Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Oceanic trench wikipedia , lookup
Northern Cordilleran Volcanic Province wikipedia , lookup
Geology of the Pyrenees wikipedia , lookup
Post-glacial rebound wikipedia , lookup
Mantle plume wikipedia , lookup
Izu-Bonin-Mariana Arc wikipedia , lookup
Great Lakes tectonic zone wikipedia , lookup
Abyssal plain wikipedia , lookup
Large igneous province wikipedia , lookup
Baltic Shield wikipedia , lookup
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. B6, PAGES 11,937-11,958, JUNE 10, 1994 A geological model for the structure of ridge segments in slow spreading ocean crust Brian E. Tucholke and Jian Lin Departmentof GeologyandGeophysics, WoodsHole Oceanographic Institution,WoodsHole, Massachusetts Abstract. First-order(transform)andsecond-order ridge-axisdiscontinuities createa fundamental segmentation of thelithosphere alongmid-oceanridges,andin slowspreadingcrustthey commonlyareassociated with exposureof subvolcanic crustanduppermantle. We analyzed availablemorphological,gravity,androck sampledatafrom the AtlanticOceanto determine whetherconsistent structural patems occurat thesediscontinuites andto constraintheprocesses thatcontrolthe patterns.The resultsshowthat alongtheir older,inside-comersides,both firstandsemond-order discontinuities arecharacterized by thinnedcrustand/ormantleexposures aswell asby irregularfault patternsanda paucityof volcanicfeatures.Cruston young,outside-comer sidesof discontinuities hasmorenormalthickness,regularfault patems, and commonvolcanic forms. Thesepatternsare consistent with tectonicth'mningof crustat insidecomersby lowangledetachmentfaultsaspreviouslysuggested for transformdiscontinuities by Dick et al. [1981]andKarson[1990]. Volcanicuppercrustaccretes in thehang'rag wall of thedetachment, is strippedfromthe inside-comer footwall,andis carriedto theoutsidecomer. Gravityandmorphologicaldatasuggestthatdetachment faultingis a relativelycontinuous, long-livedprocessin crustspread'rag at <25-30mm/yr,thatit maybe intermittent at intermediate ratesof 2540 mm/yr,andthatit is unlikelyto occurat fasterrates. Detachmentsurfaces aredissected by later, high-anglefaultsformedduringcrustaluplift 'intothe rift mountains;thesefaultscancut throughtheentirecrustandmaybe thekindsof faultsimagedby seismicreflectionproffi'mg overCretaceous NorthAtlanticcrust.Off-axisvariationsin gravityanomalies'indicatethatslow spreading crustexperiences cyclicmagmatic/amagmatic extension andthata typicalcycleis about2 m.y. long. During magmaticphasesthe footwallof the detachmentfault probably exposeslowercrustalgabbros,althoughtheserockslocallymay havean unconformable volcanic carapace.During amagmaficextensionthedetachment may dip steeplythroughthe crust, providinga mechanismwherebyuppermantleultramaficrockscanIx: exhumedveryrapidly, perhapsin aslittic as0.5 m.y. Togc•cr, detachment faultingandcyclicmagmatic/amagmatic extensioncreatestronglyheterogeneous lithospherebothalongandacrossisochronsin slow spreadingoceancrust. Introduction the discontinuities andthe spreadingaxesare common[Searle The axes of mid-oceanridges are offset by discontinuities and Laughton, 1977], and adjacentspreadingaxes often overwhich divide the oceancrust into spreadingsegments(Figure lap, with crustal ridges or septa cutting acrossthe discontinu1) [Macdonald et al., 1991]. First-order discontinuities in ities at high angles[Phillips and Fleming, 1977; Purdy et al., both slow and fast spreadingcrust are transformfaults; they 1990]. Off-axis, the traces of these discontinuities are modernormallyoffset the ridge axis by 30 km or more (>0.5-2 m.y. ately developedbathymetricvalleys which contain a series of age offset), are orientedin the directionof relative plate mo- crustaldepressions.The tracesalso show a variety of orientation, and exhibit deep transformvalleys. Beyond the ridge- tions which, although generally symmetrical about the rift transform intersections(RTIs), the fossil transform zones ocaxis, may not follow directions of relative plate motion. In cur as well defined fracture valleys that trace the history of relslow spreadingcrest the tracescan persistfor 10 m.y. or more [Tucholke and Schouten,1988; Sloan and Patriat, 1992]. ative plate motion for tens of millions of years [Menard and Atwater, 1968; Fox et al., 1969; Tucholke and Schouten, Taken together, first- and second-orderdiscontinuitiescre19881. ate a fundamentalalong-axis segmentationof slow spreading Second-orderdiscontinuitieshave smaller offsets, ranging ridges,roughly every 50 +30 km, which is expressedby largedown to <10 km. In fast spreadingcrustthey appearas over- scale crustal ridge-and-troughmorphology that trends across lapping spreadingcenters,but on slow spreadingridges they isochrons(Figure 1). Finer-scale,higher-ordersegmentation also may occur, perhapsbeing manifestedby small offsetsof are developedas deep bathymetricdepressions.Crustal morvolcanic ridges and by individual volcanoes within the rift phology in these depressionsdiffers significantly from that of normal transformfaults. Structuresoblique to trendsof both valley [Macdonald et al., 1991]. However, these small-scale features represent short-lived (< 10$years)geologic phenom- Copyright1994by the AmericanGeophysical Union. ena, and they are not known to create coherentmorphologic patternswhich can be tracedoff-axis. Only slow spreading Paper number94JB00338. ridge segmentsboundedby first- and second-orderdiscontinuities are further consideredin this paper. 0148-0227/94/94JB-00338505.00 11,937 11,938 TUCHOLKE AND LIN: GEOLOGICAL MODEL OF RIDGE SEGMENTS ., - \ Second-order Discontinuity oc oc Second-order Discontin, Transform Discontinuity Figure1. Sketch of segmented slowspreading ocean crustoffsetby right-stepping second-order disconti- nuities anda left-stepping transform discontinuity. IC, inside corner; OC,outside comer.IC crustispatterned.Thealong-isochron dimension of a spreading segment historically hasbeentermed"segment length".We usetheterm"segment ran"in reference to thetypically longer,cross-isochron dimension. The origin of ridge segmentationhas been attributedto the some instances"). In Mutter and Karson'smodel, mantle formationof spreadingcells above relatively uniformly upwellingaffectsthe thermalstructureof segments and thus spacedRayleigh-Taylor instabilities in the melt zone of the subaxialasthenosphere [Whiteheadet el., 1984; Crane, 1985; the local patternsof brittle/ductiledeformation,but it is a consequence, not a cause,of the plateboundaryconfiguration. Lin et el., 1990]. In thismodel,eachspreading segment is At the intersectionof a ridge axis with a first- or secondthoughtto mark the locusof a rising and meltingmantle order discontinuity,an inside corner (IC) is formed in the diapir. To the extentthat the meltinganomalies alongthe bightbetween the spreading axisandtheactivediscontinuity, plate boundaryare stablewith respectto a deeper,astheno- and an outside corner (OC) occurs on the other side of the spheric reference frame, the off-axis traces of nontransform spreading axis,nextto the inactivetraceof the discontinuity discontinuities may recordthe absolutemotionof the plate (Figure1). It hasbeenrecognized for a numberof yearsthat boundary [Schoutenet al., 1987; Tucholke and Schouten, 1988]. However,the along-axislengthof segments bounded by nontransform discontinuities can expandor shrinkon ~10 m.y. timescales [SloanandPatriat, 1992;Pockalnyand Gente, thereis an asymmetryin elevationbetweeninsideand outside corners in slowspreading crust[Collette,1986;Severinghaus andMacdonald, 1988].Theinsidecorners arehigher, andthey tendto create"transverse ridges"thatextendoff-axisalong the older sidesof discontinuities.Over the along-isochron 1992]. Thusobliqueoff-axistracesof thediscontinuities may reflect developmentor demiseof the spreadingsegments length of spreadingsegments,the surfaceof the oceancrust ratherthan plate motion. Furthermore, obliquespreading tendsto slopedownwardfrom insidecornerto outsidecorner, segmentshave beenobservedto migratein differentdirections provided that a segmenthas offsets of similar senseat both on a singleplateboundary [Miiller andRoest,1992],suggest- ends(segmentB, Figure 1). If the directionof offset reverses ing thatat leastsomeof themeltinganomalies arenotstably at the endof a segment(segmentC, Figure1), thenthe entire located. segmenton one side of the spreadingaxis is preferentially MutterandKarson[1992]presented an alternate hypothesis elevated,andon theothersideit is depressed. for theoriginof segmentation on slowspreading ridges.They Asymmetry in crustal compositionof inside and outside suggested thatsegmentation is mechanically controlled by the cornersalsohasbeenobserved,notablyat the easternRTI of locationsof detachmentfaults and interveningaccommoda- theKaneFractureZonein theNorthAtlantic[KarsonandDick, tion zones where extensional strain is transferred from one 1983, 1984]. The inside corner exposesplutonic (lower detachment faultto another.(Detachment faults,asdefinedby crustal, generally gabbroic) rocks and mantle ultramafics DavisandLister[1988,p. 136]have"lowangleof initialdip, (e.g., peridotites,serpentinites),whereasthe outsidecorneris subregional to regionalscaleof development, andlargetrans- uniformlycoveredby basalts.TheIC exposure of gabbros and lational displacements, certainly up to tens of kilometersin serpentinites hasbeenattributedby Dick et el. [1981],Brown 11,939 TUCHOLKEAND LIN: GEOLOGICALMODELOF RIDGESEGMENTS and Karson[1988], andKarson [1990] to the presence of a ß _ low-angle normalfault(detachment fault)thatdipsfromthe ß insidecornerbeneaththe spreadingaxis duringphasesof amagmatic extension.Moregenerally, theexposure of plutonic and ultramafic rocks is a well known phenomenonin numeroustransformdiscontinuitiesof both the Atlantic and - • _ A ß ø./ / ß A IndianOceans[e.g.,Bonatti,1968;EngelandFisher,1975; e•e _ Prinz et al., 1976]. The vastmajorityof theserockshasbeen recoveredfrom the active transformvalleys, both walls of whicharegenerated at insidecorners. ßß •ßß _ ß .• ß **/• The above observationsof crustal composition indicate thattherearesystematic geologic differences between IC and . - OC crust at first-orderdiscontinuities,and the observationsof IC crustalelevationfurthersuggest thatthesepatternsextend to second-order discontinuities.In this paper we systemati- / callyinvestigate bothfirst-andsecond-order discontinuities in slowspreading oceancrustby examining the following properties: crustal elevation, gravityfield,crustal composi- / _ '=• _/ tion,fine-scalevolcanicmorphology, andfault patterns.The resultsshowconsistent patternsfor both first- and second- Atlantic Pacific 0 order discontinuities that allow us to developa generalmodel 1 2 Rift ValleyRelief (km) for thegeologic structure of ridgesegments in slowspreading oceancrust. We also discussobservedeffectsof cyclicity in Figure2. Plotof asymmetry in inside/outside cornerelevafromSeveringhaus and magmatic andamagmatic crustal extension thatcreate off-axis tionversusrift valleyrelief,adapted errorbarsareshown.Trianvariabilityin theoceanic crustalrecord.We do notaddress Macdonald[1988]. Approximate relationsbetweensegmentation and the observedmagnetic glesdenote newdatafor 15 IC/OCpairs,derivedfromSea Ridge field in this paper. Theserelationsare complexbecause Beammapsof Purdyet al. [1990]for theMid-Atlantic crustalmagnetization is affectedby variations in tectonism axis between Kane and Atlantis fracture zones. andmagmatism andbecause therelative contributions of upper versuslowercrustalmagnetization to the totalmagneticfield are uncertain[e.g., Pariso and Johnson,1993]; a separate axis. Thus both a dynamicand a staticcomponentappearto manuscript [M.A.Tiveyet al., unpublished data,1993]will addressthis topic. Observations at First- and Second-Order Discontinuities Crustal Elevation contribute to creation and maintenanceof excesselevation at inside corners. Two kindsof dynamicuplift havebeenproposed to create highinsidecomers.In one,thedrivingforces responsible for formationof the median valley and flanking rift mountains upl•ifttheinsidecorner.IC upliftexceeds thatof OC crust becausethe mechanicalstrengthof the active discontinuity the insidecorneris lessthanthatof the inactive Severinghaus andMacdonald [1988]studied theasymmetry bordering traceof thediscontinuity [Kuoet al., 1984;Severinghaus and between crustal elevations at inside versus outside corners at 57 