JOURNALOF GEOPHYSICALRESEARCH, VOL. 105,NO. B4,PAGES 8355-8386, APRIL 10,2000

Glacial-interglacial trench supply variation, spreading-ridge subduction, and feedback controls on the Andean margin developmentat the Chile triple junction area (45-48øS)

JacquesBourgois,• Christhie Guivel,2 Yves Lagabrielle,3Thierry Calmus,4 JacquesBoulbgue,S and Valerie Daux6

Abstract. Duringthe Chile triple junction (CTJ) cruise (March-April 1997), EM12 bathymetry andseismic reflection data were collected in thevicinity of theChile triple junction (45-48øS), wherean activespreading ridge is beingsubducted beneath the Andean continental margin. Resultsshow a continentalmargin development shaped by tectonicprocesses spanning a spectrum from subduction-erosionto subduction-accretion. The Andeancontinental margin and the Chile trenchexhibit a strongsegmentation which reflects the slabsegmentation and the Chiletriple junctionmigration. Three segments were identified along the Andean continental margin: the presubduction,the synsubduction,and the postsubductionsegments, from northto south.Both climate-inducedvariations of the sedimentsupply to thetrench and the tectonic reorganization at the Nazca-Antarcticaplate boundary involving postsubduction ridge jump arethe two main factors that controlthe tectonicregime of this continentalmargin. Along the surveyarea we infer the successionof two differentperiods during the lastglacial-interglacial cycle: a glacialperiod with ice-rafteddetrital discharges restricted to the shorelinearea and low river outputand a warmer periodduring which the Andean ice capretreat allowed the Andes to be drainedoff. Duringthese warmperiods, rapid increase in trenchdeposition caused the margin to switchfrom subduction- erosionor nonaccretionto subduction-accretion:(1) along the presubduction segment after the last deglaciationand (2) alongthe postsubduction segment after the interglacialepisode at 130-117ka. Conversely,a nonaccretionor subduction-erosionmode characterized the presubduction and postsubductionsegments during glacial maximums. The majoreffects of subductionof the buoyantChile ridgeinclude a shallowtrench which diverts trench sediment supply and tectonic instabilitiesat the Nazca-Antarcticaplate boundary. We suggestthat a postsubductionwestward jump of the Chile ridgeoccurred during the past780 kyr. It producedslab fragmentation and individualizationof an ephemeralmicroplate north of the Taitaofracture zone: the Chonos microplate.In 780 kyr, two episodesof subduction-accretionseparated by an episodeof subduction-erosionoccurred in relationwith the Chonosmicroplate individualization and subduction.The currentnorthward migration of thetriple junction along the Chonosmicroplate- SouthAmerica plate boundary introduces a sharpchange in the tectonicmode from subduction- erosionto the northto subduction-accretionto the south.The datacollected along the Taitaoridge haverevealed the complexthree-dimensional structure of an accretionarywedge which includes a midslopethrust sheet exhibiting the characteristicsof an ophiolite:the TaitaoRidge ophiolite. No connectionexists between the Taitao Ridge ophiolite and the BahiaBarrientos ophiolite cropping out onlandin the Taitao peninsula.

1CentreNational de la RechercheScientifique, Universit6 Pierre et 1. Introduction Marie Curie, Laboratoire de G6odynamique,Tectonique et Environnement,Paris. The Chile triple junction (Figure 1) is the site wherethe 2Laboratoirede Plan6tologieet G6odynamique, Universit6 de Antarctica,the Nazcaand the SouthAmerica plates meet Nantes, Nantes, France. [Candeand Leslie, 1986; Cande et al., 1987;Behrmann et al., 3Institutde Recherchepour le D6veloppement,Noumea, Nueva Caledonia. 1994]. At 46ø09'S, the active spreadingcenter at the 4InstitutoGeologico, Universidad Autonoma de Mexico,Estacion Antarctic-Nazcaplate boundary is being subductedbeneath Regionaldel Noroeste,Hermosillo, Sonora, Mexico. theSouth America continental margin. The Chile margin triple 5Laboratoirede G6ochimieet M6tallog6nie,Universit6 Pierre et junction area providesthe opportunityto study the Marie Curie, Paris. petrologicaland tectonic effects of spreadingridge subduction 6LaboratoiredeG6ologie S6dimentaire, Universit6 Pierre et Marie alonga continentalmargin, a processthat has dramatically Curie, Paris. affectedthe geology of both North and South American Copyright2000 by the AmericanGeophysical Union. marginsduring the past 70 Myr [Atwater, 1970; Dickinson andSnyder, 1979; Ramos and Kay, 1992; Sisson and Pavlis, Papernumber 1999JB900400. 1993; Kay et al., 1993]. Spreadingridge subductionleaves 0148-0227/00/1999JB900400509.00 specificstructural and stratigraphicsignatures, as recently

8355 8356 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

78øW 77øW 76øW 75øW 74øW 73øW 45øS

CHILERIDGE 64 km/m.y.

859 86O 861 46øS SEGMENT' •63 78øW 76øW 74øW 72øW

NAZCA PLATE

8,0•,mlm-¾'

47øS

GP

48øS

48øS

Figure 1. Bathymetricmap of the region surveyedduring the CTJ cruise of the R/V L',4talante in the Chile triple junctionarea, between45øS and 48øS.The main seafloormorphological features include the Chile ridge axesand fracturezones, the Chile trench, and the Andean continental margin.Location map is shown in the inset.Numbers 859 to 863 referto the ODP sites drilled during ODP Leg 141. Chile triple junction (CTJ); Chiloe block (CB); Esmeraldafracture zone (EFZ); Golfo de Penas(GP); Liquine-Ofqui fault system(LOFS); Rio Baker (RB); Tres Montes peninsula(TMP); Taitao peninsula(TP).

shown along the Taitao peninsula transect, including (1) regime of the margin would have evolved from subduction- rapid uplift and subsidenceof the forearcdomain [Cande and erosion to subduction-accretion.Indeed, the rebuilding of an Leslie, 1986; Bourgois et al., 1992], (2) anomalous near- accretionary prism following the partial destruction of the trenchand forearcmagmatism [Mpodozis et al., 1985; Kaeding forearc [Behrmann et al., 1994; Bourgois et al., 1996] i s et al., 1990; Lagabrielle et al., 1994; Guivel et al., 1999], (3) associatedwith the passageof the Chile triple junction along removal of forearc material from the overriding plate [Cande the margin. This situation offersto addressthe question of and Leslie, 1986; Cande et al., 1987; Behrmann et al., 1992a relative weight of factors controlling the tectonic regime of and b; Bourgoiset al., 1996], and (4) extensionaltectonics in convergent margins (i.e. type I versus type 2 convergent relation with a subducted ridge segmentat depth [Murdie et margin [von Huene and Scholl, 1991]) and the evolution al., 1993]. Moreover, ophiolite emplacement[Forsythe and through spaceand time from one type to another. Although, Nelson, 1995], elevatedthermal gradient [Cande et al., 1987], the effects of subduction-erosion have been demonstrated at alteration, diagenesis, and mineralization [Haeussler et al., many convergent margins, the constraints of physical 1995] of forearcmaterials are expectedas a consequenceof hot conditionsthat apply at the front of convergentmargins have fluids venting from the subductingspreading ridge. been poorly investigated [Scholl et al., 1980; Hussong and Results of Ocean Drilling Program (ODP) Leg 141 Wipperman, 1981; Aubouin et al., 1984; von Huene and [Behrmannet al., 1992a; Lewis et al., 1995] suggestthat the Lallemand, 1990]. Parameters such as (1) the rate and subductionof the active spreadingcenter of the Chile ridge direction of relative plate motion, (2) the topography of the beneath the South American continent is associated with a subducting plate, and (3) the type and volume of sediment major changein the tectonicregime of the continentalmargin. need to be quantified [von Huene, 1986]. As the Chile triple junction migratedto the north, the tectonic During the CTJ cruise (March 13 to April 7, 1997) of the BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8357

78øW 77øW 76øW 46ø26'S to 47øS. (3) The typical "postsubduction segment", located south of 47ø10'S, exhibits a wide accretionary wedge (i.e., the Golfo de Penasaccretionary prism). Climate-induced variations of the sediment supply to the trench axis and the tectonic reorganization at the Nazca- Antarctica plate boundary involving postsubduction ridge jump are identified as two main factors that control the tectonicregime of the Andeancontinental margin in the Chile : [: _ triple junction area. A nonaccretion or subduction-erosion mode characterized the presubduction and postsubduction segmentsduring the two last glacial maximums,whereas warmer periods associated with rapid increase in trenct• deposition caused the margin to switch from subduction- erosion or nonaccretion to subduction-acccretion.The major effects of subduction of the buoyant Chile ridge include a shallow trench which divert trench sediment supply and tectonic instabilities at the Nazca-Antarctica plate boundary. A postsubductionwestward jump of the Chile ridge occurred during the past 780 kyr. It produced slab fragmentationand individualizationof an ephemeralmicroplate. In 780 kyr, two episodes of subduction-accretionseparated by an episode of subduction-erosion occurred along the synsubduction segment.The data collectedalong the Taitao Ridge revealsthe complexthree-dimensional structure of an accretionarywedge which includes a midslope thrust sheet exhibiting the characteristicsof an ophiolite:the Taitao Ridge ophiolite. The survey carried out by the R/V L'Atalante helps to better address the problem of mass transfer in the case of the subductionof an active spreadingcenter and examine the main _ factors determining the tectonic regime of the continental -- i i i i I i i i i i i i i i i i i i i i i i i i i i I i i i i i margin. Figure 2. Ship track during the CTJ cruise. Numbers I to 21 refer to seismic lines shot during the cruise. 2. Tectonic Setting Seafloorspreading along the Chile ridge at the Nazca- Antarctica plate boundary (Figure 1) occurredat a rate of 64 R/V L'Atalante, geophysical surveys employing EM12 km/Myr [DeMetset al., 1990] over the past 5 Myr [Herron et multibeam echo-sounding,sonar imagery,six-channel seismic al., 1981]. North of the Chile triple junction, the Nazca plate reflection, gravity, and magnetic profiling (Figure 2) were is beingsubducted beneath the SouthAmerica plate at a rate of conductedin the Chile triple junction area.A 100% coverage 80 km/Myr in a N80øE direction. To the south, the Antarctic bathymetric map (Plate 1) was obtained between 45øS and plate is subductedbeneath the SouthAmerica plate at a rate of 48øS [Bourgois et al., 1997]. Samples were collected at 24 km/Myr in an E-W direction.The SouthChile ridge (Plate numerousdredge sites located in the vicinity of the Chile 1) consistsof shortsegments (segment 1, 2, and 3) trending triple junction on both oceanic and continental sides of the N160ø-165øE, separated by a seriesof parallel fracturezones. Chile trench. In the areasurveyed during the CTJ cruisethey are the Darwin The Andean continental margin and the Chile trench andGuamblin fracture zones north of the Chile triple junction exhibit a strong segmentation which reflects the slab and the Taitao and Tres Montes fracture zones to the south. segmentationin relation with the northwardmigration of the Becausethe Chile ridge axis is -10 ø oblique to the Chile Chile triple junction. From north to south, three main trenchaxis, the Chile triple junction migratesnorthward at segmentsare defined:(1) North of the Darwin fracturezone, -160 km/Myr when in a ridge-trench-trenchconfiguration as the "presubductionsegment" which extendsfrom 45ø10'S to today [Cande et al., 1987]. When in a transform-trench-trench 45ø50'S is representativeof the marginbefore the subduction configuration,the Chile triple junction migratesslowly back of the Chile ridge.A thicktrench infill underthrustingbelow a to the south,at a rate of 10 km/Myr. narrow continental slope characterizesthis segment. (2) According to kinematic reconstruction [Forsythe et al., Between45ø50'S and 47ø10'S,a very complexsegment of the 1986], a long segmentof the Chile ridge met with the Chile margin is designatedas the "synsubductionsegment". This trench west of Tierra de Fuego, at 52øS, around 14 Ma. The segmentis divided into three subsegments.From north to northernsection of this segment,bounded to the north by the southit includesthe "Chileridge subsegment",along which Esmeraldafracture zone, subductedat about 49øS between 12 the Chile ridge is currently subducting beneath the and 10 Ma. In the surveyed area the Chile ridge segment continentalmargin between the Darwin fracturezone and the located between the Esmeralda and the Tres Montes fracture Chile triple junction; the "NorthTaitao canyon subsegment", zones was then subducted around 6 Ma. Off the Taitao and between the Chile triple junction and the South Taitao Tres Montes peninsulas, the subduction history was canyon;and the "Taitao ridge subsegment",located seawardof reconstructedin detail by Leslie [1986]. From 5-6 to 3 Ma, the Taitao and Tres Montes peninsulaswhich extendsfrom when the Tres Montes fracturezone subduction occurred,the 78•W 77ø30'w 77øW 76"30'W 76"W 75"30'W I . • • .• ?.-'>•,,•• '-' •_•..5 /•,s---• -/ z

