JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 95, NO. Bll, PAGES 17,555-17,569, OCTOBER 10, I990

Evidence for Age and Evolution of Comer Seamountsand Great Meteor SeamountChain From Multibeam B athymetry

BRL• E. TUCHOLKE

WoodsHole OceanographicInstitution, WoodsHole, Massachusetts

N. CHRtSTtAN SMOOT

U.S. Naval OceanographicOffice, StennisSpace Center, Mississippi

The Comer in the western North Atlantic and Great Meteor "chain" in the eastern North Atlantic are thoughtto progressin age from Late Cretaceousthrough late Cenozoic. They both presumablyformed by volcanismabove the New Englandhotspot when first the North Americanplate, and then the Mid-Atlantic Ridge axis and , moved over the . High-resolution,multibeam bathymetryof the seamountsshows geomorphic features such as guyors,terraces, and a baselevel plateau (Cruiserplateau) that we interpretto have formedat sealevel. We have backtrackedthese features to sea level alongthe North Arianticcrustal age-depth curve in orderto estimatetheir ages. The derivedage patternof volcanismindicates formation of the Comerseamounts at ca. 80 Ma to 76 Ma, withmigration of the Mid-AtlanticRidge plateboundary over the hotspotand formationof the Cruiserplateau about 76 Ma. Seamountages suggest that subsequentvolcanism on the Africanplate movedfirst northward,in the Late Cretaceousto early Cenozoic(Plato, Tyro, and Atlantisseamount groups), then southwardto GreatMeteor Seamountin the late Cenozoic.Recurrent volcanism appears to haveoccurred at someseamounts up to 20- 30 m.y. after their initial passageover the hotspot. It wouldthus appear that intralithosphericconduits can link the hotspotto old seamountsseveral hundred kilometers away.

INTRODUCTION It appearsto haveformed as the migrated We have known for more than 30 years that the floor of the west-northwestwardover the New Englandhotspot during the ArianticOcean exhibits numerous large seamounts[Heezen et Cretaceousperiod. Radiometricage dating of seamount aI., 1959](see Epp and Smoot [1988] for a recentsummary). samplesreported by Duncan [1984]suggests that a linear Most existing knowledge of these seamountsis based on migration rate of 47mm/yr best fits the seamountage scatteredecho sounding and seismiclines and, in a very few distributionand that the seamountsrange in age from about places,on detailed surveys. Sampling of the seamounts 103 Ma in the west()to about82 Ma at their likewiseis limited, but it has recovered both basaltic igneous southeasternterminus (Nashville Seamount). Although rocksand sedimentary samples from the sedimentarycaps that separatedfrom the New Englandseamounts, the Comer sometimesare present,particularly on flat-toppedguyors. seamountgroup farther east (Figure 1) alsois interpretedas Mostof the recoveredigneous rocks have been exposedto havingbeen formed above the New Englandhotspot. The Comer seamountshave beenthought to date to about75-70 seawaterfor long periods of time and are highly altered. Consequently,accurate radiometric age datingof the samples Ma, andthey would thus fall alongthe same age-distance trend is difficult, and the derived geochronometricages can be as the New Englandseamounts [Duncan, 1984]. Hotspot modelssuch as thoseof Duncan [1984] and of Morgan [1983] subjectto large uncertainties. Samplesof the sedimentary capsare useful in providingminimum ages for the seamounts, predictthat the westward migrating Mid-Atlantic Ridge axis overrodethe New Englandhotspot in Late Cretaceoustime; butit usuallyis difficult to estimatethe periodof time that thereafterthe hotspotvolcanism formed seamounts on the elapsedbetween emplacementof the igneousrock and accumulationof the sampledsedimentary record. Despitethe Africanplate. LateCretaceous to Recentabsolute motion of lackof detailed data on most seamounts,enough information theAfrican plate with respect to themanfie is thoughtto have been slow and in a generallynortherly and then easterly existsto suggestgeneral ages or agetrends; these data have direction [Morgan, 1983;Duncan, 1984]. The volcanic beenused to predictlarger-scale age relationshipsin the edificesof the Ariantis,Plato, Tyro, Cruiser,and GreatMeteor contextof specific models for the origin of seamounts, groupsof seamounts,which form the GreatMeteor "chain" particularlythose seamounts found in linearchains. A number (Figure1), presumablywere constructedabove the New ofmodels have been proposed for theorigin of NorthAtlantic Englandhotspot during this period. seamounts[Vogt and Tucholke,1979]; the mostsuccessful The above models imply that there should be an age models have been those that attribute formation of the progressiontoward younger volcanism from north to southon seamountchains to passageof the lithosphereover a manfie theAfrican plate, terminating in the vicinityof GreatMeteor plumeor hotspot[e.g., Morgan, 1972, 1981;Duncan, 1984]. Seamount.K-At ageson two volcanicsamples from Great The bestdeveloped and best known seamountchain in the Meteor have been reportedas 11 Ma and 16.3_+0.4Ma by NorthAtlantic is theNew Englandseamount chain (Figure 1). Wendtet al. [1976]. yon Rad [1974] deducedfrom studyof carbonatesands that the flat summit of Copyright1990 by the AmericanGeophysical Union. was in shallow water from late Miocene to early Pliocenetime Papernumber 90JB00863. (<11 Ma). Seamountsfarther north in the GreatMeteor 0148-0227/90/90JB.00863505.00 "chain"presently have no directage data, although an indirect

17,555 17,556 TUCHOLKEAND SMOOT: AGE AND EVOLUTION OFNORTH ATLANTIC SEAMOUNTS

40øN

30'

80øW 70' 60' 50ø 40' 30' 20' 10ow 20'

Fig. 1. Sketchmap showing locations of majorseamounts and volcanic islands in theNorth Atlantic and the positions of the Comerseamounts and eastern Atlantic groups of seamountswith respectto the HayesFracture Zone. Smallboxes show the areasdepicted by detailedbathymetry in Figures2-4.

age determination has been noted for Hyeres Seamount. positionsalong the Mid-Atlantic Ridge crest, based on the Fermont and Troelstra [1983] reportedlarger foraminiferaof platemotion studies of Klitgord and Schouten[1986]. Thetwo late Aquitanian/earlyBurdigalian age (ca. 23-19 Ma) in a seamountgroups lie in conjugatepositions along the Hayes turbiditc bed that was piston-cored 50 km west of the FractureZone. Althoughthe fracturezone consistedof a single seamount. It is likely that the foraminiferawere derived from fracture valley at the time that the Corner and Cruiser sedimentsdeposited in shallow water (<50 m) on Hyeres' seamounts were formed in Late Cretaceous time, it summit. subsequentlyevolved into a triplet of fracture valleys in Verhoef [1984] made indirect age estimatesof seamountsin responseto changesin relative plate motion (Figure 1; B.E. the Great Meteor "chain" based on the relations between age Tucholke and N.C. Smoot, manuscriptin preparation, 1990). and elasticthickness of the lithosphereas describedby Bodine In the ensuing discussion we briefly describe the et al. [1981]. His greatestestimated age was 65 Ma for the morphology of the Comer and Cruiser seamounts, and we Cruiser group of seamounts;estimated ages of most of the derive the apparent age (based on age-depth relations) of remaining seamounts,including Great Meteor, were in the seamountgeomorphic features that are thought to have formed rangeof 3847 Ma. Subsequently,Verhoef and Collette[1985, at sea level. From these data we examine the patternsof 1987] recognized that these ages probably are too large volcanismin the seamountgroups and considerthe questionof becausethey do not considereffects of lithospherethinning age-progressivevolcanism above the . and thermal rejuvenationat the time of seamountformation We concludethat the pattern of volcanism is consistentwith [Derrickand Crough,1978]. Verhoefand CoIlette[1985] also sequentialformation of the New England,Comer, and the Great used age-depthbacktracking to determinemaximum ages of Meteor "chain" seamountsabove the New England hotspot. several seamounts in the Great Meteor "chain", with derived However, Late Cretaceous and Cenozoic absolute motion of the agesranging from 37 to 23 Ma. There are largeuncertainties African plate over the hotspot differs significantly from in the ages derived in all these studies, so the presenceor predictionsof existing hotspot models. absenceof age-progressivevolcanism has been unresolved. SEAMOUNT MORPHOLOGY However, Verhoefand Collette [1987] preferredthe conceptof generally simultaneousvolcanism, possibly as a responseto Corner Seamounts changein plate stresspatterns, and they assumedan age of 22 Ma for the bulk of the seamountcomplex. The Cornerseamounts (Comer Rise) lie betweenabout 34øN It is possiblesignificantly to improveour understandingof to 37øN and 47ø00'W to 52ø30'W (Figure 1). The seamounts the origin and age of these seamount groups through are positionedmostly along the north side of the Hayes examination of their morphology, depths, and associations FractureZone, and in largepart they parallelthe fracturezone with one another in plate kinematic reconstructions. (Figure2). The main volcanicedifice in the eastcentral region Consequently,we have studieddetailed multibeam bathymetry is an east-westtrending ridge near 34ø45'N;it parallelsthe over the Comer, Plato, Tyro, Cruiser, and Great Meteor Late Cretaceousseafloor spreading direction and supportsa seamount groups (Figures 1-4), obtained through swath- seriesof four peaks (Figure 2). The two largest of these mappingsurveys of the U.S. Naval OceanographicOffice. seamounts(9 and 11) have roundedtops and they rise to less Aspects of the sonar data acquisitionsystem have been than 600 fm (1100 m) depth. Both seamountshave terraces discussedby Smoot [1986]. Navigationin the surveyswas developedin their flanksat depthsof about1000-1100 frn producedby merginginertial and satellitenavigation with (1875-2060 m). speedlogs to produceaccurate positioning. At the westernend of theridge is anotherwell-defined, linear We also examined plate kinematic reconstructionof the ridge that extends across the Hayes Fracture Zone at an Comer and Cruiser seamounts to their Late Cretaceous orientationof N25"W(Figure 2). Thisridge is morethan 230 TUCHOLKEAND SMOOT:AGE AND EVOLUTIONOF NORTH ATI_42qTIC SEAMOUNTS 17,557

