ContentslistsavailableatChemicalGeologyjournalhomepage:www.elsevier.co - PDF document

ContentslistsavailableatChemicalGeologyj
ContentslistsavailableatChemicalGeologyjournalhomepage:www.elsevier.com/locate/chemgeoThetraceelementandSr-Nd-PbisotopegeochemistryofJuanFernandezlavasrevealvariablecontributionsfromahigh-HemantleplumeThiB.Truong,PaternoR.Castillo,DavidR.Hilton,JamesM.D.DayScrippsInstitutionofOceanography,UniversityofCalifornia,SanDiego,LaJolla,CA92093-0212,USAARTICLEINFOCorrespondingauthor.Presentaddress:CollegeofEarth,Ocean,andAtmosphericSciences,OregonStateUniversity,Corvallis,OR,97331,USA.E-mailaddress:(P.R.Castillo).=valuerelativetoairHe)inHawaiianlavasusedtobolsterthedeepmantleorplumeoriginofOIB(CraigandLupton,1976).Inthismodel,portionsofthelowermantlehavenotyetundergonesig-cantmixingwith,and/ormayberelativelyundegassedrelativeto,theuppermantlesotheyretainagreaterproportionoftheirprimordialvolatilecontent(CraigandLupton,1981).HeliumandSr-Nd-Pbiso-topescancorrelateinOIB(e.g.,Farleyetal.,1992;HananandGraham,1996;Hiltonetal.,1999).Consequently,theplumemodelinvokesacommon,high-HedeepmantlecomponentthathasbeentermedPHEM(primitiveheliummantle-Farleyetal.,1992(common-HananandGraham,1996)ortheaforementionedFOZO(Hartetal.,);thesetermsdieronlyinthedetailsofSr,NdandPbisotopes,ifatall(Hofmann,2014).Thus,OIBwithhigh-Heoeruniquein-sightsintothemagmasourceandevolutionoflinearvolcanicchains.Theeast-westtrendingJuanFernandezlinearvolcanicchaininthesoutheasternPaciFig.1)isanimportantcasestudyforheliumisotopicvariationsinOIBbecauseitslavasspanarangeinvaluesbetween7.8and18.0RFarleyetal.,1993).ThisvariationspansthetypicalratioofMORB(8±1RGraham,2002)intotherangeofthehigherratiosmeasuredforsomeintraplatelavas�(9anduptoca.50RStuartetal.,2003).AswithmodelsforHawaiianmagmaticevolution,anearlypopularexplanationisthatastationarymantleplumeformedtheJuanFernandezlinearvolcanoesbecauseoftheeastwardmovementoftheNazcaplateovertheJuanFernandezplume,creatingnewvolcanoesinthewestandcuttingosupplytoolderonesintheeast(Bakeretal.,1987;Farleyetal.,1993Resultsoftheseearlyinvestigations,however,alsoshowedfairlyhomogeneousSr-NdisotopicratiosofJuanFernandezlavas(etal.,1986;Farleyetal.,1993).ThismeansthatdespitetheirlargerangeofHeisotopicvariation,JuanFernandezlavasshowlimitedtonocorrelationbetweenHeandSr-Ndisotoperatios.SimilartootherHeOIB(e.g.,fromHeard,Samoa,Iceland,andHawaii),however,thehighestHevaluesofJuanFernandezlavasgenerallyconvergeatthePHEM,CorFOZOmantlecomponent(e.g.,Farleyetal.,1992;HananandGraham,1996;Hiltonetal.,1999;Castilloetal.,).AsFOZOistheconvergenceofalmostallOIB,itisperhapsvolumetricallyandcompositionallythemostdominantcomponentofmantleplumes(Castillo,2015,2017)and,thus,needstobebetterunderstood.Ontheotherhand,Natland(2003)arguedthattheHecontentsofolivinecrystalsinJuanFernandezlavaswerenotderivedfrommantleplumesbutwereinsteadindicativeofvolatilesthatper-meatedshallowmagmareservoirsandsubsequentlybecametrappedincrystallizingolivine.Thisnotion,togetherwiththeatypicallackofHecorrelationwithSr-Ndisotopes,hadbeenusedasamajorargumentinfavoroftheplumecounter-proposalthatOIBcompositionisnotdenitiveofdistinctmantleend-componentsand,thus,isnotdirectlyconnectedtodeepmantleplumes(Foulger,2011andrefer-encestherein).Inthisstudy,wefurtherexaminethepetrologyandgeochemistryofJuanFernandezlavastobetterconstraintheirpetrogenesisandthenatureoftheirmantlesource.OurnewtraceelementandSr-Nd-Pbisotopicanalyses,combinedwithavailableHeandSr-NdisotopesFarleyetal.,1993),indicatethatsamplingthroughvariabledegreesofpartialmeltingandinherentheterogeneityofthenominallyHemantleplumesource(orFOZO)canexplainthedistinctlavagroupsofJuanFernandezlavasdespitetheirlimitedSr-Nd-Pbisotopevariation.2.GeologicbackgroundandsamplesanalyzedTheislandandseamountchaincomprisingtheJuanFernandezlinearvolcanoesislocatedatlatitude34°SinthesoutheasternPacisome660kmwestoftheChileancoast(Fig.1).Thechainspansabout800kmontheca.22to37MaoldNazcaPlate(Bakeretal.,1987;Corvalan,1981;GrippandGordon,2002;RodrigoandLara,2014).TheJuanFernandezIslandsconsistoftwomainislands180kmapart,Ro-binson(R.)Crusoe(alsoknownasMasaTierra,closertotheland)andAlexander(A.)Selkirk(MasAfuera,awayfromtheland),andasmallislandlocatedsouthwestofR.Crusoe,SantaClara(Bakeretal.,1987;Farleyetal.

,1993;Deveyetal.,2000;RodrigoandLara,201
,1993;Deveyetal.,2000;RodrigoandLara,2014).Twoprominentseamounts,FridayandDomingo,liewestofA.Selkirk.Thestratigraphyoftheislandsispoorlyknown,butradiometric(K-Ar)agesforR.Crusoelavasrangefrom5.8to3.1MawhereasthoseforA.Selkirkrangefrom2.5to0.85Ma(Bookeretal.,1967;Stuessyetal.,1984;Bakeretal.,1987).DredgedfreshbasaltssuggestFridayandDomingoseamountsareyoungerthanA.Selkirk(Farleyetal.,1993Deveyetal.(2000)alsonotedthatthelargerFridaySeamountappearsmorphologicallyevolvedwithevidenceofvolcanicactivityoveraFig.1.BathymetricmapofJuanFernandezIslandsfromFarleyetal.(1993)T.B.Truongetal.broadareaandislikelytobeolderthanDomingothatliestothewest.Moreover,DomingoSeamounthaslesssedimentcover(Deveyetal.,).Theagesofislandsamplesaswellasobservationaldatafortheseamountsareconsistentwiththeproposedeastwardage-progressionofvolcanismintheJuanFernandezvolcanicchain(Farleyetal.,1993;Deveyetal.,2000;GrippandGordon,2002WefocusedonthesuiteofsamplesfromR.CrusoeandA.SelkirkthatwerecollectedduringLeg1oftheScrippsInstitutionofOceanography(SIO)HYDROSexpeditionaboardR/VMelvillein1988.Majorandselecttraceelementcompositions,lithologicandeldde-scriptionsaswellasHe-Sr-Ndisotopicdatahavebeenreportedpre-viously(Farley,1991;Farleyetal.,1993).Olivinegrainsfromsomeofthesampleshavealsobeenanalyzedforoxygenisotopes(Eileretal.,).Basedonnormativemineralogy,thesamplesconsistofolivinetholeiites,alkalibasaltsandbasanites.Moreover,thesampleshadpreviouslybeenclassiedintofourgroupsnamelyI,II,III,andIVbyFarleyandco-workers(Farley,1991;Farleyetal.,1993).GroupsIandIIconsistofbasaltsandbasanites,respectively,fromR.CrusoewhereasGroupIIIconsistsofA.Selkirkbasalts,whicharelithologicallyandchemicallysimilartoGroupIbasalts.ThefewGroupIVlavasconsistofFridaySeamountalkalibasaltsthatarereportedtobelikeR.CrusoeandA.Selkirkbasalts;theyarenotincludedinthisinvestigation.Bakeretal.