Contents lists available atScienceDirect
Earth and Planetary Science Letters
www.elsevier.com/locate/epsl
Beryllium isotope signatures of ice shelves and sub-ice shelf circulation
Duanne A. White
a,∗, David Fink
b, Alexandra L. Post
c, Krista Simon
b, Ben Galton-Fenzi
d,e, Simon Foster
a, Toshiyuki Fujioka
b, Matthew R. Jeromson
a, Marcello Blaxell
a,
Yusuke Yokoyama
faInstituteforAppliedEcology,UniversityofCanberra,ACT,2617,Australia
bAustralianNuclearScienceandTechnologyOrganisation(ANSTO),PMB1,Sydney,NSW,2234,Australia cGeoscienceAustralia,Symonston,ACT,2609,Australia
dAntarcticClimateEvolutionCooperativeResearchCentre,PrivateBag80,Hobart,TAS,7001,Australia eAustralianAntarcticDivision,Kingston,Tasmania,Australia
fUnivesityofTokyo,Tokyo,Japan
a r t i c l e i n f o a b s t ra c t
Articlehistory:
Received14June2018
Receivedinrevisedform1October2018 Accepted5October2018
Availableonline25October2018 Editor:D.Vance
Keywords:
meteoric10Be marinesediments Beisotopemixing sub-iceshelfcurrents sedimentadvection
Be isotopesare ausefultracer ofsediment sourceand transport pathwaysbuthave not beenwidely testedinglacio-marineenvironments.WemeasuredBeisotopesinarangeofdepositionalenvironments fromopenmarine,sub-iceshelfandsubglacialsettingsthroughoutPrydzBay,oneofAntarctica’slargest ice drainage systems. We find that strong sub-ice shelf and bottom current circulations can advect 10Be-rich open marine sediments into an ice shelf cavity, and 10Be-poor terrestrial sediments onto the continental shelf atthe ice shelf outflow, meaning that 10Be concentrationsreflect sub-ice shelf circulationpatternsratherthandepositionalenvironment.However,HCl-extractable10Be/9Beratioscan provideamorerobustdiscrimination ofsedimentdepositedinopenmarineand sub-iceshelfsettings.
Thus,Beisotopesareausefultracerofbothenvironmentalsettingandsub-iceshelfcirculationstrength inbothmodernandpaleo-icesheetmargins.
CrownCopyright©2018PublishedbyElsevierB.V.Allrightsreserved.
1. Introduction
Berylliumisotopesdisplayarangeofpropertiesthatmakethem usefultracersofgeochemicalprocessesandsedimentsources.10Be is produced in the atmosphere (termed meteoric 10Be) by in- teraction of cosmic rays with oxygen nuclei, and decays with a 1.387-million-yrhalf-life (Chmeleff etal., 2010). Cosmicrays also interact withEarth surface rocks (termed in-situ 10Be), although production rates are orders of magnitude lower. Stable 9Be is a naturallyoccurringtraceelementincrustalrocksandaccumulates, together with meteoric 10Be, in many sediment archives in ei- thertheprimarymineralsorsecondaryweatheringproducts (von Blanckenburgetal.,2012).
Muchofthedevelopmentandapplicationofberylliumisotopes in sediment cycling to date has focused ontemperate terrestrial andmarineenvironments.Intemperateareas,meteoric10Berains down onto the land and ocean surfaces, where it is adsorbed to theoutsideof particles orincorporatedinto secondary phases suchasMn andFe-oxideswhereit mixeswith9Besourced from
*
Correspondingauthor.E-mailaddress:[email protected](D.A. White).
weatheringoftheprimary minerals(WillenbringandvonBlanck- enburg, 2010). The relatively constant flux of10Be to the earth’s surface atanyone location overtime meansthat 10Be/9Be ratios and 10Be concentrations provide useful measures of soil forma- tion (e.g. Pavich et al., 1984), sediment transport pathways and provenanceinriverineenvironments(Brown,1987).Inthemarine environment meteoric 10Be is efficiently andquickly transported fromsurfacewatersviabiologicalpathwaystotheseabed(Frank etal.,1997),apropertywhichhasbeenusedtomonitorweather- ingandsequestrationofCO2 onglacial/interglacialcycles(e.g.von Blanckenburgetal.,2015).
Ice sheetstransport and disperseberyllium isotopes ina way thatdiffersmarkedlytothatintemperateortropicalsystems.The sources andtransport pathwaysof 9Be and 10Be across the ma- jor iceoutletsofcontinentalscaleicesheetsarelargelyseparated (Graly et al., 2018). 9Be is sourced primarily by glacial erosion andistransportedinsedimentsthatareconcentratedinthebasal zone (lower few meters and the deforming bed; Fig. 1). Mete- oric10Beproducedintheatmosphereisdepositedwithsnowand then found throughout theice column(Beer etal., 1987), which is largely debris-free. During seaward transport by the ice sheet there is limited or no mixing of these two components, unlike riverinesystemswheremixingoccursbothatthesourceanddur- https://doi.org/10.1016/j.epsl.2018.10.004
0012-821X/CrownCopyright©2018PublishedbyElsevierB.V.Allrightsreserved.
Fig. 1.Conceptualmodelsoftheinfluenceofdifferentglacialandmarineprocesses onthedistributionofBeisotopesinmarginalmarineandice-shelfenvironments.