360 N

35ø

/ 34 /34

**330

52øW 510 500 Fig. 2. Bathymetryof thewestern Comer seamounts from multibeam swath mapping. Depths are in uncorrectedfathoms with a 100-fm(-190-m) contourinterval. Magneticanomaly identifications for bothobserved and rotated anomalies (bold dotsand labels) are from Klitgord and Schouten[1986]. Seafloorisochrons (short-dashed lines) older than anomaly 34 (84 Ma) are interpolatedbetween anomalies 34 andM0; all anomalyages are taken from the DNAG time scale[Kent and Gradstein,1986]. Long-dashedlines are plate flowlines from Klitgord and Schouten (1986). The southerntwo flowlines approximatethe location of theHayes Fracture Zone and an unnamedfracture zone; the exact position of a fracturezone in thevicinity of the northernflow line is uncertain.Seamounts are numbered as in Table1. Stipplinghighlights terraces around seamounts4, 9, and 11. km long,reaching from 33ø50'Nto 35ø40'N. The crestsof reachesto 400 fm (750 m). There is little seismically peaksalong the southernpart of the ridge are at depths of detectable sediment cover on any of the Comer seamounts about1400 fin (2620 m). The northwestend of the ridge is [e.g., McGregor et al., 1973], so the depthscited representthe imbeddedwithin another group of sevenmajor seamounts and depthsof the volcanic crust. connectingridges. Theselatter features have WNW andNNE to The two linear volcanic ridges in the southeastpart of the NEtrends, generally parallel and orthogonal to thedirection of Comer seamountsform a morphologicbight that opens to the seafloorspreading at the time (Figure 2). Flat-topped southeastand has seafloor depths of 2400-2500 fm (4525- seamountcrests in this area are common, and they are at 840- 4720 m; Figure 2). Sedimentthickness within this bight is up 1160fm depth(1580-2180 m). Seamount4 has a terraceat to 400-500 m (McGregor eta!., 1973; Tucholke et al., 1982). similardepth (1000 fm) (1875 m), but its crestis roundedand We can approxirnatethe unloaded depth of oceanic basement 17,558 TUCI-IOLKEAND SMOOT: AGE AND EVOLO. trION OF NORTH ATLANTIC SEAMOUNTS

340

21

•900

Plato

m m m m

HAYESF.Z.

32 • Cruiser 3O I 32 Irving

34

mm m m m mm m m m-- - ß Hyeres - ß • 33y 0534 - : •t- .. contour interval: 100fm 30øW 290 280 27o

Fig. 3. Bathymetryof the Cruiserplateau, south of HayesFracture Zone, and adjacentseamounts as determinedby multibeamswath mapping. Depths are in uncorrectedfathoms with a 100-fm(~190-m) contour interval. Magneticanomaly identifications(bold dots and labels) are fromKlitgord and Schouten[1986]. Plateflow linesfrom Klitgordand Schouten (long dashes)approximate the locationof the HayesFracture Zone and a probablefracture zone just southof Irving Seamount. Seamounts are numbered as in Table 2.

by adding 0.6 times sediment thicknessto the water depth Althoughthere is somesediment cover to masktrends, the [Crough, 1983], giving a correctedbasement depth of between fabric of normal oceanic crust can be observed at the 4800 and 5000 m. Similar considerations indicate that southeasterncomer of the map in Figure 2. In this areaa unloaded basement depth around the seamounts to the steep,east facing scarp runs N10øE for morethan 70 kin. This northwest would be about 5300-5500 m. orientationis parallelto the LateCretaceous axis of theMid- TUCHOLKEAND SMOOT: AGE AND EVOLUTION OF NORTH A•C SEAMOUNTS 17,559

AtlanticRidge. At the southernedge of the contouredarea, two reported here include negligible correction for sediment shortridges and an interveningdeep trend orthogonalto the loading. scarp;we interpretthese features as defininga fracturezone Immediatelynorthwest of the Cruiser seamountsare the Plato (Figure2). seamounts,which as a grouptrend 5ø-8 ø north of west(Figures 1 and3). PlatoSeamount has a m'mimumdepth of 260 fm (490 EasternAtlantic Seamounts m) and its easternneighbors, seamounts 12 and 7, reach 360 The area studied in the eastern North Ariantic includes several fm (680 m) and<760 fm (<1430m), respectively.The eastern groupsof seamounts(Figures 1, 3, and4). We referto the endof the Platoseamounts is abuttedby a linearseries of three large,central group just southof HayesFracture Zone (Figure smaller seamountsthat together trend N35øW (Figure 3). 1)as the Cruiser seamount group. It includesthree major peaks Theseseamounts are the southernpart of the Tyro group,and as shown in Figure 3: Hyeres Seamount in the southwest they are colinear with the northeasternmargin of Cruiser (crestaldepth 160 fro, or 300 m); the large,flat-topped Irving plateau. The large seamountat 34øN (seamount11) and its Seamountin the north central area (140 fm; 265 m); and companionto the southeast(seamount 10) are not on the same CruiserSeamount in the northeast (390 fm, 735 m). All three trend as the smaller seamounts to the south. Seamount 11 and of theseseamounts, as well as several subsidiarypeaks, sit Tyro seamountat 35øN(off the mapin Figure3) defineanother atopan oceanic plateau (Cruiser plateau). The mean,unloaded linear trend of aboutN35'E, roughly 70ø to the trend in the depthof this plateauis difficult to determinebecause the southernpart of the Tyro group. plateaulocally has accumulated large thicknesses of sediment The Ariantis seamountslie just northwestof the map in betweenthe major seamounts (Figure 5) [Verhoef, 1984; Figure3. They consistof a groupof at least10 peaksranging Luyendyket al., 1979]. By applying an averagesound in depth from about 1500 m to less than 500 m [Verhoef, velocityof 2.0 krn/s to observedreflection time thicknesses 1984]. Their detailedgeomorphology was not exam/nedin the of sediment,and correctingfor crustal loading by sedimentsas present study. notedearlier, an averagebase level of the plateau of the order Great Meteor Seamount lies 150 km south of the Cruiser of 2400-2600 m is indicated. plateau and is a large, flat-topped with a minimum There are two other levels at which terraces and seamount crestaldepth of 145 fro, or 275 m [Ulrich, 1971] (Figure 4). crestsgenerally occur on Cruiserplateau. One level is at about Great Meteor is cappedby a sedimentarysection about 400 m 850 fm (1600 m). Both Irving and Cruiser seamountshave thick [Hinz, 1969]. It is flanked on the southwestby two markedbreaks in slope that, on average, appear to define smaller seamounts,the closer of which (Little Meteor) is also terracesin this depth range (Figure 5). Two small seamounts flat-toppedand has a m'mimumdepth less than 160 fm (300 (4 and5, Figure3) on the centralCruiser plateau have roughly m). These seamountsrepresent the southernmostvolcanic similarpeak depths. At a shallowerlevel, the crest of Irving peaks along the Great Meteor seamount"chain". Seamountis nearly flat, and this volcanic feature can be All the easternAtlantic seamountgroups discussedabove considereda guyot or tablemount. It and the rounded crest of occur on a broad basement swell on the flank of the Mid- HyeresSeamount mark anotherimportant bathymetric level at Atlantic Ridge. The averagedepth anomaly of this swell is 140-160fm (265-300 m). These terracesand seamountcrests about 1500 m in the immediate vicinity of the seamount are mostly unsedimented(Figure 5), so their depth values groups [Verhoef, 1984]. Outside the region of the swell,