(1987)initiallyclassiedeightgroupsofJuanFer-nandezlavasbasedongeographicdistributionofrockunits,butlateradoptedtwobroadgeochemicalgroupsforR.Crusoe:centralandperipheral.Centralgroupconsistsofpicriticbasalts,aphyricalkalicbasaltsandquartztholeiiteswhereasperipheralgroupincludesolivinetholeiites,alkalibasaltsandbasanites.TherecentandmoredetailedworkonR.CrusoebyReyesetal.(2017),however,favorsacentralstructureschemewithagentlydippingshieldsequencecross-cutbydykecomplexes(shieldunit)thatisseparatedbyanunconformityfromtheoverlyinglavasandpyroclasticbedseruptedfromseveralsmallandisolatedvents(rejuvenatedunit).Signicantly,thebasanitesoccurasdikesandasparasiticconesalongthesouthwestpeninsula(etal.,1986).GroupIbasaltsandthelithologicallyandcompositionallyakinGroupIIIbasaltsfromA.SelkirkinvestigatedherebelongmainlytotheshieldunitwhereasGroupIIbasanitesarepost-shieldlavasthatbelongtotherejuvenatedunitofReyesetal.(2017)3.AnalyticalmethodsSeventeenlavaswereanalyzedfortheirtraceelement(Rb,Sr,Y,Ba,Pb,Th,rareearthelements(REE),andhigheld-strengthelements(HFSE)Nb,Ta,Zr,andHf)andPbisotopiccompositions.StrontiumandNdisotopeswerealsoanalyzedforonesample(PF-21)toreplicatethendingsbyFarleyetal.(1993),andforthreesamples(PF-3,MF-16,MF-6)thathadnotbeenanalyzedpreviously.Leadisotopeswerede-terminedforall17lavas.Allanalyseswereperformedonwhole-rocksamples.Neodymiumisotopesofolivineseparatesfromvesamples(MF-3,PF-21,PF-5,PF-10,PIN-8)spanningtherangeofHeratiosfrom7.8to18.0Rwerealsoanalyzed.Fortheanalyses,thefreshinteriorportionsofthesampleswerecrushedtocm-sizefragments,ultrasonicallycleanedinaca.5%HNOsolutiontocleanseanycontaminationsuchasmetalsandextrapowderfromcrushing,anddriedinanovenatca.110°C.Oncedried,thefragmentswerepowderedinanaluminaceramicshatterbox.Traceelementsweredeterminedonunleachedpowdersbyinductively-cou-pledplasmamassspectrometry(ICP-MS)usingaThermoQuestElement2instrument.Strontium,NdandPbisotopeswerealsoanalyzedonunleachedpowdersthroughthermalionizationmassspectrometry(TIMS)usingaMicromassVG54instrument.Thesampleswerenotleachedbecausemostofthemarerelativelyunaltered,asevidencedbythepresenceofgroundmassandphenocrysticolivine.Moreim-portantly,wewanttodirectlycombineouranalyseswithpreviousmajor-traceelementandSr-NdisotopedatacollectedfromunleachedsamplesbyFarleyandco-workers(Farley,

1991;Farleyetal.,1993Thesampledigestionp
1991;Farleyetal.,1993ThesampledigestionprocedureusedfortheICP-MSandTIMSanalysesislikethatdescribedinJanneyandCastillo(1996).Fortraceelementanalysis,about20mgofsamplepowderwasdissolvedinacleanTebeakerusingadoubly-distilledHF:HNO(2:1)mixture,driedoverahotplate,andevaporatedtwiceinconcentratedHNO.Thedissolvedsamplewasdilutedca.4000×witha1%HNOsolutioncontaining1ppbInasaninternalstandardtocorrectforsignalrunintensityvariationandinstrumentaldrift.ForPb,SrandNdisotopeanalysis,eachsamplewasdissolvedasaboveandthenre-dissolvedin4.5NHCl,dried,thendissolvedagainin2.5NHBr,dried,andnallypassedthroughacationcolumninaHBrmediumtocollectPb.Theleft-oversamplesolutionfromthePbcolumnwasdriedandloadedintoaprimarycationcolumninachloridemediumtocollectSrandREE.Finally,theREEcutwaspassedthroughanionexchangecolumninanalphaHIBA(alpha-hydroxyisobutyric)acidmediumtocollectNd(JanneyandCastillo,1996;Castilloetal.,OlivineseparatesanalyzedforNdisotopeswerehand-pickedunderabinocularmicroscopefromaliquotsofthecm-sizedfragmentsofwholerocksamplespreparedbyFarley(1991).Approximately70mgofseparateswerethenultrasonicallycleanedwitha1%HNOanddriedinanovenatca.110°Cpriortodissolution.Theolivinese-paratesweredissolvedinthesamewayasthewholerockpowders.Theseparateswerenotcrushedpriortodissolution.Neodymiumwasse-paratedfromtheolivinesolutionasdescribedaboveandanalyzedthroughTIMStodirectlycomparetheirNdandcompositions.DespitethelowconcentrationofNdinolivine,Ndwasmeasuredontheseparates.Themeasurements,however,tooklongerandtheratiostypicallyhavelargeruncertaintiesinstrumentalerrorupto±32Table1)comparedtothoseofthewholerocks.AdditionaldetailsoftheanalysesplusanalyticaluncertaintiesarereportedasfootnotestoTable14.Results4.1.TraceelementsThetraceelementcompositionsofthesamplesarepresentedinTable1.TheirBa/ZrratiosareconsistentwiththesubdivisionofthesamplesintoGroupI(averageBa/Zr=0.87),GroupII(averageBa/Zr=1.67)andGroupIII(averageBa/Zr=0.59;Farleyetal.,1993Fig.2A).TheonlyexceptionissamplePV-2,whichFarleyandco-workersdesignatedasbasanitebasedonitsca.5%normativenephe-line,butnewresultsshowthatitsBa/Zr(1.05)aswellasothergeo-chemicalcharacteristicsplotclosertoGroupIbasalts.Were-classifythissampleasGroupIbasanitoid.ZirconiumalsocorrelateswellwithNbandtheoverallNb/Zrrange(0.110.27)isconsistentwithhighratios�(0.1)ofOIB(Fig.2B).GroupIIbasaltshavethehighestNbandZrcontents�(58and�228ppm,respectively)whereasGroupsIandIIIhavesimilarlylowervalues(42ppmNb,253ppmZr).GroupIhaslowerNbforgivenZrvaluesthanGroupII.JuanFernandezlavashavelight-REEenrichedprimitivemantle-normalizedpatterns,alsotypicalofOIB(Fig.3A).MostofthesamplesplotwithintherangeofpreviousREEanalysesbyBakeretal.(1987)anddisplaythesamethreegroupsasproposedbyFarleyandco-workers.Samplesfrombothislandshaveroughlysimilaraveragelight-tomedium-REEslopes,buttheiraverage(La/Sm)valuesslightlydi(GroupIII=2.0±0.3,GroupI=1.9±0.3andGroupII=3.4±0.3).Moreover,thenormalizedconcentrationpatternsshowthatGroupIIIislessfractionated(average(La/Yb)=9.0)thanbothGroupsIandII(GroupIIaverage(La/Yb)=20.2).MostsamplesrevealaslightlypositiveEuconcentrationanomaly,orEuenrichmentrelativetoadjacenttraceelementsinnormalizedplots,withGroupIIIshowingthegreatestanomalywhereasGroupIhasthesmallest.Theprimitivemantle-normalizedspidergramshowstheentirelavaT.B.Truongetal.Table1TraceelementandisotopedataforJuanFernandezlavas.SamplePV-2PV-4PF-3PF-5PF-10PIN-12GroupIIIIIIRocktypeRb(ppm)0.703698(12)0.512835(5)0.512808(14)0.512827(9)RocktypeRb(ppm)0.7036500.7034600.703430(12)0.7035100.7034200.7035600.5128560.5128960.512870(6)0.5128960.5128760.5128860.512824(9)0.512837(9)T.B.Truongetal.suiteexhibitssub-parallelpatternsthatareenrichedinhighlyin-compatibleelements,alsotypicalofOIB(Fig.3B).GroupsIandIIIagainhavenearlyidenticalconcentrationpatternswhereasGroupIIisdistinctivelymoreenrichedinhighlyincompatibleelements(e.g.,BaandThuptoca.90×primitivemantle),consistentwithpreviousre-sults(Bakeretal.,1987;Farleyetal.,1993).AllgroupsalsodisplaypositiveNbandTaconcentrationanomalies.GroupsIandIIIbasaltslikewisedisplayapositiveTianomalybutGroupIIdoesnot.Inotherwords,thebulkofJuanFernandezlavasexhibitenrichmentinTi,Taan