Top(afterYokoyamaetal.,2016);ificebergcalvingisthedominantprocessofice loss,anycontinentalsedimentreleasedunderneathiceshelveswillhaveverylow 10Beconcentrations,as allterrestriallysourced10Beistransportedseaward into theopenmarineenvironment.Middle;ifsub-iceshelfmeltisactive(dottedline), 10Befrommeltingicewillbedepositedinsedimentsinthesub-iceshelfcavity, althoughthehighsedimentgenerationratesnearthegroundinglinecauses10Be concentrationstobelowerthanintheopenmarine settings,especiallycloseto theicesheet.Bottom;ifstrongbottomcurrentsareactive,sedimentsfromopen marineenvironmentshighin10Be,organicmatteranddiatomsaretransportedinto thesub-iceshelfcavity,reworkingandmixing10Beconcentrationsand 10Be/9Be ratiosinsedimentthroughoutthemarineportionofthesystem.
ingtransport.Thus,icesheetscandelivertheirisotopicberyllium fluxesfromcontinentalinteriors tocoastal marginsin a spatially segregatedmanner,withlimitedinteractionduringtransport.This means there is little opportunity for sediments entrained in the ice to mix their Be isotope fluxes and set a locally representa- tive 10Be/9Be ratio. The independent pathways in Be isotopes in glacial environments resultsin generally low-meteoric 10Be con- centrationsinsedimentsproducedbyglacialtransport,whichhas successfullybeenappliedasamarkerforpresenceandabsenceof icesheets(Schereretal.,1998).
Therehavebeenrelativelyfewstudiestodeterminehowdiffer- entmodes ofberyllium transport inice affects their distribution onglaciatedcontinentalandmarinemargins.Existingstudiesfrom theRossSeasectorofAntarctica(Sjunneskogetal.,2007) suggesta strongpositive10Be concentrationgradientbetweentheicesheet andthe continentalshelf.There are indicationsthatsub-ice shelf
sediments,traditionally challenging to identify insediment cores usingexisting proxies(Postetal., 2014),mayberecognisedfrom their relatively low 10Be concentrations (Yokoyama et al., 2016).
However, the utility of 10Be as an iceshelf proxy relieson calv- ing of icebergsat the iceshelf front transportingmeteoric 10Be deposited on the ice shelf and sheet to the open marine envi- ronment, thus preventing 10Be reaching sub-ice shelf sediments in the ocean cavity (Fig. 1, top). Other marine-margin processes such as basal ice-shelf melt and sub-ice shelf ocean circulation mayalsoinfluencespatial distributionofBe isotopes(Fig.1,mid- dleandbottom),buttherateandroleoftheseprocessesispoorly quantified.The influenceofsub-iceshelf processesmaylimit our abilitytocorrelateBeisotopesinsedimenttothedynamicsofice sheet retreat and advance, and understand the effect of glacial–
interglacialcycleson berylliumcyclinginthe globalocean.Thus, moreinformationisrequiredtobetterunderstandthedepositional environments and processeswhich control Be isotopeconcentra- tions.
In thispaperwe investigate thedistribution of berylliumiso- topes and their range in concentration in sediment from Prydz Bay, the outlet for the Lambert Glacier–Amery Ice Shelf system.
Thisareapresentsausefulcasestudy,astheprocessesofoceanic circulation(Galton-Fenzietal.,2012),sediment transport(Postet al., 2014), and ice shelf (Hemer and Harris,2003) and ice sheet history(e.g. Domacketal., 1998; Whiteetal., 2011) inthissec- toroftheAntarcticcoastalmarginarerelativelywelldefined.Itis alsoan area wherestrongoceaniccirculationsonthe continental shelfandinthesub-iceshelfcavityblurthephysicalandbiological characteristics ofdifferentglacio-marineenvironments (Hemer et al.,2007; Postetal.,2014) andthusposechallengestotraditional sedimentologicalmodels ofproximal glacio-marinesedimentation (e.g.DomackandHarris,1998) usedtoidentifypasticeshelf ex- tents.
Here,weaimtodistinguishhowsub-iceshelfcirculationaffects theBeisotopiccompositionofsedimentsindifferingenvironmen- tal settings andrelate theseto oceanicand glacial environments andprocesses.We investigatewhetherBe isotopes can begener- ally applied asa tracer ofdifferent depositional settings along a transect froma continentalicesheet, to the groundinglinezone (where groundedicebeginsto floatandinteractwiththe ocean) alongthesub-iceshelfsettingtoanopenmarinecontinentalshelf environment.
2. Environmentalsetting
At the northern extension of the Lambert Glacier–Amery Ice Shelf System, the Amery IceShelf is the drainagepoint ofeight tributary glacial basinsthe largestofwhich are theLambert and MellorBasins(Fig.2).Thesystemdrains16%oftheEastAntarctic IceSheet,transmitting88.2±2.9Gtayear,makingitasignificant contributortotheoverallicemassbalanceofAntarctica(Yuetal., 2010; Postetal.,2014).
The Amery IceShelf isapproximately 450 kmlong and thins fromamaximumofover 2500matthesouthern groundingline to around 200m atthe shelf front. The thicknessof thesub-ice cavity rangesfrom 250m to 1600m (Galton-Fenzi etal., 2008), witha highbasal melt ratecomparedto other EastAntarctic ice shelves(Pritchardetal.,2012).
TheLambert–AmerysystemdrainsintoPrydzBay,wheresteep bathymetryandtopographyallowsforcurrentstoformthestrong clockwise circulationof the Prydz Bay Gyre. ModifiedCircumpo- larDeepWaterflowsfromtheeastoverFourLadiesBank, where it intersects the East Wind Drift current, carrying fresh iceshelf water from West Ice Shelf and highly saline water produced in thesea-icefreeareaoftheBarrierPolynya(Williamsetal.,2016).