41530 S • ,"-,•-•-'•, ' ' ,,.,½'%;,••'• •'•,..?%•,.,; •:•' • •;<; !

• • •,s x "• • •• '-• /'•X • x• .... • - •

z

46ø30'S- ! • • •-, x "t %1 •-•_ n I :• _T•:/•• • • •'••.. 'i - I • •" , ', /• z

'- • ' i

! /•/ •/,,•..• I

• , ' ,;•'.•., z

-3000 Fig.19 •,½,;½'•( '½'½"-. .

o Q.

• .... "',"'ao•%.••v•,,, 7tl

Plate1. Bathymetry derived fi'om the EM12 data. Scale on the left shows the color code for water depth. Locationmaps for Figures 5, 8, 9, 13, 16, 17,19, and 22 areshown. Chile trench (CT); Darwin fracture zone (DFZ);Guamblin fracture zone (GFZ); Isla Campana (IC); North Taitao Ridge fault (NTRF); Taitao fracture zone(TFZ); Taitao Fracture Zone ridge (TFZR); Taitao Ridge (TR); Tres Montes fracture zone (TMFZ); Tres Montespeninsula (TMP); Taitaopeninsula (TP). BOURGOIS ET AL.' CHILE TRIPLE JUNCTION 8359 triple junction migrated to the south from 46ø35'S to 75ø35'W 75ø30'W 75ø25'W 75ø20'W 75 ø15'W 46ø45'S. From 3 to 2.8 Ma the short ridge segment located I I I 46ø30'S between the Tres Montes and Taitao fracture zones was then subductedin associationwith the northward migration of the triple junction, from 46ø35'S to 46ø25'S. From 2.8 to 0.2-0.3 Ma the subduction of the Taitao fracture zone occurred in an TA I TA O PENINSULA area located between 46ø25'S and 46ø30øS. Since 0.2-0.3 Ma 46o35,S the Chile triple junction defined by the ridge segmentlocated north of the Taitao fracturezone (i.e., the Chile ridge segment 1, CRS I hereafter) and the trench (Figure 1) migrated / northward to its present location, at 46ø09'S. The map patternof shallowdepth (i.e., <100 km) seismicity 46o40,S exhibits a continuous distribution along the Nazca-South America plate boundary as well as along the ridge axis and SenoHoppner transformsat the Nazca-Antarctica plate boundary [Bourgois Pluton et al., 1993]. Therefore the seismic activity allows Barazangi and Izacks [1976] to define accurately the Wadati-Benioff zone north of the triple junction. Between 33øS and 45øS 46o45'S I Rape• latitude the Nazca slab subducts at an angle of 25-30 ø. I Plutor : Although much evidence of recent surface faulting has been recognized along the Liquine-Ofqui fault system (Figure 1), ::..:...... ½. the coastal area and the forearc south of the triple junction appear devoid of significant seismic activity. As a 46o50,S consequence,the Antarctica slab geometry beneath the South America plate is poorly defined south of the triple junction. However, Cande and Leslie [1986] have proposed that the Antarctica slab is being subducted with an angle of 20 ø between 46øS and 49øS. TRES MONTES PENINSULA 46o55,S

3. Regional Geology I I Onland, the most strikingfeatures east of the area surveyed duringthe CTJ cruisebetween 45ø10'S and 48ø10'S include Figure 3. Simplified geologicmap of the Taitao peninsula after (1) the southernend of the right lateral Liquine-Ofqui fault Mpodozis et al. [1985], Forsythe et al. [1986], and Bourgois systemwhich boundsthe Andeanbatholith to the west; (2) et al. [1996]. Units are as follow: 1, pre-Jurassicmetamorphic the calcalkaline South Volcanic Zone whose southern basement;2, Bahia Barrientos ophiolite; 3, Chile margin unit; extensionis at the latitude of the Chile triple junction at the 4, main volcanic unit; 5, Pliocene plutons. Mount Hudson volcano. The 46ø09'S latitude of the Chile triple junction fits also with the northern boundary of a volcanic gap extending southwardto the adakitic Mount Lautaro volcano, at 49øS; and (3) the Chiloe block (Figure 1) The Taitao and Tres Montes peninsulas located 50 to 1 10 west of the Liquine-Ofqui fault system.The geology of the km south of the Chile triple junction form the westernmos! Chiloe block continental margin has been interpreted from promontoryof the Chile coast.Volcanic and plutonic rocks of bothfield mappingof onlandexposures in the Taitao and Tres Pliocene age exposed along this promontory are located at Montes peninsulasand ODP Leg 141 drilling. <17 km landwardfrom the trenchaxis. Conversely,these rocks Impingementof the Chile ridge has caused large-scale are buried under a thick accumulationof slope sediment in deformationof the forearcupper crust. Forsythe and Nelson 1000 to 3000 m deep water, to the north and to the south of [1985] proposedan indenterkinematic model in which an the promontory.The Taitao and Tres Montes peninsulasare a elongated forearc silver (the Chiloe block) is moving window on the geology of the lower and midslopes of the northwardaway from an extensionalzone locatedin the Chile Chile continentalmargin. Five main tectonic units (Figure 3) triplejunction area. Two majortectonic features accommodate are recognized in the Taitao and Tres Montes peninsula area this movement:(1) transtensionalfaults in the Golfo de Penas, [Mpodozis et al., 1985; Forsytheet al., 1986; Kaeding et al., which has been interpretedas a pull-apart basin, and (2) the 1990; Bourgois et al., 1993, 1996]: (1) pre-Jurassic Liquifie-Ofquifault systemalong which right-lateralslip is metamorphic basement rocks, (2) the Bahia Barrientos suggested[Herv• et al., 1979]. Three-dimensionalfinite ophiolite,(3) the volcanic-sedimentaryChile marginunit, (4) element models [Nelson et al., 1994] of the Chile ridge the main volcanicunit which correlateswith the Chile margin subductionsetting substantiate development of an arc-parallel unit, and (5) the plutonicsuite which includesthe Cabo Raper fault at about 150-200 km fromthe trench. The Liquifie-Ofqui and SenoHoppner plutons. The Chile marginunit [Bourgois fault systemis interpretedas a complexzone of strike-slip et al., 1993] consistsof interbeddedsedimentary and volcanic faulting that coincideswith the forearc-arcboundary and material. Its estimated thickness is 4-6 km. The base of the boundsthe easternmargin of the Chiloe block consideredas a sequenceunconformably overlies the pre-Jurassicmetamorphic terrane. basementof the Chile margin.Nannoplankton assemblages are 8360 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