dNb,ortheso-calledpositiveanomalies,whic
dNb,ortheso-calledpositiveanomalies,whichareperva-sivefeaturesofOIB(Jacksonetal.,2008;PetersandDay,20144.2.Strontium,neodymiumandleadisotopesTheSr,NdandPbisotopiccompositionsofJuanFernandezlavasarepresentedinTable1andshowngraphicallyinFigs.4and5.OurnewandexistinganalysesbyFarleyetal.(1993)indeedshowlimitedvariationsinbothSr(0.703410.70370)and0.51291)values(Farley,1991),butonaverageareslightlymoredepleted(lowerSr,higherNd)thantheanalysesGerlachetal.(1986Fig.4).Indetail,GroupIsamplesspantheentirerangeofSrvaluesalthough,onaverage,GroupIIhasthelowest(0.70351)andGroupIIIhasintermediate(0.70356values.GroupIalsohasthelargestrangeofNdratios,butthelowestaveragevalue(0.51286).GroupIIhasamoderaterangeofNd(0.521870.51290)whereasGroupIIIhasthehighestNdratios(upto0.51291).Overall,GroupIisrelativelythemostisotopicallyheterogeneousintermsofSrandNdisotopiccom-position;GroupIIcompletelyoverlapsitsmostdepleted(lowandhighNd)end.GroupIIIisrelativelyhomogeneousandhashigherNdforgivenSrthanGroupsIandII.AlsoshowninTable1aretheNdisotopiccompositionsofolivineseparatesfromatleastonesamplefromeachlavagroupthatspanstherangeofHe(generallyfrom8to18R).TheNdvaluesmeasuredonthreeolivineseparates(PIN-8,MF-3,PF-10)aresimilarwithintheanalyticaluncertainties(±0.000016foreach)ofrespectivewholerockvalueswhereastheothertwo(PF-5,PF-21)areoutsidetheanalyticaluncertainties.ThelattertwohaveslightlylowerNdthanthewholerockvalues.ThenewPb,andPbvaluesforJuanFernandezlavasoverlapwithpreviousanalysesofGerlachetal.(1986Fig.5).Asawhole,however,theyareslightlythoughsystematicallyhigherinPb,andPbforagivenPbthanthelatterdataset,althougharestillrelativelylimitedincomposition.Indetail,ourGroupsIandIIsamplescompletelyoverlapinPbvalues,liketheirSrandNdisotopes,andaremoreRocktypeHawaiianbasaltRb(ppm)0.703614(11)0.703601(12)0.7036700.7035700.512880(5)0.512869(14)0.5128860.5128960.512886(32)Reproducibilityofthetraceelementanalyseswasbetterthan5%basedonrepeatedanalysesofrockstandardsanalyzedasunknowns.Theaccuracyofthestandardiswithin10%ofthesuggestedvalues(BHVO-1,=33),butgenerallybetterthan±5%forrareearthelements(REEs).Strontium,NdandPbisotoperatioswereanalyzedusinga9-collector,MicromassSector54thermalionizationmassspectrometer(TIMS).Totalproceduralblanksare35pgforSr,10pgforNdand60pgforPb.Strontiumisotopicratioswerefractionation-correctedtoSr=0.1194andarereportedrelativetoSr=0.710254±0.000018(=22)forNBS987duringtheperiodofanalysis.Neodymiumisotopicratiosweremeasuredinoxideform,fractionationcorrectedtoNdO=0.72225(Nd=0.7219)andarereportedrelativetoNd=0.511856±0.000016(=19)fortheLaJollaNdStandardduringtheperiodofanalysis.NumbersinparenthesesafterSrandNdareinstrumental(2)errorsandrefertothelastsignigures.StrontiumandNdisotopevaluesfromFarleyetal.(1993)havebeennormalizedtothemeasuredstandardvalues(Sr=0.710254;Nd=0.511856).Leadisotopicratioswereanalyzedusingthedouble-spikemethodtocorrectformassfractionationduringanalyses;separatemeasurementsofspikedandunspikedsamplesweremadeondierentaliquotsfromthesamedissolution.TheSBL-74Pbdouble-spikefromtheUniversityofSouthamptonwasused.Duringtheanalysisperiod,themethodproducedthefollowingresultsforNBS981:Pb=16.9300±0.0020,Pb=15.4896±0.0027andPb=36.6999±0.0086(=11).2precisionsforindividualrunsarebetterthanthese.BSN=basanite;A.B.=alkalibasalt;O.T.=olivinetholeiite.ValuesinitalicsarefromFarleyetal.(1993)T.B.Truongetal.radiogenicthanGroupIII.ThereisnoapparentdistinctionamongthegroupsintermsofPbratios.JuanFernandezPbisotopicar-rayslayparallelandabovethenorthernhemispherereferenceline(NHRL)anddonottrendtowardthePacicDMMsource.5.DiscussionDespitelimitedSr-Nd-Pbisotopicvariation(Figs.4and5),thenewtraceelementdatashowlargevariationandsupportthepreviouschemicalclassicationschemefortheR.CrusoeandA.SelkirklavasFigs.2and3).Moreover,majorelementdataarealsovariableandshowthatJuanFernandezbasalttotrachytelavasdenecoherenttrendsorliquidlinesofdescentinmostMgOvariationdiagrams.Thesetrendsindicatethatfractionalcrystallizationofindividualbatchesofparentalbasalticmagmaseruptedatdierenttimesinthetwoislandsismainlyresponsibleforsomeofthecompositionalvariationswithinandamongthelavagroups(Truong,2015).Ontheotherhand,thetraceele

mentvariationsofthelavas(Fig.2)plustheva
mentvariationsofthelavas(Fig.2)plusthevariable,thoughlimitedSr,NdandPbisotopicvariations(Figs.4and5)suggestthatthepar-entalbasalticmagmasofJuanFernandezlavasoriginatefromamantlesourcewithaslightlyheterogeneouscomposition(seealsoFarleyetal.,).ToarstorderthemajorandtraceelementvariationsoftheparentalmagmasofJuanFernandezlavagroupscanbeascribedtopartialmeltingofacommon,slightlyheterogeneous,mantlesource.5.1.ElementalvariationsasafunctionofpartialmeltingOurJuanFernandezlavasuiteisbiasedtowardpicriticcomposi-tions�(13.5wt%MgO)duetopreferentialsamplingofolivine-accu-mulativerocksforHeisotopeworkand,thus,doesnotrepresenttheliquiduscompositionofprimarymagmas(Farley,1991;Farleyetal.,1993;Bakeretal.,1987).Qualitatively,GroupIIhasthehighestNaforgivenMgOvaluesamongthebasaltsandalsohasslightlyhigherFeO*foragivenMgOthanGroupI(Truong,2015).Thesecomposi-tionalfeaturesgenerallysuggestthatGroupIIbasanitesweremostlikelyproducedbyarelativelysmallerdegreeofpartialmeltingathighpressure(e.g.,KleinandLangmuir,1987;NiuandO'Hara,2008).Ontheotherhand,GroupIonaveragehasalargerangeofNaOandin-cludesthelowestNaOpluslowFeO*,whereasGroupIIIhasuniformandintermediateNaOandhighestFeO*.ThesecharacteristicssuggestGroupsIandIIIwereproducedbymoderatedegreesofpartialmeltingatlowtomoderatepressures(e.g.,KleinandLangmuir,1987;NiuandO'Hara,2008GroupIIbasaltsarethemostenrichedinhighlyincompatibletraceelementsandGroupIisslightlymoreenrichedthanGroupIII.TheincompatibletraceelementvariationsofJuanFernandezlavaswhenplottedintheprocessidenticationdiagramsofAllègreandMinster;seealsoFeigensonetal.,1983)showthatGroupIIbasaniteswereproducedbylimiteddegreesofpartialmeltingofthecommonsourcewhereasGroupIandIIIbasaltswereproducedbyhigherdegreesofpartialmelting(Fig.6AandB).ThisisconsistentwiththesteepREE0.250.500.751.001.251.501.752.00PV-2100150200250300350400Nb/Zr = 0.11ShieldGroup IGroup IIGroup IIIFig.2.He(R)versusBa/ZrdiagramforlavasfromJuanFernandezIslands(slightlymodiedafterFig.4AofFarleyetal.,1993),showingGroupsI,IIandIIIlavas.