Here,thecurrentsplits,withpartcontinuingwestwardacrossthe
Fig. 2.MapofLambertGlacier–AmeryBasin,PrydzBay,moderniceshelfedge(blueline)andlocationofcurrentgroundingline.Sitesdiscussedinthetext,showinglocation ofkeyiceshelves,openmarineandsub-iceshelfbottomcurrents.Notethestronggyre-likecirculationunderneathAmeryIceShelf(afterGalton-Fenzietal.,2012),andthe differingcharacteristicsofwatermassesinPrydzBay,andtheeasternandwesternsidesofthesub-iceshelfcavity.CDP,MP,DPandBPindicateCapeDarnley,Mackenzie, DavisandBarrierPolynyasrespectively.Redboxininsetonlowerrightindicateslocationofmainmap.Sea-bedelevation(metresbelowsealevel)fromIBSCOcompilation shownforareasseawardofthegroundingline.(Forinterpretationofthecoloursinthefigure(s),thereaderisreferredtothewebversionofthisarticle.)
ice shelf front, while the remainder flows south into the sub- iceshelf cavityof theAmery IceShelf (Fig. 2). Belowthe Amery IceShelf, thesetwo currentsmix andforma highly saline water mass,whichthensinks.Theback-slopingbathymetryoftheAmery IceShelf cavityandCoriolis effectdrives thesecurrentsalong an eastern channel belowthe AmeryIce Shelf towards the ground- ingline,causingsignificantbasalmelt.Here,thenowfreshermelt waterrisesandfollowstheclockwisegyrealongthewesternmar- gin of the Amery Ice Shelf, andis ejected into Prydz Bay as Ice Shelf Water, completing the ‘ice pump’ driven circulation typical of Antarcticsub-shelf circulation systems (i.e.Herraiz-Borreguero etal., 2015). IceShelf Waterfrom theAmery IceShelf is mostly recirculatedthroughthePrydzBayGyreback undertheice-shelf, butsomeflowswestwardsoutofthebay(Fig.2).
In the open marine environment, in particular on the west- ern flank of the Prydz Channel depression, currents can exceed 1cm/s (Nunes Vaz andLennon,1996). Flow speedscan be even greaterundertheAmeryIceShelf,whichhasastrongsub-iceshelf current system (∼5–10 cm/s; Fig. 2). Sub-ice shelf currents un- der the Amery are substantially more vigorous than other large ice shelves (e.g. Ross Ice Shelf, 1 cm/s; Jenkins et al., 2003;
Post et al., 2014), and is capable of depositing sediment from a marineenvironmentgreatdistancesbeyondtheicefront.
3. Methods
The PrydzBayregionhasbeenthefocusofseveraldecadesof geophysical mapping andsediment coring.Sediment grabs, grav- ityandpistoncoreshavebeencollected oncruisesby theAurora Australis(186&901),JoidesResolution(Leg188,2000),Nathaniel B. Palmer(NBP01,2001) andPolarstern (XXIIV,2007).Coreshave alsobeencollected fromunderneath theAmeryIceShelf through the AMISOR program (Allison, 2003). For this paper, we inves- tigated material from AuroraAustralis cruises 186, 901 and the AMISOR program,which hasbeen preservedat4◦C sincecollec- tion.
To investigate the modern distribution of 10Be in Prydz Bay sediments, we selected samples from core top (surficial) sedi- ments representative of different modern depositional environ- ments.Sevencore-topsamplesfromopen marine(labelledasGC) andsixsea-bedsamplesfromsub-iceshelf(labelledasAM01–06) environments were obtained from a wide geographic area (Ta- ble 1). Open marine sediment samples were restricted to areas where iceberg turbation is absent, or in the case of GC18, po- tentially limited. Core top GC27 was sampled by ship in open marineconditionsadjacenttotheiceshelfmarginbutisnowcov- eredbytheiceshelf.Observationsoftheshelf-frontpositionsince
Table 1
Berylliumisotopeandcorelocationdata.
Core name Cruise Water
deptha (m)
Core depthb (cm)
Latitude Longitude Sample mass (g)
9Be totalc (μg/g)
9BeHCl extracted (μg/g)
10BeHCl extractd (106at/g)
10Be/9 HCl ext.
(10−8at/at) Open marine core tops
GC22 901 766 0 68.065 72.2760 0.41 – 0.22 785.1±18.4 5.5
GC27 901 776 0 68.947 73.5858 0.68 – 0.25 985.6±22.7 5.8
GC29 901 789 4 68.663 76.6955 0.34 – 0.18 847.1±19.8 7.1
GC24 901 705 0 68.094 73.1893 0.46 – 0.21 892.1±20.7 6.2
GC15 186 1050 1 68.518 70.4075 0.86 – 0.12 173.2±4.5 2.2
GC16 186 726 4 68.375 71.3070 0.70 – 0.13 347.7±10.4 4.1
Sub-ice shelf core tops
AM01B AMISOR 840 0 69.431 71.4462 0.62 2.01 0.49 810±19.6 2.5
AM02 AMISOR 843 0 69.713 72.6400 0.44 – 0.64 1110±28.7 2.6
AM03 AMISOR 1339 0 70.561 70.3322 1.35 2.58 0.53 838.3±19.2 2.4
AM04 AMISOR 1002 0 69.900 70.2903 1.36 2.61 0.50 457.5±10.5 1.4
AM05 AMISOR 979 1 70.230 69.6967 1.79 2.73 0.44 248.3±5.7 0.8
AM06 AMISOR 902 0 70.250 71.3500 2.11 2.41 0.27 508±11.8 2.8
Basal diamicts (subglacial)
AM01b AMISOR 840 45 69.431 71.4462 2.75 – 0.32 31.1±1.1 0.1
AM04 AMISOR 1002 115 69.900 70.2903 2.51 – 0.24 36.2±0.9 0.2
GC16 186 726 227 68.375 71.3070 2.13 – 0.42 160.9±4.1 0.6
GC24 901 705 431 68.094 73.1893 2.32 – 0.42 86.3±2.2 0.3
GC29 901 789 351 68.663 76.6955 2.27 – 0.17 100.7±2.4 0.9
Shallow water
GC18 901 320 1 67.283 76.5703 0.71 – 0.35 776±18.2 3.3
Down-core
AM01b* AMISOR 840 0 69.431 71.4462 0.62 2.01 0.49 810±19.6 2.5
AM01b AMISOR 840 4 69.431 71.4462 0.77 – 0.45 766.1±17.9 2.5
AM01b AMISOR 840 11 69.431 71.4462 1.13 2.22 0.48 813.8±18.7 2.5
GC22* 901 766 0 68.065 72.2760 0.41 – 0.22 785.1±18.4 5.5
GC22 901 766 13 68.065 72.2760 0.57 – 0.29 757.6±18.3 3.9
GC22 901 766 22 68.065 72.2760 0.57 – 0.54 703.8±16.4 2.0
Repeat analyses
GC24-1e 901 705 0 68.094 73.1893 0.39 – 0.24 834.8±19.7 5.1
GC24-1bf 901 705 0 68.094 73.1893 N/A – – 47.1±1.4
AM05-2e AMISOR 979 1 70.230 69.6967 1.34 2.73 0.45 252.4±6 0.8
AM05-2bf AMISOR 979 1 70.230 69.6967 N/A 2.73 – 14.5±0.5
a Waterdepthforopenmarinecores,depthbelowgroundediceshelfsurfaceforsub-iceshelfcores.
