JOURNALOF GEOPHYSICALRESEARCH,VOL. 95, NO. B8, PAGES12,697-12,711, AUGUST 10, 1990 The Cobb-Eickelberg SeamountChain: Hotspot Volcanism withMid-Ocean RidgeBasaltAffinity DANA L. DESONtoAND ROBERTA. DUNCAN Collegeof Oceanography, OregonStateUniversity, Corvallis Cobbhotspot, currently located beneath Axialseamount ontheJuandeFucaridge,hasthetemporal butnottheisotopic characteristics usually attributed toa mantle plume.Theearlier volcanic products of thehotspot, fromeightvolcanoes in theCobb-Eickelberg seamount (CES)chain, showa westward age progression awayfromthehotspot anda westward increase in theagedifference between the seamounts andthecrust onwhich theyformed.These results areconsistent withmovement ofthePacific plateover a fixedCobbhotspot andeventual encroachment bythewestwardly migrating JuandeFucaridge.CES lavasare slightlyenrichedin alkaliesandincompatible elements relativeto thoseof the Juande Fuca ridgebuttheyhaveSt,Nd,andPbisotopic compositions virtually identical to those foundalongthe ridge.Therefore, Cobbhotspot is a stationary, upper mantle melting anomaly whose volcanic products showstrong mid-ocean ridgebasalt(MORB)affinity.Theseobservations canbe explained by low degreesof partialmeltingof entrained heterogeneous uppermantleMORB sourcematerialwithina thermally drivenlowermantle diapiror by anintrinsic MORB-like composition of thedeeper mantle sourceregionfrom whichnortheast Pacificplumesrise. INTRODUCTION the Endeavoursegmenthas been attributedto the contribution to theoverriding Juande Fuca Sincehotspotswere first described by Wilson [1963] and of theHecklemeltinganomaly linked to convectivemantleplumesby Morgan [1972], both spreadingridge melts [Karsten, 1988]. Axial seamount, which straddles the JDFR axis at -46øN temporaland chemicalcharacteristics of thesefeatureshave beenrecognized.The linear array of volcanoesand the age- (segment2) [Delaneyet al., 1981] is the youngestexpression progressive distribution of hotspottracks[Morgan,1972]are of volcanismover an uppermantle melting anomalytermed two of the more distinguishing featuresof volcanismresulting the Cobbhotspot(Figure 1). Older volcanicproductsof the from a fixed mantle plume. Some hotspots have been hotspot,the Cobb-Eickelbergseamount(CES) chain,includea persistentover long periods,a featureclearly seenin the line of edifices on the Pacific plate, from Brown Bear a group volumesof lava producedby the Hawaiianhotspot,while other northwestthroughCorn, Cobb, and Pipe seamounts, Pacific plate hotspots have produced shorter or more of small volcanoes known as the Seven Deadly Sins intermittent seamount or island chains. Thus far, volcanic seamounts, and finally the larger Warwick, Eickelberg,and rocks formed at hotspots have been characterized by Forster seamounts(Figure 1). Two seamountsto the east of distinctivecompositions, usuallymore radiogenicSr, Nd, and Axial seamounton the Juande Fucaplate (ThompsonandSon Pb and higher concentrationsof incompatible elements, of Brown Bear) are inferred to be related to this linear volcanic relative to basaltseruptedat spreadingridges. The isotopic province. Beyond Forster seamount,the westernmostedifice compositionof lavas eruptingover a given hotspotmay not in the Eickelberggroup,is a gap of 800 km before a line of be constantwith time, nor are all hotspotsalike isotopically, seamounts which includes Miller, Murray, and Patton but as a group these rocks have been termed ocean island seamounts.Theseolder volcanoeshave beeninterpretedto be basalts(OIB) as distinguished from basaltsthat form at mid- earlierproductsof volcanicactivity at Cobb hotspot[Duncan oceanridges(MORB). The Cobb-Eickelberg seamount chain, and Clague, 1985; Smoot, 1985], but this suggestion has been and other hotspot-associatedvolcanic lineaments in the questioned by Dalrympleet al. [1987] becauseseamountages northeastern Pacific, are unusual in that they exhibit the do not completelyagreewith rigid Pacificplatemotionover a temporalcharacteristics producedby othermantleplumes,but fixed hotspot. not the OIB isotopicsignatures. The Pacificmantlerotationsproposed by DuncanandClague Basedon differentmorphologicandtectoniccharacteristics,[1985] and Pollitz [1988] place Miller seamount(26 Ma the Juande Fucaridge (]DFR) wasdividedinto four spreading [Dalrymple et al., 1987]) directly over Cobb hotspotat the segmentsby Delaney et al. [1981]. Along-segmentand time of its formation. Basaltic rocks fxom this seamount have between-segment variationsin basaltchemistrygive evidence relativelylow alkali contentsand flat rare-earthelement(REE) of separate magmaticsystems beneaththeridge[Liias, 1986]. patterns[Dalrympleet al., 1987] similarto volcanicproducts The Endeavoursegment,northernmost of the four, whichlies of the CES lineament. Backtrackingof Murray and Patton north of the Cobb offset (Figure 1) [Delaneyet al., 1981] is seamounts,however, places their positions at origin some chemicallythe mostprimitive(highestMg-numbers)yet has 150-250km westof Cobbhotspot.The higheralkalicontents the mostincompatible elementenrichedlavasfoundalongthe and light-REE enriched nature of the Murray and Patton ridge[Liias, 1986;Karsten,1988]. The unusualchemistry of Copyright 1990 by the American Geophysical Union. Paper number 89JB03074. 0148-0227/90/89JB-03074505.()0 seamount basalts would be consistent with their formation as a productof late-stagerejuvescentvolcanismdownstreamfrom the Cobb hotspot, or with formation at another,now extinct, hotspot. Whether or not the three older seamountsformed at Cobb hotspot, there has been a lack of consistencyof volcanismover the melting anomaly;it either becameactive or it teacrivedafter a long periodof quiescence more thannine 12,697 12,698 DESONIEAND DUNCAN:• 150 ø 145 ø COBB-EICg•.T.nERO SEAMOUN•CHAIN 140 ø 135 ø 125 ß 130 ø NORTH 60 ß AMERICAN PLATE Giacomini .•mount Pratt Guyot Weaker Guyot 55 ß Patton DIckins Seamount 8eamount Denson Guyot 4• o o Guyot /Miller Seamount Davidson Parker Seamount Gu Bowie Seamount Tuzo Wilson Solmounts ß athflnder PACIFIC contour •/'d•8..mount PLATE interval T.Horton Guyot•• 50 ß = 400 ß Hock 8mt•. '• Unione • •G DE FUCA PLATE Pros. 45 ø Jackson Solmounts 40 ø ntour Fig. 1. interval Bathymetric map showingseamounts, spreading ridges(doublelines)andfracturezones(singlelines)of the northeastPacific, after Chaseet al. [1970]. Inset showsseamountsof the Cobb-Eickclberglineamentand Explorerand Union volcanoes. million yearsago with the formationof Forsterand Eickelberg seamounts. Since the hotspot began this latest pulse of activity, volcanicoutputhas been continuousbut hasvaried up to 20% [Karsten and Delaney, 1989] which is within the known variability of other hotspots [Bargar and .lackson, 1974; Lonsdale, 1988]. Within the northeast Pacific basin are several other linear originatedby differentmechanisms.The longestand best known of theseis the Pratt-Welkerseamountchain (Figure 1). Dalrymple et al. [1987] suggesteda complex history involving at least two hotspots;some of the seamounts formed in a midplate setting while volcanismat or near a spreading ridgeformedothers[Turneret al., 1980]. ThePrattWelker hotspot(s)may be located near the Tuzo Wilson volcanic chains, all of which are generally subparallelwith seamounts [Chase, 1977; Cousens et al., 1985], Bowie more prominent Pacific chains to the southwest. These seamount[Turner et al., 1980], or the Dellwoodknolls[Silver lineaments differ in length and volume and may have et al., 1974]. All three of theseyoung volcanicsites(Figure DESONIE ANDDUNC•: Tm• COBB-EICKELBERG