The Isotopic Climatic Records in the Allerdbllingyounger Dryas and Postyounger Dryas Events

GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 9, NO. 4, PAGES 557-563, DECEMBER 1995

The isotopic climatic records in the Allerid-B011ing- Younger Dryas and post-Younger Dryas events

Samuel Epstein Divisionof Geologicaland Planetary Sciences, California Institute of Technology,Pasadena

Abstract. The combinedisotopic records in the Greenlandice cores,in modernand ancientwood samples,and in marineforaminifera provide a climaticrecord for theAller6d-B011ing-Younger Dryasand post-Younger Dryas Periods that is not compatiblewith theusual climatic interpretationof the2H/1H ratios in theGreenland ice cores. For example, the Younger Dryas was not solelya NorthAtlantic event because evidence for it is alsofound in the2H/1H record in wood sampleswhich grew on thenorth slope of Alaska. The YoungerDryas was not necessarilya time of coolingover the ice caps. Moreover,it mighthave been a warmingperiod in the temperate zonesof thenorthern hemisphere continents. A betterunderstanding of theclimatic scenarios duringthese periods of timecould be ascertainedfrom additional isotopic data in plantremains whichgrew during these intervals and which cover a widerclimatic range on the continents.

Introduction lb). Five of these sampleswere dated by Acceleration Mass Spectrometer (AMS) techniques. These samples were Salix The transitionfrom the last Glacial Period to the present (willow shrubs)collected in an area borderedby north latitude InterglacialPeriod is recordedin the isotopiccomposition of 69ø 38' and 70ø 30' and west longitude151 ø 36' and 155ø 56'. the Greenlanddeep ice cores(e.g., DYE 3) [Dansgaardet al., Nonexchangeablehydrogen was extractedfrom the plant cellu- 1989]. This transitionstarts with a gradualincrease in the 8D lose [Epsteinet al., 1976]. Becauseboth the ice and plants in values[•D = (Rsample/ Rstandard - 1)x 1000 (%o), R = 2H/1H effect record the isotopic compositionof the ambient meteoric standardisStandard MeanOcean Water (SMOW)], followed b; waters [Epstein et al., 1976; White eta/., 1994], their $D a sharpincrease of the $D from -260 to -230%0(Figure l a). recordsshould be comparable. Preferential loss of cellulose in This is referredto as the Aller6d-BOllingevent. The Aller6d- old wooddue to agingdoes not affectthe $D valuesof the orig- Bolling is followed by the Younger Dryas event, which is inal cellulose [Yapp and Epstein, 1977]. The negative plant associatedwith a decreaseof the 5D value to -285%0,followed $D values(-217%o to -135%o,Figure 1) indicatethat the water by a rise to -235%0between 10.6 and 10.4 Kyr B.P. This in used by these plants must have precipitated from air masses turn is followed by a drop to -255%0and a rise to -230%0, that experienced many cycles of isotopic fractionation prior followed by the generaltrend towardthe InterglacialPeriod. to their arrival on the Alaskan north slope. The AllerOd-Bollingis generallyconsidered to be a warming TheAlaskan isotope record is plottedagainst 14C age, and event,and the YoungerDryas is a coolingevent. Broeckeret the DYE 3 ice core record [Dansgaard et al., 1989] is plotted al. [Broecker et al., 1989; Broecker and Denton, 1990] againstdepth in Figure1. The coincidenceof $D maximaand attributed the cooling to a diversion of ice sheet meltwater minimaand the similarranges of the $D valuesfor the Alaskan from the MississippiRiver to the St. LawrenceRiver drainage plant remainsand the GreenlandDYE 3 ice recordsuggests that system [Broecker eta/., 1989]. This diversion served to shut the two records are reasonablysynchronous. off the North Atlantic conveyorbelt circulationsystem which As in the case of the ice record, the Aller6d-Belling in the currently supplies an enormous amount of heat to the Alaskan wood record starts with a sharp increaseof +30%0 in atmosphereover the North Atlantic region. Consequently, the $D at approximately12 Kyr B.P., followed by a continu- "the Younger Dryas is primarily a North Atlantic and not a ous decreasefrom -150 to -220%0correlated with the Younger globalevent" [Broecker and Denton,1990, pp. 54]. Enough Dryas event at approximately 10.5 Kyr B.P. This is followed stableisotope data now existto form a basisfor reexamining by a sharprise to -135%o,followed by a post-YoungerDryas this interpretationas well as reconsideringthe significanceof oscillation extending over 1000 years, spanning a 30%0 range the stableisotope records in ice coresand in samplesof wood in $D values. The synchronousvariations in the $D records that grew duringthe late-glacialperiods. in the ice core and in the willow shrubs are impressive and indicate that the Younger Dryas event is recordedin both iso- Results and Discussion tope records and therefore was apparently not exclusively a We analyzedthe $D of a seriesof 26 well-preserved•4C- North Atlantic event. In addition, the synchronousvariations also indicate that the rate of the growth of the ice cap is rela- dated plant remainsfrom the north slope of Alaska (Figure tively uniform for at least the period between 12 and 10 Kyr B.P, suggestinga uniform temperaturein this period of time. Copyright1995 by theAmerican Geophysical Union. The smallerrange in the $D valuesrecorded in the ice core,as Papernumber 95GB02741. comparedto that recordedin the wood samplesfrom the north 0886-6236/95/95GB-02741 $10.00 slope of Alaska, is likely due to some degreeof homogeniza-