of early Pliocene age (5-5.3 Ma)at the base of the sequence 75øW 70øW 65øW and of early Pleistoceneage (1.5-1.6 Ma) in strata close to the I top of the sequence. A shallow water depositional Glacial limit environmentcharacterized the Chile marginunit throughout. duringlast glacial max. Beddingexhibits a subverticalto vertical attitudealong major ...... Dividetoday strike slip faults. From 5-5.3 Ma to 1.5-1.6 Ma, an area of the .... Rio Baker Chile margin locatedtoday abovesea level at <40 km fromthe drainagebasin 45os Chile trench axis was in upper slope to shelf water depth conditionsand had undergonea continuoussubsidence of 4-6 km. The tectonic regime, which is closely correlatedto the Chile ridge and transform migration beneath the Taitao peninsulaarea [Bourgoiset al., 1996], changedabruptly from subsidence to uplift sometime later than 1.5-1.6 Ma. Ot' particular interestis the 4-6 Ma Cabo Raper pluton located at the seaward edge of the Taitao peninsula. Its chemical Ice today characteristics,similar to those of adakitic or trondhjemite- Major tonalite-dacite suites, combined with tectonic data allowed today Bourgoiset al. [ 1996] (1) to reconstructthe paleogeometryof the Chile margin at the time of the Cabo Raper pluton 50øS emplacementand (2) to estimatequantitatively the volumeof material removedby subduction-erosionin associationwith the Chile ridge subductionbeneath the Taitao peninsula. Theseresults emphasized the particularrole of the subducting Chile ridge as an agentthat may contributeto increasethe rate of subduction-erosionat depth and thereforeto strongly influencethe tectonicregime of the continentalmargin. During ODP Leg 141, five sites (Figure 1) were drilled along the Chile continental margin.Sites 859, 860, and 861 were drilled along a transect located 10 km south of the Darwin fracturezone with the objective of providing a 55øS characterizationof the continentalslope 100 kyr priorto ridge collision. Site 863 is located few kilometers south of the Chile triplejunction, in an areawhere the axisof the spreadingridge subductedat 50 ka. Finally, Site 862 was drilled along the Figure 4. Limits of the last glacial maximumand distribution Taitaoridge interpreted to be an igneousbody that may be the of existingice field. Campode Hielo Norte (CHN); Campode seaward extension of the Bahia Barrientos ophiolite. The Hielo Sur (CHS); Cabo de Horno (CH); Chiloe Island (CI); sedimentsdrilled during ODP Leg 141 are Quaternaryto Comodoro Rivadavia (CR); Rio Baker (RB); Rio Simpson early Pliocene, the oldest recoveredsediment being 4.2-4.3 (RS); Taitao peninsula(TP). Note that the developmentof the ice cap during the Last Glacial Maximum prevents drainage to Ma in age. The lithology of rocks [Behrmannet al., 1994] the Pacific as the Rio Baker actstoday. recovered at Sites 859, 860, 861, and 863 includes detrital sedimentswith grain size ranging fromclay to gravel and conglomerate. By contrast to the Taitao and Tres Montes peninsulaarea, no volcanicrock was recoveredhere. Site 862 latitude trenches [Kulm et al., 1973; Scholl, 1974]. These located on the crest of the Taitao ridge drilled 20 m of wedges are only a few hundred thousand years and can thus sedimentsoverlying volcanic rocks. Microfossil assemblages rapidly thicken the trench infill entering the subduction. Off [Behrmannet al., 1992a]from the sedimentindicate an ageof southern Chile, Bangs and Cande [1997] argued that the late Pliocene, meanwhilerhyolitic samplesprovided Ar/Ar glacial climate and climatic variations in this region caused agesof 1.54+0.08and 2.2+0.4 Ma [Forsytheet al., 1995].New significant fluctuation in trench sedimentsupply that in turn detailed geochemical investigation on volcanic rock influenced the tectonic regime. The succession of short recoveredduring ODP Leg 141 [Guivel et al., 1999] allowed episodes of accretion, nonaccretion, and erosion observed comparisonwith equivalent volcanic suites exposedonland between35øS and 40øS is proposedto be linked to variation along the Taitao peninsula [Bourgois et al., 1993; in trench sediment supply. Lagabrielle et al., 1994; Le Moigne et al., 1996]. The During the Last Glacial Maximum, -20 kyr the Patagonian magmatic products erupted on both structures exhibit ice cap (Figure 4) extended1800 km along the Andes [Hollin similaritiesin chemicalsignature and age. and Schilling, 1981; Porter et al., 1992]. From Cap Horn at 56øS to Chiloe island at 42øS, ice totally coveredthe Andean 4. Paleoclimatic Background reliefs from the Argentina Andean piedmont to the Pacific shoreline.The last glaciation ended with massivecollapse of Glacial events which are supposedto increasethe ratesof ice lobesclose to 14 kyr ago [Lowell et al., 1995]. Since that terrigenoustrench sedimentationcan affectnot only the time, the ice cap retreatedto its presentdistribution, restricted accretedvolume [yon Huene and Scholl, 1991] but also the to the North and South Patagonianice fields. tectonicregime at convergentmargins. Wedge-shaped bodies The climateof Patagoniais dominatedby the Pacific Ocean of axial trench turbidite depositswere describedin high- to the west, the Andean Cordillera with peaks rising above BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8361

3000 m, and the dry plains of Argentina to the east. The cool which originated from a concave-upscar located 10-12 km temperatebelt extendssouth of 42øS [Miller, 1976], with the landwardfrom the subductionfront. The top of the displaced westerliesand precipitationreaching a maximum around 50øS. blocktilted landward produced a rev, erse slope. The estimated Here, meanannual precipitation may exceed5000 mm at sea displacementof the block is 5 krn as a minimum.Therefore the level. Precipitation totals decline sharply northward from Chile marginmorphology as a whole appearsto be shapedby 2000 mm at 40øSto <150 mmat 30øS. Numerical modeling to mass wasting and landslide processes. Failure of reconstructthe climate of Patagonia during the Last Glacial oversteepenedslopes is indicative of continental margin Maximum [Hulton et al., 1994] shows a northward migration subsidencesustaining slope instability. Similar featuresare of the precipitationbelt of-5 ø latitude with a decreaseof the observed frequently along margins undergoing active annual precipitationtotals at 50øS, and an increaseat latitude subsidence,considered commonly as the typical signatureof 40øS, the westerlies reaching a maximum at 45øS. The subduction-erosionat depth [Scholl et al., 1980; yon Huene topographic barrier of the Andes is expected to influence and Scholl, 1991]. Therefore we may assumethat massive atmosphericcirculation in a similar way during both glacial slumpingand masswasting documentedbetween 45ø10'S and and nonglacial periods. 45ø50'S, reflect active subduction-erosionat depth. This Therefore, along the surveyarea transect,between 45øS and tectonic regime is consistent with the narrowness of the 48øS latitude, we may infer the succession of two different continental slope along this segmentof the Chile continental periods during the last glacial-interglacial cycle: (1) a glacial margin. period with ice-rafted detrital discharge similar to the Four seismic lines (lines I to 4, Figures 2 and 5) were Heinrich events described in the northern Atlantic and low recordedalong the presubductionsegment. Line 3 (Figure 7) river output to the Pacific and (2) a warmer period during showsthe main stratigraphicand morphotectonicelements of which the ice cap retreat allowed the Patagonia Andean this area, including the downgoing Nazca plate, the trench piedmont to be drained off to the Pacific Ocean, across the axis and the lower continentalslope. Along this sectionof the Andesthrough the Rio Baker streamsystem (Figure 4). During Chile trench located between the Darwin and Guamblin the Last Glacial Maximum,the upper layer circulation of the fracture zones, the trench fill exhibits two depositional southernPacific Ocean and the westerlieshad the samestrong sequences.A 700-m-thick accumulationof well acoustically westward pattern as today. Only a modest northward stratifiedsediment overlies a more transparentseismic facies migration occurred. Therefore we infer that icebergs and some poorly defined parallel configurations.The lower originating from the Patagonian ice cap were permanently transparentlayer is-100 m thick. The horizontally stratified drifting along the Pacific South America shoreline, with trenchinfill exhibitsa total thicknessincreasing considerably Patagonian ice-rafted discharges being restricted to areas landward.It clearly onlapsthe Nazca plate sedimentarycover. close to the Chile shoreline. This demonstratesthat landward tilting of the volcanic In a recentreanalysis ofthe 15180 time series, Winograd et basement occurred before the trench sediment accumulated as al. [1997] show that the last four interglaciations, i.e., imagedalong line 3 (Figure 7). Along line 3 the fiat floor of isotopicsubstages 5e, 7e, 9c, and l 1c, rangefrom 20 to 26 kyr the Chile trench axis is -19.2 km wide. Because the in duration. Each of these warm events was followed a cold convergence rate between Nazca and Antarctica plates is periodof about60 to 80 kyr in duration.Therefore five glacial- about -8 cm/yr, subduction of the sedimentaccumulation, as interglacial cycles being 80 to 120 kyr in duration occurred shown along line 3 today, will take <250 kyr. As a during the past 500 kyr. The warmest portion of the last consequence,we assumethat line 3 exhibits trench sediment interglaciation,i.e.,substage 5e, is markedby a 15180plateau not older than 250 ka. Thereforeroughly 700 m of trench fill which documents an ice volume similar to the one observed sedimentaccumulated in <250 kyr, at a rate of 2800 m/Myr that today. Moreover, the uraniumseries dated coralsmarking the is twice as much the 1500 m/Myr rate documentedin the gulf last interglacial sea level high stand provide evidences that of Alaska at deepsea drilling Site 180 [Kulrne! al., 1973]. The sea level was above modern levels between 130 and 117 ka. trench fill increases in thickness to the north, from 700 m Thereforewe suspect that the Patagonian ice cap may have along line 3 to 1100 m north of the Guamblin fracture zone, suffereda massiveretreat between 130 and 117 ka allowing along line 1 (not shown). By contrast,the Chile ridge axial the river systems,including the Rio Baker river system,to be valley located in the southward extension of the Chile trench reactivatedas they are today in the Andes. It is commonly is nearly devoid of sediment.Active trench sedimentation in consideredthat the rate of delivery of terrigenous sediment the surveyedarea is controlledby the Chile ridge subduction increasesduring glaciation along high-latitude trenches[von that appearsto be dammingthe trench sediment supply Huene and Scholl, 1991]. In the specificsituation of southern originatingfrom the north [Bangsand Cande, 1997]. Chile we suspect that (1)major supply of sediment to the Along line 3, two reverse faults with <10 m offset are trench axis occurredduring the short interglaciation periods observed at 1.5 and 3 krn seaward of the subduction front, since 2-3 Ma and (2) conversely, glaciation periods were respectively. These high-dipping landward faults which characterizedby low sedimentsupply to the trench axis and crosscuttthe trenchsediment upper sequenceroot at depth in ice-rafted dischargesrestricted to the upslopearea, closeto the the lower transparentlayer. Similar tectonic elementsexist on shoreline. both lines 3 and 1, meanwhile lines 2 and 4 show trench sediment accumulation devoid of tectonic disruption. 5. Presubduction Segment (45ø10'-45ø50'S) However,we assumethat the two faults imagedalong lines 1 and 3 in the trenchinfill also exist alongline 2 but are covered The subductionfront (Figure 5) is sinuous resulting froma by the above described slump. Since the trench sediment successionof overlapping lobes protruding into the trench. accumulation shows little disruption and tectonism, we One of theselobes (Figure 6) between45ø23'S and 45ø32'S, attributethe trench infill to rapid influx of sediment.The-1.5 alongthe seismicline 2, exhibitsthe characteristicsof a slump krn spacing of reverse faults imaged along lines 1 and 3 8362 BOURGOISET AL.' CHILETRIPLE JUNCTION