intersections of ridgeaxeswithfirst-andsecond-order discontinuitiesin the Atlantic and Pacific Oceans. They found that insidecornersare consistentlyhigher than outsidecor- Macdonald, 1988]. A secondexplanation,proposedby Bercoviciet al. [1992], is that the excesscrustalelevation maybe generated by viscoelastic rebound accompanying cor- beneathactivediscontinuities. nersin slowspreading crust.Asymmetry in elevation ranged nerflow of thelithosphere from about400 m to 2500 m andaveragedabout1000 m. No correlationwas observedbetween elevation asymmetryand spreading rateor magnitude of offset(ageor distance), but there is a correlationbetweenasymmetryand depthof rift valley (Figure 2). Severinghaus and Macdonald[1988] alsoobserved that inside cornerstend to lie above the normal crustal age-depth Isostaticeffects also have been invoked to explain both IC upliftandcontinued IC elevation awayfromtheridgeaxis. Collette [1986], for example,observedthat topographic depression of thevalleyin an activediscontinuity oftenis greater thanthatin theinactive trace.Whenisostatic rebound of thisdepressed crest occurs, rebound of thecoup16d IC crust ispredicted tobegreater thanrebound attheoutside comer [J. curvefor oceancrust[Parsonsand Sclater, 1977], and they Lin andJ.Y.Chen,Subsurface andsurface loadings on the 1986]andtheIndianOcean[Dicket al., 1991]showthatIC elevationof IC crusttendsto be lockedinto the lithosphereon the ridge flanks. at the northern Mid-AtlanticRidge,subsuggested thatwithina fewmillionyearsthecrustsubsides to oceaniclithosphere Research Letters,1993]. IC elevation normaldepths.Thisobservation wouldindicate thatdynamic mittedto Geophysical by serpentinization of ultramarie rocksat upliftcontributes to theelevation of insidecomers.Thereare alsomaybecaused discontinuities; serpentinization reducesthe bulk densityof few detailedoff-axis recordsof crustaldepth availableto evalthelithosphere andmayleadto diapirism or isostatic uplift uatethisidea. Data from the KaneFractureZone [Tucholkeand onceIC crestspreads Schouten,1988] suggest thatIC cruststartswell abovethe [Bonatti,1976]. In anyof thesecases, pasttheactiveoffset,it becomes better coupled toOCcruston age-depth curveandinitiallysubsides at faster-than-normal sideof thediscontinuity. Thusexcess relative rates. However,otherdatafrom the NorthAtlantic [Collette, theopposite crustretains anomalouselevationfor tens of million years off 11,940 Gravity TUCHOLKE AND LIN: GEOLOGICAL Field A number of detailed surveys have acquired gravity data along the crest of the Mid-Atlantic Ridge [Kuo and Forsyth, 1988; Linet al., 1990; Blackman and Forsyth, 1991; Morris and Derrick, 1991]. These data are closely spacedand are accompanied by detailed bathymetric mapping so that the gravity values can be corrected for three-dimensional(3-D) variations in seafloor topography. In the North Atlantic, Lin eta/. [ 1990] reportedthe resultsof a survey(2-5 km line spacing) that coveredfive completespreadingsegmentsand their bounding discontinuitiesbetween 28øN and 30ø30'N; five of the discontinuities are second-order, and one is the fransform of the Ariantis Fracture Zone (Plate l a). The second-orderdis- continuitiesare expressedoff-axis by bathymetricvalleys that form irregular discordantzones. The Atlantis Fracture Zone exhibits a deep, well defined, and relatively linear trough at the fransform and in the adjacent inactive fracture zone valleys. To reveal density variationsbeneaththe seafloor,the gravity data were reduced to residual gravity anomaly (Plate lb) [Kuo and Forsyth, 1988; Linet a/., 1990]. The residualgravity anomaly was calculated by removing from the observed free-air anomaly the 3-D effects of seafloortopographyand the crust-mantle interface (assuming uniform, 6-km-thick crus0, as well as the effects of lithosphericcooling basedon the 3-D passive-upwelliffg model of Phipps Morgan and Forsyth [1988]. Seawater, crustal, and mantle densities of 1.03,2.7, and3.3 Mg m'3, respectively, wereassumed. The residual gravity anomaly shows clear banding across the ridge axis, and this banding correlatesclosely with ridge segmentation. Inside corners at second-orderdiscontinuities consistentlyexhibit residual gravity highs of 20-40 regal (red arrows), and anomalies at inside corners of the Atlantis frans- form reach 40-50 regal. The margins of thesegravity highs commonly extend beneath the axes of the boundingdiscontinuities, and they locally impinge on the flanks of adjacent outside corners. In contrast, OC crust and segment centers have a much reducedresidual gravity anomaly,typically 5-25 mGal. The resultingasymmetryin IC versusOC residualgravity anomaly, whether compared acrossthe spreadingaxis or only within a segmenton one side of the axis, averages10-20 regal and locally reaches35 mGal. This asymmetryskewsthe gravity anomalies clockwise from a ridge-orthogonaldirection in locations where discontinuities offset the rift valley right-laterally (red arrows, Plate lb). The pattern of gravity highs on IC crust and gravity lows on OC crust continuesoffaxis to the limits of the survey, notably along the Atlantis FractureZone (FZ). A similar pattern of residual gravity anomaliesis observed in the vicinity of the TAG area on the Mid-Atlantic Ridge axis at 25ø to 27øN (Plate 2 and Figure 3). In this area, threecom- MODEL OF RIDGE SEGMENTS plete ridge segmentsand their boundingsecond-orderdiscontinuitieswere surveyedwith a narrowbeamecho-sounding system and gravimeter at a line spacing averaging about 10 km [Rona and Gray, 1980; Zervaseta/., 1990]. The skewedbanding of gravity anomalies acrossdiscontinuitiesat the ridge axis is clearly expressed,and bandsof residual gravity highs and lows follow IC and OC crust,respectively,for at least 100 km off-axis. Segmentcenters,like OC crust, tend to have low residual gravity anomalies. IC and OC residual gravity anomalyvaluestypically differ by 15-25 mGal (Figure 3). The bands of gravity anomalies also show significant internal variability along the run of a segment,with local, closed,relative highs and lows that differ by about5-15 mGal. Asymmetries in IC/OC residual gravity anomalies also appearat other locationsin the Atlantic Ocean where gravity and bathymetryhave beenmappedin detail. In the MARK area south of the eastern RTI of Kane Fracture Zone [Morris and Detrick, 1991], the gravity field is asymmetricbut the asymmetry is reducedcomparedto the examplescited above;the IC residual gravity anomaly averagesabout 10 mGal higher than that over OC crust near the RT!. Stronger asymmetry is observed at a second-orderdiscontinuity farther south in the MARK area (23ø15'N), whereresidualgravity anomaliesover IC crustexceedthoseover OC crustby 10-20 mGal. In residual gravity maps of the South Atlantic [Kuo and Forsyth, 1988], two transform faults and three second-order discontinuities also show similar patternsof asymmetry;the IC residualgravity anomaliesvary widely but range 10-50 mGal higher than their OC counterparts. With the exception of a second-order discontinuityat 33ø35'S, however, the IC/OC gravity asymmetry generally is less pronouncedthan in the North Atlantic. The observed variations in residual gravity anomaly are thought to result from variations in density (i.e., temperature, composition)of the upper mantle, variations in thicknessor density of the ocean crust, or some combinationof both [Kuo and Forsyth, 1988; Lin et al., 1990]. Theoretical modeling which is consistentwith available data suggeststhat the contribution of upper mantle thermal structure to the gravity anomaliesis very limited. Lin and Phipps Morgan [1992] showedthat if mantle upwelling is driven only by plate separation, then large gravity anomalies should be limited to large-offset transforms (>2 m.y. offset); at smaller, secondorder discontinuities(<1 m.y. offset) only small anomaliesare predicted.Sparkset al. [1993] extendedthe passive-upwelling model to include buoyancy-drivenmantle upwelling. They showedthatwith focusedupwelling,along-axischangesin the gravity field are strongly dominated by increased crustal thicknesswhich results from enhancedmelting beneath segment midpoints, rather than by mantle-densityvariations. The asymmetry of the gravity field across the ridge axis also is difficult to explain only by variationsin uppermantle Plate 1. (a) Bathymetry of the Mid-Atlantic Ridge axis near Atlantis Fracture Zone (FZ) in the North Atlantic Ocean,from Purdy eta/. [1990]. Bold linesmark therift axis, anddashedlinesfollow axesof maximum depth(AMD) alongtracesof second-order discontinuities; the AMD is an objectiveapproximationof the former positionof the plate boundary. (b) Residualgravity anomaliesfor the sameregion, with ridge axis and discontinuitieslocated as in (a). Note residualgravity highs at insidecorners(arrows). (c) Relative crustal thickness calculated from residual gravity anomalies. Note thin crust beneath inside corners (arrows). (d) Sketchmap of ridge segmentationand major fault zones. Long, relatively linear fault zones commonlycover the full along-isochronlengthof segments,and more irregularfault patternstend to occur at inside corners,particularly along Atlantis FZ. TUCHOLKEAND LIN: GEOLOGICAL MODELOFRIDGESEGMENTS 44oW 43ow 42øW 44øW 11,941 43ow 42oW S • ATLANTIS F.Z. ,,, .. • 28øN RESIDUALGRAVITYANOMALY BATHYMETRY 1 lll ! (MGAL) 44oW 43ow I I ', 44øW 42oW ATLANTIS' -' 0 0 0 0 43øW ' (. ,, 28 øN CRUSTAL THICKNESS 1 (KM) • • • o • • TECTONIC PATTERNS 42øW 11,942 46øW TUCHOLKE AND LIN: GEOLOGICAL 45ow 44øW MODEL OF RIDGE SEGMENTS 46øW 45øW 44øW e?øN B A s 26øN 25øN BATHYMETRY RESIDUAL GRAVITY ANOMALY ß (M)• o o Plate 2. (a) Bathymetry of the Mid-Atlantic Ridge axis in the vicinity of the TAG area [from Rona and Gray, 1980]. The ridge axis and discontinuitiesare marked as in Plate 1. (b) Residualgravity anomaliesof the sameregion,markedas in Plate I [Zervaset al., 1990]. Note the bandsof residualgravityhighs(arrows) and lows extendingoff-axis along IC and OC crust,respectively,and the variation of anomalyvalues within each of the bands. Fine white lines locate axes of gravity highsand lows profiled in Figure 3. G and S identify dredgedgabbrosand serpentinites[Tiezzi and Scott, 1980]. density. If the gravity anomaliesare attributedto temperature or compositionaldifferencesin the mantle, then some process is required in the subaxial asthenosphereto partition cooler, denser upper mantle to inside corners and hotter, less dense upper mantle to outsidecornerson the oppositeside of the rift axis. We know of no reasonable physical mechanism to achieve such partitioning. The strong asymmetry in IC/OC residual gravity anomaly thus appears best to be explained by differences in crustal thickness and/or density. It seems unlikely that crustal density can be the sole control; it is iraplausible that a mechanism exists to generate IC and OC crust with similar thicknesses,but with differing densities, in the absenceof crustal thinning. We therefore hypothesize that the high residual gravity values at inside cornersare causedby crustal thinning and consequent elevation of the Moho; accompanying increasesin average crustal density are permissible but are not required. Relative crustal thicknessesalong the Mid-Atlantic Ridge near the Atlantis FZ were calculated using a referencecrustal thickness of 6 km and a crust/mantle density contrast of 0.6 Mg m'3 (Platelc) [Lin et al., 1990]. The totalrangeof calculated crustal thicknesses is 6 km, with minimum values con- sistently occurring at IC crust. IC crust averages2 to 3 km thinner than crust at adjacent outside corners, both at the Atlantis transform the south. and at the second-order However, relative discontinuities inside and outside corners at the Ariantis FZ are a kilometer more less than those around to crustal thicknesses of both the second-order or discontinuities. The thermaledgeeffect of this large transform[Fox and Gallo, 1984] may have reduced the magma budget and the overall crustal thickness compared to the smaller-offset discontinuities farther south. As noted earlier, the residualgravity highs observedalong IC crust commonly extend well beneaththe axes of the adjacent first- and second-orderdiscontinuities,implying that the discontinuities themselves are floored by thin ocean crust (Plate l c). This observationis consistentwith previousgravity studiesof both transformvalleys and second-orderdiscontinuities which concluded that crust in discontinuities is anomalously thin [Louden and Forsyth, 1982; Prince and Forsyth, 1988; lfuo and Forsyth, 1988; Lin et al., 1990; Morris and Detrick, 1991; Blackman and Forsyth, 1991]. It also agrees with seismic refraction experiments which have shown that discontinuities typically have reduced crustal thicknessesand/or anomalousseismic velocities [e.g., White et at, 1984; Cormier et al., 1984;Ambosand Hussong,1986; Minshull et al., 1991; Detrick et al., 1993]. Thin crust TUCHOLKE AND LIN: GEOLOGICAL MODEL OF RIDGE SEGMENTS l1,943 beneathridge axis discontinuitiesis generally attributed to diminishedvolcanismcausedeither by hydraulic head loss as 0(2 IC E 2:6 ø 11 'N meltsintrudealong-strikeaway from magmaticloci near segment centers[Whiteheadet al., 1984;Macdonald, 1986] or by acceleratedheat loss to the cold truncatinglithosphereat the offset [Fox et al., 1980; Fox and Gallo, 1984]. Crustal Composition 26ø03'N The inference from gravity data that crust at inside corners and in troughsof the adjacentdiscontinuitiesis thinner than OC crustcan be examinedin the resultsof seafloorsampling throughoutthe Atlantic Ocean. A thinner, and possibly , œ 25ø49'N denser,crustal section shouldresult in increasedexposureof plutonic and ultramafic rocks at inside corners. OC seafloor, which from gravity data appearsto have a more normalcrustal 25ø42'N thickness,should be dominated by basaltic rocks and should only rarely exposeplutonic or ultramafic sections. To examine the distribution of rock types in relation to E 25ø34'N IC/OC tectonicsetting,we compiledthe resultsof dredge,drill site, and submersiblesample recoveries at 655 stations reported in the literature and in the dredgecollectionof Woods I [ I I I I [ I Hole OceanographicInstitution (Table 1). The recoveries - 100 - 50 0 50 100 were divided into three categories:(1) Stations recovering only a lower crustal to upper mantle suite of plutonic and Distance (krn) ultramafic rocks; the rocks are primarily gabbros and periFigure 3. Profiles of residual gravity anomaly along the dotites (peridotites typically 10-90% serpentinized);(2) staaxes of gravity highs over IC crust and the axes of gravity tions recovering only an upper crustal suite of volcanic and lows overconjugateOC crustin the TAG areaof Plate2. Latiintrusiveigneousrocks, predominantlybasaltsand subsidiary tude labels show where each profile crossesthe Mid-Atlantic diabases;and (3) stations recovering a mixed suite of these Ridge axis (vertical line). Note that all profiles are oriented rocks. We noted only the presenceor absenceof each rock with OC cruston the left. Residualgravity valuesof OC crust type at the stations;data generallywere not availableto quanare routinely lower than IC values, suggestingconsistently tify relative weights or volumes of rock types. These three thinner crust along inside corners. categorieswere further subdivided accordingto the tectonic settingfrom which the sampleswere recovered: """'" . Table 1. Distributionof RockTypesin Dredge,Drill Site, and SubmersibleSampleRecoveriesFrom Slow Spreading Crust of the Atlantic Ocean Tectonic Ultramafic/Plutonic Setting Inside corners Ultramafic/Plutonic RocksOnly B as alt/Di abase + Basalt/Diabase 142 91 125 (38 P, 80 U, 24 UP) (16 PB, 1 PD, 15 PBD, (116 B, 1 D, 8 BD) 17 LIB, 3 LIBD, 27 UPB, 1 UPD, 11 UPBD) Outside coruers Undetermined Totals Only 11 12 109 (8 U, 3 UP) (10 LIB, 2 UPB) (105 B, 2 D, 2 BD) 21 12 58 (2 P, 7 U, 12 UP) (5 PB, 1 PBD, 5 UPB, 1 UPBD) (55 B, 1 D, 2 BD) 358 IC recoveries, including 233 recoveries of U/P rocks(65%) 132 OC recoveries, including23 recoveries of U/P rocks(17%) 91 undetermined recoveries,including 33 recoveries of UfP rocks (36%) Special tectonic setting 23 (4 P, 17 U, 2 UP) 7 (2 PB, 2 LIB, 1 UBD, 2 UPB) 44 Totals 197 recoveries of 122 recoveries of U/P + B/D 336 recoveriesof only B/D rocks (51% of only UfP rocks (30% of total) rocks (19% of total) 74 STS recoveries, (44 B) total) including 30 recoveries of U/P rocks (41%) 655 total recoveries, including319 recoveries of U/P rocks (49%) Compilationexcludesrecoveriesfrom seamountsand from drill siteson >20 Ma crust,exceptingplutonic and ullxamaficrocks recoveredin specialtectonicsettings(STS). U, ultramafic;P, plutonic;B, basalt;D, diabase. Referencesusedfor the compilation are availableon requestfrom the seniorauthor. 11,944 TUCHOLKE AND LIN: GEOLOGICAL 1. The inside corner group includescrustcurrentlygenerated at inside cornersalong the ridge axis, as well as IC crust which extends off-axis along the run of a spreading segment (i.e., along the old side of the ridge axis discontinuity;Figure 1). The IC categoryis restrictedto crust within 30 km of the axis of maximum depth (AMD) of the bounding discontinuity (approximately half the along-isochronlength of an average spreading segment). For segmentslonger than 60 kin, we includedsamples>30 km from the AMD but within the IC half of the segment. 2. The outsidecorner group includesOC crust both adjacent to the ridge axis and extendingoff-axis along the run of a segment (i.e., on the young side of the ridge-axis discontinuity). 3. Samples of unknown morphotectonic affinity were recovered where existing bathymetric data are inadequate to define sample location with respect to traces of ridge axis discontinuities. Most samplesare from regions where secondorder discontinuities occur. The small offsets and intermittent bathymetric valleys of these discontinuitiesseldom are clearly defined in any bathymetric maps except those surveyed with multibeam bathymetric systems. We also assignedplutonic and ultramaficrocks from a number of sample sites to the category of "special tectonic setting". In these environments the tectonism responsible for exposureof the rocks is unrelated to original ridge segmentation. Most of these samplescome from Kings Trough, which is thoughtto be a failed rift [Srivastavaet al., 1990]; from the Galicia Bank margin, where peridotitesbecameexposedduring continental breakup [Boillot et al., 1989]; from Gorringe Bank, which is a compressionalplate boundary[Klitgord and Schouten, 1986]; and from the Puerto Rico Trench or the south wall of Cayman Trough. Stations recovering plutonic and ultramafic rocks in the Atlantic and Caribbean are shown in Figure 4. The great majority of the samples come from IC crust. However, this reflects sampling bias toward the walls of active discontinuities, both walls of which are inside corners. To correct for this bias, we examinedoccurrenceof plutonic/ultramaficrocks versus occurrenceof only upper crustal rocks at both inside and outside corners(Table 1). At inside corners,two thirds of the total recoveriesincluded plutonic and/or ultramafic rocks; in this group, ultramafic rocks were recovered slightly more often than plutonic rocks (163 versus 133 occurrences). At outside corners, in contrast,only 17% of 132 samplerecoveries contain plutonic/ultramafic rocks; ultramafic samples are present in all of this group (23 occurrences), and plutonic rocks are relatively uncommon(five occurrences). From these data it is clear that lower crustal and upper mantle lithologies are sampledmuch more consistentlyon IC crust than they are on OC crust. Although there is uncertaintyaboutthe representativenessof recovery by various sampling tools, particularly dredges,it is clear that uppermantle rocks are widely exposed at inside corners. We interpret this to mean that the crustal sections there are thinned, in accordance with the gravity observations noted above. The common occurrence of lower crustal rocks also suggeststhat the averageIC crustaldensity is higher than that of outside corners. OC crust is dominated by upper crustal lithologies, indicating that relatively thicker and more continuous volcanic sections are present in these settings. This is consistentwith the presenceof thicker crust inferred from gravity data. It is noteworthy that at least 58% of the samplerecoveries from inside corners included upper crustal basaltsor diabases MODEL OF RIDGE SEGMENTS (92% at outside corners). These percentagesare minima; the publishedpapersfrom which the data were derived do not necessarily note corecovery of volcanic or intrusive rocks if deeper crustal and mantle rocks were the subject of study. Conversely, some published papers dealing with the basaltic crust also fail to report coincident recovery of plutonic or ultramafic rocks. These latter rocks, however, generally have been noted in other publications,and any effects causedby their under reporting probably is negligible. Despite these uncertainties,we can safely state that it is not uncommonfor an upper crustalcarapaceto be presenton IC crust. The common occurrenceof lower crustal and upper mantle lithologies, however, indicates that this carapace must be thin and/or intermittent. An example of the differentiation of rock types between inside and outside corners is shown in Figure 5 for the area around Kane FZ. The easternridge-transformintersectionhas been extensivelystudiedby dredging,submersibledives, and bottom photography[Karson and Dick, 1983; 1984; Mevel et al., 1991]. These studies have documentedthat lower crustal gabbros cropoutoveran areaof at least50 km2 alongtheeast face of the IC high betweenabout2500 m and 6000 m depth. Parts of the crest and west flank of the IC high have basalts exposedat the seafloor, as do the rift valley axis and the outside corner. A well defined neovolcanic basalt ridge with active hydrothermalvents is developedin the center of the rift valley [Karson et al., 1987; Gente et al., 1991]. At the south end of this ridge segment, serpentiniteshave been dredged [Gente et al., 1989] and sampled by submersible[Karson et al., 1987; Mevel et al., 1991] from the outside corner (west wall of rift valley); this is a clear exceptionto the observation that outside cornersrarely expose deeper lithologies. Ocean Drilling Program (ODP) Site 670 was drilled on the inside corner of the next segmentto the south(Figure 5), and it recovered serpentinites and serpentinized harzburgites [ShipboardScientificParty, 1988]. Deep Sea Drilling Project (DSDP) Site 395, to the southwest,also recovered serpentinites and gabbros interbedded between basalts at-160-170 m subbottom[Shipboard Scientific Party, 1979]. This site was drilled in a depressionthat can now be recognizedas part of an east-westtrending bathymetricvalley marking the trace of a former second-orderdiscontinuity. The discontinuity becameindistinctat about 5 Ma, and it is not readily recognizable at the presentridge crest. Offset of magneticanomalies immediatelynorth and east of Site 395 [Hussonget al., 1979] suggeststhat the site was drilled on IC crust. At the western ridge-transform intersection of Kane FZ, gabbros have been reported from submersibledives on the inside corner [Zonenshainet al., 1989] and serpentiniteshave been dredgedthere [B. Tucholkeand H. Dick, unpub. data]. The axis of the rift valley contains a neovolcanic ridge [Karson and Dick, 1983; 1984; Zonenshain et al., 1989] from which extensivedredging has recoveredbasalts. No samples have beenreportedfrom the outsidecomer. Gabbrosand serpentinites have been dredged from the walls on the old, IC sides of the Kane fracture valley both within the transform region and along the inactive fracture valley (Figure 5) [Miyashiro et al., 1969; Dick, 1989]. Fine-Scale Hundreds Volcanic of volcanic Morphology cones have been identified in multi- beam bathymetricdata along the axis of the Mid-Atlantic rift valley betweenthe Kane and Atlantisfracturezones[Smithand 11,945 TUCHOLKEAND LIN: GEOLOGICALMODELOF RIDGESEGMENTS 90ø $0ø 70ø 60ø 50ø 40ø 30ø 20ø I0ø 60 ø 0ø I0ø 20 ø 30 ø ,. •ALIClA glNGœ T•OUGH ß 30 ø 20 ø VœMA F.. Z. IOø o CORNERS CORNERS PLUTONIC IOø 20 ø ULTRAMAFIC o o /x ß ,0, ß "' PLUTONIC PLUS ULTt?AMA FIC PLUTONIC + ULTt?AMAFIC IN SPECIAL TECTONIC SET TING •10GRAND• 30o[ 40 ø - 60 ø Figure 4.Occurrence ofplutonic andultramafic rocks indredge, drillsite, andsubmersible sample stations fromtheAtlantic Ocean.Symbols show composition andtectonic setting; all samples illustrated arefrom <20Macrust, except those noted forspecial tectonic settings. Compiled fromthepublished literature and thedredge collection of WoodsHoleOceanographic Institution. Cann, 1992],andtheyform a moreor lesswell definedneo- well understood,but recent, detailed near-bottomsurveyssugvolcanoes maysurvive faulting well volcaniczonein nearlyeveryspreading segmentsurveyed. gestthatmanyof these toberecognized off-axis [Smith andCann,1993]. This observation suggests that a neovolcanic zoneis a com- enough Multibeam bathymetry andsidescansonarrecords provide monfeaturealongmostsegments of slowspreading ridges. evidence that the volcanic upper crust is better preserved on Theprocesses attending dissection andtransport of thisvol- caniccarapace intotherift mountains andridgeflanksarenot outside corners than on inside corners. Figure 6 illustrates 11,946 TUCHOLKE AND LIN: GEOLOGICAL 46øW MODEL OF RIDGE SEGMENTS 45 ø 44 ø 24 ø N 23 ø Figure 5. Structurecontourson basementaroundKane FZ [from Tucholkeand Schouten,1988]. Contour interval is 250 m; depths>4000m are shaded. Symbolsshowrecoveriesof plutonicand ultramaficrocks as in Figure4; recoveriesof only basaltsare indicatedby solid squares.DSDP andODP drill sitesare numbered. Second-orderdiscontinuitiesand ridge axis are markedas in Plate 1. 