557 558 EPSTEIN: ISOTOPIC CLIMATE RECORDS IN YOUNGER DRYAS EVENTS

gradual increaseinto the Holocene. This is usually explained as a rapid swing in air temperaturesover the Greenlandice sheet. There is an alternativeexplanation for the existingiso- -230•.• tope recordsfor the Aller6d-B011ing-YoungerDryas and post- Younger Dryas events. An alternativescenario is suggestedby the excessdeuterium values(d = •SD-8•5180, where d = excessdeuterium) for the DYE 3 ice samplescovering the transition from the Younger -260 - Dryas to the Preboreal[Dansgaard eta/., 1989]. As suggested by Dansgaard et al. [1989], these data show that the marine -270 - sourceof moisturefor the Younger Dryas in the polar regions was warmer than that for the Preboreal. They also suggested -28O - that this change was causedby a shift of a "dominatingmois- I 1780 1790 1800 1810 ture source in the subtropical part of the Atlantic Ocean" Depth in ice core (meters) [Dansgaardeta/., 1989, pp. 533] and presumablyby a warmer marine source of the moisture. Petit et al. [1991] show that I I I I the major effect on the magnitude of the excessdeuterium in B x - -140 the Antarctic snow is directly related to the sea surfacetemper- ature of the source of the moisture. /,x• / \ The effectof surfaceocean temperatures on the •SDrecord of _/x// • /•, ,/ x\ X/-'x Ii.--160 the ice caps is illustratedin Figure 2. This figure showsthe / b..,'x\•_xx/x , I Rayleigh-typecurves, similar to thosepublished by Yapp and x ,X Epstein [1977] that illustrate the relationship between iso- - X ,0' x- -180 I x/ topic compositionand temperatureof condensationfor precip- I / itation originatingfrom oceanwater with a fid of 0%0,but I x/ x -200 with different initial oceansurface temperatures (20 ø, 10ø and I I - IO/x 0øC). These curvesillustrate the effect of ocean surfacetem- peratureon the relationshipbetween temperature and the iso- -220 topic composition in the ice caps. For example, arrow A in Figure2 showsthat the fid recordin the ice cap will remain 8 9 10 11 12 constantat-300%0, in spite of a cooling of 10øC on the ice Years x103 BP cap, if at the sametime the marine sourceof water feedingthis ice cap coolsby approximately20øC. On the other hand,ar- Figure 1. (a) The õD recordof the 14C-datedwillow shrub row B showsthat an increasein fid of approximately130%o, samplesfrom the north slopeof Alaska, as comparedto (a) an normally indicating a warming trend, will be recordedon the exact copy of an enlargedversion of the publishedDYE 3 ice core data from Greenland[Dansgaard et al., 1989]. The •SD ice cap if the overlying air temperature remains constant, valuesof the ice coresare plottedagainst depth of the ice core while the temperatureof the ocean'ssurface decreases from 20ø sampleson the ice cap. (b) The points marked with crosses to 0øC. It is the unique and transienttemperature history of have a suggestedanalytical uncertainty of +160 years,whereas midlatitudeocean surfaces during the Aller6d-B011ing-Younger the open circles in Figure lb have been determinedby the Dryas and post-YoungerDryas that could play an unusualbut AMS techniquewith an error of _+80years. important role in determining the short-term isotopic variations on the ice caps at that time. These•SD values recorded in the ice capdepend on thediffer- tion that must have taken place on the ice cap at this great ence in temperaturebetween the air massesover the ice and the depth in the core. marine source of moisture. If simultaneous cooling of the The good correlationbetween the fid recordsof the wood ocean'ssurface and the atmosphereover the ice caps(DYE 3) is and the ice core betweenthe period 12.3 to 10 Kyr B.P. sug- assumedto be responsiblefor a largedecrease in •SDon the ice, gestssimilar variationsin their respectivemarine sourcesand certain circumstanceshave to exist. For example, assume an cooling historiesof the air massesproviding the moisturefor oceantemperature of 20øCand a temperatureover the ice cap of thesesites. It is difficult to imaginethe distantAtlantic Ocean -8øC. The cooling history of an air massfeeding the ice cap as the source of precipitation for the north slope of Alaska will resultin a fid valueof -200%0for the snowof the ice cap, becausethe fid valuesof the northslope shrubs are systemati- accordingto Figure 2. If the ocean temperatureremains at cally ß-80%o higher than thoseof the ice in the DYE 3 core in 20øC,but the temperatureof the ice capdrops to -15øC,the •SD Greenland, which borders the Atlantic Ocean. It seems more of the snowon the ice cap will decreaseto -265%0,a changeof reasonable to assume that the Pacific Ocean off the western -65%0. However, if the oceantemperature decreases by 10øC North American continent was the dominant, ultimate source and the temperatureon the ice cap dropsto -13'C, there should of the rain and snow that fell on the north slopeduring that be no changein the •SDof the snowon the ice cap. In orderto time. causethe decreasein the fid of -65%0on the ice cap, the tem- One of the unusual characteristics of the fid record for the peratureon the ice cap would have to cool further to -22øC. Greenlandice cap [Dansgaardet al., 1989] is the abruptshift The minimum•SD value of theYounger Dryas in theDYE 3 core to lower valuesat approximately10 Kyr, followedby a more already approachesthe •SDvalue for the Glacial Period. It EPSTEIN: ISOTOPIC CLIMATE RECORDS IN YOUNGER DRYAS EVENTS 559