76ø20'W 76ø 10'W 76øW 75o 50'W 75ø40'W

Fig. 6

Fig•7

[5 ø40'S /

/

o

- o50' S

,T-,r-- Figure5. Bathymetricmap(contours areevery 20 m)of the Chile trench and continental margin along the presubductionsegment. Locations of Figures6 and7 areshown. Darwin fracture zone (DFZ); Guamblin fracturezone (GFZ); volcanicridge (VR). Locationis shownon Plate 1. documentsthat accretion is currentlyactive along the trench theDarwin fracture zone to the northto theChile triple axis.Nearly all of the 700 m of trenchsediment is accretedto junctionto thesouth, is characterizedbya prominent scarp at the continentalmargin in two separatedsheets. The-10 m thetoe of the continental margin. This 250 to 750 m high scarp upwarddisplacement of the horizontallystratified sediment of (Figure8a) allowsus to easilyidentify the subduction front. thehanging wall shows no significant variation with depth Theacoustic imagery (Figure 9) andthe analysis ofbathymetry indicatingthat fault propagation occurred recently, after the allowto accuratelylocate the Chiletriple junction. The trench sequenceaccumulated. subductionof the Chile ridge spreadingcenter occurs at 46ø09'S,in an areaclose to a volcanicedifice which has been 6. SynsubductionSegment (45ø50'-47ø10'S) sampledby dredgingduring the CTJ cruise.This volcano protrudingwithin the CRS 1 axial valley is currently TheChile ridge subsegment (Figures 8a and8b), which underthrustingbelow the Chile continental margin or venting extendsfrom 45ø50'S to 46ø09'Sover a distanceof 35 km from throughthe toe of thecontinental wedge. BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8363

76øW 75ø55'W 75ø50'W The CRS I shows (Figure 10) a bathymetrictransverse sectionasymmetric with a landwardrift valley wall lying -100-150m deeperthan its seawardconjugate flank reflecting theaccumulated overburden of the lowercontinental slope as

A ' '•'- _ 45ø20'S enteringthe subduction[Bangs et al., 1992].Conversely, the seawardrift valley wall is a prominentrelief with rectilinear normal fault scarpsoffsetting the ridge flanks. This relief boundsa domainwith 55 to 120 m thick sedimentarycover west of the Chile ridge active domain which is devoid of sediment. The fabric of the ocean crust that extends 20-25 km west of the rift valley is dominatedby seawardtilted blocks _ 45ø25'S (Figure 10). Sedimentsaccumulated in the half graben •ne 2 structures exhibit fan-shaped turbidite accumulation onlappingolder tilted blocks. This areaof theAntarctic plate appearsto have been the locus of polyphasedextensional tectonics.At leasttwo pulsescan be recognizedin this area closeto the activespreading axis. Thetectonic history of the continentalmargin along the 45ø30'S Chileridge subsegment is well constrainedby ODP drilling data acquired at Sites 859-861 located on multichannel seismicline 759 (Figure8b). Accordingto Behrmannet al. [1994], subduction-accretionceased during the late Pliocene afteran importantphase of forearcbuilding which occurredin the earlyto late Pliocene.During the past-2 Myr, sediments of the downgoing oceanicplate have been subducted.The • 45ø35's subduction-erosion inferred at depth is consistentwith not o 5 km only the tectonic evolution [Bangs et al., 1992] of the I I I , I I seawardmostbasin sediment and normal faults identified along line 745 (Figure 8b) but also the tectonicregime of the continentalmargin to the north. Figure 6. Structural sketch of an area of the presubduction segment.Line with tick marksshows normalfault, line with The North Taitaocanyon subsegment extends from 46ø09'S open triangles shows thrust fault including the major to 46ø26'S(Plate 1). As to the north,a 100 to 150 m high decollement at the subduction front, and line with crosses scarpwith high backscatterin the acousticEM12 imagery showsanticline dome axes.A, B, and C indicate line drawing (Figure9)allows us to defineclearly the subductionfront, of profiles (thick line). Reversefaults imaged along profiles A alsocrosscutting the moutharea of the North Taitaocanyon. (seismicline l) and C (seismicline 3) are supposedto exist At 46ø12'S, the high-reflectivityterrane of the CRS I to the (dash line) along profile B (seismic line 2) beneath a slump north and the thick sediment accumulation which lobe protruding into the trench. See text for more details. Location is shown on Figure 5. characterizesthe Chile trenchto the southexhibit a sharp contactin the EMI2 acousticimagery (Figure 9). Indeed,line 9 (Figure 11), which lies closeto line 751 (Figure 8b) describedby Bangset al. [1992], exhibitsa trenchfill with a The axial valley of the CRS I (Figure 9), characterizedby considerably greaterthickness than to the north. This trench high-reflectivityterranes, is an active spreadingcenter. The fill is boundedseaward by a setof parallelnormal faults which averagewidth of the active domain is 3 to 6-7 km and offsetsthe top of the oceaniccrust by -500 m downward.The correspondsto a deepelongated graben devoid of significant trenchfill thickenssouthward exhibiting a-500 m maximum sedimentaccumulation. The trench axis on lines 6 (not shown) thicknessaccumulation along line 10, at the moutharea of the and7 (Figure10), which lies 5 and 3 km eastfrom the axis of North Taitao canyon. To the west, the sedimentaccumulation the spreadingridge, respectively,is also relatively free of is 100-180 m thick. As in the north, this sediment cover sedimentas it is just north along line 745 [Bangs et al., recordeda tectonicevolution through two extensionalpulses. 1992]. This situation is all the moreremarkable because the However,the fault systemhere trends -N0 ø, makingan angle Chile ridge axial grabenand the CRS 1 connectnorthward to of---20ø with theaverage direction of faultsbounding the CRS the Chile trench axis where sediment infill is hundreds of 1. Structuresalong line 10 (Figure12) aremore complex than metersthick. South of the Chile triple junction, the Chile to the northalong line 9. They includecompressional tectonic trench axis, which connects to the North Taitao canyon at featuresalong the trench axis associatedwith the N90-120øE 46ø21'S, is in the exact southward prolongation of the Chile trending ridge uplifted along a reverse fault at 46ø23'S ridge axial valley. Becausea N90-120øE trending ridge (Figures8a and 8b). This 100-120 m high ridge is associated (Figures8a and 8b) obstructsthe Chile trenchaxis at 46ø23'S, landwardwith minor ridgeswhich outline morphologically the detrital material bypassing through the North Taitao thrustsheets displaying an imbricatestack at depthsimilar to canyonhas to be carriedto the north along the Chile trench a flower structure(Figure 12). The reversefaults at the thrust axis and the Chile ridge axial valley. Since these two sheet boundariesroot at depth along a major normal fault morphologicaldeeps show no significant accumulationof belonging to a set of normal faults which marks the seaward sediment, we assume that little detrital material transits wall of the trench axis at depth. This indicates that tectonic throughthe North Taitao canyontoday. inversionoccurred along one of thesefaults, at least. 8364 BOUROOIS ET AL.: CHILE TRIPLE JUNCTION

^ ^ 3 CT IIL

VOLCANIC RIDGE '"'-- LINE3 A•• I1•_•. V.E. atseafloor--8 decollement• • • Nazcaplatesedimentarycover reversefaults .••' f--••'.•• •• onlapi horizontally trenchinfillstratified •' _

:• '•- '••• transparentlayer

(km) -10 o Figure 7. Trench section of the seismicline 3. Horizontal scalemarks distance from the subductionfront in kilometers. Note the reverse faults in the trench fill associatedwith incipient accretion. See text for more details. Line drawing inset shows Chile trench (CT); 1, oceaniccrust; 2, Nazca plate sedimentarycover; 3, trench infill. Location is shown on Figure 5.

The deformationfront along the North Taitao Canyon segments.The northernsegment (NS, Figure 8b) is in the subsegment(Figures 8a and 8b) has a sigmoidalshape with a southernextension of the Chile ridgespreading center and left lateraloffset of 2 km alongthe compressionalridges of the trendssubparallel to it. Thesouthern segment (SS, Figure 8b), Chile trench at 46ø23'S. The EMI2 imagery data (Figure 9) whichtrends nearly perpendicular to the northernsegment, show a clear connection between the main reverse fault parallelsa slightly bentridge located along the northernrim bounding these ridges to the south and the left-lateral strike- of the North Taitaocanyon. The morphology considered as a slip fault offsettingthe decollementlandward. This is in good whole,including the MSD andthe two US segments(NS and agreementwith the flower structure imaged along line 10 SS) thatboth exhibit a seawardconcave shape and a decrease (Figure 12). Along the continentalmargin this transpressional in heightto the northand to thesouth, is typicalof a slump. fault systembounds a prominentE-W trending ridge forming TheMSD resulted fi'om a slopefailure along a concave-upslip to the north the southernwall of the North Taitao canyon. surthceti'om which the US scaroriginated. The top of the dome North of the North Taitao canyon, the EMI2 bathymetric tilted backwardresulting in the quasi-reverseflat surfaceat map displaysthree main morphotectonicfeatures. From west to the top of the MSD; meanwhile,the hummockydisorganized east (Figure 8b) it includes the frontal ridge along the main topographyat the toe of the seawardslope protruded into the decollement,a wide dome-shapedrelief (i.e., the middleslope trench floor and was subsequentlyuplifted along the dome(MSD, Figure8b)) anda majorscarp (i.e., theupper scarp subductionfront. Small irregular-shapedmounds dot the hereafter(US, Figure8b)) with surfaceslope steepness of---12 ø, uplifted flat trench floor betweenthe subduction front and the dipping twice muchas to the north. Along seismicline 751 seawardslope of the dome-shapedrelief. The morphological [Bangset al., 1992], the frontalridge appears to be pushedup protrusionand the isolatedmounds on the flat upliftedtrench at the base of the trench slope. About 6-8 km toward the trench floor indicate a debris flow in the trench axis with isolated on the lower slope is a 600-700 m high scarp along the outrider blocks. However, since the US shows no clear seawardedge of the MSD which exhibits a strong asymmetry signatureat depth in seismicline 751, we assumethat these in sectionwith a peculiarflat plateaubelow the US. The US at morphological characteristicsare those of a detachmentfault the upper middle slope boundary exhibits two different dippingseaward, as previouslyproposed by Bangs et al. BOURGOIS ET AL.' CHILE TRIPLE JUNCTION 8365 m

J_l_..-!- 8366 BOURGOIS ET AL.' CHILE TRIPLE JUNCTION

76øW 75ø50'W 75ø40'W 45ø50'S

46øS

46ø10'S -•-:::•:•:•.•a'..", -...... -• ,•a."½•z'•- "*•.':*."• •":•:•:?•'*•;'• ::;•:j;•'•;:::...... • •*:.• ' ' :': e2. ":=•:.,,•.•.•..:.-•.•,•..,•:,. •... •:•::•..:..:,:•:•-,:::•:•.• .•s'.•a•.•.•Z•.•.•..•: .... •:-':.•-•.. •-..•:•:•:•=.'

