A 135,000-YEAR VOSTOK-SPECMAP COMMON TEMPORAL FRAMEWORK 3Lamont-Doherty Earth Observatory Palisades, New

PALEOCEANOGRAPHY,

VOL. 8, NO. 6, PAGES 737-766, DECEMBER 1993

A 135,000-YEAR VOSTOK-SPECMAP COMMON

TEMPORAL

FRAMEWORK

Todd Sowers,1 Michael Bender,l,2

LaurentLabeyrie,2DougMartinson,3 JeanJouzel,4,5 DominiqueRaynaud,6 JeanJacquesPichon,7and YevgeniySergeevich Korotkevich8

Abstract.The objectof thepresentstudyis to introducea meansof comparingtheVostokand marinechronologies.Our strategyhasbeento use

surfacewatersof the world'soceans.We compare ourrecordof •518Oatm to the •518Osw recordwhich hasbeendevelopedfrom studiesof theisotopic

the•5180of atmospheric 0 2 (denoted •518Oatm) from

composition of biogenic calcite(•5•8Oforam) in deep-

theVostokice coreasa proxyfor the•5180of seawater (denoted •S18Osw). Ourunderlying premise in using•518Oatm asa proxyfor •518Osw is thatpast variationsin •5•8Osw (anindicatorof continental ice volume)havebeentransmitted to theatmospheric 02 reservoirby photosynthesizing organismsin the

seacores.We havetied our •5•8Oatm recordfrom Vostokto the SPECMAP timescalethroughoutthe

•Graduate Schoolof Oceanography, Universityof RhodeIsland,Narragansett. 2CentredesFaiblesRadioactivites, Laboratorie Mixte CNRS-CEA, Gif Sur Yvette, France.

3Lamont-Doherty EarthObservatory Palisades, New York.

4Laboratoire de Modelisation du Climat et de

l'Environnement,Gif Sur Yvette Cedex, France.

5AlsoatLaboratoire deGlaciologie etGeophysique de l'Environnement, St. Martin d'Heres Cedex, France.

6Laboratoire deGlaciologie et Geophysique de l'Environnement, St. Martin d'Heres Cedex, France.

7Department Gdologieet Ocdanologie, Universitd Bordeaux, Talence Cedex France.

8ArcticandAntarcticResearchInstitute,St. Petersburg,Russia. Copyright1993 by theAmericanGeophysical Union. Papernumber93PA02328. 0883-8305/93/93 PA-023 28510.00

last135kyr by correlating •S18Oatrn witha •5•8Osw record from V 19-30. Results of the correlation indicate that 77% of the variance is shared between these two records. We observed differences between

the•518Oatm andthe•518Osw recordsduringthe coldestperiods,whichindicatethattherehavebeen subtlechanges in thefactorswhichregulate•5•8Oatm otherthan•518Osw. Ouruseof •518Oatm asa proxy for •518Osw mustthereforebe considered tentative, especiallyduringtheseperiods.By correlating •518Oatm with •518Osw, we providea common temporalframeworkfor comparingphase relationships betweenatmospheric records(from ice cores)and oceanographic recordsconstructedfrom deep-seacores. Our correlatedage-depthrelationfor the Vostok core should not be considered an absolute

Vostoktimescale.We considerit to be the preferred timescalefor comparingVostokclimaterecordswith marineclimaterecordswhichhavebeenplacedon the SPECMAP

timescale.

We have examined the

fidelity of this commontemporalframeworkby comparingseasurfacetemperature(SST) records from sedimentcoreswith an Antarctictemperature record from the Vostok ice core. We have demonstrated that when the southern ocean SST and

Antarctictemperaturerecordsarecomparedon this commontemporalframework,they showa high degreeof similarity. We interpretthisresultas supportingouruseof the commontemporal frameworkfor comparingotherclimaterecordsfrom

738

Sowers et at.: A 135-ka Vostok-SPECMAP

Correlation

the Vostokice corewith any climaterecordthathas beencorrelatedinto the SPECMAP chronology.

timescale[Imbrie et al., 1984] or the higherresolutionshorttimescale[Martinson et al., 1987]). The stratigraphic basesfor thesetimescales arethe

INTRODUCTION

SPECMAPstacked•518Oforam records(longrecord

Studiesof ice coreshavesuppliedvastamountsof information

[Prell et at., 1986] and short record [Pisias et at., 1984]) which allow directtransferof the timescales

about late Pleistocene climate.

throughcorrelation withanyother•518Oforam versus

Variationsin the isotopiccompositionof the ice from

agerecords.The basisfor thesetwo SPECMAP chronotogies is thatvariationsin theEarth'sorbital geometryinfluenceincomingsolarradiationwhichin turn forces,or at leastpaces,variationsin the sizeof thecontinentalice sheets[Hayset al., 1976a;Imbrie et at., 1984]. The higher-resolution shortSPECMAP chronology(usedhere)was constructed usingfour different"tuning"strategies involvingfive proxy climateindicators (includingthe•S18Oforam record) from RC11-120 [Hayset at., 1976a,b] (Figure 1) as well asthe SPECMAP stacked•518Oforam record

Greenland

and Antarctic

ice cores have been used to

constructtemperaturerecordsfor thesesites [Dansgaardet at., 1985;Loriuset at., 1985;Hammer et at., 1986; Jouzet et al., 1987; Johnsenet at.,

1992]. Otherimportantrecordswhichhavebeen constructed from ice coreanalysesincludethe compositionof the pateoatmosphere [Neffelet al., 1988; Raynaudet at., 1988; Staufferet at., 1988; Chappettazet at., 1990; Bamotaet at., 1991; Etheridgeet at., 1992] and a recordof aerosol transportto the ice sheets[Raisbecket al., 1987; Petit et at., 1990].

The originalchronologyfor the Vostokice core was constructed usinga two-dimensionalrheotogic modelof the ice flow upstreamfrom Vostokin conjunctionwith a recordof the pateoaccumutation rate [Loriuset at., 1985]. The chronologyfor the trappedair parcelswasderivedby calculatinga recordof the ice age-gasagedifference(Aage) [Bamotaet at., 1987,1991]andsubtracting it from the ice ageversusdepthrelationof Loriuset at. [ 1985]. The approachadoptedhereis to correlate

ourrecordof the•5180of atmospheric 0 2 (•518Oatm) from Vostokwith the •5180of seawater(1518Osw) derived from studies of the/5180 of foraminiferal

calcite.Ourunderlying premisein using•518Oatm as a proxyfor •518Osw is thatpastvariations in •518Osw havebeentransmitted to the atmospheric 0 2 reservoir by photosynthesizing organismsin the surface watersof the world'soceans. We compareour recordof •518Oatm to the •518Osw recordwhichhas beendevelopedfrom studiesof the isotopic composition of biogeniccalciteof benthic foraminifera(/518Oforam) in deep-sea cores[Labeyrie et at., 1987; Shackleton,1987]. In making sucha correlationwe derivea directestimateof a gasage versusdepthrelationfor Vostokthatis consistent with the marinechronostratigraphy. Our correlation providesthebasisfor a moredirectcomparison of the atmospheric pCO2 andpCH4 recordsfrom Vostok with climaterecordsobtainedfrom deep-sea sedimentstudiescovetingthe last 140 kyr.

[Pisiaset at., 1984]. Errors in this timescale,

estimatedusingvariousclimateindicators,deepsea cores,andtuningstrategies,average+3.5 kyr and rangefrom 1 to 7 kyr. The shortSPECMAP timescalehasalsoservedasthe basisfor correlating otherrecordssuchasthe China toessdeposits [Kukla and An, 1989; Hovan et al., 1991] andpollen recordsfrom North AmericaandEurope[Guiotet at., 1989].

Ice Core Chronostrati graphies

Datingice coresretrievedfromtheEastAntarctic plateauhasprovendifficultbecauseof thelack of annualstratigraphic layersin the ice. Loriuset at. [ 1985] establisheda timescalefor the Vostok ice core with the use of a two-dimensional

ice flow model.

Their approachutilizeda recordof the paleoaccumutation ratein conjunctionwith model calculations of the degreeto whichannuallayers havebeenthinnedduringtransportwithintheEast Antarcticice sheet(commonlyreferredto asthe thinningfunction). Recentsensitivitystudies suggestthatan ice age-depthprofilefor an ice coreis largelydependenton thepaleoaccumutation record used[Ritz, 1992]. The originalVostokchronology wasconstructed by assuming thatthe accumulation rateupstreamfrom Vostokwasdirectlyrelatedto the

•SDor •5180of theprecipitation.Thisassumption waslatersupported by 10Bemeasurements [Yiouet at., 1985; Raisbeck et at., 1987; Jouzet et at., 1989; Raisbeck et at., 1992]. On the other hand, the

average•SDice fromtheDomeB core(located~300 km upstreamfrom Vostok)duringtheHoloceneis ICE CORE AND DEEP-SEA CHRONOSTRATIGRAPHIES

CORE

Deep-SeaChronostrati graphies Most records of Pleistocene climate constructed

from deep-seacorescanbe placedinto oneof the two SPECMAP timescales(thelower-resolution long

7%0higherthanthatfortheHolocene section of the Vostokice core[Kottyakov,1990]. The endof the lastglacial-interglacial transition(•-10ka) is about 140 m deeperat Dome B thanat Vostok. The inferredaccumulationrateat Dome B overthisperiod is thus60% higherthanat Vostok. This resultis muchhigherthanthe 10% differencewhichwould be derivedfrom thedifferencebetweenthe •SDice

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

739

dustrecordwith the magneticsusceptibilityrecord from RC11-120 (takento be a dustproxy). They suggested that the Loriuset al. [ 1985] agefor the marinestage5e was about10 kyr olderthanthe

SPECMAPage,andnotedthat23øTh]234U dating studiesof coralsfavoredthe SPECMAP age [Bard et al., 1990b] (but seeLambeck and Nakada [ 1992] for 60 ø 70" 70 ø

8O*

Rc• 9O ø 100 ø

too*

110 ø 120 ø

120 •

130"

140 ø

130 •

170 ø

I•0 ø

170 ø

I•0 ø

150 •

140 •

Fig. 1. Index map showingthe locationof thedeepseasedimentcore (MD88-770, 46os, 96OE) as well

astheAntarcticice coresincludedin our study. Also plottedarethe summerseasurfacetemperature isotherms for the southern ocean.

anotherview). Finally, Pichonet al. [ 1992] compareda recordof seasurfacetemperature(SST) from the southernoceanwith the Vostoktemperature recordto derivea new age-depthcurvefor Vostok. Pichonet al. [1992] alsosuggested thatthe Loriuset al. [ 1985] age for stage5e was about6 kyr older thanthe corresponding SPECMAP age. The net result of these correlation

studies is to

proposevariousice age-depthprofilesfor Vostok which are generallywithin the stateduncertainties of the originalLorius et al. [ 1985] chronology.The presentwork providesanothermethodof correlating the Vostokandmarinetimescales.We emphasize thatthe goalof recentcorrelationstudiesby Petitet al. [ 1990], Shackleton et al. [ 1992], and Pichon et

al. [ 1992] hasbeento placethe Vostokclimate recordsinto the SPECMAP chronology.None of the correlationstudiesgivesany indicationabout which absolutechronologyis superior.In the next sectionwe useourrecordof •jl 8Oatmto correlatethe Vostok and SPECMAP

timescales.

We focus on

developinga commontimescalewhichis suitablefor comparingthe Vostokandmarineclimaterecords. The resultsfrom this studyhaveno bearingon the absolute nature of the Vostok timescale itself.

valuesof the ice at the two sites. Ritz [ 1992] concluded from this observation that the

accumulation ratenearVostokis not solelycontrolled by thetemperatureof theinversionlayer. Other factors,suchasthe orographicandradiativecooling appearto significantlyinfluencethe accumulation rate overEastAntarctica(C. Ritz, Chronologyof the Vostokice corebasedon precipitationandice flow modeling,submittedto Journalof Glaciology, 1993). Furthermore, Jouzel et al. [1992] showed

that,whenthe sameglaciological modelwasapplied to boththe VostokandDome C cores,the timingof theminimumin the dustflux nearthebeginningof thelastglacialterminationat thetwo siteswas differentby about1 kyr. Theseresultshavebeen incorporatedinto a revisedflow modelfor the Vostok area which now extends to 2546 meters

below the surface [Jouzel et al., 1993]. Minor modificationsto the Lorius et al. [ 1985]

Vostokchronologywere proposedin threelater studies. Jouzel et al. [1987] used a more detailed

Finally, we compareour correlationresultswith previouscorrelationsanddiscussthe implicationsfor leadsandlagswithin the climatesystem.

COMPARINGTHE •j180OF ATMOSPHERIC0 2

WITH

THE •j180 OF SEAWATER

A Recordof the t5180of SeaWater We haveconstructed a recordof •j18Osw from the benthic•j18Oforam record(Uvigerinasenticosa) from V 19-30 (3ø2I'S, 83ø2I'W, 3091-m water depth) [Shackletonand Pisias, 1985]. We chosethe V19-30

corefor two reasons.First,the •j18Oforam record from thiscorewasusedin constructing the SPECMAP stackedrecord [Pisiaset al., 1984]. Consequently,the SPECMAP agemodelwas directlyappliedto thiscoreduringthedevelopment of the chronostratigraphy. Second,the long

•i18Oforam record(>300kyr) fromthiscoreprovides a targetcurvefor correlating •j18Oatm variations from

•iDicerecordin placeof the•i18Oice recordfor

deeper(yet to be drilled) sectionsof the Vostokcore

estimatingthe inversionlayertemperature above Vostokandthe paleoaccumulation rate. They recalculated the Vostokage-depthcurvebasedon the revisedpaleoaccumulation ratesandfoundthatthe revisedchronologyagreedwell with thatof Loriuset

into the SPECMAP

al. [1985]. Petit et al. [ 1990] correlatedthe Vostok

foraminiferal •i180 record itself. The first is the

timescale.

