TECTONICS, VOL. 5, NO. 7, PAGES 1145-1160, DECEMBER1986 TECTONICALLY CONTROLLED ORIGIN OF THREE UNUSUAL ROCK SUITES IN THE WOODLARK BASIN Dallas Abbott and Martin Fisk Collegeof Oceanography, OregonStateUniversity, Corvallis,Oregon Abstract. We proposealternativemechanisms for theoriginof threeunusualrocksuites,high-Mg andesites,NaTi basalts,and arclike rocks,that have beendredgedfrom the Woodlark basin,southwest PacificOcean. We showthatthehigh-Mgandesites andNaTi basaltsareassociated with anunusually coolridgeenvironment.The coolingis dueto increased hydrothermal circulation, stimulated by an unusuallyhighcrustalpermeability.The high permeabilityis mainlydueto crackingof the Woodlarkbasinlithosphere asit passes overthe flexuralbulgein front of the subductionzone, althoughlocalprocesses, suchasfaultingin fracture zones,may also make somecontribution.This increased convectivecoolingaffectsmagma dynamicsandchemistryat theridgecrest.The high-Mgandesites occurwherethehydrothermal circulation lowersthetemperatures in theupper oceaniccrust,causingthemagmachamberto sitin themantleratherthanin thecrustandpromoting the interaction of basaltmagmawith harzburgite. The NaTi basalts are also the indirect result of increased crustalcooling,whichcausesanomalously low degrees of partialmeltingof theirdepletedmantle source.The arclikerocksarecausedby interaction withvolatilesfromlithosphere thatwasemplaced beneath theedgeof thebasinlessthan6 m.y. agoby a now inactive subduction zone to the north of the currentlyactiveone.Simplethermalmodelsindicate Copyright1986 by the AmericanGeophysicalUnion. Papernumber6T0474. 0278-7407/86/006T-0474510.00 thatthisformerlysubducting plateis still cold enoughto retainvolatilesandto remainseismically active. As the old platelosesvolatiles,theyriseinto the mantle convection cell which feeds the Woodlark basinridge crest. Becauseridge subduction increases the likelihoodof ophioliteobduction,these observations andexplanations of Woodlarkbasin tectonicsarepotentiallyimportantfor ophiolites. INTRODUCTION The Woodlark Basin, locatedin the southwest PacificOceaneastof PapuaNew Guinea,contains an activelyspreadingridge,thatis beingsubducted beneaththe New Georgiaislandarc to the northeast (Figures1 and 2). A secondlithosphericslab,a relic from a previousperiodof southward subduction of thePacificplate,is emplacedbeneath the island arc from the north. Althoughoceaniccrustfurtherthan 150km from the subductionzoneis normalmid-oceanridge basalt (MORB), the other rocksfrom the Woodlark basin spreading centerbelongto threedifferentsuites:(1) arclikerocks,basaltsthroughdaciteswith a high contentof arclikecomponent;(2) high-Mg andesites, andesites with a highmagnesiumcontent,and(3) NaTi basalts,basaltswith a high sodiumand titaniumcontent[Perfitet al., 1986]. The high-Mg andesites and NaTi basalts formed at transform zones within 50 km of the trench. The arclike rocks were generatedat ridgecrestswithin 150 km of the trench[Taylor and Exon, 1986]. The threespatiallyrestrictedsuitesareall unusual in theirtectoniclocation.High-Mg andesites have beenpreviouslyfoundin islandarc andforearc regions[Johnsonet al., 1983; ReaganandMeijer, AbbottandFisk:TectonicallyControlledRockSuites •55 156 157 I KEY & _ DREDGE •TRENCH • LOCATION AXIS RIDGE CREST ol•':•'J _< 2. M.Y.OLD [7"/'/"/•> 2 M.Y. OLD I0 155 156 157 158 ø Fig. 1. The Woodlarkbasin:an areawherelithospheric flexuremaybe affectinghydrothermal and volcanicactivity(mapafterTaylorandExon [ 1986]). The spreading centermarkedby double dashedlines(Ghizoridge)is eruptingislandarc-typetholerites,dacites,andandesites andis within the flexuralbulge. Trianglesaredredgehaullocations.Dredges31 and32 areon the Simbofracture zoneandtransformfault.Dredge31 recoveredhigh-Mgandesites andarclikerocks. Dredge32 recoveredNaTi basaltsandarclikerocks. Dredge26 recoverednormalMORBs [afterTaylor, 1986]. 1984], but have never before been found on a transform fault. NaTi basalts have been found at five other sites,four of which are far from subducti6nzones[CAYTROUGH, 1979;Schilling The factorswhichdistinguish theWoodlarkbasin from othermarginalbasinsarethe subduction of very youngoceaniccrest(0-5 m.y.), a rapid convergence rate(> 10 cm/yr), anda polarityreversal et al., 1983; Cousenset al., 1984; W.G.Melson and of subduction(late Miocene) [Colemanand T. O'Heam, unpublishedcompilationof glass analysesof MORB, 1985]. All four of thesesites are (or were)ridgesegments bracketedat bothends by relativelyold (10-30 m.y.)oceaniclithosphere. TheNaTi basaltsfromtheWoodlarkbasinerupted alonga transformfault with an agecontrastof only 2.6 m.y., muchyoungerthanin the otherlocations. Finally,the arclikerocksarecommonlyrestrictedto Kroenke, 1981;Weisselet al., 1982;Taylor, 1986; Figure2]. The presenceof ridgesubduction, the rapidconvergence rate,andthespatialrestriction of thedifferentrocktypesimpliesthattheunusual volcaniccharacterof theWoodlarkbasinmaybe causedby tectonicprocesses.Therefore,we examinedtwo processes whichcouldrestrict unusualrocktypesto thevicinityof thetrench:(1) unusuallystrongcoolingof the lithosphereby hydrothermalcirculation,whichcouldbe responsible for the high-Mg andesites andNaTi basalts,and (2) volatile lossfrom the relict subducting plate,whichcouldbe responsible for the back-arc basinsand island arcs, and the Woodlark basin is neither. The origin of theserocksis relevantnot only in thecontextof Woordlarkbasinpetrology,but alsoto the muchmoregeneralconcernof whetherterrestrial ophiolitesreallyrepresent typicaloceaniccrustal rocks[Miyashiro,1973;MooresandJackson,1974; Mooreseta!., 1984]. Ridge subduction,suchas is occurringin theWoodlarkbasin,may increasethe chancesof ophioliteemplacement [Mooreset al., 1984]. Consequently, theremaybe petrologicand chemicalsimilaritiesbetweenophiolitesandthe unusual Woodlark basin rocks. arclike rocks. LITHOSPHERIC COOLING PLATE FLEXURE AND The oceaniccrestcreatedat mid-oceanridgesis cooledby a combinationof conductionand AbbottandFisk: TectonicallyControlledRock Suites 1147 valuesat depthat a rate of 0.222 kbar/km. We assume this model in the discussion to follow. Lithosphericflexureis onesourceof tensile stress.Lithosphericplatesbendastheyenter subductionzones. This flexureproduceslarge tensionalstresses in the upperhalf of the subducting oceanicplate (Figures3 and4). Thesehigh tensile stresses can cause new cracks to form on the ß.-.: /..o ß ,. surfaceof the flexuralbulge. Thesecrackswill propagatedownwarduntil the breakingstrengthof the rock exceedsthe tensilestresses inducedby Young Plate o o- flexure (Table 1). These new cracksadd to cracks causedby thermoelasticstresses. Old Plate o ,60 Woodlark Plate ;$o Pacific 460 Plate Fig. 2. SeismicitybeneaththeNew Georgiaisland arc [afterCooperandTaylor, 1985]. Both the old, previouslysubductingPacificplate(fight) andthe presentlysubducting Woodlarkbasinplate(left) are clearlyvisible. The seismicdataimply thattheolder subducting plate is intactandhasnot brokenapart. convection. The relativeimportanceof conduction andconvectiondependsprimarilyon thebulk permeabilityof the crust-sediment system. Becausemarinesedimentsare usuallyseveralorders of magnitudelesspermeablethanbasalt[Abbottet al., 1981; Andersonet al., 1985], the depositionof 200-300 metersof sedimenton top of thebasalt basementnormally causescrustalcoolingto be dominatedby conduction ratherthanconvection [Andersonet al., 1979]. However, the intensityof Increasedhydrothermalcirculationproduceslarge chemicalandthermalchangeswithinthe oceanic crust. If flexurally inducedcrackingandfaulting increases thepermeabilityor thetopographic roughness of an unsealedplate,hydrothermal circulationwill intensify. Althoughoceaniccrestas old as 80 m.y. is