(B)Nb(ppm)versusZr(ppm)diagram.PV-2isaGroupIbasanitoid.Rock / ChondriteHo ErYb LuGroup III A. Selkirk (shield)10100YTbHfZrCeLaNbUThBaRbRock / Primitive MantleGroup I basaltsGroup II basanitesGroup III basaltsRejuvenatedShieldABFig.3.(A)Chondrite-normalizedREEconcentrationpatternsofJuanFernandezlavas.ShadedregionistherangeofvaluesforgroupIolivinetholeiites.(B)Primitivemantle-normalizedincompatibletraceelementconcentrationpatterns.RegionswithdishadesofgrayrepresentrangesofGroupsIandIIIlavas.Averageconcentrationsareshown.DatafromFarleyetal.(1993)areincludedandnormalizingvaluesarefromandMcDonough(1989)T.B.Truongetal.patternsandgreatestenrichmentsinhighlyincompatibleelementsofGroupII(Figs.3and4).Moreover,GroupIIIappearstohaveundergoneaslightlylargerdegreeofpartialmeltingthanGroupI.TheplotsalsoshowthatalargepartofthetraceelementvariationsofGroupII,eventhoughtheseareallhighMgO�(11wt%)basanites,falltotherightofthebatchpartialmeltingcurve,indicatingenrichmentduetovariabledegreesoffractionalcrystallizationorolivineaccumulation.Thelatterprocessmostprobablyoccurredduringtransitfromthedeepmantletothesurfaceasgeothermobarometricandtexturaldataindicatepoly-bariccrystallizationatawiderangeofpressures(Reyesetal.,2017FractionalcrystallizationisalsoconsistentwiththedistinctinversecorrelationbetweenMgOandNbinthebasanitesincontrasttothenearlyconstantNbcontentsfortherangeofMgOinGroupsIandIIIbasalts(Truong,2015).Signicantly,asimilardierenceinin-compatibletraceelementconcentrationsexistbetweencurrentGalá-pagosvolcanoesandtheirapparenttrackalongtheCocosRidgeandthishasalsobeeninterpretedasduetodierencesinextentofpartialmeltingofacommon,heterogeneoussource(Hoernleetal.,2000ModelcalculationsusingthevariationofLa/YbversusDy/Ybil-lustratethatGroupsIandIIIbasaltscanbeproducedby5to�10%Pitcairn-GambierHawaiiIcelandHIMUDMMEM1EM2SamoaMarquesas San Felix-St. Helena- 0.70330.70350.70360.7020.7030.7040.7070.708Group I basaltGroup II basaniteGroup III basaltFig.4.NdversusSrdiagramforJuanFernandezlavasrelativetosomeOIBfromtheGEOROCdata-base().InsetshowsdetailsoftheNdversusSrforJuanFernandez

eldintheinsetenclosesdatafromGerlachetal
eldintheinsetenclosesdatafromGerlachetal.,1986TheexpandedFOZOeldincludestheeldsofHofmann(1997)Stracke(2012);seetextforadditionalDMMEM1EM2HIMU19.119.3DMMEM1EM2HIMU19.119.31819HawaiiHawaiiPitcairn-GambierPitcairn-CanaryCanary St. Helena- Cook-Autral St. Helena- Cook-Autral San Felix- San Felix-Kerguelen-Kerguelen-FOZONHRLFig.5.PbversusPbandPbdia-gramsforJuanFernandezlavasfromthisstudyandGerlachetal..SymbolsasinFig.4.InsetsshowdetailsofJuanFer-nandezanalyses(eldintheinsetenclosesdatafrometal.,1986).FieldsforotherOIBfromtheGEOROCdatabase)areshownforcomparison.NHRListheNorthernHemispherereferencelineHart,1984).Thelocationsofthefourproposedmantleend-membersDMM,HIMU,EM1,andEM2arefromZindlerandHart;theexpandedeldsforFOZOaremodiedafterT.B.Truongetal.polybaricpartialmeltingofaspineltogarnetlherzolitesourceFig.6C).Onaverage,GroupIIIisproducedathigherdegreesofpartialmeltingthanGroupI.Ontheotherhand,GroupIIbasaniteshavethehighestLa/Ybandindicatethattheycanbeproducedatrelativelylowdegrees(26%)ofpartialmelting.Thus,modelcalculationresultsareconsistentwithGroupIandIIIJuanFernandezlavasbeingproducedbyahigherdegreeofpartialmeltingthanGroupIIbasanites.Insummary,themajorandtraceelementvariationsofJuanFernandezlavascanbeexplainedthroughvariationsofdegreeand,toacertainextent,pressure(depth)ofpartialmeltingofacommon,slightlyheterogeneousmantlesourcecharacterizedbyalimitedSr,NdandPbisotopicvariation(Figs.4and5).ThegeochemistryofGroupIIcanbemodeledastheresultofasmalldegreeofpartialmelting(+fractionalcrystallizationand/orolivineaccumulation)whereasthoseofGroupsIandIIIrepresentlargerdegreeofpartialmelting.5.2.TheSr-Nd-PbisotopiccompositionofthemantlesourceDespitethecleartraceelement-HeclassicationofJuanFernandezlavas(Fig.2a),thereislimitedSr-Nd-PbisotopicvariationFigs.4and5).Indetail,however,thecombinedSr,NdandPbisotopedatashowtwoslightlydierentthoughdistinctcomponentsintheirmantlesource(seealsoGerlachetal.,1986;Farleyetal.,1993).GroupIrepresentsthemostenrichedcomponent,whichonaveragehastheNdandmostradiogenicPbratios.AlthoughGroupsIandIIoverlap,GroupIIgenerallyoccupiesthelessradiogenicendoftheGroupIisotopicrange.Ontheotherhand,GroupIIIre-presentstheleastenrichedcomponent,whichonaveragehastheNdandleastradiogenicPbandofthethreegroups.FromSr-Nd-Pbisotopicvariations,thelong-livedradiogeniciso-topicratiosofJuanFernandezindicateamantlesourcethatiswithinorapproachestheputativeFOZOcomposition(Figs.4and5).TheoriginofFOZOiscontroversial(Hartetal.,1992;Haurietal.,1994;Hiltonetal.,1999;Strackeetal.,2005;HananandGraham,1996;Hofmann,1997;Stracke,2012;Hofmann,2014;White,2015;andreferencestherein).Forexample,FOZOmaybearegioninthemantlecontainingrecycledcrustalprotolithswithcontinentalPbandlocatedinthetransitionzoneabovethe670-kmdiscontinuity(HananandGraham,).Itmayalsorepresentancientrecycledoceancrustandlitho-spherethatareeectivelymixedwiththemantle(Dayetal.,2010Alternatively,astheNdandSrofFOZOarebetweenbulksilicateEarth(BSE)andDMM(Fig.4),itmayrepresentthelower,cportionofpreviouslysubductedoceanicslabs(2015,2016).Thatis,FOZOcouldbearepresentationofthelong-term(Ga)time-integratedrecordofthegeochemicaldepletionofuppermantlewithtime,fromBSEtomodernDMM.IntermsofPbisotopes,FOZOisayoungHIMUthatalsohasarangeofradiogenicPbisotopicratiosbetweenDMMandclassicHIMU(Fig.5Vidal,1992;Thirlwall,1997;Strackeetal.,2005;Castillo,2015,2016).Inotherwords,thevariableSr,NdandPbisotopiccompositions,aswellasthedistinctivelyhighandvariableHeratiosofFOZOcanbeacollective,inherentfeatureofsubductedoceaniclithosphericmantle.Insummary,thecombinedHe-Sr-Nd-Pbisotopesindicatethatpar-entalmagmasofGroupsIandIIlavasmostlikelyoriginatedfromFOZO.ParentalmagmasofGroupIIIbasaltsmostlikelyalsooriginatedfromFOZObutadditionallycontainedarelativelyless-enrichedorcomponent.Notably,Gerlachetal.(1986)suggestedasi-milar,dualmantlesourceforJuanFernandez.TheirsuggestionwasdeducedfromthecombinedisotopicdataforJuanFernandezplusthoseoftheaforementionedSanFelixandSanAmbrosioislands(Figs.5andGerlachetal.(1986)showedthattheseislandsdeneanarrayre-presentingamixtureofEM1andclassicHIMUend-components.Inthisstudy,however,theisotopicdataforJuanFernandezaswellasthoseforS