SEAMOUNI' CHAIN 12,699 1) are proposedto reflecthotspotactivity,althoughthe Tuzo lavas. We discussthesedata in termsof possibledynamic Wilsonseamounts and Dellwoodknollsmay haveoriginated models of mantle plume and asthenosphere mixing, and by partialmeltingin a pull-apartbasin[Allanet al., 1988]. A concludethat the deep mantle under the northeastPacific moresoutherlyline of seamounts in the northwestern Gulf of region is intrinsicallymore MORB-like than the centraland Alaska, the Horton-Pathfinder-Parker group, cannot be southPacific [Hart, 1984, 1988]. attributedto a currently active hotspot. Severalshorterseamount chainslie closeto andintersect SAMPLE DESCRIFrION$ the JDFRfromthe west. Theseappearto reflecta much Basaltsamples from the Cobb-Eickelberg andExplorershorter-lived andshallower-origin meltingphenomenon. The Unionseamounts were obtainedfrom J. Delaneyand P. HeckandHeckleseamount chains, for example, maybe the Johnson of theUniversity of Washington andwereoriginally result of passiveupwellingand partial melting of dredged duringR/VThomasThompson cmises Tr063,Tr080, heterogeneous upper mantle in advance of thenorthwestwardly andTI'175. TheCobbseamount sample wasobtained by a migrating JuandeFucaspreading ridge[DavisandKarsten,diverfromthepinnacle of thevolcano.Dredge locations are 1986]similarto thesetting of theLamont seamounts nearthe givenin Table1. EastPacificRise[Fornariet al., 1988a]. Otherseamounts and Samplesused in this study for age determinations and small seamountchains,includingExplorerand Union compositional analyses werethefreshest available, withlittle seamounts whichlie to thenorthof theCES(Figure1), areof to no interstitial glassandonlyslightto moderate alteration. unknown origin. All aforementioned seamount chainsare CES basaltsamplesare generallymicrocrystalline and foundon the Pacificplate;only two edifices,Thompson and aphyric. In some samples,phenocrystsof plagioclase, Son of Brown Bear, which are interpretedto be relatedto clinopyroxene, andFe-Ti oxidesare foundin an intergranular volcanismat Cobb hotspot[Desonie and Duncan, 1986; or intersertalmatrix. Plagioclasephenocrysts are generally KarstenandDelaney,1989]wereformedon theJuande Fuca freshbut may be slightlyresorbedor zoned. Both samples plate. from Eickelberg seamountused in this study contain In this paper we report new nøAr-39Al'and K-Ar age plagioclase glomerocrysts. Severalbasaltsamplesarehighly determinationsfrom eight volcanoesof the CES which vesicular,indicatingthat the magmasdegassed as theycooled documenta clear age progressionalong the lineament,in andwereprobablyeruptedat a relativelyshallowdepthor had concertwith platemotionover a stationaryhotspot. Major high ascent rates [Dixon et al., 1988]. Secondary andtraceelementcontentsare slightlyenrichedover thosefor mineralization is slight; the matrix of some samples is heterogeneous MORB eruptedalongthe Juande Fucaspreadingpartially altered to reddish-brown clays. One Warwick ridge. In St, Nd, and Pb isotopiccompositions, the CES seamountsample(TI'080 DH04-4) has zeoliteslining some basalts are indistinguishable from nearby spreading ridge vesicles. TABLE 1. DredgeSampleLocationsandWholeRockDescriptions for Seamounts of theCobbEickelbergSeamount ChainandExplorerandUnionSeamounts Utilizedin This Study. Sample Seamount Name LatitudeøN LongitudeøWNatureof sample TI'080 DH 01-1 Eickelberg(E) 48.29 133.09 v, d TI'080DH 02A TI'080 DH 02-8 TI'080 DH 04-4 Eickelberg (E) Eickelberg (E) Warwick(W) 48.32 48.53 48.02 133.10 133.10 132.46 m, pp v, pp m, pm TT080 DH 05-14 Warwick(W) 48.01 132.53 v, d TI'080 DH 08-8 TI'080 DH 09-1 Waxwick(W) Waxwick(W) 48.03 48.02 132.45 132.47 m, pp pm TI'080 TI'080 TI'175 TI'175 unnamed(X) unnamed(X) Sloth (S) Lust(L) 48.28 48.28 47.62 47.50 133.22 133.22 131.70 131.51 fo v, fo v, mx v, mx TI'175 DH 74-17 Gluttony(G) 47.14 131.46 v, pp CB-1 TT080 DH 14A TI'080 DH 14-5 Cobb (C) Thompson(T) Thompson(T) 46.77 46.03 46.03 130.83 128.63 128.63 v, mx v, mx m, mx TI'063 DH 34 TI'063 DH 35 TI'063 DH36A Explorer(Ex) Explorer(Ex) Union(U) 48.96 48.98 49.55 131.03 130.98 132.73 op, pp v, op, pp v, pp DH DH DH DH 10A 10-7 70-2 71-4 A symbolfor eachseamount is includedin parentheses behindseamount nameandwill be used throughoutthis study. Sampledescriptions: v = vesicular, m = massive, mx=microcrystalline, d = devitrified glass, pp = plagioclase phenocrysts, pm = plagioclase microlites,op = olivinephenocrysts, fo = Fe-Ti oxides. 12,700 DESONIE ANDDUNCAN:Tim COBB-EICKELBERG SEAMOUNT CHAIN exhibit greater variability than the JDFR basalts (stippled field), althoughthe older Pacific plate CES have ratheruniform AnalyticalProcedures REE concentrations. Cobb through Eickelberg seamount Major and some trace elementanalysesfor dredgedbasalts basalts show slight enrichment in light-REEs relative to the were determined from pressed powder pellets by X-ray JDFR field (including the Endcavour segment which lies fluoresencespectroscopy[Hooper, 1981]. Rare-earthelement virtually within the field for segments1-3) and a steeperlight(REE) and Sc concentrations were determined from to heavy-REE pattern (and thus greater (La/Sm)N) than the instrumentalneutron activation analysis(INAA) [Laul, 1979]. spreadingridge samples. CrossingREE patternswithin the Isotopedilution techniqueswere employedto measureSt, Rb, Pacific plate CES (Figure 5, inset) may indicatethat the basalts Cs, K, Sm, and Nd concentrations[White et al., in press]. originated from variable partial melting of a mildly Where any of thoseelementsis availableby anotheranalytical heterogeneous source or dynamic partial melting of a homogeneous source[Langmuiret al., 1977]. technique,the isotopedilution valuesare preferred. MgO content and Mg-number Sr, Nd, and Pb isotopicratioswere measuredon a VG Sector The moderate thermal ionization mass spectrometerat Cornell University, [=100Mg/(Mg+Fe2)] of many CES and Explorer and Union following the procedureof Whiteet al. [in press]. All samples samplesindicate that the basaltshave undergonelow pressure were leached in 6 N HC1 to eliminate effects of seawater fractionationaway from primary melt compositions. Olivine alteration. In order to comparethe CES resultswith thosefrom phenocrystsare present only in the sample from Explorer However, low Ni contents and a constant previously publishedstudiesof other northeastPacific basalts, seamount. only samples that were similarly leached before isotopic CaO/A1203 with decreasingMg-number indicate significant olivine removalfrom the mantle-derivedmelts (Figure 6). The analysiswere considered. absenceof a Eu anomalyin the REE patternsof CES samples Major and Trace Elements (Figure 5) and roughly constantSr/Zr ratios (Table 2) provide The older Pacific plate CES lavas (Cobb throughEickelberg evidencethat althoughplagioclasewas on the liquidus,a large seamounts)show a narrow range of variation with respectto amount of it was not removed from the melt during many chemical parameters, often smaller than previously crystallization. From the lack of correlation between Sc and analyzed samples from the JDFR (Table 2) [Liias, 1986; CaO/A1203 with Mg-number (Table 2 and Figure 6) it appears Karsten, 1988]. However, a larger variation in chemical that clinopyroxenewas not an important fractionatingphase. parametersexists in seamountsof the CES chain that lie close REE patterns (Figure 5) and heavy-REE ratios show no to the spreadingridge (Figure 2) [Morgan, 1985]. Variations evidence for garnet in the CES source. Probably, as with in Brown Bear seamountare reportedby Rhodeset al. [this