sometimeshydrothermallyactive [Embleyet al., 1983], theseolder,hydrothermally active areas are located in oceanic areas with unusuallylow sedimentation rates. Because sedimentation ratesneartrenchesareusuallyquite high,the sedimentcoveron mostsubducting oceanic crust reaches a thickness of 200-300 meters when the subductingplateis only 0-20 m.y. old. The subducting platein the Woodlarkbasinis 0-5 m.y. old and is coveredby lessthan200 metersof sediment;thushydrothermalcirculationand consequent lithosphericcoolingshouldbe intensified by flexurallyinducedcracking. EVIDENCE FOR INC•ASED CRACKING DUE TO LITHOSPHERIC FLEXURE Three observations indicate that new cracks form hydrothermal convection increases with increasing totalavailableheat,topographic roughness, and systempermeability[Stefanson,1983;Fehnet al., 1983]. Thus, changingany one of thesevariables canchangethe amountof convectivehydrothermal coolingat mid-oceanridges. Fracturingincreases thenumberof interconnected at flexuralbulges: normalfault earthquakes, new fault blocks,andporewaterheliumanomalies. Many trencheshavereactivatedfault blockson the flexurallyformedseawardtrenchslope[Schweller voids within oceanic crust and thus increases the alignedparallelto the trendof thetrench,ratherthan perpendicular to theoriginalspreading direction permeabilityof the oceaniccrust. The increased permeabilitystimulates hydrothermal circulation, especiallyin thelowerlayersof theoceaniccrust. The primarycauseof thesecracksis thoughtto be tensilethermoelastic stresses inducedby the cooling andcontraction of agingoceaniccrust[Lister,1974; Bratt et al., 1985]. Laboratoryexperimentson small mineralsamplesindicatethatcrackscanpropagate at cleviatoficstresses of 500 barsor greater[Martin, 1972]. However, the strengthof the oceaniccrustis probablylessthan500 bars,sinceanysufficiently largepieceof materialwill containmanypreexisting and Kulm, 1978; Watts et al., 1980]. In the Peru-Chile trench at 220-27 ø S and in the Guatemala trenchat 12ø-14ø N, someof thesefault blocks are [Von Huene and Aubouin, 1982; Schwelleret al., 1981,Figure6]. Theseobservations imply that lithosphere flexurecancausetheformationof new faultsandthatthedeviatoricstresses in theupper partof theflexuralbulgeexceedthetensilestrength of therock. Becausecrackspropagateat roughly half the stressthatis requiredto breaktherock [Verhoogenet al., 1970], new crackswill form at greaterc•epth•s thannewfaults. The OHe/'*Heratio of gascontainedin mid-ocean ridgerocksderivedfrom the mantlesourceof flaws and zones of weakness. Bodine et al. [1981] normal MORBis~1.15x 10-5 [Lupton andCraig, modeledthe strengthof the crustundertensionas increasingfrom nearzeroat the surfaceto higher 1975;R. Barnes,unpublished manuscript,1986]. When the rocksarecracked,juvenileheliumis 1148 AbbottandFisk: TectonicallyControlledRock Suites go has height A trench I • bulge atkx=3w/4 -,.,'"P•.'."• "':•i!i!i!i:i:i:i:i:!:i:i:i:i:' ...... "' ' I" ::i:i:!:i:i•'',it Fig. 3. Diagramof lithospheric flexureat a trench(notto scale):h, elasticthickness; x, horizontal distancefrom the load;z, distanceaboveor belowtheneutralsurface;y(x), flexureof theplate;A, maximumheightof thebulgeabovethe surfaceof zerodeflection(measured from theseawardside of theplate). The bulgehasits maximumheightat kx = 3•r/4. Themaximumstressis at kx = •r/4. The upperhalf of theplateis in tension;thelowerhalf of theplateis in compression. concentration, of H•erelativeto therockandhasthe depthcouldcomefrom newly crackedrock [R. Barnes,unpublishedmanuscript,1986]. al., 19801,the releaseof heliumby the crackingof •ewly,emplacedrock causesanomaliesin the TENSILE STRESS CAUSED LITHOSPHERIC FLEXURE ventsat mid-oceanridges. Fracturingof older, previouslyuncrackedoceaniclithospherewill also releasejuvenileheliumgasinto the surrounding water. The onlyDeep SeaDrilling Project (DSDP) sitedrilled on top of a flexuralbulge,site436, locatedon 110-m.y._-old lithosphereenteringthe Thickerplatesaremoredifficultto bend,and consequently, stresses andcrackinginducedby flexuralbendingof theplateareproportional to the thicknessof theplate. Becauseplatesincreasein thicknessasthey age,maximumflexuraltensile stresses increasewith increasingageof theplate. the porewater [R. Barnes,unpublishedmanuscript, 1986]. To producethisratio, 10% of theporewater heliummustcomefrom newly crackedrocks. As partof theplatewhichbehaveselastically.The mechanicalthickness of theplateis thethickness of thelayerthatbreaksratherthanflowsunderthe influenceof largeflexural bendingstresses, andis approximately givenby thedepthto the600 øC released. Because most seawater has a low atmosphericOHePHe ratioof 1.4x 10-6[Lupton et BY He/•4He ratioofpore fluids emitted athydrothermal Japan trench, hasa 3He/4He ratioof4.40x 10-6in thel•eliu•ndiffusesslowlythrough thesediments, the •He/'*He ratio of the pore water decreases. Consequently,100% of theporewaterheliumat These flexural tensile stresses are concentrated in the isotherm [McNutt, 1984]. Mechanical thickness shouldnot be confusedwith the plate'sthermal thickness,whichis roughlyequalto thedepthto the 1o a• 1300 øC isotherm [Parsonsand Sclater, 1977]. If o 5 i o 1oo DISTANCE, 2oo km Fig. 4. The tensilestressat thetopof theplate(z = 1/2 h) versusdistancefrom thebeginningof the trenchwithinlithospheric platesof differentages(in millionsof years). Note thatthemaximumtensile stressexceedsthe breakingstrengthof the rock (1 kbar)andthecrackpropagation stressevenin the youngestsubducting plates.Seeappendixfor derivationof the figure. tensionalcracksanddeepwatercirculationare present,the convectivetransferof heatwill increase thedepthto the 600 øCisothermandwill consequently increasethemechanicalthickness. The portionof theplatethathasincreased surface tensilestresses inducedby flexureis calledthe flexuralbulge. Our calculations (seeappendixfor details)of the tensilestressat thetopof theplate within the flexuralbulgeindicatethatstresses will locallyexceedthestrengthof therock (Figure4). The depthof crackingvarieswith plateage. For a 5-m.y.-oldplate,cracksmay extendto 3.7 km depth nearthe flexuralbulge (Table 1). Figure5 shows thedepthof crackingandconsequent hydrothermal alterationwhichcanbe producedby lithospheric flexure. On platesolderthan20 m.y., thisdepthis greaterthanthe 4- to 8-km depthof hydrothermal AbbottandFisk:TectonicallyControlledRock Suites 1149 TABLE 1. Elastic Half-Thickness,Maximum TensionalStressat the Top of the Plate near the FlexuralBulge,andMaximum Depthof Flexure-Induced Cracking,of the OceanicLithosphereasa Functionof PlateAge* PlateAge, m.y. ElasticHalf-Thickness, km 1 5 10 20 40 80 140 2.1 4.6 6.6 9.3 13.1 18.6 24.6 MaximumStress,kbar 3.1 4.2 5.2 6.2 7.2 8.6 10.5 Depth,km 1.8 3.7 5.1 6.9 9.2 12.3 15.6 *The plateis assumed to be conductively cooled. alteration inferred from marine seismic studies [Lewis, 1978; Abbott andLyle, 1984] and from ophiolites[GregoryandTaylor, 1981; Cockeret al., 1982]. This impliesthatthe crackingof the oceanic crustin the vicinity of flexuralbulgeswill very quicklyexceedthenormaldepthof the hydrothermal cracking and will resultin thereleaseof juvenile 3He. Ourcalculations aretherefore consistent with the observationof porewaterheliumanomaliesin the sedimentoverlyingthe flexuralbulgenext to the Japantrench. CALCULATION HYDROTHERMAL OF TIlE DEPTH CIRCULATION OF Hydrothermalconvectionlowersthe temperatures in the upperpart of the plate (Figure6). Becausethe mechanicalthicknessof theplateis determinedby the depthto the 600øC isotherm,theselower temperatures will alsoincreasethethickness of the HALF-ELASTIC • •o o I I I I I 0 30 60 90 120 I PLATE AGE, m.y. Fig. 5. Elastichalf-thickness of a conductively cooledplateandcrackedthickness of the sameplate at the