b Sedimentsub-sampledepthbelowcoretop.
c Total9BecontentfromcompleteHFdissolutionofdrysedimentaliquot.
d 10Beuncertaintiesare1σstandarderrors.
e Repeatfromasecondsedimentaliquotandfullprocessing.
f RepeatleachingofleachedsedimentusedinfirstGC24-1andAM05-1.
* Duplicateddataentries(initalics)fromthemarineandsub-iceshelfcoretopsforeaseofcapturingthedowncoretrends.
1936suggests thissite is likely open marine for atleast 80% of the 60–70 yr calving cycle (Fricker et al., 2002). Core top sedi- ments from open marine andsub-ice shelf environments are all finegrained,withvarying siliceouscontent inopen marine envi- ronmentsof 1to 50%, andof diatom abundancein sub-iceshelf sedimentsof0to160×106 valves/g(Postetal.,2014).
Thenatureofgravitycoringmeansthesurficialsedimentsmay havebeendisturbed ornot retained,andthus there issome po- tentialoursamplesmaynotrepresentthemostrecentlydeposited sediment.Availableradiocarbondating (e.g.Postetal.,2014)and comparisonwith ages of known surface samples (Domack et al., 1991) indicatesthat ifthishasoccurred, oursedimentsarelatest Holoceneinage,likelythelast∼1ka.Tounderstandthetemporal variationsthatmayhaveinfluencedtheBeisotopeconcentrations, wesupplemented the coretopsediments withalimitednumber ofdown-coresamples.Holocenesiliceousmarineooze(SMO)from open marine (GC22, depths of 13 and 22 cm) andsub-ice shelf (AMO1B,4 and11 cm, Table1). The limitedvariation presentin BeisotopeconcentrationacrossHolocenetimescalesprovidescon- fidencethatthedifferencein10Beconcentrationsinourcoretops across the region reflect geographic rather than temporal varia- tions.
Depthsamples were also takenfrombasal sands and diamict facies interpretedasgrounding lineorsubglacial sedimentsfrom a range of sub-ice shelf and open marine cores (AMO1B [depth 45 cm],AMO4[115 cm],GC16[227 cm],GC24[431 cm]andGC29 [351 cm]).
WeusedaHClextractantchemistryprocesstoreleaseboth9Be and10Befromthesediments(Knudsenetal.,2008).Thisapproach effectivelyremovesthefullauthigenicmeteoric10Befractionwith- outaccessingin-situ10Be.Sedimentwascollectedfrom1cmthick sections,dried overnight at100◦C, andthen reactedwith10 ml of6 MHClfor3h atroom temperature.Thesolutionswerethen splitintotwoaliquotstoseparatelymeasureconcentrationsof9Be by ICP-OESandICP-MSatthe UniversityofCanberraandcosmo- genic10Be byAcceleratorMassSpectroscopy.Thealiquotremoved for9Bewas∼20%oftotalextractedsolutionandwassufficientto achieve arepeatabilityof<5%for9Be assaysintherange100to 500ppb.
FortheAMSanalysis,thesecond aliquotwasspikedwith0.22 or 0.45 mg of 9Be prepared from beryl crystal with a negligi- ble10Becontent.Berylliumwas isolatedfromthesolutions using methods described in Child et al. (2000), oxidised, mixed with Nb powder and pressed into AMS targets (Fink et al., 2000). Fi-
nal 10Be/9Be ratios were corrected by full chemistry procedural blanks andnormalized to the NIST-4325 10Be/9Be AMS standard usinga nominal ratioof27,900 × 10−15 (FinkandSmith, 2007;
Nishiizumi et al., 2007). Full procedural blanks gave a mean 10Be/9Be ratio of 20 ± 10 × 10−15 (n=2), which was equiva- lentto 0.5% the ratiomeasured forthe lowest sediment sample.
Finalanalyticalerrorinconcentrations(atoms/gram)werederived fromaquadraturesumofthelargerofthetotalstatisticalerroror standardmeanerrorfromrepeatAMSratios(typically1%),2%for AMSstandardreproducibilityand1%inBespike.
Toassessthat our chemistrymethods were reproducible with respect to equivalent extraction efficiencies, full replicate proce- dural measurements were carried out on two core-top samples (GC24, AM05) starting from unprocessed sediment sub-samples resulting in 10Be concentrations consistent within 1 and 5% re- spectively.While6 MHClhasbeenshowntoquantitativelyextract themeteoric10Becomponentinmarinesediment(Knudsenetal., 2008),wefurthertestedthequantitativenatureoftheHClextrac- tionby conductinga second6 MHCl leachonthe sameleached sedimentaliquotsofGC24-1andAMO5-1usedabove.Thesecond repeat leach extracted a further ∼1–3% of the 10Be extracted in thefirstleach.Asimilarresultwasgivenforwhole-sedimentcon- centrationsofmostmajor elements,although thiswashigher for themorerefractoryelements,suggestingmostofthemoremobile fractions of the sediment had been dissolved by the first leach.
