTMg/Ca (red) records from MD97-2120 (25); mid-depth southwest Pacific δ13Cb record from MD97-2120 (blue); deepwater δ13Cb records from the North Atlantic [MD95-2042 (37) for TI and TII; Site 980 (41) for TIII and TIV] and South Atlantic [Site 1089 (35)]; and δ18Ob records from core MD97-2120 (dark blue), North Atlantic core MD95-2042 and Site 980 (green), and South Atlantic Site 1089 (brown). Brackets indicate abrupt positive δ13Cb excursions that coincide with the deglacial SSTMg/Ca increase. Numbers 1 to 3 above δ13Cb are (1) early deglacial δ13Cb maxima at times of reduced NADW formation (H1 in TI); (2) and yellow bars, transient δ13Cb minima during Southern Hemisphere cold reversals (ACR in TI) and Northern Hemisphere warm anomalies (B/A in TI); (3) δ13Cb increases in pace with Southern Hemisphere warming and Northern Hemisphere cold reversals (YD in TI).

Once started, AAIW production was closely coupled with rapid climatic rebounds in both hemispheres. The sequence of negative and positive δ13Cb anomalies during TI (labeled 1 to 3 in Fig. 4) is tied to the North Atlantic H-event 1 [H1, <17.6 ky B.P. (40)], the Antarctic cold reversal [ACR, 14.1 ky B.P. (38)], and the subsequent Younger Dryas cold spell (YD, 13 ky B.P.). Within this TI sequence, the Northern Hemisphere cold episodes (H1 and YD) consistently correlate with δ13Cb maxima in our record, whereas the Southern Hemisphere ACR is marked by a salient δ13Cb minimum (Fig. 4A). A similar alternation between positive and negative mid-depth δ13Cb excursions and anticorrelation with North Atlantic deepwater δ13Cb (41) is seen during TII and TIII in association with two-step warmings that are interrupted by midtermination cold events in the Southern Hemisphere (25, 42) (Fig. 4, B and C). Coupling of the δ13Cb swings with temperature in the south confirms the pattern observed along the rest of the records that links positive/negative δ13Cb anomalies with warm/cold climates on orbital and suborbital time scales. The ventilation minimum in the south during the ACR correlates with strengthened NADW formation during the Northern Hemisphere Bølling/Allerød (B/A) warm period (7), whereas the subsequent δ13Cb increase coincides with the YD cold reversal when NADW production was reduced. Such anticorrelation of water mass conversion during deglaciations corroborates similar phasing of NADW and AAIW production observed during glacial and interglacial periods and demonstrates bipolar anticorrelation of THC forcing to be a persistent feature across different climate states.

The final phases of the glacial terminations are marked by brief negative δ13Cb anomalies during or immediately after early-interglacial peak warmth (Fig. 4). These anomalies are exceptional in that they document ventilation minima during Southern Hemisphere warm episodes, whereas along the rest of the record ventilation minima correlate with cold events. Similar peak-interglacial ventilation minima are indicated at a bottom-water core site in the South Atlantic (ODP 1089, 4621-m water depth) (Fig. 2) (22, 35), which suggests that the events affected the entire water column at high southern latitudes. We are not aware of similar δ13Cb anomalies at core sites upstream along the deepwater flow path in the Atlantic basin (14, 41), which suggests that the anomalies were forced regionally in the Southern Ocean. Teleconnections with coeval maxima in tropical monsoons and reduced southward salt transports may be indicated, plausibly in connection with northward displacements of the intertropical convergence zone that drives precipitation patterns, and thus ocean salt budgets, in the tropics.

The high-amplitude variability seen along our records demonstrates that water mass conversion in the Southern Hemisphere oceans is as sensitive to rapid climate change as in the North Atlantic. Although this does not seem unexpected, it casts new light on the role of Southern Hemisphere ocean THC in climate change. The incursion of abrupt increases of AAIW renewal during periods of thermohaline slowdown in the North Atlantic is intriguing in that it demonstrates synchroneity of changing THC forcing in both hemispheres. The lack of high-resolution records from other sectors of the Southern Ocean does not enable us to unambiguously assess whether the abrupt swings document changes of localized convection in the South Pacific or whether they are representative of mid-depth ventilation changes in wider SAZ. Yet, antiphased δ13Cb patterns between the hemispheres during some glacial, interglacial, and deglacial periods (Figs. 2, 3, and 4) suggest that the THC seesaw operated across different climate states and is a robust feature of the global THC. This then raises the possibility that decreasing AAIW production that is observed during the past few decades (43), through its influence on basin-wide salt budgets and density gradients (4, 6, 7), will have the potential to stabilize thermohaline overturn in the North Atlantic, with implications for climate stability in the course of increasing atmospheric greenhouse-gas concentrations and global warming.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S5


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article