apexof theflexuralbulgeversusageof the platein millionsof years. The crackedlayerhas flexurallygeneratedtensilestresses whichexceedthe breakingstrengthof therock. The dashedline indicatesthe approximateincreasein elastic half-thickness of theplatecausedby heatremoval duringhydrothermal circulation. Seeappendixfor derivationof the figure. layer whichhastensilestresses (Figure5). Tensile stresses will occurin the upperhalf of thislayer, which is boundedroughlyby the 300ø C isotherm. Althoughwe haveno dataon the temperatures within ridge crestcrustaffectedby flexure,Fehnet al. [ 1983] modeledthe intensityof hydrothermal circulationat the CostaRica rift by usingcrustal modelswith differentratesof exponentialdecrease of permeabilitywith increasingdepth. Subsequent in situ measurements in the same area found that the permeabilityof the oceaniccrustdoesdecrease exponentiallywith increasingdepth[Andersonet al., 1985]. Fehn et al. [1983] foundthat a changein thicknessfrom ~2.5 km to ~4.2 km of the layer with a highenoughpermeabilityto supportactive hydrothermalcirculationchangedtheratio between minimum and maximum values of heat flow in one off-ridgeconvectioncell from about1:2 to about1:8 (Figure7). The modelwith a maximumdepthof circulationof ~4.2 km producedthebestfit to surfacemeasurements of heatflow. Subsequent, independentobservationsof the heatflow in this areaindicatea maximumdepthof hydrothermal coolingof 4-6 km [Hobart et al., 1985]. If we use the isothermsin Fehn'smodels,the depthto the 300ø C isothermon crustlessthan 1 m.y. old with activehydrothermal circulationis at least3 km, equivalentto the thicknessof the layerwith tensile flexuralstresses within a conductivelycooled 5-m.y.-old plate (Figure 5). The maximumdepthof significanthydrothermalcirculationis shownby the streamlinesto be 4.2 km. Becausehydrothermal crackingis causedby the temperaturecontrast betweenthe water andthe surrounding rock [Lister, 1983], thisgreaterdepthis actuallya betterestimate of the maximumdepthof cracking. We can estimate the effect of flexure on the hydrothermalcirculationby notingthatlithospheric flexureroughlydoublesthenumberof largefaults on the trenchwall. Otherfactorsbeingequal,a doublingof surfacefaultswouldquadruplethe averagesurfacepermeabilityof theoceaniccrustand 115o AbbottandFisk: TectonicallyControlledRock Suites Temperature 600 1200 i Oceanic Crust With maximum deviatoric stresses near transform faults couldbe ashighas200-400bars:a significant fractionof thetotalstress requiredto propagate cracks. Because of thesmallsizeof theplate,theregional stress field in the Woodlark basin has not been Hydrothermal Circulation T Temperature 600 1200 Oceanic Crust With No Circulation T Fig. 6. Schematiccartoonshowingthechangein thethermalprofile(fight) andconsequent increase in the elasticthickness, E, of younglithosphere (left) asa resultof hydrothermal circulation.Circulating fluidsareestimated to havea maximumtemperature includedin globalsolutions of intraplatestress [Richardson et al., 1979]. Theseglobalsolutions indicatethatslabpull andridgepushcontribute significantly to intraplatestresses andthatslabpullis about2-3 timesstrongerthanridgepush. Consequently, theregionalaxisof tensionalstress withintherapidlyconverging Woodlarkbasinplate is mostlikely within 30ø of the trendof the Simbo transform fault. Thisorientation of theregional stress field near the transform trace would cause localtensilestresses thatexceedtheregional tensionalstresses by asmuchas50-100% [Fujita andSleep, 1978]. Therefore,tensionalstresses alongthe Simbotransformfault will havethree additivecomponents: thefirstinducedby therm•lasticstresses, thesecond by lithospheric flexure,andthe thirdby thetransformorientation. A ridgecrestor transformlocatedon a flexural bulgeshouldhaveanincreased depthof hydrothermal circulation andincreased coolingof the crust. The cool crust would result in short-lived crustalmagmachambers, andconsequently, any steadystatemagmachamberswouldlie withinthe mantle.Deepmagmachambers mayoccuralong of 400ø-500øC. In this cartoon,the total thermal thickness, T, doesnot change.If flexurallyinduced coolingis profound,the thermalthicknesswill also •o ß C increase. woulddoublethe maximumdepthof crack penetration[Lister, 1983]. This, in turn, would roughlydoublethe maximumdepthof hydrothermal circulationandalterationto 8.4 km (Figure5). bJ T O k(z)= 5 x10-15exp(-z/434)m 2 ' TRANSFORM FAULTS AS A SECONDARY CAUSE OF HIGH PERMEABILITY In additionto flexurallyinducedtensilestresses, transform faults can cause local increases in tensile stresswhichdependuponthe orientationof the transformwithin theregionalstressfield. If the regionaltensilestressaxisis parallelto the orientationof the transform,the maximum tensile stressnearthe transformwill be 50% greaterthan the regionalstress[Fujita andSleep,1978]. However, if the transform fault is oriented at an angleof 30ø to theregionaltensilestress,the maximum tensile stress near the transform will be twice the regionalstress[Fujita andSleep,1978]. The regionaldeviatoricstressfieldswithin large plateshavea maximummagnitudeof 100-200bars [Richardsonet al., 1979]. This impliesthat Fig. 7a. Top: Calculatedheatflow versusdistance from theridgecreston a ridgewith active hydrothermalconvection.Solidlines:totalheat flow. Dashed lines: conductive heat flow. Bottom: Streamlinesof water flow (dashedlines) and isotherms(solidlines)versusdistancefrom theridge crest. Streamlinesare unevenlyspacedto better delineate areasof weakcirculation.Thefunction in., themiddleof thefigureis thepermeability,k, in m•, versusdepth,z, in the oceaniccrust(in meters) [after Fehn et al., 1983]. AbbottandFisk:TectonicallyControlledRock Suites increased hydrothermal coolingalongthe Simbo transform E • fault. The otherknownoccurrences of high-Mg andesites in arcvolcanossupporttheideathata magmachamberin contactwith harzburgite can producehigh-Mg andesites.Many islandarcsand most continental arcs have crustal thicknesses in moJk(z)=5 T x10-15exp(-z/.724)m 2 excessof 30 km [Gill, 1981]. In thosearcs,magma chambers at lessthan30 km wouldnot bringbasaltic magmaintocontactwith harzburgite, andhigh-Mg andesites couldnot form by themechanism proposedby Fisk [1986]. The two arcswith documented occurrences of high-Mgandesites (boninites),theIzu-Boninarc [HickeyandFrey, 1982] and the New Britain arc [Johnsonet al., 1983], have crustal thicknessesof 12-15 km and 0 10 DISTANCE, 20 Km Fig. 7b. SameasFigure 7a, exceptfor different permeabilityversusdepthfunction. This model providesthe bestfit to heatflow measurements from the CostaRica rift, an area of normal,unflexed oceanic crust [Fehn et al., 1983]. 