Thesetests suggest that the HClextraction issufficiently aggres- sivetoquantitativelyremovethemeteoric(mobile)componentof thesediment.
To assess that fraction of element concentrations leached via our chemistry procedures to their total whole sediment concen- tration, regardlessofaffinity, we comparedextractedsolutions to assaysinHF-basedtotaldigest (Bourlesetal., 1989).Between10 and30%(meanof23%)ofthetotalBepresentinthesedimentwas extractedbythe 6 M HClleachon freshsediment,valuessimilar to that obtained via weaker extractiontechniques used to target
‘reactive’ 9Be in Antarcticsediments (Vallettaet al., 2018). There werenoobvioustrendsbetweenBeextractionefficiencyandsam- ple characteristics such as mineralogy or total Be concentration.
The first HClleach typically extracted∼30% ofthe more soluble major elements (Fe, Mn, Na, K), and 10% of more refractory el- ements(Ti,Al),consistentwithexistingstudies (e.g.Larneretal., 2007)indicatingitiscapableofextractingtheadsorbed,oxide,car- bonateandorganicfractionsofthesediment.
Lastly,we comparedourberylliumisotopemeasurements toa range of sediment characteristics to better understand the pro- cesses that influence isotopeconcentrations andratios. Data was sourcedfromavailablecore-topanalysesofgrainsize,diatomabun- dance, opalandpercentmodern radiocarbon (fromcore-topbulk organic radiocarbon ages) previously analysed from Prydz Bay (O’Brienetal.,1995; Harrisetal.,1998; TaylorandLeventer,2003) andunder the ice shelf (Post et al., 2014). Holocene sedimenta- tionratesateachsitewereestimatedbasedonradiocarbondating whereavailable (Domack et al., 1998; Taylor andMcMinn, 2002;
Post etal., 2014), orthe assumption that biogenic sedimentation began shortly after ice decoupled from the bed across the re- gionaround thebeginningofthe Holocene(Domack etal.,1998;
HemerandHarris,2003).
4. Results
In the Prydz Bay region, we observed distinct differences in 10Be and 9Be concentrationsofsediments depositedin thethree majorice-marginaldepositional environments(open marine,sub- ice shelf andinferred subglacial/grounding line; Fig. 3, Table 1).
Be-isotope concentrations and ratios were generally within the range of those reported elsewhere on the Antarctic continental
Fig. 3.10BevsHClextractable9BeinPrydzBay.Openmarineandsub-iceshelfare core-topsedimentsintheirrespectiveenvironmentsinPrydzBayandunderthe AmeryIceShelf,while‘subglacial’sedimentsarebasaldiamictsinselectedcores bothundertheiceshelfandinPrydzBay.Notethatbothisotopesareneededto discriminatesamplesfromthethreeenvironments,thoughallsubglacialsediments canlargely bedifferentiatedby10Bealone.Errorbarsrepresent 1σ standarderrors.
shelfandslope(Franketal.,1995; Sjunneskogetal.,2007; Valletta etal., 2018).However, therelationshipbetweenenvironmentand Be isotope concentrations was substantively more complex than previously reported fromthecontinentalshelf (Sjunneskog etal., 2007), with isotope concentrations varying by up to a factor of five within each environment. Apart fromlower 10Be concentra- tionsinsubglacialenvironments,nosingle concentrationprovided a unique signature ofdepositional environment.Spatial trends in theberylliumisotope concentrationsinPrydz Bayprovideinsight into the different sources andprocesses that influence 10Be and 9Beconcentrationsinicemarginalmarinesediments.
4.1. 10Be
Modern, core-top10Be concentrationsinopen marine (∼200–
1000 × 106 at/g) and sub-ice shelf environments (250–1100 × 106 at/g) displayed substantial variation within, andoverlap be- tween,thetwogeographicallydifferentdepositionalenvironments.
Biogenic poor, basal diamicts inferred to have been deposited in subglacialsettings(30–160×106 at/g)weredistinctlylowerthan modern(coretop)sediments.
Thespatialvariationincore-top10Beconcentrationsfollowsthe pattern ofopen marine and sub-iceshelf circulation. Higher val- uesarepresentinthecentralandeastern partsofPrydzBay,and the‘inflow’sectorunderneaththeiceshelf.Muchlowervaluesare presentonthewestern,‘outflow’sideoftheiceshelf,andtheopen marine immediatelynorth ofthe calvingmargin of theice shelf, wheresub-iceshelfwaterdischargesintoPrydzBay(Fig.4).
VariationoverHolocenetimes(i.e.down-core)withintheSMO horizons was limited, confirming the environmental conditions duringthisperiodwerebroadlystable(Table1).Thisconfirmsthat core-topsedimentsprovidean accuraterepresentationofmodern depositional processes forboth open marine (GC22) and sub-ice shelfsettings(AM01B).
4.2. 9Be
In contrast, HCl extractable 9Be variability is more subdued, with intra and inter-group variability halved that seen in 10Be.
HClextractable9Be wasgenerallyhigherunderneaththeiceshelf
Fig. 4.Corelocations(left)andcore-topresultsforHClextractable10Be(106 at/g),9Be(μg/g)and10Be/9Beratio(10−7at/at)concentrations.Ineachplot,thesamples dominatedbyaterrestrialsedimentsourceofBeisotopeconcentrationsorratiosareshownaswhitedots,andsedimentsdominatedbyanopenoceansourceofBeisotopes areshownasblackdots,witha3levelgrey-scaleforvaluesbetweenthetwolimits.NotethattheiceshelfhasexpandedovercoresiteGC27sincecollection,aspartofa 60–70yr calvingcycle(Frickeretal.,2002).
