ARTICLES

Abrupt changes in Antarctic IntermediateWater circulation over the past25,000 years

KATHARINA PAHNKE1*, STEVEN L. GOLDSTEIN1,2 AND SIDNEY R. HEMMING1,2

1Lamont-Doherty Earth Observatory of Columbia University, Palisades, New York 10964, USA2Department of Earth and Environmental Sciences, Columbia University, New York 10027, USA*e-mail: kpahnke@ldeo.columbia.edu

Published online: 23 November 2008; doi:10.1038/ngeo360

The circulation of Antarctic Intermediate Water is thought to make an important contribution to the global ocean–climate system,but the details of this interaction are not fully understood. Furthermore, the behaviour of Antarctic Intermediate Water underglacial and interglacial conditions is not well constrained. Here we present a 25,000-year-long record of neodymium isotopicvariations—a tracer of water-mass mixing—from the middle depths of the tropical Atlantic Ocean. Our data reveal abruptly enhancednorthward advection of Antarctic Intermediate Water during periods of reduced North Atlantic overturning circulation during thelast deglaciation. These events coincide with an increase in the formation of Antarctic Intermediate Water and warming in thesouthwest Pacific Ocean, which suggests a tight link with Southern Hemisphere climate. In contrast, the initial incursion of southernsource water into the North Atlantic ∼19,000 years ago coincided with weak Antarctic Intermediate Water formation in the Pacific andreduced overturning in the North Atlantic. We conclude that reduced competition at intermediate water depth at this time allowedexpansion of Antarctic Intermediate Water into the North Atlantic. This early incursion of Antarctic Intermediate Water may havecontributed to freshening of the North Atlantic, perhaps spurring the subsequent collapse of North Atlantic deep convection.

Close correlation of past changes in the meridional overturningcirculation (MOC) of the ocean and climate oscillations suggestsa tight and mechanistic coupling between these two systems1.The most notable of these MOC–climate changes occurred duringthe last glacial and deglacial period, when episodes of icebergand meltwater discharges from the Northern Hemisphere icesheets (Heinrich events) were closely coupled to the collapse ofNorth Atlantic Deep Water (NADW) formation, cooling in largeparts of the Northern Hemisphere1, warming in the SouthernHemisphere2,3 and an increase in Antarctic Intermediate Water(AAIW) formation4. These observations suggest a possible linkbetween AAIW and NADW circulation. Such coupling, with acritical role for AAIW in contributing to or even triggering abruptMOC reorganizations, is also increasingly being suggested bymodel studies5–8. Available mid-depth records from the Atlanticof benthic foraminiferal δ13C and Cd/Ca show nutrient increasesduring Heinrich events and the Younger Dryas, presumably linkedto increased presence of nutrient-rich southern source water atintermediate water depth9–12. But to fully establish the sourceof these δ13C and Cd/Ca anomalies, and for a comprehensiveunderstanding of the behaviour and role of AAIW under differentclimate states and different scenarios of deep overturning in theNorth Atlantic, extra mid-depth palaeo-records are required. Herewe present two records of neodymium isotope ratios (143Nd/144Nd,expressed in εNd notation) measured on the authigenic Fe–Mnoxide fraction of cores from the mid-depth Atlantic Ocean. εNd inauthigenic Fe–Mn oxides dispersed in marine sediments directlyreflects the Nd isotopic composition of the sea water from whichthey precipitated13. Several tests allow verification of the seawater

integrity of εNd (see Supplementary Information) for applicationas a palaeo-water-mass tracer14,15. εNd is imprinted on sea waterthrough weathering and erosion of the continents, and its useas a water-mass tracer stems from the geographic distribution ofthe continents with different lithologies and ages. North Atlanticwater masses (NADW/Glacial North Atlantic Intermediate Water,GNAIW) reflect input of old crustal Nd and have a typical εNd

of −13.5, whereas higher Pacific εNd values of about −4 reflectinput from young volcanic rocks. Southern Ocean and IndianOcean εNd ranges from −6 to −9, reflecting the mixing betweenAtlantic and Pacific source waters. This variability reflects therelatively short residence time of Nd in the ocean of ∼500–2000years16–18. Chemical and biological processes do not fractionateεNd values. In the deep ocean today, εNd therefore behaves as aquasi-conservative water-mass tracer, approximating the mixingratio of source waters13. This makes εNd a powerful tool to constrainpast changes in water-mass mixing, driven by reorganizations of theMOC (refs 14,15).