20-30 km, respectively[Gill, 1981]. Consequently,bothof thesearcshavelocations wherebasalticmagmacouldequilibratewith harzburgiteat pressures of lessthan 10 kbar. ORIGIN OF NaTi BASALTS Dredgehaul32 (Figure1) sampledvolcanic rocks formed at the intersection of the Simbo the Simbo transform where it crosses the flexural bulge(station31, Figure 1). Recentvolcanismon the Simbo transform occurs at two locations: near transform with theWoodlarkbasinspreading center, about33 km fromthetrench.Theflexuralbulge produced by a 1- to 2-m.y.oldsubducting plateis the ridge-transformintersectionandat the Simbo volcano.This "leak•" behaviorimpliesthatthe 50-60 km wide (Figure3); thusthe Simbo ridge-transform intersectionis within the flexural bulge. This dredgerecoveredrockswith an unusual Simbotransformis not likely to haveunusuallythin crust,whichcouldalsoproducea subcrustal magma lessthan0.3%K20. Perfitetal. [1986]hav• chamber. Simbo volcano has been characterized as a forearcvolcano,butthelocationsof deep earthquakes indicatethatit is seawardof the trench [CooperandTaylor, 1985]. As discussed earlier, the Simbotransformis in theproperorientationto magnifythe regionaltensilestresses inducedby subduction. A DEEPER MAGMA CHAMBER AND THE FORMATION OF HIGH-Mg ANDESITES Fisk [1986] foundthattheinteractionof n-type MORB andharzburgiteat pressures of lessthan 10 kbar canproducehigh-Mg andesites[Johnsonet al., 1986]. He infersthat a subcrustal magmachamber canresultin thegenerationof high-Mgandesites with boniniticaffinities. High-Mg andesites with alkalienrichments similarto theexperimentally producedandesiteswere foundin a dredgehaul northwest of the Simbo volcano in the Woodlark basin [Johnsonet al., 1986]. In addition, comparison of experimentallyproducedandesites [Fisk, 1986] with naturallyoccurringoceanic andesites (Figure8) showsthattherearemajor elementchemicalsimilarities. Consequently, the Woodlarkbasinhigh-Mg andesitescouldhave formedin a subcrustal magmachamberproducedby composition: over4% Na?.O,over1.5%TiOg,and termed these rocl•s NaTi basalts. We use a less restrictive definition for NaTi basalt of over 3.4% Na2 O,over1ß5%TiO2 , andlessthan0.5%K20. TheNaTi basaltmustalsohavea Mg number indicatingthatit is notextensively fractionated anda low (lessthan 1%) water content. Basedon this definition,all of thebasaltglasses dredgedfromthe Caymanspreading centerin theCaribbeanareNaTi basalts[CAYTROUGH, 1979]. The Cayman spreading centeris 110km longandis bounded by two transformsover 1000km long. At the ridge-transformintersections, eachof these transforms juxtaposesnew oceaniclithosphere against34-m.y.-oldlithosphere,producingage offsetsof 34 m.y. [Fox and Gallo, 1984]. Because the Caymanridgeis relativelyshort,muchor all of theridgecrestwill "see"the thermaleffectsof the two boundingtransformfaults. The old oceanic lithosphereon the othersideof the transformfaults couldindirectlyproducerapidcoolingof any shallowmagmachambersaswell ascoolingof the local asthenosphere. Enhancedcrustalcoolingis alsoassociated with the unusualbasaltglassesfrom the 40-km-longridge segmentin theCharlieGibbsfracturezone(Figure 9). The ageoffsetsof the two adjoiningtransform segmentsare 16 m.y. 'and10 m.y. [Searle, 1981]. 1152 AbbottandFisk: TectonicallyControlledRock Suites 6 5 - •xe\.•Q_ • • Fieldforlower ,•' %t',•/ •,%,.oXx/) //•: alkali series of.,' •o• •t• • / Woodlark Basin 0 _ 4 ,,-' + 2 -I 40 I 45 I 50 .... 5•5 I 60 I 65 7•0 SiO2 wt % Fig. 8. Alkali-silicadiagramfor oceanicandesites andexperimentally synthesized high-magnesium andesites.Triangles,Hart [ 1971];squares,Byerly et al. [1976];diamonds,experimentally producedliquidsat 1250ø C; starredhexagon(MORB), whichwasthe startingbasaltcomposition in experiments,Fisk [1986]. Field for low silica andesitesof Johnsonet al. [1986] is shown. Fractionalcrystallization of olivinealonefromtheexperimental andesite produces thesilicaand alkali enrichmentshownby the line of crosses. The threeglasssamplesfrom the CharlieGibbs ridgesegment have3.4-3.7%Na20, 1.6-2.0% TiO , and0.3-0.5%K20 [Schillinget al., 1983]. The2water contents, which range from 0.1to0.6%, aretoolow for thesamplesto havebeenextensively altered. The samplesalsohaveMg numbersfrom 57 to 63, andthuscouldnot havebeenderivedby fractionation•Consequently, two of the documented occurrences of NaTi basaltsarefrom smallridge segmentssandwichedbetweentwo longtransform faults. In additionto the coolingeffectscausedby thelong ageoffsets,the hightopographic relief of long ageoffsettransforms[Fox andGallo, 1984] will alsoincreasetheintensityof hydrothermal circulation [Stefanson,1983] and hencethe efficiencyof coolingof theoceaniclithosphere. Another nonsubduction zone area where NaTi basaltshavebeenfoundis at two topographic steps on theSouthwest Indianridgewherethetriple junctionhasbeenmigratingeastwardin successive ridgejumps. The largetopographic steps[Anderson et al., 1977] andanomalously gmatdepths[Sclater et al., 1981]observedat theboundarywith the Antarcticplateon thispartof theridgecrestare evidencethatsuchridgejumpshaveoccurred. NaTi basaltglass(8.5% FeO, 7.6% MgO, 4.2% Na:•O,0.2% K2 O, 1ß7% TiO2 ) [WßGß Melsonand T. O' Heam, unpublished compilationof glass analysesof MORB, 1985] occursat 28.85ø S, 61.92øE, in an areawheretriplejunctionmigration anda ridgejump havejuxtaposed crustof about anomaly19 andanomaly8 ages[Sclateret al., 1981]. The jumping SW Indianridgesegmentwas thusthermallyequivalentto a shortridgesegment sandwichedbetweentwo piecesof 19-m.y.-old oceaniclithosphere.NaTi basaltglass(8.5%FeO, 7.6%MgO,3.7%Na20,0.2%K20,and1.5% TiO2.) [W ' G ß Melson andT ' O' I-Fearn,unpublished compilation of glassanalyses of MORB, 1985] also occursat 26.62ø S, 67.53øE, wherethetriple junctionmigrationboundaryliesbetweencrustof anomaly7 and 5 age [Sclateret al., 1981]. This locationwasthermallyequivalent to a smallridge segmentbetweentwo transforms with ageoffsetsof 18 m.y. Thus, the NaTi basaltson the Southwest Indianridgewerealsogenerated at a ridgesegment betweentwo piecesof unusuallyold crust. Reducedpartialmeltingof a depletedmantle sourcehasbeenproposedas a causeof theNaTi basaltsin the Woodlark basin [Perfit et al., 1986]. A mid-ocean ridgetype,depleted [%urce forthese rocksis supportedby measuredœ1•uvaluesof +9.2 to + 10.0 [Staudigelet al., 1986]. The reduced partialmeltingof themantlecouldbeindirectly causedby theincreased crustalcoolingwhichis inferred for all three of the tectonic locations of NaTi basaltsmentionedpreviously.However,if one acceptsall of theCaymantroughglassesasNaTi basalts, thereducedpartialmeltingmustoccurin the mantleupwellingcell beneaththeridgecrest. AbbottandFisk:TectonicallyControlledRock Suites 1153 theflexuralbulgeof the subducting Explorerplate [Riddihoughet al., 1980; Cousenset al., 1984],is divertedby the Paul Reveretransformridge (age offset, 4.5 m.y.), which parallelsthe trench. Thus, the hydrothermalsystemin the Explorerrift should not be sealed,andflexurallyinducedcrackingand increasedcoolingcannotbe ruledout. One sample from the Explorerrift hasa low watercontent (0.11%), an unfractionated Mg number(61), anda CAYMAN TROUGH N-MORB Nap_O,% Fig.9. Ti02 versus Na20content of NaTi basalt glassesfrom differentareascomparedto normal MORB with the sameMg number. Dottedline: Field of analysesfrom the Caymantrough. Triangles:Woodlarkbasin. Crosses:CharlieGibbs ridge. Squares:SouthwestIndianridge. Open circle: Explorerridge. The ridge segmentin the Caymantroughis 110 km long,but conductivecoolingof theoceaniccrust near transform faults is calculated to extend less than 15 km from ridge-transformintersections [Forsythe andWilson, 1984]. However,convectivecooling of theupwellingcell in the asthenosphere near transformfaultscanextendto distancesequalto the thicknessof the transformlithosphere[Forsytheand Wilson, 1984]. On a ridgesegmentsandwiched betweentwo transformswith ageoffsetsof 34 m.y., theconvectioncell beneathanentire110-km-long ridgesegmentcouldbe affected. Consequently, the upwellingcell or cellsfeedingtheridgecrestin the Caymantroughupwellmoreslowlythanthose beneathmore typicalridge segments, thereby producingdecreased decompressive meltingof their MORB source(Figure 10). However, the age offset on the Simbotransformis only 2.6 m.y., muchless thanat the Caymantroughtransforms, the Charlie Gibbs transforms, and the old SouthwestIndian ridgetriplejunction. If increased coolingcauses reducedpartialmeltingof the asthenosphere, some other tectonic factor must act to cause anomalous crustalcoolingon the Simbotransform. Two features of the Woodlark basin could cause anomalous crustalcoolingalongtheSimbo transform:the longdurationof subduction in this areacouldcoolthe underlyingasthenosphere [Perfit et al., 1986] or the additionalhydrothermalcracking andcoolinginducedby