Fig. 5.RelationshipbetweenpercentModernCarbonand10Beconcentration(leftpanel;RossSea,Sjunneskogetal.,2007; Yokoyamaetal.,2016andPrydzBay),and<63 μm mudfractiontoBeisotopesvalues(rightpanel;PrydzBay)fromcoretop(i.e.modern)sediments.VerticaldottedlineinlefthandpanelrepresentsthepercentModern Carbon(pMC)ofpre-bombwatersandtypicalcarbonateshellsontheAntarcticcontinentalshelf(BerkmanandForman,1996),includingunderthemodernAmeryIceShelf (Postetal.,2014).
(range = 0.27–0.64 μg/g) than in open marine (0.12–0.25 μg/g), withsubglacialsedimentsshowingintermediatevalues(0.17–0.42 μg/g,Fig.3).
4.3.10Be/9Beratios
The 10Be/9Be ratio inthe HClextractable portion ofthe sedi- mentprovidedimproveddiscriminationbetweenthethreedeposi- tional environments than either of these values alone. The ratio also allowed discrimination of additional geographic trends that werenotobservableintherawconcentrations.
The lowest 10Be/9Be ratios were observed in subglacial sedi- ments(0.1–0.9×10−8atoms/atom),whilethehighestvalueswere foundintheopen marine environmentofeastPrydz Bay(5–7 × 10−8). Sediment in the eastern, inflow side of the iceshelf pro- ducedanarrowrangeof10Be/9Beratios(2.5–2.8×10−8),despite widevariationintherawconcentrationsofboth.Sedimentonthe western,outflowsideoficeshelfcavity(<1×10−8)andinopen marineenvironmentsimmediatelynorthofcalvingmarginonthe outflow(2.2–4.1×10−8)providedlowervaluesthantheirequiva- lentsintheeastern/inflowingsideofthecirculation.
4.4. Comparisontoothersedimentcomponents
Our sampling was biased toward the mud-rich surface sed- iments, namely siliceous muddy ooze. However, within the ge- ographic range that we have covered, there is little observed correlation between grainsize and either 10Be concentrations or 10Be/9Beratios.HClextractable10Beand9Bearepoorlycorrelated with abundance of the mud fraction (R2=0.1 and 0.3 respec- tively,Fig.5).Inaddition,inallthreemeasures(9Be,10Be,andthe 10Be/9Be ratio), sampleswith 100% mud fraction exhibit the full range of values observed in the dataset. The limited correlation between Be isotope values and the sediment grainsize, suggests that althoughsome partitioningmaybe present(e.g.Wittman et al., 2012), the delivery and deposition of Be isotopes in sub-ice shelfandopenmarineenvironmentsinthissectoroftheAntarctic continentalshelfarenotcontrolledprimarily bygrainsize.Inturn thisimpliesthattheBeisotopesextractedfrombulksediment(i.e.
the HClextractable or‘reactive’ Be) via ourHCl leachingprocess were well equilibrated within the water column prior to being incorporated in the authigenic phase of the sediment. Similarly, theabsenceofcorrelation(R2 of0.1orless)betweensedimenta- tionrateandeitheroftheBeisotopeconcentrationsindicatesthat
Fig. 6.Sedimentadvectionandmixingalongabottomcurrentflowlineasinferredfromatwo-componentmixingmodelusingboth10Beand14Casproxiesforsediment source.FlowlinedistancefollowsthetrajectoryofbottomcurrentsinFig.2,startingfromoffthecontinentalshelf,movingclockwisetothesoutherngroundinglineunder theiceshelf,andthenreturningnorthalongthewesternboundaryoftheiceshelf.Theoverallreductionintheopenmarinesedimentfractionalongtheflowlinemoving frominflowtooutflowzones(i.e.withincreasingdistanceunderneaththeshelf)suggests14Cand10Beconcentrationsareprimarilycontrolledbysedimentadvection,andan increasingadmixtureofcontinentalsediment.Secondaryprocesses(suchasverticalmixing,storage,crosscurrentflows)mayexplaintheoffsetfromasimpletwo-member linearmixingmodel.
core-sitesedimentationratesare nottheprimarycontrol onlocal 9Beor10Beconcentrations.
Thecorrelation betweenthe10Be concentrationswas explored against several measures of the biogenic componentof the sed- iment, including diatom abundance, opal and radiocarbon reser- voir ages. Diatom varve counts are available for most sub-ice shelf (Post et al., 2014) and a few open marine sites (Taylor and Leventer, 2003), while measurements of biogenic opal are only available for open marine sediments (O’Brien et al., 1995;
Harris et al., 1998). Both measures (diatom count and biogenic opal) correlate only weakly with both 10Be concentration (R2= 0.2)and10Be/9Beratios(R2=0.3).Berylliumconcentrationsand ratios are much more strongly correlated with the radiocarbon
‘age’, perhaps best expressed by the linear correlation between percentModernCarbon(pMC;definedasthemeasuredsamplera- diocarbonnormalisedtoatmosphericradiocarboncontentat1950 referencedtotheOxalic acid-IIstandard)and10Be concentrations (R2=0.5)and10Be/9Beratios(R2=0.8;Fig.5).
Tellingly,thelineartrendbetween10BeconcentrationandpMC insurfacesedimentsintersectstheorigin(Fig.5).Thisresultwould be expectedif 10Be concentrationsare a product of two compo- nent mixing, withendmembers beingdominatedby terrigenous sediments(low/nearzero10BeandpMC)andbiogenicmarinesed- iments(high10BeandpMC)respectively.Withinthesub-iceshelf cavity,both the 10Be concentrationsandpMCdecrease alongthe flow-pathofsub-iceshelfcirculation(Fig.1,Postetal.,2014),with thelowestvaluesofbothfoundatAM05ontheoutflowsideofthe cavity.