basalts of the JDFR [Liias, 1986], the depth of melt issue]. segregationfor CES lavas was within the spinel stabilityfield. The basaltsfrom Thompsonseamount,consideredpart of Determining the petrogenesis of this basaltic suite is the CES chain [Delaney et al., 1981] although moving limited by the small numberof analysesfrom severalvolcanic eastwardwith the Juande Fuca plate, are more primitive and edifices erupted in a changing tectonic setting over 9 m.y. depleted in incompatible elements than other CES lavas However, the compositional similarity of the older Pacific (Figure 3) and, in some chemicalparameters,JDFR lavas. plate CES allow a commonhistory for theseseamounts,which Chemical analyses of the Seven Deadly Sins cluster are could range from multiple stages of melting of a single reportedby Rhodeset al. [this issue]. For almostall chemical homogeneoussource(dynamic partial melting) to similar melt criteria, the two samplesfrom Explorer and Union seamountsconditionsacting on a heterogeneoussource. In their studies (TI'063 DH 34 and 36A, respectively)have a far greater of the JDFR and near-ridgeseamounts(includingBrown Bear compositionalrange than do the Pacific plate CES; e.g., the and Son of Brown Bear), Morgan [1985], Liias [1986], and most extreme Mg-numbers and (La/Sm)N analyzedare from Karsten [1988] concluded that the mantle beneath the Juan de thosetwo volcanoes(Table 2), and they encompassa greater Fuca region must be heterogeneous on a small scale(hundreds range than the older CES on a plot of incompatibleelements of metersto a few kilometers). The variability of lavaswithin (Figure 3). Brown Bear seamount[Morgan, 1985] is one line of evidence Although many northeast Pacific seamount basalts are for heterogeneity of the upper mantle beneath the Cobbstrongly enriched in alkali and incompatible elements and Eickelberg region. Overlapping REE patterns (Figure 5) in ratiosrelativeto JDFR lavas[Cousenset al., 1985;Dalrymple CES basaltswith similar chemical characteristicsand a range et al., 1987; Cousens, 1988], the Cobb-Eickelberg,Explorer, in TiO 2 and other chemical parametersat constantM gO for and Union seamountsamplesstraddlethe line dividingalkalic basalts of the Juan de Fuca region (Figure 7) allow us to from tholeiitic basalts on a plot of alkalies versus silica conclude that the entire CES lineament was producedby (Figure 4) and are transitionalin their incompatibleelements preferential melting of compositionally similar, andratiosas well (Table 2). In somechemicalparameters (i.e., incompatible-element rich pods within a heterogeneous Zr/Nb, Zr/Y, P20s, andK20) CES lavasresemblethe Endeavour mantle source. segmentbasalts,which are enrichedrelativeto segments 1-3 Ratio-ratioplots involving four different elementsthat are of theJDFR (Table3) [Liias, 1986;Karsten,1988]. However, incompatiblein fractionatingphases(i.e., Zr/Y versusTi/Nb) ontheaverage CESbasalts areenriched in St,Zr, andP205but indicate mixing of two sources or magmas if sample aredepleted in Nb andhavehigherZr/Nband(La/Sm)Nrelative compositionsare linked by a curve [Langmuiret al., 1978]. In to Endcavour segmentlavas(Table3). CES lavasderivefroma a companion plot, one in which the ratios shown have the source (or sources) enriched over the JDFR source that same incompatible element in the denominator (i.e., Zr/Nb resembles but is distinguishable from the Endcavour segmentversus Ti/Nb), sample compositions will fall on a line if source. mixing has occurred. In both plots the samplecompositions On a chondrite-normalized REE plot (Figure5), the CES should lie in the same relative positionsunlessfractionation COBB-EICKEI•ERG SEAMOUNT COMPOSITIONS DESONIE AND Dt•c•: o o T• COBB-EICKELBERGSEAMOUNTCHAIN 12,701 12,702 DESONIE AND DUNCAN: TI-m COBB-EICKELBERG SEAMOUNt CHAIN o 1.4 0 O0 ,t basalts (Figures 8a and 8b). However, scatter within these plots precludessimple mixing of two discreteend-members. Rather, the patterns seen are more compatible with partial melting of incompatible-element enrichedsegregations within the heterogeneousnortheastern Pacific upper mantle and variablemixing with a relatively depletedJDFR source. To a first order,mixing betweenenrichedand depletedsourcescan also be seen along strike of the CES lineament, with more enriched values of incompatible elements (K20 + Na20) and ratios(La/Sm)N at the olderendof the chainandmorevariable (Brown Bear seamount)or less enrichedvalues (Cobb or Axial seamounts) at the youngerend (Figures2a and 2b). 0.2 Sr, Nd, and Pb Isotopes Although CES lavas are slightly enrichedin incompatible element concentrationsrelative to normal-MORB (n-MORB) from the JDFR (Table 3), Sr, Nd, and Pb isotopicratiosfor the CES and Explorer and Union seamountsplot well within the rangeexpectedfor n-MORB. Samplesfrom the seamounts fall virtually within the JDFR field in their Sr and Nd isotopic ratios (Figure 9a) [Eaby et al., 1984; Morgan, 1985; Liias, ß ß 8 1986], and there is little isotopic variation along the seamountlineament (Figure 9b). Only in Pb isotopicratios is there a difference in composition between the seamount 400 3•0 2•0 1•0 6 160 200 basaltsanalyzedand the Juande Fucaridge (Figures10a and 10b). All CES samplesfall in or near the JDFR field [Church Distance from Cobb hotspot, km and Tatsumoto,1975; Hegner and Tatsumoto,1987] but range Fig. 2. Variation in alkali and light-REE enrichmentwith distance to somewhat elevated 206Pb/204pbbut well within the larger from Cobbhotspot(Axial seamount).Basaltsdredgedfrom seamounts farthest west of Cobb hotspot show greater enrichment and less field for northeastPacific seamounts.The single samplefrom variability than seamountbasalts dredged near the ridge. Solid Union seamountis clearly distinctfrom the CES basaltsin Pb trianglesrepresentanalysesof Cobb, BrownBear, Axial, and Son of isotopic composition. BrownBear seamounts by Morgan [ 1985]. GEOC/mONOLOGY has affectedthe less incompatibleelements[Langmuiret al., 1978]. On most ratio-ratio plots (i.e., Figures 8a and 8b), CES lavas and near-ridgeCES [Morgan, 1985] plot closeto a mixing curve or line. These data suggestmixing of a low alkaliesand (La/Sm)N, high Zr/Nb end-memberexpressedmost stronglynear the spreadingridge, with a higher alkaliesand (La/Sm)N, lower Zr/Nb end-memberseenin the older CES 100 Eleven of the freshest basalt samples from the CobbEickelberg, Explorer, and Union seamountswere chosenfor age determinations by 4øAr/39Artotal fusion and conventional K-Ar radiometric dating methods [McDougall and Harrison, 1988; Dalrymple and Lanphere, 1969]. Becauseof the young ages and low potassium concentrationsin these lavas and samplesize limitations for irradiation, not enoughradiogenic ß u 10 O-- DH01-1 0 DH02A [-]-- DH02-8 l m DH04-4 ] DH05-14 X DH08-8 A DH10-7 ß CB-1 [•-- DH14-5 • DH34 []]-- DH36A 6sl•bl•a l•lbK I•a•e•r l•Idi• S'm•r•i•f •b Fig. 3. Geochemical patterns normalized to n-MORB[afterSunandNesbitt,1977];incompatibility of an elementin nMORB decreases to the right. DESONIEAND DUNCAN:THE COBB-EICK•.RERG SEAMOUNT CHAIN 12,703 10 in Table 4. The relatively large age uncertaintiesresult from measuringsmall amountsof radiogenic,,oArwithin a total ',OAr signal of largely atmosphericorigin. A replicate 40Ar-39Ar total fusion analysis was performed on one sample from Eickelbergseamount(Table 4). Agescoincidewithin their two sigma errors. Reactor irradiation is known to increase the atmosphericargon component [McDougall and Harrison, 1988] so conventionalK-At ageswere determinedfor Explorer seamountwhere sampleswere expected to be the youngest analyzed(i.e., smallestradiogenic•øAr) and Explorer samples were absolutelyfresh (no alteration). Crustal ageswere based ' I ' I ' I ' on magnetic anomaly interpretationsby Wilson et al. [1984] 30 40 50 60 70 and tectonicreconstructions by Karsten and Delaney [ 1989]. Figure 11 shows the variation in age of eight seamounts SiO 2wt.