ß*[• [•:' ,' ....;•;• .•[• *:.•: .:;•'-...'•:'•... .•½,•,.. •..:[•.•:.;.-•..•,...'r;.-...;.•..•.)•.-•:'-:•=.::•::-•.-::,.•,'.....*':.:':x-:,....'•.•' ..--•,z•.

'*•* , ...... ,..• .- ..... ß...... */X•:-.,.-.... , .... •...... ,...... , ...... ?**•,,...... •... *x•,• .... ;. - f• ...... '.,:',..,:..'z:..,.-..::"-•,.:,', -'..:,.•:S, •,...... '.:,.;•,,:...... -=• ...... ::. - •;.'**...... -;• ...... '•:•t,• •:•-•...... :.•. ' *•ß...... --,ß '-•'•'::*. •½•'.-•...... •.' ':.'..•-::'"-.'.*:*•. :. *..•:. :';*•'• .,,;' •.:::. '-• :..• ...... '- "•:'<*:: *.½.i:•'-..::*,'"".'*:'::* ß...r•.•.:•-•.:...... *::::;•:'=*•'••.• :• • r'.'..':•:.. '•...'•..... '" •. -- -. •., .. .• ...... , ...... ,..;..;$...... '•.•:...... ,:,...... •... ;...... -...... ";..'•;.-...... *:*'::*'::::":':**"*:"* ""•:":":'"* ...... *""'" "'"•' f•*;'"*:• ...... **$•":•;;•'5=•?"%:;.'•:. _

I 46ø20'S

Figure 9. EM12 imageryof the Chile ridge subsegment.Note the high backscatterof the axial valley of the CRS 1. Chile ridge(CR); Chile trench (CT); Chile triple junction (CTJ); decollement(D); normal fault (NF); upperslope scarp (US). Seetext for moredetails. Location is shownon Plate 1.

[1992] and Behrmannet al. [1994]. As a consequence,they structures of the Chile continental margin. It exhibits an speculatethat outcropof the continentalmargin metamorphic asymmetricaltransverse cross section. The southernwall of the basementshould exist along the upperscarp. This assumption canyon is 750-1250 m higher than its northern wall, was verified during the CTJ cruise.Dredging (CTJ23, Figure suggesting that subsidence occurred to the north. Five 8b) along the US allowed us to recovermetamorphic rocks dredges(Figures 13a and 13b) have beenperformed along the similar to the pre-Jurassiccontinental basementof the Tres canyon walls. Two dredges(42 and 44, Figure 13b) located Montes and Taitao peninsulas(Figure 3). along the southern wall at 10 and 15 km landward from the The E-W trending middle course of the transverseNorth trenchaxis recoveredpre-Jurassic metamorphic rocks similar to ßTaitao canyon (Figure 13a) is 1250-2000 m deep. This those sampledonland [Mpodoziset al., 1985; Forsythe et al., prominentmorphological feature opens a window into deep 1986; Bourgois et al., 1993] in the Taitao and Tres Montes BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8367

w E CRSl LINE 7 V.E. at seafloor = 8

top oceanic decollement•

4

5 - 30 (km) - 20 - 10 0

Figure 10. CRS I and adjacentseafloor sections along seismic line 7. Horizontalscale marks the distancefrom the decollementin kilometers.Note that the landwardrift valley wall is deeperthan the seawardwall. The tilted blockswest of the Chile ridge evolvedthrough two tectonicpulses (NF1, normalfault of the first extensionalpulse, and NF2, normalfault of the secondextensional pulse). Line drawinginset: 1, oceanic crust;2, sedimentarycover of the oceaniccrust; 3, old turbidites(syn-NF1); 4, youngturbidites (syn-NF2). Seetext for moredetails. Location is shownon Figure8b.

peninsulas (Figure 3). Two other dredges (19 and 43, Figure canyon,recovered volcanic rocks and sediments[Guivel et al., 13b) located along the northernwall of the canyonat the same 1999] similarto thosesampled along the northernwall of the distances from the trench axis have recovered volcanic rocks canyon.The sedimentswhich unconformably overly the Chile and sediments having affinities with rocks of the Pliocene continentalmargin basementextend to within <5 km landward suites of the Chile Margin Unit and the Main Volcanic Unit fromthe trenchaxis. Becausethe North Taitaocanyon fault described onland in the Taitao peninsula [Bourgois et al., showsno offsetof the subductionfront along its westward 1992, 1993; Guive! et al., 1999]. This is in good agreement prolongation,it musthave no significanttectonic activity with subsidence of the northern wall of the canyon with today. respect to the southern wall. We therefore assumethat a The Taitao ridge subsegmentlocated off the Taitao and Tres northward dipping major fault with a normal dip-slip Montes peninsulas extends from 46ø26'S to 47 ø10'S. It componentof•l km as a minimumfollows the canyon course. includesthe prominentTaitao ridge (Figure 13a) regarded as a A dredge (17, Figure 13b) located only 5 km landward from fragmentof oceaniccrust in theprocess of emplacement[Leslie, the trench axis, along the southernwall of the North Taitao 1986]. Seismicdata along line 762 (Figure13b) revealthat the

w 3. 2 '• --1 E ^ V.E.atseafloor =8

:'.•.."-'::•:' % . decollement•CT ,

topoceanic cm•t - 20 -10 0 Figure11. Chiletrench (CT) andadjacent seafloor sections along seismic line 9. Horizontalscale marks the distancefrom the decollement in kilometers. Note the highly deformed thick accumulation ofsediment along thetrench axis. The tilted blocks west of the Chile ridge evolved through two tectonic pulses (NF1, normal faultof the first extensional pulse, and NF2, normal fault of the second extensional pulse). Line drawing inset: 1,oceanic crust; 2, sedimentarycover of the oceanic crust; 3, oldturbidites (syn-NFl); 4, youngturbidites (syn-NF2).See text for moredetails. Location is shownon Figure8b. 8368 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

2 W g

g V.E.atseafloor: 8

(km) -20 -10 0

Figure 12. Chile trench (CT) and adjacentseafloor sections along seismicline 10. Horizontal scale marksthe distance fromthe decollementin kilometers.Note the thick accumulationof sedimentalong the trench axis. The set of activereverse faults (thick line) exhibitsthe characteristicsof a flower structureparticipating in the accumulationof thrust sheetsresulting in the constructionof a 100-120 mhigh ridge which obstructs the trench(see Figure 8). The oceaniccrust is down offset alongmajor normalfaults (thin line). Inversion tectonic occurredalong one of thesefault, at least. Line drawing inset:flower structure(FS); normal fault (NF); reverse fault (RF); 1, oceaniccrust; 2, trenchinfill and sedimentarycover of the oceaniccrust. See text for more details. Location is shown on Figure 8b.

oceaniccrust west of the section is coveredby at least 700 m of theAntarctic plate at thetoe of theNorth TaitaoRidge fault of sediments.To the east,the Taitao Ridge appears to represent is characterizedby a gentleslope dipping steadily to the west, a majortectonic slice of oceaniccrust that may be the seaward from2500 to 3250m depth.As the Antarcticseafloor deepens extensionof the Bahia Barrientosophiolite [Bangs et al., to the west, the fault scarpincreases in height along strike, 1992]. Accordingto Behrmannet al. [1994]the volcanicpile fromno significanttopographic signature at point X (Figure drilled at ODP Site 862 (Figure 13b) at <14 km seawardfrom 13b) to the eastto >1200 m at point Y. Bahia San Andres was probably formedat a ridge-transform The EMI2 map (Figure 13a) strongly suggeststhat the intersectionand representsa possiblecandidate to becomean Taitao Ridge consistsof accretedtectonic slices markedin the ophiolite body emplacedinto the SouthAmerican forearc. morphologicaldata by two ridges(RI and R2, Figure 13b). The North Taitao Ridgefault (NTRF, Figure13b) bounding The RI and R2 ridgeswhich trendNI45øE, are 7.5 and 18 km the Taitao Ridge to the north trends N55øE. Becauseit is long, respectively,and exhibit a smooth morphological rectilinearin the bathymetricmap, the North TaitaoRidge fault signature. They can be interpreted as two thrust sheets of must have a subvertical attitude. This major fault exhibits a trench sedimentstacked at the toe of the main scarp(MS, scarpwith decreasingoffset to the east and finally no clear Figure 13a) of the Taitao Ridge. Becauseit exhibits a lobe morphologicalsignature along the Chile continentalmargin. shapein map view and closely follows the 3500-m isobath, Conversely,it connectsseaward to an E-W trending flexure the frontal decollement must have a subhorizontal attitude at (Plate 1) that is a conspicuousfeature of the Antarcticaplate depth, at least beneaththe two frontal thrust sheets,i.e., R1 extending at distances as far as 100 km to the west, at least. andR2, Figure13b. Along line 12 (Figure 15), the top of the North of the North Taitao Ridge fault, line 11 (Figure 14) oceaniccrust can be seendipping landward,suggesting that showsthat the oceaniccrust is coveredby sediments400-450 the oceaniccrust is underthrustedunder the TaitaoRidge. In m thick. Insteadof dippinglandward beneath the Taitao Ridge this area,active trench floor deformationincluding a reverse as seealong line 762 [Bangs et al., 1992], here the top of faultat 3 km seawardfrom the subductionfront strongly oceaniccrust can be seento tilt seawardas it approachesthe supportsthat subductionis activebeneath the TaitaoRidge. ridge. Thesetwo differenttectonic responsesof the oceanic This high-dippinglandward fault, which roots at depthin a crust along the North Taitao Ridge fault suggestthat strike- acoustically"transparent layer" (called unit T in section7), is slip movementoccurred along it. The along-strikebathymetry partof a fold andthrust belt developingseaward. Because the BOURGOIS ET AL.' CHILE TRIPLE JUNCTION 8369