INTERMEDIATE-WATER εNd IN THE TROPICAL AND SOUTHWEST ATLANTIC

We studied core MD99-2198 (12.09◦N, 61.23◦W, 1,330 m waterdepth) from the Tobago Basin in the western tropical Atlanticand core KNR159-5-36GGC (27.51◦ S, 46.47◦W, 1,268 m waterdepth) from the Brazil margin in the southwest Atlantic overthe past 25,000 years (25 kyr) (Fig. 1). The latter core has beenstudied previously11,19. Both cores are located at the lower limitof the AAIW core layer and within the transition to underlyingdeep-water masses. These are Upper NADW in the Tobago Basin

870 nature geoscience VOL 1 DECEMBER 2008 www.nature.com/naturegeoscience

© 2008 Macmillan Publishers Limited. All rights reserved.

ARTICLES

0.250.250.5

0.5 0.750.75

0.751

11

1.25

1.251.25

1.5

1.5

1.5

1.75

1.75

1.75

2

2

22

2

2.25

2.25

2.252.25

Ocea

n Da

ta V

iew

5,000

4,000

3,000

2,000

1,000

0

60° S 40° S 20° S EQ

Latitude

20° N 40° N 60° N

Dept

h (m

)

NADW

AAIW

KNR159-6-36GGC MD99-2198

AABW

Figure 1 Locations of intermediate-water cores MD99-2198 andKNR159-5-36GGC. The core locations are superimposed on an Atlantic section ofphosphate concentrations49 that depict the different water masses (using the OceanData View software50). AABW: Antarctic Bottom Water.

(MD99-2198) and a mixture of Upper NADW and CircumpolarDeep Water (CDW) on the Brazil margin. With εNd values of AAIWand CDW of −6 to −9 (ref. 20), and NADW of −13.5 (ref. 21),the Tobago Basin core should be very sensitive to changes in therelative influence of Northern versus Southern Component Water(NCW versus SCW), whereas such changes can be expected to bemore muted in the Brazil margin record (Fig. 1).

The core-top εNd in both cores (Tobago Basin -11, Brazil margin−7.3) is within 1 εNd unit of seawater εNd at nearby stations20–22.Continental material at both core sites that could potentially haveoverprinted the authigenic signal shows much lower εNd of −12and −10, respectively (see Supplementary Information, Fig. S1).Moreover, both core-top values fall into the seawater εNd–salinityfield for deep to mid-depth Atlantic sea water, supporting theintegrity of the εNd in terms of reflecting sea water. The downcorerecords show short-term excursions of more positive εNd valuesduring the past 25 kyr, but no discernible long-term trend (Fig. 2).In the Tobago Basin, εNd values were more positive (εNd −10 to−8) than today at around 19–15.3 kyr ago and 13.4–11.5 kyr ago.Similar events are recorded at the Brazil margin at 17–15.7 kyrand 13.5–11.3 kyr, but with lower amplitudes than in the TobagoBasin. These excursions indicate shifts towards enhanced influenceof a water mass with high εNd signature. Possible candidates areSCW and Mediterranean Outflow Water. Although the latter couldpotentially affect εNd in the Tobago Basin, it has εNd values between−9 and−10 (ref. 23), too low to account for εNd values of up to−8in our core. This leaves increases in the relative influence of SCW(AAIW) at the core sites as the most plausible explanation for theεNd excursions.

CHANGES IN NCW VERSUS SCW INFLUENCE

Before interpreting the εNd changes at our core sites strictlyin terms of variations in the mixing ratio of NCW and SCW(NADW/GNAIW and AAIW), we have to consider the stabilityof the endmember εNd values. Several studies indicate thatNorth Atlantic and North Pacific source waters have been nearlyconstant on glacial–interglacial to millennial timescales throughthe Pleistocene epoch (at εNd ∼ −14 and −4, respectively)24–26.In contrast, the Southern Ocean εNd signature, which reflectsthe balance between inputs of North Atlantic and Pacific waters,is likely to have changed with the varying export of NCW(NADW/GNAIW) to the Southern Ocean14,15,27. More specifically,Southern Ocean εNd seems to have shifted towards more Pacific-like(higher) values at times of reduced North Atlantic overturning(for example, Heinrich events, Younger Dryas)28. During the

–8

–6

–7

–8

–9

–10

–11

–12

2

3

4

0 5 10

Age (kyr)

15 20 25

0.4

0.3

0.2

0.1

1

0.8

0.6

0.6

0.4

0.2

ε Nd

ε Nd

δ18O

(‰ V

PDB)

AAIW fraction

AAIW fraction

Cdw (nm

ol kg–1)