o lO 20 I 1 I I I I 0

-lOO

-2oo

-300

-400

-5oo

ß.30 -20 -10 0 10 20 T'C

Figure 2. The relationshipsbetween the temperatureand the isotopic compositionof precipitationorigi- nating from an air mass which obtained its moisture from the surfaceof the ocean at three different ocean surface temperatures,20 ø, 10ø and 0ø C., were calculatedusing the Raleighequation where the valuesof the iso- topicfractionation factors were adjustedfor the temperatureof precipitation.Line A illustratesthat the •SDof the snow for the ice cap will remain constantif both the temperatureof the air mass over the ice cap and the temperatureof the surfaceof the oceandecrease simultaneously. Line B illustratesthe •SDrecord of the ice at a specificlocation at a constanttemperature, when the temperatureof the marine sourceof water changesfrom 20ø to 0ø C. Line C is the rangeof the •SDfor the DYE 3 ice core,assuming that the range15ø-8øC is entirely due to the temperaturechange of the oceansurface. Line D is similarto Line C but coversthe •SDrange for the samplesfrom the north slope of Alaska.

seemsunlikely that the temperatureof the Younger Dryas of followed by an increasein oceanwater temperatureof-7øC to the ice cap would be lower thantemperature during the Glacial 10.5 Kyr B.P., then followed by a decreasein ocean water Period. temperatureof-6øC to 10 Kyr B.P. This whole scenariofrom In orderto calculatea priorithe •SDof the ice capsfrom pre- 12.3 to 10 Kyr B.P. could be accomplishedby respective sumed changesin the ocean temperature,some initial condi- injection of cold water to the oceansurface followed by the in- tions must be specified. A temperatureof 15øC for the ocean corporationof the cold surfacewater into the dominantprein- surfacewas assignedto a specific•SD value of-300%0 in the ice jection water whosetemperature was 15øC,and then followed core. This will define a constanttemperature of-20øC over the by anotherinjection of cold water to the ocean surface. ice cap. If the õD of the ice is increasingas time progresses Figure 2 can also be applied to the data from the north slope from 12.3 Kyr B.P. towardyounger ages while air temperature of Alaska. The only difference involves the assignmentof a over the ice remainsfixed at -20øC,the surfaceocean tempera- constanttemperature of this sourceof water used for the plant ture would decrease from 15øC to 8øC. The broken line in growth. The minimum•SD of thiswater is -215%oand the tem- Figure3a was calculatedfrom line C in Figure 2 for a decrease peraturerange for the oceanis between17 ø and 8øC,as previ- in oceanwater temperaturefrom 15ø to 8øC at 12.3 Kyr B.P., ously assigned. As a consequence,the temperatureof the ice 560 EPSTFJN:ISOTOPIC CLIMATE RECORDS IN YOUNGER DRYAS EVENTS cap in Alaska which provides the water for the plant growth bonatesreflects warmer ocean temperatures,consistent with would be -12øC. Line D in Figure2 couldbe usedto definethe the conventionalinterpretation of the Greenlanddata that the rangeof the 5D recordfrom the northslope of Alaska. Aller6d-B•lling Period representsa warming temperature. A The temperatureof the oceansis a major factor affectingthe rapidrise in oceansurface temperatures at about12.3 Kyr was isotopiccomposition of rain and snow and thereforethe 5D inferredfrom "a paleontologicaltransfer function" [Bard et al., recordsin the ice caps and wood samples[Yapp and Epstein, 1987, pp. 792], as well as the 5180 values for the 1977]. Fairbanks [1989] (Figures 3 and 4) and Bard et al. foraminifera. However, it is difficult to imagine a rise of about [1987] studiedthe implication of the injection of meltwater to 7øC in the temperatureof the oceanin 200 years,from 12,460 the surfaceof the ocean and the ocean temperatureduring the to 12,26014C years B.P. [Bard et al., 1987],at thetime when Aller6d-Bolling-Younger Dryas periods. Here we attempt to global sea level apparentlyrose by some24 