% with distancealong the seamountchain from Cobb hotspot, Fig. 4. Alkaliesvs. silicadiagramfor the CES, Explorer,andUnion assumingits location is defined by the center of Axial volcano seamounts. Northeast Pacific seamount field includes data from seamounts of the Pratt-Welker[Dairytopiceta!., 1987],Tuzo Wilson, (zero age). Distanceswere measuredalong the lineament,in Bowie, and Dellwoodknoll seamountchains[Cousenset al., 1985; the directionof Pacific plate motion over the mantle [Duncan Cousens,1988]. Near-ridgeCES field includesdatafrom BrownBear, and Clague, 1985; Pollitz, 1988]. Also shownon this figure Sonof BrownBear,and Axial seamounts [Morgan,1985]. TheJDFR are the ages determinedfor Union and Explorer seamounts field is from Liias [1986] and Karsten [1988] and includesthe Endcavour segment.Alkali basalt-tholeiite boundaryfromMacdonald plotted againstdistancefrom Endcavourand Middle Valley and Katsura [ 1964]. spreading segments, respectively (Figure 1). Volcanic activity may continue for as long as 3 m.y. at a given seamount(e.g., Pratt-Welkerseamounts[Turner et al., 1980]); SOAr waspresentfor analysis by n0Ar_39Ar incremental heating a 2-m.y. spanwas recordedin the threesamplesanalyzedfrom '•:• '"'"'";", "",, -"'•:' ' near-ridge CES / ••"-Juan de Fuca Ridge experiments. Basaltsamples werecrushed andsievedto obtain Eickelbergseamount.Therefore,we haveusedthe oldest the0.5and1.0mmsizefraction, thenwashed ultrasonically in sample fromeachseamount wheremorethanonesample was distilledwater. For '*0Ar-ZgArtotal fusion experiments,analyzedas an estimateof the time whenthe volcanowas samples were irradiatedfor 6-10 hoursin the coreof the locateddirectlyover the hotspot. Replicateanalyses for Oregon StateUniversity TRIGAreactor, wheretheyreceived a Eickelberg sampleDH02Awereaveraged.Oldestseamount neutronflux of 0.6-1.0x 1017nvt. Isotopes of argonwere ageswereusedin a least-squares linearregression to determine measured on an AEI MS-10Smassspectrometer, aftersample an averagerate of platemotionoverthe hotspotof 43 + 3 fusion via radio frequencyinductionheatingand gas km/Ma. A least-squares regression of all PacificplateCES purification; potassium contents weredetermined by atomic includingall threeEickelberg pointsgivesa rate of plate absorptionspectrophotometry.Analytical data and age motionoverthehotspot of 47 + 3 km/Ma. AgesfromUnion calculations fromthe'mAr-a9Ar andK-Ar experiments aregiven andExplorerseamounts werenot includedin the regression, TABLE 3. AverageCompositions of SelectedChemicalElementsandRatiosfor Different TectonicSettingsin the Juande FucaRegion Chemical JDFR EndcavourNear-Ridge Axial Near-Ridge PacificPlate ParameterSegments 1-3a Segment b Seamounts c Seamount a CES d CES e K20 (wt.%) P205(wt.%) 0.18 0.17 0.42 0.23 0.12 0.07 0.15 0.13 0.17 0.13 0.43 0.28 Sr ppm Zr ppm Nb ppm 113 104 5.3 211 134 13.9 158 54 2.0 136 92 4.3 159 93 4.2 271 145 10.4 Zr• Zr/Y 22.1 3.10 9.7 4.58 28.1 2.17 23.3 3.26 28.1 3.66 14.0 3.84 (La/Sm) N 0.86 a0.70 0.76 0.74 1.24 --- a FromLiias [1986]. b From Karsten [1988]. c Includes Heck,HeckleandSpringfield seamount chains, fromKarsten[1988]. d Includes Cobb, Brown Bear, andSon ofBrown Bear seamounts, data from Morgan [1985], averagedby Liias [1986]. e Includes Eickelberg, unnamed (except forP205wt.%),Warwick, andCobb seamounts, from this study. 12,704 DESONIEAND DUNCAN:THE COBB-EICKF•BERG SEAMOUNT CHAIN 100 O-- 1o i C½ i i La I I i La Ce Nd I Nd I Sm Eu Smi Eui t DH01-1 • DH02A {"]• DH02-8 m_ DH04-4 i DH05-14 X DH08-8 A DH10-7 ß CB 1 [] DH14-5 • v DH34 [] DH36A t Lui Tb Yb I I Tb I Yb Lu Fig. 5. Chondrite-normalized [afterNakamura,1974]REEpatterns for seamounts of theCES,Explorer, andUnion seamounts. JDFR(includingEndeavour segment) field shownin stipplepauem[Wakeham,1978;Liias, 1986]. Analyses fromsamples dredged neartheBlanco fracture zone(dredges 6 and7) fromWakeham [1978]werenotincluded. Insetshows REE patterns for threebasalts fromWarwick seamount. beingfromseparate provinces, but thesedataaregenerallyThis changein the crustalage at the time of seamount consistent with thePacificplatemotioninferredfromtheCES volcanismis due to the westwardmigrationof the JDFR, age-distance relationship. toward Cobbhotspot, sinceatleast10Ma [Riddihough, 1984; Two agepatterns canbe seenin Figure11: KarstenandDelaney,1989]. 1. Agesincrease tothenorthwest alongtheCESchainfrom The line of seamounts from Axial throughEickelberg, Axialseamount at zeroageonthespreading ridge,to 3.3Ma at althoughshort,is subparallelwith other more prominent Cobbseamount, some105 km from the ridge,andon to 9.0 Pacificchainsthat are associated with stationarylong-lived Ma at Eickelberg seamount, about340 km fromthe ridge. mantlehotspots. Thevolcanic migration rateof 43+ 3 km/Ma Union seamountis one memberof a shortPacific plate is consistentwith a plate-mantlevelocity of 44 km/Ma volcanicchainfartherto thenorth. The zeroagepositionof predictedfrom the latestPacificplaterotationpole at 62øN, theassociated meltinganomaly is unknown; we haveassumed94øWandrotationrateof 0.95ø/Ma[Pollitz, 1988], which it tobetheEndearour ridgesegment (Table4). applies fortheperiod0-9Ma. NewagedatafromtheCES,as 2. Thereis a northwesterly increase in thedifference in the wellasthePratt-Welker seamounts [Dalrymple et al., 1987], ageof theseamounts andthecruston whichtheyformed.Samoa [McDougall, 1985],andtheLouisville seamount chain Axialseamount is currently forming ontheridgeonzero-age[Wattset al., 1988]support the ideaof rigidPacificplate crustbutEickelberg seamount formed oncrustthatwas2-3Ma. motionovera fixedconstellation of hotspots (Figure12). 1.0 3.0 •ø!• cpxPlag o 0.9 o o 2.0- 0.8 o o o o fl o o ß ß []0 *•1 oO 0.7 o 0.6 40 1.0- o ' 5'0 ' Mg-number 6'0 ' 70 MgO wt. % MgOvariation diagram forbasalts fromtheJDFR Fig. 6. Fractionation trendsfor olivine(ol), clinopyroxene (cpx), Fig.7. TiO2 versus andplagioclase (plag)andvariation of of CaO/A1203 withMg-number (solid squares)[Liias, 1986], Axial and the near-ridgeCES (open squares) [Morgan,1985],andtheolderCES(circles,thisstudy). for Cobb-Eickelberg, Explorer,andUnionseamount basalts. DESONIE AND DUNCAN: THE COBB-EICKELBEROSEAMOUNTCHA• 7 (a) 6o 5- 4- 3- 12,705 usually distinguishhotspot from spreadingridge lavas, the CES basaltscannotbe separatedfrom MORB compositions. The lack of distinguishing St, Nd, Pb, andHe isotopicratiosis a feature of other northeast Pacific seamountand spreading ridgelavas[Churchand Tatsumoto,1975; Lupton, 1982;Eaby et al., 1984; Morgan, 1985; Liias, 1986]. No significant isotopicvariationhas yet been found in basaltsfrom the PrattWelker [Churchand Tatsumoto,1975;Hegner and Tatsumoto, 1989], Heckle [Hegner and Tatsumoto, 1989], Tuzo Wilson [Cousenset al., 1985], Bowie [Cousenset al., 1985; Cousens, 1988], Dellwood knolls [Cousenset al., 1984], or President Jackson (W. M. White, personal communication, 1988) 2 seamount groups. 60- Models 50- Several models may explain the age distribution and chemicalcompositionsof the CES basalts. 