anFelix-SanAmbrosioislandsrepresentamixt
anFelix-SanAmbrosioislandsrepresentamixtureofadepletedandFOZO(oryoungHIMU)components.100% Spinel100% Garnet0.1%0.5%1%2%3%5%10%02040601.01.41.82.22.6Dy/YbNLa/YbNN-MORB4080400600P. M.P. M.F. C.F. C.Fig.6.A.andB.,MagmaticprocessidenticationdiagramsafterAllègreandMinsterforJuanFernandezlavas.SymbolsasinFig.4.PMrepresentsequilibriumbatchpartialmeltingandFCrepresentsfractionalcrystallization.C.La/YbversusDy/YbforJuanFernandezlavas.Apartialmeltingmodelforaprimitivemantlesourceisshown,usingtheparametersgiveninDayetal.(2010).Estimatesofpartialmeltingaremadeassumingaprimitivemantlesourcewithgarnetperidotitestartingmodalproportionsofolivine,orthopyroxene,clinopyroxene,andgarnetof0.598/0.211/0.076/0.115andaspinelperidotitewitholivine,orthopyroxene,clinopyroxeneandspinelof0.578/0.21/0.119/0.033,withappropriatephasesmeltingandpartitioning.T.B.Truongetal.5.3.GeochemicalandvolcanologicevolutionofJuanFernandezIslandsManymajorgeologicfeaturesofJuanFernandezIslandsaresimilartothoseofotherlinearislandvolcanicchains.Forexample,thegeo-chemicalevolutionofJuanFernandezIslandsgenerallyresemblesthatofJasperSeamount,whichbelongstotheFieberling-GuadalupeSeamountTrailintheeasternPacicthat,inturn,displaysmanyas-pectsofHawaiianevolutiondespiteitsmorelimitedsize(Konteretal.,).Fieberling-Guadalupedemonstratesthreeindividualstagesthatrepresentdecreasingdegreesofpartialmeltingofitsmantlesourcewithtime.AtJasper,themostvoluminousshieldbuildingstagewasfol-lowedbytheyoungerandsubstantiallylessvoluminousankseriesnallybytherejuvenatedsummitalkalicseriesthatarehighlyenrichedinincompatibletraceelements(Konteretal.,2009).Toaorder,thecombinedR.CrusoeandA.SelkirkdataindicateahistoryroughlyfollowingthesestagesofJasperevolution.ThegeochemicalevolutionofR.CrusoealsobearsgrosssimilaritieswiththatofTahaa,thelargestoftheSocietyIslands(WhiteandDuncan,1996).Tahaahasanagerangeof2.3Mawithavolcanichiatusof1.2M.y.betweentheshieldbuildingandpost-erosionalstages.Al-thoughthelatterstageinTahaahasaless-enriched(orrelativelyde-pleted)isotopicsignaturecomparedtotheearliershieldbuildingstage,whichisoppositeoftheshieldrejuvenated(orpost-shield)isotopicrelationshipinJuanFernandez,themainpointisthatSocietyIslandlavaswerecreatedbytwomantlecomponents.Aspresentedearlier,compositionalsignaturesofFOZOanddepletedcomponentsarere-cognizedinJuanFernandezlavas(seealso,Gerlachetal.,1986;Bakeretal.,1987BasedonthegeochemicalfeaturesofJuanFernandezlavasandtheirsimilaritieswithotheroceanicislandspurportedlygeneratedbymantleplumes,togetherwitheldobservationsandpetrographicdataBakeretal.,1987;Farley,1991;Farleyetal.,1993;Natland,2003;Gerlachetal.,1986;Truong,2015;Reyesetal.,2017),thefollowingmodelforthegeologicevolutionofJuanFernandezvolcanicchainisproposed(Fig.7).Theearliestphaseisthepre-shieldbuildingstage.Magmatismassociatedwiththisphaseisnotwelldened,butin-ferencesfromothervolcanicislandchainssuggestthatthisstagemostprobablyeruptsaheterogeneousmixtureoflavasthatprecedesthearrivalofthemainbodyofanupwellingmantleplume.Forexample,Deveyetal.(2000)reportedthatmagmasfromthesubmarineFridayandDomingovolcanoesarehighlyalkaline,mostprobablyduetosmalldegreesofpartialmeltingoftheplumesource.Thus,lavasatthisstagehaveavariablecompositionbecausethecorecomponentoftheplumehasnotbeensampledand/orthemainbatchofplumemelthasnotyetimpingedthelithosphere.Thisstagewouldberepresentedbythein-accessible,notyetsampled,lowermoststructuresofbothislandsandthesmall,western-mostseamountsalongthechain.Thenextphaseisthemainshield-buildingstageandisrepresentedbyA.SelkirkGroupIIIbasalts.Thebasaltsareproducedbyarelativelylargedegreeofpartialmeltingofthecoreofthemantleplume.ThisiscompatiblewithapreviousproposalthatthelowBa/ZrratioA.SelkirkGroupIIIbasaltsweretappingasourcesomewhatdierentfromthatofR.Crusoebasalts(Farleyetal.,1993).Atthisstage,themantleplumeyieldsmagmasthataremainlytholeiitictotransitionaltoalkalicbasaltsrelativelydepletedisotopicandincompatibleelementcomposi-Thelaterphaseoftheshield-buildingstageisrepresentedbyR.CrusoeGroupIbasalts.TogetherwithGroupIIIbasalts,theycomprisetheshieldunitbasaltsofReyesetal.(2017).However,GroupIbasaltswereproducedthroughlater,smallerdegreeofpartia

lmeltingoftheFOZOcomponentofthemantleplu
lmeltingoftheFOZOcomponentofthemantleplume.Assmallerdegreepartialmelts,theselavasarerelativelyenrichedinincompatibletraceelementsthanGroupIIIlavas.Finally,therejuvenatedstageisrepresentedbyminorintrusivesandlatevolcanics,suchastheGroupIIbasanitesinR.Crusoe.Thelatterlavasarethesmallestdegreepartialmeltscomingfromthetailendofthemantleplume,experiencedpolybaricfractionalReyesetal.,2017)andwereeruptedasparasiticconesandintrudedasdikesthatcross-cuttheunderlyingshieldlavas(etal.,1986;Bakeretal.,1987;Reyesetal.,2017).Accordingly,theycomprisethesmallvolume,moreSi-undersaturatedandincompatibleelement-enrichedlavas.AsGroupIIlavasintrudedpre-existinglavaformations,theyalsoexhibitthemostalterationorcrustalmodifeatures.ThisisconsistentwiththeslightlyhigherOcontentofthebasanitesrelativetootherJuanFernandezlavas;ahigherthannormalOcontentisgenerallyattributedtoassimilationofalteredlavasfromhigh-levelsinthevolcanicedice,assimilationoftheoceaniccrustunderlyingthevolcano,orincorporationofsubductedoceaniccrustinmantlesources(Eileretal.,19975.4.ImplicationsforthevariableHeofJuanFernandezlavasTherangesofHeinJuanFernandezlavasare13.6to18.0R A. Pre-shieldFig.7.ProposedstagesofthevolcanologicevolutionofJuanFernandezIslands.A.Pre-shield,B.EarlyPhaseshield,C.LatePhaseshield,D.Rejuvenated(orpost-shield).Seetextfordiscussion.T.B.Truongetal.inGroupI,11.212.5RinGroupIIand7.89.5RinGroupIII.ToexplainthelowerHeofGroupIIbasanitesrelativetoGroupIFarleyetal.(1993)proposedbinarymixingbetweenaHeprimitive,undegassedandgeochemicallyenrichedJuanFernandezmantleplumeend-memberandthegeochemicallydepleteduppermantle,inafashionlikethatpreviouslyproposedforseveralHawaiianvolcanoes(Kurzetal.,1983).Notably,however,Farleyandco-workersdidnotproposeanasthenosphericoriginfortheGroupIIIbasaltsthatactuallyhaveMORB-likeHe.Moreover,theirpro-posalisclearlyatoddswithmorerecentinterpretationsofavailabledataforthedepletedcomponentandvariableHenatureoftheHawaiianmantleplume.Incontrasttothepreviousideathatthede-pletedcomponentcomesfromtheasthenosphere(e.g.,Kurzetal.,1983;Kelleretal.,2000Regelousetal.