metersas a con- determineif we can createthe type of calculatedfid record sequenceof a rapid influx of glacialmeltwater [Fairbanks, shownin Figure 3a as a model for the DYE 3 ice core by using 1989]. The physicalbasis of Fairbanks'results seems to be a completelydifferent source of information,namely the 5180 sound, and his conclusions about times and rates of sea level record reported by Bard et al. [1987] for the pelagic rise are acceptedhere as valid in the subsequentdiscussion. foraminifera in a sedimentary core taken off the coast of If a rapid variation of the meltwater input into the North Portugal.We plotthe 5180 of Bardet al. [1987](Figure 4a) as Atlantic during the glacial-interglacialtransition (Figures. 1 a relationshipbetween the 5180 and 14C dates, rather than the and 4a) is the commoncause of the rapid and synchronousiso- variationof the 5180 recordwith depthin the core. The topicvariations in theGreenland DYE 3 ice core[Dansgaard et foraminifera5180 values before 10.5 Kyr (Figure 4a) appear to al., 1989] and in the planktonic foraminifera off Portugal vary synchronouslywith the fid recordsof theDYE 3 ice core [Bard eta/., 1987] (see Figures. 2 and 3a), the following can from Greenlandand the wood remainsfrom the north slope of reconcile the isotopic variations with material balance con- Alaska (Figure 1). Under normal conditions of a relatively straints. If a mixture of meltwater and cold surface ocean water constant/5180 of oceanwater, a lower/5180in themarine car- from the North Atlantic was injectedinto the surfaceocean water off Portugalat ~ 12.4 Kyr B.P. (Figure4a), it would have produceda new mixture. The coldwater mixture from the North A /" Atlantic(~0 ø C) couldhave had lower 5180 values, because of its meltwater component. The normally warmer (15øC) -240 [CLIMAPProject Members, 1976] and higher 51aO (~ 2.0%0) surface ocean waters in the midlatitudes would thus have been isotopicallyand thermallymodified by the introductionof the North Atlantic water. A value of 2.0%0is probably a reason- ICE able estimate of "normal" lower latitude surface ocean water o -260 51•O value,because of the 180-depletedice capsduring the Glacial Period as well as the normal excessevaporation of surface water at thoselatitudes [Sverdrup eta/., 1942]. Lower latitude surface ocean waters modified by the mixing would markedlyaffect the calculated temperature recorded in the 5180 -280 of the Portugueseforaminifera. The effect of suchmixing on the midlatitude surface oceans would have been to lower their \ / temperaturesfrom 15ø to 8øCand 5180 valuesfrom 2.0 to \,.// -0.6%0. We have basically selectedthe combinationof temperatureand the isotopiccomposition of water mixtureswhich I I B Figure 3. (a) The 5D recordmeasured in the DYE 3 ice core sample (solid line) comparedto that calculatedfor specified 14 variationsof the surfaceocean temperature(broken line) of the sourceof the moisture feeding the ice caps. (b) The presumed surfaceocean temperature in the sourceregion is shown. The temperatureof theice capand the 5D of theocean surface were assumedto be constantat -20øC and 0%0,respectively, for the generatedrecord. The differencein the curvesof Figure3a is dueto thefact that at 12.3Kyr B.P.,the 5180 of theocean which yieldedthe moisturethat actuallyaccumulated in the ice core (.qolidline) was probably about 2.0%o,and not 0ø/_oo,as assumedfor the calculatedline. This explains why the solid line begins at 12.3 Kyr B.P., with a 6D value 20%0 higher than that of the broken line at 12.3 Kyr B.P. The measured5iD value that is lower by 5%0than that calculatedat 12 Kyr B.P. is assumedto be a consequenceof the additionof cold water with a 5180value of approximately-3.5%0 which changed the 5180 10 11 12 of the surface ocean water from 2.0%0 at 12.3 Kyr B.P. to YEARSx 103 BP -0.6%0at 12 Kyr B.P. (see text). EPSTEIN: ISOTOPIC CLIMATE RECORDS IN YOUNGER DRYAS EVENTS 561