YD HE1

SCW

NCW

ε σNd (2 )

a

b

c

d

Figure 2 Atlantic intermediate-water records. a,b, Tobago Basin (MD99-2198) (a)and Brazil margin (KNR159-5-36GGC) (b) εNd records (circles; externalεNd error±0.35 (2σ )) and relative AAIW influence (no symbols).c, Intermediate-water benthic Cd/Ca from cores NEAP4K (black circles)12,OC205-2-103GGC (grey circles)9 and KNR159-5-36GGC (crosses)11.d, Intermediate-water benthic δ18O from M35003-4 (circles)38 andKNR159-5-36GGC (no symbols)19 (both on the Uvigerina scale). The symbols alongthe age axis are age control points for both cores (see the Methods section). Thearrow shows the direction of relative increase in NCW and SCW influence. HE1:Heinrich event 1, YD: Younger Dryas.

Last Glacial Maximum (LGM), on the other hand, GNAIW wasexported from the North Atlantic, partly compensating for the lackof formation and southward transport of NADW (refs 29,30).

To account for these changes in the Southern Ocean εNd

endmember, and to thus permit predictions about changesin the relative influence of NCW versus SCW at the coresites, we use an εNd record from the deep Indian Ocean28.The core V19-188, (6.87◦ S, 60.67◦ E, 3,356 m water depth, seeSupplementary Information, Figs S3,S4) is bathed today by CDW,and its Late Holocene value is consistent with Southern OceanεNd. Sea water and Fe–Mn crusts in the deep Indian Ocean havebeen shown to carry a Southern Ocean εNd signature, pointing toa direct source from the Southern Ocean and its stability in theIndian Ocean31,32. A low-resolution record from the deep SouthAtlantic (42.9◦ S) shows similar absolute values and changes, butthe temporal resolution and position of the Indian Ocean coreoutside the direct influence of NADW make it the best monitor ofSouthern Ocean εNd variations available at present. We thereforeconsider the record a viable profile to account for Southern OceanεNd changes in our intermediate-water records and to carry outfirst-order estimates of changes in the presence of AAIW at thecore sites. The Indian Ocean record shows that Southern OceanεNd changed on glacial–interglacial and millennial timescales, with

nature geoscience VOL 1 DECEMBER 2008 www.nature.com/naturegeoscience 871

© 2008 Macmillan Publishers Limited. All rights reserved.

ARTICLES

shifts to more radiogenic and hence more Pacific-like values before22 kyr as well as during Heinrich event 1, when the formationand southward export of NADW were markedly weaker thantoday (see Supplementary Information, Fig. S2). The LGM (around20 kyr), on the other hand, was marked by more Atlantic-likevalues, consistent with increased southward export of GNAIW atthis time29,30.

In the present day, Antarctic Bottom Water/CDW (εNd =−8to −9) has lower Nd isotope ratios than AAIW (εNd =−6 to −7).Because our Indian Ocean core samples the deep circum-Antarcticwater masses, for mass balance purposes we assume it is 2εNd

units more negative than the coeval intermediate water. Usingthis adjusted record, we estimate the varying Nd concentration inthe Southern Ocean using a simple two-endmember model withtypical Nd concentrations and εNd values for the North Atlanticand North Pacific of 17 pmol kg−1, −13.5 and 40 pmol kg−1, −4,respectively. The εNd and calculated Nd concentration records forthe Southern Ocean were then used to estimate the varying fractionof AAIW at our Atlantic core sites with a two-endmember (NCWand SCW) calculation. The estimated AAIW changes in both coressuggest increased presence of NCW and reduced influence of AAIWduring the LGM compared with today, consistent with evidencefor increased formation and southward transport of GNAIW(refs 29,30) and reduced formation of AAIW (ref. 4) (Fig. 2). Thisweakened inflow of nutrient-rich AAIW into the North Atlanticduring the LGM could account for the lower observed nutrientconcentration of North Atlantic waters at this time33.

On millennial timescales, variations in the Brazil marginrecord disappear when normalized to the temporally changingSouthern Ocean endmember (Fig. 2b). This is probably due tothe fact that we cannot differentiate between the two differentdeeper and shallower Southern Ocean water masses with similarεNd signatures at the core site (CDW and AAIW, respectively)and suggests that the variations seen in the raw εNd data areentirely due to past changes in the Southern Ocean εNd signature(see Supplementary Information, Fig. S2) and not to changes inAAIW circulation. The Tobago Basin record, on the other hand,documents short-term increases in AAIW presence at times thatcoincided with Heinrich event 1 and the Younger Dryas whenNorth Atlantic overturning was weak. These positive excursionsare in contrast with lower AAIW presence during the LGM and at∼14 kyr. The presence of these short-term changes in the TobagoBasin highlights the high sensitivity of this site to changes inthe northward extent of AAIW. In our following discussion ofthe short-term changes, we will therefore focus on the TobagoBasin core.