We have normalized

the•jl8Oforam valuesby arbitrarilyremoving3.46%0 from all samples,sothatthe averageHolocene •jl 8Oforam valueis 0%0. There are two factors which influence the

740

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

•jl8Osw,andthesecond isbottom-water temperature.

Results are listed in Table 1.

To correctfor bottom-water temperature changes, we havemadethefollowingassumptions: (1) glacial bottomwaterswere 1.7øCcolderthaninterglacial temperatures [ChappellandShackleton,1986; Birchfield,1987;Shackleton, 1987],(2)deepwaters warmedduringthe terminationsthemselvesand cooledduringthetransitionfromthelastinterglacial (stage5e) to theglacialperiod(stage5d), startingat about120 ka in the SPECMAP chronology [FairbanksandMatthews,1978;Labeyrieet al., 1987;Shackleton,1987],and(3) thetemperature changeoccurredovera 5-kyr periodcenteredwithin the terminationsor 5e-5dboundary.To correctfor thesetemperature variations, we subtract 0.4%0from all glacial•518Oforam data[Shackleton andOpdyke, 1973] andapplythe correctiongraduallyoverthe 5kyr transitionperiods.The exacttimingof the deepwatertemperature changeswithina transitionis difficultto assess on a globalscale. Giventhatthe averagelengthof the last two terminationsis 10-12 kyr, ourplacementof thetemperature changesin the middleof the transitions introduces a 1-3 kyr uncertaintyin the exacttimingof thechangesin deepwatertemperature.This uncertaintyis onelimit on the accuracyof ourcommontemporalframework for theseperiods.

We sampledthe Vostok3F coreroughlyevery25 m startingat 140 m belowthe surface(mbs)and continuingdownto 2058 mbs. Given an ageof 160

A Recordof the(5180ofAtmospheric 02 Fromthe Vostok Ice Core

We havemeasured the isotopiccomposition of trapped0 2 andN 2 from95 discretedepthsalongthe Vostok 3F (2083 m) and BH-1 (178 m) cores. Fossilair sampleswereextractedfrom 12- to 16-g ice samplesby allowingtheice samples to meltinvacuo.The -1-cm3 (STP) air samplewasthen quantitatively transferred to a stainless steelsample tubeimmersedin liquidhelium. The air samplewas theneitherequilibrated intoa glassampouleand sealedfor lateranalysisor introduceddirectlyintothe samplereservoirof a FinniganMAT 251 isotope ratiomassspectrometer whereit wasanalyzed againstan aliquotof dry air [Sowerset al., 1989, 1991]. We reportour resultsusingthe delta notation,wherethe referenceis thepresentday air:

8002 ice ={[ 180160/1602 (air) /]_ 180160/1602 (sa) 1}103 (1)

Results of the•515N of trapped N2 (•j15NN2 ice)and •5180 oftrapped 0 2(•518Oo2 ice)arereporte-d inTable 1. The •j15NN9 'iceresultsareusedto correctthe

•j18Oo2 icevalu-es forgravitational fractionation of air in the tim priorto occlusion[Craiget al., 1988; Schwander,1989; Sowerset al., 1989] accordingto the followingequation:

•jl8Oatm = •j18002 ice- 2(•j15NN2 ice)

(2)

ka for the bottom (2083 mbs) of the 3F core [Lorius

et al., 1985], our samplingfrequencyfor the deep corecorresponds roughlyto onesampleevery2,000 years. This valueis comparableto the current turnovertime of atmospheric 0 2 [Benderet al., 1985]. In additionto samplingevery25 m, we performedhigh-resolution samplingduringperiods of majorclimatechange.On the otherhand,poor corequalitylimitedoursamplingto 50 m between 500- and 700-m depth. We alsoanalyzed17 samplesfromthe 114.8-to 171.1-mdepthintervalin theBH-1 shortcore(178 m) whichwasdrilledduringthe 35th Soviet Antarcticexpeditionin 1989-1990.Thesesamples provideda high-resolution recordcoveringthelast3 kyr. Both the 3F and the BH-1

short cores were drilled

usinga thermaldrill. As a result,corequalityis poor at shallowdepths.Therearetwo indications thatthe accuracyof our resultsfor samplesabove1000 mbs may havebeenslightlycompromised by thispoor

corequality. First,•jl8Oatmin theHolocenesamples of Vostokis significantlygreaterthanthe modem atmospheric valueadoptedasthereferenceand thereforedefinedto be 0%0. The mean•jl 8Oatmof samples with ages< 2.5 kyr is 0.19+ 0.09%0(1 = 21). Analysesof samplesin this agerangefrom thehigh-quality GISPII core,indicatethatthe•5180 of atmospheric 02 hasbeen0.1+0.1%0(1(5,n= 32) duringthis interval (T. SowersandM. Bender, unpublished data, 1993). Second,someadjacent samples indicateabruptchanges in •5180whichare geochemically unreasonable giventhe 2-kyr turnover time of atmospheric 02. Thustheuncertainty in •jl8Oatm in theupperpartof theVostokcoreis greaterthanourprecision;we assigna valueof + 0.2%0.Thereis no masonto believethatthereis any sucherrorin samplesbelow 1,000-mdepth(-60 ka), wherepreservation is very good. A plotof •518Oatm versusdepthin theVostok coreswas constructed by connectingaverage •j18Oat mvaluesat eachdepth(Figure2). The averagestandarddeviationaboutthemeanof all 95

depthsis _+0.05%0. We compare the•JDice record from Jouzelet al. [1987] with the •518Oatm recordin Figure2. In general,high•jl 8Oatmvaluesare associated with low •JDice valueswhicharegenerally thoughtto be indicativeof colderperiods. In orderto compareour•jl8Oatm datawiththe •jl 8Oswrecordin the time domain,we utilizethe Loriuset al. [1985] ice chronologyandthe Aage estimatesfrom Bamola et al. [ 1991]. Today,in the Vostokregionof Antarctica,air parcelsarebeing occluded into bubbles between 95 and 105 mbs

[Bamolaet al., 1987; Bamola et al., 1991]. The age of the ice at thesetwo depthsis 2,600 and3,000

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

741

TABLE 1. Resultsof TrappedGasAnalysisFromtheVostokIce Core •j180 of

Sample Depth,IceAge,a GasAge,b Trapped Gases . mb• ka ka 815N. • õ•80.• 114.8a 114.8b 116.3a 116.3b 120.3a 120.3b 125.2a 125.2b 130.1a 130.1b 135.3a 135.3b 139.7a 139.7b 139.8a 139.8b 143.7a 143.7b 149.6a 149.6b 152.8 156.9 161.8a 161.8b 165.1a 165.1b 166.8a 166.8b 169.7a 169.7b 169.7c 174.6 177.1a 177.1b 184.3a 184.3b 213.8a 213.8b 269.5a 269.5b

3.6

0.8

3.7

0.9

3.8

1.1

4.0

1.3

4.2

1.5

4.4

1.7

4.6

1.9

4.6

1.9

4.7

2.1

5.0

2.3

5.1 5.3 5.5

2.5 2.6 2.8

5.6

3.0

5.7

3.1

5.8

3.2

5.9 6.1

3.3 3.5

6.4

3.8

7.7

5.1

10.2

7.6

294.5a 294.5b 303.6a 331.4a 331.4b 356.8a

11.5

8.7

11.9 13.5

9.1 10.2

15.1

11.5

Atmospheric O•..•

0.48 0.39 0.46 0.40 0.49 0.49 0.43 0.45 0.37 0.39 0.50 0.45 0.48 0.43 0.39 0.38 0.50 0.49 0.43 0.47 0.47 0.51 0.49 0.44 0.49 0.49 0.49 0.46 0.48 0.63 0.46 0.47 0.54 0.45 0.40 0.39 0.43 0.42 0.45 0.41

1.10 0.93 1.12 1.06 1.09 1.15 1.07 1.05 1.01 1.09 1.11 1.27 1.04 1.08 1.11 1.11 1.11 1.11 1.02 1.08 1.05 1.14 1.03 1.00 1.05 1.06 1.13 1.07 1.04 1.29 0.97 1.00 1.33 1.14 1.04 1.24 0.81 0.88 0.87 0.95

0.13 0.15 0.20 0.25 0.11 0.16 0.20 0.16 0.26 0.31 0.10 0.37 0.08 0.22 0.33 0.35 0.10 0.12 0.15 0.15 0.11 0.12 0.05 0.13 0.07 0.08 0.15 0.15 0.07 0.04 0.04 0.06 0.24 0.24 0.25 0.46 -0.05 0.04 -0.03 0.13

0.42 0.48 0.46 0.43 0.43 0.37

1.17 1.31 1.42 1.74 1.85 1.95

0.33 0.34 0.50 0.89 1.00 1.21

742

Sowers et al.: A 135-ka Vostok-SPECMAP

TABLE 1. (continued)

5180 of Atmospheric

Sample Depth,IceAge,a GasAge,b Trapped Gases ka ....... •._ka__•_ ......515...N.•..__qzo_ 356.8b

381.5a 381.5b

16.9

409.0a

19.0

409.0b 434.2a

12.8 14.1

21.1

15.3

21.9

15.8

25.2

19.1

29.4

24.0

33.4

28.2

434.2b 443.7a

443.7b 483.7a 483.7b 534.5a 534.5b 585.5a 585.5b

09•

0.39

2.11

1.33

0.38

2.15

1.39

0.35

2.20

1.49

0.38

2.05

1.29

0.38

2.10

0.42

2.06

1.35 1.23

0.41

2.11

1.30

0.54

1.94

0.86

0.39

1.92

1.15

0.48 0.41 0.51

2.06 1.93 1.49

1.10

0.46

0.47

1.54

0.60

0.41

1.69

0.87

0.39

1.72

0.95

1.11

656.5

39.0

33.7

0.39

1.40

0.62

694.5a

41.9

36.6

0.35 0.29

1.28 1.24

0.58

759.8a 759.8b

46.8

41.6

0.38 0.37

1.24 1.28

0.48

788.5a

49.0

44.0

0.48

1.30

0.35

0.50

1.38

0.39

0.38

1.23

0.48

0.38

1.32

0.56

0.32

1.22

0.58

0.35

1.32

0.63

0.39

1.23

0.45

0.32

1.26

0.61

694.5b

788.5b 806.2a

50.3

45.6

806.2b 834.2a

52.4

47.8

834.2b

857.6a

54.2

49.6

857.6b

885.6a

0.66 0.53

56.3

51.9

0.31 0.31

1.34 1.33

0.71

908.2

58.1

53.7

0.35

1.58

0.87

934.4a

60.3

55.6

0.33

1.62

0.97

0.33

1.76

1.09

0.40

1.70

0.89

885.6b

934.4b 937.3

60.6

55.8

957.4a

62.3

57.3

957.4b 982.3a

64.5

59.0

982.3b

0.71

0.33

1.57

0.90

0.34

1.61

0.93

0.41

1.61

0.79

0.30

1.62

1.02

1003.4a

66.3

60.7

0.38

1.67

0.92

1003.4b 1017.2a

67.5

61.9

0.37 0.38

1.73 1.71

1.00 0.96

0.33

1.73

1.08

69.2

63.8

0.33

1.37

0.72

0.39

1.67

0.89

71.2

65.9

0.35

1.59

0.88

0.32

1.67

1.02

0.36

1.39

0.68

1017.2b 1037.7a 1037.7b

1060.5a 1060.5b 1082.0a

72.9

67.9

__

Correlation

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

743

TABLE 1. (continued)

•5180of

SampleDepth, Ice Age? GasAge,b mbs

1082.0b 1108.4a 1108.4b 1134.6 1157.7a 1157.7b 1162.5a 1162.5b 1181.3a 1181.3b 1205.5a 1205.5b 1233.8a 1233.8b 1265.2a 1265.2b 1290.4 1307.7a 1307.7b 1309.5a 1309.5b 1333.3a 1333.3b 1363.7a 1363.7b 1391.2a 1391.2b 1419.7a 1419.7b 1443.0a 1443.0b 1467.7a 1467.7b 1494.3a 1494.3b 1505.0a 1505.0b 1514.3 1566.2a 1566.2b 1592.0a 1592.0b 1636.3a 1636.3b 1656.1a 1656.1b 1683.3a

ka

ka

75.1

70.4

77.2 79.0

72.7 74.7

79.4

75.2

80.8

76.9

82.7

79.0

84.9

81.3

87.3

83.3

89.2 90.6

85.0 86.2

90.7

86.3

92.6

88.4

95.0

90.9

97.2

93.2

99.5

95.8

101.5

97.9

103.6

100.1

106.1

102.3

107.1

103.2

108.0 113.0

104.0 108.3

115.3

110.5

119.0

115.1

120.6

117.2

122.6

119.6

Trapped Gases Atmospheric •]•N,.•_..... õ]80,_•........ O9.,_•_= 0.34 0.37 0.34 0.33 0.37 0.31 0.37 0.33 0.36 0.32 0.32 0.28 0.29 0.33 0.37 0.31 0.30 0.35 0.29 0.39 0.31 0.27 0.29 0.34 0.36 0.32 0.32 0.37 0.36 0.33 0.32 0.29 0.27 0.28 0.29 0.41 0.35 0.41 0.40 0.41 0.42 0.39 0.38 0.36 0.58 0.40 0.38