25 ø 05' N 45ø40'W 45ø30 ' 45ø20 ' 25 ø 00' 24 ø 55' Figure 6. Multibeam bathymetry of an inside/outsidecorner pair next to a second-orderdiscontinuityat the Mid-Atlantic Ridge axis. Contourinterval is 100 m; depths>3500 m are shaded. Triangleslocaterelatively circular volcanic conesidentified in 20-m contourdata, basedlargely on work by Smith and Cann [1992, also unpublisheddata, 1992]. Heavy lines are boundaryfaults demarcatingthe inner floor of the rift valley, and the dashedline marks the AMD of the discontinuity. Note that IC crustexhibitsblocky morphologyand uncommonvolcaniccones;OC crustis more lineated,with commonvolcaniccones. TUCHOLKEAND LIN: GEOLOGICALMODEL OF RIDGE SEGMENTS 11,947 volcanicfeatures,althoughcommonelongationof conespar- this observationfor multibeambathymetricdata in the North Atlantic. The OC crustis dominatedby ridge-paralleltopog- allel to the faultssuggests thattheirshapehasbeencontrolled or modifiedby faulting. IC crusthas a distinctlydifferent raphyandfaults,andit exhibits numerous conical features that appearto be small volcanoes.Someof theseconesare bounded by steep,linearslopes whichparalleltheridgeaxis andmaybe normal-fault zones.Thevolcanoes alsopopulate therift valley[Smithand Cann,1992],andtheyarecommon withinthebathymetric depression thatdefinesthe discontinuity between spreading segments. TheIC crusthaslesslineated topography andshowsonly uncommon occurrences of vol- character;it shows dominantly blocky, irregularly faulted topography with onlyrareindications of volcanoes or hummocky volcanicmorphology. The few volcanicfeatures observedon IC crusttypicallyoccurdeepwithin the fracture valley,closeto the axisof the discontinuity. Fault Patterns Detailed studiesof fault patternsby submersible observation and in bottomphotographs showconsistent differences Side scansonardata show a similar patternin distribution in the natureof faultingat insideandoutsidecorners[Karson and Dick, 1983, 1984;OTTER, 1984]. Directobservations of of volcanicfeatures,as illustratedfor the Kane FZ in Figure7. canic cones. Most of the coneson IC crust occur within 3-5 km of the axis of the discontinuity. Along this transformdiscontinuity the OC crustexhibits the basalticcrust at outsidecornersindicate that faults tend to ø) withthrowsusuallylessthana few tensof abundantevidenceof hummockyvolcanicterrainandvolcanic be steep(700-90 meters,and the faults locally groupto form fault zonesof cones. The volcanoesrangein diameterfrom about0.5 to average slope(•30ø). Mostof thefaultsfaceinward nearly2 km, andtheyvaryfromsingle,isolated conesto clus- moderate ters. There are no clear-cutexamplesof faults incisingthe towardtherift valleyaxisandstrikeparallelto it; backtilting 47o45,W I / , ,.o •x I" •. , :::?::::?:::::::::::::::::::::::::::::::: ...... '.... i •-•• • •- ;:•E:•::..,• -•-'•-; ••, ß :•:•:•:•:•:• .... • •' ':.. •/ , ...v.....:.:.:.:.:.:, x :-:-:.:-:.. ..... '"................ I [ • / ' :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: •% • • '...................................................... ==================================================== ................... '"[h................. [?•:• 5 30' :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: •. .•.::•::..•t • [ • .... ::•::•i•::.. •• • I , ( J • I ........................ •:•:•::.• ,• ., 44o45'W .... :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: • - ,, I ....... •:•:•:•:•:•:•:•:•:•:•:• ................ ::::::::::::::::::::::::::: • • • 44ø30 ' 44ø15 ' Figure7. Tracings of faults andvolcanic features observed in SeaMARC II sidescan sonar images along KaneFZ. Shaded areashavenodata.(top)Western limb,-8-12 Ma crust(young, OC side).Axisof fracture valleyis indicated bydashed line. OCcrust tothenorth exhibits hummocky volcanic morphology, volcaniccones, andcoherently oriented faultsthatparallel theridgeaxisandcurvetoward theKanetransform. IC crustto the southhasfew volcanicfeaturesandshowsmoreirregularfaultpatterns.(bottom)Eastern limb,-2-6 Ma crust(young, OC side);OC andIC crustexhibitthesamekindsof features asthewestern limb,butwithpositions of IC andOC crustreversed. TUCHOLKE 11,948 AND LIN: GEOLOGICAL MODEL OF RIDGE SEGMENTS of the crestsof fault blocksby 5ø-15ø away from the rift valley along strike; thesefaults are developedalmostentirely in the is common. In contrast, faults on inside comers occur in two subvolcanic rangesof dip values[Karson and Dick, 1983; 1984]. One suite The samekinds of observationsappearconsistently,but at larger scales,in long-rangeside scan sonarimagesand multibeam bathymetricmaps of Atlantic Ocean crust(Figure 7 and 8). The slopemap in Figure8 is a usefulproxyfor understanding fault distribution,with certain limitations. Comparison betweenfaults mappedfrom submersibledives aroundthe Kane of faults dips at low to moderateangles(20ø-60ø),with dip directionstoward the rift axis and locally toward the ridgetransform intersection or the transform fault zone. These faults are intersectedby more steeplydipping (700-90ø) faults with throws of a few tens of meters to 150 m or more. Most of the s.teepfaults dip towardthe rift valley, the RTI, or the transform valley, and their orientationoften is highly irregular 43øW crust. FractureZoneandmultibeamslopemapsof the sameareasuggeststhat the steeperslopes(>_30 ø) are createdmostly by sets 42øW 30øN 25øN 46øW 45ow Figure 8. Maps of seafloorslope(top) aroundthe Atlantis FZ and (bottom) at a spreadingsegmentfarther south,derivedfrom 200-m griddedSeaBeambathymetryof Purdyeta/. [ 1990]. Gray-scaleintensityis proportional to slope value; steepestslopes(black) are greater than 30ø. Bold line marks the axis of the rift valley. Note zonesof steep,irregular slope(taken to be an indicationof irregularfaulting) on IC crust. Finer-scale,linear tracesof lesserslopesare interpretedas coherentlyorientedzonesof smaller faults that crosssegmentcentersand outside corners. TUCHOLKE AND LIN: GEOLOGICAL of closely spaced, steep faults and not by individual, lowerangle faults. Furthermore, orientations of individual faults in local, submersibleobservationsoften are significantly different from orientations suggestedfrom the slope maps, although average fault orientationsare comparable. As might be expected,then, the slope maps filter out details of individual faults, but they provide a relatively accuratedepictionof overall fault-zone structure at scales of hundreds of meters or more. At thesescales,the slopemapsclearly show significantdifferencesin fault patternsbetweenIC and OC crust (Figure 8). These differences are strongestadjacentto transformdiscontinuities like those at the Atlantis and Kane Fracture Zones, and MODEL OF RIDGE SEGMENTS l1,949 However, all the available data on seafloormorphology,composition, and gravity signaturediscussedabove suggestthat there is consistentcrustal thinning and exposureof plutonic and ultramafic rocks at inside comers of both first- and second- order discontinuities. We present here a general model of detachment faulting in slow spreading ocean crust that explains available observationsat these discontinuities. In the proposed model, a low-angle normal (detachment) fault dips below the rift axis from the surfaceof each inside corner(Plate3) [Tucholkeand Lin, 1992]. Thus,wherea ridge segment is bounded by discontinuities with consistent sense of offset, there is cross-rift switch of detachment footwalls and they are less obvious but still apparentat second-orderdiscontinuities. OC fault zones have generally lower slope values hanging walls over the along-axis length of the segment. Where a segment is bounded by discontinuitiesof opposite offset (segmentC, Figure 1), the entire segmenton one side of and narrower the rift is an IC footwall widths than fault zones on inside corners. Fault zones on outside cornersalso strike parallel to the rift axis in long, relatively linear trends, although they often curve in the directionof the ridge axis near the ends of segments(Figure 7 and 8). At insidecorners,the averagetrendof fault zonesparallels the rift axis, but secondaryorientationsparallel some ridge axis discontinuities,particularly at transforms. In addition, IC fault zonescommonlyexhibit irregularcuspate,arcuate, and convoluted patterns. The similar fault characteristics observed in fine-scale data (submersible,photographs)and inferred from large-scaledata (multibeam slope maps, side scan sonar) indicate that the latter usefully characterizefault patternsin the crust. Furthermore, thesefault patternsappearto correlatewith the natureof the crust being faulted (volcanic versus subvolcanic [Karson and Dick, 1983]). Thus, to a first order, the fault patterns within spreadingridge segmentssuggestwidespreadoccurrences of subvolcanic crust at inside corners, in contrast to occurrence of volcanic crust on outside corners. This is con- sistentwith, and extendsthe resultsof, the gravity, sampling, and fine-scale morphologystudiesnoted above. Geological Model of Ridge Segments First-Order Considerations The observation of low-angle normal-fault surfaces at the eastern inside corner of the Kane FZ, combined with informa- tion aboutexposuresof lower crustaland uppermantlerocksat inside comers of other North Atlantic transform faults, led to the hypothesisthat extensionon the low-angle faults could be responsiblefor exhumingplutonic and ultramafic rocks [Dick et al., 1981]. Brown and Karson [1988] and Karson [1990] developed this concept, and they noted a strong structural analogybetweenobservationsof low-angle normal faulting in slow spreadingocean crust and the developmentof large-scale detachmentfaults that exposemetamorphiccore complexesin the Basin and Range of the western United States [e.g., Wernicke, 1981; Davis and Lister, 1988]. The essenceof the oceanic detachment fault model is that OC crust in the rift val- ley constitutesa hanging wall (above the detachmentsurface) that is coupled acrossthe adjacent,inactive segmentboundary to IC crust in the neighboringsegment(Plate 3). IC crust in the rift valley forms the footwall block below the detachment fault, andit hasdecoupledboundariesat the spreadingaxis and along the active discontinuity. To date, developmentof this hypothesishas been based on limited examples of first-order (transform) discontinuities. block beneath a detachment fault sur- face, and the opposite side is entirely an OC hanging wall block. The detachment fault dips beneath the intrusive and volcanic carapace that constitutesthe neovolcanic zone, and it bottoms out in a zone of distributed shear beneath the brit- tle/ductile transition. Near a segmentcenter, where melt supply and thermal gradient are thoughtto be higher than at segment edges [e.g., Fox et al., 1980; Macdonald, 1986], the detachment probably soles out in a shear zone at a shallower subsurfacelevel. In addition, axial seafloor at segmentcenters tends to be shallower than at segmentends, giving the rift an hourglassshape [Phillips and Fleming, 1977], and rift-mountain seafloor at segmentcentersusually is deeper than that at inside corners. These topographiceffects, together with the shallower brittle/ductile transition, reduce the dip of a detachment fault near a segmentcentercomparedto its dip at an adjacent inside corner. Over the along-isochronlength of any segment,extensionat the inside corner is taken up largely by low-angle detachment faulting, but extension at the outside corner is accommodatedonly by higher-angle faults. For any segment having consistent sense-of-offset at its bounding discontinuities, the segment center is the location where the IC-to-OC, (Plate 3). detachment