(2003;seealsoKurzetal.,2004;Freyetal.,2005)proposedthatthedepletedcomponentispartofthenominallyenrichedHawaiimantleplume,similartoproposalsfortheGalapagosmantleplume(e.g.,Hoernleetal.,2000)andSamoanmantleplume(Jacksonetal.,2014).TheMORB-like(8±1RHeofHawaiianlavascouldthereforebeanintrinsicpartoftheheterogeneousmantleplume(Regelousetal.,2003;Kurzetal.,2004;Freyetal.,2005;seealsoHahmetal.,2009).ItcouldalsorepresentaHeresidueduetopreviousdegassing(e.g.,Hiltonetal.,2000;DePaoloetal.,2001)orduetoearliercarbonatitemeltseparationofthehighHeplumecomponent(Hofmannetal.,2011).Sig-cantly,almostallgeochemicaltracersindicateadeepmantleplumesourceforHawaiianlavasalthoughtheirHevariesfrom6to34DePaoloetal.,2001;Kurzetal.,2004JuanFernandezexhibitsalargerangeinHesimilartoHawaiianlavasdespiteitslimitedrangesinSrandFig.8).Moreover,boththeirhighHelavasareassociatedwithPbandlowNdandZr/NbrelativetootherJuanFernandez(Figs.24and5)andMaunaKealavas(Kurzetal.,).Thus,basedonmajorandtraceelements,Sr-Nd-PbisotopesandgeologicevolutionoftheJuanFernandezlinearvolcanicchainpre-sentedabove,aswellasontheoverallcompositionalsimilaritybe-tweenJuanFernandezandHawaiianlavas,weproposethatthelow(orHeisalsoanintrinsicpartoftheJuanFernandezmantleplume.Inotherwords,thelowHeofJuanFernandezlavasdoesnotcomefromthegeochemicallydepleteduppermantle(cf.Farleyetal.,1993AnasthenosphericmantlesourceforthelowerHeofGroupIIbasaltsisalsohighlyinconsistentwiththeirextremeenrichmentinincompatibletraceelementsandthesimilarityoftheirOIB-likeSr-Nd-PbisotopicsignatureswithbothGroupsIandIIlavas(Figs.2to5,8Asdiscussedintheprevioussection,rejuvenatedGroupIIbasanitesweremostprobablyderivedlaterfromthesamemantleplumesourceasGroupIbasalts.Suchasourcemayhaveupwelledfrombelowandunderplatedthelithosphereduringthetransitionfromshieldtore-juvenatedstageofthevolcanicevolutionofJuanFernandezlavasFig.7).ThisargumentisclearlysupportedbyeldobservationsshowingthattheGroupIIbasanites,consistingmainlyofparasiticconesanddikesintrudingbasalticows,aresmallerinvolumeandca.1M.y.youngerthanGroupIbasalts(Bakeretal.,1987;seeals

o,Gerlachetal.,1986;Reyesetal.,2017).Bei
o,Gerlachetal.,1986;Reyesetal.,2017).Beingsmallerdegreepartialmeltsandhavingsueredpolybariccrystalfractionationorolivineaccumulation(Fig.6),GroupIIbasanitescontainhigherincompatibletraceelementsuchasThandUthanGroupIbasalts.Consequently,highUandThcouldhaveloweredtheHeratioinGroupIIba-saniteswhilemaintaininganearlyconstantSr-Nd-Pbisotopiccompo-sitionduringtheroughly1M.y.transitionorhiatusfromshieldtore-juvenatedstages.Thekeyissueiswhether1M.y.issucienttimetogrow-intherequisiteHetolowerHevalues.ModelcalculationsTable2)usingthefollowingequationfromGrahametal.(1987)=×+He2.8010{[U](4.35Th/U)}T(cmSTPg)483showthattheHe(11.212.5R)rangeofGroupIIbasanitescanindeedbeattainedwithin1M.y.fromGroupIbasaltPIN-8(withanHeof17.2R)bysimplytrapping3%ofHe*pro-ducedbythehighUandThcontentsoftheGroupIIbasanites.Moreimportantly,degassingofJuanFernandezvolcanoesmostprobablyhelpedlowertheconcentrationoftheoriginalHeand,thus,decreasedthetimerequiredtolowerHe(e.g.,Hiltonetal.,2000).Inshort,theaboveanalysessuggestthatitisplausibletoproducethelowerHevaluesofGroupIIbyradiogenicin-growth±degassing.Ifcorrect,thenthereisnorequirementfortheinvolvementof0.70340.70400.51280.5129Fig.8.He(R)versusSrandNddia-gramsforJuanFernandezlavas.FieldsforvolcanoesintheislandofHawaiionly(afterKurzetal.,2004)areshownforreference.SymbolsasinFig.4.NotethatlikeJuanFernandez,individualHawaiiansamplesuiteshavealargerangeofHe(R),butlimitedrangesofSrandGroupIIbasanitesareasubsetofGroupIthatexperiencedproductionofHe*fromitsUandThcontents.Seetextforadditionaldiscussion.Table2PercentHefromGroupIbasaltneededtoproduceGroupIIHeafter1M.y.hiatus.hiatus.PIN-8[He*]3He/4He%He*cc/gcc/g(cc/g)(R/R)Needed132.69E083.75E0712.5132.69E083.75E0711.2Notes:PIN-8hasthehighestHe(17.2R/R)amongGroupIsamples.He*istheamountofHeproducedafter1M.y.usingtheaverageU+ThofGroupIIbasanites.T.B.Truongetal.asthenosphericmaterialingeneratingtheGroupIIbasanites(cf.,etal.,1993AsinthepreviousSr-Nd-Pbisotopicargument,theparentalmagmasforGroupsIandIIhighHelavasoriginateatornearFOZOasitisthecommonsourceofhighHe�(9R)OIB(e.g.,HananandGraham,1996;Hiltonetal.,1999;Castillo,2015,2016).AlthoughGroupIIIparentalmagmasalsooriginatefromthesamesource,thelatterhasMORB-likeHeastheJuanFernandezmantleplumeiscompositionallyheterogeneousandmostlikelycontainsaless-enrichedordepletedcomponentwithlowerHe(9R),similarto,e.g.,MaunaKea.MajorandtraceelementdatashowthattheJuanFer-nandezparentalmagmaswerederivedthoughsequentialandvariabledegreesofpartialmeltingofsuchacommonmantleplumesource.Initial,relativelylargedegreeofpartialmeltingsampledFOZOplusthedepletedcomponentwithaMORB-likeHeof8±1R,produ-cingGroupIIIbasaltswhereaslater,smallerdegreeofpartialmeltingsampledmainlytheFOZOsourcewithhighHe&#x-303;&#x.7 0;9R,produ-cingGroupIbasaltsandGroupIIbasanites.Insum,noneofthetwosourcecomponentsoftheJuanFernandezmantleplumerequireex-tremeend-componentsandinsteadfallwithinorclosetoFOZO.Theaboveproposalisconsistentwiththegeochemistryofseveralcriticalsamples.ThePV-2basanitoidisamixtureofbasaltandbasanitemagmasproducedduringthetransitionfromtheclimaxoftheshield-buildingstagetotherejuvenatedstageand,thus,ischemicallyandisotopicallytransitionalbetweenGroupsIandII.Unsurprisingly,PV-2alsohasHe(13.6R)transitionalbetweenGroupsIandIIFig.2).Ontheotherhand,sampleMF-C2,theonlyGroupIIIsamplefromA.SelkirkanalyzedherethathasHehigherthantheMORBrange,at9.5R,wascollectedfromtheisland'ssummit(Farley,1991;Farleyetal.,1993).Hence,itisayoung,ifnottheyoungest,eruptiveatA.Selkirkandrepresentsthetransitionfromthemainshield-buildingA.Selkirk(orGroupIII)tolateshield-buildingR.Crusoe(orGroupII)stage,whencethemoredepletedcomponentwasgenerallyexpendedand/oroverwhelmedbytheinuxofhigherHeFOZOplumeFinally,thesimilarityoftheNdratiosofthreeoutoftheveolivinesandhostrocks(Table1)suggeststhatthevaluespreservedbytheolivinesareinisotopicequilibriumwiththoseintheirrespectivehostmagmas.Thatis,theolivinephenocrystsmayhaveequilibratedwiththelastdropofmeltsinJuanFernandez,despitetheircomplexcrystallizationhistoriesasclearlyevidencedinthepet-rographyandchemistryoftheseolivines(Natland,2003).Neodymium,similartoHe,occursintraceamountsinolivineasitsdis

