Earth and Planetary Science Letters 518 (2019) 67–75
Contents lists available atScienceDirect
Earth and Planetary Science Letters
www.elsevier.com/locate/epsl
Deep ocean 14 C ventilation age reconstructions from the Arctic Mediterranean reassessed
Mohamed M. Ezat
a,b,c,∗, Tine L. Rasmussen
a, Luke C. Skinner
b, Katarzyna Zamelczyk
aaCAGE–CentreforArcticGasHydrate,EnvironmentandClimate,DepartmentofGeosciences,UiT,TheArcticUniversityofNorway,Norway bGodwinLaboratoryforPalaeoclimateResearch,DepartmentofEarthSciences,UniversityofCambridge,UK
cDepartmentofGeology,FacultyofScience,Beni-SuefUniversity,Beni-Suef,Egypt
a r t i c l e i n f o a b s t ra c t
Articlehistory:
Received16July2018
Receivedinrevisedform19February2019 Accepted17April2019
Availableonline13May2019 Editor: J.Adkins
Keywords:
FramStrait NordicSeas ArcticOcean oceancirculation radiocarbon lastglacialclimate
The present-day ocean ventilation in the Arctic Mediterranean (Nordic Seas and Arctic Ocean), via transformationofnorthwardinflowingwarmAtlanticsurfacewaterintocolddeepwater,affectsregional climate, atmospheric circulation and carbon storage in the deep ocean. Here we study the glacial evolutionoftheArcticMediterraneancirculationanditsinfluenceonglacialclimateusingradiocarbon reservoir-agereconstructionsondeep-seacoresfromtheFramStrait thatcoverthelateglacialperiod (33,000–20,000 yr ago;33–20 ka).OurresultsshowhighBenthic-Planktic14Cagedifferencesof∼1500
14C years 33–26.5 kasuggesting significantwater columnstratification between∼100–2600 mwater depth, and reduction and/or shoaling of deep-water formation. This phase was followed by break- up ofthestratification duringthe LastGlacialMaximum (LGM;26–20 ka),with Benthic-Planktic14C agedifferencesof∼25014C years,likelyduetoenhancedupwelling. Theseocean circulationchanges potentiallycontributedtothefinalintensificationphaseofglaciationviapositivecryosphere-atmosphere- ocean circulation-carboncycle feedbacks.Ourdata alsodo not support‘extremeaging’of >600014C yearsinthedeepArcticMediterranean,andappeartoruleouttheproposedoutflowofveryoldArctic Oceanwatertothe NordicSeasduringthe LGMandto thesubpolarNorthAtlanticOceanduringthe deglacialperiod.
©2019TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
TheAtlanticMeridionalOverturningCirculation(AMOC)playsa keyroleintheEarth’sclimatesystemthroughitscontrolonheat, freshwater, nutrient and greenhouse gas transports (Ganachaud and Wunsch, 2000; Rahmstorf, 2000; Takahashi et al., 2009).
At present, the northern limb of the AMOC constitutes water masstransformationofnorthwardinflowingwarmAtlanticsurface water to cold deep water through local overturning. This well- ventilated deep waterin the ArcticMediterranean has a ventila- tionageof∼500 yr,whichisclosetothesurfacewaterreservoir age of ∼400 yr (e.g., Hansen and Østerhus, 2000; Broecker and Peng,1982; Mangerudetal., 2006). Thewell-ventilateddeepwa- teroverflowstheGreenland-ScotlandRidgeintotheNorthAtlantic Ocean,formingamajorpartofthelowerNorthAtlanticDeepWa- ter(NADW)(e.g.,HansenandØsterhus,2000).
*
Correspondingauthor at:GodwinLaboratory forPalaeoclimateResearch, De- partmentofEarthSciences,UniversityofCambridge,UK.E-mailaddresses:me416@cam.ac.uk,mohamed.ezat@uit.no(M.M. Ezat).
Paleoclimate research seeks a better understanding of AMOC- climate interactions through the study of the behavior of the AMOC and its controlling mechanisms under different climate statesandondifferenttime scales(e.g.,Clarketal.,2002; Skinner etal., 2014; Pena andGoldstein, 2014). In thisregard, thestruc- ture and vigor of the AMOC during the Last Glacial Maximum (LGM; 26–19 ka, e.g., Clark et al., 2009) have been intensively investigated (e.g., Sarnthein et al., 1994; Curry and Oppo, 2005;
Otto-Bliesneretal.,2007; KeigwinandSwift,2017).However,sev- eralaspectsoftheAMOCoperationduringtheLGMremainlargely unconstrained. For example, contrasting scenarios of deep ocean circulationintheArcticMediterraneanduringtheLGMhavebeen proposed,rangingfromnear-cessationtovigorous,aspresent-day, ocean ventilation (e.g., Veum et al., 1992; Bauch et al., 2001;
Hoffmann et al., 2013; Thornalley et al., 2015; Meinhardtet al., 2016).Manypaleoceanographicstudies ofthisissuearebasedon radiocarbondating ofmarinecarbonates(e.g.,foraminifera,corals and mollusks) to infer past changes in ocean ventilation, based on the premise that these marine carbonates record the radio- carbon activity of ambient seawater (e.g., Broecker et al., 2004;
Adkins and Boyle, 1997). Radiocarbon age offsets between con- https://doi.org/10.1016/j.epsl.2019.04.027
0012-821X/©2019TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Fig. 1. Majorsurfaceanddeep-watercurrentsintheNordicSeas(e.g.,HansenandØsterhus,2000)andsedimentcorelocations.Yellowstarreferstolocationofinvestigated sedimentcoresHH12-946MCandHH12-948MC(thisstudy)andsedimentcoreJM06-16MCpublishedbyZamelczyketal. (2012).Whitecirclesrefertoothercoresites discussedinthetext.MapismodifiedfromEzatetal. (2014,2017).(Forinterpretationofthecolorsinthefigure(s),thereaderisreferredtothewebversionofthisarticle.)