of temperatureTo injectedinto thatlocation. X + Y = 1, To < 15øC,and •5180 < -0.6%o.For T o = 0øCand •5180 of cold water equals-3.5%o, X = 0.47and Y = 0.53.For T o = 4.3øCand •5180 of cold water equals -2.0%o,X = 0.35 and Y = 0.65. In princi- • 12 ple, a large variety of mixtures of the two sourcesof water can

10 be usedto definethe temperatureof 8øCand the •5180of -0.6%o. We have defined reasonableconditions whereby a decrease 8 of temperatureof 7øC off Portugal at 12.4 Kyr B.P. could be recordedby the foraminiferawhose 5180 actuallychanged --1• I I ', from2.4 to 1.0%o,a 5180change which would normally be as- sociated with a rise in temperature. Following this cooling event at 12.4 Kyr B.P., a cutoff of the cold high-latitude water would allow the temperatureof the ocean surface off Portugal O- B to return fairly close to its presumedoriginal temperatureof approximately15øC (Figure 3c) andthe surface water 5180 to increase(Figure 3b). The 5180 of theforaminifera increases from +1.0%o at -12.3 Kyr B.P. to +1.7%o at 10.6 Kyr B.P. 1 (Figure3a). Asdeeper more 1SO-rich water mix,• withthe surfacewater, the/5180 of the waterhas to increasefrom -0.6 to +1.2%o to be consistentwith the model and the hypothesized rise in temperature. The second relatively rapid addition of 2- cold water began at -10.5 Kyr B.P. [Fairbanks, 1989]. The /5180 of the foraminifera at this time decreased from 1.7 to 0.8%oat about10 KyrB.P. Fora valueof/5180 = -3.5%oand a temperature= 0øC for the cold water injectedat 10.5 Kyr B.P., addedto waterwith a /5180value of 1.2%o(Figure 4b) anda lO 11 12 13 temperatureof 15øC (Figure 4c), this secondaddition of cold Years x lO3 B P meltwater would have lowered the surface ocean water tempera- tureoff thecoast of Portugalto about9 ø or 10øCand the 5180 Figure 4. Curve C representsthe temperatureof the surface value of ocean water to -0.8%0 correspondingto a foraminifera oceanshown in Figure 3b, which matchesquite well the tem- 5180 of +0.7%o. peraturerecord calculated from the difference between the •5180 A comparisonbetween Figures 3 and 4 illustrates that the recordof theforaminifera (curve A) andthe õ180 of thewater oceantemperatures used to calculatethe fid of the ice coresare (curve B) off the coast of Portugal. It was assumedthat the similar to those calculated from the foraminifera data, if the coolingof thissurface ocean water and increase in its 5180 at foregoing assumptionsare made regarding the effect of cold 12.3 Kyr B.P. resultedfrom the initial addition of cold oceanic water injection into the area of the foraminifera growth. The surfacewater from higher latitudes with a 5180of-3.5%o and a temperatureof 0øC. The