40- Passivelyupwellingheterogeneous mantleand a migrating spreading ridge. The CES may be the result of the same processesthat form near-ridge seamountsalong the East 3020- (b) 100 0 1000 2000 3000 4000 5000 6000 Ti/Nb Pacific Rise [Batiza and Vanko, 1983, 1984; Zindler et al., 1984; Fornail et al., 1988a, b] and Juan de Fuca and Gorda ridges[Davis and Karsten,1986; Karstenand Delaney, 1989]. Most of thesesmall near-ridgeseamountclustersare thought to be the result of volcanism above a short-lived melting anomalyembeddedin the uppermanfielateral flow, ratherthan Fig. 8. Incompatible elementratio-ratiomixingplots[afterLangmuir a deep mantle plume. Following the Batiza and Vanko [1983, et al., 1978]. Axial seamountsamplesare shownin solid squares [Morgan,1985],near-ridgeCES (includingBrownBear,Cobb,Axial, and Son of Brown Bear seamounts)are shown in open squares [Morgan, 1985], and older CES analysesare shownin opencircles 0.5133 . (this study). JDFR and Endcavoursegmentfields are labelled[Liias, 1986; Karsten, 1988]. NE Pacific Se DISCUSSION Summaryof Temporaland ChemicalCharacteristics of the CES The CES chain has the temporal and morphologic characteristicsof volcanismgeneratedover a hotspotfed by a mantleplume. The seamountchain is age progressiveand the linear geometry and age distribution of volcanoes are compatiblewith the motion of the Pacific plate over other well-establishedhotspots. Age relationshipsalong the CES 40 39 chain, determined by Ar- Ar and K-Ar analyses, in conjunctionwith the tectonic reconstructionof the Juan de Fuca region of Karsten and Delaney [1989] require a fixed source for the CES lineament and a westwardly migrating spreading ridge. Our data demonstrate that the melting anomaly currently beneath Axial seamountis a stationary, persistentfocusof unusuallyhigh magmasupply,even though volcanic activity along the lineament may have been intermittent, with the Eickelberg to Axial seamountsection being the latest "pulse"of plume flow. In most respects,CES lavas are transitionalbetweenbasalts characteristicof hotspotsand those of segments1-3 of the JDFR; they are similar to those of the Endeavoursegment. Compositional variability among CES basalts probably 0.5132 0.5131 (a) 0.5130 i 0.7024 0.7023 ß0.7 b25 ' 0.7026 I ß0.7027 I ß0.7028 87Sr/86Sr 0.7028 (b) 0.7027 0.7026 0.7025 coø ß o o o 0.7024 0.7023 .... 400 30'0 2to 0'0 6 ' ' 200 Distance from Cobb hotspot,km reflects arange ofpartial melting ofthemantle supplied tothe Fig.9. (a)143Nd/•44Nd versus S7Sr/S6Sr forCES and Explorer and meltinganomaly,which is heterogeneous with respectto Union samples withJDFR[Hegner andTatsumoto, 1987]andother NE incompatibleelements[Morgan, 1985;Liias, 1986; Karsten, Pacificseamount [HegnerandTatsumoto, 1985;Cousens et al., 1985; 1988]. The eastward trendtowarddepletedcompositions Cousens, 1988)fieldssuperimposed. Range inisotopic ratios shown indicates thatprogressively greater proportions ofthesource is ton-MORB field. (b)equal 87Sr/a6Sr versus distance fromCobbhotspotfor samples fromthe for JDFRbasalts participated in meltingfor CESbasalts.In CESlineament. Solid triangles represent analyses ofBrown Bear and terms of the St, Nd, and Pb isotopic compositions,which Axialseamounts by Eabyet al., [1984]. 12,706 DESONIEnN• DUN•: T•m COBB-EIC•:•ERO SEAMOUNTCHAIN 15.60 1984] and Zindler et al. [1984] model for small seamountsnear the East Pacific Rise, a somewhatsmaller degree of partial melting of the sameheterogeneous upper mantle sourcethat is producing the JDFR lavas would result in melts somewhat enriched in incompatible elements and alkalies but with essentially MORB-like isotopic composition. This mechanismcould easily explain the modest enrichment in alkalies of the CES lavas and might. with an even smaller degree of partial melting, explain the alkali basalts and 15.55 15.50' 15.45 hawaiites of other northeastern Pacific seamounts. The scale of volcanic activity at such near-ridge seamounts is,however, much smaller than intheCESprovince. A •q•ical 15.40 near-ridgeseamount•s 100 km, comparedwith 1000 km for a Cobb-Eickelberg cone. The duration of volcanism for an 38.50 entire near-ridge seamountchain can be brief: the Heckle chainnear the Juande Fuca ridge beganactivity at ~2 Ma and ceasedat 0.5 Ma [Karsten and Delaney, 1989]. Volcanism alongthe CES chainhaspersistedfor at least9 m.y. Although it is possible that a large passiveupper mantle melting anomalycould accountfor the increasedvolume and duration of volcanism at Cobb hotspot, we believe that a melting anomaly that has been fixed with respect to the hotspotreference frame for at least 9 m.y. must be rooted 37.30' , , , , - , ' 18.4 18•618.819.019.2 19.4 19.6 below the asthenosphere.A melting anomaly, of the type responsible for volcanism at small near-ridge seamount chains,wouldbe sweptalong with uppermantleflow over the time scalerequired. Therefore,we believe that Cobb hotspot ' 10.(a) •7pb/2ø4Pb versus •6pb/•ø•Pb and(b) •Pb/•ø•Pbversus has been maintained by a mantle plume during the CES PbforCES,Explorer, andUnionsamples compared withJuan formation. Also, near-ridge seamountbasaltsin the Juan de de Fucaridge [Churchand Tatsumoto,1975; Hegner and Tatsumoto, Fuca region, i.e. Heck, Heckle and Springfield [Karsten, 1987] andNE Pacificseamounts [Churchand Tatsumoto,1975;Hegner 1988], are more depletedin alkali and incompatibleelements and Tatsumoto,1989; Cousens,1988] fields. The samplefrom Union seamount(DH36A) is distinct. Range in isotopicratios shown is at a given MgO content than the spreadingridge basalts,in •.•38.10 "-':• 37.70 •' .................... 'NEPacific '"• Seamounts 206pb•04pb contrast with the CES lavas. slightlylarger thanvaluesseenin n-MORB basalts. TABLE 4. 40Ar-39Ax andK-AtTotal Fusion Ages andAnalytical DataforWhole Rock Basalt Samples From theCobbEickelbergand ExplorerandUnion Seamounts Sample J-Factor 40Ar/39Ar 36Ar/39Ar37ArC/39Ar %K •r080 DHOl-1 •r080 DH02A replicate •r080 DH02-8 •r080 DH08-8 •r080 DH 10-7 •r175 DH 70-2 •r175 DH 71-4 •r175 DH 74-17 CB-1 •r080 DH 14-5 •r063 DH 34 0.001957 0.002446 0.002446 0.002446 0.001957 0.001292 0.003050 0.003050 0.003050 0.003050 0.002446 0.003050 21.269 15.865 40.990 21.768 28.394 23.862 35.811 29.516 55.674 11.161 163.580 30.274 0.06435 0.05083 0.13615 0.07212 0.09047 0.77980 0.12171 0.10218 0.19423 0.04158 0.56704 0.11003 22.052 15.574 12.812 14.542 16.109 22.765 14.064 18.908 25.713 22.013 60.010 29.057 •r063 DH 35 0.003050 35.303 0.13252 50.211 Tr063 DH 36A 0.003050 8.875 0.29922 *40Armol/g%40At Age(:L1sigma)Ma (xlO-9) 11.91 12.69 4.16 7.35 6.85 12.61 2.64 2.72 0.51 0.44 0.07 0.666 4.375 0.489 6.510 0.16 8.696 1.14 9.03 8.98 7.52 7.05 6.91 7.73 5.20 4.40 1.55 3.27 3.19 0.12 0.17 0.31 0.35 5.78 (0.41) (0.75) (0.46) (0.15) (0.30) (0.33) (0.32) (1.07) (1.40) (0.30) (1.30) (0.57) (0.03) (0.37) (0.04) (0.65) Ages calculated using thefollowing decay and abundance constants: •œ=0.581x10 -10yr'l;)•[1=4.963x10-10 yr-1; 40K/K=l.167x10-4 mol/mol.Neutronfluxmonitored withhornblende standard MMhb-1. Correctedfor decaysinceirradiation. DESON[E AND DU•C•: TI• COBB-EICKELBERG S.EAMOUNTCHAIN 12,707 10.0 Eickelberg ß unnamed 8.0 x\\ Warwick \xxx"x 6.0 U t xx 4.0 2.0 Gluttony J • Kilometers Exp•'or,er from ridge Fig.11.Seamount ageversus distance fromspreading ridgeforCESandExplorer andUnionseamounts. K-ArageofCobb seamount by Dymond et al. [1968]shown astriangle.Measured distance forCESisfromCobbhotspot, defined asthe centerof Axial volcano(46øN,130øW),alongthedirection of Pacificplatemotion[Pollitz,1988]. Crustalages[after Wilsonet al., 1984]areshown by thedashed line;a breakindicates location of a pseudofault. Average rateof motionof thePacificplateoverthehotspot wascalculated to be 43 +_.3 km/Ma. Replicate analyses of sampleDH02Afrom Eickelbergseamountwere