1.42 1.28 1.35 1.26 1.00 1.10 0.89 0.95 0.73 0.79 0.74 0.96 0.87 1.03 1.41 1.37 1.29 1.19 1.13 1.27 1.15 1.12 1.28 1.38 1.56 1.27 1.33 1.16 1.17 0.89 0.94 0.69 0.68 0.77 0.81 0.80 0.75 0.97 1.58 1.63 1.71 1.76 1.67 1.68 1.60 1.53 1.22

0.74 0.53 0.67 0.60 0.26 0.48 0.15 0.30 0.0! 0.15 0.11 0.41 0.29 0.37 0.67 0.75 0.70 0.49 0.56 0.50 0.52 0.58 0.70 0.70 0.84 0.63 0.69 0.41 0.45 0.23 0.30 0.11 0.14 0.21 0.22 -0.03 0.05 0.16 0.77 0.80 0.88 0.97 0.90 0.96 0.44 0.73 0.47

744

Sowers et al.: A 135-ka Vostok-SPECMAP

TABLE 1. (continued)

•j•80 of

Sample Depth,IceAge,a GasAge,b Trapped Gases Atmospheric mbs. ka ka 8•N, .,•...,_•j•8.O, q0o O•.,• . 1683.3b 1693.1a

123.3

120.4

1693.1b

1716.7a

125.0

122.3

1716.7b

0.38

1.26

0.50

0.50

1.14

0.13

0.50

1.14

0.14

0.46

0.86

-0.05

0.42

0.89

0.04

1732.7

126.1

123.5

0.43

0.88

0.01

1757.0a

127.8

125.2

0.43

0.78

-0.08

0.35

0.68

-0.03

0.40

0.61

-0.18

0.41

0.71

-0.11

0.41

0.76

-0.05

0.40

0.80

0.00

0.56

1.03

-0.08

0.51

1.06

0.05

1757.0b 1784.3a

129.7

127.3

1784.3b 1823.5a

132.5

130.3

133.2

131.0

1823.5b 1831.9a 1831.9b 1845.2a

134.1

132.1

1845.2b 1858.0a

135.1

133.0

1858.0b

0.43

1.19

0.34

0.40

1.20

0.41

0.50

1.42

0.43

0.44

1.39

0.51

1868.5

135.9

133.7

0.48

1.81

0.85

1882.3a

137.0

134.6

0.44

2.03

1.16

0.39

2.02

1.23

0.47

2.03

1.08

0.49

2.10

1.12

0.47

2.24

1.30

0.43

2.19

1.33

0.43 0.41

2.22 2.22

1.36 1.40

1882.3b 1883.4a

137.1

134.7

1883.4b 1907.6a

139.2

136.4

1907.6b 1907.6c 1934.5a

141.9

138.3

0.40

2.26

1.45

144.2

139.8

0.34

2.04

1.37

0.38

2.21

1.46

1955.8a

144.3

139.9

0.39

2.08

1.31

1955.8b 1965.5a 1965.5b

145.5

140.7

0.34 0.35 0.33

2.01 2.01 2.08

1.32 1.31 1.41

1982.3a 1982.3b

147.9

142.4

0.36 0.35

1.94 1.94

1.22 1.24

1990.8a 1990.8b

149.2

143.4

0.47 0.44

1.98 1.97

1.04 1.09

2016.4a

153.4

147.6

0.41

1.90

1.09

0.29

1.71

1.14

0.45

1.64

0.74

0.43

1.71

0.84

1934.5b 1954.6a 1954.6b

2016.4b 2058.8a

160.5

154.4

2058.8b

Columns4 and 5 are the isotopiccompositionof trappedN2 and 02, respectively,relativeto the present-dayatmosphere. The lastcolumnis

•5•8Oatm, thepaleoatmospheric •5•80of 02 aftercorrection forgravitational fractionation(equation(2)). Note that mbsindicatesmetersbelow the surface.

a Data are from Lorius et al. [ 1985].

bDataarefromBarnolaet al. [ 1991].

Correlation

Sowers etal.'A 135-ka Vostok-SPECMAP Correlation

0.0

x

• 0.5

745

•180atm

x

x

x

x

-420

x

1.o

x x x

x

-44O

4••Dice

1.5

-460

2.0

-480

I

500

I

1000 Depth (mbs}

1500

2000

Fig.2. Isotopic composition ofpaleoatmospheric 02 (•j18Oalm) based ontheisotopic composition oftrapped 02 intheVostok icecore(top).Results arereported relative tothepresent-day atmosphere. Theresults fromeachanalysis areplotted (withcrosses) alongwitha linejoining all theaverage values ateachdepth.Alsoplotted isthefiDice record fromJouzel etal. [1987] (bottom). Low1518Oath values aregenerally associated withhighfiDice values.

yearsB.P. respectively[Jouzelet at., 1987],which

implies thatthebubbles ofairwhichhaverecently

/5180of atmospheric 0 2 todayis+23.5%0 relative to standard mean ocean water. This 23.5%0 difference

formednearVostokarebetween2,600and3,000 yearsyoungerthanthesurrounding ice. Forthis

between atmospheric 0 2 andseawater isreferred to

study, Aageistakenastheageof thetim atmidpoint

theM-D effect).TheM-D effectresults fromoxygen

of thecloseoff depthinterval.

Our•518Oatm dataareplotted inFigure3 versus ageaccording to theVostokgaschronology of Bamotaet al. [ 1991]. We alsoplotthe/518Osw record, discussed previously, versus ageaccording totheSPECMAPchronology. Therangeof/5180 valuesfor the two recordsis similar,1.3%oand

1.6%o for/518Osw and•518Oatm, respectively. The general formsof thetworecords arestrikingly

similar.The divergence betweenthetworecordsfor ages> 110ka caneasilybeaccounted forgiven

uncertainties in the timescales.

Geochemical Controls on the 8180

ofAtmospheric 02: TheMorita-DoleEffect Morita [1935],Dole [1935],Dole et al. [1954], andKroopnickandCraig[ 1972]haveshownthatthe

asthe Morita-Doteeffect(hereinafterreferredto as

isotopefractionationassociated with the

photosynthetic production of 0 2 bothonlandandin

the ocean, and from fractionationassociatedwith the

consumption of 0 2 duringaerobicrespiration. As notedby Sowerset at. [ 1991], themare at least

fourfactors whichmayhavecaused/518Oat min the pastto be differentfrom today'svalue. First,

changes in/518Osw resultfromchanges in continental icevolumeanddirectlyimpactthe/5180 of photosynthetic 0 2 produced in theocean. Changes in /518Osw alsoaffectthe/5180of water evaporating fromtheocean,the/5180of precipitation onthecontinents, andultimately,the•5180of photosynthetic 0 2 produced onthecontinents. Second,changes in thehydrologic cycleon glacial/interglacial timescales influencethe/5180of

leafwaterandthereby the/5180 ofphotosynthetic 02

746

Sowerset at.: A 135-ka Vostok-SPECMAP Correlation -0.5

0

25

50

75

100

125

150

Lorius or SPECMAP Age (ka) Fig.3. Records of theõ18Oatm (dashed line)andõ18Osw (solidline)fromV19-30plottedonthe Barnolaet al. [ 1991] and SPECMAP [Martinsonet al., 1987] timescales,respectively.The

•j18Oatm recordisreported relativetothepresent-day atmosphere. Theõ18Osw recordhasbeen constructed fromthe•j18Oforam recordfromV19-30 asnotedin thetext.It hasbeennormalized by subtracting 3.46• fromall õ18Osw values,sothattheaverage õ18Osw duringtheHolocene is 0•.

Someof thediscrepancies betweenthe two recordsmay be due to chronological

inconsistencies between thetwotimescales. Between warmperiods, theõ18Oatm recordshows muchlargervariations thantheõ18Osw record,suggesting thatõ18Oatm maynotbeanidealproxy for õ18Osw, especially duringtheseperiods. producedon thecontinents.Third,ecological changesoverglacial/interglacial timescales causethe globalaverageof the respiratoryisotopeeffectto change. One canimaginesuchchangesresulting eitherfrom changesin the make-upof themarineand continentalbiospheres or from changes in theisotope effectson the specieslevel. Finally, changesin the ratio of continentalto marinegrossprimary productivity mayhavehada largeeffecton•518Oatm, because the •j180 of leaf wateris generallyheavier than that of seawater. For reference, the residence

timeof atmospheric 0 2 today,with respectto photosynthesis andrespiration,is estimatedto be between2,000 and 3,000 years[Benderet at., 1985].

We cangainsomeinsightintothefactorswhich

haveinfluenced •j18Oat mby examining Figure3 in more detail. If we assume,for the moment, that

someof theoffsetbetween the•jl8Os wand•j18Oat m

is the resultof chronological inconsistencies, thenthe highdegreeof covariation betweenthetwo records

impliesthatthemajorfactorinfluencing •jl8Oatm is changes in •jl8Osw.As we mentioned previously, therangeof •j18Oat mvaluesis similarto thefi18Osw valuesoverthe last 150 kyr. This resultsuggests thatthebiologicandhydrologicfactorswhich influence •jl8Oatm haveremained closeto their present-day values.For example,a 30% changein theratioof marineto terrestrial productivity would producea changeof about0 .5%0 •n ' •5180atmretat•ve '

to •j18Osw. Theonlytimewhena change of this magnitudeis suggested by thedatais duringmarine isotopestage5d (110 ka). We cannotruleoutthe possibilitythatthereweresubstantial changes in the biologicandhydrologiccycleswhichhadno net effecton•jl8Oatm ' butwe consider it unlikelygiven the similarnatureof the two curvesin Figure3. Fromthe similarityof thetwo curvesin Figure3

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

we tentativelyconcludethat(1) themajorfactor

influencing •jl8Oatm appears tobe•j18Osw, (2) there areperiodswhentheM-D effectwassubstantially differentthanit is today,and(3) thedegreeto which thefirst two conclusions maybe considered valid dependson how well we cancorrelatethe Vostok andSPECMAP chronologies. We will nowturnourattentionto thedevelopment of the gasageversusdepthrelationfor Vostok. As mentionedpreviously,we derivethisrelationby

correlating ourmeasured •j18Oat mrecordwiththe

.b18Osw record fromV19-30(Figure 3)using the

•nversecorrelationmethoddevelopedby Martinson et al. [ 1982]. We thentestthe fidelity of this cormnonframeworkby comparingtemperature records from the two media.

CONSTRUCTION OF A "COMMON TEMPORAL FRAMEWORK"

Establishinga GasAge-Depth Curvefor Vostok Basedon the goodagreementbetweenthe

•jl8Oatm_ ageand•jl8Os w- agerecords, weconclude thatthemajorfactorcausing changes in •jl8Oatm is variabilityin •S18Osw. We nowassume thattheM-D effect has been constant at the modem value of

+23.5%oduringtheperiodof ourstudyandderivea curveof gasageversusdepthfor the Vostokice core

by correlating theVostokcurveof •jl8Oatm versus depthintotheV19-30curveof 1518Osw versus age [Martinson et al., 1982; Martinson et al., 1987].

Beforeproceeding,we notethattherewere times whenthe M-D effectclearlydeviatedfrom the modem value.

Thus the correlation that we derive is

basedon an assumption whichis not completely correctandwill undoubtedlyneedto be further refined.

The first stepin derivingour age-depthcurvefor Vostokwas to choosethe deepestand shallowest depthswe wishedto correlateand assignagesto thesetwo depths. As an upperbound,we choose our shallowestsamplefrom 114.8mbs. The ice age at 114.8 mbsis 3.6 ka [Loriuset al., 1985]; the age of the trappedair parcelsat 114.8mbs,calculated usingthe Aagevaluesof Bamolaet al. [1991] (2.8 kyr), is then0.8 ka (3.6 ka- 2.8 kyr). Becausethe turnovertime of atmospheric 0 2 is about2 kyr, the

•jl8Oatm of trappedgasdatedto be0.8 ka reflectsthe •5•80 of seawaterat 2.8 ka [Sowerset al., 1991]. We choosethe depthof 1934.5mbsasthe bottom

of ourcorrelation interval.The•jl8Oatm reaches a maximumat thisdepthwhichcorresponds to the

•518Osw maximum, datedat 135kain theSPECMAP chronology.This periodcorresponds to the glacial maximumprecedingthepenultimatedeglaciation. After tying the end-pointsof the two records,we

applieda 1.0-cycle/kyr filterto the•jl8Oatm record

747

andcorrelatedthe two recordsusingthe protocol developedby Martinsonet al. [ 1982]. The correlationinvolvesthe construction of a mapping function(in our casea correlatedgasage-depth curve)for the Vostokice core. The mapping function consistsof a sum of sinusoids(of increasingorder)whichhaveamplitudes chosento maximize the amount of shared variance between the two •5180 records. The number of sinusoids added

to the mappingfunctiondetermines theresolutionof the correlation.In our case,we stoppedthe correlationat a resolutionof- 17 kyr (8 sinusoids). We chose to correlate the two records at this modest

level of resolutionbecausethe characteristic response time for an ice sheetwhichhasbeensubjectedto changesin accumulation rateis approximately17 kyr [Imbrie et al., 1984]. The correlationcoefficient between the correlatedrecordswas 0.88 (77% sharedvariancebetween the correlatedrecords).