Great16 +30027ø Meteor 3o. Guyot

330

34

contour Interval: 100 fm

Fig. 4. Bathymctryof the GreatMeteor seamountgroup from multibeamswath mapping. Depths are in uncorrcctedfathoms with a 100-fm (-190-m) contourinterval. Magnetic anomalyidentifications are basedon Klitgord and $chouten[1986]. Numbersidentify seamountsin Table 2. 17,560 TUCHOLKEAND SMOOT: Ao-n5 AND EVOLLrrION OF NORTH ATLANTIC SEAMOUNTS

3

HYERES

31¸

29øW 28ø 27ø Fig. 5. Tracingsof seismicreflection profiles (vertical exaggeration -7) acrossthe Cruiserplateau region, adapted from Verhoef [1984]. Light, straightlines showship tracks. Profilesare plottedwith respectto thesetracks, with an offset of 3333 ms (2500 m). Basement(bold lines) shallowerthan the 2500 m level is shaded;arrows provide reference between other tracksand their associatedprofiles. The averagedepth of the baselevel Cruiserplateau basement is about2500 m whencorrected for sedimentthickness and sedimentloading. Seamountson Cruiserplateau are identifiedby name and numberas in Table 2. Unshaftedarrows around Cruiser and Irving seamountsshow 850-fro level (1600 m) where,on average, terracing is observed.

approximately50-150 km to the east, the depth of oceanic underlyingcrust, and the age-depthrelation for AtlanticOcean basement corrected for sediment overburden is between 5000 crust,we can backtrackthe featuresfrom their presentdepths and 5500 m. This is the normal depth for oceaniccrust of to derivethe timeswhen they were at sea level, and thusthe middle to Late Cretaceousage. approximateages of their origin. The threegeomorphic features thought to have originatedat DETERMINATION OF SEAMOUNT AGES or nearsea level are (1) fiat-toppedguyots or tablemounts,(2) Among the geomorphic features of the seamounts,three terraces,and (3) the Cruiserbase level plateau. Bevellingof kinds appearto havebeen formednear sealevel eitherduring seamountcrests at wave base, often accompaniedby reef or immediatelyfollowing major volcanic events. Becausewe growth,is a well documentedphenomenon in the Pacific[e.g., know the present depth of these features, the age of the Hamilton, 1956; Schlanger et al., 1987] but is less well TUCHOLKEAND SMOOT: AGE AND EVOLUTIONOF NORTH ATLANTIC SEAMOUNTS 17,561

knownin the Atlantic. The bestexamples in the presentstudy according to tectonic interpretations of Tucholke and areGreat Meteor and Little Meteor seamounts(Figure 4), Mountain [1986], and their profiles extendedout only to 10- IrvingSeamount (Figure 3), andseamounts 1, 2, 5, and6 in the 15 Ma crust. On longer profiles discussedby Sclater et al. Comergroup (Figure 2). Someseamounts have more rounded [1975] for the samearea, the subsidenceconstant is very close tops(e.g., seamounts 9 and 11, Figure2); we inferthat these to 350. Furthermore,lithosphere that is thinnedand "reset"to werealso eroded at sea level, but they were not beveledflat or a new thermal age by passageover a hotspotappears to follow did not support levelling reef growth. Both the flat- and the Parsons and Sclater [1977] depth relation for the new rounded-topseamounts differ markedly in cross sectionfrom thermalage [Dotrick and Crough, 1978]. Basedon the above morepeaked seamounts which probably never were subaerially considerations,we concludethat use of the Tucholkeand Vogt exposed(e.g., seamounts 7, 13, and14, Figure2; seamounts8 [1979] age-depthrelation should yield reliable estimatesof and9, Figure3). agesof seamountfeatures which were developedat sealevel. The observed terraces are relatively level benches in flanks Seamount ages are constrained only by minimum and of seamounts(Figures 2, 3, and 5). They are presumedto have maximum values, derived in the following way. Minimum originatedby erosionat sea level in one of two ways: (1) ages are determinedby assumingthat the seamountfeature bevellingof a seamounttop to form a guyot, with subsequent subsided at the maximum allowable rate, that is as if the volcanismthat built the seamountto a higher level but did not surroundingcrust had been totally reset to zero thermal age; in entirely eradicate the existing guyot morphology, thus such instances we would expect to see concomitant leavingit manifestedas a terrace,or (2) terracecutting in the development of a thermal swell. Conversely, derived seamountflank during a period when the seamounteither was maximum ages are based on subsidence at the rate of onlyvery slowly subsidingor was beinguplifted, followedby surroundingnormal ocean crest; consequently,they assumeno rapidsubsidence and drowning. In either casethe terraces thermalreset and imply no developmentof a thermalswell. imply thermal and/or volcanic events that allowed the In making age determinationsfor the Comer seamounts seamountto be erodedat wave base. Terracesare developedat (Table 1), we assumedthat subsidencefollowed the age-depth seamounts4, 9, and 11 in the Comer group (Figure 2) and on relation for the underlyingcrust at each seamount,and that no the flanks of Irving and Cruiser seamounts in the eastern uplift or thermal age reset of the crustoccurred during volcanic Atlantic(Figures 3 and 5). episodes(Figure 6). This assumptionmay seem surprising, The Cruiser base level plateau around Cruiser, Irving, and but it appears to be reasonable because the adjacent oceanic Hyeresseamounts averages about 2500 rn (1330 fm) deep and crest is at the expecteddepth (correctedfor sedimentloading) coversan areaof approximately14,000 km 2 (Figures3 and 5). for its age. Any uplift or thermal reset would have causedthe Thereis little diagnostic bevelling to support an origin near crust to have a positive depth anomaly. We also assumethat sealevel, althoughwe assumethis to be the casebecause it is crustalloading by each volcanic edifice was synchronouswith sucha large and level volcanic plateau. Subsequentarguments seamount construction and that no significant load-induced will be made about the position of the plateau in plate subsidence occurred after construction. It is possible that reconstructionsand the resulting implications for probable limited thermal uplift in fact did occur during volcanic agesof volcanism;these argumentssupport but do not prove constructionand that this was roughly offset by load-induced our assumption. subsidenceto give the correct crustal depth for crustal age. Usingthe empirical age-depthcurve for crest More rapid subsidencewould accompany any such thermal derivedby Tucholke and Vogt [1979], we have backtracked reset, and it would have the effect of decreasingthe calculated sea-level-related features of the seamounts to determine their ages of volcanism. For this reason we consider the ages in apparentages. The empirical curve used is virtually identical Table 1 to be maximum ages,although they probablyare close to Parsonsand Sclater's [1977] relation of crustal depth with to true ages. For peaked seamountsin the Comer group, only age,d(t)=2500+350t •/2 (whered is depthin meters,and t is maximum ages can be given (parentheses,Table 1), since the timein m.y.), but it assumesan initial ridge-crestelevation of seamountscould have formed below wave base at any time after 2700 m. At the time that most of the seamounts being their supporting crust subsidedto depths greater than their consideredhere were formed, long-term level was about heights. 200-&50m higherthan at present[Haq et al., 1987]. Thus the Calculation of ages of volcanism in the eastern Atlantic practicaleffect of using the Tucholke and Vogt [1979] (Table 2) is more involved becausethe seamountsare located subsidencecurve (initial depth 2700 m, or 200 m below on a positive depth anomalyof about 1500 m [Verhoef, 1984] Parsonsand Sclater'sinitial depth) is that no correctionfor and because some seamounts appear to exhibit several long-termsea level changeis necessary,or justifiable. With volcanicphases. Figure 7, constructedfor Irving Seamounton respectto higher-frequencychanges in eustaticsea level, our the Cruiser plateau, illustrates our methodology. The Cruiser resolutionof seamountages is inadequateto allow assessment plateau (base level) is assumedto have formed near sea level. of their impact,and we thereforehave ignoredany effectsof It has a maximum average age of 78 Ma, which is the average morerapid sealevel variations. The actualerror that would be age of the underlying crust; this age is permissiblebecause the introducedby sealevel fluctuations of the orderof +50 m would anomalouslyshallow depth of the crust could have been caused beabout +5% in derivedages of seamountfeatures. by uplift and thermal reset during succeeding volcanic It couldbe questionedwhether the subsidenceconstant (350) episodes. The minimum age of the plateau is about 65 Ma, in the age-depthrelation is appropriateto use where the determined by assuming that the crust experienced a total lithospherehas been affected by hotspot volcanism. For thermal reset (and thus subsidedlike new lithosphere)and had example,Johansen et al. [1984] derived a constantof 2734-30 no subsequentthermal reset (dottedPMIN line, Figure 7). We for the Reykjanes Ridge near Iceland. However, in estimatean actualage of 76 Ma, basedon likely distributionof determiningthis value they assumed that the ridge axis had not volcanic patterns in the Late Cretaceous (see Plate subsidedin the past 15 m.y., a questionableassumption Reconstruction). We further assumethat the massivevolcanic 17,562 TUCHOLEEAND SMOOT:AGE AND EVOLUTION OF NORTH ATLANTIC SEAMOUNTS