averaged. Compositionally buoyant mantle plume and a migrating spreading ridge. Compositionalbuoyancyoccurswhen plume flow is driven by intrinsic density differencesbetween the diapir and its surroundings[Griffiths, 1986a]. Chemical diffusion proceeds extremely slowly so that compositional diapirs maintain their chemical identity relative to the upper mantle material throughwhich they rise [Griffiths, 1986a]. A compositionalplume that has moderatelyenrichedtrace element contentsbut a nonradiogenicisotopiccharactercould originate in several ways. For example, the MORB-like isotopic composition of all seamountsof the northeastern Pacific may reflect a broad regional difference in mantle compositionfrom the equatorialoceanicregions where Dupal hotspotsare found [Hart, 1984]. Hart [1988], who noticed a poleward trend toward less radiogenicisotopic compositions in hotspot-associatedbasalts, termed the regional isotopic homogeneity of the northeastern Pacific the anti-Dupal anomaly. He proposed that the lower mantle which feeds plumes may be compositionally variable on a very long -16 -14 -12 lengthscale(10'• kin) and someregionsmay be isotopically MORB-like if large amountsof melts have been previously extracted from them. A metasomaticevent that enrichedthe mantle in tb.eregion 20 40 60 80 100 120 140 160 in Rb, Sm, and U, along with other incompatible elements AngularDistance from RotationPole (62o N, 94ø W) [Menzies and Murthy, 1980], but so recently that their Fig. 12. Volcanicmigrationrate as a functionof rotationpole located elevated concentrationshave not yet significantly shifted the by at 62ø N, 94ø W [Pollitz,1988]. Thedottedcurveis a least-squares Sr andNd isotoperatiosfrom the MORB field wassuggested bestfit withan angular rotationrateof 0.95+ 0.02ø/Ma. Cousenset al. [1985] to explain the alkalic but isotopically i i i i i i i i 12,708 DESONIE AND DUNCAN: T}ta COBB-EICK•RERG SEAMOUNTCHAIN MORB-like compositions of the Tuzo Wilson seamounts. argued that the unusual isotopic structure of the Galapagos This explanation could also apply to the transitionallavas of archipelago,in which Sr isotopic ratios decreasetoward the the CES chain. centerof the hotspot,may result from the thermalentraimment The proximity of Cobb hotspotto the Juan de Fuca ridge processproposedby Griffiths [ 1986a, b]. Thermal entrainmentcan also occur in a continuousplume may provide anotherexplanationfor the unusualchemistryof the CES basalts. A weak supply of relatively enrichedplume which is deflected by mantle shear flow beneatha spreading material to a hotspot in a spreading ridge setting may be ridge [Richards, 1988]. In this case,as well as for discrete diluted by mixing with the MORB partial melt zone. That is, diapirs,the most MORB-like magmasare to be expectedfrom the zone of partial melting below the ridge may extendat least the centerof the plume. Therefore,the isotopiccomposition 130 km away from the spreadingaxis. This model of hotspot of eruptedmagmasmay be an important distinctionbetween and spreadingridge interactionis easilytestable. Older Gulf of chemical and thermal plumes, regardlessof possible smallAlaska seamounts(e.g., Miller, Patton, and Murray) proposed scale heterogeneity and variable source. A large volume of to have formed at hotspotsfar removedfrom a spreadingridge entrained material implies a sharply reduced temperature at the time of seamount formation should exhibit a more contrastbetweenplume and normal manfie. In reality plumesmay occur from a combinationof thermal typical OIB signatureif they formed from a plume with OIB composition. If basaltsfrom suchseamountsare found to have and compositionalinstabilities,but one effect may dominate. MORB compositions they would almost certainly have The hotspotthat producedthe Louisville seamountchain (LSC) resulted from a plume with MORB chemical characteristics may have been supplied by a largely compositionalplume. since melts from those hotspotscould not easily have mixed Like many hotspot tracks, the LSC has an associated with spreadingridge melts. The alkalic lavasfrom Murray and topographic swell, in this case several hundredsof meters Patton seamounts,which erupted through 11-12 Ma oceanic high, and a volcanic eruption rate of 3-4000 km3/Ma from 70 crust [Dalrymple et al., 1987], would have been 550 km west Ma until 20 Ma, when it waned sharply[Lonsdale, 1988]. The of the spreading ridge system and isotopic data could be LSC shows a very narrow range of St, Nd, and Pb isotopic expected to detect an OIB isotopic plume signature, if one values over the 65-m.y. range of available samples. These existed. basaltsshow radiogenicSr, Nd, and Pb isotopiccompositions, Interaction of a plume of enrichedmantle compositionwith typical of OIB lavas [Cheng et al., 1987]. Based on the oceanic lithosphere cannot explain CES chemistry. For the consistently enriched compositions of Louisville samples, Hawaiian Islands, a model of mixing between the enriched Cheng et al. [1987] suggestedthat the LSC was formed by a mantle (EM) plume and a small degreeof melting of oceanic compositional plume that originated from a long-lived, mantle source. This plume was lithospherethrough which the plume rises, can explain the stationaryand homogeneous distinct incompatibleelement and isotopic compositionsof not noticeably contaminated by overlying asthenosphereor tholeiitic and alkalic lavas [Chen and Frey, 1983]. A similar lithosphereas were plumes that form other hotspot chains mixing model can also be appliedto the CES lavas and other [Cheng et al., 1987, and referencestherein]. In contrast,the more intermittentand consistentlyMORBseamountlavas of the northeastPacific [Cousenset al., 1985]. A plot of La/Ce vs 87Sr/86Sr[Chen and Frey, 1983, Figure 3] like melts from Cobb hotspot perhaps manifest a thermal shows that CES lavas could result from 0.25% to more than plume. Such a plume may rise throughand entrainthe MORB3.0% partial melt of lithospheric(MORB) material mixed with like upper mantle, supplyingthe melting anomalyat the base EM, in proportionsof about 1:4, a larger MORB component of the lithosphere. To produce the observed isotopic than is seen in Hawaii. Samplessuch as the basalt from Cobb compositions,we would expect the proportion of MORBseamountcould be generatedonly if the La/Ce of the upper componentto be large. If a large amount of the final plume mantle beneath the northeast Pacific is lower than the one volume is entrained material, the plume might be broad but postulatedin the Chen and Frey [1983] model for Hawaii. would be thermally dilute (cooled); thus, resulting volcanism However, the required variation in degree of melting of the might be volumetrically small compared with that of other lithosphere, followed by mixing of nearly constant hotspots. Furthermore, thermal entrainment increases proportions of MORB and EM componentswould result in strongly with decreasing thermal plume Rayleigh number 87Sr/86Srenrichmentwith increasingdegreeof melting of the [Griffiths, 1986b]. Plumes that start off weaker (smaller lithosphere, a trend not seen in the CES lineament. The diameter and lower temperature contrast) suffer stronger observedcompositionsseem to point strongly to the absence entrainment. This is consistentwith a relatively "weak" of a long-termenrichedmantle componentin lavasof the CES (thermal)plume sourcefor the CES. and other northeast Pacific lineaments. Thermally buoyantplume and a migratingspreadingridge. SummaryModel of the Formation of a plume that has MORB isotopiccharacteristics The major and