temporarybenthic (B)andplanktic(P) foraminifera(B-P age dif- ferences) have been used as a proxy for ocean mixing between (near)surfaceandbottom waters(e.g.,Broecker etal.,2004).This ventilationproxyis bestapplied in regions wherelocalized deep mixingexertsasignificantinfluenceonlocaldeep-waterradiocar- bonactivity.When independentcalendar agesformarinerecords orknowledge aboutpastsurface reservoirages are available,B-P age differences can be converted into Benthic-Atmosphere (here referredtoasB-A)agedifferences.Thisallowsfordeep-oceanra- diocarbon activities from contrasting hydrographic regimes to be correlatedtoacommonatmosphericreferenceandprovideamea- sureofthelocaldeepocean-atmosphereradiocarbonisotopedis- equilibrium (e.g.,Skinner etal., 2010; for details about different methodsforreconstructing14C-ventilationagesseeCookandKeig- win,2015).
Benthic-atmosphere 14C ventilationreconstructionsfrom 2711 m water depth in the northern Norwegian Sea have been em- ployed to suggest an extreme aging (up to 10,000 14C years) and near-isolation of the deep Arctic Mediterranean during the LGM(Thornalleyetal.,2015).EpisodicoverflowsofthisagedArc- tic/Nordic Seas reservoir into the subpolar North Atlantic across the Greenland-Scotland Ridgemay haveoccurred during the last deglaciation(Thornalleyetal.,2011,2015;Ezatetal.,2017).How- ever,thisscenarioisinconsistentwithreconstructionsbasedonNd isotopeandPa/Thproxies that suggestvigorous ocean circulation intheArcticOceanduringthelateglacialperiod(Hoffmannetal., 2013; Meinhardtetal.,2016).
In thisstudy, we investigateventilation andocean circulation changesinthecentralFramStrait leadingtoandduring theLGM
relatedtoclimatevariabilityusing14Cdatesonbenthicandplank- tic foraminiferal species. Thornalley et al. (2015) suggested that theinferredextremelyagedwaterinthedeepNorwegianSeadur- ingtheLGMwassourcedfromthecentralArcticOcean.TheFram Strait,theonlydeepgatewayforexchangeofdeepwaterbetween the Arctic Ocean andthe Nordic Seas,must therefore record the signature ofthis hypothesizedArctic outflow tothe Nordic Seas.
ThismakestheFramStraitanideallocationtotestthehypothesis ofexistenceofanArcticdeep-waterendmemberwithaventilation age>6000 yr anditssubsequentinflowtotheNordicSeasduring theLGM.
2. Materialsandmethods
Two deep-sea multicores HH12-946MC (78◦53N; 01◦45W;
2637 m waterdepth, 42cmlength)and HH12-948MC(78◦52N;
00◦21E;2542 mwaterdepth,44.5 cmlength)wereretrievedfrom thecentralFramStraitduringacruisewithR/VHelmerHanssenin July 2012(Fig.1). Thetwo sediment coreshavebeensampledat 0.5 cmintervalson-board.Thesampleswereweighedandfreeze- dried.Subsequently,thedrysampleswereweighedandwetsieved through1000,500, 100and63 μmmesh-sizes.Theresidueswere dried and weighed. More than 300 benthic foraminiferal speci- mensincoreHH12-946MCfromthesizefraction>100 μmwere counted and identified to species level. The number of benthic foraminifera in the deglacial part of the record was too low for quantification.
Forstablecarbonandoxygenisotope analyses∼30specimens ofthe plankticforaminiferal speciesNeogloboquadrinapachyderma
M.M. Ezat et al. / Earth and Planetary Science Letters 518 (2019) 67–75 69
Table 1
Benthicandplanktic 14CdatafromsedimentcoreHH12-946MC.Theδ13Cwasmeasuredbyagassourcemassspectrometer.Ben- thic14C-datesfromtopcentimeterofthecoreyieldedtooyoungagesandarenotplottedinFig.4.Abbreviations:Neogloboquadrina pachyderma(N.p),Oridorsalisumbonatus(O.u),Pyrgodepressa(P.d),Benthic-planktic14Cageoffset(B-Poffset),MarineIsotopeStage (MIS).Asterisksreferto14Cdatesgraphitizedatthe14ChronoCentre,Queen’sUniversityBelfast,whereasallother14Cdateswere graphitizedattheUniversityofCambridge.Allradiocarbondateswereanalyzedatthe14ChronoCentre,Queen’sUniversityBelfast.