TABLE 1. Comer Seamount CharacteristicsIn Western North Atlantic

Crustal Depth,* Depth,* Agefl' Peak+ Nature Age,Ma fm m Ma

1-Rockaway Fiat top 99 920 1730 82 2-CastleRock Fiat top 98 960 1805 82 3 Flat top 95 1160 2180 86 4 Roundedpeak 95 400 750 50 4 Terrace 95 1000 1875 81 5 Fiat top 93 840 1580 73 6 Fiat top 94 900 1690 77 7 Peak 90 1450 2720 (89) 8 Peak 88 1400 2620 (86) 9 Roundedpeak 86 600 1130 56 9 Tenace 86 1000 1875 74 10 Flat top 84 1150 2160 76 11-Caloosahatchee Roundedpeak 82 600 1130 53 11 Terrace 82 1050 1970 72 12 Peak 81 1200 2250 (76) 13 Peak 83 1400 2620 (82) 14 Peak 80 1400 2620 (79)

+ Numbersidentify volcanic edifices from Figure2. * Depthsin fathomsto nearest20 fm; correspp_.ndingdepths in metersrounded to nearest5 m. •' Only maximumages, in parentheses,are kidicatedfor peakedseamounts. episodewhich constructedthe plateau also createdmost of the observed850 m constructionof the terrace above the plateau observeddepth anomaly (Figure 7). (dot-dashed C850 m line, Figure 7). Note that for this A secondvolcanic or thermal event, during which the Irving maximum age to be applicable, the crust would later have to terraceformed, has a maximumage of 65 Ma; if the eventwere experience thermal reset or it could not reach its present any older, the Cruiser base level plateau could not have elevation (e.g., dot-dashed TMAX line, Figure 7). The subsided far enough below sea level to accommodatethe m'mimumage of the terrace is 30 Ma, assumingtotal thermal resetof the crust(dot-dashed TMIN line, Figure 7); this ageis unrealisticallyyoung, however, becauseuplift of the plateauto o its formersea-level position, which might be expectedin the case of total thermal reset, could not have occurred. The final event, associatedwith formation of the flat top on CREST:• 750m Irving guyot,has minimum and maximumages of 3 Ma and17 • Ma,respectively (GMIN and GMAX, Figure 7), controlledby the maximum and minimum possible subsidencerates. The minimum age again is unrealistically young since total TERRACE• 1875m thermal reset of the crust is unlikely to have occurred. Likewise,the maximumage may be unrealisticallyold, sinceit assumesno thermaleffects at all (i.e., subsidenceat the rateof • normalocean crust of that age). Within the acceptableage • rangesfor theirformation, the solidterrace and guyot lines in 3q Figure7 give an example ofpartial thermal uplift and crustal z_ age reset accompanyingeach phase of volcanicconsauction. x The specificages of the eventsimplied in this exampleare not 4• significant;the ages will shift dependingon how thermal effectsare apportionedbetween the two events. An independentestimate of the age of formationof Irving guyot can be derivedfrom the known age and depth of Great -5 Meteor Seamount(guyot) farther southin the Great Meteor ...... CORNER "chain". The unloadeddepth of basalticbasement at the crest NORMAL '"' SEAMOUNT 4 CRUST= of the guyotis about515 m (Table2). If we take 11 Ma asthe 555Om age of mostrecent volcanism on the seamount[Wendt et al., I I I ,I I I I I I 0 10 20 30 40 50 60 70 80 90956 1976], then the subsequentaverage subsidence rate was47 Ma m/m.y. GreatMeteor Seamountis on 86 Ma crust[Klitgord Fig. 6. Age-depth reconstructionfor seamount 4 in the Comer andSchouten, 1986]. Practicallyspeaking, this crust would be seamounts. The subsidence curve used is for 95 Ma crest and assumes subsidingat the same rate as the 78 Ma crust beneath the no thermal reset during constructionof the seamount(see text). Cruiserplateau, in, the absenceof any thermalperturbations. TUCHOLKEAND SMOOT: AGE AND EVOLUYION OF NORTH All. ANTICSEAMOUNTS 17,563

TABLE 2. SeamountCharacteristics In EasternNorth Atlantic

CrustalAge, Depth,* Depth,* Maximum Minimum ActualMinimum Peak+ Nature Ma fm m Age,Ma Age,Ma Age,• Ma

, 1-Hyeres Roundedpeak 76 160 300 19 3 6 2-Irving Flat top 78 140 265 17 3 6 3-Cruiser Peak 78 390 735 37 0 <16 2/3 Terrace -7 8 850-900 1600-1690 65 30 - 2/3 Basalplateau -78 1300-14005 2440-26205 -78 65 4 Peak 80 700 1320 57 0 - 5 Peak 78 900 1690 64 0 - 6 Peak 66 850 1600 53 0 - 7 Roundedpeak 61 7 60 1430 47 24 - 8 Peak 61 900 1690 52 0 - 9 Peak 59 800 1500 47 0 - 10 Peak 59 800 1500 47 0 - 11 Roundedpeak 57 440 83 0 31 11 18 12 Flat top 57 360 680 26 9 14 13-Plato Roundedpeak 53 260 490 20 6 10 14 Roundedpeak 52 880 1655 45 30 - 15 Flat top 51 1180 2220 51 51 - 16-GreatMeteor Flat top 86 2705 515• 33 6 11 17-LittleMeteor Flat top 83 160 300 21 3 6 18 Peak 83 7 00 1315 59 0 -

+Numbersidentify volcanicedifices from Figures3 and 4. *Depthsin fathtxnsto nearest20 fm unlessspecified more precisely from published literature; corresponding depths in metersrounded to nearest 5 m. tBasedon 47 m/m.y. linear subsidencerate derivedfrom Great Meteor Seamount. •:Basementdepth correctedfor sedimentloading.