trace element chemicalcharacteristics might result from thermal entrainmentof MORB uppermantle CES are best explainedby mixing of enrichedand depleted by a plume of lower mantle composition. Thermally buoyant lavasresultingfrom partial meltingof a heterogeneous upper plumesare driven entirelyby viscosityand densitydifferences mantle in both a hotspotand spreadingridge meltingregime. createdby a temperaturegradient acrossa thermal boundary Like JDFRmelts,CES liquidsprobablysegregate in the spinel layer [Griffiths, 1986a]. Diffusion of heat from suchthermal lherzolite stability field. CES lavas are also similar to all diapirs warms the surroundingmantle and lowers its density northeasternPacific seamountand spreadingridge basalts sufficiently for the plume to entrain its initially cooler analyzed thus far in that they show no enrichment in surroundingsbut conserve its total heat content, buoyancy, radiogenic isotopesover normal-MORB. Thus, there is no and driving force. The thermalplume will be a mixture of the evidenceto suggestthat Cobb hotspotis a compositionally diapiric material and surroundingmaterial entrainedtoward its buoyantmantleplume. Although,the chemicalcharacteristics center [Griffiths, 1986a]. Recently, Geist et al. [1988] have of the lavascan be explainedby a largepassiveuppermantle DESONIEAND DUNCAN:• COBB-EICKELBERG SEAMOUNT CHAIN 12,709 melting anomaly, we believe that, to have remained 1984]. At 8 Ma the JDFR lay about 130 km east of Cobb stationary,the sourceof the CES lineamentmustbe rooted hotspot(Figure 13). Eickelbergseamount was thenforming beneath upper mantle lateral flow. The most likely abovethehotspoton seaflooraged2-3 m.y. Proximityof the explanationfor the unusualgeochemicalcharacteristics of Cobb hotspotto the JDFR may have resultedin dilutionof Cobb hotspotis a deeplyrooted,thermallybuoyantdiapir plumematerialby mixingwith the MORB partialmelt. The whichhasentrained nonradiogenic uppermantleMORBsourcenextseamount to theeast(unnamed) hadbegunformingwhile material. The large,thermallydiluteplumesuppliesthe Eickelberg seamount was still active,-7.8 Ma. By 6 Ma, hotspotwith volumesof warm mantlein excessof that Eickelberg seamount hadmoved88 km westof thehotspot, upwelling alongthespreading ridge,resulting in thebuildup theJuande Fucaridgehadmoved35 km closerto thehotspot of thevolcanic cones of theCESchain.Progressively greaterandWarwickseamount wasforming on-2 Ma crust.At -4 Ma, mixingof plumeand spreading ridgemantlesources has Slothseamount hadformedon 1.6-m.y.crustandhadmoved occurred astheIDFR approached thehotspot. morethan50 km to the west;Lustseamount had formedon Agerelationships alongtheCESlineament since9 Ma 1.8-m.y.crustwhiletheridgewasonly65 km fromthe permitgreater resolution of theKarsten andDelaney [1989] hotspot.Although a largeuncertainty is associated withthe modelforhotspot-spreading ridgeinteraction. In thismodela ageof Thompson seamount it is consistent withanoriginat westward migration of theIDFRtoward a fixedCobbhotspottheridgeat~3.2Ma,followed bytransport eastatarateof -34 wasdetermined fromanalysis of thepropogating rift and km/MawiththeJuan deFuca plate[Desonie andDuncan, 1986; spreading history of theridge[Wilson,1984;Riddihough, Karstenand Delaney,1989]. Cobbseamount formedat approximatelythe same time as Thompsonseamounton crust of age 0.4 Ma. Interaction of spreading ridge and hotspot partial melt zonesresultedin lessenriched(Cobb seamount)or more variable (Brown Bear seamount)chemical characteristics in the seamountsnear the Juan de Fuca spreadingridge. Axial seamountis currentlyforming over the ridge-centeredhotspot. E • 8MaI 400 • 300 200 100 0 100 200 EXW ?-•,','.....'•?•'4'•:,:•:•'?•'•' ..... I 400 I 300 I-'•:' 200 I 100 EXW 300 200 EXW .... the guidance of M. Cheatham. We thank R. Walker of the Radiation Center at OSU for prompt INAA results. Age 6 Ma 400 ".....:'•-.."!•t ::•'"" '"'1•'"' ' ' I 0 100 S 100 S L I 200 0 C 100 were obtained with the assistance of L. G. REFERENCES 200 Allan, J. F., B. L. Cousens,R. L. Chase,M.P. Gorton, and P. J. Michael, Alkaline lavas from the Tuzo Wilson seamounts, NE Pacific:Volcanismat a complexspreadingridge-transform fault intersection,Eos Trans.AGU, 69, 1503, 1988. Bargat, K. E., and E. D. Jackson,Calculatedvolumesof individual shieldvolcanoes alongthe Hawaii-Emporer seamount chain,J. Res. T ,,•,,•,, U.S. Geol. Surv.,2,545-550, 1974. I 2Ma• 3o0 determinations Hogan, also of OSU. Applicationof the thermal entrainment model developed from discussionswith M. Richards of the University of California, Berkeley. Reviews by A. Grunder, M. Richards,and W. Bryan were extremely valuable. An especially thoroughreview by J. Karsten led to substantial improvementof the manuscript. L ,,,•,•,,,,,,,,,,,,,,,...,,•,,,, 400 Acknowledgments. We are grateful to M. Perfit of the University of Florida (OCE-8716827) and P. Hooper of WashingtonState University for providingXRF data. W.M. White allowedus the use of his massspectrometry facilitiesat Cornell University. Isotopic analyseswere performedunder 200 •oo 0 100 C A T 200 Batiza,R., and D. Vanko, Volcanicdevelopment of smalloceanic central volcanoes on the flanks of the East Pacific Rise inferred EXW S L from narrow-beamecho-sounder surveys,Mar. Geol.,54, 53-90, 1983. ::':!i.!!:!;;i;!!:iii::; :,/' Batiza,R., andD. Vanko,Petrologyof youngPacificseamounts, J. Geophys. Res., 89, 11,235-11,260, 1984. Chase,R. L., J. TuzoWilsonKnolls:Canadian hotspot, Nature,266, 344-346, 4oo 3o0 200 • oo 0 100 200 1977. Chase,T. E., H. W. Menard,andJ. Mammerickx, Bathyrnetry of the NorthPacific,charts3 and4, Instituteof MarineResources, Scripps Instituteof Oceanography,La Jolla, Calif., 1970. Fig. 13. Locationsof Cobb hotspotand Juande Fuca ridge, in 2-m.y. Chen, C. -Y., and F. A. Frey, Origin of Hawaiian tholeiite and alkalic time incrementsfrom 8 Ma to present [after Karsten and Delaney, basalt,Nature, 302,785-789, 1983. 1989] with improved resolutionby 39Ar-39Ar total fusion age Cheng,Q., K.-H. Park,J. D. Macdougall,A. Zindler,G. W. Lugmair, determinationson basaltsof the CES chain. Large arrows indicate H. Staudigal,J. Hawkins,and P. Lonsdale,Isotopicevidencefor a motion of the Pacific and Juande Fucaplates. Age of crustat the time hotspot origin of the Louisville seamountchain, in Seamounts, of eruptionof each seamountis given. Symbolsfor seamountnames Islandsand Atolls, Geophys.Monogr. Ser., vol. 43, editedby B. as in Table 1. H. Keatinget al., pp. 283-296,AGU, Washington, D.C., 1987. 