Time interval Depth (cm)
Calendarage (ka)
Species 14Cage (yr)
Error (yr)
B-Poffset (yr)
δ13C (h)
Mid 1.25 4.9 N. p 4667 28 0.06
Holocene 1.25 4.9 O. u 2884 30 −1783 −0.42
1.25 4.9 P. d 1362 28 −3305 −0.43
1.25 4.9 P. d 1297 28 −3370
3.75 6.0 N. p 5620∗ 30 −0.01
3.75 6.0 O. u 5835∗ 30 215 −0.8
Early 24 21.0 O. u 18335 80 367 −1.7
MIS 2 24.25 21.2 N. p 17968 80 −0.78
25.25 21.4 N. p 18280∗ 75 −0.64
25.25 21.4 N. p 18075 90
25.25 21.4 O. u 18580 88 505 −1.9
25.75 21.8 N. p 18399 75
25.75 21.8 O. u 18511 82 112 −1.83
27.25 22.3 N. p 18765 73 −0.64
27.25 22.3 O. u 18787 95 22 −1.92
28.25 22.8 N. p 19335 77 −0.58
28.25 22.8 O. u 19232 148 −103 −1.86
Late 30.75 26.1 N. p 22372 108 −0.88
MIS 3 30.75 26.1 P. d 32400 329 10028 1.04
31.25 26.4 P. d 33434∗ 459 10711 0.72
32.75 27.1 N. p 23016∗ 140 −0.61
32.75 27.1 N. p 23076 131
32.75 27.1 O. u 25327 161 2251 −1.49
32.75 27.1 P. d 31975 312 8899 0.71
33.75 28.4 N. p 24741 134 −0.83
33.75 28.4 O. u 26118 170 1377 −1.35
34.75 30.0 P. d 38639∗ 832 12102
35.75 31.6 N. p 28333 225 −0.37
35.75 31.6 O. u 29715 340 1382 −1.21
37.25 33.4 N. p 29497∗ 274 −0.41
37.25 33.4 P. d 40749∗ 1075 11252 1.1
(size fraction 125–150 μm), ∼20 specimens of the shallow in- faunal benthic foraminifera Oridorsalis umbonatus (size fraction 150–250 μm),∼5specimensoftheepifaunalbenthicforaminifera Cibicidoideswuellerstorfi (size fraction 250–500 μm) and 1 speci- menoftheinfaunalbenthicforaminiferaPyrgodepressa(sizefrac- tion500–1000 μm)werepicked.Theplankticstableisotopeswere measuredatLeibnizLaboratoryforRadiometricDatingandStable IsotopeResearchinKiel,Germany,whilethebenthic(O.umbonatus andC.wuellerstorfi)stableisotopesweremeasured attheGodwin LaboratoryforPaleoclimateResearch,UniversityofCambridge,UK forsedimentcoreHH12-946MCandatUiT,theArcticUniversityof Norwayforsediment coreHH12-948MC. Stable isotope measure- mentsin P.depressawere performedatUiT, theArctic University ofNorway.Manystudiesaccountforthevitaleffects(deviationof foraminiferalδ18Ofrom‘equilibrium’values)usingspecies-specific corrections(e.g.,Bauchetal.,2001).Inthisstudy,wedidnotapply anycorrectionsbecausewe onlydiscussrelative changesin δ18O records(inparticularδ18OdifferencesbetweenN.pachydermaand O.umbonatus),which are notaffected by corrections. Inaddition, correctionsforN.pachydermavaryfrom0.0to−1.6hindifferent studies(seee.g.,Padosetal.,2015).
WepickedsamplesofO.umbonatus(sizefraction150–250 μm;
2to 3 mg carbonate),N.pachyderma (sizefraction 250–500 μm;
3 to5mgcarbonate)andwhereavailable,mixedspeciesofgenus Pyrgo (mainly Pyrgodepressa; size fraction >500 μm) from sed- iment core HH12-946MC for radiocarbon dating. All specimens were carefully inspected under the microscope in order to in- spectforanysignsofcontamination andonlypristine specimens wereincludedintheanalyses.Foraminiferalsamplesweregraphi- tized at the University of Cambridge (see Freeman et al., 2015
fordetailed methodology) andsubsequently analyzed by acceler- ator mass spectrometry atthe 14Chrono Centre, Queen’s Univer- sityBelfast,NorthernIreland.Some additionalsampleswereboth graphitized and analyzed at the 14Chrono Centre, Queen’s Uni- versity Belfast (Table 1), and two new samples (Pyrgorotalaria andN.pachyderma)fromsedimentcoreLINK15(61◦45N;2◦40W, 1602 m water depth; Olsen et al., 2014) were also graphitized and analyzed there. No pretreatment was applied for the sam- ples graphitizedattheUniversity ofCambridge,whereas samples graphitizedattheQueen’sUniversityofBelfastwerecleanedprior to the graphitization using 2 ml 15% hydrogen peroxide (with ultra-sonication for 3 min). Samples with and without pretreat- ment yieldedindistinguishableresults(Table 1). Wealso crushed thespecimensofPyrgo tochecktheinside oftheshellsforpres- enceof diageneticcarbonatefillings inside different chambersas in Keigwin (1979). All specimens analyzed were pristine alsoon theinsideoftheshells.
2.1. Agemodels
WeconstructedtheagemodelsofsedimentcoresHH12-946MC and HH12-948MC based on twelve and four calibrated planktic foraminiferal14Cdates,respectively(Fig.2).Theplankticradiocar- bondateswerecalibratedusingtheCalib7.04programandtheMa- rine13 calibrationdataset (Reimeretal., 2013). We implemented the reservoir age correction of405 yr inherent in the Calib pro- gram (StuiverandReimer, 1993),whichisclosetomodern reser- voirageofthesurface oceanintheopenNordicSeasof∼400 yr (e.g.,Mangerudetal.,2006).Thesurfacereservoiragesvarieddur- ing the glacial period due to changes in ocean circulation and
Fig. 2. AgemodelsforsedimentcoresHH12-946MC(A)andHH12-948MC(B)based oncalibrated 14CdatesmeasuredinplankticforaminiferalspeciesNeogloboquad- rinapachyderma.Sedimentationratesareshownatthebottomofthefigurepanels.