However,the Great Meteor crustprobably experienceda major LATE CRETACEOUS PLATE RECONSTRUCTION thermalreset when it was formed, thus accounting for its AND P^TrERN OF VOLCANiSM present1500 m depth anomaly; this reset was youngerthan Late Cretaceous reconstruction of the Corner seamounts and that of Cruiser plateau crust, so its subsidencerate may be Cruiserplateau at anomaly33y (-74.5 Ma) is shownin Figure greaterthan that of Irving guyot. Thus, if we apply the 47 8, based on plate motion studiesof Klitgord and Schouten m/m.y.rate to Irving guyot,we obtainan actualminimum age [1986]. The northernmargin of Cruiser plateau,bounded by of 6 Ma for Irving volcanism (Table 2). The applied the Hayes FractureZone (FZ), closesto the southernedge of subsidence, of course, is a linear rate rather than the the east-westridge in the east central Comer seamounts. The exponentialrate that would be expectedfrom thermal decay; fracturezone inferredto be presentat the southernedge of the however, over the short time span involved in the Comer seamounts aligns with the large structural offset calculations,it provides a reasonable approximation that through the Cruiser plateau between Hyeres and Irving probablyis within about 10% of values that would be derived if seamounts. We presume that this offset marks the it werepossible to define an exponentialcurve. continuationof a fracturezone trace that now is mostlyburied Similaranalyses for the otherseamounts in the easternNorth below volcanicrocks of Cruiserplateau. Atlanticyield the agessummarized in Table 2. We pointout The illustratedposition of the spreading-ridgeaxis between thatages derived for the final phaseof volcanismat Irving and Hayes FZ and the fracture zone immediatelyto the south is Cruiser seamounts and for volcanic events at all other conjectural, and it is not symmetrically situated between seamountsare straightforwardcalculations that are not affected conjugateanomalies 33o (Figure 8). It is, however,midway by thecomplexities in the foregoingexample. However,both betweenthe steep,apparently fault-controlled western scarp of theminimum and maximum ages are consideredunrealistic in Cruiser plateau and the similarly steep-wailed southern each case, the former because total reset of the subsidence volcanic ridge of the Comer seamounts. More complete relationwould be requiredwithout associated uplift to former closure of the seamount groups, to about 76 Ma, would butt crustallevels, the latter because subsidence would be assumed thesefeatures against one another. We infer that a spreading to havebeen unaffected by thermaluplift, eventhough thermal ridge jump occurred~76 Ma, abandoninga spreadingcenter upliftis knownto havecaused a depthanomaly of- 1500m. that was formerly beneathCruiser plateau midway between The"actual minimum" ages (Table 2) arederived by applying anomalies330 (dashedlines, Figure 8). The ridgejump may theGreat Meteor subsidencerate, as noted above,to seamounts havebeen facilitatedby the fact that relative platemotion was youngerthan about 30 Ma. changingat this time [Klitgord and Schouten,1986] and by Asidefrom uncertaintiesin thermalhistory, all agesin the the presence of the New England hotspot beneath the aboveanalyses also are subjectto errorscaused by practical spreadingaxis, as discussedbelow. uncertaintiesin age of the underlying crust and in absolute There is a clear trendin apparentages of volcanismin the depthof seamount features on which the calculations were Comer seamounts,progressing from maximum agesof 86-81 based.Potential errors caused by theselatter factorsare Ma in the northwestto 76-72 Ma in the southeast(Table 1 and conservativelyestimated to be +3 m.y. Figure 2). We interpret this trend to be causedby north- 17,564 TUCHOLKEAND SMOOT: AGE AND EVOLUTION OFNORTH A•AN'rlC SEAMOUNTS

separatedCruiser plateau from •e Comerseamounts. Hotspot volcanismappears to have continuedon the North American platefor severalmillion years more, to agesat leastas young as72 Ma on thenorth side of theHayes transform (Table 1). Two unsurveyedseamounts at the northeasternend of the Comergroup (dashed contours, Figure 8) may havebeen includedin this late volcanic phase. If so, they imply eithera slightnorthward shift in locusof Late Cretaceousvolcanism, or an expansionof the hotspotfootprint to -150 km radius. The former seems more likely because, following initial constructionof the base level Cruiserplateau, there appearsto have been little significant volcanism on the African plate within the remainderof Late Cretaceoustime (Table 2).

CENOZOIC VOLCANISM Determiningpatterns of volcanismon the Africanplate is / complicatedby thelarge uncertainties in calculatedages (Table / 2). Nonetheless,there are consistenttrends. The oldest / Cenozoicvolcanism is interpretedto be that which formedthe DEPTH / ANOMALY• / Irvingand Cruiser terraces sometime between 65 and30 Ma 3900m (Figure9). As notedabove, northward migration of vol- canismmay have commenced during the La•e Cretaceous.If this occurred, and if the Irving/Cruiser seamountson the Africanplate and the unsurveyed eastern Comer seamounts on the North Americanplate were containedwithin a -100-km- radius volcanic footprint, then these seamounts cannot NORMAL IRVING CRUST:• SEAMOUNT 54.00m postdatethe earliestCenozoic (-64-65 Ma); with subsequent seafloorspreading the seamountlocations became separated well outsidethis foot'printsize. Subsequentvolcanism more I I I , i I I i 6 0 10 20 30 40 50 60 70 78 clearlymigrated northward, eventually forming the Tyro group Ma of seamounts,and part of the Plato groupof seamounts,at ages Fig. 7. Age-depthreconstruction for Irving Seamounton the Cruiser lessthan about50 Ma. The northwardmigration also accounts plateau. The solidline at the top of the graphgives an exampleof the for formation of the Ariantis seamounts, which must be age-depthhistory of the seamountcrest, including construction(C) youngerthan the crust on whichthey reside (40-31 Ma north and net uplift (U) at the time of each volcanicepisode; net uplift includesthe effectof crustalloading by the volcanicpile. The deeper, of OceanographerFracture Zone and 48-42 Ma southof the dashedlines trace the age-depthhistory of deeperlevels on the fracture zone; Figures 1 and 9). Subsequentsouthward seamount and of normal, 78 Ma ocean crust. Construction of Cruiser migrationcaused construction and rejuvenatedvolcanism in plateau(C), with totalthermal reset of crustand associated net uplift the Plato seamounts,as well as rejuvenated volcanismon (U), is thoughtto haveoccurred ~76 Ma (labelledP at top). Maximum age of construction(PMAX) is age of underlyingcrust (~78 Ma); Cruiser plateau (Irving, Hyeres, and probably Cruiser minimum age (PMIN) is 65 Ma, based on total thermal reset and seamount)sometime after about25 Ma. Furthersouthward allowingfor no subsequentuplift/thermal reset (dottedPMIN line). migrationformed Great Meteor Seamount by 16-11 Ma and Maximumage of constructionof shallowerterrace is 65 Ma (TMAX at constructedits southerlycompanions at probably younger top), limited by 850 m of construction(C850 m line) betweenthe ages. The agespread of recurrentvolcanism in the seamounts plateau and terrace levels and assuming no thermal reset; this maximumage would require later uplift/thermalreset to bring the on Cruiserplateau is at least60 m.y.,ranging from the ca. 76 terrace to its present level (see, e.g., dot-dashedTMAX line). M a basalplateau to 17 Ma or lesson Irving guyot. Minimum terraceage (TMIN at top) assumestotal thermalreset (dot- Recurrentvolcanism also appeared about 50-56 Ma in the dashedsubsidence curve). Maximumand minimumages of guyot Comer seamountsat peaks4, 9, and 11 (Table 1). These construction(GMAX, GM!N at top) assumeno thermalreset and total thermal reset, respectively(dotted subsidencecurves). See text for eventspostdate the originalvolcanism by some 18-31 rn.y. fuller discussion. The peakswould have been about800 km from the New Englandhotspot at the time, andon the oppositeside of the Mid-Atlantic Ridge axis. westwardmigration of theNorth American plate over the New Englandhotspot, with arrivalof the plateboundary over the DISCUSSION AND CONCLUSIONS hotspotabout 76 Ma. Thepart of theplate boundary that first Detailedmorphology of the Comer seamountsand the crossedthe manfie plume was the westernridge-transform seamountsin the Great Meteor "chain" shows that local trends intersectionof the HayesFZ (Figure8). Volcanismprobably of volcanic ridges and individual seamounts largely are occurred both north and south of Hayes FZ on the North controlledby the structuralfabric of the supportingoceanic Americanplate but only southof the fracturezone on the crust. Primarytrends in crustalstructure that are paralleland Africanplate. The Mid-AtlanticRidge crest presumably was perpendicularto plateflowlines, for example,are mirroredin uplifted,and volcanism at theridge axis formed the Cruiser volcanic edifices of both the Comer- and Plato-group baselevel plateau near sea level. The "volcanicfootprint" of seamountsnorth of HayesFracture Zone (Figures2 and3). thehotspot at thistime appears to havebeen about 100 km in However,there also are well developedoblique edifices that radius. The inferred ridge jump shortly following76 Ma typically strike 30' to 45' to plate flowlines. The best TUCHOL.KEAND SMOOT:AGE AND EVOLUTIONOF NORTH A'H_ANTIC SEAMOUNTS 17,565