to no-detachment transition occurs As a result of this fault geometry, the basaltic neovolcanic zone forms the hanging wall above the detachment,and it is transportedprimarily to the outside corner. The IC footwall, stripped of the basaltic crustal carapace, exposes gabbroic lower crust. Toward the segment center, where extension is taken up increasingly by high-angle faults and less by detachment faulting, the stripping mechanism is less effective. Segment centers probably also accrete thicker crust because of increased melt supply [Fox et al., 1980; Macdonald, 1986; Linet al., 1990]. For these two reasons, crust extending off-axis along the middle of segmentsshould exhibit a fairly thick and complete volcanic section with comparatively few seafloor exposuresof lower crustal rocks (Plate 3). Conceptually the detachment-faultingmodel has two endmembers. In one, the detachment fault at each inside corner is constantlyactive and long-lived. As a result, the detachment surface would be continuous over the full inside comer run of a ridge segment. For a spreadingsegmentthat has existed from the time of initial continental separation,the breakaway zone for the detachmentwould be at the continent-oceanboundary in the continental margin. In the other end-member, detachment faulting is an episodicphenomenonand thusmay or may •,95o TUCHOLKE AND LIN: GEOLOGICAL MODEL OF RIDGE SEGMENTS I TUCHOLKE AND LIN: GEOLOGICAL not be currently observed at any given spreading segment [Brown and Karson, 1988; Karson, 1990; Dick et al., 1991]. The existing gravity data suggestthat detachmentfaulting can be relatively continuous. Residual gravity profiles over conjugate IC and OC partsof spreadingsegmentsin the TAG survey (Plate 2) show that gravity values are consistentlyhigher over IC crustthan over OC crust(Figure 3). This implies that the IC crust is systematicallythinner and therefore that the thinning mechanism(detachmentfaulting) has been continuously operative. Second-Order Effects Two second-ordereffectsmodify the detachmentmodel outlined above. First, it is apparent from the common recovery of volcanic rocks on inside corners(Table 1) that strippingof upper crust to the outside corner is an imperfect process. It may not be uncommon for short, high-angle faults in the young hanging wall volcanics to fail more easily than the underlyingdetachmentfault. Volcanic parcels overlying the detachmentsurfacethus could be rafted episodically onto the inside corner as klippen. Such rafting may be most pronouncedduring periodsof peak magmaticactivity when newly emplacedvolcanicslap farthestonto the detachmentfootwall. However, becausedetachmentfaulting is thoughtto be a relatively continuousprocess,we expectthat the basalticcarapace will lie unconformablyon the detachmentsurface. The contact may be either tectonic (active faulting between hangingwall and footwall rocks) or depositional(flows on top of the once exposedfault surface). We interpretthe high-anglenormal faults (-700-90 ø) which dissectthe original detachment(Plate 3) as another secondary effect. These faults are generallycolinearwith the high-angle faults observedat segmentcentersand at outsidecorners;thus they can extend with remarkablecontinuity acrossnearly the entire along-isochronlength of a segment(Figure 8). The faults are developed primarily in responseto uplift of axial crust into the rift valley walls and adjacent rift mountains [e.g., Macdonald, 1986], and they developsignificantthrows beginning<0.5 m.y. off-axis. On inside cornersthesefaults curve into the active discontinuities and exhibit orientations obliqueor even normalto the adjacentridge axis; however,the faults appear to have only dip-slip movement which accommodatesuplift of IC crust with respectto crust in the discontinuity [OTTER, 1984;Fox and Gallo, 1986]. Cye!lelty In Magmatle/Amagmatle Extension It is now generally recognizedthat there is no steady state magma chamberbeneathslow spreadingridges and that magmatism must be an intermittentphenomenon[e.g., Derrick et al., 1990]. However, the timescales and relative volumes of melt involved in magmaticphasesare poorly understood. On the shortest(smallest) scale, individual basalt flows are separated in time by thousandsof years [e.g., Ballard and van Andel, 1977]. At the 100 kyr to 1 m.y. timescale,Pockalny et al. [1988] have suggested that individualabyssalhills may be formed by magmaticpulsesseparatedby periodsof extension. This sub-million-year alternation between phases of essentially amagmaticor weakly magmaticextensionand phasesof magmatically robust extension has also been suggestedto explain the variable exposureof basaltic, plutonic, and ultramafic rocks on the walls of rift valleys [e.g., Cannat, 1993]. Such cyclicity will have a profoundeffect on the structureand composition of the ocean crust along the run of spreading segments. MODEL OF RIDGE SEGMENTS 11,951 Residual gravity anomalies (Plate 2b) show significant variability within the bands of gravity highs and lows that extend off-axis along IC and OC crust, respectively. This variability appears as closed relative highs and lows. It rangesfrom 5 to 15 regal (rarely 20 regal) over IC crust and slightly less (typically -10 mGal) over OC crust in the TAG gravity data. The variation is quasi-periodic, with wavelengths of the order of 2.0-J:0.4m.y. [Tucholke and Lin, 1992]. Longer off-axis gravity records,out to -28 Ma, show similar wavelengthsbut also suggestadditional,longer-wavelength variability with periods up to -9 m.y. [Lin et al., 1993]. We interpret this variability to representchangesin crustal thicknessand density. Assuming standardcrustal den- sitiesof 2.7 Mg m'3, the amplitudes of thicknessvariations are up to about2 km in IC crustandtypicallyabout1 km in OC crust. The primary mechanism controlling this thickness variation is thought to be cyclicity of magmatic and amagmatic extension, superimposedon detachmentfaulting. A schematicevolution of detachment faulting over a full magmatic/amagmaticcycle is illustrated in Figure 9 for an IC/OC pair. In Figure 9a the spreadingsegmentis in a magmatic phase. A magma body may or may not be presentin the position of this cross section near the segmentedge, but the thermal budgetof the segmentis high, the brittle/ductiletransition is relatively elevated beneath the spreading axis, and the detachment fault soles out at a shallow level. Volcanic rocksof the neovolcaniczone are depositedunconformablyon the detachment surface; they are extended in synthetic and antithetic faults, and part of this hanging wall may subsequentlybe rafted as klippen onto the insidecorner. Figure 9b representsthe early part of amagmaticor weakly magmatic extension following the magmatic phase. As the crust cools and isothermsare depressed,the detachmentfault extendsthrough the crust following a shear zone beneath the brittle/ductile transition. The toe of the fault probably will dip significantly, following the steepest thermal gradient beneath the spreading axis. Thus within a relatively short time, deep-crustalrocks are exhumedin the IC footwall. As amagmaticextensioncontinues(Figure 9c), ultramafic rocks becomewidely uplifted and exposedon the insidecomer. Late in the amagmaticphase, crust in the OC hanging wall also becomes highly extended. Faults there may provide deep pathwaysfor seawaterpenetrationinto the uppermantle, facilitating serpentinizationand serpentinitediapirism which can carry ultramaficrocks and allochthonouslower crustalrocks to the seafloor [e.g., Bonatti, 1976; Francis, 1981]. Serpentinization of the footwall mantle (not shown in Figure 9) probably also occurs, but it may be less intense becausethe footwall is not as stronglyextendedand faulted. At the beginningof a new magmaticphase(Figure 9d), the thermal budget increases,the brittle/ductile transition rises, and the detachment fault again soles out at a shallow level. Generation of a lower crustal gabbroic section may recommence, with a normal crustal section being developed at the outside corner. Discussion Structure and Stratigraphy of Inside Corner Crust The detachmentmodel, and the cyclicity of magmaticversus amagmatic extension, have important implications for the structure and stratigraphy of inside corner crust on slow spreadingridges. During a phaseof magmaticextensionthe 11,952 TUCHOLKE OUTSIDE AND LIN: GEOLOGICAL MODEL CORNER OF RIDGE INSIDE SEGMENTS CORNER Detachment Basalt Di,kes J J ..... •:;i•:;i•i::i::i:;::i i:;ii::ili::i:.:,•i::i:;::! •......... ::::::::::::::::::::::::::::::::::::::::::::::::: -- A. Magmatic Surface Neovolcanic Zone =============================================== ':::•i::i::iii•::i::•iii:i?:!:;•ii:.'•'•::•,.,• •" /%. / orittl•_ •, Melt Zone •'uetile .... ......... ,-" ' Gabbro Successor Klippe Faults / • B. Amagmatic " Serpentinites ................... ,•................. ....... ................ •.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:...:.:.:.:.:.:.:.:.:•:.:.:.:•.:•:.:.:.:•.:.:.:.:.:.:.:.:.:.:.::.•:.:.:.:.:.:.:.:.:.:.:.:.:.:.:...:.:.:.:.:.:.:.:.:.:.:.:.:. :.:.:.:.:.:.:.:.:.,•:.:.:.E:•:E:•:•:•:E:•:•.•:•:E: :•:::•::::•:::::•:•:•:•:•:-,•:•:•:•-•:•::.•:•:•:•:•:•:•::::: .•::::::•:•::::•:•:E:•:•:•:•:•:•::-:--..':•:•:•:'-•:•:• .. "::::•:• ........... ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::i:?::::::;::•:::: i:•::::.:.. ".'.'." '""•:::::::.'"-•-•.•.:-:.:.:.".'.'.•:.:::::•:•::-'•3: !::.::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: •i.".' .... i':M"!"'"'i!::ii!!!!!::!!i::!::!ii::!::i;.•iii!!::::./'.:i:: ß ...... Neovolcanic Zone Figure 9. Cross section acrossrift axis through an inside/outsidecorner pair, showing schematicevolution of a detachmentfault througha cycle of magmatic/amagmatic extension. Sectionextendsto crest of rift mountains~2 m.y. to either side of rift axis. H is hangingwall above detachmentfault, and F is footwall. See text for discussion. detachmentfault presumablysolesout near the top of gabbroic crust that is forming around the perimeterof the subaxialmelt zone. It also generallyunderliessheeteddikes (and the overlying volcanics), to the extent that the dikes are formed by intrusion of basaltic melts into brittle upper crust. This is consistent with sample recoveries (Table 1) which suggest that diabase dikes are relatively uncommon on IC crust (40 occurrencesin 358 recoveries, or 11%). Thus the normal IC crustal section generated during magmatic extension ideally consistsof gabbroic rocks that immediately underlie the detachment surface. As magmatic extension wanes and the detachment dives into the lithosphere, successivelydeeper lithofacies become exposed in the footwall (e.g., cumulate ultramafics followed by residual ultramafics of the upper mantle). These rocks should exhibit a wide variety of compositions(dunites, lherzolites, etc.) and structures(e.g., mylonites, cataclasites)that representall the complexitiesof subaxial anatexisand deformation. An attractive feature of the diving-detachmentconcept is that only a shortperiodof time is requiredfor the fault to reach and exhume upper mantle rocks. The dip angle of the toe of the detachmentfault during these episodesis unknown. With extreme (vertical) dip, deep structurallevels could rise in the footwall at a rate equal to the spreadinghalf rate (~10 to 17 km/m.y. in the Atlantic). This value would be reducedto about 7 to 12 km/m.y. for a dip of 45ø. Even at suchreducedangles, significantexposureof upper mantle rocks could occur in 0.3- TUCHOLKE AND LIN: GEOLOGICAL MODEL OF RIDGE SEGMENTS 0.6 m.y., assumingan average IC crustal thicknessof 4 kin. An importantimplication of a steeplydipping detachmenttoe is that originally horizontal lower crustal and upper mantle structures, where preserved during exhumation along the detachment, will be rotated toward the vertical as the footwall is exposed;these structuresshould intersect the detachment surfaceat high angles. Detailed near-bottomstudiesof lower crustal and upper mantle exposureson inside-cornercrest can test these postulatedrelations. The footwall exhumation described above is consistent with 11,953 diapirism, late within a phase of amagmaticextension. It is possible that under such conditions, low-angle faults may develop above shallow shear zones within the serpentinites. With renewedmagmatism,OC crustcould developa complex record of serpentiniteswhich are intruded by dikes [e.g., Mevel et al., 1991] and unconformablycappedby volcanics. High-Angle Faults The long, relatively linear faults that often extend over the along-isochronlength of spreading segments(Figure 8) appear to form during uplift of rift valley crustinto the bordering rift mountains. Most of thesefaults are steeplydipping (70 ø90ø) at the seafloor, and on inside cornersthey are interpreted as successorfaults that dissectthe detachmentsurface. Dips of the high-anglefaults at depth are uncertain. The fault blocks typically are rotated, as indicated by backtilting of block crests by 5ø-15ø, and this rotation must be accomplishedby flattening of the zone of failure at depth. Both listric faults (curvilinear,concave-upward)and planar faults which intersect a deep, more level zone of sheardeformation,meet this criterion. Seismic reflection profiles oriented along plate flow lines over CretaceousAtlantic crust show apparentfault surfaces of both listric and planar aspect which dip toward the the deformationhistory that is commonly observedfor ultramafic tectonites in the slow spreading Atlantic and Indian Oceans [e.g., Aumento and Loubat, 1971' Bassias and Triboulet, 1992; Cannat et al., 1992]. These rocks typically have an initial history of low-stress,high-temperatureplastic deformationthat probablyoccurredat depthsof 10 km or more in the mantle. They subsequentlyrecord hydration (e.g., amphibolization, serpentinization) and cataclasis, mylonitization, or shearing under progressively lower-temperature, higher-stressconditions. We attribute this later history to .