76ø10'W 76øW 75ø50'W 75ø40'W 75ø30'W

46ø20'S

D

46o25'S

MST

46o30'S

TR

BA MS 46ø35'S

STC TP

A

NTC

46o20'S \ \ 42

Fig.1,,• 46o30'S

MSTS

\

decollem•

15

Figure13. The Taitao Ridge subsegment along the synsubduction segment. (a)Bathymetric map (contours are every20 m).(b) Structural sketch of the area shown in Figure13a. Note that the structural sketch includes EM12imagery data shown in Figure16. Line with tick marks shows normal fault; line with open triangles showsthrust fault including the major decollement atthe subduction front; solid dot marks the ODP Site 862. LocationsofFigures 14 and 15 are shown. Bahia San Andres (BSA); Chile trench (CT); main scarp (MS); middleslope terrace (MST); main scarp thrust sheet (MSTS); North Taitao canyon (NTC); North Taitao Ridge fault(NTRF); ridge 1 (R1);ridge 2 (R2);South Taitao canyon (STC); Taitao peninsula (TP); Taitao ridge (TR). Dredgesites (11, 13, 14, 16, 17, 18, 19, 42, 43, 44, 45, 46, and the thick short lines in Figure13a). Letters X andY referto descriptionsin the text. See text for more details. Location isshown on Plate 1. 8370 BOURGOIS ET AL.' CHILE TRIPLE JUNCTION

w 862 E SOUTH AMERICA PLATE

well-stratified sediment LINE 11 V.E. at seafloor = 8

acoustic basement

/ North Taitao Ridge fault

.E • ANTARCTIC PLATE 4-- > NORTH TAITAO RIDGE

topoceanic crust 0 10 km I I

Figure 14. The Taitao Ridge and the adjacentseafloor along seismicline 11. Note that a well-stratified sedimentarycover overlies the Taitao ridge basement. Site 862, ODP site drilled during Leg 141.See text for more details.Location is shownon Figure 13b.

w E W E

LINE 12 V.E. at seafloor = 8 decollement reverse faults

A A A A A A A \ i I i (km) -10 topoceanic crust0 (kin) -10 0

Figure15. Chile trench along (left) seismic line 12 and (right) line drawing showing the oceanic crust and the trenchinfill activelyunderthrusted beneath the TaitaoRidge. Horizontal scale marks the distancefrom the decollementin kilometers. See text for more details. Location is shownon Figure 13b. BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8371

76 ø O'W I_1 I I I I I I i_ [ i I I LI _1 _l I I I I L_I I I I I I I .

0'S

F

}'S

TR

Figure 16. EM 12 imageryof the Taitao Ridge subsegment.Note the low reflectivityof mostof the area covered by the Taitao Ridge, including the scarpalong the North Taitao Ridge fault. Acoustic basementof volcanic origin (AB); Chile trench (CT); decollement(D), fault (F); North Taitao canyon (NTC); North Taitao Ridge fault (NTRF); ridge I (R 1); ridge 2 (R2); Taitao Ridge (TR). See text and Figure 13 for more details.Location is shown on Plate 1.

seaward thrust sheet shows no disruption and tectonism at the terrace(MST, Fig. 14A) that parallelsthe North Taitao Ridge trench floor, we assumethat the thrust sheet belt developed in fault. To the southeastthe midslopeterrace also parallels a two phases separatedby a rapid influx of sediment in the scarp which bounds the acoustically structurelessmound trench. Instead, the trench fault and thrust belt are in the drilled during ODP Leg 141. The midslopeterrace deepens southeastward prolongation of the Taitao Ridge frontal landward along strike, from 1500-1750 to 2100-2300 m. In the accretionary prism (RI and R2, Figure 13b), they do not EM 12 imagery(Figure 16) the scarpalong the North Taitao connect with each other; the frontal decollement divides them. Ridge fault (NTRF, Figure 16) exhibits the same low- A major feature imagedalong line 11 (Figure 14) consistsof reflectivity.signature as the midslopeterrace, evidencing that the well-stratified pile of sediment covering the acoustic the Taitao Ridge sediment comes into contact with the basementof the Taitao Ridge, west of ODP Site 862. In the Antarctic plate sedimentarycover east of 75ø56'W. Line I 1 EMI2 bathymetric map (Figure 13a), this sedimentary (Figure 14) shows the acousticbasement of the midslope sequence is associated with a N55øE trending midslope terracecropping out alongthe scarpof the North Taitao Ridge 8372 BOURGOIS ET AL' CHILE TRIPLE JUNCTION

76ø20'W 76ø10'W 76øW 75ø50,W 75ø40'W . I I I I I . I I I I I I I I I I I I I I I I . 46ø40'S

46ø50'S

47øS

47ø10,S

Figure17. Bathymetricmap (contours are every 20 m)ofthe Chile trench and continental margin along the TaitaoRidge subsegment off Taitaoand Tres Montes peninsulas. Location of Figure18 is shown.This area fits a majorreentrant of the continentalmargin between the TaitaoRidge to the north and the Golfo de Penas accretionaryprism to the south. See text for more details. Location is shownon Plate 1.

faultat 75ø58'W. This acousticbasement (AB, Figure 16) those cropping out in the Taitao peninsula [Guivel et al., exhibitsa high backscattersignature in the EMI2 imagery 1999]. They include lavas with normal(N-type) mid-ocean thatcan be followedsoutheastward along the mainscarp (MS, ridge basalt to calc-alkaline, andesite, and dacite Figure13a) of theTaitao Ridge that trends parallel to the two compositions. The volcanic basement and associated ridges(RI andR2, Figure13b) of the accretionaryprism west sedimentsdrilled at Site 862 arethe seawardprolongation of of the TaitaoRidge. the volcano-sedimentarycover of the Taitao peninsula,i.e., ODP Site 862 drilled a 20-m-thicksedimentary layer of Chile Margin unit and Main Volcanicunit (Figure3). When Plioceneage overlyingvolcanic rocks of Plioceneage associatedwith the EM12 acousticand bathymetricdata and [Forsytheet al., 1995] with chemicalcharacteristics similar to seismicrecords along lines 11 and762, theseresults suggest BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8373

decollement E w LINE 13 Morningtoncanyo V.E. at seafloor = 8

seaward tilted blocks oceanic crust

(km) -30 -20 -10 0 Figure 18. Chile trenchalong seismicline 13. Horizontalscale marks the distancefrom the decollementin kilometers.The trenchinfill onlapstwo major seawardtilted blockscovered by a transparentunit. Seetext for more details. Location is shown on Figure 17.

that the Chile margin sedimentary cover extends seaward to contrast,the Antarctic plate basalts beneath the trench fill are the inner scarpbounding the Taitao Ridge midslope terrace. block faulted togetherwith a thin cover of sediment.Since the Becausethe midslope terrace sedimentsexhibit a consistent trench sedimentclearly onlapsthese extensionaldeformations, along-strike dipping attitude (55øE) and a seismic facies we may assumethat basement faulting predates the trench (Figure 14) similar to thoseof the adjacentAntarctic plate, we accumulationand therefore is not subduction-related.Finally, assumethat the Taitao Ridge accretionaryprism includes a the oceanic crust can be seen to dip consistently landward, thick midslopethrust sheet(MSTS, Figure 13b). Since dolerite and the trench sediment is little disrupted in the approach of and basalt were dredged along the scarp of the North Taitao the decollement supporting the supposition that subduction Ridge fault at 75ø58'W, the acousticbasement of the midslope is active in a nonaccretion mode along the Taitao and Tres thrust sheetimaged along line 11 is assumedto be of volcanic Montes peninsula transect. The Pliocene Cabo Raper pluton origin. (Figure 3) which is locatedwest of the Taitao and Tres Montes The section of the Chile margin south of the Taitao Ridge peninsulas, has a culminating point (800 m above sea level) (Figure 17) is characterizedby a major reentrant between <2 km from the shoreline. The dip slope between the 46ø35'S and 47ø10'S (Plate 1). As a consequence,at 46ø50'S culminating point and the shoreline is along the extension of the Chile trench axis lies <10 km seaward of the shoreline of the continental slope between 1750 and 2250 mwater depth the Taitao peninsula.This reentrantand the associatednarrow (detail map not shown),along the sametransect. This suggests continental margin coincide with prominent morphotectonic that the Cabo Raper pluton extends seawardto within 6-7 km anomalieson both seawardand landward sides including the landward from the trench axis. Taitao Fracture Zone ridge (Plate 1) and the Taitao and Tres Montes peninsulas, respectively. The Taitao Fracture Zone 7. Postsubduction Segment (47-48øS) ridge is a conspicuous E-W trending featureof the Antarctic plate. It extendswest of 76ø10'W throughout the surveyed Seven seismiclines (15 to 21, Figure 2) were shot along the area being at least 100 km long and culminates 250-300 m Golfo de Penas accretionary prism (Figures 19 and 20) in an abovethe adjacentseafloor. This ridge is a southwarddipping area extending fromthe Tres Montes peninsula (47øS) in the monocline bounded to the north and to the south by the north to the Isla Campanaat 48øS (Plate 1). The area includes Taitao and the Tres Montes fracturezones, respectively. Line the Humboldt abyssal plain, the Mornington channel, deep- 13 (Figure 18) located on the landward prolongation of the sea fan sedimentary bodies along the Chile trench axis, the Taitao Fracture Zone ridge exhibits a 750-1000 m thick lower slope accretionaryprism, the upper slope ridges,and the accumulationof trench sedimentunconformably overlying the upper slope basin. When compiledwith the EM12 bathymetry oceanicbasement. Although the trench seafloor shows a clear and the acousticimagery (not shown), the seismicdata permit seawarddip, the trench sequenceexhibits a consistent quasi a three-dimensional interpretation of the structures and their horizontal attitude and little tectonic deformation. By evolution through time (Figures 20, 21a, 2lb and 22). 8374 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

76ø40'W 76ø20'W 76øW 75ø40'W 77'W

Fig. 20

47ø40'S

Fig. 20

Figure 19. Bathymetricmap (contours are every 50 m)of the Golfo de Penas accretionaryprism along the postsubductionsegment. Location of Figure 20 is shown.Location is shownon Plate 1.