THE DEGLACIATION: AAIW INCURSIONS INTO THE TROPICAL ATLANTIC

The broad positive εNd excursion in the Tobago Basin around19–15.3 kyr, suggesting increased AAIW presence, is in line with asmall increase in benthic Cd/Ca in core M35003-4 (ref. 10) (samelocation as core MD99-2198), which spans the same time intervaland was also interpreted as an increase in the presence of AAIWat the core site. Other North Atlantic mid-depth Cd/Ca recordsalso indicate the presence of nutrient-rich SCW at intermediatewater depths during Heinrich event 1 and the Younger Dryas(refs 9,11,12) in agreement with the εNd changes in the TobagoBasin (Fig. 2). This scenario is further supported by increasedradiocarbon reservoir ages of North Atlantic surface waters duringthe Younger Dryas that have been attributed to the incursion of old,southern-sourced waters at intermediate water depth34, and by 14Cdata from the North Atlantic that show the presence of 14C-depletedwater similar to present-day AAIW during Heinrich event 1, theYounger Dryas and at 15.4 kyr (ref. 35).

2.0

1.6

1.2

0.8

0.4

0

0

0.4

0.8

1.2

–12

–11

–10

–9

–8

0 5 10 15 20 25

0.1

0.2

0.3

0.4

0.06

0.08

0.10

Age (kyr)

ε Nd

δ13C

(‰ V

PDB)

δ13C

(‰ V

PDB)

AAIW fraction

231Pa/ 230Th

SCW

NCW

ε σNd (2 )

a

b

c

d

YD HE1

Figure 3 Comparison of Atlantic and Pacific intermediate-water records.a, Benthic δ13C from North Atlantic cores M35003-4 (black circles)38, MD99-2198(grey crosses) and SO75-26KL (no symbols)36. b, Deep-water 231Pa/230Th fromOCE326-5GGC27. c, εNd record from MD99-2198 (circles; external εNd error±0.35(2σ )) and relative AAIW influence (no symbols). d, Benthic δ13C from MD97-2120,southwest Pacific4. The grey bars mark Heinrich event 1 and the Younger Dryas; thelight grey bar denotes the increase in the northward extent of the AAIW and reducedNorth Atlantic MOC before Heinrich event 1. The arrow shows the direction ofrelative increase in NCW and SCW influence.

Given the reduced formation of NADW and GNAIW duringHeinrich events and the Younger Dryas, as indicated by a largearray of palaeo-data27,29,36, the increased presence of AAIW inthe North Atlantic at these times could imply that AAIW simplyfilled the resulting void at mid-depth without a change in itsformation rate. However, independent evidence from benthic δ13C(refs 4,37) in the southwest Pacific indicates that the formationand northward penetration of AAIW was enhanced during timesof millennial-scale warm anomalies in the Southern Hemispherethat coincided with Heinrich events and the Younger Dryas (ref. 2).Comparison of the benthic δ13C record from the southwest Pacificwith the εNd changes in the Tobago Basin shows correlation ofincreased AAIW formation in the southwest Pacific (high δ13C)with increased AAIW presence in the tropical Atlantic (high εNd),notably at the time of Heinrich event 1 and the Younger Dryasevent (Fig. 3). This obvious correlation suggests that northwardexpansion of AAIW in the Atlantic was linked to changes in AAIWformation, and in turn to Southern Hemisphere climate. Thissupports model studies that indicate an enhanced Southern Oceanintermediate-water cell when NADW formation was reduced5–8.

ENHANCED NORTHWARD PENETRATION OF AAIW AT∼19 kyr BP

In contrast to the events during the deglaciation, the earliestincrease in AAIW presence in the Tobago Basin at ∼19 kyr wasnot matched by a simultaneous increase in AAIW formation inthe southwest Pacific, where benthic δ13C indicates that AAIW

872 nature geoscience VOL 1 DECEMBER 2008 www.nature.com/naturegeoscience

© 2008 Macmillan Publishers Limited. All rights reserved.