5. Discussion
5.1. Whatdeterminessediment10BeconcentrationsinAntarcticshelf environments?
Thehigh10BeconcentrationsfromunderneathAmeryIceShelf (AM01–6corestops)whichoverlapwithopenmarinevaluesare incompatible with a ‘calving dominant’ process of 10Be distribu- tionasshowninFig.1(top).Thiscontrastswiththatobservedby Yokoyamaetal. (2016) inastudyofRossIceShelfretreat.Here,we considerthatallthreeprocessesdepictedinFig.1–calving,sub- glacialmelt, andstrongbottom(sub-surface cavity)currents may allhavearoleininfluencing10BeconcentrationsinthePrydzBay sub-iceshelfregion.
Toaidinterpretationofthe10Bedistribution,weuseatwoend- membermixingmodel,where sedimentis derived fromtwopri- marysources:(1)openmarinesettingsand(2)continentalsources derivedfromiceerodingbedrockandoldersediments.Thepropor- tion of sediment mixing is estimated using end-member source values for 10Be and 14C as proxies along the clockwise current gyre commencing atthe continental shelf break to the southern groundingline,returningnorthwardstotheshelfbreakforatotal distance of about 1400km. The model successfully captures the first-order mixingbehaviour ofthe two isotopesseen in theraw dataofFig.6,suggestingatwo-componentmixingistheprimary controlonthe14Cand10Beconcentrationsinthesediments.How- ever,giventhepotentialcomplexityinthesystemwealsoidentify instanceswhereother factorsmayhaveinfluencedconcentrations oftheseisotopesinsedimentsinPrydzBay.
Sedimentoriginatingfromtheopenmarineenvironmentispri- marily biogenic, with thepotential fora mineral component de- rivedfromicerafteddebrisanddust.10Beconcentrationsinthese environments are a function ofthe 10Be deposition rate(derived from direct atmospheric deposition and ice melt) and the sedi- ment mass flux. The maximum core top 10Be concentrations in open marine settings in Prydz Bay (∼800–1000 × 106 at/g) are likethosedepositedinthemodernopenocean(vonBlanckenburg etal.,2015).However,theaverageconcentrationinPrydzBaysed- imentsarelowerthanthoseobservedintheRosssea(Sjunneskog etal.,2007) (Fig.5).Thus, thetrue10Beconcentration ofrecently producedbiogenicsedimentinPrydzBaymaybepartiallymasked bybottomcurrentsredepositingreworkedsedimentfromthecon- tinental shelfwithlower 10Beconcentrations.However, thesimi- larityinthecore-topsedimentsacrosseastern PrydzBaysuggests thatifthisisthecase,thesedimentspresentlybeingdepositedon the continental shelf are relatively well mixed, eitherin the wa- tercolumnorbypost-depositionalreworkingbybottomcurrents.
Basedonthisinterpretation,we assigntheendmembervaluesin our mixingmodelforopen marinesedimentsof 74%modern for 14Cand877×106 at/gfor10Be,calculatedastheaverageofcore top-sedimentsfromPrydzChannel(GC22–27).
Sedimentderivedfromprimary continentalsourcesisassumed to be dominantly lithogenic, sourced by glacial erosion of pre- Cenozoic (and largely Precambrian) bedrock that lies underneath most of the modern Lambert Glacier–Amery Ice Shelf drainage system (Mikhalsky et al., 2001). The subglacially derived sedi- mentisexpectedtohaveverylow 10Be concentrations,reflecting the lithogenic sources, although subglacial melt of meteoric ice
may contribute some 10Be to the sediment (Graly et al., 2018).
Thesesourcesedimentsarealsoexpectedtohaveabiogeniccom- ponent derived from Cenozoic and Mesozoic sediments present underneath the modern outlet glaciers (e.g. Mishra et al., 1999;
Whiteetal., 2011),which wouldincludean organiccarbon frac- tionwitheffectivelyzeropMC.Sub-glacialbasaldiamictssampled fromthePrydz Baycontinentalshelf have10Be concentrationsof
∼100 × 106 at/g (Table 1, GC cores) withlittle spatial variation ortemporalvariationduring formerglacialadvances acrossPrydz Channel (Guitard, 2015). Concentrations from the basal diamicts further south underneath the Amery Ice Shelf (∼30 × 106; Ta- ble 1, AM0 cores) are lower than in Prydz Bay, likely reflective of an increased continental component, which dilutes the open marine signal or alternatively a reducedinput ofreworked open marine material. None of the subglacial sediments we measure frombeneath theAmery Ice Shelf have10Be concentrations that areaslowasthosemeasuredinmodernsubglacialsettings(∼8× 106atoms/g,Sjunneskogetal.,2007),suggestingthatwemaynot havesampledthisprimary source.Forourcontinentalendmem- berinourmixingmodel,weutilizethebasaldiamictsfromunder theiceshelf(AM01&4,10Beof34×106 at/g,assumedradiocar- bondead).
Weinterpret 10Be and 14C concentrationsatour samplinglo- cationsalong theclockwisebottomcurrentflow-linethatfall be- tweenthesetwocomponentsources (i.e.subglaciallithogenic,34
×106 at/g,andzeropMCwithopen marine877 ×106 atoms/g and74%pMC)tobeprimarilytheresultofatwo-sourcesediment mixing process (see Fig. 6). Mixing could occur either by (hori- zontal)mixingofprimarysourcesinthemodern-dayenvironment viaoceancirculation,orby(vertical)mixingofsedimentbyglacial deformationofunderlying,andthuschronologicallydifferinghori- zons.