By correlating the•jl8Oatm versus depthcurve fromVostokintothe•jl8Oswversusagerecord,we infer a gasage-depthrelationship for theVostokice core. The impliedageof the trappedgassample, whichwe defineasthe correlationage,is equalto the gasageplusthe atmospheric 0 2 turnovertime,2 kyr. Globalproductivityandhencethe atmospheric 0 2 turnovertimeundoubtedly variedduringglacialinterglacialcycles. Estimatesvary widelybut are generallywithin 25% of present-day values[Lyle, 1988;Meyer, 1988;Mix, 1989]. Changesof this magnitudewould introducean errorof + 0.5 kyr to

ourgasages.In Figure4, we plot•j18Oat mversus theVostokcorrelation ageand•518Osw versus the SPECMAP age. This figureshowsthatonecan derivea correlationage-depthcurvefor Vostokin

which•jl8Oatm and•jl8Os w trackeachothervery well.

Figure4 alsoshowsthat•j18Oat mdoesnotfollow •jl8Os wexactlyandthattheM-D effectmusthave variedduringthe last 135 kyr. A detaileddiscussion of the variableM-D effectwill be presented in a forthcomingpaper. The rangeof variabilityin the

M-D effect(0.9%o)is about75% aslargeasthe

rangeof 1518Osw (1.3%o), andit mayclearlyimpart significanterrorsto our Vostokage-depthcurve, whichis derivedsimplyby assumingthattheM-D effect has been constant.

It is for this reason that we

consider•5180to be an equivocalcorrelation tool [Sowers et al., 1991].

Establishingan Ice Age-Depth Curvefor Vostok In the previoussectionwe deriveda gasage-depth

relationby correlating the•jl8Oatm recordfrom VostokintotheSPECMAP•jl8Osw record.In order to comparethe isotopicandchemicalcomposition of thepaleoprecipitation at Vostokwith themarine climaterecords,we needto calculatean ice age-depth curvefor this core. Put anotherway, we needto

748

Sowerset al.' A 135-kaVostok-SPECMAPCorrelation

Lorius Age (ka) -o.5

o --+

25

50

75

100

125

•180sw

0.5

•180atm 1.5

I 0

25

50

75

100

125

150

SPECMAP Age (ka) Fig.4. Records of the•j18Oatm (thinline)and•j18Osw (thickline)plotted ontheSPECMAP

[Martinson et al., 1987]timescale.Theõ18Oatm recx)rd hasbeencorrelated intotheSPECMAP

chronology using theinverse correlation method ofMartinson etal.[1982].Wehave purposely nottriedtocorrelate thedetailed features which areshorter thanabout 5 kyr.Webelieve thelarger õ18Oatm variations aretheresult ofchanges in180fracfionation which aremost likelyassociated withtheterrestrial biosphere. TheLoftus etal.[1985]timescale isplotted ontheupper axisfor

comparison. Theõ18Osw record isidentical tothatplotted inFigure 3.

account forthefactthattheairparcels weretrapped

at least 100 m below the surfaceof the ice sheetand

a detailedexplanation of thesefactorsandthe associated errors.Theresulting iceagesfor eachof

arethusyoungerthanthesurrounding ice. We refer to thisagedifference astheiceage-gas agedifference

our sampledepthsare listedin Table 2.

(Aage). We haveutilizeda densification model [HerronandLangway,1980],therecordof the

IMPLICATIONS FOR CLIMATE LEADS AND LAGS

paleoaccumulation ratefromourcorrelated gasagedepthcurve,andanestimate of thethinning function [Ritz,1992]to calculate a recordof thepaleo-closeoff depthandAageall alongthecore. Resultsof thesecalculations suggest thatpaleo-close-off depths varied between 100 and 125 m below the surface

withcorresponding Aagevaluesrangingfrom2.9to 7.0 kyr,respectively [Sowers et al., 1992]. Deeper close-offdepthsandlargervaluesof Aageoccur duringglacialperiods whenthetemperature and accumulation ratewerelow. The appendix contains

Sowerset al. [ 1991]compared thedepths at

whichthe•180 of 0 2 andpCO2 began tochange at thebeginning ofthepenultimate termination. They foundthat/3180 of 0 2began tofallapproximately 6 kyrbeforepCO2 beganto rise. Theyconcluded that thepCO2 riseledthe•jl8Os w decrease by 4 + 1.7

kyr(theresponse timeofthe/3• 80 ofatmospheric 02 toa change in/5180of seawater is about2 kyr). The age-depth curveof Pichonet al. [1992]supports this conclusion, asit putsthestartof thepCO2 rise(and

Sowers et al.: A 135-ka Vostok-SPECMAP Correlation

749

TABLE 2. VostokIce CoreCorrelationAges SampleDepth, mbs 114.8 116.3 120.3 125.2 130.1 135.3 139.7 139.8 143.7 149.6 152.8 156.9 161.8 165.1 166.8 169.7 174.6 177.1 184.3

213.8 269.5 294.5 303.6 331.4 356.8 381.5 409.0 434.2 443.7 483.7 534.5 585.5 656.5 694.5 759.8 788.5 806.2 834.2 857.6 885.6 908.2 934.4 937.3 957.4 982.3 1003.4 1017.2 1037.7 1060.5 1082.0 1108.4 1134.6

Gas Age,

Ice Age, ka

ka 0.8 0.9 1.1 1.3 1.5 1.7 1.9 1.9 2.0 2.3 2.4 2.6 2.8 2.9 3.0 3.1 3.3 3.4 3.7 5.0 8.0 9.5 10.1 11.9 13.7 15.6 17.9 20.2 21.0 24.9 29.8

Correlated 3.9 4.0 4.1 4.3 4.5 4.8 4.9 4.9 5.1 5.3 5.5 5.6 5.8 6.0 6.0 6.1 6.4 6.5 6.8 8.1 11.7 14.0 14.8 17.4 20.4 23.9 25.8 27.8 28.7 32.0 36.3

34.5 40.4 43.3 47.9 49.9 51.1 52.9 54.5 56.4 57.9 59.6 59.8 61.2 62.9 64.4 65.3 66.8 68.4 69.9 71.8 73.6

40.2 45.7 48.3 52.5 54.3 55.8 57.3 58.5 60.9 63.2 65.4 65.6 66.8 67.9 68.5 69.4 71.2 72.7 74.3 76.1 77.4

Lorius a 3.6 3.6 3.8 4.0 4.2 4.4 4.6 4.6 4.7 5.0 5.1 5.3 5.5 5.6 5.7 5.8 5.9 6.1 6.4 7.7 10.2 11.5 11.9 13.5 15.1 16.9 19.0 21.1 21.9 25.2 29.4

33.4 39.0 41.9 46.8 49.0 50.3 52.4 54.2 56.3 58.1 60.3 60.6 62.3 64.5 66.3 67.5 69.2 71.2 72.9 75.1 77.2

CorrelatedIce Age Lorius Ice Age, ka 0.3 0.4 0.3 0.3 0.3 0.4 0.3 0.3 0.4 0.3 0.4 0.3 0.3 0.4 0.3 0.3 0.5 0.4 0.4 0.4 1.5 2.5 2.9 3.9 5.3 7.0 6.8 6.7 6.8 6.8 6.9 6.8 6.7 6.4 5.7 5.3 5.5 4.9 4.3 4.6 5.1 5.1 5.0 4.5 3.4 2.2 1.9 2.0 1.5 1.4 1.0 0.2

750

Sowers et al.' A 135-ka Vostok-SPECMAP Correlation

TABLE 2. (continued)

Sample Depth, mbs 1157.7 1162.5 1181.3 1205.5 1233.8 1265.2 1290.4 1307.7 1309.5 1333.3 1363.7 1391.2 1419.7 1443.0 1467.7 1494.3 1505.0 1514.3 1566.2 1592.0 1636.3 1656.1 1683.3 1693.1 1716.7 1732.7 1757.0 1784.3 1823.5 1831.9 1845.2 1858.0 1868.5 1882.3 1883.4 1907.6 1934.5

GasAge, IceAge.ka Correlated IceAgeka Correlated _.Lorius_ a LoriusIceAge,ka 75.3 75.7 77.0 78.8 80.9 83.2 85.1 86.4 86.5 88.3 90.5 92.5 94.6 96.3 98.1 100.0 100.8 101.5 105.1 106.9 110.0 111.3 113.2 113.9 115.6 116.8 118.6 120.7 123.7 124.4 125.5 126.6 127.4 128.6 128.7 130.7 133.0

78.7 79.1 80.8 82.7 85.3 87.9 89.5 90.1 90.2 92.4 94.7 96.5 98.3 100.0 102.2 104.3 105.3 106.3 109.3 110.4 112.8 114.1 116.1 116.8 118.5 119.7 121.5 123.6 126.4 127.0 128.2 129.5 130.4 131.9 132.0 134.7 136.8

79.0 79.4 80.8 82.7 84.9 87.3 89.2 90.6 90.7 92.6 95.0 97.2 99.5 101.5 103.6 106.1 107.1 108.0 113.0 115.3 119.0 120.6 122.6 123.3 125.0 126.1 127.8 129.7 132.5 133.2 134.1 135.1 135.9 137 137.1 139.2 141.9

-0.3 -0.3 0.0 0.0 0.4 0.6 0.3 -0.5 -0.5 -0.2 -0.3 -0.7 -1.2 -1.5 -1.4 - 1.8 - 1.8 - 1.7 -3.7 -4.9 -6.2 -6.5 -6.5 -6.5 -6.5 -6.4 -6.3 -6.1 -6.1 -6.2

- 5.9 -5.6 -5.5 -5.1 -5.1 -4.5 -5.1

Gasagesarethosederived fromtheinverse correlation of the•j18Oat mrecordwiththe

fi18Osw record fromV19-30core.Thethirdcolumn liststheiceages fromthecorrelation (seetextfordetails). Thelastcolumn liststhedifference between thecorrelated iceages

andthosefromtheLoriuset al. [ 1985].Notethatmbsindicatesmetersbelowthe surface.

a Data arefrom Loriuset al. [ 1985].

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

warmingat Vostok) at about139 ka in the SPECMAPchronology, while the15180of seawater beginsto decreaseat about135 ka in this chronology.Both the radiotarianchronology [Shackletonet at., 1992] andthe diatomchronology [Pichonet al., 1992] indicatea lag at the endof the termination: CO2 reacheditsmaximumvalueat about 129 ka in thesechronotogies (andabout126 ka in

ourchronology), whilethe1518Osw minimumwas about123-125ka accordingto the SPECMAP chronology.The work summarizedherethus indicates thatCO2 changes ledtheice volume decreasethroughoutthe penultimatetermination. COMPARISON OF AGE-DEPTH PROPOSED FOR THE VOSTOK

751

proposedby Loriuset at. [ 1985]. The correlatedice age-depthcurvedeviatesfrom theLoriuset al. [1985] curveby morethan2 kyr between350 and 1000 mbsandagainbelow 1500 mbs(Figure5). Our correlation of Vostok into the SPECMAP

chronologyrequiresincreasingthe Loriuset al. [ 1985] agefor the beginningof terminationI by 5 kyr anddecreasingthe agefor the startof termination II by 5 kyr. Our chronologyis in excellent agreementwith the Loriuset al. [ 1985] chronology for the intervalbetween70 and 110 ka (roughly1050 to 1580 mbs).

Our age-depthcurveis comparedwith similar curvesrecentlyproposedby otherworkers(Figure 5). Petit et at. [ 1990] pickedcontrolpointsby aligningthe "magneticflux" recordfrom themarine

CURVES ICE CORE

sediment core RC 11-120 with dust accumulation

Our age-depthcurve,like othersrecentlyproposed for the Vostok ice core [Petit et al., 1990; Pichon et

at., 1992; Shackletonet at., 1992], generallyagrees very well with the "flow model"chronology

nearVostok(we refer to thiscorrelationtechniqueas the dustcorrelation).Pichonet al. [ 1992] picked controlpointsby aligningthe temperature- depth recordfor Vostok with the recordof SST versusage

160

Lorius et al. [1985]

140

-

120

-,

100

-

This Work

Lorlus (1985)

80 -

Pichon (CaI.Age) Shackleton (1992)

60

-

40

-

zO

-

0

•

This paper "Best

Petit (1990)

estimate"

500

1000

1500

2000

Depth (mbs) Fig. 5. Variousice ageversusdepthprofilesfor the Vostokice core. The curvelabeled"Best

estimate" is based onour•5180correlation. Alsoplottedaretheestimate byLoriuset al. [1985] andthe "controlpoints"of Pichonet al. [ 1992](diatom),Shackletonet al. [ 1992] (mdiolaria),and Petitet al. [1990] (dust). In general,all recentcorrelations agreewith theoriginalmodelof Lorius et al. [ 1985] within stateduncertainties.

752

Sowerset al.' A 135-ka Vostok-SPECMAP Correlation

for southern ocean core MD 84-551.