ARTICLES

formation was still at its reduced last glacial level4 (Fig. 3). Instead,the early AAIW increase in our record occurred at a time ofreduced GNAIW formation in the North Atlantic36,38,39 that hasbeen associated with increased iceberg discharge from the Europeanice sheets40,41 and a rise in sea level by some 10 m due to thedisintegration of the Northern Hemisphere ice sheets at ∼19 kyr(ref. 42). A Pa/Th record from the deep northwest Atlantic alsoindicates an early demise of the Atlantic MOC (Fig. 3; ref. 27).Diminished competition at intermediate water depth resultingfrom weak GNAIW/NADW formation would have facilitatedincreased northward penetration of AAIW. Accordingly, this couldexplain the early increase in AAIW influence in the Tobago Basinindependent of changes in AAIW formation strength and SouthernHemisphere forcing.

INTERHEMISPHERIC NADW–AAIW COUPLING

Our mid-depth εNd record documents marked increases in thenorthward penetration of AAIW during the past 25 kyr that werein pace with increases in AAIW formation in the southwest Pacificand Southern Hemisphere warming and occurred at times ofreduced overturning in the North Atlantic (Heinrich event 1,Younger Dryas). In contrast, the early increase in AAIW northwardpenetration at ∼19 kyr was not matched by a similar event in thesouthwest Pacific.

The observed inverse relationship of millennial-scale changesin NADW formation and the northward penetration of AAIWprovides strong evidence for the NADW–AAIW see-saw suggestedby several models5–8. This coupling is maintained by oppositedensity gradients between the high-latitude formation regionsand hence AAIW and NADW in the models. Perturbations ofthis gradient through negative or positive freshwater forcing inone hemisphere trigger a reversal in water-mass formation in theother hemisphere. Our data demonstrate a strong link betweenSouthern Hemisphere climate anomalies and AAIW changesin the Atlantic and Pacific. The former has previously beentaken to suggest a direct response of mid-depth overturning toclimate and surface ocean processes in the high-latitude SouthernHemisphere4. Although our data do not allow us to single outthe ultimate driver of these variations, if increases in AAIWformation were driven by negative freshwater forcing duringSouthern Hemisphere warming, as suggested by simultaneousAAIW and climate changes in the southwest Pacific4, incursionsof AAIW into the North Atlantic could have had a dampeningeffect on North Atlantic overturning6,7. This see-saw did not operateat ∼19 kyr, when northward incursion of AAIW into the NorthAtlantic must be attributed to freshwater-induced reduction ofGNAIW formation36,38–41 and resulting relaxation of mid-depthcompetition between northern and southern source waters. Thisintroduction of fresh AAIW to the upper water column in theNorth Atlantic may have been involved in the freshening of theNorth Atlantic and thereby contributed to the collapse of NADWformation during Heinrich event 1.

SUMMARY

Our Nd isotope data provide unequivocal evidence for changesin the northward extent of AAIW in the Atlantic and an inversecoupling of NADW/GNAIW and AAIW circulation. The observedmid-depth εNd changes thus strengthen the notion of a key rolefor AAIW in abrupt reorganizations of the global MOC. Takingthis into consideration, along with the observed rapid response ofAAIW to current climate warming43,44, the possibility of a swiftand substantial impact of present and future climatic changes onthe global MOC is a real concern. Monitoring of AAIW changes,

both in the past and today, may therefore hold important clues topast and future MOC scenarios and the link between climate andocean circulation.

METHODS

Cores MD99-2198 (12.09◦ N, 61.2◦W, 1,330 m water depth) andKNR159-5-36GGC (27.51◦ S, 46.47◦W, 1,268 m water depth) were sampled at5–10 cm and 1–16 cm intervals, respectively (∼4 cm3) and the <63 µm fractionwas separated through wet sieving. The Fe–Mn fraction was extracted fromthe <63 µm size fraction following the procedures of refs 14,15. Briefly, afterremoval of all carbonate from the sediment, the samples were leached with0.02 M hydroxylamine hydrochloride to recover Fe–Mn oxides. For Nd andSr isotope analyses on the detrital fraction, the residual sediment was leachedwith hydroxylamine hydrochloride another 2–3 times to remove all remainingFe–Mn oxides. Some 50–100 mg of dry sediment was then digested in 3:1nitric acid and hydrofluoric acid. Separation of Nd and Sr was achieved usingEichrom Tru-Spec resin. Nd was further isolated using α-hydroxyisobutyricacid and cation resin. Eichrom Sr-Spec resin was used to isolate Sr. Nd and Srisotopes were measured on a VG (Micromass) Sector 54 thermal ionizationmass spectrometer at LDEO by dynamic multicollection. 143Nd/144Nd ratioswere normalized to 146Nd/144Nd= 0.7219 to correct for mass fractionationand adjusted to the value of the La Jolla standard of 146Nd/144Nd= 0.511860.87Sr/86Sr ratios were normalized to 87Sr/86Sr= 0.1194 and adjusted toNBS987= 0.710230. External reproducibility from routine runs of the La Jollaand NBS987 standards was 143Nd/144Nd= 0.5118597±0.000018 (2σ, n= 154)and 87Sr/86Sr= 0.710265±0.000024 (2σ, n> 22), respectively.