12,710 DESONIEAND DUNCAN: THE COBB-EICK•.RERG SEAMOUNTCHAIN Endcavoursegment,Juan de Fuca Ridge, Ph.D. thesis,329 pp., ridgebasaltsfrom theJuande Fuca-Gorda Ridgearea,N. E. Pacific Univ. of Wash.,, Seattle, 1988. Ocean,Contrib.Mineral. Petrol.,53, 253-279, 1975. Karsten,J. L., and J. R. Delaney,Hot spot-ridgecrestconvergence in Cousens,B. L., Isotopically depleted, alkalic lavas from Bowie the northeastPacific, J. Geophys.Res., 94, 700-712, 1989. Church, S. E., and M. Tatsumoto,Lead isotoperelationsin oceanic Seamount,NortheastPacific Ocean, Can. J. Earth Sci., 25, 1708- Langmuir, C.H., J.F. Bender, A.E. Bence, and G.N. Hanson, Petrogenesisof basalts from the FAMOUS area: Mid-Atlantic Cousens,B. L., R. L. Chase,and J. -G. Schilling, Basaltgeochemistry Ridge, Earth Planet. Sci. Lett., 36, 133-156, 1977. of the ExplorerRidgearea,northeastPacificOcean,Can. J. Earth Langmuir,C. H., R. D. Vocke, G. N. Hanson,andS. R. Hart, A general Sci.,21, 157-170, 1984. mixing equation with applications to Icelandic basalts, Earth Planet. Sci. Lett., 37, 380-392, 1978. Cousens,B. L., R. L. Chase, and J. -G. Schilling, Geochemistryand origin of volcanicrocksfrom Tuzo Wilson and Bowie seamounts, Laul, J. C., Neutron activation analysisof geologicalmaterials,At. Energy Rev., 17, 603-695, 1979. northeastPacific Ocean, Can. J. Earth Sci., 22, 1609-1617, 1985. Liias, R. A., Geochemistry and petrogenesis of basaltseruptedalong Dalrymple, G. B., D. A. Clague, and T. L. Vallier, 40Ar/39Ar age, the Juan de Fuca Ridge, Ph.D. thesis, 264 pp., Univ. of Mass., petrology,and tectonic significanceof some seamountsin the Gulf Amherst, 1986. of Alaska, in Seamounts,Islands, and Atolls, Geophys. Monogr. Lonsdale, P., Geographyand history of the Louisville HotspotChain Ser., vol. 43, edited by B. H. Keating et al., pp. 297-315, AGU, in the SouthwestPacific,J. Geophys.Res.,93, 3078-3104, 1988. Washington,D.C., 1987. Lupton,J. E., Helium isotopevariationsin Juande FucaRidgebasalts, Dairytopic,G. B., and M. A. Lanphere,Potassium-ArgonDating: Eos Trans. AGU, 63, 1147, 1982. Principles, Techniquesand Applicationsto Geochronology,258 Macdonald,G. A., and T. Katsura,Chemical compositionof Hawaiian pp., FreemanCooper, San Francisco,Calif., 1969. lavas, J. Petrol., 5, 82-133, 1964. Davis, E. E., and J. L. Karsten, On the cause of the asymmetric McDougall, I., Age and evolution of Tutuila, America Samoa,Pac. distributionof seamountsaboutthe Juande Fuca Ridge: Ridge-crest Sci., 39, 311-320, 1985. migration over a heterogeneousasthenosphere, Earth Planet. Sci. McDougall, I., and T. M. Harrison, Geochronology and 1716, 1988. Lett., 79, 385-396, 1986. Thermochronology by the '•øAr-439ArMethod, Monogr. Geol. Delaney, J. R., H. P. Johnson,and J. L. Karsten, The Juan de Fuca Geophys., no. 9, 272 pp., Oxford University Press,New York, 1988. ridge-hotspot-propogating rift system:New tectonic,geochemical, and magneticdata,J. Geophys.Res.,86, 11,747-11,750, 1981. Menzies, M. A., and V. R. Murthy, Nd and Sr isotopegeochemistry of Desonie, D. L., and R. A. Duncan, Spreadingridge and hotspot hydrousmantle nodulesand their host alkali basalts: Implications contributions to seamounts nearthe Juande FucaRidge,Eos Trans for local heterogeneitiesin metasomaticallyveined mantle, Earth AGU, 67, 1231, 1986. Planet. Sci. Lett., 46, 323-334, 1980. Dixon, J. E., E. Stolper,and J. R. Delaney, Infrared spectroscopic Morgan, C., Geochemistryof basaltsfrom the Cobb hotspotastride measurementsof CO2 and H20 in Juan de Fuca Ridge basaltic the Juan de Fuca ridge, M.S. thesis, 138 pp., Univ. of Mass., Amherst, 1985. glasses,Earth Planet. Sci. Lett., 90, 87-104, 1988. Duncan,R. A., and D. A. Clague,PacificPlatemotionrecordedby Morgan, W. J., Deep mantle convectiveplumes and plate motions, Am. Assoc. Pet. Geol. Bull., 56, 201-213, 1972. linearvolcanicchains,in The OceanBasinsand Margins,vol. 7A, editedby A. E. Narin et al., pp. 89-121, Plenum,1985. Morgan, W. J., Rodriguez,Darwin, Amsterdam.... A secondtype of Dymond,J. R., N. D. Watkins, and V. R. Naydu, Age of the Cobb hot spotisland,J. Geophys.Res., 83, 5355-5360, 1978. Seamount,J. Geophys.Res.,73, 3977-3979, 1968. Nakamura, N., Determinationof REE, Ba, Mg, Na and K in Eaby, J., D. A. Clague,and J. R. Delaney,Sr isotopicvariationsalong carbonaceous and ordinarychondrites, Geochim.Cosmochim. Acta, 38, 757-775, 1974. the Juande Fuca Ridge, J. Geophys.Res.,89, 7883-7890, 1984. Fornari, D. J., M. R. Perfit, J. F. Allen, R. Batiza, R. Haymon, A. Pollitz, F. F., EpisodicNorth America and Pacific plate motions, Tectonics, 7, 711-726, 1988. Barone,W. B. F. Ryan, T. Smith, T. Simkin, and M. A. Luckman, Geochemical and structural studies of the Lamont seamounts: Seamountsas indicatorsof mantle processes,Earth Planet. Sci. Lett., 89, 63-83, 1988a. Fornari, D. J., M. R. Perfit, J. F. Allan, and R. Batiza, Small-scale Rhodes,J. M., C. Morgan,andR. A. Liias, Geochemistry andaxial seamount lavas: Magmaticrelationship betweenthe Cobbhotspot andthe Juande Fucaridge,J. Geophys. Res.,this issue. Richards, M. A., Hotspots, plates,andplumes:Towardan integrated heterogeneitiesin depletedmantle sources:Near-ridge seamount theory, Eos Trans. AGU, 69, 1413, 1988. lava geochemistryand implicationsfor mid-oceanridge magmatic Riddihough, R., Recentmovements of theJuande FucaPlatesystem, processes,Nature, 331,511-513, 1988b. J. Geophys.Res., 89, 6980-6994, 1984. of Geist, D. J., W. M. White, and A. R. McBimey, Plume-asthenosphere Silver,E. A., R. von Huene,andJ. K. Crough,Tectonicsignificance mixing beneaththe Galapagosarchipelago,Nature, 333, 657-660, theKodiak-Bowie seamount chain,northeastern Pacfic,Geology, 2, 1988. Grdfiths, R. W., The differing effectsof compositional and thermal buoyancieson the evolutionof mantle diapirs,Phys. Earth Planet. Inter.,43, 261-273, 1986a. Griffiths, R. W., Thermalsin extremelyviscousfluids, includingthe effects of temperature-dependent viscosity, J. Fluid Mech., 166, 147-150, 1974. Smoot,N. C., Observations on Gulf of Alaskaseamount chainsby multi-beamsonar,Tectonophysics, 115, 235-246, 1985. Sun, S. -S., and R. W. Nesbitt, Chemicalheterogeneity of the Archaeanmantle, compositionof the Earth and mantle evolution, Earth Planet. Sci. Lett., 35, 429-448, 1977. 115-138, 1986b. Turner,D. L., R. D. Jarrard,andR. B. Forbes,Geochronology and origin of the Pratt-Welker Seamountchain, Gulf of Alaska: A new Hart, S. R., A large-scaleisotopeanomaly in the SouthernHemisphere mantle, Nature, 309, 753-757, 1984. pole of rotationfor the Pacificplate,J. Geophys.Res.,85, 65476556, 1980. Hart, S. R., Heterogeneous mantledomains:Signatures,genesis,and Wakeham,S. E., Petrochemicalpatternsin young pillow basalts mixing chronologies,Earth Planet. Sci. Lett., 90, 273-296, 1988. dredged fromtheJuandeFucaandGordaridges, M.S. thesis, 95 pp., Hegner,E., and M. Tatsumoto,Pb, St, and Nd isotopesin sea•nount Oreg. State Univ., Corvallis, 1978. basaltsfrom the Juande Fuca Ridge and Kodiak-BowieSeamount Watts,A. B., J. K. Weissel,R. A. Duncan,andR. L. Larson,Originof Chain, NortheastPacific, J. Geophys.Res.,94, 17,839-17,846, 1989. the Louisvilleridge and its relationshipto the Eltaninfracturezone system,J. Geophys.Res., 93, 3051-3077, 1988. Hegner,E., and M. Tatsumoto,Pb, Sr, and Nd isotopesin basaltsand sulfidesfrom the Juande Fucaridge,J. Geophys.Res.,92, 11,380- White, W. M., M. M. Cheatham,and R. A. Duncan, Isotope geochemistry of Leg 115 basaltsand inferenceson the historyof 11,386, 1987. the Reunionmantleplume,in Proceedings of the OceanDrilling Hooper,P. R., The role of magneticpolarityandchemicalanalysisin Program, ScientificResults, vol. 115, editedby R. A. Duncanet establishingthe stratigraphy,tectonicevolution,and petrogenesis al., CollegeStation,TX (OceanDrilling Program),in press. of the Columbia River Basalts,Mem. Geol. Soc. India, 3, 362-376, 1981. Karsten, J. L., Spatial and temporal variations in the petrology, morphology and tectonics of a migrating spreadingcenter: The Wilson, D. S., R. N. Hey, and C. Nishimura, Propogationas a mechanismof reorientation of the Juande FucaRidge,J. Geophys. Res., 89, 9215-9225, 1984. DESON-m AND DUNCAN:T•m COBB-EICKm-nERG SEAMOUNT CHAIN Wilson, J. T., A possibleorigin of the Hawaiian Islands, Can. J. Phys.,41, 863-870, 1963. Zindler, A., H. Staudigel,and R. Batiza, Isotope and trac• el•n•nt geochemistryof young Pacific seamounts: Implicationsfor the scaleof uppermantle heterogencity,Earth Planet. $ci. Lett.,70, 175-195, 1984. 12,711 D.LiDesonie and R.A.Duncan, College ofOceanography, Oregon State University,Corvallis, OR 97331. (ReceivedApril 10, 1989; revisedSeptember25, 1989; acceptedSeptember26, 1989.)