SST for this

corewasinferredfrom diatomtaxonomyusinga transferfunction(the diatomcorrelation). Shackleton et al. [ 1992]pickedcontrolpointsby aligningthetemperature-depth recordfor Vostok with the SST recordof core RC11-120, determined

fromradiolariantaxonomy(theradiolarian correlation)[Hayset al., 1976a,b]. For thislatter recordwe haveplottedonlycontrolpointsassociated with maximumSST values;SST minimamightnot be faithfullyreflectedgiventhelimitedtemperature rangeof radiolarianspecies. Betweenabout290- and360-m depth,theinterval coveringthelastdeglaciation, ourproposed agedepthcurvedivergesstronglyfromthatof Loriuset

al. [ 1985]. We assignthe357-mdepthanageof 20 kyr, compared to theflow modelageof 15 kyr. As notedabove,thePetitet al. [ 1990]chronology (dust) is in fair agreement withtheflow chronology over thisinterval.The diatomtimescale in thisdepth interval is based on 14C dates rather than the short

SPECMAPchronology (whichis independent of the

14Cchronology). To allowa moredirect comparison,we note that Martinsonet al. [1987]

assigned an ageof 17.9kyr for thelastglacial maximum(event2.2 in Table2). Thisageis in good agr•ment with a siderealageof about18 kyr for the

•j18Ofora mmaximum, basedon 14Cdatingof many foraminiferal•180 curves[Samtheinet al., 1992]

afterconverting •4Cagesto absolute agesbasedon the work of Bard et al. [ 1990a]. When we convert

theMD84-551timescale from•4Cagesto absolute ages,the startof the last terminationmovesfrom 15

ka (•4C) to 17ka (calendar).If we wereto recorrelate the MD84-551

SST and Vostok

temperaturerecordson the calendartimescale,the

latitudesimulatedprecipitation calculated by Prelland Kutzbach [1987] to coincide with the maximum in

northernhemispheresummerinsolationat 31 ka.

Thereis a strongprecession signalin theCH4 concentration variations[Chappellazet al., 1990]. Chappellazet al. [1990]explainedtherelationship betweenCH4 maximaandmaximain simulated precipitation by suggesting thathigherprecipitation leadsto higherratesof CH4 production.All other CH4 concentration maximain theVostokrecordare closelyalignedwith northernhemisphere summer insolationmaximagivenourcorrelatedages[M. Bender et al., The Dole effect and its variations

duringthe last 130,000yearsasmeasuredin the Vostokice core,submittedto GlobalBiogeochemical

Cycles,1993). The factthattheCH4 maximumat 625-mdepthprecedes theinsolationmaximumby 6 kyr is anothercausefor concernabouterrorsin our correlatedchronologybetween25 and55 ka. Our correlatedagesarelessreliablefor thissectionthan for anyotherpartto the Vostokcore. In the depthintervalfrom 800 m to 1200 m (55-80

ka), agesinferredfrom•180 of 0 2, diatoms, radiolaria,anddustscatterconsiderably but areall on averageabout5 kyr olderthanagesat the samedepth estimatedfrom the flow model(Figure6). In this depthinterval,noneof the correlationmethods wouldbe expectedto work well, becausevariations

in paleotemperature, •j180of 0 2 andseawater, and dustflux areall smallandthereforedo notprovidea strongbasisfor intercomparingrecords.In general, it canbe saidthatthevariouscorrelationapproaches give aboutthe sameaccumulationratesbetween800

and1200m astheflow modelbutgiveolder absoluteages. Between1200 and 1600m, agesfrom the flow

Vostoktemperature-age recordwouldbe in better agreementwith the temperature-age recordfrom the Byrd ice core,wherethebeginningof thetermination is datedbetween16 and 18 ka [Beeret al., 1992]. Betweenabout400- and800-mdepth,

dateddepthin thisintervalis at about1200m, where theradiolarianrecordshowsa distincttemperature

corresponding to correlated SPECMAPagesof 2555 ka, ourproposed correlation agesarehighly

corresponding to stage5a. At thisdepth,datesfrom

uncertain. Comparing •j18Oat mand•18Oswvalues overthisintervalis problematicfor two reasons.

First,•18Osw variations aresmall(totalrangeof 0.35%o), sothatthereisnota strong signal.Second,

modeland15180 stratigraphy arein goodagr•ment, and agesfrom dust,diatoms,and radiolariaare

higherby about4 kyr. Perhaps themostaccurately

maximaand•180 of 0 2 shows clearminima theflow model,from•180 of 0 2 , andfrom radiolariaagreeto within3 kyr. Diatomagesare

higherthan•80 agesby about3 kyr,andthismay reflectthefactthatthediatomtemperature record showsmorevariabilitythan,andis noteasily

sampleresolutionis poorbecauseof substandard ice qualityin thisinterval;between500- and650-m

correlated with, the ice core record in this time

depthwe haveanalyzedonly2 samples.Thispoor

interval.

basis for correlation translates into two anomalous

By 1700m, corresponding approximately to the beginningof stage5d, thevariousage-depth curves showincreased divergence.Diatomagesagr• well with theflow chronology, while•180 agesareabout 7 kyr younger.We believethatthe•j180agesare

results in thecorrelated chronology. Thefi18Oat m minimumat 533-mdepth,derivedfroma single sample, isnotcorrelated intothe•jl8Os wminimumat 32 ka butprecedes it by about5 kyr (Figure4). Additionally,theCH4 concentration maximumat 625-m depthin theVostokcore,datedat 32 ka in the Loriuset al. [ 1985]chronology, is redatedto 37 ka by ourcorrelation withtheSPECMAPchronology. This ageis 6 kyr olderthanthemaximumin low-

lessreliable,because the•518Oat misclearlynot followingchanges in •jl8Osw suchthatmuchof the variabilityin •j18Oat m mustbeduetochanges in the M-D effect.The radiolarianchronology of Shackleton et al. [1992]bringsthetimeof the

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

753

10

g

,,,,,

[,,"", O

This work

.... I---

•• '

--

,,

,

,n

', •

,

,

-'-•,

I

Pichon(1992)

o

Shackleton(1992)

A

Petit (1990)

,,•!

Pichon (1992) This

work

-10

0

500

1000

1500

2000

Depth (rnbs) Fig. 6. DifferencesbetweentheoriginalLoriuset al. [ 1985]ageestimates andthosefromrecent correlationstudies.The thickline illustrates theresultsfrom theõ180 (ice)correlation.The dashedlineillustrates thepaleoaccumulation rateprexticted fromthePichonet al. [ 1992](diatom) correlation.Alsoplottedarethe"controlpoints"fromShackleton et al. [ 1992](circles)andPetitet al. [ 1990](triangle_s). Terminations I andII aremarkedfor reference.Duringtermination I, both thediatomandõ180correlations indicatethattheiceagesareolderthancorresponding agesfrom theLoriuset al. [ 1985]timescale.Fortermination II, all correlated agesareall about7 kyr youngerthantheLoriuset al. [ 1985]timescale.Duringthelastglacialperiod,thecorrelated ages varyby lessthan+6 kyr fromtheoriginalLoriuset al. [1985]chronology.

decrease in pCO2 observed at Vostokintolinewith

thet•me ß of theA •513C (plank tom'c-ben thic)increase observed in both their stacked record as well as their V 19-30 record. This last result must be viewed with

caution,asthe Aageestimates usedby Shackletonet al. [ 1992] were not consistentwith their modified ice

age-depthcurve. At 1850-m depth(stage5e), radiolarians,diatoms,

and/5180all giveconcordant ageswhichareyounger thanthat of the flow modelchronology.

Radiolariansanddiatomsgive agesof about128 ka for the temperature minimumat stage5e, in good

agreement with the/5180ageof about126ka. The ageestimatedfrom the flow modelis about 133 ka. The dustageat a slightlydeepercontrolpointis even youngerthanthe/5180age. Pichonet al. [1992] and Shackletonet al. [ 1992] have shownthat Vostok

temperature andpCO2 valuescanbe broughtinto phasewith southernoceanSST andforamA•13C valuesby invokingan agefor thetemperature

754

Sowers et al.: A 135-ka V6stok-SPECMAP

minimumwhichis about5-10 kyr youngerthanthat of the flow chronology.Our work supports their conclusions.

Correlation

from the •5180 correlation to transfer the SST record

from the depthdomainto thetime domain. Glacial/interglacial seasurfacetemperature changes inferred from the RC 11-120 and MD88-770 are about 6o-8oc.

COMPARISON OF OTHER DEEP-SEA VOSTOK CLIMATE RECORDS

AND

SouthernOceanSeaSurfaceTemperature Records Correlated

to the Vostok

records

Sea surfacetemperatureestimatesfor RC 11-120 are systematicallyhigher,by about4øC,than temperaturesestimatedfor MD88-770. One reason is thatdespitetheir similarlatitude,MD88-770 is actuallylocatedabout3ø closerto the PolarFront than RC 11-120.

Another reason is that the

InversionLayer Temperature

radiolariantransferfunctionreflectstemperatures below 7øC very poorly,while the diatomtransfer

We now comparetemperature recordsfrom Antarctica andthesouthern oceantoinvestigate the validityof ourunderlyingpremisewhichis that

function loses discrimination

fi18Oat mmaybeusedasa proxyfor•518Osw. Becausethetemperature abovetheAntarctic

inversion layeris controlled by thetransport of latent and sensibleheat from the southernocean,the record

of Antarcticinversionlayertemperature is believedto have been similar to the SST record for the southern

oceanover the sameperiod [Pichonet al., 1992].

The•SDof theprecipitation overAntarctica hasbeen shownto becontrolled by thetemperature atthetop of theinversionlayer[JouzelandMerlivat, 1984]. The recordof •SDvariations alongtheVostokicecore

above about 10øC. We

suspectthatthe RC 11-120recordmay be overestimating glacialtemperatures andadoptthe SST recordfrom MD88-770 for comparisonwith Vostok.

Both RC11-120

and MD88-770

are

presentlylocatedin thesubantarctic watermass which is boundedto the southby the AntarcticPolar Frontandto the northby the Subtropical Convergence.Summerwatertemperatures between

10øCand14øCalongwithsalinities of 34.3-34.9%0 normallycharacterize thenorthernboundaryof subantarctic waterwhichis presentlylocatedat about 40øS.

The Antarctic

Polar Front has water

temperatures of between4 and 10øCwith salinities

have been used to construct a record of the inversion

between 34.0 to 34.6%0. The location of the

layertemperatureovertheVostokarea[Jouzelet al.,

AntarcticPolarFrontis presentlylocatedjust south of 50øSlatitude [LozanoandHays, 1976]. Interpretingseasurfacetemperature recordsfrom thisregionof the southernoceanis problematic, becausesomevariabilitymaybe relatedto localized changesin the hydrographic setting,ratherthan large-scalechangesin SST. In particular,migration

1987].

We assume thatthereis a strongrelationship betweensouthern oceanseasurface temperature and thetemperature abovetheinversion layer.We then testtheage-depth curvewe derivedfor Vostokby comparing theVostokcurveof inversion layer temperatureversusageagainstthe southernocean seasurfacetemperature-age curve(inferredfromthe temperaturedependence of planktonicmicrofossils). The degreeof concordance is a measureof the

fidelityof ourage-depth curveandthevalidityof the assumption thattemperature is faithfullyrecorded by thedifferentproxydata(diatomandradiolarianSST

and•SDice). Seasurfacetemperature recordsfromRC 11-120 (44ø3I'S, 79o52'E)[Hays et al., 1976a,b; Lozano andHays, 1976] andMD88-770 (46os, 96oE) (L. Labeyrie et al., Changesin southern ocean hydrologyoverthelast230 kyr asrecordedin core MD88-770, manuscript in preparation,1993)are plottedin Figure7. RelevantdatafromMD88-770 arealsotabulatedin Table3. The temperature recordsareestimated fromdatarelatingthe

of the Polar Front acrossour site, rather than

regionalchangesin SST, maybe responsible for muchof the variabilitywe observe.Two studiesin particularhavelookedat the movementof the Polar Frontin this regionof the southernocean;Hayset al. [ 1976b] studiedthe distributionof radiolaria,while Howard and Prell [ 1992] looked at the foraminiferal

assemblages to determinethepositionof thePolar Frontduringthelastglacialmaximumperiodrelative to thepresentdaylocation.Bothstudiesconcluded that the Antarctic Polar Front was located about 40-7 ø

northof its present-day positionduringthe last glacialperiod. GiventhatMD88-770 is located about6ø northof thepresentdayAntarcticPolar Front, it is conceivable that some of the SST

variabilityobservedin Figure7 is relatedto the passageof the frontoverthe site. Nevertheless, we

distribution of microfossils recovered from sediment

believe that the influence of the Polar Front has not

coretopsto the overlyingseasurfacetemperature.

compromised this SST recordsubstantially for three reasons:(1) The Subantarctic assemblage has dominatedthe diatompopulation throughout the 150ka record,asexhibitedby its high-factorloadings

The diatomrecordis believedto represent summer SST to about_+loC[Pichonet al., 1992]. For both

records, weestimated agesbycorrelating fi18Ofora mdepthcurves intotheSPECMAPstacked fil8Oforam_ age curve [Pisiaset al., 1984; Martinsonet al.,

1987]. We thenusetheageversusdepthrelation

(>62%), (2) the amountof ice-rafteddebrisdoesnot

increaseduringglacialperiods,and(3) the abundance of the fourdiatomspecieswhichare

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

755

14

RCl1-120

I

;

11 II

.... 0

SST

It

/, I

! 0

Radiolarian

II

I• I

25

....

I

'

'

'

50

'

t

tI

\

MD88-770 Diatom SST I

'

'

'

'

75

I 100

....

I

'

'

125

150

SPECMAP AGE (ka)

Fig.7. Seasurface temperature (SST) records from RC11-120 (44øS, 80øE) (solid line)and

MD88-770 (46os,96o'E)(dashed line) bothplottedon theSPECMAPtimescale.The RC11-120 recordis fromHayset al. [ 1976a]basedon a radiolarian transferfunction.The SSTrecordfrom MD88-770 is basedon a diatomtransferfunction[Pichonet al., 1992]. The two recordsshow

similarglacialto interglacial temperature shiftsof 6ø-8øC.The SSTrecordfromRC11-120is systematically higherthanthatoYMD88-770. Oneexplanation for thissystematic differenceis that theradiolariantransferfunctionis notreliableat temperatures below7øC. associated with seaice (Nitzschiacurta,Actinocyclus actinochilus,Nitzchia sublineata, and Nitzchia

cylindrus) remainslow. Basedon this evidencewe believe that the SST record from MD88-770

can be

usedas a masonableapproximation of theSST record of the subantarctic waters north of the Polar Front.