subsequentincrease in the 5180 was gradual warming over the ice cap due to the combined forcing of the Milankovitch effects could cause a gradual increase in due to the shutoffof the cold high-latitudewater and remixing of the modified Portuguesewater with ocean water that had an the fid of the ice cap during the glacial-interglacialtransition. initial5180 of 2.0%oand temperature of 15øC. This remixing The large AllerOd-B01lingand YoungerDryas fid oscillations is thoughtto have yielded Portuguesesurface ocean water with in the ice cap record appear as extraordinary transient phe- a 5180of 1.2%oand a temperatureof about 15øC. nomena. This paper proposes that meltwater-inducedrapid changesin ocean surface temperaturesat the moisture source mixed with the warm surfacewater off Portugalto conform to can accountfor the ice cap fid oscillations. The similar fid the previouslypresented model (Figure 3a) which accountsfor record from the Alaskan samplescould be in responseto simi- the•SD record in theDYE 3 icecore and in thewood which grew lar variations in Pacific ocean temperature. in the north slope of Alaska. The same combinationwill also One of the apparent inconsistenciesin the curves labeled conform with our interpretationof the ocean surfacetempera- "ice cap" in Figure 3a is the difference between the calculated turerecorded by the•5180 record in thepelagic foraminifera •SDrecord obtained from the temperaturesof the ocean(broken [Bard et al., 1987]. A variety of mixturescould be injected line) and that measuredin the Greenlandice cap (solid line) for into the ocean water off the coast of Portugal which would the changeat ~ 12.4 Kyr B.P. The model oceanwater tempera- meet these conditions. turedecrease of 7øCat thattime yieldsa calculatedice core•SD Because the specific heat of this injected water and the increaseof 50%o,but the correspondingmeasured increase in oceanwater are very similar, the generalmaterial balanceequa- •SDon the ice cap is only 29%o. However, theseapparently tion applies to this case: inconsistentresults actually reinforce our scenariobecause the differencein the •SDrecords is whatis expectedat that timedue Fortemperature: 15.0(X) + TO (Y) = 8øC (1) to the proposedchanges in the fid valuesof the oceansurface Forisotopic of oceanwater: 2.0 (X) +/5180(D = -0.69'00(2) waters. Associatedwith an ocean temperaturedrop at 12.4 Kyr where X and Y are, respectively,the fractionsof the contribu- B.P. is the decreaseof 2.6%0 in the 5180 of the ocean surface tion of the water off the coastof Portugaland of the cold water water (equivalent to 21%oin the 5iD) (curve B in Figure 4). 562 EPSTEIN: ISOTOPIC CLIMATE RECORDS IN YOUNGER DRYAS EVENTS