Both the 10Be and 14C concentrations incore topsunder and nearthe moderniceshelf provideevidenceforhorizontalmixing ofthe two primary sediment sources on length scales ofseveral hundredkilometres(Fig.6).Someofoursub-iceshelfsitesonthe eastern,inflow side ofthe shelf (AM01, 2and3) have 10Be con- centrations identical to those deposited in modern open marine environments in Prydz Bay (GC24, 27 and 29), and thus are in- ferred to have sourced almost all their sediment fromadvection fromtheopenmarineenvironment.Thesehighconcentrationsare primarilytowardthefrontoftheiceshelf,butalsointhecentral portion(AM03)wheremodelledsub-iceshelfcirculationisdirectly fromPrydz Bay(Galton-Fenzi et al., 2012) with little interaction withthe ice shelf margins. Other proxies in these sites such as diatomconcentrations (Postet al., 2014) indicatethat sediments deposited at AM01, 2 and 3 are also compatible with a wholly openmarinecomponentofthesediment.
Mixing-modelvaluesusingtheradiocarbonconcentrationssug- gestalower openmarinecomponentinsomeofthesub-iceshelf core-topsthan10Be(AM01 & 3,Fig.6), whichmaybe afunction ofeitherdifferentialtransportpathwaysfor10Beand14C,orsome reworking of late Quaternary open marine sediment fromwhich 14Chasdecayed.
Furthersupportforthebottomcurrentadvectionmodelispro- vided by the 10Be and 14C concentrations at AM05, which sits on a modelled outflowzone (∼1200km along the mixing line).
Whileconcentrationsarelow(∼250×106at/g),theyneverreach theverylowconcentrationsseeninmodernsubglacialconditions in west Antarctic outlet glaciers (Sjunneskog et al., 2007). There are two possibilities for thelack of a dominantcontinental 10Be signature in the sub-ice shelf sediments atthis location;(1) ad- vection ofsediment fromtheopen marine settingextends asfar south as∼ 200km from themodern iceshelf edge against the outflow zone, so none of our sites are truly dominated by sub- glacialsediment(2)10Beisreleasedfrommeltofthebaseofthe
iceshelf,andattachestocontinental-derivedsedimentdeliveredat thegroundinglineandmovedbyoutflowcavitycurrentsinquan- tities sufficient to provide a ‘mixed’ 10Be signal (i.e. the sub-ice shelfmelt processofFig.1).Weconsiderthereissupportingevi- denceforbothprocesses.AM05haslittletonodiatomabundance, suggestinglimitedadvectivetransportofclayorsiltsizedparticles fromopenmarinesettings,whichwouldmeanthe10Beatthissite isderivedfromlocalsources(i.e.sub-iceshelfmelt).However,ra- diocarbonconcentrationsatthissiteimplyasmallfractionofthe organiccomponentofthesediment (∼10%)isderivedfrommod- ernsources,supportingadvection(Postetal.,2014).Ineithercase, themeasured10Beconcentrationsintherangeof∼200×106at/g observed atAM05inthe outflowarea, andthecore-topforRISP 79–14at the rear ofthe RossIce Shelf (Sjunneskog etal., 2007) mayapproximatethelowerrangeof10Beconcentrationsbeingde- positedundermodernAntarcticiceshelves.
Thisdistributionof 10Be, incombinationwithsimilar patterns for other proxies of sediment transport (diatoms & radiocarbon), providesupportfortheconceptualmodelthat strong-bottomcur- rentsareakeyprocessindetermining10Beconcentrationsinsub- iceshelfenvironments.
5.2. Beisotopesastracersofpaleo-iceshelvesandsub-iceshelf circulation
For ice-shelveswith morelimited sub-ice shelf current circu- lationandwherecurrentvelocitiesarelowerthanneededforsilt suspension and transport, the dilution of the ice shelf 10Be sig- nalmaynot beassignificant aswe observeinPrydz Bay.Hence, in those regions where advection is low and/or circulation does nottransfersedimentfromtheiceshelfcalvingmargintowardthe groundingline,shiftsin10Be concentrationsthroughtimemaybe representative of,andthus interpreted as, a changefrom sub-ice shelftoopenmarinesettings(e.g.Yokoyamaetal.,2016).
In contrast to the above, in regions withstrong sub-ice shelf currentsandsediment advection, itmaynot be possibletoiden- tifyaspecific10Beconcentration(orrange)thatisuniquetoeither openmarine orsub-iceshelfsettings.Thisisthecasewe observe in theAmery IceShelf – Prydz Baydata.In thiscase, the distri- butionof 10Be isprimarily affectedbythe strengthanddirection ofsub-iceshelfcirculation.Thus,thereispotential forusing10Be to identify whether or not changes in sub-shelf circulation have influencedpaleo-icesheetfluctuations.
Where strongsub-shelf circulations arepresent, the combina- tion ofboth Be isotopes, inthe Prydz Baydata, i.e.the 10Be/9Be ratio, appears to provide a better distinction between the three environments than 10Be in isolation. Due to the lower ‘reactive’
9Be concentrations (i.e.HCl extractable)inopen marine coretop sediments, corresponding 10Be/9Be ratios are twice that of sub- iceshelfsediments,whichinturnarefivetimeslargerthanratios insubglacialbasalsediments.Moreover,the10Be/9Be displayless variability within each environment than either of the isotopes alone. We attribute both the spatial pattern and relatively low variability of 10Be/9Be to (a) input of 9Be into the sub-ice shelf cavityfromaterrestrialsource(b)thelongresidencetimeofbot- tomwaters (2 yrinthecavity)andpotential foreddycirculation (Galton-Fenziet al.,2012), whichcan thusproduce a well-mixed watermassinthecavity,and(c)equilibrationofthe10Be/9Beratio characterizing the sediment source withthe(dissolved) 10Be/9Be of the water column while the sediments are transported (e.g.
Brown etal., 1992). Thus, the combination ofthe two process – mixingandequilibration–resultsinsediment10Be/9Beratiosthat aremorehomogeneous withineachenvironmentthanacrossenvi- ronments.Thisexplainstheconsistencyof10Be/9Bewithineachof thethreeenvironmentsdespiteclearlyfarlargervariabilityin10Be abundances. Thus, we conclude that 10Be/9Be ratios are a more