Ice Core Recordsof AtmosphericTemperatures Over the Antarctic continent, there is a 300- to 700

m layer of air in whichthe temperature increases with increasing altitude.Abovethistemperature inversion,temperatures decrease according to the adiabaticlapserate. The strengthof theinversion layer,as measuredby the differencebetweenthe surfacetemperatureandthetemperature just above the inversionlayer, is about5oc nearthecoast [PhillpotandZillman, 1970]. Maximum temperaturedifferencesreach20oc nearVostok. As notedabove,the accumulation rateandtheisotopic composition of theprecipitation arerelatedto the

temperature justabovetheinversion layerwherethe precipitation is formed[Robin,1977;Jouzeland Medivat, 1984]. We adopttheinversionlayer temperature recordfromJouzelet al. [1987]whichis basedonthemeasured •SDprofilefromtheVostok corefor comparison withtheSSTfromMD88-770. In Figure8 we haveplottedtheVostoktemperature recordusingourcorrelated iceage-depth relation. The SSTrecordis plottedusingtheSPECMAP timescaleasin Figure7. Thesetwo recordsshow glacial/interglacial temperature changes of 6ø-8øC. Plotted in this fashion, the two recordshave a

correlationcoefficientof 0.72, implyingthat52% of the varianceis sharedby thetwo records.When thesetwo datasetsarecompared usingtheLoriuset al. [ 1985]chronology, thecorrelation coefficient is 0.54 (i.e., only-30% sharedvariance).We interpret thehigherdegreeof correlation between thetwo recordsusingourcommontemporalframework as evidence that the Vostok record is better correlated

withthedeep-sea recordusingthechronology derivedherefrom•518Oat mdata.We makenoclaim for havingimprovedtheabsolute chronology.

756

Sowers et al.: A 135-ka Vostok-SPECMAP

TABLE 3. Data From Deep-SeaSedimentCore MD88-770 SampleDepth,

cm 0 10 20 30 40 50 60 70 80 90 100 110 120 130 136 140 150 154 156 158 160 162 164 166 170 174 176 178 180 184 186 190 194 196 200 204 210 214 220 224 226 230 234 236 240 244 246 250 254 260 264 270

Age,

Benthic

SeaSurface

ka

•18.O,%oPDB

Temperature, oC

0.0 6.0 7.7 9.4 11.1 13.0 15.0 17.0 17.3 17.7 18.0 18.3 18.6 18.9 19.1 19.3 19.6 19.7 19.8 19.9 19.9 20.0 20.0 20.1 20.2 20.4 20.4 20.5 20.6 20.7 20.8 20.9 21.0 21.1 21.2 21.3 21.5 21.7 21.9 22.0 22.1 22.2 22.3 22.4 22.5 22.6 22.7 22.8 23.0 23.2 23.3 23.5

3.55 3.50 3.55 3.83 3.87 4.49 4.72 5.16 5.01 5.10 4.92 5.02 5.14 5.03 5.19 4.90 5.10 5.13 5.09 5.18 5.12 5.10 5.16 5.02 5.15 5.06 5.04 5.05 5.04 5.06 4.98 -

9.2 10.1 9.9 8.8 9.0 9.9 6.9 5.1 3.2 2.9 3.0 2.0 3.6 1.2 3.1 2.7 2.2 2.4 3.3 3.2 3.1 3.0 4.1 2.7 3.6 3.2 2.9 2.3 1.9 3.3 1.8 2.7 2.6 2.7 3.2 2.9 2.8 3.5 2.0 2.1 2.4 2.4 2.8 1.0 1.4 2.2

Correlation

Sowerset al.: A 135-ka Vostok-SPECMAP Correlation

757

TABLE 3. (continued)

SampleDepth, cm

Age, ka

Benthic õ•80, %oPDB

274 280 284 286 290 294 298 300 302 306 310 316 320 330 340 350 360 366 370 376 380 386 390 396 400 406 410 420 426 430 434 440 446 450 456 460 466 470 476 480 486 490

23.6 23.8 23.9 24.0 24.1 24.3 24.4 24.5 24.5 24.6 24.8 25.0 25.1 25.4 26.4 27.4 28.3 28.9 29.3 29.9 30.3 30.9 31.3 31.8 32.2 32.8 33.2 34.2 34.8 35.1 35.5 36.1 36.7 37.1 37.7 38.1 38.6 39.0 39.6 40.0 40.6 41.0

5.08 5.05 4.99 4.92 4.94 5.04 4.98 4.96 4.80 4.86 4.95 4.82 4.65 4.39 4.69 4.76 4.83 4.83 4.70 4.80 4.78 4.77 4.78 4.74 4.70 4.72 4.72 4.82 4.86 -

496

41.6

4.76

500

41.9

-

510 516 520 526 530 536 540

42.9 43.5 43.9 45.2 46.0 47.3 48.1

4.76 4.71 4.68 4.70 4.70 -

Sea Surface

Temperature,ø.C 2.4 2.8 2.6

2.3 1.6 2.8 2.8 3.1 3.2 2.4 3.3 2.4 2.6 1.8

2.3 2.1

1.6 3.9 2.3 2.9 3.4 3.5 4.5

3.8 3.2 2.1 -

3.8 3.2

3.8 2.7 3.2

758

Sowers et al.: A 135-ka Vostok-SPECMAP Correlation

TABLE 3. (continued)

SampleDepth, cm

550 556 560 570 580 590 600 610 616 620 626 630 640 644 646 650 652 656 660 670 680 690 696 700 704 710 714 716 720 726 730 736 740 750 754 760 762 764 766

768 770 774 780 784 790 794 798 800 802 808 810 814

Age,

Benthic

Sea Surface

ka

õ18Q,%9.PDB

Temr>emmre, øC

50.2 50.6 50.9 51.5 52.2 52.8 53.5 54.1 54.5 54.8 55.2 55.5 56.0 56.3 56.4 56.6 56.7 57.0 57.2 57.8 58.4 59.0 64.1 68.8 69.6 70.8 71.2 71.4 71.7 72.2 72.6 73.1 73.5 74.4 74.8 75.4 75.6 75.8 76.0 76.1 76.3 76.7 77.3 77.7 78.3 78.7 79.1 79.3 79.8 81.6 82.2 83.4

4.60 4.64 4.63 4.66 4.72 4.68 4.74 4.52 4.71 4.60 4.60 4.66 4.70 4.62 4.70 4.71 4.86 4.74 4.89 4.74 4.54 4.44 4.56 4.68 4.53 4.40 4.19 4.29 4.32 4.32 4.21 4.18 -

3.4 5.5 2.8 2.6 2.1 3.7 2.4 4.4 4.2 4.4 5.1 2.9 3.0 5.4 3.3 4.9 2.1 3.4 4.0 2.8 1.7 3.0 1.2 1.1 1.6 3.2 2.9 1.8 1.7 1.7 4.0 3.1 3.1 2.6 7.7 2.2 6.1 3.9 6.3 8.7 8.5 8.0 7.4 6.8

Sowerset al.: A 135-ka Vostok-SPECMAP Correlation

759

TABLE 3. (continued)

SampleDepth, cm,

Age, ka

Benthic õ180,%0PDB

SeaSurface Temperature, oC

820 824 830 832 834 836 838 840 842 844 846 848 850 860 870 880 890 894 900 902 904 906 910 920 930 930 940 950 960 970 980 990 1000 1010 1020 1030 1040

85.1 86.3 88.0 88.6 89.2 89.8 90.4 91.0 91.2 91.5 91.7 92.0 92.3 93.6 94.9 96.2 97.8 98.4 99.4 100.2 100.9 101.7 103.3 104.8 106.3 106.3 107.8 109.3 110.8 114.7 118.6 122.6 125.2 127.8 130.3 132.7 135.1

4.18 4.28 4.34 4.30 4.23 4.25 4.16 4.12 4.18 4.13 4.33 4.32 4.28 4.24 4.30 4.05 3.71 3.28 3.38 4.00 4.64 4.32 5.02

8.3 6.2 8.8 7.7 7.3 8.8 6.3 6.9 4.4 3.5 2.1 2.9 5.8 4.6 2.9 7.3 6.9 6.2 7.0 5.7 7.9 6.9 8.4 6.3 4.8 4.7 2.3 4.6 4.4 7.7 9.3 8.2 8.8 7.2 3.7 3.2

1050 1060 1070 1080 1090 11O0 1110 1120 1130 1140 1150 1160 1170 1180 1190

136.0 136.9 137.8 138.7 139.6 140.5 141.4 142.3 143.6 144.9 146.1 147.4 148.7 150.0 151.3

4.89 5.00 4.94 5.01 4.91 4.92 4.75 4.84 4.86 4.84 4.86 4.88 -

2.6 2.8 2.2 3.6 2.2 3.2 2.0 3.6 3.1 3.1 2.7 3.8 2.9 2.9 3.9

760

Sowers et al.: A 135-ka Vostok-SPECMAP Correlation

TABLE 3. (continued)

SampleDepth, cm 1196 1200 1210 1220 1226 1230 1240 1250 1250 1256 1260 1260 1270 1270 1280 1290 1300

Age, ka 152.1 152.6 153.4 154.2 154.7 155.0 155.8 156.6 156.6 157.0 157.4 157.4 158.2 158.2 159.0 159.7 160.5

Benthic õ180.%•PDB 4.92 4.74 4.43 4.58 4.72 4.70 -

SeaSurface Ternptralur½.øC 3.1 2.0 1.5 1.5 0.5 1.7 1.5 1.9 2.8 1.7 3.1

Ages are basedon the correlationinto the stackedSPECMAP

•518Oforam record.The •518Oforam datais a composite recordof the Cibicidesand Uvigerina species.The last columnliststhe seasurfacetemperatures asdeducedfrom the diatomtransfer function D 166/34/4 [Pichon et al., 1992]. PDB indicatesPee Dee belemnite.

CONCLUSIONS

lastglacialmaximumandsubsequent deglaciation is as much as 50% lower than that for the Lorius et al.

We haveconstructed a recordof •il8Oat mfromthe analysesof trappedgasesin the Vostokice core

covering thelastfull glacialcycle.Our•j18Oat m recordis verysimilarto theseawater •j180record overthe sameperiod. We interpretthisstrong similarityasanindicationthatchanges in the•i180of seawater, which are known to be associatedwith

changesin the volumeof continentalice, havebeen

transmitted to theatmospheric 0 2 reservoirby photosynthesizing organismsnearthe surfaceof the oceanand on the continents.Furthermore,it appears

thatthemajorfactorinfluencing •j18Oat misthe•i180 of seawater.

Based on the similar behavior of these

two•i180records, we havecorrelated the•j18Oat m recordfromtheVostokicecoreintothe•jl8Oforam recordfrom V 19-30 usingan inversecorrelation method. Resultsof the correlationwith eight coefficients

show that 77% of the variance is shared

betweenthe•j18Oat mand•i18Osw records. We have utilized the results of the correlation to

derivean ice age-depthrelationfor theVostokice core. The record of accumulation

rate deduced from

our correlatedice age-depthrelationdiffersfromthat developedby Loriuset al. [ 1985]. For thecorrelated ice age-depthcurve,the accumulation rateduringthe

[ 1985] agemodel. For the restof the core,the correlatedrecordof accumulationrate showslarge oscillationswhich are mostlikely the resultof 1- to 3-kyr errorsin our ice age-depthrelation. Such

errorsmay resultfrom errorsin (1) the SPECMAP timescale which are calculated to lie between 1.1 and

7.7 kyr [Martinsonet al., 1987], (2) estimatesof Aagewhichareprobablylessthan+ 2 kyr, and(3) estimates of theturnovertimeof atmospheric 02 (_+0.5kyr). Errorsmay alsoresultfrom the fact that

•jl8Oatm Canbeinfluenced by hydrologic and biologicvariabilityin addition tochanges in •i18Osw. We haveinvestigated thefidelityof thecorrelated ice age-depthrelationby comparinga seasurface temperature recordfromthe southern oceanwith the recordof Antarctictemperature fromtheVostokice core. The highdegreeof covariationbetweenthese tworecordssupports theuseof •jl8Oatm asa proxy

for •i18Osw andindicates thatourcorrelated agedepthcurvefor Vostokgivesa chronologywhichis more consistentwith the SPECMAP chronology. Becauseof poorsampleresolutionbetween400-

and800-mdepthandthesmallchanges in •il8Osw between 25 and 55 ka in the SPECMAP record, we

areunableto reliablycorrelate thetwo•j180records overthistime/depthinterval.

Sowerset al.:A 135-kaVostok-SPECMAP Correlation

761

Lorius ice age (ka) 0

25

50

75

100

125

2 Vostok

o

-

I I I I

I I

>'-6 --

-8

0

25

50

75

100

125

150

SPECMAP age (ka) Fig. 8. Recordsof theVostokinversionlayertemperature andtheseasurfacetemperature (SST) recordfromMD88-770 plottedon thecommontemporalframework.The agemodelfor the Vostokinversionlayertemperature is ourbestestimatefromFigure5. The SSTrecordfrom MD88-770 hasa stateduncertainty of + IøC. The plotshowsa higherdegreeof similaritythana similarplotmadewiththeVostoktemperature recordplottedontheLoriuset al. [ 1985] chronology.We interpretthisresultassupport for ourcommontemporalframeworkandtherefore

ouruseof õ18Oatm asa proxyforõ18Osw. Forreference, wehaveplottedtheLoriuset al. [1985] timescaleon theupperaxisfor comparison.