Errorbarsassociatedwithradiocarbondatesrepresentonestandarddeviation.
carboncycle(e.g.,Bard,1988).Welackinformationabouttemporal changesinreservoiragesatthecentralFramStraitduringthelast glacial.However,wediscussindetailtheeffectsofpastchangesin reservoiragesonourresultsaswellasontheagemodels(seeDis- cussion).Followingthesameapproach,were-calibratedpreviously published planktic14C dates fromsediment coresMSM5/5-712-2 (78◦55N; 6◦46E; 1487 m water depth; Zamelczyk et al., 2014;
Müller andStein, 2014), JM06-16MC (78◦54N; 01◦69E, 2546 m waterdepth; Zamelczyketal.,2012), NP90-39(77◦16N;09◦06E, 2119mwaterdepth;Hebbelnetal.,1994; DokkenandHald,1996) andPS1243(69◦37N;06◦55W,2711 mwaterdepth;Bauchetal., 2001; Thornalley et al., 2015). Accordingly, the age models were re-calculatedtoobtaincomparabletime-scaleswithoursediment coresHH12-946MCandHH12-948MCignoringpossible,butprob- ablysmall,spatialvariabilitiesinreservoiragesatagiventimefor thesenearbycoresites.Theoriginallypublishedagemodelsforthe timeintervalsofinterestinthisstudy(seesection 2.2) arebased onplankticforaminiferal14Cdates exceptforcorePS1243, which is basedon alignment of marine plankticforaminiferal δ18Oand Greenlandicecoreδ18O(Thornalleyetal.,2015).Thus,themodi- ficationtotheseagemodels,exceptforcorePS1243,onlyincludes theuse oftheradiocarbon calibrationof Reimeretal. (2013) in- stead ofusing no(e.g., Hebbeln etal., 1994) orother calibration curves(e.g., Müller andStein, 2014). We also discussthe poten- tialimpact of usingthe age modelfromThornalley etal. (2015) (seeSupplementary Fig. 2).Thesedimentationratesinbothcores rangefrom1to3 cm/kyr, exceptforthe interval19.7–19.5 ka in coreHH12-946MC,wherethecalculatedsedimentationis6cm/kyr (Fig.2).Thus, a0.5 cmsamplespans ∼300–500 yr,whichpoten- tiallycaninducealargeerroron14Cdates(e.g.,PengandBroecker,
Fig. 3. OxygenisotoperecordsfromtheFramStraitandthecentralNordicSeas.
ForaminiferalspeciesusedareNeogloboquadrinapachyderma(N.p;red),Oridorsalis umbonatus(O.u;black)andCibicidoideswuellerstorfi(C.w;blue).
1984). However, because the distribution patterns ofthe benthic foraminiferal speciesO.umbonatus,C.wuellerstorfi and Pyrgospp.
showsomeabruptchanges(SupplementaryFig. 1)andbecausethe 14CdatesofN.pachydermaandO.umbonatusare inchronological order(Table1),weassumetheerrorisinsignificant.
2.2. Selectionofsedimentcores,foraminiferalspeciesandtime-slices
Theδ18OrecordsofthreesedimentcoresfromthecentralFram Strait (Zamelczyk etal.,2012,2016andthisstudy)andone sed- iment corefrom thecentral Nordic Seas (Bauchet al., 2001) are compared(Fig.3).Allplanktic,epifaunalandinfaunalbenthicδ18O records from both areas display the same glacial to interglacial variability(Fig.3).Bauchetal. (2001) observedthattheδ18Omea- suredintheepifaunalbenthicspeciesC.wuellerstorfifromtheLGM sedimentsin thenorthernNorwegian Seaare lowercompared to theHoloceneincontrasttotheinfaunalbenthicδ18Orecords.The same feature is found in the two sediment cores fromthe cen- tral Fram Strait with average LGM δ18O of ∼3.4h and ∼4.8h for C. wuellerstorfi and O.umbonatus, respectively, whereas both speciesshowsimilarHoloceneδ18Oof∼4.1h (Fig.3).At2781 m water depth in the southern Norwegian Sea (Veum etal., 1992) both C. wuellerstorfi andO.umbonatus show the same LGM δ18O (=∼4.8h) as recorded in our records by O.umbonatus (Fig. 3).
The ‘peculiar’ low LGM δ18O in C. wuellerstorfi is therefore re- stricted to the deep central and northern Nordic Seas (Bauch et al., 2001; this study). Deep-sea records from the central Arctic Ocean do not contain LGM sediments (e.g., Hanslik et al., 2010) andthuswe arenot abletotrace thenorthernextentofthelow C. wuellerstorfi δ18O. The relative abundance of C. wuellerstorfi in the LGM sediment is very low (SupplementaryFig. 1) with only
∼7 specimens per sample on average. Low δ18O in C.wueller- storfihasbeenexplainedbyinfluenceofanephemeralwatermass
M.M. Ezat et al. / Earth and Planetary Science Letters 518 (2019) 67–75 71
withverylowδ18O,thoughthepossibilityofunknownC.wueller- storfi-specific‘vitaleffects’occurringunderLGMconditionscannot be excluded (Bauch et al.,2001). We are not ableto assessthese differentexplanationsandwesuggestthatfutureworktakeadvan- tageof recentlydeveloped radiocarbonanalytical techniques that arecapableofdatingafew hundredsofmicrogramsofcarbonate (e.g.,Gottschalk etal., 2018) to ensure that these C.wuellerstorfi with low δ18O are of LGM age and not transported specimens orreworked. We emphasize that the ‘peculiar’ low LGMδ18Oof C.wuellerstorfi are consistent betweenrecords in thecentral and northernNordicSeas(Fig.3),whichmeritsfurtherinvestigation.
Similar to the central Nordic Seas (Bauch et al., 2001), our recordsshow thatthebenthicforaminiferalassemblagesfromthe deepFramStraitare dominatedalmostexclusively byO.umbona- tusduringtheLGM(SupplementaryFig. 1).Wethereforefocusthe discussiononO.umbonatusasthisspeciesismostlikelytorecord theaveragelong-termbottomwaterconditionsduringtheLGMin thedeepFramStraitandthenorthern/centralNordicSeas(seealso Bauchetal.,2001).