Since the calculatedcurve of Figure 3a is basedon the assump- A comparisonof the •D recordsof Germanpine trees tion that the 5D of the ocean water is 0%0, the fact that the [Becker et al., 1991] with the Alaskan wood samplesand the measured5D is 21%ohigher than the calculated5D is dueto a DYE 3 ice core shows a relationship dissimilar to that 5D valueof +21%o(8 x 2.6 51SO)for theocean prior to 12.4 observedfor the •D values of the wood from the polar region Kyr B.P. For that5D valueof surfaceocean water, a decrease and the ice core samples.A õD increaseof 30%0in the pine in surfaceocean water temperature of 7øCwould cause a calcu- trees is correlated with a decrease in õD of 30%0 for the lated rise in the 5D on the ice cap from -270 to -220%0. Alaskan record, and a 25%0decrease for the DYE 3 record. By However, since the cooling of the ocean's surfaceis proposed thistime the õD or •lSO of theoceans were already quite simi- to have been causedby the injectionof low- õD cold water lar to modem values, and Becker et al. [1991] presentevidence whichresulted in a decreaseof the õD of the Portuguesesurface that their •D values are directly related to continentalwarm- oceanwater of about 21%o,the 5D of the ice on the ice cap ing. would also decreaseby about 20%0,resulting in the observed As we have already pointed out, a warming ocean surface •D value of about -240%0, instead of-220%0. couldresult in a correspondinglowering of the õD valuesin Up to this point, we have confinedour discussionsto the the polarregions as the cold water injectionis shutoff at -10 isotoperecords between 12.4 and 10 Kyr B.P. This wasdue to Ky• B.P. The questionthen is, wasthere a smallinjection of the absenceof convincing evidence that rapid injection of coldwater at -9.5 Kyr B.P., as indicatedby the rise in •D at cold water, followed by a cutoff of the cold water, occurreddur- thispoint in time in the Alaskanwood samples and in the DYE ing the period of 10 to 8.8 Kyr B.P. We wish to includethe 3 ice core? Such an injection of cold polar water was not isotope data to suggestfurther investigationsregarding the obvious in the data of Fairbanks [Fairbanks, 1989] or in the isotopicand otherclimatic records for this period(Figure 5). õ1•Odata for thepelagic foraminifera reported by Bardet al. [1987]. However, in the case of the foraminifera data, there -60 areonly two data points for the õ1•O values for the time period 10 to 8.8 Kyr B.P. It would be interestingto invest additional effortin obtainingmore detailed records for 14C-datedtree rope samplesand foraminifera for possible identification of the -80 - presenceof periodicinjections of cold meltwaterfrom the ice caps. Althoughthere might be small magnitude,short-term fluctuations in the temperatureof the surface ocean waters, over -100 - the longer term the surface ocean temperaturesmay remain I I • stable (within a few degrees)or changegradually in response -4o - B to worldwide climate changes. At such times, corresponding temperature changes will occur in the ice caps as well. Consequently,the õD recordsin the ice cap shouldusually Alaska reflect primarily the air temperatureover the ice. This empha- sizes the point that the model presentedhere concerningsur- -160 face ocean temperaturesis for the specialcircumstances asso- ciated with a glacial to interglacial climatic transitionas rep- resented by the Bolling-Younger Dryas and post-Younger Dryas oscillations. On the other hand, if the ice recordscon- -180 - tain relatively rapid fluctuations in their isotopic composi- tions, it is relevant to consider the possible role that ocean temperaturesmight play in determiningthese records. It is Dye 3/ obviously important to complement this isotopic record with -200- - 245- / \ / temperature-sensitivefid recordsof differentmaterials found at \ / other locations on the continents. These will be critical to the \ / understandingof the climate dynamicsresponsible for the his- - -255 - •_../ tory of the climate on Earth. The impact of the isotopic -220 I ] • recordsof the few nonglacialsources used in this paper illus- 8 9 10 11 trateshow importantthese records can be in providing infor- Years x103 BP mation about past climates. Figure 5. Comparisonof the õD recordsof the !4C-dated Alaskan wood samples, the 14 C-dated German pine samples, Summary and the DYE 3 ice core samplesbetween the period of 10 to 8.8 In spiteof the uncertaintiesassociated with determiningthe Kyr B.P. 'l'-he14 C datesfor the Germanpine samples were exact simultaneityof eventsrecorded in thesedifferent, widely estimated from the relationship between the absolute and the 14Cdates by Bardeta/. [1990],using the Becker absolute tree separatedsystems, the temporalcorrespondence of the large ring chronology[Becker et al., 1991]. The õD recordfor the isotopicvariations between - 10 and 12.5 Kyr B.P. is striking DYE 3 ice core samplesis presentedin this figure only for the and has importantimplications. The õD data on the wood purposeof the •D comparisonwith that of the post-Younger samplesfrom the north slope of Alaska suggestthat the Dryas event recordedin the Alaskan and Germandata. Aller6d-Bollingand YoungerDryas were widespreadevents in EPSTEIN:ISOTOPIC CLIMA• RECORDSIN YOUNGERDRYAS EVENTS 563

the northern hemisphere. The combinationof isotope data Bard, E., B. Hamelin, R.G. Fairbanks,and A. Zindler, Calibration of the from the Alaskan wood, the Greenlandice cap and pelagic 14Ctimescale over the past 30,000 years using mass spectrometric U- foraminiferawith sea level data from Barbadossuggests that Th agesfrom Barbadoscorals, Nature, 345, 405-410, 1990. Becker, B., B. Kromer, and P. Trimborn, A stable-isotopetree-ring the Allerfd-B011ingevent was initiated by rapid cooling, and timescaleof the Late Glacial/Holoceneboundary, Nature, 353, 647- the YoungerDryas event by a rapid warmingin the midlatitude 649, 1991. North Atlantic Ocean and probablyin part of the North Pacific Broecker,W.S., and G.H. Denton, What drives glacial cycles?,Sci. Am• Ocean. These proposedtemperature changes could be due to 262 (1), 48- 56, 1990. Broecker, W.S., J.P. Kennett, B.P. Hower, J.T. Teller, S. Trumbore, G. rapid additionand subsequentshutoff of meltwater-augmented, Bonani, and W. Wolfli, Routing of meltwater from the Laurentide high-latitudesurface ocean waters. Consequently,the large, Ice Sheetduring the Younger Dryas cold episode,Nature, 341, 318- rapid isotopic oscillations that were recorded on the north 321, 1989. •-•,t-...... ro during ...... - Climate:Long-Range Investigation. Ma.•Ding. and Prediction (CLIMAP) ProjectMembers, The surfaceof the Ice-Age Earth, Science,191, Interglacialtransition need not necessarilybe associatedwith 1131-1137, 1976. temperaturevariations at theselocations. Isotope analysesof Dansgaard,W., J.W.C. White, and S.J. Johnsen,The abrupttermination accurately14C-dated wood samples from variouscountries of the YoungerDryas climate event, Nature, 339, 532-534, 1989. would provide critical informationabout the climatic history Epstein,S.; C.J. Yapp, andJ.H. Hall, The determinationof the D/H ratio of theseareas during the Allertd-B011ing-YoungerDryas and of non-exchangeablehydrogen in celluloseextracted from aquatic andland plants,Earth Planet. ScœLett., 30, 241-251, 1976. post-YoungerDryas events. Outsideof the time spancovering Fairbanks, R.G., A 17,000-year glacio-eustatic sea level record: these events, the general ice core isotopic record for the tran- Influenceof glacial melting rates on the Younger Dryas event and sitionbetween the Glacial and Interglacialperiods could be due deep-oceancirculation, Nature, 342, 637-642, 1989. mainly to warmingof the air over the ice caps,if the tempera- Petit, J.R., J.W.C. White, N.W. Young, J. Jouzel,and Y.S. Korotkevich, Deuterium excess in recent Antarctic snow, J. Geophys. Res., 96, ture of the ocean sourcefor the moisturefeeding the ice caps 5113-5122, 1991. remained relatively constant. Sverdup,H.U., M.W. Johnson,and R.H. Heming, The Ocean, Their Physics,Chemistry and General Biology, Prentice-Hall,Inc., New Acknowledgments.It is a pleasuretoacknowledge David Carter, York, 1942. USGSAlaska Branch, for providingthe inc -datedsamples from the White, J.W.C., J.R. Lawrence, and W.S. Broecker,Modeling and inter- northslope of Alaska and for his helpful remarkson earlier versionsof preting D/H ratios in tree rings: A test case of white pine in the this manuscript. Crayton Yapp made critical contributionsto this northeasternUnited States,Geochim• Cosmoctu'm• Acta, 58, 851-862, manuscriptwith his cogentand greatly appreciatedremarks. We also 1994. acknowledgethe technicalassistance of EleanorDent, XiahongFeng, Yapp,C.J., and S. Epstein,Climatic implicationsof D/H ratiosof mete- XiaomeiXu, and Meg Garstang.This researchwas supported by NSF oric water over North America (9500-22,000 B.P.) as inferred from ATM9219891 to SamuelEpstein. ancient wood cellulose C-H hydrogen.Earth Planet. Sci. Lett., 34, 333-350, 1977.

References S. Epstein,Division of Geologicaland Planetary Sciences, 170-25, California Instituteof Technology,Pasadena, CA. 91125. (e-mail' Bard, E., M. Arnold, P. Maurice, J. Duprat, J. Moyes, and J.-C. epstein@ arms. gps. caltech. edu) Duplessy,Retreat velocity of the NorthAtlantic polar front during thelast deglaciation determined by 14Caccelerator mass spectrom- (ReceivedJuly 23, 1995' revisedAugust 30, 1995; etry, Nature, 328, 791-794, 1987. acceptedSeptember 5, 1995.)