tributioncientinolivineisverylow(K=0.006
tributioncientinolivineisverylow(K=0.006;Rollinson,1993).ThefactthatwecouldmeasureNdfromtheolivineseparatesindicatesthatNdisalsotrappedlikeHeincrystallizingolivinephe-nocrysts,asNdispresentintinymeltinclusionsandsilicateorspinelcrystalstrappedinsidecavities,bubblesandthermalstressfracturesintheolivine(cf.,Natland,2003).Ontheotherhand,theslightlylowerNdratios(whichgenerallytrendtowardhigherFig.8;seealso,Farleyetal.,1993)oftwooutoftheveolivinesre-lativetohostrocksmayalsoindicatethattheinitialprimarymagmahashigherHethantheresidualdierentiatedand,mostlikely,moredegassed(e.g.,Hiltonetal.,2000;DePaoloetal.,2001)meltthatcrystallizedtoformthebulkoftherock.Eitherway,thesepreliminaryresultssuggestthatolivinereectstheinherentheterogeneitiesofbothHeandNdinthesourceofJuanFernandezlavas.Thus,thepetrographiccharacteristicsofHe-bearingolivinesgenerallydonotsupporttheideathattheHecontent(and,hence,e.g.highHe)ofolivinescomesfromimplicitlyseparate,late-arrivingmantlevolatilesthatpermeatedintoshallowmagmareservoirswheremagmaticdif-ferentiationwerefairlyextensive(cf.,Natland,2003).Instead,theHeandbulkcompositionofJuanFernandezlavasreectthoseoftheirsourceandthedierentiationprocessesexperiencedbythepri-marymeltsderivedfromthissource.6.ConclusionsThisstudypresentsnewtraceelementandSr,NdandPbisotopicanalysesofJuanFernandezlavastofurtherconstraintheirpetrogenesisandmantlesourcecomposition.Variationsinmajorandtraceelementcompositioninthelavasaremainlycontrolledbydierencesindegreesofpartialmeltingofacommon,slightlycompositionallyheterogeneousmantlesource.Overall,majorandtraceelementdataandmodelingsuggestthatA.SelkirkGroupIIIbasaltswereproducedbythelargestdegreeofpartialmeltingatmoderatelylowpressure,followedbyR.CrusoeGroupIbasalts.R.CrusoeGroupIIbasaniteswereproducedbythesmallestdegreeofpartialmelting.Fractionalcrystallizationac-countsforsomeofthecompositionalvariationsoftherockgroups,particularlythoseintheR.CrusoeGroupIIbasanites.ThenewdataconrmthelimitedrangeofSr,NdandPbisotopicratiosofJuanFernandezlavasdespitetheirrelativelylargerangeofHevalues.Aswithmajorandtraceelementanalyses,thelimitedisotopicrangeofradiogenicisotopesindicatesthattheparentalmagmasforJuanFernandezlavasarederivedfromacommon,thoughslightlyheterogeneousmantleplumesource.ThissourceresidesatorneartheFOZOregioninsideSr-Nd-Pbisotopicspace.Asmallbutdis-tinctcompositionalvariabilityexistswithinthissource.AFOZOmantlesource,withayoungHIMUPbisotopiccomposition,ispreferentiallysampledbyR.CrusoemeltswhereasA.SelkirkmeltssampleFOZOwithanadditionalgeochemicallydepletedavor.Signicantly,theobservedgeochemicalvarianceofJuanFernandezlavasisconsistentwiththetemporalevolutionofJuanFernandezvolcanoesastheyprogressthroughshield-buildingandrejuvenatedstages.Resultsofthecombinedmajor-traceelementandSr-Nd-Pbin-vestigationdonotindicatesubstantialcontributionsfromastheno-sphericmantle,andthereislittleevidenceofcrustalcontamination.Instead,thecombinedresultsshowthattheearlyphaseofshieldbuildingstage(GroupIIIlavas)sampledbothFOZOandarelativelydepletedcomponentwithHe=8±1Rwhereasthelaterphasesampled(GroupIlavas)FOZOcomponentwithHe&#x-303;&#x.7 0;9.GroupIIbasanitesarelatestageGroupIlavasthatsueredin-growthofHeand,thus,haveHelowerthanGroupI,butstillhigherthanGroupIII.JuanFernandezisunlikeotherhighHeOIBlinearvolcanicchainsbecauseofthedominanceoftheFOZOmantleplumecomponentinitsmantleplumesource.WethankJ.KonterandananonymousreviewerfortheirhelpfulreviewsandcommentsandC.MacIsaacandL.Finninfortheirassis-tanceinthelaboratory.AppendixA.SupplementarydataSupplementarydatatothisarticlecanbefoundonlineatAllègre,C.J.,Minster,J.F.,1978.Quantitativemodelsoftraceelementbehaviorinmagmaticprocesses.EarthPlanet.Sci.Lett.38,1Baker,P.E.,Gledhill,A.,Harvey,P.K.,Hawkesworth,C.J.,1987.GeochemicalevolutionoftheJuanFernandezIslands,SEPacic.J.Geol.Soc.Lond.144,933Booker,J.,Bullard,E.C.,Grasty,R.L.,1967.PalaeomagnetismandageofrocksfromEasterIslandandJuanFernandez.Geophys.J.Roy.Astr.Soc.12,469Castillo,P.R.,2015.TherecyclingofmarinecarbonatesandsourcesofHIMUandFOZOoceanislandbasalts.Lithos216,254Castillo,P.R.,2016.AproposednewapproachanduniedsolutiontooldPbparadoxes.Lithos252,32Castillo,P.R.,201

7.AnalternativeexplanationfortheHf-Ndman
7.AnalternativeexplanationfortheHf-Ndmantlearray.Sci.Bull.62,T.B.Truongetal.Castillo,P.R.,Scarsi,P.,Craig,H.,2007.He,Sr,Nd,andPbisotopicconstraintsontheoriginoftheMarquesasandotherlinearvolcanicchains.Chem.Geol.240(3),Corvalan,J.,1981.PlateTectonicMapoftheCircum-PacicRegion,SoutheastQuadrantCraig,H.,Lupton,J.E.,1976.Primordialneon,helium,andhydrogeninoceanicbasalts.EarthPlanet.Sci.Lett.31,369Craig,H.,Lupton,J.E.,1981.Helium-3andmantlevolatilesintheoceanandtheoceaniccrust.In:Emiliani,C.(Ed.),TheSea.7.Wiley,NewYork,pp.391Day,J.M.D.,Pearson,D.G.,Macpherson,C.G.,Lowry,D.,Carracedo,J.C.,2010.EvidencefordistinctproportionsofsubductedoceaniccrustandlithosphereinHIMU-typemantlebeneathElHierroandLaPalma,CanaryIslands.Geochim.Cosmochim.Acta74,6565DePaolo,D.J.,Bryce,J.G.,Dodson,A.,Schuster,D.L.,Kennedy,B.M.,2001.IsotopicevolutionofMaunaLoaandthechemicalstructureoftheHawaiianplume.Geochem.Geophys.Geosyst.2(2000GC000139)Devey,C.W.,Hémond,C.,Stoers,P.,2000.Metasomaticreactionsbetweencarbonatedplumemeltsandmantleharzburgite:theevidencefromFridayandDomingoSeamounts(JuanFernandezchain,SEPacic).Contrib.Mineral.Petrol.139,68Eiler,J.M.,Farley,K.A.,Valley,J.W.,Hauri,E.,Craig,H.,Hart,S.R.,Stolper,E.M.,1997.Oxygenisotopevariationsinoceanislandbasaltphenocrysts.Geochim.Cosmochim.Acta61,2281Farley,K.A.,1991.RareGasesandRadiogenicIsotopesinSouthPacicOceanIslandBasalts(Ph.D.Thesis).UniversityofCaliforniaFarley,K.A.,Natland,J.H.,Craig,H.,1992.Binarymixingofenrichedandundegassed(primitive?)mantlecomponents(He,Sr,Nd,Pb)inSamoanlavas.EarthPlanet.Sci.Lett.111,183Farley,K.A.,Basu,A.R.,Craig,H.,1993.He,SrandNdisotopicvariationsinlavasfromtheJuan-FernandezArchipelago,SEPacic.Contrib.Mineral.Petrol.115,75Feigenson,M.D.,Hofmann,A.W.,Spera,F.J.,1983.Casestudiesontheoriginofbasalt.Contrib.Mineral.Petrol.84(4),390Foulger,G.R.,2011.Platesvs.Plumes:AGeologicalControversy.JohnWiley&Sons,Frey,F.A.,Huang,S.,Blichert-Toft,J.,Regelous,M.,Boyet,M.,2005.OriginofdepletedcomponentsinbasaltrelatedtotheHawaiianhotspot:evidencefromisotopicandincompatibleelementratios.Geochem.Geophys.Geosyst.6,Q02L07.Gast,P.W.,Tilton,G.R.,Hedge,C.,1964.IsotopiccompositionofleadandstrontiumfromAscensionandGoughIslands.Science145(3637),1181Gerlach,D.C.,Hart,S.R.,Morales,V.W.J.,Palacios,C.,1986.Mantleheterogeneitybe-neaththeNazcaplate:SanFelixandJuanFernandezislands.Nature322,165Graham,D.W.,2002.Noblegasisotopegeochemistryofmid-oceanridgeandoceanislandcharacterizationofmantlesourcereservoirs.Rev.Mineral.Geochem.47,Graham,D.W.,Jenkins,W.J.,Kurz,M.D.,Batiza,R.,1987.Heliumisotopedisequilibriumandgeochronologyofglassysubmarinebasalts.Nature326,384Gripp,A.E.,Gordon,R.G.,2002.Youngtracksofhotspotsandcurrentplatevelocities.Geophys.J.Int.150(2),321Hahm,D.,Castillo,P.R.,Hilton,D.R.,2009.AdeepmantlesourceforhighHeoceanislandbasalts(OIB)inferredfromPacicnear-ridgeseamountlavas.Geophys.Res.Lett.36(2),L20316.Hanan,B.B.,Graham,D.W.,1996.Leadandheliumisotopeevidencefromoceanicbasaltsforacommondeepsourceofmantleplumes.Science272(5264),991Hart,S.R.,1984.Alarge-scaleisotopeanomalyintheSouthernHemispheremantle.Nature309,753Hart,S.R.,Hauri,E.H.,Oschmann,L.A.,Whitehead,J.A.,1992.Mantleplumesanden-trainment:isotopicevidence.Science256(5056),517Hauri,E.H.,Whitehead,J.A.,Hart,S.R.,1994.Fluiddynamicandgeochemicalaspectsofentrainmentinmantleplumes.J.Geophys.Res.99,24,275Hilton,D.R.,Porcelli,D.,2014.Noblegasesastracersofmantleprocesses.In:Carlson,R.L.(Ed.),TheMantleandCoreVol.3,TreatiseonGeochemistry,2ndedition(Eds.Turekian,K.K.,Holland,H.D.).Elsevier-Pergamon,Oxford,UK,pp.327Hilton,D.R.,Gronvold,K.,Macpherson,C.G.,Castillo,P.R.,1999.ExtremeHera-tiosinnorthwestIceland:constrainingthecommoncomponentinmantleplumes.EarthPlanet.Sci.Lett.173,53Hilton,D.R.,Macpherson,C.G.,Elliott,T.R.,2000.Heliumisotoperatiosinmacphe-nocrystsandgeothermaluidsfromLaPalma,theCanaryIslands(Spain):implica-tionsforHIMUmantlesources.Geochim.Cosmochim.Acta64,2119Hoernle,K.,Reinhard,W.,Morgan,J.P.,Garbe-Schönberg,D.,Bryce,J.,Mrazek,J.,2000.ExistenceofcomplexspatialzonationintheGalápagosplume.Geology28(5),Hofmann,A.W.,1997.Mantlegeochemistry:themessagefromoceanicvolcanism.Nature385(6613),219Hofmann,A.W.,201

4.SamplingMantleHeterogeneitythroughOcea
4.SamplingMantleHeterogeneitythroughOceanicBasalts:IsotopesandTraceElements,2nded.TreatiseGeochem3.pp.67Hofmann,A.W.,Farnetani,C.G.,Spiegelman,M.,Class,C.,2011.DisplacedheliumandcarbonintheHawaiianplume.EarthPlanet.Sci.Lett.312,226Jackson,M.G.,Hart,S.R.,Saal,A.E.,Shimizu,N.,Kurz,M.D.,Blusztajn,J.S.,Skovgaard,A.C.,2008.Globallyelevatedtitanium,tantalum,andniobium(TITAN)inoceanislandbasaltswithhighHe.Geochem.Geophys.Geosyst.9,Q04027.Jackson,M.G.,Hart,S.R.,Konter,J.G.,Kurz,M.D.,Blusztajn,J.,Farley,K.,2014.HeliumandleadisotopesrevealthegeochemicalgeometryoftheSamoanplume.Nature514,Janney,P.E.,Castillo,P.R.,1996.BasaltsfromtheCentralPacicBasin:evidencefortheoriginofCretaceousigneouscomplexesintheJurassicwesternPacic.J.Geophys.Res.101(B2),2875Kay,R.,Hubbard,N.J.,Gast,P.W.,1970.Chemicalcharacteristicsandoriginofoceanicridgevolcanicrocks.J.Geophys.Res.75(8),1585Keller,R.A.,Fisk,M.R.,White,W.M.,2000.IsotopicevidenceforLateCretaceousplu-ridgeinteractionattheHawaiianhotspot.Nature405(6787),673Klein,E.M.,Langmuir,C.H.,1987.Globalcorrelationsofoceanridgebasaltchemistrywithaxialdepthandcrustalthickness.J.Geophys.Res.92(B8),8089Konter,J.G.,Staudigel,H.,Blichert-Toft,J.,Hanan,B.B.,Polvé,M.,Davies,G.,Shimizu,N.,Schiman,P.,2009.GeochemicalstagesatJasperSeamountandtheoriginofintraplatevolcanoes.Geochem.Geophys.Geosyst.10(2),Q02001.Kurz,M.D.,Jenkins,W.J.,Hart,S.R.,Clague,D.,1983.HeliumisotopicvariationsinvolcanicrocksfromLoihiSeamountandtheIslandofHawaii.EarthPlanet.Sci.Lett.66,388Kurz,M.D.,Curtice,J.,Lott,D.E.,Solow,A.,2004.RapidheliumisotopicvariabilityinMaunaKeashieldlavasfromtheHawaiianScienticDrillingProject.Geochem.Geophys.Geosyst.5(4),Q04G14.Morgan,W.J.,1972.Deepmantleconvectionplumesandplatemotions.AAPGBull.56(2),203Natland,J.H.,2003.CaptureofheliumandothervolatilesduringthegrowthofolivinephenocrystsinpicriticbasaltsfromtheJuanFernandezIslands.J.Petrol.44(3),Niu,Y.,O'Hara,M.J.,2008.Globalcorrelationsofoceanridgebasaltchemistrywithaxialdepth:anewperspective.J.Petrol.49(4),633Peters,B.J.,Day,J.M.D.,2014.AssessmentofrelativeTi,Ta,andNb(TITAN)enrich-mentsinoceanislandbasalts.Geochem.Geophys.Geosyst.15(11),4424Regelous,M.,Hofmann,A.W.,Abouchamis,W.,Galer,J.G.,2003.GeochemistryoflavasfromtheEmperorSeamounts,andthegeochemicalevolutionofHawaiianmagma-tismfrom85to42Ma.J.Petrol.44(1),113Reyes,J.,Lara,L.E.,Morata,D.,2017.ContrastingP-TpathsofshieldandrejuvenatedvolcanismatRobinsonCrusoeIsland,JuanFernándezRidge,SEPacic.J.Volc.Res.341,242Rodrigo,C.,Lara,L.E.,2014.PlatetectonicsandtheoriginoftheJuanFernándezRidge:analysisofbathymetryandmagneticpatterns.Lat.Am.J.Aquat.Res.42(4),Rollinson,H.R.,1993.UsingGeochemicalData:Evaluation,Presentation,Interpretation.LongmanGroupUKLtd.,LondonStracke,A.,2012.Earth'sheterogeneousmantle:aproductofconvection-driveninter-actionbetweencrustandmantle.Chem.Geol.330,274Stracke,A.,Hofmann,A.W.,Hart,S.R.,2005.FOZO,HIMU,andtherestofthemantlezoo.Geochem.Geophys.Geosyst.6,Q05007.Stuart,F.M.,Lass-Evans,S.,Fitton,J.G.,Ellam,R.M.,2003.HighHeratiosinpi-criticbasaltsfromBanIslandandtheroleofamixedreservoirinmantleplumes.Nature424,57Stuessy,T.F.,Foland,K.A.,Sutter,J.F.,Sanders,R.W.,Silva,M.,1984.Botanicalandgeologicalsignicanceofpotassium-argondatesfromtheJuanFernandezIslands.Science225(80),49Sun,S.-S.,McDonough,W.F.,1989.Chemicalandisotopicsystematicsofoceanicbasalts:implicationsformantlecompositionandprocesses.In:Saunders,A.D.,Norry,M.J.(Eds.),MagmatismintheOceanBasins.GeologicalSocietySpecialPublication,London,vol.42.pp.313Thirlwall,M.F.,1997.PbisotopicandelementalevidenceforOIBderivationfromyoungHIMUmantle.Chem.Geol.139(1),51Truong,T.B.,2015.TraceElementAbundancesandSr-Nd-PbIsotopicConstraintsonthePetrogenesisofJuanFernandezLavas(MSThesis).UniversityofCalifornia,SanDiego(59pp.)Vidal,P.,1992.Mantle:moreHIMUinthefuture?Geochim.Cosmochim.Acta56,White,W.M.,2015.IsotopeGeochemistry.JohnWiley&Sons,ChichesterWhite,W.M.,Duncan,R.A.,1996.GeochemistryandgeochronologyoftheSocietyIslands:newevidencefordeepmantlerecycling.Geophys.Monogr.Ser.95,183Wilson,J.T.,1963.ApossibleoriginoftheHawaiianislands.Can.J.Phys.41,863Zindler,A.,Hart,S.,1986.Chemicalgeodynamics.Annu.Rev.EarthPlanet.Sci.14,T.B.T