We mainlybase the reconstruction of thepast ocean ventila- tion in the Fram Strait on sediment core HH12-946MC, because itisthe longestrecord (spanningthepast ∼40kyr)ofourthree sedimentcores.Inthisstudy,wefocusonthepast33kyr.Thesed- imentsdating fromthe deglaciation >2500 mwater depth from theFramstraitinbothcoresHH12-946MCandHH12-948MCcon- tain only a few calcareous and agglutinated foraminifera or are completelybarren(Fig.3).ThiswasalsofoundbyZamelczyketal.
(2012) innearbycoreJM06-16MC(Fig.3).Whethertheabsenceof foraminifera resulted fromunfavorable environmental conditions, suchasextremelyhighsalinity,lowoxygenorlowfoodsupply,or frompost-mortemdissolutionrequiresfurtherinvestigation.How- ever,we notethat thisfeaturehas notbeenrecorded fromother deep areas (>2500 m water depth) further south in the Nordic Seas(e.g.,Bauchetal.,2001; Thornalleyetal.,2015),orfromshal- lower areas in the Fram Strait (Bauch et al., 2001). In deep-sea records from the central Arctic Ocean, sediments from the late deglaciation (∼13.5–10 ka) also contain calcareous foraminifera (e.g.,Hansliketal.,2010).Thus, itseemsthat thisfeaturewasre- strictedto thedeep central FramStrait. Futurework focusing on biomarkersandsediment provenancemayhelp toclarify theori- ginofthislargelycarbonate/foraminiferafreesedimentinterval.In this study, we therefore focus on three time slices: late Marine IsotopeStage3(MIS3)(33–26 ka),LGM(26–20 ka)andearly/mid Holocene(10–5 ka).
3. Resultsanddiscussion
3.1.OceancirculationchangesintheFramStraitleadingtoandduring theLGM
The14C agedifferences betweenthe co-existingbenthicfora- miniferalspecies O.umbonatus andthe plankticspeciesN.pachy- derma (BOu-PNp age difference) range from ∼2200 to 1400 14C years during late MIS 3 (∼33–26.5 ka) (Fig. 4e). These BOu-PNp agedifferencesareanorderofmagnitudehigherthanthemodern deeptosurface water14Cagedifference inthe region(∼100 yr) indicatinga strong stratification ofthe watercolumnduring late MIS3andpointingtoreduceddeepwaterformationintheNordic Seas.However,presenceofanactive,butshallowerconvectioncell cannot be ruled out. The absence of independent constraints on thecalendaragechronologyandsurfacereservoirageoffsetspre- ventsan exact estimate ofthe ventilationage ofthe deepwater (i.e., B-A 14C ventilation age). If near-surface reservoir age esti- matesfrom thecentral Nordic Seas(∼400–650 yr; Thornalley et al., 2015; see Fig. 4d) are assumed to apply, the B-A age differ- encein the deepFram Strait would have been ∼2000 14C years
Fig. 4. OceancirculationandprimaryproductivityreconstructionsfromtheFram Strait. (A) Concentration of plankticforaminifera from sediment core NP90-39 (Hebbeln et al., 1994). (B) C25 isoprenoid lipid (IP25) from sediment core MSM5/5-712-2(MüllerandStein,2014).HighconcentrationofIP25suggestspres- enceofseasonalseaice,whereasabsenceofIP25suggestseitherpermanentsea- ice cover (when the concentration ofphytoplankton-induced sterols is low)or openoceanconditions(whentheconcentrationofphytoplankton-inducedsterols ishigh) (seeMüller and Stein,2014fordetails).(C)Concentration ofdinosterol fromsedimentcoreMSM5/5-712-2(MüllerandStein,2014).(D)Near-surfacereser- voiragesfromsedimentcorePS1243(Bauchetal.,2001; Thornalleyetal.,2015;
seeSupplementaryFig. 2).(E)Benthic-Planktic 14Cagedifferencefromsediment core HH12-946MC.Error barsare combined1σ errors inplankticand benthic 14Cdates.(F)δ13CmeasuredinOridorsalisumbonatus(black)andNeogloboquadrina pachyderma(red)fromsedimentcoreHH12-946MC.(G)δ18OmeasuredinOridor- salisumbonatus(black)andNeogloboquadrinapachyderma(red)fromsedimentcore HH12-946MC(Zamelczyketal.,2016andthisstudy).
during late MIS 3. TheseB-A age difference estimates should be takenwithagreatcaution,becauseofthelargedistancebetween thetwo areas(Fig. 1)andthelarge uncertainties associatedwith the surfacereservoir ages inThornalleyetal. (2015). It hasbeen suggestedthat surface advectionof Atlanticwater tothe eastern Fram Strait took place during late MIS 3 based on high abun- danceofplankticforaminifera (Hebbelnetal., 1994; Dokkenand Hald,1996;seeFig.4a).Studiesbasedonphytoplankton-produced sterolsandC25isoprenoidlipid(IP25)fromtheeasternFramStrait however, suggest that perennial seaice conditionsgenerallypre- dominatedintheperiod∼30–26.5 ka(e.g.,MüllerandStein,2014;
see Fig. 4b, c). It is therefore possible that the Fram Strait dur- ing late MIS 3 was similar tothe Arctic Ocean today, wherethe seaicecoverisinsulatedfromsubsurfaceAtlanticwaterbyacold halocline(e.g.,Aagaardetal.,1985),butwithsignificantlyreduced deep-waterformation.