•hc•allng, cooling, and de.fc•rrnatlc•n of the. IC footwall Seawaterpenetrating the rocks through fractures and faults appearsto be the primary sourceof water for the hydration reactionsthat form serpentinites[Bonatti et al., 1984]. In a succeedingphaseof magmatic extension,the exhumed ridge axis [e.g., Whiteet al., 1990;Morris et al., 1993]. Both kinds of faults commonly sole out in banded, subhorizontal lower crustal or upper mantle rocks in the rift axis should be zones of high reflectivity in the lower crust or upper mantle. intrudedby gabbrosand basalticmelts. Ultramafic rocks cut These basal zones have been variously interpretedas sills or by gabbroicdikeletshave beenrecoveredfrom slow spreading variationsin fluid content[White eta/., 1990], as relic magma crust in both the Atlantic and Indian Oceans [Mevel et al., chambers [Mutter et al., 1985], or as zones of deformation at 1991; Cannat et al., 1992; Fisher et al., 1986], and they may the basesof fault blocks [Mutter and Karson, 1992]. Whatever be examplesof such intrusiverelations. If rejuvenatedmagtheir origin, they must have accommodatedsome shear strain matism is so robust that a magma chamber forms, gabbros if the intersecting,dipping reflectionsare indeed faults. may replaceand unconformably underlieuppermantlefootwall We assumethat the crustal faults sole out just beneath the rocks (Figure 9d). brittle/ductile transition. The depth of this transitionis controlled by the thermal state of the crust [e.g., Chen and Structure and Stratigraphy of Outside Corner Morgan, 1990], and it consequentlywill vary with positionin Crust a spreadingsegment. Two factorssuggestthat isotherms,and therefore the depth of the brittle/ductile transition, may be During phasesof magmaticextension,OC crustdevelopsa elevated at inside corners compared to outside corners: (1) "normal" crustalsequenceconsisting(from bottom to top) of Becauseof thermal edge effects, isothermsare skewed where cumulateultramafics,gabbros,sheeteddiabasedikes, and pillow basalts. We suggestthat during phases of amagmatic they crossridge axis discontinuities,and IC crust is warmer than OC crust [Phipps Morgan and Forsyth, 1988; Lin and extensionthis sectionis thinnedin two ways (Figure 9). One is extensiontaken up by high-angleOC faults. Movementon Phipps Morgan, 1992; Sparkset al., 1993]. (2) Detachment faulting exhumesdeep, warm lithosphereat insidecorners. In these faults is unlikely to thin the crust significantly or to addition, significant asymmetriesmay exist in hydrothermal causemuch exposureof deepcrustallevels, exceptperhapslate in a long phase of amagmaticextension. Second,the crust circulationthroughIC and OC crust. If, for example,the upper will be effectively thinned and intruded by serpentinite crustal volcanics and sheeted dikes of outside comers are more highly fractured,faulted, and fissuredand thusmore permeable diapirs. Thinningby thesemechanisms shouldresultin OC seafloor than the plutonic/ultramaficrocks at inside corners,then cooloutcropsprimarily of volcanic rocks, with local occurrences ing of OC crust by hydrothermal circulation would be enhanced. Unfortunately, little is known about the bulk perof serpentinizedultramafics. This is essentially what we observe in sampling results onoutside corners (Table1). Plu- meability of OC and IC crust, so the effects of hydrothermal circulationcurrently are unresolved. tonic rocks and diabaseshave rarely been recovered,and we Elevation of the brittle/ductile transition beneath inside infer that they are only infrequentlyexposedon steep fault corners also may be enhancedby the occurrenceof serpenscarpsor carried to the surface as allochthonousblocks by serpentinite diapirism. Serpentinized ultramafic rocks, tinites. Faults penetratingthe thinnedIC crustprovide potential pathwaysfor seawaterto reach the upper mantle and to althoughuncommon,have been sampledmore frequentlyon form serpentinitesin hydrationreactions. Geologicallyrapid OC crust, and there are some well documented OC areas where serpentinizationunder these conditionsoccursat temperatures suchrocks are widely exposed. One is in the MARK area at 23ø20'N (Figure 5). Othersoccur at the outsidecornersof the from about 100ø to 500øC [O'Hara, 1967; Macdonald and Fyfe, 1985]. Serpentinization could facilitatedevelopmentof weak, Fifteen-Twenty FractureZone [Caseyet al., 1992; H. Dick et al., unpublishedmanuscript,1990]. Our model suggeststhat shallow zones of shear strain that take up the motion of northese rocks have been emplaced primarily by serpentinite mal faults in the overlying britfie crust. 11,954 TUCHOLKE AND LIN: GEOLOGICAL MODEL OF RIDGE 71o SEGMENTS 70 ø 69ow 0C EAST FLOWLINE PROFILE 711 WEST 5-] ISOCHRON PROFILE SOUTH 699 NORTH Figure 10. Block diagram (top left, looking away from rift axis) showingpossibleconfigurationof highanglefaults that extendover the full along-isochron lengthof a singlespreadingsegment;at insidecorners thesefaults dissectthe detachmentsurface(tick pattern). The faults soleout at shallowlevels on insidecorners and near the baseof full thicknesscrust at outsidecorners. Fault surfacesdip from IC to OC along isochronlines, and along flow lines they dip toward the ridge axis. Compare with similar geometry of apparentfaults in multichannelseismicreflection profiles over Cretaceouscrust in the western North Atlantic (bottom and fight, tracedfrom profiles by White et al. [1990]). On isochronprofile 699, the positions of inside and outsidecornersare labeled,basedon offsetsin M seriesmagneticanomaliesmappedby Minshull et al. [1991](inset). The questionedsegmentboundaryis inferredon the basisof seismicstructure and has no apparentoffset in magneticanomalies;it may be a zero-offsetdiscontinuity[Schouten and White, 1980]. BSFZ is Blake Spur FractureZone. In flow line profile 711, note that faults usually are not imagedin the uppercrust. From all these considerations, tle/ductile transition at inside we infer corners that the brit- is elevated relative to that at outsidecorners. Outsidecorner crustprobablyhas relatively normal thickness,with the brittle/ductile transition in the lower crust [Chen and Morgan, 1990]. Thus, over the along-isochron length of a spreading segment the brittle/ductile transition will deepen from inside corner to outside corner. A suite of high-anglefaults which fit this patternand which flatten with approachto the brittle/ductile transition is illustrated schematicallyin Figure 10. Imaged along an isochron line, the faults dip gently from shallow levels at the inside corner to near the base of the crust at the outside corner. Imaged across isochrons, the faults dip somewhat more steeply toward the ridge axis. This picture is in general agreementwith dip direction,dip magnitude,and lateral extent of apparentfaults in seismicreflection profiles acrossCretaceous crust in the western North Atlantic (Figure 10). In isochron-parallel profilesthe faultsdip at about20ø towardOC crust, and togetherwith their apparentbasal shearzones,they extend over nearly the full lengthsof segments. Faults facing the ridge axis have a steeperaveragedip where they are imaged in the central and lower crust (200-40 ø [Morris et al., 1993]), and they typically can be tracedfor 10 or 12 km. These faults normally are not imaged in the upper part of the crust, possibly becauseof steep dips, distributeddeformation in heterogeneous rocks, or masking by hydrothermalalteration [White et al. , 1990]. Mutter and Karson [1992] proposeda different scenariofor fault patterns within a slow spreading segment. They envisioned oppositely dipping detachment faults (across-rift) much as we described above, but with the detachments con- nectedthrough arcuatefaulting which defines an accommodation zone and creates oblique structuresat the segmentcenter. In their geometry the detachment surfaces have alongisochron dips from segment centers toward inside corners, together with complex strain in the central accommodation zone. This geometry is difficult to reconcile with dips of along-isochronfaults toward outside corners(Figure 10, bottom) and with the commonly observedcontinuity and linearity of fault zones that pass segment centers in rift valley walls (Figure 8). It is possiblethat some of the high-angle successorfaults on inside corners may develop into low-angle faults akin to detachmentfaults (Plate 3). Their developmentcould be facilitated by the presenceof relatively shallow serpentinites,and TUCHOLKE AND IzIN: GEOLOGICAL MODEL OF RIDGE sitional tratedby Brownand Karson[1988],Karson[1990],andDick outside corners of ridge segments in slow spreading ocean crust. These differences exist at both first-order (transform) and second-order (nontransform) discontinuities. Crust at inside corners appears to be tectonically thinned, and it consistsof a complex of lower crustal and upper mantle rocks that is capped by a thin and intermittent carapaceof basaltic intrusive and volcanicrocks. On average,OC crustprobablyhas a more normal thickness (-6 kin) and a relatively complete crustal stratigraphy. The characterof crustal faulting and the development or lack of volcanic morphology are quite sensitive to IC or OC tectonic setting; these characteristicstherefore can be useful for interpretingsenseof offset in ridge flank crust where classical indicators such as magnetic anomalies may be ambiguous. We have outlined a model of low-angle, normal detachment faulting wherein inside cornerscomprisefootwalls and outside corners represent hanging walls. Available gravity data, which show footwall crust to be consistently thinner than hanging wall crust, suggestthat the detachmentfaulting is a continuous or nearly continuous process in slow spreading crust (<-25-30 mm/yr full rate). In these environmentsthe basaltic carapace on IC crust, where present, is expected to occur as klippen in tectonic or depositional contact with an underlyingdetachmentsurface. At faster spreadingrates of 2540 mm/yr, but where a rift valley is still developed, the detachmentprocessmay become intermittent, with inside cor- inside corners. At both inside and outsidecomers, high-angle faults tend to curve near ridge-axis discontinuities(Figures 7 and 8). This curvatureis thoughtto reflect rotation of the directionof maximum tensile stressat the offsets [e.g., Searle and Laughton, 1977; Fox and Gallo, 1984]. However, significantadditional irregularity of fault patterns is observed on IC crust. This irregularitymay be a consequence of differencesin IC versus OC composition. Inside corner lithosphereis an aggregateof variably deformedand intrudedlower crustaland uppermantle rocks, and patternsof failure in this medium probably will be complex. Crust at segment centers and outside corners, in contrast, is thought to have more uniform composition, and its failure in relatively coherentfault patternsmay reflect this greater uniformity. Relation of Structure to Spreading Rate The occurrenceof detachmentfaulting and the cyclicity of magmatic/amagmaticcrustal