The dataandarguments presented heredonotgive anyreasonto believethatourchronostratigraphy is any betteror worsethan that of Lorius et al. [ 1985]. Futureeffortsto dateGreenland ice coresby countingannualdustlayersbackthroughthe penultimate deglaciation will helpin constructing absoluteice corechronologies.As these"absolute" chronologies becomeavailable,we will thenbe able to establishtherelationship betweentheice core climaterecordsandanyclimaterecordswhichcanbe 230Th- 234U dated.

APPENDIX: CONSTRUCTING AN ICE AGEDEPTH RELATIONSHIP FOR THE VOSTOK ICE CORE

We usedthe followingapproachto calculatean ice age-depthcurvefor the Vostokice core.

1. Calculatethecorrelationgasageof eachtrapped gassamplein the Vostok ice core. The correlation ageis ourestimateof thegasageat eachsample depthaccordingto the SPECMAP chronology.It is calculated as described in an earlier section.

2. Calculatetheice age-gasagedifference(Aage) for eachsampleusingthedensification approach outlinedby Sowerset al. [1992]. Briefly, air is assumed to mix rapidlythroughthetim to thedepth of bubbleclose-off.We assumethatthedensityof ice in thebubblecloseoff regionincreases with decreasing temperature accordingto theempirical relationof Martinerieet al. [ 1992]. The depthat whichfirn reachesthiscloseoff densityis givenasa functionof accumulation rateandtemperature by the empiricalequationsof HerronandLangway[ 1980]. The ice ageat thisdepthis theice age-gasage differenceandequalsthemassof theoverburden per unit areadividedby the meanaccumulation rate.

762

Sowers et al.: A 135-ka Vostok-SPECMAP

This approachgivesAagevaluesvery closeto those calculatedusingthe methodof Bamolaet al. [ 1991]. 3. Calculatetheice ageat eachdepthby adding Aage(calculatedin step2) to thecorrelationgasage for eachdepthcalculatedin step1. 4. Calculate ice accumulation rate as a function of

depth. Accumulationrateis equalto the slopeof the ice age-depthcurvedividedby thethinningfunction. The thinningfunctionis definedasthemassper unit areaof an annuallayerof ice at depthz in theice sheetdividedby the massper unit areaat thetime of deposition.For this calculationwe haveadoptedthe thinningfunctioncalculatedby Ritz [1992]. 5. RecalculateAagevaluesfor eachgassample usingthe curveof accumulation rateversusdepth calculatedin step4. The pointof thisrecalculation is to estimateAagevalueswhichareconsistent with the correlatedice agedepthrelation. 6. Calculateanotherestimateof the ice ageversus depthrelationby addingthe Aagevaluesfrom step5

Correlation

to the correlatedgasage at eachdepthfrom step1. 7. Calculatethe "correlationice ageversusdepth" relationby averagingthe ice agesfrom steps3 and6. We averagethe ice agesbecausethe iterative proceduredescribedabovecausesice agesof some samplesto divergewith successive iterations.Our ice agecurvesdid not convergewith successive iterations because of errors in the initial correlated

gasage-depthrelationwhichareon theorderof a few centuries.

These errors result in anomalous

accumulation rate estimates,which lead to errors in

subsequent Aageestimatesover someportionsof the core. We notethatice agescalculatedin steps3 and 6 differ by an averageof 0.2 kyr, with a maximum differenceof 2 kyr. Doesour chronologygive reasonable curvesof accumulation

rate? One test of our curve of

correlatedice ageversusdepthis whetheror notit providesreasonable estimates of accumulation rate throughoutthe studyperiod. In Figure9, we plot

3.5

_

2,5

-

•3180 Correlation (ice)

Lorlus et. al. [1985]

_

1,5

-

0,5

Pichon et al. [1992]

3180 Correlation (gas)

0

500

1000

1500

2000

Depth (mbs) Fig. 9. Recordsof the surfaceaccumulation rate at Vostok. The thick line is derivedfrom the Loriuset al. [ 1985] ice ageversusdepthrelationandhasbeencorrectedfor thinning.The other

threecurvesarefromtheõ180correlation (gasandice)andthediatomcorrelation (dashed line). Notetheloweraccumulation ratescalculated fromtheõ180-correlated iceage-depth relations between300 and 500 mbs. Also notethe largeoscillationsin the correlatedaccumulationrate profile below 700 rn below surface(mbs).

Sowerset al.: A 135-ka Vostok-SPECMAP Correlation

accumulation rateversusdepthfor all correlative age modelsfor Vostok. Forcomparison, we alsoplot the record of accumulation rate of Lorius et al.

[1985]. Therearetwo curvesfromthisstudywhich aremarked/5180correlation (ice)and/5180 correlation(gas). The curvelabeled/5180correlation (gas)is thederivativeof themappingfunctionfrom

the/5180 of 0 2 and•518Osw correlation. Thecurve labeled/5180 correlation (ice)istheslopeof the correlationice ageversusdepthrelationdescribed in theprevioussection.Our/5180correlation (ice) curveoscillateswith a highfrequencywhilethe paleoaccumulation recordfromLoriuset al. [ 1985] variesslowly. We attributemuchof thehighfrequency oscillations to smallerrorsin ourage estimates.

The high-frequency variationsin accumulation ratesimpliedby ourchronology (Figure8) are glaciologically improbable.However,theydo not implylargeerrorsin estimatedages.Considertwo levelswithinthecorewhichhavebeenassigned ice agesdifferingby 2 kyr. If the errorin the relative agesof theselevelsis 1.0 kyr, thenthe accumulation ratecalculatedfor theintervalin question wouldbe higherthantheoriginalvalueby a factorof 2 or lowerby a factorof 0.67. Thisrangeof accumulation ratesis comparable to theoscillations in theaccumulation ratewhichwe observe in Figure9. We thusattributea largeportionof thedifference between the two curves of accumulation rate shown

in Figure9 to uncertainties in thecorrelated ice age assignedto eachdepth. Datingerrorscouldbe the resultof smallerrorsin the SPECMAP chronology, uncertainties in calculatedvaluesof Aage,or variations in the M-D

effect.

With our chronology,the depthintervalbetween 290 and 360 m (terminationI) corresponds to a periodof 6 kyr, which is 2.8 kyr longerthanthat estimatedusingthe flow modelapproach[Loriuset al., 1985]. The longerduration,inferredfrom our correlation,corresponds to accumulation rateswhich are55% lowerthanthoseestimatedby Loriuset al. [1985]. Accumulationratesfor thisdepthinterval from the Pichonet al. [ 1992] chronologyare25% higherthan thoseinferredfrom the flow model. Accumulation rate estimates from the flow model are consistent with 10Be data for the Vostok ice core if one assumes that the flux of 10Be to Vostok was

about38% higherduringthelastglacialmaximum [Raisbeck et al., 1992]. There are three factors

whichinfluencethe [10Be]in ice at Vostok:(1) the production of 10Bein theatmosphere (whichis dictatedby the flux of cosmicraysto the outer portionof the Earth'satmosphere andthe intensityof the Earth'sgeomagnetic field), (2) the accumulation rate at the time of deposition,and (3) variationsin localprecipitation patternson theEastAntarctic plateau[Raisbecket al., 1992]. We haveestimated

theflux of 10Beto theVostokareausingthe accumulation ratecurvesshownin Figure9 andthe

763

Vostok [10Be]recordfrom [Raisbecket al., 1992]. Ourcalculations indicatethattheglacial10Beflux to Vostokwas 19% higherfor our correlatedice age depthrelation,38% higherfor the Loriusagemodel, and83% higherfor the Pichonagemodelrelativeto Holocenevalues. Theseresultsarein qualitative agreement with thehigherflux of 10Beto thePacific Oceanduringthe sametime [Lao et al., 1992]. Furthermore,if we assumethatthe localprecipitation patternsaroundVostok remainedconstantoverthe

last25 ka, thentheenhanced flux of 10Beduring glacialperiodssupports thehypothesis that10Be productionwasenhanced duringthelastglacial maximum.

Acknowledgments. We wouldlike to thankall the membersof the SovietAntarcticExpeditions,Terres Australeset AntarctiquesFran•aises(TAAF), ExpeditionsPolairesFran•aises(EPF), andthe Office of Polar Programs(NSF) who helpedretrieve andtransportthe Vostok core. J. Orchardokindly providedhis analyticalexpertise.We are also gratefulto J. M. Bamola, E. Bard, C. Lorius, C. Ritz, J. Imbrie, and all the SPECMAP heroes for

theirusefulcomments.This researchwas supported on the Americansideby the NSF Division of Polar Programs(DPP grants 8820807 and 8822020) and on the Frenchsideby TAAF andProgramme Nationald'Etudede la Dynamiquedu Climat. DGM's researchwas supportedby NSF grant 8312637 (SPECMAP). This is Lamont-Doherty EarthObservatorycontribution5081. REFERENCES

Bard, E., B. Hamelin, R. Fairbanks, and A. Zindler,

Calibration of the14Ctimescale overthepast 30,000 yearsusingmassspectrometric U-Th ages from Barbados corals, Nature, 345, 405-410, 1990a.

Bard, E., B. Hamelin, and R. G. Fairbanks, U-Th

agesobtainedby massspectrometry in coralsfrom Barbados:Sealevel duringthe past130,000 years,Nature, 346, 456-458, 1990b. Bamola, J. M., D. Raynaud, Y. S. Korotkevich, and C. Lorius, Vostok ice core provides160,000year recordof atmospheric CO2, Nature, 329, 408-414, 1987.

Bamola, J. M., P. Pimienta,D. Raynaud,and Y. S. Korotkevich,CO2-climaterelationship asdeduced from the Vostok ice core: A reexamination based on new measurements and on a reevaluation of the

air dating, Tellus, Ser. B, 43, 83-90, 1991. Beer, J., S. J. Johnsen, G. Bonnani, R. C. Finkel,

C. C. Langway, H. Oeschger,B. Stauffer,M.

Suter,andW. Woelfli,løBepeaksastime markersin polarice cores,in TheLast Deglaciation:AbsoluteandRadiocarbon Chronologies,editedby E. Bard andW.

764

Sowers et al.: A 135-ka Vostok-SPECMAP

Broecker,pp. 141-153, Springer-Verlag,New York, 1992.

Bender,M. L., L. D. Labeyrie,D. Raynaud,and C. Lorius,Isotopiccomposition of atmospheric 0 2 in ice linkedwith deglaciation andglobalprimary productivity,Nature, 318, 349-352, 1985. Birchfield, G. E., Changesin deep-oceanwater

/5•80andtemperature fromthelastglacial maximumto thepresent,Paleoceanography, 2, 431-442, 1987.

Chappell,J., andN.J. Shackleton,Oxygenisotopes and sea level, Nature, 324, 137-140, 1986.

Chappellaz,J., J.-M. Bamola, D. Raynaud,Y. S.

Correlation

densification: An empiricalmodel,J. Glaciol.,25, 373-385, 1980. Hovan, S. A., D. K. Rea, and N. G. Pisias, Late Pleistocene continental climate and oceanic

variabilityrecordedin northwest Pacific sediments,Paleoceanography, 6, 349-370, 1991. Howard, W. R., and W. L. Prell, Late Quaternary surface circulation of the Southern Indian Ocean

andits relationship to orbitalvariations, Paleoceanography,7, 79-118, 1992. Imbrie, J., J. D. Hays, D. G. Martinson,A.

Mcintyre,A. C. Mix, J. J. Morley, N. G. Pisias, W. L. Prell, and N.J. Shackleton,The orbital

Korotkevich, andC. Lorius, Atmospheric CH.4

theoryof Pleistocene climate:Supportfroma

Vostok ice core, Nature, 345, 127-131, 1990.

Milankovitchand Climate,editedby A. Bergeret al., pp. 269-305, D. Reidel,Norwell, Mass.,

recordoverthe lastclimaticcyclerevealedby ttie

Craig, H., Y. Horibe, and T. A. Sowers, Gravitationalseparationof gasesandisotopesin polaxice caps,Science,242, 1675-1678, 1988. Dansgaard,W., H. B. Clausen,N. Gundestrup,S. J. Johnsen,and C. Rygner,Dating andclimatic interpretationof two deepGreenlandice cores,in GreenlandIce Core: Geophysics,Geochemistry, and the Environment,Geophys.Monogr. Ser., Vol. 33, editedby C. C. LangwayJr., et al., pp. 71-76, AGU, Washington,D.C., 1985. Dole, M., The relativeatomicweightof oxygenin water and air, J. Am. Chem. Soc., 57, 27312732, 1935. Dole, M., G. A. Lane, D. P. Rudd, and D. A.

Zaukelies,Isotopiccompositionof atmospheric oxygenand nitrogen,Geochim.Cosmochem. Acta., 6, 65-78, 1954.