Stable carbon isotope ratios (δ13C) were measured on Cibicidoides spp.along a section in core MD99-2198 of large δ13C changes (LDEO stable isotopefacility). We used this benthic δ13C section to fine-tune the age scale for coreMD99-2198 (see below).

The age scale for MD99-2198 is based on four 14C accelerator massspectrometry (AMS) dates (NOSAMS/NSF facility at WHOI) measured onmixed planktonic foraminifera Globigerinoides ruber and G. sacculifer andconverted to calendar ages using the Calib5.0.1 program45 and the Marine04data set46 (see Supplementary Information, Table S1). In addition, we correlatedthe magnetic susceptibility47 record and the benthic δ13C record of coreMD99-2198 to that of core M35003-4 from the same location using five andtwo tie-points, respectively (see Supplementary Information, Fig. S1). The agescale of core M35003-4 in this interval is constrained by 10 14C AMS dates38,48

that were updated using the Calib5.0.1 program and the Marine04 calibrationcurve46. Two 14C AMS dates were disregarded, 14.75 kyr and 15.85 kyr (seeSupplementary Information, Table S1). The former is within the uncertainty ofthe age model of core M35003-4 that we used to constrain the age model forcore MD99-2198 (by tuning magnetic susceptibility and benthic δ13C records,see above). The latter shows an age reversal and deviates considerably from theotherwise linear age versus depth correlation.

The age scale for core KNR159-5-36GGC is based on 15 14C ages11,19.

Received 25 April 2008; accepted 21 October 2008; published 23 November 2008.

References1. Broecker, W. S. Massive iceberg discharges as triggers for global climate change. Nature 372,

421–424 (1994).2. Blunier, T. & Brook, E. J. Timing of millennial-scale climate change in Antarctica and Greenland

during the last glacial period. Science 291, 109–112 (2001).3. Pahnke, K., Zahn, R., Elderfield, H. & Schulz, M. 340,000-year centennial-scale marine record of

Southern Hemisphere climatic oscillation. Science 301, 948–952 (2003).4. Pahnke, K. & Zahn, R. Southern Hemisphere water mass conversion linked with North Atlantic

climate variability. Science 307, 1741–1746 (2005).5. Keeling, R. F. & Stephens, B. B. Antarctic sea ice and the control of Pleistocene climate instability.

Paleoceanography 16, 112–131 (2001).6. Saenko, O. A., Weaver, A. J. & Gregory, J. M. On the link between the two modes of the ocean

thermohaline circulation and the formation of global-scale water masses. J. Clim. 16,2797–2801 (2003).

7. Weaver, A. J., Saenko, O. A., Clark, P. U. & Mitrovica, J. X. Meltwater pulse 1A from Antarctica as atrigger of the Boelling/Alleroed warm interval. Science 299, 1709–1713 (2003).

8. Stouffer, R. J., Seidov, D. & Haupt, B. J. Climate response to external sources of freshwater: NorthAtlantic versus the Southern Ocean. J. Clim. 23, 436–448 (2007).

9. Marchitto, T. M. Jr., Curry, W. B. & Oppo, D. W. Millennial-scale changes in North Atlanticcirculation since the last glaciation. Nature 393, 557–561 (1998).

10. Zahn, R. & Stuber, A. Suborbital intermediate water variability inferred from paired benthicforaminiferal Cd/Ca and δ13C in the tropical West Atlantic and linking with North Atlantic climates.Earth Planet. Sci. Lett. 200, 191–205 (2002).

11. Came, R. E., Oppo, D. W. & Curry, W. B. Atlantic Ocean circulation during the Younger Dryas:Insights from a new Cd/Ca record from the western subtropical South Atlantic. Paleoceanography 18,PA1086 (2003).

nature geoscience VOL 1 DECEMBER 2008 www.nature.com/naturegeoscience 873

© 2008 Macmillan Publishers Limited. All rights reserved.

ARTICLES

12. Rickaby, R. E. M. & Elderfield, H. Evidence from the high-latitude North Atlantic for variations inAntarctic Intermediate water flow during the last deglaciation. Geochem. Geophys. Geosyst. 6,Q05001 (2005).