The smoothflat seafloorimaged west of line 19 is part of the can be divided into three units with different acoustic Humboldt abyssal plain which extendsbetween the Pacific signatureslabeled unit 1, 2, and 3 frombottom to top. Unit 1 rise and the SouthAmerican continentalmargin southof 47øS. can be divided into subunits 1A and lB from bottom to top. Three volcanoes located at 77øW, 47ø30'S (Figure 22) They are 100-105 and 55-60 m thick, respectively.Subunit I A underline the crest of a N-S trending low ridge bounded by consistsof parallel strong reflectionsand constantthickness the depressionparalleling the Tres Montes fracture zone directly overlying the basement.Top of subunit lB is (TMFZ, Figure22) to the north.In line 18 (Figure 2), at about underlineby a strongreflection extending from lines 16 to 19 77øW longitude (Figure 23), the sedimentarycover of the (not shown) over a distanceof-50 km. SubunitlB exhibits an Antarcticplate exhibitsa constantthickness of-750-775 m. It irregular thickness. It appearsas channels scooping out the BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8375

76ø30W 76øW I

• STRUCTURESDIAPIR

47ø30S

•w•e19

48øS

Figure 20. Sketchmap showingthe major morphostructuralelements of the Golfo de Penasaccretionary prism. Golfo de Penasaccretionary wedge (GPAW). The area locatedeast of lines 19 and 20 with irregularopen dots showsa hummocky morphology interpreted as the signature of ice-rafteddischarges. Location is shown on Figure 19.

subunit IA. Unit 2 is composedof two subunits which are Golfo de Penas and the Cabo Bynoe canyons to the south. acousticallysemitransparent. From bottom to top,subunits 2A Since the South Taitao canyon and the Cabo Raper canyon and 2B are 70-75 and 100-105 m thick, respectively.Subunit show no connection to onland river systems,we assumethat 2A shows short-spacedcoherent and thin reflections, while most of the detrital material bypassing the Mornington subunit 2B shows no coherent reflection. Unit 3 consists of a channel originates from the Golfo de Penas area today. Very 425-430 m thick accumulationof mainly parallel and constant strong reflections characterize not only the present-day thickness reflections. In most cases, reflections are continuous sedimentation of the Mornington channel but also the over distances up to tens of kilometers. According to the sequence located beneath the channel. This 30-35 m thick magneticage of the oceanicbasement [Leslie, 1986; Tebbenset sequenceresulted fro TM an accumulation of sediment infill of 1- al., 1997] we assumethat units 1, 2, and 3 show a sediment 2 to 10 m deep channelsburrowed by sedimentoriginating record of the past 10 Myr. from the Mornington channel meanderingalong the Chile The high backscatterwhich characterizesthe Mornington trench axis. Along line 19 (Figures 21a and 2lb), the strong channelin the acousticEMI2 imagery [Bourgoiset al., 1997] reflectorsproduced by the sedimentaccumulated along the is interpreted as the signature of the rough sediment surface Mornington channel can be followed landward not only developed in association with the erosion and transport of beneathdeep-sea fan IV but also farther to the east below the detrital material along the channel.The Mornington channelis lower slope. The Mornington channel sequenceappears as an typically 5-10 km wide, exhibiting levees on either side. The heterochronoussequence that migrated seawardin relation to main tributaries (Figure 22)are identified to be the South the developmentof both deep-seafan IV and the lower slope Taitao canyon and the Cabo Raper canyonto the north and the accretionarywedge. 8376 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

LINE19 I V.E.at seafloor = 8 .! "•--"--•.•---.-•----.4•.-?.•....••

'.' '.-•:-,--" -• -:...-•_.' - • ?--• -.'• -• - -.... '•-:'_--:•.•.-..... " - '-- -, m,.•...... --•-.--•'-.. -- -•,:-•:-"':--,-'-&' ..... ,'. --- - '•.C•-g•.- ',•.• -•-- -•t•.•'-.'-•--•-:•-,•-•..-,•...... -'-.•.,[.•2•--:•-----:½: ....•-• .... .•:.----:•-•'z•_2}..-•__•-,: ~ =. -;.'_,

- 26 - 20 - 10 0 10 20 24 30 40 (kin), .... I I i I i I I I I

.o o UpperSlope ridges '-- & innerridge 2-- CHILETRENCH .9 •FOLowerDEPENAS Slope ACCRETIONARY• outer PRISM,.._•

__ Mornington channel Humbold I abyssalplain!• deepseafanlV • •

.•--.---• decollement B olderMornington channel deposit •-- topoceanic crust

Figure21. (a) Seismicline 19 along the Chile trench and Golfo de Penasaccretionary prism. Horizontal scale marksthe distancefrom the decollementin kilometers.(b) Line drawing showing the main sedimentbodies andmorphostructural elements of this area.Location is shownon Figure22.

In the EM12 bathymetry(Figure 19) deep-seafan IV (Figure better preservation than those of deep-sea fans I and il 22) exhibits a typical lobe shape. It consists of a package of reshapedby the Mornington channel cutoff. The crest of the continuous reflections (Figures 21a and 2lb) that range from fan-shapedbody of deep-seafans II, IV, and V that do show the subhorizontal to gently seaward dipping extending from the distributary canyon outflow exhibits no connection to active Mornington channelto the decollementat the toe of the lower canyonsupstream. The decollementand the associateddiapirs slope. The seaward end of the fan wedge exhibits cut off the distributarychannels of the deep-seafans built into interfingering of fan turbidites with the Mornington channel the abyssal plain. EM12 bathymetry allows us to follow the sequences.To the east,the seismicfacies of deep-seafan IV can morphologicalcrest of the deep-seafan landward, along the be followed through the lower slope to the backstopfront, i.e., Lower Slope accretionary wedge. Thereforewe assumethat the front of the Upper Slope ridges domain (see below) acting building of the deep-seafan building predatesthe tectonic as a buttress[von Huene and Scholl, 1991], over a distance of developmentof the Lower Slope accretionarywedge. At the 50 km. To the east, the corresponding sequencesare 1600 to time the Lower Slope accretionarywedge began to form,Cabo 1900 m thick. The along-strikeextension of deep-seafan IV is Bynoe canyon, considered as the main tributary outflow for •40-50 km. Five imbricated deep-sea fans having a similar deep-seafans III and IV, changed its courseto flow along the location both in space and time with respect to the main present-day channel. structural features of the area are identified. From north to Below deep-seafan IV (Figures21a and 2lb), the Antarctic south,they are termed deep-seafans I, II, III, IV, and V (Figure plate sedimentarycover exhibits a mostly transparentseismic 22). The typical high-amplitude reflectorsof deep-seafans III, facies (unit T hereafter) with some parallel configurations of IV, and V exhibit the original morphological signature with poor continuity and low amplitude in its lower part. Unit T BOURGOIS ET AL.' CHILE TRIPLE JUNCTION 8377

77øW 76ow

'rFZ

47øS

Fig.

Fig. 21

... • 48os øoø --• o

GOLFO DE PENAS ACCRETIONARY PRISM I I I Figure22. Main drainage (line with solid arrows) bypassing theGolfo de Penas accretionary prismtoday and deep-seafans (open dots are seaward limit and open arrows are crest of fan body) along the Chile trench south ofthe Taitao Ridge. Cabo Bynoe canyon (CBC); Cabo Raper canyon (CRC); deep-sea fans I to V (DSFI to DSF V,open dots are seaward limit); Golfo de Penas canyon (GPC); North Taitao canyon (NTC); South Taitao canyon(STC); Taitao fracture zone (TFZ); Tres Montes fracture zone (TMFZ). Crosses along line show the seawardlimit of theMornington channel deposit; Dash-dotted line shows the seaward limit of thetransparent layerunit T, darkpattern isthe main morphostructural reliefof the Golfo de Penas accretionary prism, diagonal shadingshows the main Antarctica plate relieves. Locations ofFigures 21and 23 are shown. See text for more details. Location is shown on Plate 1.

extendswidely (Figure 22) throughoutthe surveyedarea: (1) and lr identifiedby Leslie [1986] and Tebbenset al. [1997] at from line 12 to 21 over a distance of •-150 km in a N-S 46øS extends southward to the Taitao fracture zone, in an area directionand (2)from the backstopfront to a seawardlimit where unit T exists. We infer that unit T is Lower Pleistocene determinedby theAntarctic plate morphology. North of the at the oldestand can be possiblyas youngas 1 Ma. Taitao fracturezone, unit T overlies an oceaniccrust younger The Golfo de Penasaccretionary prism (Figures 19, 21, 22, thananomaly 2. Moreover,the boundarybetween Chrons 1 n and 23) exhibits two differentdomains with distinctive 8378 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

sedimentcovering the deep-seafans, i.e., deep-seafan IV along line 19. Moreover, the two main basins located along line 19 show no significant differencesin their deformation,both sufferinga similar amountof deformationand uplift. Tectonic 5 deformationof the Lower Slope accretionarywedge is mainly concentratedalong the diapir structures.They accommodated most of the convergence between Antarctic and South American plates that occurredafter deep-seafan accumulation. The uplift of the frontal diapir structure associatedwith the decollement(Figures 21a and 2lb) can be estimatedfrom the upward shift of the Mornington channel sequenceat depth. The amountof displacementalong the decollementis -•40-50m at this site. When associated with the state of deformation of thefrontal diapir, it suggeststhat deformationbegan recently at this site. Conversely, the more evolved deformation of diapirslocated landward, as shown by no clear reflection occurringat shallowdepth, suggests that deformationbegan 2B earlierlandward. Although the frontaldiapir showsa more evolvedsignature along strike (for example,along line 20, not shown),we assumethat deformation began earlier landward as lB also suggestedby a higher uplift of inner diapir structure. However,the EM12 imageryshows active erosion occurring 6 1A along all elongatedridges, suggestingthat all diapir structures are suffering active uplift. Therefore active shorteningexists not only along the decollementbut also throughoutthe Lower Slope accretionarywedge, along the diapirsstructures. The meantotal net shorteningof the Lower Slopeaccretionary wedge calculated along lines 17 to 21 is -3 km. If we considerthat the lower slope shortening accommodatesthe whole 2.4 cm/yrnet convergencebetween the Antarcticaand the South Americaplates, its deformation would begin at 125 ka at the earliest. The Outer ridge and the Inner ridge of the Upper Slope Figure 23. Seismostratigrafyof the Antarctica plate ridgesdomain form a prominentrelief which is a majorbarrier sedimentary cover from seismic line 18. See text for more for detrital materialoriginating from the Golfo de Penasarea. details.Location is shownon Figure 22. The Golfode Penasand the CaboBynoe canyons (Figure 22)