accretionmust vary with rate of seafloor spreading;the primary indicators of crustal thinning (plutonic/ultramaficexposures,gravity banding, morphological variations)observedin slow spreadingcrust in the Atlantic and Indian Oceans are strongly subduedor wholly absent on fast spreadingridges suchas thosein the Pacific [e.g., Fox and Gallo, 1989; Madsen et al., 1990]. To a first order, we infer that detachment faulting is restricted to slower spreading ridges where a rift valley is developed;this is suggestedby the correlation between rift valley depth and the degree of asymroetry in IC/OC crustal elevations,which would appear to be linked to the detachmentprocess(Figure 2 [Severinghausand Macdonald, 1988]). Rift valleys normally are presenton midoceanridges having full spreadingrates less than about 25-30 mm/yr and are absent at rates above about 40 mm/yr [Macdonald, 1986]. At rates in the 25-40 mm/yr range, the development of detachments may be intermittent, perhaps varying inversely with the intensity and duration of phasesof magmatic extension. In such cases,detachmentfaulting could be very similar in characterto the models proposedby Karson [1990] and Dick etal. [1991]. The limited developmentof IC/OC differences in relative crustal thickness in the South Atlantic (spreadingrate 33 mm/yr [Kuo and Forsyth, 1988]) could be the expression of such intermittency. At slower spreadingrates, the prevalence of long-lived master detachments is expectedto increase, and banded, flow line patterns of crustal thickness variation should be developed more consistently. Conclusions Data from a wide variety of sources,including multibeam bathymetry, side scan sonar, crustal sampling, and gravity field, indicatethat there are fundamentalstructuraland compo- between 11,955 their intermittencyof occurrencewould be similar to that illus- et al. [1991]. This processof incisementwould transferrocks that were formerly in the footwall of the main detachmentinto the hanging wall of the new fault [Davis and Lister, 1988]. Strictly speaking,it may not be appropriateto call these features detachment faults, since they are of limited extent and have small translational displacements[cf. Davis and Lister, 1988]. In our model, these incised low-angle faults are considered to be subsidiaryto the master detachmentas a mechanism for exposing lower crustal and upper mantle rocks at differences SEGMENTS the crustal sections at inside and ners locally exhibitinga completecrustalsectionand no detachmentfaulting. We suggestthat detachmentfaulting is unlikely to occur in crust spreading at rates greater than 40 mm/yr where no rift valley is present. The masterdetachmentfaults at insidecornersare incisedby high-anglesuccessorfaults that are createdby uplift of rift valley crust into the bordering rift mountains. The locations and orientations of these faults are controlled both by the local stressfields along the rift axes and at the ridge-discontinuity intersections and by variable lithospheric composition created by cyclicity in crustal accretion. Consequently,the structure and igneous stratigraphyof inside cornersis expectedto be complex. The high-angle successorfaults at times may also develop into local, low-angle normal faults above a shallow brittle/ductile transition at inside corners. This incise- ment, as well as possible subsequentexcisement (transfer of hangingwall rocks to the footwall of a new detachment[Davis and Lister, 1988]), will further complicatethe structuralframework. The high-anglefaults associatedwith uplift into the rift valley walls commonly extend linearly and quasi-continuously over the complete along-isoehronlengths of spreading segments. The faults most likely have decreasingdips with depth in the crust and sole out in a zone of shear deformationjust below the brittle/ductile transition; this zone may be relatively shallow beneath inside corners but it deepensto near the base of normal-thickness crust beneath outside corners. We suggestthat it is these faults, and not the master detachment faults, which have been imaged by reflection profiling in Cretaceous crust of the western North Atlantic Ocean. Cycles of magmatic/amagmatie extension strongly modulate the composition, thickness, and structural framework of both IC and OC crust. Available residual gravity data over both inside and outside corners suggestthat a full cycle of magmatie/amagmatieextension may average about 2 m.y. in duration but that longer- and shorter-period variations also 11,956 TUCHOLKE AND LIN: GEOLOGICAL may exist. During phases of magmatic extension, the IC footwall of the master detachment exposes largely gabbroic lower crustalrocks that are intermittently cappedby basalt. It is during thesemagmaticphasesthat the OC hangingwall will most closely approximate a normal crustal section having standardigneousstratigraphyand velocity structure. As a spreadingsegmententersa phaseof amagmaticextension, the crust at the spreadingaxis cools, the brittle/ductile transitiondeepensinto the uppermantle, and the toe of the detachmentdives to follow. With continuedamagmaticextension, footwall uplift rapidly exhumes progressively deeper rocks from the lithosphere. In addition, through-crustfaults in the hangingwall facilitate seawaterpenetrationand serpentinization of the upper mantle, and this may eventually stimulate serpentinitediapirism in the hanging wall. The longer theperiodof am•gmatic extension, themorepronounced will be the effective thinning of the hanging wall by faulting and serpentinitediapirism, and the greaterwill be the possibility o• OC eicposu•e of uppermantlerocks.Fromtheseconsiderations, we infer that variability in crustal thicknessand compositionwill be greatestin IC crust and will be complementary but lesswell developedin OC crust. The long-term operationof detachmentfaulting at the ridge axis results in creation of ribbons of thin crust that extend off- axis along the old, IC sidesof first- and second-orderdiscontinuities,togetherwith ribbonsof more normal crust at segment centers and along young, OC sides of the discontinuities [Tucholkeand Lin, 1991]. Thus significantalong-isochron variability in crustal composition,extendingwell beyond the bounds of known anomalous crust within discontinuities, MODEL OF RIIX]E SEGMENTS Acknowledgments. This researchwas supportedby Office of Naval Researchgrants N00014-90-J-1621 and N00014-91-J-1433 and by NationalScienceFoundationgrantsOCE 8716713 and OCE 9020408. Wethank H.J.B.Dick,G.Hirth,J.Karson, andM. Kleinrock forhelpful discussions andG. HirthandR. Pockalny for thoughtful reviewsof the manuscript. Contribution 8369 of Woods Hole Oceanographic Institution. References Ambos, E. L., andD. M. Hussong, Oi:eanographer transform fault structurecomparedto that of surrounding oceaniccrust:Resultsfrom seismicrefractiondataanalysis,J. Geodyn.,5, 79-102, 1986. Aumento,F., and H. Loubat,The Mid-Atlantic Ridge near 45øN, XVI, Serpentinizedultramarieintrusions,Can. J. Earth $ci., 8, 631-663, 1971. Ballard, R. D., and T. H. van Andel, Morphologyand tectonicsof the innerrift valleyat lat. 36ø50'NontheMid-AtlanticRidge,Geol.$oc. Am. Bull., 88, 507-530, 1977. Bassias, Y., andC. Triboulet,Petrology andP-T-t evolutionof theSouth West Indian Ridge peridotites. A easestudy: East of the Melville FractureZoneat 62øE,Lithos,28, 1-19, 1992. Bercovici,D., H. J. B. Dick, andT. P. Wagner,Nonlinearviscoelasticity andthe formationof transverse ridges,J. Geophys.Res.,97, 14,19514,206, 1992. Blackman,D. K., and D. W. Forsyth,Isostaticcompensation of tectonic featuresof the Mid-Atlantic Ridge: 25ø-27ø30'S,J. Geophys.Res., 96, 11,741-11,758, 1991. Boillot, G., D. Mougenot, J. Girardeau,and E. L. Winterer, Rifting processes onthewes•Galiciamargin,Spain,in Extensional Tectonics and Stratigraphyof the North Atlantic Margins, editedby A. J. Tankard and H. R. Balkwill, AAPG Mem., 46, 363-377, 1989. Bonatti,E., Ultramaficrocksfrom the Mid-AtlanticRidge,Nature,219, 363-364, 1968. shouldexistin old oceancrustalongthe flanksof slowspread- Bonatti, E., Serpentiniteprotrusionsin the oceaniccrust,Earth Planet. ing mid-ocean ridges. Optimal operation of the detachment processcould result in up to half of the oceancrustexhibiting thinned or missing gabbroic or volcanic sections. Ophiolites in the FrenchWestern Alps and the Apenninescommonlydis- $ci. Lett., 32, 107-113, 1976. Bonatti,E., J. R. Lawrence,' andN. Morandi,Serpentinization of oceanicperidotites: Temperaturedependence of mineralogyand boron content,Earth Planet. $ci. Lett., 70, 88-94, 1984. play thesefeatures[Lagabrielleand Cannat,1990;Tricart and Brown,I. R., andJ. A. Karson, Variations in axialprocesses ontheMid- Lemoine, 1991], and it thereforeis possiblethat the Ligurian ocean crust from which the ophiolites were derived was accreted at slow spreadingrates (<25 mm/yr). If so, these ophiolites are a potentially rich source of information about the structure and stratigraphy of ocean crust affected by detachmentfaulting. Cannat,M., Emplacementof mantlerocksin the seafloorat mid-ocean ridges,J. Geophys.Res.,98, 4163-4172, 1993. Cannat,M., D. Bideau,and H. Bougault,Serpentinizedperidotitesand In the ocean basins, two kinds of fundamental studies are Atlantic Ridge: The median valley of the MARK area, Mar. Geophys. Res.,10,109-138, 1988. gabbros in the Mid-AtlanticRidgeaxial valleyat 15ø37'Nand 16ø52'N,Earth Planet. $ci. Lett., 109, 87-106, 1992. Casey, J. F., M. Cannat, and FARANAUT Scientific Party, Crustal architectureand mechanismsof ultramafic and gabbroicexposure alongrift valley walls of the Mid-Atlantic Ridge northand southof the 15020' transform (abstract), Eos Trans. AGU, 73 (43), Fall MeetingSuppl.,537, 1992. Chen, Y., andW. J. Morgan, A nonlineartheologymodelfor mid-ocean ridgeaxistopography, J. Geophys. Res.,95, 17,583-17,604,1990. Collette,B. J., Fracturezonesin the North Atlantic:Morphologyand a needed to assessthe patterns of crustal heterogeneity,both acrossand along isochrons,that are suggestedby the detachment model. First, long (20-30 m.y.) recordsof conjugateIC and OC crustmust be examinedwith detailedmorphological and potential-fieldsurveyson the two flanks of a slow spreadling ridge. These surveyscan examine the degreeof variability in crustal compositionand relative crustal thicknessand can model, J. Geol. $oc. London, 143, 763-774, 1986. determinewhetherthesefactorsvary sympatheticallyat inside Corntier,M. H., R. S. Detrick,andG. M. Purdy,Anomalouslythin crust and outside corners. Second, refraction seismic 'studies are needed to test the validity of crustal thicknesspatterns suggestedby the geologicaland geophysicaldata. However, serpentinizationis likely to have altered upper mantle and deepest crustal ultramafic rocks, and it will strongly affect seismic velocity structure. This difficulty can best be overcome by carefully siting the seismicexperimentswithin the context of detailed geophysicalsurveysand by analysisof crustalthicknesspatternsusingcombinedmodelingof seismicand gravity data. Fine-scale, near-bottomgeologicalmapping and sampling will provide the ultimate tests of the observationsand predictionsmade from theselarge-scaledata. in oceanic fracture zones: New seismic constraints from the Kane FractureZone,J. Geophys.Res.,89, 10,249-10,266,1984. Crane, K., The spacingof ridge-axishighs:Dependenceupon diapiric processes in the underlyingasthenosphere?, EarthPlanet.Sci.Lett., 72, 405-414, 1985. Davis, G. A., and G. S. Lister, Detachmentfaulting in continental extension: Perspectives fromthesouthwestern U.S. Cordillera, Spec. Pap. Geol. $oc.AM, 218, 133-159, 1988. Detrick, R. S., J. C. Mutter, P. Buhl, and I. I. Kim, No evidence from multichannelreflection data for a crustal magma chamberin the MARK areaon the Mid-AtlanticRidge,Nature,347, 61-64, 1990. Derrick,R. S., R. S. White, and G. M. Purdy,Crustalstructureof North Atlanticfracturezones,Rev. Geophys.,31, 439-458, 1993. TUCHOLKE AND LIN: GEOLOGICAL MODEL OF RIDGE SEGMENTS •_!,957 Dick, H. J. B., Abyssal peridotites, very slow spreadingridges and oceanridge magmatism,in Magrnatisrn in the OceanBasins,edited topography of the crestal mountains near a ridge-transform by A.D. Saunders andM. J. Norry, Geol.Soc.Spec.Publ. London, Kuo, B. Y., and D. W. Forsyth,Gravity anomaliesof the ridge-transformsystemin theSouthArianticbetween31 and34.5øS:Upwelling centersand variationsin crustalthickness,Mar. Geophys.Res., 10, 42, 71-105, 1989. Dick, H. J. B., W. B. Bryan,andG. Thompson,Low-anglefaultingand steady-stateemplacementof plutonic rocks at ridge-transform intersections,Eos Trans.A GU, 62, 406, 1981. Dick, H. J. B., H. Schouten,P.S. Meyer, D. G. Gallo, H. Bergh, R. Tyce,P. Patriat,K. T. M. Johnson, J. Snow,andA. Fisher,Tectonic evolution of the Atlantis II Fracture Zone, Proc. Ocean Drill. Program,Sci.Results,118, 359-398, 1991. Engel,C. G., and R. L. Fisher,Graniticto ultramaficrock complexesof the Indian Oceanridge system,westernIndian Ocean, Geol. Soc. intersection,Eos Trans. AGU, 65, 274, 1984. 