13. Goldstein, S. L. & Hemming, S. R. in Long-Lived Isotopic Tracers in Oceanography, Paleoceanography,and Ice-Sheet Dynamics Vol. 6 (eds Holland, H. D., Turekian, K. K. & Elderfield, H.) 453(Elsevier, 2003).

14. Rutberg, R. L., Hemming, S. R. & Goldstein, S. L. Reduced North Atlantic Deep Waterflux to the glacial Southern Ocean inferred from neodymium isotope ratios. Nature 405,935–938 (2000).

15. Piotrowski, A. M., Goldstein, S. L., Hemming, S. R. & Fairbanks, R. G. Intensification and variabilityof ocean thermohaline circulation through the last deglaciation. Earth Planet. Sci. Lett. 225,205–220 (2004).

16. Goldstein, S. L. & O’Nions, R. K. Nd and Sr isotopic relationships in pelagic clays andferromanganese deposits. Nature 292, 324–327 (1981).

17. Jeandel, C., Bishop, J. K. & Zindler, A. Exchange of neodymium and its isotopes between seawater andsmall and large particles in the Sargasso Sea. Geochim. Cosmochim. Acta 59, 535–547 (1995).

18. Tachikawa, K., Athias, V. & Jeandel, C. Neodymium budget in the modern ocean andpaleo-oceanographic implications. J. Geophys. Res. 108, 3254 (2003).

19. Oppo, D. W. & Horowitz, M. Glacial deep water geometry: South Atlantic benthic foraminiferalCd/Ca and δ13C evidence. Paleoceanography 15, 147–160 (2000).

20. Jeandel, C. Concentration and isotopic composition of Nd in the South-Atlantic Ocean. Earth Planet.Sci. Lett. 117, 581–591 (1993).

21. Piepgras, D. J. & Wasserburg, G. J. Rare earth element transport in the western North-Atlanticinferred from Nd isotopic observations. Geochim. Cosmochim. Acta 51, 1257–1271 (1987).

22. Tachikawa, K., Jeandel, C. & Roy-Barman, M. A new approach to the Nd residence time in the ocean:The role of atmospheric inputs. Earth Planet. Sci. Lett. 170, 433–446 (1999).

23. Spivack, A. J. & Wasserburg, G. J. Neodymium isotopic composition of the Mediterranean outflowand the eastern North Atlantic. Geochim. Cosmochim. Acta 52, 2767–2773 (1988).

24. Abouchami, W. et al. Secular changes of lead and neodymium in central Pacific seawater recorded bya Fe–Mn crust. Geochim. Cosmochim. Acta 61, 3957–3974 (1997).

25. van de Flierdt, T. et al. Temporal stability of the neodymium isotope signature of the Holocene toglacial North Atlantic. Paleoceanography 21, PA4102 (2006).

26. Foster, G. L., Vance, D. & Prytulak, J. No change in the neodymium isotope composition of deepwater exported from the North Atlantic on glacial–interglacial timescales. Geology 35, 37–40 (2007).

27. McManus, J. F. et al. Collapse and rapid resumption of Atlantic meridional circulation linked todeglacial climate changes. Nature 428, 834–837 (2004).

28. Goldstein, S. L. et al. Quantifying changes in the global thermohaline circulation: A circum-Antarcticperspective. US Geol. Survey . The National Academies USGS OF-2007-1047, ExtendedAbstract 209 (2007).

29. Boyle, E. A. & Keigwin, L. North Atlantic thermohaline circulation during the past 20,000 yearslinked to high-latitude surface temperature. Nature 330, 35–40 (1987).

30. Curry, W. B. & Oppo, D. W. Glacial water mass geometry and the distribution of δ13C of 6CO2 in thewestern Atlantic Ocean. Paleoceanography 20, PA1017 (2005).

31. Bertram, C. J., Elderfield, H. & Shackleton, N. J. Cadmium/calcium and carbon isotopereconstructions of the glacial northeast Atlantic Ocean. Paleoceanography 10, 563–578 (1995).

32. Albarede, F., Goldstein, S. L. & Dautel, D. The neodymium isotopic composition of manganesenodules from the Southern and Indian oceans, the global oceanic neodymium budget, and theirbearing on deep ocean circulation. Geochim. Cosmochim. Acta 61, 1277–1291 (1997).

33. Sigman, D. M., Lehman, S. J. & Oppo, D. W. Evaluating mechanisms of nutrient depletion and 13Cenrichment in the intermediate-depth Atlantic during the last ice age. Paleoceanography 18,PA1072 (2003).