50 km

Fig. 8. Platereconstruction of the Comerand Cruisergroups of seamountsto the time of magneticanomaly 33y (-74.5 Ma) basedon Klitgord and Schouten[1986]. Trianglesidentify picks of anomaly33y, and anomaly33o picksare shownby dots (observed)and squares(rotated). Seamountfeatures younger than 74.5 Ma are shownby lightly stippledcontours. The positionof the Hayes FractureZone is approximatedby a syntheticflow line of relative plate motion (bold line), and the Mid-Atlantic Ridge axis is indicatedby parallellines. Dashedparallel lines on Cruiserplateau mark the inferredposition of a defunct spreadingaxis, presumedto be abandoned-76 Ma (see text). Note the coincidenceof the structuraloffset in Cruiserplateau south of Irving Seamountwith the presumedfracture zone at the southernedge of Comer seamounts.Dashed contoursoutside the area of multibeambathymetry are 1000 and 2000 fm depthsderived from Canadian Hydrographic Service [1984]. examplesof thesepuzzling trends are the volcanicridge at and observed volcanism. However, Verhoef felt that the extension southof Hayes FZ in the Comer seamounts(Figure 2), the was localized and may have been related to the presenceof the southernTyro groupseamounts (Figure 3), and the seamount nearby Europe-Africa plate boundary along the - marginand volcanic ridges along the western side of Great Gibraltar line. The fact that we see the same phenomenon Meteorguyot (Figure 4). The samekinds of trendsalso appear around Great Meteor Seamount farther south, and around the in Hyeres,Irving, and Cruiser seamounts on Cruiserplateau. Comer seamountsin the western North Atlantic, suggeststhat The originof the obliquetrends is problematic.Verhoef the oblique trends are more widespreadand that large-scale [1984]suggested that the easternNorth Atlanticregion north lithospheric stresses(e.g., from plate motion change or of theCruiser plateau experienced NE-SW orientedtension as a lithospherecooling) may be responsible. On the other hand, result of horizontal contraction of the cooling lithosphere, we cannotdismiss the possibilitythat the obliquetrends are and tensional gashes could have provided sites for the uniqueto seamountsand that they are generatedby only local 17,566 TUCHOLKEAND SMOOT: AGE AND EVOLUI•ON OF NORTH ATLAN•C SEAMOUNTS

DISTANCE IN KM 300 200 100 0 100 200 300 40O 90

80

,- 70 ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::.,.:RU I S E R BASAL PLATEAU :,:,:.:.:,:.:,:.:.:.:,:,:,:,:,:,'.-,',-,-,-,',-.'.'.'.-,'.-.'.'.'.-,

I :"'.". 60 I :7;:; õ0

z_ 4.0 N

30 14 >' 2/3T

I 20

I ATL 10 N 11 12 I 13 2 i 16 17 o Fig. 9. Plot of easternNorth AQanticseamount ages (from Table2) versusdistance north and southof HayesFracture Zone. Vertical barsextend between age minima andmaxima for eachseamount, with "actualminimum" age indicatedby a black diamondalong the bar whereappropriate. Seamounts are numbered as in Table2; the Cruiserand Irving terracesare indicated by 2/3T. The Atlantisseamounts north and south of OceanographerFracture Zone are ATL N and ATL S; the upperpans of the bars, betweenhorizontal dashes, show the age range of oceaniccrust beneaththe seamounts. The only existing age constrainton theseseamounts is that they are youngerthan their supportingcrest. Other seamountsfor which we have calculatedonly maximumages •able 2) are not shownin this figure. The possibleage range and the latitudinalextent of Cruiserbasal plateau are indicatedby the dot-shadedbox. "Forbiddenzone" signifiesthat seamountscannot be older than the crest on which they reside. The broad dot-shadedbar showsour interpretationof first northward,then southward migration of volcanism. Northward and southwardcomponents of volcanic migration rates are 10 and 56 mm/yr, respectively,but theserams are not well constrainedby existingdata on seamountages. Only the age datumfor seamount 15 in the Plato groupis unexplainedby the proposedpattern of volcanicmigration (although it couldfit wkhin the kind of 200 km volcanicfootprint noted in text); the seamountappears to have formedat the Mid-Atlantic Ridge axis at 51 Ma. lithosphere response to passage over the hotspot. We the end of the Cretaceouslong normal-polarity interval [e.g., unfortunatelyhave little information about the distributionof Klitgord and Schouten,1986; Tucholke and Schouten,1988]. oblique features in the adjacentbasins. Until the necessary Application of the 112 mm/yr rate also places the Mid- detailed surveys are conductedin these basins,we can only AtlanticRidge plate boundaryover the hotspotat -•76 Ma. speculateabout whether the oblique trends have plate-wide The Late Cretaceousplate boundaryintersected the New significance. Englandhotspot at the westernend of the -50-km offset which On the seamounts themselves, we have interpreted thencomprised the Hayes transformfault. The principallocus geomorphicfeatures such as terracesand guyorsto have formed of volcanism appears to have been on the ridge axis at sea level, concurrent with major episodes of volcanic immediatelysouth of the fracture zone (Figure 8). In that construction. By tracking these features from their present location extensivevolcanism and uplift of the Mid-Atlantic depthback along the North Atlantic crustalage-depth curve, Ridgeaxis is interpreted tohave formed the 14,000 krn 2 basal we have calculatedthe approximateages of theseevents. The Cruiserplateau near sea level. Shortlythereafter, a westward ages of initial volcanism at the Comer seamounts are jump of the spreadingridge axis just south of Hayes FZ generally consistentwith models of plate probablymoved the plate boundarywest of the hotspotcenter motion acrossthe New Englandhotspot (Figure 10) [Duncan, (thus increasingthe transformoffset to -120 km) and caused 1984]. The rate of migration of volcanism from Nashville final separationof the Comer andCruiser seamount groups. Seamount(southeast end of New Englandseamounts) through It is interestingto note that there was very little volcanic the Comer Seamounts,however, must have been greaterthan constructionwithin the Hayes FZ. An exception is the the averagerate of 47 mrnJyrthat Duncan [1984] determined southernarm of the Comer seamounts,which probablywas within the New Englandseamounts. A rate of 112 mm/yrbest constructedacross the fracturevalley at or slightlywest of the fits the age data of both the easternNew Englandseamounts westernridge-transform intersection (Figure 8). In contrast, and the Comer seamounts(Figure 10). This interpretation the lithosphereof the adjacentNorth Americanplate supported implies that a significant rate changeoccurred -85 Ma. The developmentof the major east-westvolcanic ridge just to the timingof sucha rate changeis consistentwith a largechange north of the fracture zone. This observation is somewhat in relative plate motion (and necessarilyin absoluteplate counterintuitivebecause the NorthAmerican lithosphere would motion) that most likely occurred in the North Atlantic near have been colder and thicker than that within the fracture zone TUCHOLKEAND SMOOT: AGE AND EVOLUTIONOF NORTH ATt.AwrIC SEAMOUNTS 17,$67

130 WMIP

120

110

•100 :::::::::::::::::::::: ß MAR All z

• 90 NEW ß ENGLAND • A M SMTS. 80-

70- ß CORNER SMTS.

60-

50 1500 1000 500 0 500 1000 DISTANCE IN KM Fig. 10. Plotof seamountages versus distance for theNew Englandand Comer seamounts in the westem North Atlantic. Soliddots are 4øAr-39Ar ages of New EnglandSeamounts from Duncan [1984]. Seamountsare B, Bear,A.if, Ariantis H; G, Goshold;A, Allegheny;M, Michael;and N, Nashville. WMIP is the youngestseries in the White MountainIgneous Province.Duncan's [1984] best fit linearmigration rate of 47mm/yrfor the New EnglandSeamounts is shownby a solid line. The dashedline indicatesa 112mm/yrmigration rate to theComer Seamounts, intersecting the Mid-Afiantic Ridge plateboundary about 76 Ma (arrow,see text). Agesshown for the Comer seamounts are from Table 1, with3-m.y. error bars as notedin text. All distancesare measuredfrom NashvilleSeamount, except that distanceswithin Comer Seamountsare measuredalong plate flow lines beginningat the westernmostseamount.