Etheridge,D., G. Pearman,and P. Fraser,Changes in tropospheric methanebetween1841 and 1978 from a high accumulation rateAntarcticice core, Tellus, Ser. B, 44, 282-294, 1992. Fairbanks, R. G., and R. K. Matthews, The marine

oxygenisotoperecordin Pleistocene coral, Barbados,West Indies, Quat. Res. N.Y., 10, 181-196, 1978. Guiot, J., A. Pons, J. L. de Beaulieu, and M. Reille,

A 140,000-yearcontinentalclimatereconstruction from two Europeanpollenrecords,Nature,338, 309-313, 1989. Hammer, C. U., H. B. Clausen, and H. Tauber,

Ice-coredatingof the Pleistocene/Holocene

boundary appliedto a calibration of the14Ctime scale, Radiocarbon, 28, 284-291, 1986. Hays, J. D., J. Imbrie, and N.J. Shackleton, Variations

in the Earth's orbit: Pacemaker of the

ice ages,Science,194, 1121-1132, 1976a. Hays, J. D., J. Lozano, N. Shackleton,and G. Irving, Reconstruction of the AtlanticandWestern Indian Ocean sectorsof the 18,000 B. P. Antaxctic

Ocean,in Investigationof Late Quaternary Paleoceanography andPaleoclimatology, edited by R. M. Cline, pp. 337-372, GeologicalSociety of America, Boulder, Colo., 1976b.

Herron, M. M., and C. C. Langway, Jr., Fim

revised chronology ofthemarine/5180 record, in 1984.

Johnsen,S. J., H. B. Clausen,W. Dansgaard,K. Fuhrer, N. Gundestrup,C. U. Hammer, P. Iversen, J. Jouzel, B. Stauffer, and J.P.

Steffensen, Irregularglacialinterstadials recorded in a new Greenland ice core, Nature, 359, 311313, 1992.

Jouzel,J., and L. Mefiivat, Deuteriumand oxygen

18 in precipitation: Modelingof theisotopic effectsduringsnowformation,J. Geophys. Res., 89, 11,749-11,757, 1984. Jouzel, J., C. Lorius, J. R. Petit, C. Genthon, N. I.

Barkov, V. M. Kotlyakov, and V. M. Petrov, Vostokice core:a continuousisotopetemperature recordoverthe lastclimaticcycle(160,000years), Nature, 329, 403-407, 1987. Jouzel, J., G. Raisbeck, J.P. Benoist, F. Yiou, C.

Lorius, D. Raynaud,J. R. Petit, N. I. Barkov,Y. S. Korotkevitch,and V. M. Kotlyakov,A comparison of deepAntarcticicecoresandtheir implications for climatebetween65,000and 15,000 yearsago, Quat. Res.N.Y., 31, 135-150, 1989.

Jouzel, J., J. R. Petit, N. I. Barkov, J. M. Barnola,

J. Chappellaz,P. Ciais, v. M. Kotkyakov,C. Lorius, V. N. Petrov,D. Raynaud,and C. Ritz, The lastdeglaciation in Antarctica:Further evidenceof a "YoungerDryas"typeclimatic event.,in TheLastDeglaciation:Absoluteand RadiocarbonChronologies,editedby E. Bard and W. Broecker,pp. 229-266, Springer-Verlag, New York, 1992. Jouzel, J., N. I. Barkov, J. M. Barnola, M. Bender,

J. Chappellaz,C. Genthon,V. M. Kotlyakov,V. Lipenkov,C. Lorius, J. R. Petit, D. Raynaud,G. Raisbeck, C. Ritz, T. Sowers, M. Stievenard, F. Yiou, and P. Yiou, Vostok ice cores:extending

theclimaticrecordsoverthepenultimate glacial period,Nature, 364, 407-412, 1993. Kotlyakov,V. M., Globalchangesoverthe last climaticcyclefromAntarcticice corerecords,in Glaciers-Ocean-Atmosphere Interactions,edited by V. M. Kotlyakkov,pp. 15-27,International

Sowers et al.: A 135-ka Vostok-SPECMAP Correlation

Associationof HydrologicalSciences,St. Petersburg,Russia, 1990. Kroopnick,P., and H. Craig, Atmosphericoxygen: Isotopiccomposition andsolubilityfractionation, Science, 175, 54-55, 1972.

Kukla, G., andZ. A. An, Loessstratigraphyin centralChina,Palaeogeogr.Palaeoclimatol. Palaeoecol., 72, 203-225, 1989.

Labeyrie,L. D., J. C. Duplessy,and P. L. Blanc, Variationsin modeof formationandtemperature of oceanicdeepwatersoverthepast125,000 years,Nature, 327, 477-482, 1987. Lambeck, K., and M. Nakada, Constraints on the

ageanddurationof thelastinterglacialperiodand on sea-level variations, Nature, 357, 125-128, 1992.

Lao, Y., R. F. Anderson, W. S. Broecker, S. E. Trumbore, F. J. Hofmann, and W. Wolfli,

Increased production of cosmogenic 10Beduring the last glacialmaximum,Nature, 357, 576-578, 1992.

Lorius, C., J. Jouzel, C. Ritz, L. Merlivat, N. E.

Barkov, and Y. S. Korotkevich,A 150,000-year climatic record from Antarctic ice, Nature, 316, 591-595, 1985.

Lozano,J., and J. D. Hays, Relationshipof radiolarianassemblages to sedimenttypesand physicaloceanography in theAtlanticandWestern Indian ocean sectorsof the Antarctic Ocean, in

Investigationof Late Quaternary Paleoceanography and Paleoclimatology, edited

by J. D. Hays,pp. 303-336, GeologicalSociety of America, Boulder, Colo., 1976.

Lyle, M., Climaticallyforcedorganiccarbonburial in equatorialAtlanticandPacificoceans,Nature, 335, 529-532, 1988.

Martinerie, P., D. Raynaud,D. Etheridge,J.-M. Bamola, andD. Mazaudier,Physicalandclimatic parameters whichinfluencetheair contentin polar ice, Earth Planet. Sci. Lett., 112, 1-13, 1992. Martinson, D. G., W. Menke, and P. Stoffa, An

inverseapproachto signalcorrelation,J. Geophys.Res., 87, 4807-4818, 1982. Martinson, D. G., N. G. Pisias,J. D. Hays, J. Imbrie, T. C. Moore Jr., and N.J. Shackleton,

Age datingandtheorbitaltheoryof theiceages: developmentof a high-resolution 0 to 300,000year chronostratigraphy,Quat. Res.N.Y., 27, 127, 1987.

Meyer, M. K., Net primaryproductivityestimates for the last 18,000yearsevaluatedfrom simulationsby a globalclimatemodel,M. S. thesis, Univ. of Wisconsin, Madison, 1988.

Mix, A. C., Influenceof productivityvariationson long-termatmospheric CO2, Nature,337, 541544, 1989.

Morita, N., The increaseddensityof air oxygen relativeto water oxygen,NipponKagakuKaishi, 56, 1291, 1935.

765

Neftel, A., H. Oeschger,T. Staffelbach,and B. Stauffer,CO2 recordin theByrd ice core50,0005,000 yearsBP, Nature, 331,609-611, 1988. Petit, J. R., L. Mounier, J. Jouzel, Y. S.

Korotkevich,V. I. Kotlyakov,andC. Lorius, Paleoclimatological andchronological implications of the Vostok core dust record, Nature, 343, 5658, 1990.

Phillpot, H. R., and J. W. Zillman, The surface temperatureinversionoverthe Antarcticcontinent, J. Geophys.Res., 75, 4161-4169, 1970. Pichon,J.-J., L. D. Labeyrie, G. Bareille, M. Labracherie,J. Durpat, and J. Jouzel,Surface watertemperaturechangesin the highlatitudesof the southernhemisphere overthelastglacialinterglacialcycle,Paleoceanography, 7, 289-318, 1992.

Pisias, N. G., D. G. Martinson, T. C. Moore, N.J.

Shackleton,W. Prell, and B. Boden, High resolutionstratigraphic correlationof benthic oxygenisotopicrecordsspanningthe last 300,000 years,Mar. Geol., 56, 119-136, 1984. Prell, W. L., and J. E. Kutzbach, Monsoon

variabilityover the past 150,000years,J. Geophys.Res., 92, 8411-8425, 1987. Prell, W. L., J. Imbrie, D. G. Martinson, J. J.

Morley, N. G. Pisias,N.J. Shackleton,and H. F. Streeter,Graphiccorrelationof oxygenisotope stratigraphy: Applicationto thelateQuaternary, Paleoceanography, 1, 137-162, 1986. Raisbeck, G. M., F. Yiou, D. Bourles, C. Lorius, J. Jouzel, and N. I. Barkov, Evidence for two

intervals of enhanced løBedeposition in Antarctic ice duringthe lastglacialperiod,Nature, 326, 273-277, 1987. Raisbeck, G. M., F. Yiou, J. Jouzel, J. R. Petit, N.

I. Barkov,andE. Bard,løBedeposition at Vostok,Antarcticaduringthe last 50,000 years andits relationshipto possiblecosmogenic productionvariationsduringthisperiod,in The LastDeglaciation:AbsoluteandRadiocarbon Chronologies, editedby E. Bard andW. Broecker,pp. 125-139, Springer-Verlag,New York, 1992.

Raynaud,D., J. Chappellaz,J.-M. Bamola, Y. S. Korotkevich,andC. Lorius,ClimaticandCH4 cycleimplications of glacial-interglacial CH4 changein the Vostokice core,Nature, 333, 655657, 1988. Reeh, N., S. J. Johnsen, and D. Dahl-Jensen,

DatingtheDye 3 deepice coreby flow model Calculations,in GreenlandIce Core: Geophysics, Geochemistry,and the Environment,Geophys. Monogr. Ser., Vol. 33, edited by C. C. Langway Jr., et al., pp. 71-76, AGU, Washington,D.C., 1985.

Ritz, C., Flow modelingthe Vostokregion,Ph.D. dissertation, Domaine Univ., Grenoble, France, 1992.

766

Robin, G. D. Q., Ice coresand climatic changes, Philos. Trans.R. Soc. London, Ser. B, 280, 143168, 1977. Sarnthein, M., E. Jansen, M. Arnold, J. C.

Duplessy,H. Erienkeuser,A. Flatoy, T. Veum,

E. Vogelsang, andM. S. Weinelt,•5180time-slice reconstruction of meltwater anomalies at termination I in the North Atlantic between 50 and

80øN,in TheLast Deglaciation:Absoluteand RadiocarbonChronologies, editedby E. Bard and W. S. Broecker,pp. 183-200, Springer-Verlag, New York, 1992. Schwander, J., The transformation of snow to ice

andthe occlusionof gases,in TheEnvironmental Recordin Glaciersand Ice Sheets,editedby H. Oeschgerand C. C. Langway,pp. 53-67, John Wiley, New York, 1989. Shackleton,N.J., Oxygenisotopes,ice volumeand sea level, Quat. Sci. Rev., 6, 183-190, 1987.

Shackleton,N.J., and N. D. Opdyke,Oxygen isotopeandpaleomagnetic stratigraphy of equatorialPacificcoreV28-238: Oxygenisotope

temperatures andicevolumesona 105and106 year scale,Quat. Res.N.Y., 3, 39-55, 1973. Shackleton,N.J., and N. G. Pisias,Atmospheric carbondioxide,orbitalforcing, andclimate,in TheCarbonCycleandAtmospheric CO2Natural VariationsArcheanto Present,Geophys.Monogr. Ser., vol. 32, editedby E. T. Sundquistand W. S. Broecker,pp. 303-317, AGU, Washington, DC, 1985. Shackleton, N., J. Le, A. Mix, and M. A. Hall,

Sowers et al.: A 135-ka Vostok-SPECMAP

Correlation

duringthepenultimatedeglaciation, Paleoceanography, 6, 679-696, 1991. Sowers,T., M. Bender,D. Raynaud,and Y. S.

Korotkevich, The•515N of N2in airtrapped in polarice:A tracerof gastransport in thefirn anda possibleconstrainton ice age-gasagedifferences, J. of Geophys.Res., 97, 15,683-15,697, 1992. Stauffer,B., E. Lochbronner,H. Oeschger,and J. Schwander,Methaneconcentration in the glacial atmosphere wasonly half thatof thepreindustrial Holocene, Nature, 332, 812-814, 1988. Yiou, F., G. M. Raisbeck, D. Bourles, C. Lorius,

andN. I. Barkov, løBein ice at VostokAntarctica duringthe lastclimaticcycle,Nature,316, 616617, 1985.

M. Bender and T. Sowers, Graduate School of

Oceanography, Universityof RhodeIsland, Narragansett,RI 02882-1197. J. Jouzel, Laboratoire de Modelisation du Climat et de l'Environnement Batiment 709, Orme des

Meriseeirs,CE Saclay91191 Gif Sur Yvette Cedex, France.

Y. S. Korotkevich, Arctic and Antarctic Research

InstituteBeringaStreet38, St. Petersburg, Russia. L. Labeyrie,CentredesFaiblesRadioactivites Laboratorie

Mixte

CNRS-CEA

Parc du CNRS

91190 Gif Sur Yvette, France.

D. Martinson,Lamont-DohertyEarthObservatory Palisades, NY 10964.

J. J. Pichon,DepartmentG6ologieet Oc6anologie,

Carbonisotoperecordsfrom Pacificsurface watersand atmosphericcarbondioxide,Quat. Sci.

URA 197, Av des Facult6s Universit6 Bordeaux 1,

Rev., 11, 387-400, 1992.

33405 Talence Cedex France.

Sowers,T. A., M. L. Bender, and D. Raynaud, Elementalandisotopiccompositionof occluded 0 2 andN 2 in polarice, J. Geophys.Res.,94,

D. Raynaud,Laboratoirede Glaciologieet Geophysiquede l'EnvironnementBP 96-38402 St. Martin d'Heres Cedex, France.

5137-5150, 1989.

Sowers,T., M. Bender, D. Raynaud,Y. S. Korotkevich,andJ. Orchardo,The •5180of atmospheric 0 2 fromair inclusions in theVostok ice core:Timingof CO2 andicevolumechanges

(ReceivedJanuary14, 1993; revised June 17, 1993;

acceptedAugust 17, 1993.)