At the transition from MIS 3 to LGM at ∼26 ka, all proxy recordsshowdistinctchanges(Fig.4).TheaverageBOu-PNpagedif- ferencedecreasesfrom∼150014CyearsinlateMIS3(32–26.5 ka) to∼25014C yearsduring thelate LGM(23–20 ka)(Fig. 4e). This isassociatedwithanincrease inthedifference betweenN.pachy- dermaδ13CandOumbonatus δ13C(PNp-BOu δ13C)andadecrease in the difference between O.umbonatus δ18Oand N.pachyderma δ18O(BOu-PNp δ18O) (Fig. 4f, g); indicative of the biological and physicalstateintheFramStraitduringtheLGM.Weinterpretthe decrease in BOu-PNp age difference and the decrease in BOu-PNp
δ18Oasreductioninstratificationandamorehomogeneous water column.The increase inPNp-BOu δ13C isindicativeofpresence of anenhancednutrientgradientinthewatercolumn(cf.,Kroopnick, 1985) andpointstoincreaseinprimary productivity.Theincrease in the concentration of phytoplankton-produced sterols, though highlyvariable,intheeasternFramStraitduringtheLGMrelative tolate MIS3supports thesuggestedincrease inprimary produc- tivityduring the LGM(Müller andStein,2014; Fig.4c).Together, theobservedchangeslikelyindicateupwellingofnutrient-richwa- ter.
Ifthe watercolumnde-stratification during theLGM (evident from the decrease BOu-PNp 14C age difference)was achieved via downwelling(e.g.,openoceanconvectionorbrineformation),the decrease in the BOu-PNp δ18O would have been caused mainly by a change in benthic δ18O (i.e., transfer of surface water sig- nal to deep water) rather than a change in planktic δ18O (i.e., transfer ofdeep water signal to surface water). At the transition from late MIS3 to the LGM, O.umbonatus δ18O remain almost constant,whilst N.pachydermaδ18Oincreaseby ∼0.4h (Fig. 4g).
During the LGM, both N.pachyderma andO.umbonatus δ18O in- creasedby∼0.2h.Ifsome ofthe changeinBOu-PNp δ18Oisdue to changes in local seawater δ18O, a part of the increase in N.
pachyderma δ18O from MIS 3 to the LGM is likely due to mix- ing withupwelled deep waterwithhigher δ18O. The ∼0.2h in- crease in both N.pachyderma andO.umbonatus δ18O during the LGMmayhavebeencausedbythecontinuinggrowthoftheLGM icesheetsuntil21 ka(e.g.,Hughesetal., 2016). Possiblechanges in the surface-deep water temperature and/or salinity gradient mayhavecaused/contributedtotheobservedchanges inBOu-PNp δ18OdifferencefromlateMIS3toLGM, whichrepresentsasource of uncertainty in our interpretations of BOu-PNp δ18O difference changes.
Thornalley et al. (2015) observed an increase in near-surface reservoirages inthe centralNordicSeas from∼45014Cyears in lateMIS3to∼140014CyearsduringtheLGM(Fig.4d).Upwelling ofdeep aged watercould have caused thisincrease in the near- surface reservoir age. If this is correct, it would imply that the decreasein the BOu-PNp agedifference resulted froman increase inthenear-surfacereservoirageandagingoftheentirewatercol- umn,ratherthanfromadecreaseinthedeepwaterreservoirage.
Taking these near-surface reservoir ages into account, the venti- lation age of deep water in the Fram Strait may have remained around200014CyearsduringthelateMIS3andtheLGM,though thewatercolumnmixingprocessesanddynamicswereverydiffer- entduringthesetwotimeperiods.Weacknowledgethelargeun- certaintyinourestimatesforLGMdeepwaterreservoiragesgiven the poor control on the calendar age chronology in our records, and the large uncertainties associated withthe surface reservoir ageestimatesinThornalleyetal. (2015). Nevertheless,theseesti- matesareconsistentwithdeep-oceanreservoir-agereconstructions fromthe southern Norwegian Seaandthe North Atlanticforthe LGM(Ezatetal.,2017; Skinneretal.,2014).
3.2. ReassessmentofArcticMediterraneanventilationage reconstructions
Our result from the Fram Strait has significant implications for previous 14C ventilation-age reconstructions from the Arctic Mediterranean and the subpolar North Atlantic. As mentioned above, Thornalleyetal. (2015) published datathat indicatedthat ArcticMediterraneandeepwateragedbyupto>10,00014Cyears during theLGM.Alaterstudyby Ezatetal. (2017) basedonsed- iment coresfrom ∼1200 m water depthfrom thesouthern Nor- wegian SeaandtheIceland Basinshowedthat radiocarbondates measured in Pyrgo and other miliolids apparently confirmed old agesofdeep-wateroverflowsof>6000 yr fromtheArcticMediter- raneantothesubpolarNorthAtlanticduringthelast deglaciation.
However,radiocarbondatingofotherbenthicforaminiferalspecies from the same sample set yielded ventilation ages <2500 14C years(Ezatetal.,2017).Twoscenariosfortheinterspeciesbenthic foraminiferal14Cageoffsetswereproposed.First,differentbenthic foraminiferal speciesdatedfromthe samesample that mayspan tenstoseveralhundredsofyearscouldrepresentephemeralevents within the time period that a given sampling depth represents.
This interpretation implies persistent high frequency (centennial todecadal)andhighamplitude(>5000 yr)variabilityintheven- tilation ageofNordicSeas overflows during thelast deglaciation.