[e.g.,Fox and Gallo, 1986], and the thickerlithosphere could the GreatMeteor seamountgroup probably occurred after about be expectedto be lessreadily penetratedby mantlemelts. A 25 Ma and at a rate of 50-60 mm/yr (southerly component). possibleexplanation is that becausethe fracturevalley is These observationsconflict with existing models of African highly faulted, its crust and upper manfie were penetrated, plate motionin the hotspotreference frame [Morgan, 1983; cooled, and altered by seawater, effectively insulating the Duncan, 1984], which suggestonly southwardmigration of valley from significant magmatism. Indeed, such invasive volcanism over the New England hotspot (Figure 11). The alterationof the upper lithospherehas been used to explain existingmodels also have the shortcomingthat at 70-80 Ma anomalouslylow seismic velocities in transformvalleys, as they transfer the hotspot from the North American to the well as the occurrenceof median ridges that presumablywere Africanplate at the latitudeof the Platoand Cruiser seamounts; createdby serpentfinitediapirism [Fox and Gallo, 1989]. thus they cannot explain either the position or the Another factor that may have hindered fracture valley unquestionableyouth (<48 Ma) of the Ariantisseamounts. We volcanismwas plate-motion change. Counterclockwise concludethat the existing hotspotmodels requirerevision for changesin relativeplate motionwere occun4_ngat the time the motionof the Africanplate at agesyounger than about 80 Ma. Hayestransform valley overrodethe hotspot [Tucholke and As to whetherthe northwardand then southwardmigration of Schouten,1988]; this would have subjected the right-lateral volcanismin Figure9 indeedreflects "absolute" motion of the offsetof the transformto compression,perhaps limiting the African plate, the answeris not clear. There are two other potential for intratransform volcanism. possibilities:(1) while the African plate was "stalled"over Followingthe initial, -76 Ma volcanismthat constructedthe the New Englandhotspot, the hotspotwas able to expandits Cruiserbase level plateau, there appearsto have been no sphereof influenceto a radiusof more than 300 kilometers significantvolcanism on the African plate until the Irving and (i.e. Ariantis,Plato, and Tyro seamountgroups), or (2) that the Cruiserterraces were formed<65 Ma (Figure 9). The fact that New Englandhotspot migrated independently with respectto thisvolcanism recurred on Cruiserplateau near Hayes FZ otherhotspots beneath the Africanplate. The first alternative suggestsrelative stability or slow southwardmotion of the does not seem reasonable. Although Schilling [1986] has Africanplate with respectto thehotspot for at least10 m.y. A suggestedthat hotspotinfluence accounts for geochemical slow, northward shift of volcanism also could be responsible anomaliesup to 700 km from actual plume location,this for formationof the two easternmostCorner seamounts north relationappears to applyonly along"hotspot flow lines"that of HayesFZ on the NorthAmerican plate (Figure 8). connecta hotspot to a spreading-ridgecrest. Such would not Followingthe Irving-Cruiser terracing event, the seamounthave been the case for the northerly positioned Atlantis and agedata clearly indicate northward migration ofvolcanism to Tyro seamounts. In the present instance it is also difficult to theAriantis seamounts at a rate (northerlycomponent) of explainhow a hotspotthat was centered roughly beneath the about10mm/yr (Figure 9). Subsequentsouthward migration to Cruiserplateau would have engendered extensive volcanism 17,568 TUCHOLKEAND SMOOT: AGE AND EVOLIYrlON OF NORTH ATLANTIC SEAMOUNTS

., i I I i I I i volcanicphases at IrvingSeamount and by the>6 m.y.age differencebetween seamounts 14 and15 in thePlato group. ATLANTIS - 350 This,of course,is no surprise,since hotspot traces commonly N are manifestedby chainsof individualvolcanoes rather than by continuousvolcanic ridges, and episodic volcanism

TYRO .. spanningup to 20 m.y. has beendocumented elsewhere in the Ariantic [e.g., Baker, 1973; Vogt and Tucholke, 1979]. One

PLATO factor we cannotevaluate with presentlyavailable data,

ES F however,is howcontinuous or discontinuouslower-intensity volcanismmay have been at anygiven location. For example, it is possiblethat minor volcanic activity occurredmore or -- less continuouslyat Cruiser plateau between about 76 and 55 Ma; it could be manifested in such subsidiary volcanoesas CRUISER seamounts4 and 5 (Table 2 and Figure 3), or in other basement irregularitiesoccurring roughly at plateau "baselevel". '- e70 GREAT '1 A second implication of observed recurrence is that METEOR -- 300 significant,edifice-building volcanism can occur long aftera

\1. region has passedover a hotspot. The best examplesin the A•IS, ,FZ. present study are seamounts4, 9, and 11 in the Comer 10Ma . seamountgroup (Figures 2 and 10) where late stagevolcanism I .•:b55 Ma postdatedpassage of the hotspotby 18-31 m.y. and was some 800 km distant from it. Within the age uncertaintyof events :'e60 I in the Great Meteor "chain", it is also possible that the final ..- . volcanicphase at Hyeres, Irving, and Cruiser seamountswas occurringsimultaneously with the-16-11 Ma constructionof I the Great Meteor seamounts,which presumably were abovethe O50Ma New Englandhotspot at the time. These edifices are 200-300 / km apart. These examplessuggest that lithospherepassing •.040 over a hotspot is conditioned so that plume flow can reach -- 25ø large distancesback along the hotspot track to engenderlater •_.---.-- ©-•'- 20 0 10 volcanic episodes. I i i A phenomenonslightly different in mechanism,but similar 250 in scale, has been suggestedby Morgan [1978] to explain geochemicalanomalies on spreadingridges that once passed Fig. 11. Schematizedmigration of volcanismon the African plate, beginningon the central Cruiser plateau,as suggestedby this study over hotspots. In this model a sublithospheric channel (bold arrows; see Table 2). The track of the New Englandhotspot on develops and extends to maintain continuity between the the African plate as calculatedfrom the model of Morgan [1983] is hotspotand the ridge crest as the two migrate away from one shown by a dashedline, and the dotted line shows the track based on Duncan's [1984] model. For ease of comparison both are also another. For the South Ariantic, ridge-crest geochemical constrainedhere to pass through central Cruiser plateau at 76 Ma. anomaliesand hotspotsas far as 700 km apart have been cited Irrespectiveof startingand endpointsof the hotspoton the African by Schillinget al. [1985] in supportof this model. Schilling plate, however, neither model satisfactorily approximates the [1985, 1986] suggestedthat with increasingdistance between observeddistribution of volcanismin spaceand time. hotspotand ridge crest, both geochemicaland ridge elevation anomalies decrease,the latter disappearing first, until the hotspot manifestationbecomes totally intraplate. Duncan 200-300 km to the north but apparently did not cause [1984] and Schilling [1986] also noted that geochemical concurrentvolcanism on the Cruiserplateau itself or within a anomaliesoccur in basaltsalong the "flow line" of the New similar radius to the south(Figure 9). England hotspotto the Mid-Atlantic Ridge crest at 35øN The second alternative is more difficult to evaluate. latitude;the presentseparation of ridge crest and hotspotis Northward migration of volcanismcould have been as slow as more than 1000 km. In the context of these indications of 6-8mm/• (Figure 9). This is closeto the ,--5mm/yr limit that long-distancesublithospheric and intralithosphericplume normally is consideredto define "fixed" hotspots[Morgan, flow, it wouldbe worthwhileto samplesuccessive episodes of 1981], and it is well within the 10-20 mm/yr rate at which a volcanism in the Comer Seamounts and Great Meteor number of widely separatedhotspots have been calculated to seamount"chain" and to comparethem geochemicallywith move with respectto the Hawaiian hotspot[Molnar and Stock, ridge-crestand ridge-flankbasalts along the flow line from 1987]. Thus it is conceivablethat the northwardmigration hotspotto ridge crest. representsindependent motion of the New England hotspot. In conclusion,we have investigated a little used, Evaluationof this possibility,as well as improvedmodelling geomorphologicmethod of estimating ages of seamount of post-80 Ma African plate motion, will require use of both volcanismin the North Atlantic. The techniqueclearly lacks better age data and improved morphologic analysis for precision,but just as obviouslycan providesome important volcanic edifices along other Ariantic hotspottraces. constraintson patterns of volcanism and migration of The local recurrenceof volcanismsuggested by the present hotspots.There is a needto determineages along a large study has two important implications. First, it is apparent numberof hotspottraces in order to examinequestions of that volcanism above the hotspot is episodic, illustrated, for hotspotfixity and to optimize models of "absolute"plate example, by the hiatus between the plateau and terrace motion. Interpretationof detailedseamount geomorphology TUCHOLKEgaND SMOOT: AGE AND EVOLUTION OF NORTH ATLANIIC SEAMOUNTS 17,569