lithosphericflexureand transform fault orientation could increase the thickness of thelithosphere at theridge-transform intersection.The relativeimportanceof thesetwo mechanisms couldbe testedby drillingfor Na-Ti basaltsat flexuralbulgeswhicharecoveredby over 300 m of sedimentandareunequivocally hydrothermallysealed. Much of the Explorerplate,locatedoff the coast of southwestern Canada,is coveredby over300 m of sediment. However, the continentalsediment supplyto the Explorerrift, a ridgesegmentwithin NaTi basaltcomposition of 0.28%K20, 3.52% Na20, and1.82% TiO2 [Cousens et al., 1984]. -The secondhypothesisfor the originof NaTi basalts,increasedasthenospheric coolinginducedby long term subduction,couldalsoproduceNaTi basaltson the Explorerridge. However, the cooling inducedby longterm subduction is not requiredto remainconfinedwithin theflexuralbulgeandshould not be affectedby thick sedimentcover. Despite extensiverecentstudyof the Gorda-Juande Fuca-Explorerplatesystem[Clagueet al., 1984; International N.E. Pacific Activities Consortium (INPAC), 1985], we know of no NaTi basalts dredgedat locationsoutsidetheflexuralbulgewith thin ridge crestsedimentcoveror at locationswithin the flexuralbulgewith thickridgecrestsediment cover. Therefore,additionalasthenospheric cooling causedby lithosphericflexureand/ortransformfault orientationis the bestexplanationfor theWoodlark basinandthe ExplorerridgeNaTi basalts.This increasedcoolingwouldcausethe SimboandPaul Reveretransformsto behavethermallylike much longertransformsandwouldproducetheNaTi basalts. VOLATILES THE RELICT AND SEISMICITY WITHIN SUBDUCTED PLATE Perfit et al. [ 1986] suggested thatthe arclike rocksin theWoodlarkbasinarecausedby volatile additionto theWoodlarkbasinasthenosphere during thepreviousphaseof subduction to thenorth (Figure2). To testthisidea,we calculatedthe lengthof time thatvolatilescouldbe heldin stable mineralphasesin a descending slab. The mantle retention times of volatiles contained within a subducting plateare a functionof the ageof theplate at the initiation of subduction,the thicknessof the volatile-richlayer, andthe breakdowntemperatures andpressures of volatile-richminerals. The maximumthicknessof the volatile-richlayer of normal, unflexed oceanic crustis ~8 km [Abbott and Lyle, 1984]. Oncethisvolatile-richlayerreachesa temperatureand/orpressuresufficientto causeeither partialmeltingor completebreakdownof volatile-richphases,the subducting platecanno longercontributevolatilesto theoverlyingmantle. The mantle retention time is defined here as the total lengthof theportionof the subducting slabwhich retainsvolatilesdividedby the convergence rate. We have calculated mininum mantle retention AbbottandFisk:TectonicallyControlledRockSuites PRODUCES NA-TI RIFT VALLEY I I^ A ? A BASALT have shorter mantle residence times than in other AXIS platesof the samethermalage. However,these largehorizontaltemperature gradientswill not be presentin the deeperportionsof a rapidly subducting platewhichsuddenlyceased convergence.Consequently,we canusea thermal modeldevelopedfor fasterconvergingplatesto predictthe thermalbehaviorof a platewhich suddenlyceasedconvergence. r'- 4x -n • A I + ' • ^AI TF I I T.,.F, ';::.':' ;:':-f;:':'f-:: ':'.')]:' :': •:' •:5':': .::-':': )2;': .:.'.'..': .:L-':.".:' ;:."4': !.:i' ;:-:5:' :.: .f:'5:'.::.i. :':':! .:". ':! .5' :.i.5 ':• .:'.: :•. !:':.: .:(!.:.:_(:.!.:)::.:((:!((:i':?](:' PRODUCES N-MORB RIFT VALLEY AXIS T.F-t t t t t ttT. F. ttttttt ttttttt In order to facilitate observational tests of our model, we have also calculatedminimum mantle "retentiontimes"for seismicitywithin a subducting plate. Seismicity"retentiontimes"dependon the timerequiredto heatthecenterof theplateto a temperature toohighfor brittlefailureto occur. Althoughthebrittle-to-ductile transitionis clearly dependenton strainrates,the transitionfrom seismic to aseismicbehaviorcanbe described reasonably accurately by usinga cutoffpotentialtemperature of 750ø C at depthsof 0-125 km [Molnar et al., 1979; Jarrard, 1986; Chen and Molnar, 1983; Weins and Stein,1983]. Becausethecenterof a subducting plateis usuallyabout100-125km beneaththe volcanicline of an island arc [Gill, 1981; Jarrard, 1986],seismicactivitywill not be presentbeneath the arcif the entireplateis heatedabove750øC. Volatileretentiontimesin theplatedependon the pressureand temperatureof breakdownof volatile-rich mineralsand/orinitiationof partial melting. The top 8 km of a subducting plateis usuallyabout 100 km beneaththe arc. At 30 kbar (roughly100 km), partialmeltingandthe ..T:.•:::..::::.:.:.;`..•.::::z....•.:.:.:(.:.•.::?:.:....:.;.•:.:........ ;.. ...:.5.•...:....:.t:....•:.:....:.?..::•.:.:?:&.!•.i?.h?.:•.th.i&?.•:.?.i?.!?.i?.i?.!?.! Fig. 10. Schematiccartoonsof slowed asthenospheric upwellingbeneaththe axialvalleyof a mid-oceanridgeproducingNaTi basalts(top) comparedto fasterasthenospheric upwellingbeneath a normalridgecrestproducingMORBs (bottom). Boundariesare transformfaults (TF). Area inside blockslabeledTF represents theadditional lithosphericthicknessin the transformfault zones. Dottedareais the zoneof partialmeltin the asthenosphere. Lengthof arrowsis proportionalto thespeedof asthenospheric upwelling. timesfor volatilesin a subducting plateusingthe thermal model of McKenzie [1969] and an 8-km-thickvolatile-richlayer. This thermalmodel doesnot includethe effectsof the thermalgradients in the overlyinglithosphericplate,whichwould increasetheretentiontime. Althoughtheactual retentiontimesare somewhatlongerthanour calculatedvalues,relative retentiontimes are portrayedquiteaccurately.Becausehorizontalheat conductionalongthe axisof the slabmakesan insignificantcontributionto reheatingof rapidly subducting plates,minimumretentiontimes calculatedfrom thismodelusingconvergence rates from 2 to 12 cm/yr areconvergence-rate-independent to within 1% for 10- to 100-m.y.-oldplatesand to within 10% for 1-to 10-m.y.-old plates. The reheatingof slabssubducting at lessthan2 cm/yris compressed into a relativelysmallregionby the slow convergence, producinglargehorizontaltemperature gradientsalongthe axisof the slab. Becauseof theselargehorizontaltemperature gradients, horizontal and vertical heat conduction are both important,and volafilesin slowlysubductingslabs breakdown of richterite-tremolite commence at ~1000øC [Green,1973;Hariya andTerada,1973]. Thus,within a subducting platewhichsuddenly ceasedconvergence, the lastvolatilesin theportion of theplatebeneaththe volcanicline wouldbe lostat temperatures of ~ 1000øC, andthelastseismicity at temperatures of ~750ø C. ORIGIN OF THE ARCLIKE ROCKS Thevolatileandincompatible element enrichments in the arclike lavas from the Woodlark basinhavebeenproposedto be theresultof a polarityreversalof thetrench-arcsystem[Perfitet al., 1986] about 8- to 10-m.y.ago [Colemanand Kroenke, 1981; Cooperand Taylor, 1985]. However,the relictslabcontinuedto convergeand to producethrustingon the Solomonislandarcuntil at least6 m.y. ago [ColemanandKroenke, 1981]. Thus,thepolarityreversalof the arcandsubduction of therelict slabendedlessthan6 m.y. ago. The model of Perfit et al. [ 1986] doesnotrely uponfluids comingdirectlyfrom the older, previouslysubductingslabandenteringtheisland arc magmas,but insteadreliesuponsubduction and associated mantlecontamination priorto thepolarity AbbottandFisk: TectonicallyControlledRock Suites o 750oc x• •)o_ '• • O0 (center) /.