Alternatively, the agesgiven by Pyrgo speciesmaysimply be too old. Thisenigmarepresentsa criticalobstacleto makerobust in- ferencesaboutoceancirculationchangesandNorthAtlantic-Arctic Mediterraneandeep-waterexchangesinthelightofbottomwater 14Cventilation-agereconstructions. Itisthereforecrucialtoinves- tigate the reliability of the radiocarbon dates measured in Pyrgo species from the glacial and deglacialsediments from the Arctic MediterraneanandthesubpolarNorthAtlanticOcean.
A key observation in the study of Thornalley et al. (2015) is the abrupt aging of deep waterin the northern Norwegian Seas to764014Cyears(withB-Pageof621014C years)at∼22 ka(at uncalibrated planktic14C date of19,790 yr before present (BP)).
Theauthorssuggestedthatpre-agedwaterfromthecentralArctic OceanflowedintotheNorwegianSeathroughtheFramStraitdur- ingtheLGM.SedimentsdatingfromtheLGMfromtheFramStrait should therefore record the hypothesized outflow of ‘extremely aged’ Arctic deep water. However, our LGM B-P age differences from the Fram Strait do not support an outflow from the Arctic Oceanwith>600014CyearsoldwatertotheNordicSeas(Fig.4e, Table 1).Thesecontrasting resultsare mostlikelyduetothe use of Pyrgo for 14C measurements in previous reconstructions (Ezat et al., 2017). It is important to note that the LGM sediments in ourrecordsfromtheFramStraitdonotcontainanyPyrgospecies and radiocarbon dates could only be obtained on O.umbonatus (theonlyabundantcalcareousbenthicforaminifer;Supplementary Fig. 1).ThereisasmallpeakintheabundanceofPyrgospeciesin coreHH12-946MCduringlateMIS3(33.5–26 ka),andtheirradio- carbondatesare6000–8000 yr olderthanthedatesperformedin O.umbonatus(SupplementaryFig. 1;Table1).Whiletheδ13CofP.
depressaandO.umbonatus areverysimilar intheHoloceneinter- val,theδ13CofP.depressaaresignificantlyhigherinMIS3(Table1, SupplementaryTable 1).Thisrelativeincreaseinδ13CofP.depressa is,atfacevalue,inconsistentwiththeideathatP.depressarecords anolderwatermassthanO.umbonatus.
To explore further the reliability ofPyrgo 14C dates, two new radiocarbondatesweremeasuredinN.pachydermaandPyrgorota- lariaspecimensfromoneglacialsamplefromthesouthernNorwe- gianSea(coreLINK15;Fig.1).Theuncalibratedradiocarbonagesof N.pachydermaandPyrgorotalariaare21,750and29,395 14Cyears BP,respectively,withaB-Pagedifference of764214Cyears.This agedifferenceof764214CyearsbetweenP. rotalariaandN.pachy- dermafromthesouthernNorwegianSeaalsocannot beexplained
M.M. Ezat et al. / Earth and Planetary Science Letters 518 (2019) 67–75 73
by an agedoutflow fromthe Arctic/northern Norwegian Sea,be- cause the previously suggested ‘extreme aging’ in the northern Norwegian Seastartedlater (at uncalibrated planktic14C dateof 19,79014CyearsBP;Thornalleyetal.,2015;seeabove).Wewould expect that thesignature ofthe extremelyaged waterto appear earlierornearlyatthesametimeinthenorthernNorwegianSea compared to the southern Norwegian Sea to be consistent with thescenarioproposedbyThornalleyetal. (2015).Thus,thisresult supportthe evidencethat theolder14Cdates measured inPyrgo speciescomparedtoother speciesinthesamplesarenota result ofrecordingbottom-waterradiocarbonactivity.
Aremaining question iswhatdid causethetooold Pyrgo 14C dates?Maganaetal. (2010) proposedthatsimilarlyoldPyrgoages fromdeglacialsedimentsfromtheSantaBarbaraBasinintheeast- ern Pacific Ocean were caused by presence of radiocarbon-free hydrocarbonsinthecalcifyingenvironment.Thiswasbasedonthe associatedlowPyrgoδ13C(−4to−8h).Incontrast,theoldPyrgo
14C ages fromthe Nordic Seas are associated with high δ13C of 0.6to 1.1h,whichare ∼1.5h higherthanδ13Cofcoevalinfau- nal benthicspecies of much younger 14C ages (Table 1,see also Thornalleyetal., 2015; Ezat etal., 2017). At332 m waterdepth offNorthernNorway,samplesof deglacialagedatedon Quinque- loculinasp. andmixedmiliolid species yielded14C ages that are upto5000 yr olderthantheirstratigraphicchronologyandcoeval bivalveshells(Vorren andPlassen, 2002).In Holoceneshelf sedi- mentsfromSkawSpit, Denmark,mixedmiliolidspeciesalsogave up to 5000 yr older ages than coeval other benthic foraminifera andbivalve shells (Heier-Nielsen et al., 1995). This was herein- terpreted as miliolids had been laterally transported from older sediments. Lateral or downslope reworking of older fossil Pyrgo specimensseemalso aplausibleexplanation forthetooold ages recordedforsometimeintervalsintheNordicSeasandthesubpo- larNorthAtlantic(Ezatetal.,2017andreferencestherein).Indeed, Vorren and Plassen (2002) and Heier-Nielsen et al. (1995) stud- ies suggest that 14C dating of miliolid species other than Pyrgo speciesmayalsobe questionable, butthisrequiresadditionalin- vestigation.Ezat etal. (2017) speculated that thetoo old agesof Pyrgo speciesin the glacial Arctic Mediterranean mighthave re- sultedfromunknown‘vitaleffects’adoptedbyPyrgospeciesunder specificenvironmentalconditions.Forexample,microbialfermen- tation (e.g., by endosymbiotic microbes) of old organic matter, which may have been more readily available during glacial ice- raftingand/ormeltwaterdischargeevents,willresultinhighδ13C and14C-depleted bicarbonate (see Carothers and Kharaka,1980).