is a presentlyunderutilized resource that can contribute Morgan,W.Y., Hotspottracks and the openingof the Atlantic and significantlyto this process. Indian oceans,in The Sea, vol. 7, editedby C. Emiliani,pp. 443- 488, John Wiley, New York, 198!. Acknowledgments. We thank H. Schoutenfor providing the Morgan, W.J., Hotspottracks and the early rifting of the Atlantic, calculatedhotspot traces in Figure 11, and P. Fosterfor typing the Tectonophysics,94, 123-139, 1983. manuscript.Reviews and commentson this manuscriptby J. Stock,J. Parsons,B., and J.G. Sclater,An analysisof the variationof ocean Verhoef,and P. Vogt are appreciated.This work wassupported in part floor bathymetryand heat flow with age, J. Geophys.Res., 82, byONR Contract N00014-82-C-0019 to B. Tucholkeat WoodsHole 803-827, 1977. OceanographicInstitution. ContributionNo. 7377 of WoodsHole Schilling,J.-G., Upper-mantleheterogeneities and dynamics,Nature, OceanographicInstitution. 314, 62-67, 1985. Schilling,J.-G., Geochemicaland isotopicvariation along the Mid- AtlanticRidge axisfrom 79'N to 0'N, in Geologyof North America, REFERENCES vol. M, The WesternNorth Atlantic Region,edited by P.R. Vogt Baker,P.E., Islandsof the South Atlantic, in The Ocean Basinsand and B.E. Tucholke,pp. 137-156, GeologicalSociety of America, Margins,vol. 1, The SouthAtlantic, edited by A.E.M. Naim and Boulder, Colo., 1986. F.G. Stehli, pp. 493-553, Plenum, New York, 1973. Schilling, J.-G., G. Thompson, R. Kingsley, and S. Humphris, Bodine,J.H., M.S. Steclder,and A.B. Watts, Observationsof flexure Hotspot-migrating ridge interaction in the South Atlantic: andthe theologyof the oceaniclithosphere, J. Geophys.Res., 86, Geochemical evidence,Nature, 323, 187-191, 1985. 3695-3707, 1981. Schlanger, S.O., J.F. Campbell, and M.W. Jackson, Post-Eocene CanadianHydrographic Service, GEBCO, Generalbathymetdc chart of subsidenceof the Marshall Islands recordedby drowned atolls on the oceans,5th ed., scale 1:10 million, Ottawa, 1984. Hattie and Sylvania guyors, in Seamounts,Islands, and Atolls, Crough,S.T., The correctionfor sedimentloading on the seafloor,J. Geophys.Monogr. Ser., vol. 43, edited by B.H. Kenting,P. Fryer, Geophys.Res., 88, 1236-1244, 1983. R. Batiza, and G.W. Boehlert, pp. 165-174, AGU, Washington, Detrick,R.S., and S.T. Crough, Island subsidence,hot spots,and D.C., 1987. lithosphericthinning, J. Geophys.Res., 83, 1236-1244, 1978. Sclater, J., L.A. Lawyer, and B. Parsons, Comparison of long- Duncan,R.A., Age progressive volcanism in the New England wavelengthresidual elevation and free air gravity anomaliesin the Seamountsand the opening of the central Atlantic Ocean, J. North Atlantic and possibleimplications for the thicknessof the Geophys.Res., 89, 9980-9990, 1984. lithosphericplate, J. Geophys.Res., 80, 1031-1052, 1975. Epp,D., and N.C. Smoot,Distribution of seamountsin the North Smoot,N.C., Seamountsby SASS - Chainsthrough forearc seamounts, Atlantic, Nature, 337, 254-257, 1989. in ProceedingsMarine Data SystemsInternational Symposium'86, Fermont,W.JJ., and S.R. Troelstra,Early Miocene larger foraminifera pp. 470-479, Marine Technology Society, Gulf Coast Section, from the Cruiser-Hyeres Seamount complex (eastern North NSTL, Miss., 1986. Atlantic),Proc. K. Ned. Akad. Wet., Ser. B, 86, 243-253, 1983. Tueholke,B.E., and G.S. Mountain, Tertiary paleoceanographyof the Fox, P.J., and D.G. Gallo, The geology of North Atlantic transform westernNorth Ariantic Ocean, in The Geology of North America, plateboundaries and their aseismicextensions, in The Geologyof vol. M, The Western North Atlantic Region, edited by P.R. Vogt NorthAmerica, vol. M, The WesternNorth Atlantic Region,edited and B.E. Tucholke, pp. 631-650, Geological Society of America, by P. R. Vogt and B. E. Tucholke,pp. 157-172, GeologicalSociety Boulder, Colo., 1986. of America, Boulder, Colo., 1986. Tucholke,B.E., and H. Sehouten,Kane FractureZone, Mar. Geophys. Fox, P.J., and D.G. Gallo, Transforms of the eastern central Pacific, in Res., 10, 1-39, 1988. The Geologyof North America, vol. N, The Eastern Pacific Ocean Tucholke, B.E., and P.R. Vogt, Westem North Ariantic: Sedimentary and Hawaii, edited by E. L. Winrefer, D. M. Hussongand R. W. evolution and aspectsof tectonic history, Initial Rep. Deep Sea Decker, pp. 111-124, Geological Society of America, Boulder, Drill. Proj., 43, 791-825, 1979. Colo., 1989. Tucholke, B.E., R.E. Houtz, and WJ. Ludwig, Sedimentthickness and Hamilton, E.L., Sunken islands of the Mid-Pacific Mountains, Mere. depth to basementin western North Ariantic Ocean basin, AAPG Geol. Soc. Am., 64, 97 pp., 1956. Bull., 66, 1384-1395, 1982. Haq,B.U., J. Hardenbol,and P.R. Vail, Chronologyof fluctuatingsea Ulrich, J., On the topographyand morphology of the Great Meteor levels since the Triassic, Science, 235, 1156-1167, 1987. Seamount,Meteor Forschungsergeb.,Reihe C, 6, 48-68, 1971. Heezen,B.C., M. Tharp, and M. Ewing, The floors of the oceans,I., Verhoef, J., A geophysicalstudy of the Ariantis-Meteor seamount The North Ariantic,Spec. Pap. Geol. Soc.Am., 65, 122 pp., 1959. complex, Geologica Ultraiectina, Meated.Inst. Aardw. Rij•univ. Hinz, K., The Great Meteor seamount. Results of seismic reflection Utrecht 38, 151 pp., 1984. measurementswith a pneumaticsound source and their geological Verhoef, J., and B.J. Collette, A geophysical investigation of the interpretation,Meteor Forschungsergeb.,Reihe C, 2, 63-77, 1969. Atlantis-Meteor seamount complex, Proc. K. Ned. Akad. Wet., Johansen,B., P.R. Vogt, and O. Eldholm,Reykjanes Ridge: Further Series B, 88, 427-479, 1985. analysisof crustal subsidenceand time-transgressivebasement Verhoef, J., and B.I. Collette, Lithospheric thinning under the topography,Earth Planet. Sci. Lett., 68, 249-258, 1984. Atlantis-Meteor seamountcomplex (North Atlantic), in Seamounts, Kent,D.V., and F.M. Gradstein,A Jurassicto Recentchronology, in Islands, and Atolls, Geophys. Monogr. Ser., vol 43, edited by B. TheGeology of North America,vol. M, The WesternNorth Atlantic H. Keating,P. Fryer, R. Batiza and G. Boehlert,pp. 391-405, AGU, Region, edited by P.R. Vogt and B.E. Tucholke, pp. 45-50, Washington, D.C., 1987. GeologicalSociety of America,Boulder, Colo., 1986. Vogt, P.R., and B.E. Tucholke,The New EnglandSeamounts: Testing Klitgord,K.D., and H. Schouten, Plate kinematics of the central origins,Initial Rep. Deep Sea Drill. Proj., 43, 847-856, 1979. Ariantic,in The Geologyof North America,vol. M, The Western von Rad, U., Great Meteor and Josephineseamounts (eastern North NorthAtlantic Region, edited by P.R. Vogt andB.E. Tucholke,pp. Atlantic): Compositionand origin of bioclastic sands,carbonate 351-378,Geological Society of America,Boulder, Colo., 1986. and pyroelasticrocks, Meteor Forschungsergeb.,Reihe C, 19, 1- Luyendyk,B.P., et el., Site414, Initial Rep.Deep Sea Drill. Proj.,49, 61, 1974. 407-411, 1979. Wendt, I., H. Freuzer,D. Muller, U. yon Red, and H. Rasehka,K-At age McGregor,B.A., P.R. Betzer, and D.C. Krause, Sedimentsin the of from Great Meteor and Josephine seamounts(eastern AtlanticComer Seamounts:Control by topography,palco-winds North Atlantic), Deep-Sea Res., 23, 849-862, 1976. andgeochemically-detected modem bottom currents,Mar. Geol., 14, 179-190, 1973. B.E. Tucholke, Woods Hole OceanographicInstitution, Woods Molnar,P., andJ. Stock,Relative motions of hotspotsin the Pacific, Hole, MA 02543. Ariantic,and Indian oceanssince Late Cretaceoustime, Nature, 327, N.C. Smoot, U.S. Naval OceanographicOffice, Stennis Space 587-591, 1987. Center, MS 39522. Morgan,W.J., Plate motion and deep manfie convection, Mere. Geol. Soc.Am.,, 132, 7-22, 1972. (Received November 15, 1989; Morgan,W.J., Rodriguez,Darwin, Amsterdam, a secondtype of revised April 12, 1990; hotspotisland, J. Geophys.Res., 83, 5355-5360, 1978. acceptedFebruary 11, 1990.)