-"melt,•ng Aor amph!b.o. le. out • uo- Subducting Plate Age, m.y. Fig. 11. Minimum mantleretentiontimesof volatilesand seismicitywithinplateswhichhad differentagesat the onsetof subduction.Dotted curve:Time neededto warm centerof a subducting plateto 750ø C, as a functionof subducted plateage. Seismicactivityis thoughtto ceaseabovethis temperature.Hachuredcurve:Time neededto warm the upper8 km of theplate above1000ø C. Most hydrothermallyemplacedvolatileswill havebeen lostwhentheplatereachesthistemperature. Minimum mantleretentiontimesarecalculatedusing theformulain McKenzie [ 1969]. The trianglewith errorbarsshowsthe estimatedage andtime since polarityreversalof thepreviouslysubducting Pacific platewhichlies beneaththeNew Georgiaarc. reversaland consequentcessationof subduction. Our simplethermalmodels(below)indicatethatan old, previouslysubducting platecanremaincold enoughto contributevolatilesto the surrounding mantlefor millionsof yearsaftersubduction has ceased.Consequently,the slabwhich stopped subducting lessthan6 m.y. agocouldbe presently contributing volatilesto themantleconvection cell thatfeedstheWoodlarkbasinridgesegmentclosest to the trench. The minimum the seismicactivitynearthevolcanoesis shallowand sparseandcanbe attributedto surfacedeformation ratherthanto a subductingplateat depth [Drummondet al., 1981; Kellogg et al., 1985]. The subducting platesin thesethreeareasare all Pleistoceneto Miocenein age [HeezenandFornari, 1976; Sugi and Uyeda, 1984], consistentwith our modelpredictions. Thesemodelsalsoimply thatvolatilescould continueto be derivedfrom the old, nonsubducting slabin zonesof polarityreversalof the arc. The previouslysubducting Pacificplatewhichunderlies theNew Georgia/Solomon islandarcswasof Lower Cretaceousage (100-136 m.y.) at the initiationof (8km) ,,--, •:x • [ • 1000øC 80 120 160a&O • -m , 40 1155 mantle retention times for volatiles andseismicityat depthsof 100-125km show severalinterestingtrendswhichprovide observational testsof our results(Figure 11). Older subducting plateswill remainseismicallyactivefor a longertime thantheyretainvolatiles. Conversely, youngersubductingplates(lessthan30 m.y. old uponenteringthe trench)will retainvolatilesfor a longertime thantheyremainseismicallyactive. Volatilesderivedfrom the subducting plateare widelybelievedto be an importantcontributorto islandarc volcanism[McBimey, 1969;Boettcher, 1973;Nichols andRingwood, 1973; Kushiro, 1973]. Consequently,theseresultsimply that young,activelysubducting plateswill sometimes produceactiveislandarcswithoutthepresenceof subcrustal seismicitybeneaththevolcanicline. The Cascadesarc and the Andes in southernChile, whichhaveno seismicactivitybeneaththe volcanic line [BarazangiandDorman, 1969], areexamplesof this. In southernColumbia and northernEcuador, subduction[Heezen et al., 1976]. The minimum volatileretentiontime in a plateof thisageis 6 m.y. (Figure 8). The relict slabceasedsubductingless than6 m.y. ago. Consequently, the incompatible element enrichments in the rocks of the Woodlark basinspreadingcentermay be theresultof active volatileandincompatibleelementlossby an immobileslabof Pacificplate(Figure2) andarenot requiredto be causedby a previousphaseof mantle enrichment. CONCLUSIONS Lithosphericflexureat trenchescausestensionin theupperpart of the subducting plate,whichcracks new rock and forms new fault blocks. Further idenceforflexurallyinduced cracking is thehigh e contentof porefluidsfrom DSDP site436, locatedon the crestof the flexuralbulgeof 110-m.y.-oldoceaniccrustenteringthe Japan trench.If the flexedplateis youngandhasthin sediment cover, the new cracks and increased permeabilityshouldalsoincreasetheintensityof hydrothermalactivity. This increasein hydrothermal activitywouldbe mostpronounced on a ridgecrest or transformfault sittingon a flexuralbulge. Increasedridgecresthydrothermal circulationwould lower near-surface isotherms and could cause a deepermagmachamber.A magmachamberlocated in the uppermostmantle,wherebasaltmagmacould reactwith harzburgite,explainsthe anomalous high-Mg andesiteson Simbotransformfault, which lieswithin the flexuralbulgeof theWoodlarktrench. The NaTi basalts,also extrudedon the Simbo transform,maybe theresultof eruptionthrough anomalously thick,youngoceaniclithosphere.The coldneighboring platescausemantleupwellingto slow,thusproducingdecreased decompressive partialmeltingof therisingasthenosphere. NaTi basaltsalsooccurat shortridgesegments sandwichedbetweenlong ageoffsettransformsand at largeageoffsetridgejumps,bothof whichare logicallocationsfor slowedmantleupwellingdueto anomalouslycoldneighboringlithosphere. Mathematical models of minimum mantle 1156 AbbottandFisk:TectonicallyControlledRock Suites retentiontimesof seismicityandvolatilesare consistentwith the followingobservations: (1) active Holocenevolcanismandno deep,subarcseismicity APPENDIX: CALCULATION OF STRESS AND THE WDTH OF FLEXURAL BULGE in the Cascades,northernEcuador, southern Colombia, and the southemAndes; and (2) At smalldeflections of theplate,theelastic thickness is approximately equalto themechanical thickness.Thisassumption is particularlygoodfor incompatibleelementenrichmentin therocksof the Woodlarkbasinspreadingcenterlessthan6 m.y. aftertherelictPacificplatestoppedsubducting. Consequently, theincompatibleelementenrichments in the rocksof the Woodlarkbasinspreadingcenter arenot requiredto be causedby a previousphaseof mantle enrichment but could continue to be derived from the old, previouslysubducting plate. The unusualrock typesfoundin theWoodlark basinmay alsobe foundin ophiolitesemplacedasa resultof ridge subduction and/ortrenchreversal. The Troodosophiolite,whichis postulatedto have formedas a resultof a combinationof ridge subduction andseafloorspreadingbehinda trench [Moores et al., 1984], has somearclike rocks trenches[McNutt, 1984]. Therefore, all of the followingcalculations aremadeassuming thatthe elasticandmechanicalthickness areroughlyequal. The thickness, h, whichis equivalentto thedepthto the600ø C isotherm[McNutt, 1984],is proportional to the squareroot of the plate age: h(t)= 7.4x 10-2t1/2cm (1) wheret is theageof the oceanicplatein seconds. The loadat thetrenchcausestheplateto deformwith thefollowingflexure,y(x): [Miyashiro,1973;Robinsonet al., 1983]. The NaTi basalts found in the Woodlark basin would be easyto dismissasspilitesand/oralteredbasaltif theywerefoundin anophioliteandarenotlikelyto havebeenreported.Boninitesoccurin thepillow lavasof the Troodosophiolite[Cameronet al., 1979;Robinsonet al., 1983]. Consequently, two of the unusualrock typesreportedfrom theWoodlark basin,thehigh-Mg andesites (boninites)andthe arclikerocks,arefoundin an ophioliteemplacedasa resultof ridge subductionand trenchreversal. The researchon theWoodlarkbasinshouldbe applicable to otherophiolitesemplacedduringridgesubduction A(O y(x) = cos[k(t) x] exp[ -k(t) x] (2) 0.067 where k(t) is the flexural wavenumber,x is distance from the trench,andA(t) is the maximumflexure inducedby the load (which occursat kx=3x/4) (Figure3) [Caldwell and Turcotte,1979]. We estimateA(t) from McNutt's [1984] data: and/or trench reversal. A(t)= 5.2x 10-4 t1/2cm Thesemodelsof the effectsof lithospheric flexurecanbe testedby dredgingandby heatflow surveysin anotherareaof activeridge subduction. The Woodlarkbasinis not the optimumlocationfor relatedultimatelyto theplatethickness throughthe such a test because of the small amounts of sediment relations: covetingthe flexuralbulgeandbecauseof the complicatingeffectsof thepolarityreversalof the arc. The requirementsfor sucha testareaare k(t)= [Apg/4D] 1/4 50-200 m of sediment cover in the area of the (3) where t is in seconds. The flexural wavenumber is (4) flexuralbulgeanda ridgeorientationnearlyparallel to the trench. The Peru-Chileridgeis mostlike the whereAp is the differencein densitybetween _ Woodlark basin in its orientationto the trench, and it alsohas sufficientsedimentcover (50-100 m) for a heat flow survey. Furthermore,due to somewhat is theacceleration dueto gravity,andD is theplate's flexuralrigidity,givenby: unusualcircumstances, areaswith roughlyequal sedimentcoveron the sameageof oceaniccrustcan be foundbothwithin andoutsidethe flexuralbulge seawater andmantle (assumed tobe2.3gm/cm3), g D = Y h3(t)/ 12(1-v2) (5) [Cande and Leslie, 1986]. Other areasof active ridge subductionhavethe wrongsediment distributionandridgeorientation(theExplorerridge [INPAC, 1985]), are undergoinga ridgejump (the eastPacific rise at 18ø N [Luhr et al., 1985]), or are perpendicular to thetrenchaxis(Ayu trough [Weissel and Anderson, 1978]). where Y is Young'smodulus(assumedto be 1.8 x 12 2, , 10 dynes/cm ) and v is Poissons ratio (assumed to be 0.25). The flexural wavenumberis therefore: k(t)= 5.4x 10-2t-3/8cm-1 AbbottandFisk: TectonicallyControlledRock Suites The flexure inducesstress,(5,in the lithosphere: (6) (5= (•+2g)zO2y/ Ox 2 1157 Anderson, R. N., M.D. Zoback, S. H. Hickman, andR. L. Newmark, Permeabilityversusdepth in the upperoceaniccrust:In situmeasurements in DSDP hole 504B, easternequatorialPacific, J. Geor)hvs.Res., 90, 3659-3669, 1985. Barazangi;M., andJ.Dorman, Worldseismicity where• andg are.tJ•eLam• colqstants (assumed •+2g = 2.17x 10•z dynes/cm z) andz is distance abovethe plate'sneutralsurface(thatis, middle). The maximum stress,which occursat z=h/2 and kx=•/4 is then: mapscompiledfrom ESSA, CoastandGeodetic Survey,epicenterdata, 1961-1967, Bull. S½ismol.Soc. Am,, 59, 369-380, 1969. Bodine, J. H., M. S. Steckler, and A. B. Watts, Observationsof flexureandtherheologyof the oceaniclithosphere,J. Geophys.Res., 86, 3695-3707, 1981. Oma x= 1.16tl/4bars (7) Boettcher,A. L., Volcanismandorogenicbelts--The origin of andesites,Tectonophysics, 17, 223-240, 1973. The stressdecreases linearly with depthin the lithosphere,while the strengthof therock increases linearlywith depth. Thusthe toppartof the 1/thosphere is crackeddown to somedepthwherethe stressand strengthare equal. Using the Bodine et al. [1981] estimateof 0.222 bars/mfor the strength, the depthof crackingis: Bratt, S. R., E. A. Bergman,and S.C. Solomon, Thermoelasticstress:How importantas a cause of earthquakes in youngoceaniclithosphere?, J. Geophys.Res., 90, 10,249-10,260, 1985. Byerly, G. R., W. G. Melson, andP. R. Vogt, Rhyodacites,andesites,ferrobasalts,and oceanictholeiitesfromtheGalapagos spreading center,Earth Planet. Sci. Lett., 30. 215-221, d= 3.7x 10-2t1/2{1-[14100t-1/4+1] -1} cm (8) Caldwell, J. G., andD. L. Turcotte,Dependenceof thethicknessof theelasticoceaniclithosphere on age, J. Geophys.Res,, 83, 7572-7579, 1976. Equations7 and 8 areplottedin figures4 and 5. 1979. Cameron, W. G., E.G. Nisbet, and V. J. Dietrich, Acknowledgments.We thankM. Petit, R. Barnes,B. Taylor, andH. Staudigelfor providing uswith preprintsof their unpublishedmanuscripts. We thank W. White, W. Menke, S. Hoffman, P. Hoffman, B.Taylor andN.Sleep for their critical comments.Acknowledgmentis madeto theDonors of the PetroleumResearchFund, administeredby theAmericanChemicalSocietyfor thesupportof Abbottduringthisresearch.Fisk was supportedby OCE84-12395. REFERENCES Abbott,D. H., andM. Lyle, Age of oceanicplatesat subductionandvolatilerecycling,Geophys. Res. Lett., 11,951-954, 1984. Abbott, D. H., W. H. Menke, M. Hobart, and R. N. Anderson,Evidencefor excesspore pressuresin southwestIndian Oceansediments, J. Geophys.Res., 86, 813-828, 1981. Anderson,R. N., M. G. Langseth,and J. G. Sclater, The mechanismsof heat transfer throughthe floor of the Indian ocean,J. Geophys.Res., 82, 3391-3409, 1977. Anderson, R. N., M. A. Hobart, and M. G. Langseth,Convectiveheattransferin the Indian Ocean, Science, 204, 828-832, 1979. Petrographic dissimilarities betweenophiolite andoceanfloor basalts,in Ophiolites,editedby A. Panayoitou,pp. 182-192, CyprusGeological SurveyDept., Nicosea,Cyprus,1979. Cande, S.C., and R. B. Leslie, Late Cenozoic tectonicsof the southernChile trench, J_ Geophys.Res,, 91, 471-496, 1986. CAYTROUGH, Geologicalandgeophysical investigationof the mid-CaymanRise spreading center: Initial results and observations,in Deer) Drilling Resultsin the AtlanticOcean:Ocean Crust,MauriceEwing Ser.,vol. 2, editedby M. Talwani, C. G. Harrison,andD. E. Hays,pp. 66-93, AGU, Washington,D.C., 1979. Chen,W. P., andP. Molnar, The depthdistribution of intercontinental andintraplateearthquakes and its implicationsfor the thermalandmechanical propertiesof the lithosphere,J. Geophys.Res., 88, 4183-4214, 1983. Clague,D. A., W. Friesen,P. Quinterno,M. Holmes, J. Morton, R. Bouse, L. Mortenson, andA. Davis, Preliminarygeological, geophysical,andbiologicaldatafrom the Gorda Ridge, U.S. Geol. Surv.OpenFile Rep. 84-364, 1984. Cocker, J. D., B. J. Griffin, and K. Muehlenbachs, Oxygenandhydrogenisotopeevidencefor seawater-hydrothermal alterationof the 1158 MacQuadeIslandophiolite,EarthPlanet.Sci. Lett., 61, 112-122, 1982. Coleman, P. J., and L. W. Kroenke, Subduction without volcanism in the Solomon islandsarc, Geomarine Lett., _1,129-134, 1981. Cooper,P. A., and B. Taylor, Polarityreversalin the Solomon Islands arc, Nature, 314, 428-430, 1985. Cousens,B. L., R. L. Chase,andJ. G. Schilling, Basaltgeochemistry of theExplorerRidgearea, northeastPacific Ocean, Can. J. Earth Sci,, 21, 157-170, 1984. Drummond, K. J., G. W. Moore, X. Golovchenko, R. L. Larson, W. C. Pitman, III, W. A. Rinehart, T. Simkin, and L. Siebert, Plate tectonicmap of the circum-Pacificregion, northeastquadrant,Circum-Pac.Counc.for Energy and Miner. Resour.,Am. Assoc.of Pet. Geol., Tulsa, Okla., 1981. Embley,R. W., M. A. Hobart,R. N. Anderson, AbbottandFisk: TectonicallyControlledRock Suites richteritc50tremolite .50 solidsolutionat high . pressuresandpossiblepresenceof sodium calcicamphiboleunderuppermantleconditions, Earth Planet. Sci. Lett., 18, 72-76, 1973. Hart, S. R., K, Rb, Cs, Sr, and Ba contentsand Sr isotoperatiosof oceanfloor basalts,Philos. Trans. R. Soc, London, $er. A., 268,573-587, 1971. Heezen, B.C., and D. J. Fomari, Pacific Ocean map, in Atlas Geologiquedu Monde, UNESCO, Paris, 1976. Hickey, R. L., andF. A. Frey, Geochemical characteristics of boninite series volcanics: Implicationsfor their source,Geochim. Cosmochim. Acta, 46, 2099-2115, 1982. Hobart, M. A., M. G. Langseth,andR. N. Anderson,A geothermalandgeophysical survey on the south flank of the Costa Rica rift: Sites 504 and505, Init. Rcpts.Deep SeaDrill. Pr0j., !•_, 379-404, 1985. hydrothermalcirculationin 80-m.y. crest,J. Geophys.. Res., 88, 1067-1074, 1983. INPAC, Workshopreport on drilling in the N.E. Pacific, Univ. of Wash., Seattle, 1985. Jarrard,R. D., Relationsamongsubduction parameters,Rev. Geophys.,24, 217-284, Fehn, U., K. E. Green, R. P. Von Herzen, and L. M. Cathles, Numerical models for the Johnson,R. W., R. P. Macnab, R. J. Arculus, R. and D. Abbott, Anomalous heat flow in the northwest Atlantic: A case for continued hydrothermalfield at the Galapagosspreading center,J. Geophys.Res.,88, 1033-1048, 1983. Fisk, M. R., Basaltinteractionwith harzburgiteand the formationof high-Mg andesites,Geophys. Res. Lett., 13,467-470, 1986. Forsythe,D.W., andB. Wilson,Three-dimensional temperature structureof a ridge-transform-ridge system,EarthPlanet.Sci. Lett,, 70, 355-362, 1984. Fox, P. J., and D. G. Gallo, A tectonic model for ridge-transform-ridge plateboundaries: Implicationsfor thestructure of theoceanic lithosphere,Tectonophysics, 104, 205-242, 1984. Fujita, K., andN.H. Sleep,Membranestresses nearmid-oceanridge-transform boundaries intersections, Tectonophysics, 50, 207-221, 1978. Gill, J., OrogenicAndesitesandPlate Tectonics, 390 pp., Springer-Verlag,New York, 1981. Green,D. H., Experimentalmeltingstudieson a modeluppermantlecomposition at high pressureunderwater-saturated and water-undersaturated conditions,Earth Planet. Sci. Lett., 19, 37-53, 1973. Gregory,R. T. andH. P. Taylor,An