However, Pyrgo species from deglacial and glacial sediments do not always date tooold. For example,Pyrgo dates fromthe cen- tralNordicSeasat∼28–23 ka and∼16.5 kagiveventilationages
<2000 yr (Thornalleyetal.,2015).Weare thusnotableto spec- ify the exact mechanisms responsible forthe old Pyrgo 14C ages (indeed,differentmechanismsmightprevailindifferentcontexts), howeverwecanconfirmthatanArcticdeepoutflowof>6000 yr oldwatertotheNordicSeasduringtheLGM(26–20 ka)couldnot betracedtoitsputativesourceinourFramStraitrecords.
3.3.ImplicationsfortheLGMclimate
Bottom water temperature reconstructions from the Faroe- Shetland Channel at 1200 m water depth from the southern NorwegianSea showpersistently highbottomwatertemperature (∼3◦Chigherthanmodern)at26–20kacomprisingtheLGM(Ezat etal., 2014). Thisincrease inbottomwatertemperaturehasbeen explainedbyinflowofwarmAtlanticwaterbelowabuoyantlayer (Ezatetal.,2014) aspreviouslysuggestedforglacialstadialperiods (e.g.,RasmussenandThomsen,2004).Upwellingofdeepwaterin theFramStraitandprobablyinotherareasinthenorthernNordic Seasprovidesanoutflowpathwaythatmaybenecessarytomain-
tain thesubsurface inflowofAtlantic waterintothe NordicSeas.
ThissubsurfacenorthwardinflowofAtlanticwaterandsubsequent upwelling during the LGM may have provided a source of local moisture for the Svalbard-BarentsSea Ice sheet (e.g., Hebbeln et al.,1994),whichcontinuedtogrowuntil21 ka(e.g.,Hughesetal., 2016).ThisisinagreementwithbiomarkerstudiesfromtheFram StraitandtheeasternYermark Plateauthatshow mainlyseasonal sea-iceduringtheLGMandnotapermanentsea-icecover(Müller andStein, 2014; seeFig. 4b, c).The increase in primary produc- tivityduetoupwellingofnutrient-richwatermayhaveenhanced photosyntheticremovalofCO2 fromthe surface oceanenhancing local oceanicCO2 uptakefromthe atmosphere. However, a more directproxyforsurface ocean pCO2 wouldbe neededtoconfirm this, as the pCO2 drawdown via increased primary productivity mayhavebeenbalancedorexceededbytheeffectsofhighertotal dissolvedCO2 intheupwelleddeepwater.
Furthermore, our finding that the Arctic Ocean was not ex- tremelyagedduringthelateglacialhasimportantimplicationsfor the interpretation of ocean circulation changes in the North At- lantic.Forexample,KeigwinandSwift (2017) foundindicationsof presence of a relatively young andnutrient-depleted water mass duringtheLGM(21–18 ka)at5000 mwaterdepthinthesubtrop- icalwesternNorthAtlantic.Theauthorsofthisstudyexcludedthe ArcticMediterraneantobethesourceofthiswatermassbecause ofprevious inferencesofan extremelyaged (and thuslikely iso- lated)deepArcticMediterranean(Thornalleyetal.,2015).Instead, they suggested the LabradorSea asthe mostlikely source. With our new results, the Arctic Mediterranean should not be ruled out asa possible sourceof thisdeep-water massin theWestern North Atlantic. In principle, water mass source proxies (e.g., Nd isotopes)frombothregions couldpotentially revealthesource of the5000 mdeepwatermassfoundinthewesternNorthAtlantic duringtheLGM.
Finally,the closetiming betweenthe decreasein atmospheric radiocarboncontent∼26–20 ka (e.g.,Reimeretal., 2013) andour
‘inferred’ upwelling of aged (∼2000 yr old) deep water in the FramStrait (andprobablyinother areasin theNordicSeas) sug- gestsapossiblenon-negligibleimpactonthe(radio-)carboncycle viathehighnorthernlatitudes.Thesefindingsthereforeencourage additionalworktofurther explorethespatial extentofupwelling and/or deepmixingintheNordicSeas,andtoassesstheimpacts onocean-atmosphereexchangesofheatandcarbon.
4. Conclusions
In this study, we have reconstructed deep ocean circulation changes inthecentral FramStrait intwo recordsfrom>2500 m waterdepth,using14Cdatesonbenthicandplankticforaminiferal species alongside multispecies carbon and oxygen stableisotope measurements. We show that the hypothesized model of Arctic purgingofextremelyaged deepwater(>6000 yr oldwater) into theNordicSeasduringtheLGMandtothesubpolarNorthAtlantic duringthelastdeglaciationislikelybiasedduetotheuseofPyrgo speciesfor dating.Ourventilation ageestimatesshow significant aging (∼2000 yr old water) and strong stratification during late MIS 3 in the Fram Strait suggesting a dramatic reduction/shoal- ing of deep-water formation inthe Arctic Mediterranean. During theLGM,oceanstratificationbrokedownduetoenhancedvertical mixing,whichresultedinan increaseinprimary productivity.We show that thephysical oceanographic andbiological state inthe FramStrait andnorthernNordicSeasduring theperiod26–21 ka could have contributed to the final intensification phase of the lastglacialvia localocean-atmosphereexchangeofheat,moisture and carbon,thus underlining the climatic importance ofpositive feedbacks betweenthe atmosphere, ocean circulation,carboncy- cle,andcryosphereinhighnorthernlatitudes.