34. Cao, L., Fairbanks, R. G., Mortlock, R. A. & Risk, M. J. Radiocarbon reservoir age of high latitudeNorth Atlantic surface water during the last deglacial. Quat. Sci. Rev. 26, 732–742 (2007).

35. Robinson, L. F. et al. Radiocarbon variability in the western North Atlantic during the lastdeglaciation. Science 310, 1469–1473 (2005).

36. Zahn, R. et al. Thermohaline instability in the North Atlantic during meltwater events: Stable isotopeand ice-rafted detritus records from core SO75-26KL, Portuguese margin. Paleoceanography 12,696–710 (1997).

37. Bostock, H. C., Opdyke, B. N., Gagan, M. K. & Fifield, L. K. Carbon isotope evidence for changes inAntarctic Intermediate Water circulation and ocean ventilation in the southwest Pacific during thelast deglaciation. Paleoceanography 19, PA4013 (2004).

38. Huls, M. & Zahn, R. Millennial-scale sea surface temperature variability in the western tropical NorthAtlantic from planktonic foraminiferal census counts. Paleoceanography 15, 659–678 (2000).

39. Hall, I. R. et al. Accelerated drawdown of meridional overturning in the late-glacial Atlantic triggeredby transient pre-H event freshwater perturbation. Geophys. Res. Lett. 33, L16616 (2006).

40. Knutz, P. C. et al. Multidecadal ocean variability and NW European ice sheet surges during the lastdeglaciation. Geochem. Geophys. Geosyst. 3, Q1077 (2002).

41. Peck, V. L. et al. High resolution evidence for linkages between NW European ice sheet instability andAtlantic Meridional Overturning Circulation. Earth Planet. Sci. Lett. 243, 476–488 (2006).

42. Clark, P. U., McCabe, A. M., Mix, A. C. & Weaver, A. J. Rapid rise of sea level 19,000 years ago and itsglobal implications. Science 304, 1141–1144 (2004).

43. Bindoff, N. L. & Church, J. A. Warming of the water column in the southwest Pacific Ocean. Nature357, 59–62 (1992).

44. Banks, H. T. et al. Are observed decadal changes in intermediate water masses a signature ofanthropogenic climate change? Geophys. Res. Lett. 27, 2961–2964 (2000).

45. Stuiver, M. & Reimer, P. J. Extended 14C database and revised CALIB radiocarbon calibrationprogram. Radiocarbon 35, 215–230 (1993).

46. Hughen, K. A. et al. Marine04 marine radiocarbon age calibration, 0–26 cal kyr BP. Radiocarbon 46,1059–1086 (2004).

47. Labeyrie, L., Cortijo, E., Jansen, E. & Shipboard Science Party, MD114/IMAGES V cruise report,Institut Polaire Francais Paul-Emile Victor (IPEV), les rapports de campagnes a la mer, OCE/2003/02,850 (2003).

48. Ruhlemann, C. et al. Warming of the tropical Atlantic Ocean and slowdown of thermohalinecirculation during the last deglaciation. Nature 402, 511–514 (1999).

49. Garcia, H. E., Locarnini, R. A., Boyer, T. P. & Antonov, J. I. World Ocean Atlas 2005, Vol. 4: Nutrients(Phosphate, Nitrate, Silicate), NOAA Atlas NESDIS 64, (ed. S. Levitus) 396 (US Government PrintingOffice, 2006).

50. Schlitzer, R. Ocean Data View, <http://odv.awi.de> (2007).

Supplementary Information accompanies the paper at www.nature.com/naturegeoscience.

AcknowledgementsWe thank R. Zahn (Universitat Autonoma de Barcelona, Spain) and D. Oppo (Woods HoleOceanographic Institution, USA) for providing core material, S. Gitt (UNC Chapel Hill) for laboratoryassistance and R. Mortlock (LDEO) for running the stable isotope mass spectrometer. K.P. thanksT. van de Flierdt, D. Zylberberg and Y. Cai for introducing her to laboratory procedures and methods.This study was supported by a Lamont Postdoctoral Fellowship to K.P., and grants from the USNational Science Foundation and National Oceanographic and Atmospheric Administration. The datawill be made available on the NOAA Paleoclimatology website (www.ngdc.noaa.gov/paleo/data.html).This is LDEO contribution no. 7214.

Author contributionsSamples were processed and analysed by K.P. All authors contributed to the discussion andinterpretation of the results.

Author informationReprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions.Correspondence and requests for materials should be addressed to K.P.

874 nature geoscience VOL 1 DECEMBER 2008 www.nature.com/naturegeoscience

© 2008 Macmillan Publishers Limited. All rights reserved.