oxygen isotopeprofilein a sectionof Cretaceous oceanic ••l Samail ophiolite, Oman: Evidence buffering of the oceans by deep(>5for kin) seawater-hydrothermal circulationat mid-ocean ridges,J. Geophys.Res., 86, 2737-2755, 1981. Hariya,Y., andS. Terada,Stabilityof 1986. J. Rybum, R. J. S. Cooke,andB. W. Chappell,Bamusvolcano,PapuaNew Guinea: Dormantneighborof Ulawan, and magnesian-andesite locality,Geol.Rundsch,, 72, 207-237, 1983. Johnson,R. W., A. L. Jacques,C. H. Langmuir,, M. R. Perfit, H. Staudigel,P. N. Dunkley,B. W. Chappell,andS. R. Taylor, Igneous petrologyandgeochemistry of volcanic-rock dredgesamplesfrom theNew Georgiaforearc area:The magmaticeffectsof ridgesubduction, in Marine Geology. Geor>hvsics. and Geochemistry of•heWobdiark Basin-Solomon IslandsRegion,Circum-Pac.Counc.for Energy andMiner. Resour.,Earth Sci. Ser., in press, 1986. Kellogg,J. N., I. J. Ogujiofor,andD. R. Kansakar,Cenozoictectonicsof the Panamaand North Andes blocks, in Memoirs of the Sixth Latin AmericanCongresson Geology,vol. !, editedby J. Valdiri Wagner,pp. 40-59, Consejo Consultivo de Directores de Servicios Geologicosde Latinoamerica y el Cadbe, Bogota,Columbia,1985. Kushiro,I., Originof somemagmasin oceanicand circum-oceanic regions,T½ctonophysics, 17, 211-222, 1973. Lewis, B. T. R., Evolutionof oceancrustseismic velocities, Annu, Rev. Earth Planet. Sci., 6, 377-404, 1978. Lister,C. R. B., On thepenetrationof waterinto hot rock, Geophys,J. R. Astron.Soc.,39, 465-509, !974. Lister,C. R. B., The basicphysicsof water AbbottandFisk: TectonicallyControlledRock Suites penetrationinto hotrock,in Hydrothermal Processesat SeafloorSpreadingCenters,edited by P. A. Rona, K. Bostrom,L. Laubier, and K. L. Smith, Jr., pp. 141-168, Plenum,New York, 1983. Luhr, J. F., S. A. Nelson, J. F. Allan, and I. S. E. Carmichael,Activeriftingin southwestern Mexico: Manifestations of anincipienteastward spreading-ridge jump, Geology, 13, 54-75, 1985. Luptcn> J.E.,andH.Craig, Excess 3Heinoceanic basalts-Evidencefor terrestrialprimordial helium, Earth Planet. Sci. Lett., 26, 133, 1975. Lupton, J. E., G. P. Klinkhammer, W. L. Normark, R. Haymon, K. C. MacDonald, R. F. Weiss, andH. Craig, Helium-3 and manganeseat the 21o N EastPacificRise hydrothermalsite,Earth Planet.Sci. Lett., 50, 115-127, 1980. Martin, R. J., III, Time-dependent crackgrowthin quartzand its applicationto the creepof rocks, J. Geophys.Res., 77, 1406-1419, 1972. McBimey, A. R., Compositionalvariationsin calc-alkalinesuitesof Central America, in Proceedingsof the AndesiteConference,Bull. Oreg. Dep. Geol. Miner. Ind., 65, 185-189, 1969. McKenzie, D. P., Speculationson the causesand consequences of platemotions,.G_ eophys.J. R. Astron. Soc., 1_•8,1-32, 1969. McNutt, M. K., Lithosphericflexureandthermal anomalies,J. Geophys.Res., 8._•9, 11,18011,194, 1984. Miyashiro,A., The Troodosophiolitewasprobably formed in an island arc, Earth Planet. Sci. Lett., 1._•9, 218-224, 1973. Molnar,P., D. Freeman,andJ. Shih,Lengthsof intermediateanddeepseismiczonesand temperatures in downgoingslabsof lithosphere, Geophys.J. R. Astron.Soc.,4_!3, 763-774, 1979. Moores,E. M., andE. D. Jackson,Ophiolitesand oceanic crust, Nature, 250, 136-139, 1974. Moores,E. M., P. T. Robinson,J. Malpas,andC. Xenophonotos, Model for theoriginof the Troodosmassif,Cyprus,andothermideast ophiolites,Geology,1.•2,500-503, 1984. Nichols,I. A., andA. E. Ringwood,Effect of water on olivinestabilityin tholeiitesandthe productionof silica-saturated magmasin the island-arcenvironment,J. Geol., 8_21,285-300, 1973. Parsons,B. A., andJ. G. Sclater,An analysisof thevariationof oceanfloorbathymetry andheat flow with age,J. Geophys.Res., 82, 803-827, 1977. Perfit,M. R., C.H. Langmuir,M. Baekisapa,B. Chappell,R. W. Johnson, H. Staudigel,andS. R. Taylor, Geochemistryof volcanicrocksfrom theWoodlarkBasin:Addressing thequestions 1159 of ridge subductionanda reversalin the polarity of subduction, in Marine Geology. Geophysics. ,and Geochemistry of th• Woodl•kBasih-' SolomonIslandsRegion, Circum-Pac.Counc. for Energy andMiner. Resour.,Earth Sci. Ser., in press, 1986. Reagan,M. K., and A. Meijer, Geology and geochemistryof early arc-volcanicrocksfrom Guam, (3eol. Soc, Am. Bull., 95,701-713, 1984. Richardson, R. M., S.C. Solomon, and N.H. Sleep,Tectonicstressin the plates,Rcv, Geophys.,1_27, 981-1019, 1979. Riddihough,R. P., R. G. Currie, andR. D. Hyndman, The Dellwood knolls and their role in triplejunctiontectonicsoff northernVancouver Island, Can. J. Earth Sci., 1.27, 577-593, 1980. Robinson, P. T., W. G. Melson, T. O'Heam, and H. U. Schmincke,Volcanicglasscompositions of the Troodosophiolite,Cyprus,Geology, l_j_l, 400-404, 1983. Schilling,J. G., M. Zajac, R. Evans,T. Johnson, W. White, J. D. Devine, andR. Kingsley, Petrologic andgeochemical variations oalong the Mid-Atlantic Ridge from 29ø N to 73ø N, Am. J. Sci., 283,510-586, 1983. Sclater,J. G., R. L. Fisher,P. Pathat,C. Tapscott, andB. Parsons,Geophys.J. R. Astron.Soc., 64, 587-604, 1981. Schweller, W. J., and L. D. Kulm, Extensional raptureof oceaniccrestin the Chile trench,Mar. Geol., 2_•8, 271-291, 1978. Schweller, W. J., L. D. Kulm, and R. A. Prince, Tectonics,structure,andsedimentary framework of the Peru-Chile Trench, Mem. Geol. Soc. Am., 154, 323-350, 1981. Searle,R. C., The activepartof theCharlieGibbs fracturezone:A studyusingsonarandother geophysical techniques, J. Geophys.Res.,86, 243-262, 1981. Staudigel,H., M. McCulloch, A. Zindler, andM. Perfit, Complexridge subduction andislandarc magmatism:An isotopicstudyof theNew Georgiaforearcand the Woodlarkbasin,in Marine Geology,Geophysics,and Geochemistryof the Woodlark Basin-Solomon IslandsRegion,Circum-Pac.Counc.for Energy andMiner. Resour.,Earth Sci. Ser.,in press, 1986. Stefanson,V., Physicalenvironmentof hydrothermalsystemsin Icelandandon submergedoceanicridges,in Hydrothermal Pr•ocesses at SeafloorSpreadingCenters,edited by P. A. Rona, K. Bostrom,L. Laubier,andK. L. Smith,Jr., pp. 321-360, Plenum,New York, 1983. Sugi,N., andS. Uyeda,Subduction of young oceanicplateswithoutdeepfocusearthquakes, Bull. Soc. Geol. Ft., 26, 245-254, 1984. Taylor,B., A geophysical surveyof the 1160 Woodlark-Solomons region,in M.arine Geology,Geophysics,andGeochemistry of the Woodlark Basin-Solomon Islands Reeion. Circum-Pac.Counc.for EnergyandMiner. Resour.,Earth Sci. Ser., in press,1986. Taylor, B., andN. F. Exon, An investigationof ridge subduction in theWoodlark-Solomons region:Introductionandbackground, in Marine Geoloev. Geovhvsics. and Geochemistry of the W09d[•kBas•n-'S010mon Islands Regi6n, Circum-Pac.Counc.for EnergyandMiner. Resour.,Earth Sci. Ser., in press,1986. Verhoogen,J., F. J. Turner,L. E. Weiss,C. Wahrhaftig, andW. S. Fyfe, The Earth, 748 pp., Holt, Rinehart,andWinston,New York, 1970. Von Huene,R., andJ. Aubouin,Summary-Leg67, Middle America Trench transectoff Guatemala, Init. Repts.Deep SeaDrill. Proj., 67, 775-793, 1982. Watts, A. B., J. H. Bodine, and M. S. Steckler, Observations of flexure and state of stress in the AbbottandFisk: TectonicallyControlledRock Suites oceaniclithosphere,J. Geophys.Res.,85, 6369-6376, 1980. Weins,D. A., andS. Stein,Age dependence of oceanicintraplateseismicityandimplications for lithosphericevolution,J, Geophys.Res.,88, 6455-6468, 1983. Weissel, J. K., and R. N. Anderson,Is there a Carolineplate?,EarthPlanet.Sci. Lett., 41, 143-158, 1978. Weissel,J. K., B. Taylor, andG. D. Karner,The openingof the Woodlark basin,subduction of theWoodlarkspreadingsystem,andthe evolution of northern Melanesia since Mid-Pliocenetime, Tectonophysics, 87, 253-277, 1982. D. Abbott andM. Fisk, Collegeof Oceanography,OregonStateUniversity,Corvallis, OR, 97331. (ReceivedJune23, 1986; revisedAugust11, 1986; acceptedAugust11, 1986.)