205-232, 1988. Lagabrielle,Y., and M. Cannat,AlpineJurassic ophiolitesresemblethe moderncentralAtlanticbasement,Geology,18, 319-322, 1990. Lin, J., and J. PhippsMorgan,The spreadingrate dependence of threedimensionalmid-oceanridge gravity structure,Geophys.Res. Lett., 19, 13-16, 1992. Lin, J., G. M. Purdy, H. Schouten, J.-C. Sempere, and C. Zervas, Evidence from gravity data for focusedmagmatic accretionalong Am. Bull., 86, 1553-1578, 1975. the Mid-AtlanticRidge,Nature, 344, 627-632, 1990. Fisher, R. L., H. J. B. Dick, J. H. Natland, and P.S. Meyer, MaficLin, J., B. E. Tucholke,and M. C. Kleinrock, Off-axis "boudin-shaped" ultramafic suitesof the slowly spreadingSouthwestIndian Ridge: gravity lows on the western flank of the Mid-Atlantic Ridge at PROTEA explorationof the Antarcticplate boundary,24øE - 47øE, 25ø25'-27ø10'N:Evideacefor long-termpulsesin magmaticaccretion Ofioliti, 11, 147-178,1986. in spreadingsegments(abstract),Eos Trans.AGU, 74 (16), Spring Fox, P. J., and D. G. Gallo, A tectonicmodelfor ridge-transform-ridge Meetingsuppl.,380, 1993. plate boundaries: Implications for the structure of oceanic Louden, K. E., and D. W. Forsyth, Crustal structure and isostatic lithosphere,Tectonophysics, 104, 205-242, 1984. compensationnear the Kane FractureZone from topographyand Fox, P. J., and D. G. Gallo, The geologyof North Atlantic transform gravitymeasurements, I, Spectralanalysisapproach,Geophys.J. R. plate boundariesand their aseismicextensions,in The Geologyof Astron. Soc., 68, 725-750, 1982. North America;vol. M, The WesternNorth AtlanticRegion,editedby Macdonald,A. H., and W. S. Fyfe, Rate of serpenfinizafion in seafloor P. R. Vogt and B. E. Tucholke,pp. 157-172,GeologicalSocietyof environments, Tectonophysics, 116, 123-135,1985. America, Boulder, Colo., 1986. Fox, P. J., and D. G. Gallo, Transformsof the easterncentralPacific,in The Geologyof North America,vol. N, The EasternPacific Ocean and Hawaii, edited by E. L. Winterer, D. M. Hussong,and R. W. Decker,pp. 111-124,GeologicalSocietyof America,Boulder,Colo., 1989. Fox, P. J.,W. C. PitmanIII, andF. Shephard, Crustalplatesin the central Atlantic:Evidencefor at leasttwo polesof rotation,Science,165, 487-489, 1969. Fox, P. J., R. S. Detrick, andG. M. Purdy,Evidencefor crustalthinning near fracture zones: Implicationsfor ophiolites,in Ophiolites: ProceedingsInternationalOphioliteSymposium,Cyprus1979, pp. 161-168,CyprusGeologicalSurveyDepartment, Nicosia,1980. Francis,T. J. G., Serpentinization faultsandtheir role in the tectonicsof slowspreading ridges,J. Geophys.Res.,86, 11,616-11,622,1981. Gente, P., L. P. Zonenshain, M. I. Kuzmin, A. P. Lisitsin, Y. A. Bogdanov,and B. V. Baranov,Geology of the axis of the MidAtlantic Ridge between23øN and 26øN: Preliminaryresultsof the 15thcruiseof R/V AkadetnikMstislavKeMysh(March-April 1988), C. R. Acad. Sci. Ser. 2, 308, 1781-1788, 1989. Gente,P., C. Mevel, J. M. Auzende,J. A. Karson,and Y. Fouquet,An exampleof a recentaccretionon the Mid-AtlanticRidge:The Snake Pit neovolcanicridge (MARK area,23ø22'N),Tectonophysics, 190, 1-29, 1991. Macdonald, K. C., The crest of the Mid-Atlantic Ridge: Models for crustalgeneration processes andtectonics, in The Geologyof North America,vol. M, The WesternNorth Atlantic Region,editedby P. R. Vogt andB. E. Tucholke,pp. 51-68, GeologicalSocietyof America, Boulder, Colo., 1986. Macdonald,K. C., D. S. Scheirer,and S. M. Carbotte,Mid-oceanridges: Discontinuities, segmentsand giant cracks,Science,253, 986-994, 1991. Madsen,J. A., R. S. Detrick, J. C. Mutter, P. Buhl, and J. A. Orcutt, A two- and three-dimensional analysisof gravity anomaliesassociated with the East Pacific Rise at 9øN and 13øN, J. Geophys.Res., 95, 4967-4987, 1990. Menard, H. W., and T. M. Atwater, Changesin directionof sea floor spreading, Nature,219, 463-467,1968. Mevel, C., M. Cannat, P. Gente, E. Marion, J. M. Auzende, and J. A. Karson,Emplacementof deepcrustaland mantlerockson the west median valley wall of the MARK area (MAR, 23øN), Tectonophysics, 190, 31-53, 1991. Minshull, T. A., R. S. White, J. C. Mutter, P. Buhl, R. S. Detrick, C. A. Williams, and E. Morris, Crustalstructureat the Blake Spur Fracture Zone from expandingspreadprofiles,J. Geophys.Res.,96, 99559984, 1991. Miyashiro, A., F. Shido, and M. Ewing, Compositionand origin of serpentinitesfrom the Mid-Atlantic Ridge near 24ø and 30ø north latitude,Contrib.Mineral. Petrol., 23, 117-127, 1969. Hussong,D. M., P. B. Fryer,J. D. Tuthill, andL. K. Wipperman,The geologicaland geophysicalsetting near DSDP Site 395, North Morris, E., and R. S. Detrick, Three-dimensionalanalysisof gravity AtlanticOcean,Initial Rep.DeepSeaDrill. Proj., 45, 23-37, 1979. anomaliesin the MARK area,Mid-Atlantic Ridge 23øN,J. Geophys. Karson, J. A., Seafloor spreading on the Mid-Atlantic Ridge: Res., 96, 4355-4366, 1991. Implicationsfor the structureof ophiolitesand oceaniclithosphere Morris, E., R. S. Detrick, T. A. Minshull, J. C. Mutter, R. S. White, W. producedslow-spreadingenvironments,in Proceedingsof the Su, and P. Buhl, Seismic structure of oceanic crust in the western NorthAtlantic,J. Geophys.Res.,98, 13,879-13,903,1993. Symposium TROODOS1987, editedby J. Malpaset al., pp. 547-555, GeologicalSurveyDepartment,Nicosia,Cyprus,1990. Miiller, R. D., and W. R. Roest, Fracture zones in the North Atlantic Karson,J. A., and H. J. B. Dick, Tectonicsof ridge-transformintersecfrom combinedGeosatand Seasatdata, J. Geophys.Res.,97, 3337- tionsat the Kane FractureZone, Mar. Geophys.Res.,6, 51-98, 1983. Karson,J. A., and H. J. B. Dick, Deformedand metamorphosed oceanic crustonthe Mid-AtlanticRidge,Ofioliti,9, 279-302, 1984. Karson,J. A., et al., Along-axisvariationsin seafloorspreadingin the MARK area, Nature, 328, 681-685, 1987. Klitgord, K. D., and H. Schouten,Plate kinematicsof the central Atlantic, in The Geologyof North America, vol. M, The Western North Atlantic Region,editedby P. R. Vogt and B. E. Tucholke,pp. 351-378,GeologicalSocietyof America,Boulder,Colo., 1986. Kuo, B.-Y., J. Phipps Morgan, and D. W. Forsyth, Asymmetryin 3350, 1992. Mutter, J. C., and J. A. Karson, Structuralprocessesat slow-spreading ridges,Science,257, 627-634, 1992. Mutter,J. C., andNAT StudyGroup,Multichannelseismicimagesof the oceaniccrust'sinternal structure:Evidence for a magma chamber beneaththe Mesozoic Mid-Atlantic Ridge, Geology, 13, 629-632, 1985. O'Hara, M. J., Mineral facies in ultrabasicrocks, in Ultramafic and Related Rocks,edited by P.J. Wyllie, pp. 7-18, JohnWiley, New York, 1967. 11,958 TUCHOLKE AND LIN: GEOLOGICAL OTTER, The geology of the Oceanographertransform: The ridgetransformintersection, Mar. Geophys.Res.,6, 109-14l, 1984. Pariso, J.E., and H.P. Johnson,Do layer 3 rocks make a significant contributionto marinemagneticanomalies?In situ magnetization of gabbrosat OceanDrilling ProgramHole 735B, J. Geophys.Res.,98, 16,033-16,052, 1993. Parsons,B., and J. G. Sclater,An analysisof the variationof oceanfloor bathymetryand heat flow with age, J. Geophys.Res.,82, 803-827, 1977. Phillips,J. D., andH. S. Fleming,Multibeamsonarstudyof the Mid-AtlanticRidgerift valley36ø-37øN,Geol.$oc.Am.Bull.,88, 1-5, 1977. PhippsMorgan,J., andD. W. Forsyth,Three-dimensional flow andtemperatureperturbationsdue to a transformoffset:Effects on oceanic crustalanduppermantlestructure,J. Geophys.Res.,93, 2955-2966, 1988. Pockalny,R. A., R. S. Detrick,andP. J. Fox, Morphologyandtectonics of the Kane transformfrom Sea Beambathymetrydata,J. Geophys. Res., 93, 3179-3193, 1988. Pockalny,R. A., and P. Gente,A modelfor the segmentation characteristicsof a slowly spreadingmid-oceanridge (abstract),Eos Trans. AGU, 73 (43), Fall Meetingsuppl.,569, 1992. Prince,R. A., and D. W. Forsyth,Horizontalextentof anomalously thin crust near the Vema Fracture Zone from the three-dimensional anal- ysisof gravityanomalies, J. Geophys. Res.,93, 8051-8063,1988. Prinz, M., K. Keil, J. A. Green, A.M. Reid, E. Bonatti, and J. Honnorez, Ultramafic and mafic dredge samplesfrom the equatorial MidAtlantic Ridge and fracturezones,J. Geophys.Res.,81, 4087-4103, 1976. Purdy,G. M., J.-C. Sempere,H. Schouten,D. L. Dubois,and R. Goldsmith,Bathymetryof the Mid-AtlanticRidge,24ø - 31ø N: A map series,Mar. Geophys.Res.,12, 247-252, 1990. Rona, P. A., and D. F. Gray, Structural behavior of fracture zones symmetric and asymmetricabout a spreadingaxis: Mid-Atlantic RidgeOat.23øNto 27øN),Geol.Soc.Am. Bull.,91, 485-494,1980. Schouten,H., and R. S. White, Zero-offsetfracturezones,Geology,8, 175-179, 1980. Schouten,H., H. J. B. Dick, and K. D. Klitgord,Migrationof mid-ocean ridgevolcanicsegments, Nature, 326, 835-839, 1987. Searle, R. C., and A. S. Laughton,Sonar studiesof the Mid-Atlantic Ridge and KurchatovFractureZone, J. Geophys.Res.,82, 53135328, 1977. Severinghaus, J.P., and K. C. Macdonald,High insidecornersat ridgetransformintersections, Mar. Geophys.Res.,9, 353-367, 1988. ShipboardScientificParty,Site 395: 23øN, Mid-AtlanticRidge,Initial Rep.Deep SeaDrill. Proj., 45, 131-264, 1979. ShipboardScientificParty,Site 670, Proc. OceanDrill. Program,Part A, Initial Rep.,109, 203-237, 1988. Sloan, H., and P. Patriat, Kinematics of the North American--African plateboundary between28øand29øduringthelast 10 Ma: Evolution of the axial geometryandspreadingrate anddirection,Earth Planet. Sci. Left., 113, 323-341, 1992. Smith, D. K., and J. R. Cann, The role of seamountvolcanismin crustal MODEL OF RII2X3E SEGMENTS construction at the Mid-AtlanticRidge(24ø-30øN),J. Geophys.Res., 97, 1645-1658, 1992. Smith, D. K., and J. R. Cann, Building the crust at the Mid-Atlantic Ridge,Nature,365, 707-715,1993. Sparks,D. W., E. M. Parmentier,and J. PhippsMorgan,Three-dimensional mantle convectionbeneath a segmentedspreadingcenter: Implications for along-axisvariationsin crustalthickness andgravity, J. Geophys. Res.,98, 21,977-21,995,1993. Srivastava,S. P., H. Schouten,W. R. Roest, K. D. Klitgord, L. C. Kovacs,J. Verhoef, and R. Macnab, Iberian plate kinematics: A jumping plate boundarybetweenEurasiaand Africa, Nature, 344, 756-759, 1990. Tiezzi, L. J., andR. B. Scott,Crystalfractionation in a cumulategabbro, Mid-AtlanticRidge,26øN,J. Geophys.Res.,85, 5438-5454,1980. Tricart,P., and M. Lemoine,The Queyrasophiolitewestof Monte Viso (WesternAlps): Indicatorof a peculiaroceanfloor in the Mesozoic Tethys,J. Geodyn.,13, 163-181,1991. Tucholke,B. E., and J. Lin, Asymmetriccrustalstructureat ridge-axis offsets,Mid-Atlantic Ridge (abstract),Eos Trans.AGU, 72 (44), Fall Meetingsuppl.,467, 1991. Tucholke,B. E., andJ. Lin, Detachmentfaultingandits relationto geologic structureat first- and second-order offsetsof slow-spreading ridges(abstract),Eos Trans.AGU, 73 (14), SpringMeetingsuppl., 286, 1992. Tucholke,B. E., and H. Schouten,Kane FractureZone,Mar. Geophys. Res., 10, 1-39, 1988. Wernicke, B., Low-angle normal faults in the Basin and Range Province:Nappetectonicsin an extendingorogen,Nature, 291, 645648, 1981. White, R. S., R. S. Detrick, M. C. Sinha, and M. H. Cormier, Anomalous seismiccrustalstructureof oceanicfracturezones,Geophys.J. R. Astron. $oc., 79, 779-798, 1984. White, R. S., R. S. Detrick, J. C. Mutter, P. Buhl, T. A. Minshull, and E. Morris, New seismicimagesof oceaniccrustalstructure,Geology, 18, 462-465, 1990. Whitehead, J. A., H. J. B. Dick, and H. Schouten, A mechanismfor magmaticaccretionunderspreadingcenters,Nature, 312, 146-148, 1984. Zervas,C., J. Lin, and P. Rona,AsymmetricV-shapedgravitystripesat the Mid-Atlantic Ridge 26øN (abstrac0,Eos Trans.AGU, 71, 1572, 1990. Zonenshain,L. P., M. I. Kuzmin,A. P. Lisitsin,Y. A. Bogdanov,andB. V. Baranov,Tectonicsof the Mid-Atlantic rift valley betweenthe TAG and MARK areas (26-24øN): Evidence for vertical tectonism, Tectonophysics, 159, 1-23, 1989. J. Lin and B. E. Tucholke,Departmentof GeologyandGeophysics, Woods Hole OceanographicInstitution,Woods Hole, MA 02543 (email:Internetjlin @whoi.edu;btucholke @whoi.edu) (ReceivedMay 3, 1993;revisedJanuary24, 1994; acceptedJanuary31, 1994.)