morphologicalsignatures: (1)the Lower Slopeaccretionary DISTANCEFROM THE TRENCH (km) 0 20 40 60 wedge located seawardshows a structuralpattern that I I consistsof elongatedridges separated by relativelyflat basins G.P. ACC. PRISM w trendingN-S with a 3ø meanslope angle and (2) to theeast the • UJ •Z _ 1.5 Upper Slope Ridgesdomain exhibits a N I5-20øE trend and a 6ø meanslope angle. The Upper SlopeRidges domain which IlL I (,• ß rfl .....':i:!:' extends over a distance of-•100 km consists of two massive o• ,,.,I •; n ca•'1 ' wI.i.'--/ n-• •'•'."'::': I ...... ':'::::::.':'"'...... "'•"'"'"Z ridges:the Outer and the Inner ridges.The crestof the Outer z: • • I •- /.'"J'..'""'•• ridgeis in waterdepth of about2000 m belowsea level, i.e., 1000-1250m higherthan the basinsof the Lower Slope accretionarywedge; meanwhile, the Inner ridge crestis in -750-1000 m water depth. The elongatedridges of the Lower Slope accretionary wedge(line 19, Figures21a and 2lb) showroots in diapirsat depth. The transparentseismic facies with no clear reflector makingthe mainbody of the diapirsconnects downward to unit T whichprobably consists of undercompactedmud. The frontaldecollement also clearly connects to a diapirstructure at depth,showing that unit T played a peculiarrole in the tectonicstyle and evolution of the LowerSlope accretionary Figure24. Longprofile of the Golfode Penas canyon (GPC) wedge.The flat basins located between the elongatedridges andCabo Bynoe canyon (CBC) alongtheir coursesacross the exhibitvery weak deformationat depth and no significant Golfode Penasaccretionary prism. Back stop front(BSF), differencesin the amount of deformationwith depth indicating subductionfront (SF). Seetext for moredetails. Locationsof that the deformationof this areapost datesthe mostrecent canyonsare shownon Figure 22. BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8379 cutting acrossthe Upper Slope ridges belong to one of the have operated along this segment of the margin: (1) the major transverse routes to sediment feeding into the Chile episodicinput of sedimentto the trench axis in relation with trench.They connectto the east to the onland river systemof the evolution of the Andean ice cap, and (2) the tectonic the Rio Baker (Figure 4)which drains the Argentina Andean reorganization at the Nazca-Antarctica plate boundary foothills. Thereforethe Rio Baker river systemand its offshore associated with the Chile ridge subduction. These two distributary prolongation are antecedent drainage with processesand their interaction largely determinethe tectonic respectto the uplift of both the Andes and the Upper Slope regime of the continental margin (subduction-erosion versus ridges.The long profiles (Figure 24) of the Golfo de Penasand subduction-accretion) and its evolution through time. Cabo Bynoe canyon (Figure 22) show three main downstream However, the topography of the subductingAntarctica plate slope breaksbounding four different sectionsthat characterize is suspectedto have importance in both creating the major the morphotectonic domains of the Golfo de Penas reentrantsouth of the Taitao ridge and the active uplift of the accretionary prism, from east to west: (1) a section with a Taitao and Tres Montes peninsulas. Also, the thick rathersteadily decrease of the slopedownstream typical of the accretionary prism imaged along the Taitao ridge transect Upper Slope basin, (2) a section with a steepermean channel allows us to better identify the processes that induced slope and major differencesin the general shape of the long ophiolite emplacement,i.e., obduction, into the continental profile of the two canyonsthat characterizedthe Upper Slope margin.Another major thematicquestion of the Chile margin ridgesdomain, (3) a steepsection with an upwardconvex long triple junction area is that of the Bahia Barrientos ophiolite profile acrossthe lower slope, and (4) a flat sectionalong the emplacementas highlightedfrom the Taitao ridge structure. Chile trenchaxis. The long profile of the Upper Slope basin is consistentwith relative subsidenceof this areaas regardthe 8.1. Sediment Supply adjacent downstream domain. The sedimentsare actively trappedin the Upper Slope basin. Along the Golfo de Penas The seismic lines along the postsubduction segment canyon,the steepgradient of the slopeand the upward convex documentan episodic development of the Golfo de Penas shapeof the long profile of the Upper Slope ridges domainare accretionary prism. The accumulation of more than 50% of indicative of active deformation and uplift. Along this deep-seafans I to V alongthe Lower Slopeaccretionary wedge transect, the major fault of the Golfo de Penas accretionary began at 125 ka at the earliest. The tectonic style which is prism, evidencedfrom the main slope break, is located at the similar to other rapidly accreting margins with thick trench back stopfront. Along the Cabo Bynoecanyon, the main slope accumulation,for example,along the Aleutian arc [McCarthy break occurs west of the Lower slope accretionary wedge at and Scholl, 1985], is deeply marked here with diapirism. the subductionfront. Here the tectonicdeformation is actively Diapirs root at depth in a layer with a transparentseismic transferred westward, from the back stop front to the signature, i.e., the unit T, probably composed of subductionfront. The major fault at the Inner ridge-Outerridge undercompactedmud. Along line 19 (Figure 21), frontal boundaryis not associatedwith a break in the slope gradient accretion added almost all the trench fill section to the lower along the streamof both canyons. We assumethat this fault slope. Offset of trench turbidites clearly defines the frontal has no significant tectonic activity today. The Outer and Inner thrust. This thrust deeply penetratesthe trench sequence ridgesare acting as a distinct block with regard to the active down to unit T where it is supposed to sole into the uplift of the Upper Slope ridges domain. Since the Outer and decollement.We assumethat much of the incoming trench the Inner ridges are along the southward prolongation of the materialwas accretedin front of the Upper Slope ridge back Taitao and Tres Montes peninsulas,respectively, we speculate stop. Only the Antarctic plate sedimentcover below unit T that basementrock cropping out on land may extend offshore waseventually underthrust beneath the backstop.Along line along the upper sloperidges. It should also be noted that the 19, frontal accretion added 0.8 to 1.8 km of the trench fill Taitao and Tres Montes peninsulasare both actively uplifting sedimentto thetoe of the continentalmargin. Reconstructing [Bourgois et al., 1996]. the situationbefore tectonic deformation ofthe LowerSlope accretionarywedge (Figure 25) allows us to show that the 8. Discussion horizontally stratified turbidites extended from the .Morningtonchannel to the back stop front at the toe of the The previous investigations of the Chile triple junction Upper Slope ridges.In map (Figure 26), this reconstruction area indicated that the landward trench slope is severely showsthat deep-sea fan IV connectedupslope directly to the eroded by the approachof the Chile ridge crest and that the outlets of the Golfo de Penasand Cabo Bynoe canyons. continental margin is rebuilt following ridge subduction Moreover, since the convergencebetween the Antarctic and [Cande and Leslie, 1986; Cande et al., 1987; Behrmannet al., SouthAmerica plates left no significanttectonic recordin the 1992a; Behrmann et al., 1994]. This is apparent from (1) the trench sectionbeneath unit T, we infer a very rapid trench southward narrowing and steepening of the continental sedimentaccumulation. We suggest(see section 4) that deep- margin as the triple junction is approachedfrom the north and seafans (i.e., at leastdeep-sea fans II, III, andIV) formedduring the broadening of the margin along the Golfo de Penas the interglaciationisotopic substage 5e, circa 130 to 117 ka, accretionaryprism south of the triple junction, and (2) the when the Rio Bakerriver systemwas free of ice allowing reconstructionof an accretionaryprism along the North Taitao sedimentsupply to the trench.The continentalmargin at 130- canyonsubsegment immediately south of the triple junction. 117 ka (Figures 25 and 26) which extendedseaward to the The data collected during the CTJ cruise documenta much backstopfront, i.e., thetoe of theUpper Slope ridges, was only morecomplex evolution. During the past 1 Myr, the tectonic 30 to 35 km in width, about half the presentwidth. Before regime along the Andeanconvergent margin in the areaof the deep-seafans II-IV accumulated(Figure 25c), during the Chile triple junction experiencedmajor instabilities both glacial8180 isotope stage 6 (circa150 to 200ka) when the throughspace and time. Two main non-steadystate processes Patagonianice sheet prevented massivetrench fill, the trench 8380 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

-20 -10 0 (Km) 10 20 24 27 30 40 I • I i I ' ' I ' LINE 19 _

_ >• MC ',

_ CT

_ ', backstop Oceaniccrust • ', A

MC

5

6

Figure 25. Tectonicand detrital input history of the Chile trench(CT) along the postsubductionsegment (reconstructionalong line 19, see Figure 22 for location).Back stop front (BSF); Subductionfront (SF); Morningtonchannel (MC). Note that(1) the subductionfront at 150 ka (Figure25c) and 130-117 ka (Figure 25b) coincideswith the backstop front today(Figure 25a) and (2) the subductionfront migrated24 kanto the west during the past 130-117 kyr. See text for more detail.

floor was deeper than today, maybe as much as 1-1.5 km. nonaccretionary or erosional mode characteristics. The During/5180 isotope stage 6 theSouth America convergent massive input of detrital sedimentto the trench axis during margin south of the Chile triple junction was narrower and interglaciation isotopic substage5e appearsto be the major much deeper at its seaward edge than today. It exhibited factor determining the tectonic regime change from subduction-erosion to subduction-accretion. Along this specificsection of the Chile trenchthe tectonic regimeof the Andean continentalmargin is stronglydetermined by climatic factors.During the past 150 kyr, the stress regime change along the postsubductionsegment was coeval with a major uplift of the Upper Slope ridges domain with a rate of about 1 cm/yr, requiredto sustainthe 6ø value for the meanslope angle of this domain today. Along the presubductionsegment, there is a conspicuous discrepancy between the compressiveChile trench and the adjacentextensional continental margin tectonic regimes.One of the critical factorsthat could help to explain the apparent paradox is a major increase in trench sedimentation.Recent and current accretion in the trench would reflect a recent and rapid influx of trench sediment,while the extensional tectonic regimeof the continentalmargin which is generally supposed to be associatedwith subduction-erosionat depth would be a Figure 26. Reconstructionof the Chile continentalmargin and remnant of the previous situation with a low supply of trench along the postsubductionsegment at 130-117 ka. At sedimentto the trench axis. Today the tectonic regime of the that time, the decollement was located along the toe of the Chile continentalmargin north of the triple junction is at the Upper Slope ridges. The deep-seafans (DSF) III and IV connected with the outflow of the Golfo de Penas and Cabo turning point from subduction-erosion to subduction- Bynoe canyons, respectively.The dashed lines show the accretion with both regimesactive at the sametime. Massive presentlocation of the subductionfront (SF) and the back stop accretion of trench sediment will occur in the near future, front (BSF) resulting from 3 km of convergencebetween the reproducingthe situation describedto the north [Bangs and Antarctic and South America plate during the past 130-1 17 Cande, 1997] at 38øS. In this area located 800 km north of the kyr. Cabo Bynoe canyon (CBC); Golfo de Penas canyon Chile triple junction we assumethat the influx of sediment (GPC). Seetext for more detail. beganearlier in relationwith an older retreatof the ice capthat BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8381 resulted in a 2-3 times thicker trench sediment accumulation. As proposedby Bangs and Cande [1997], we suspectthat W bulge-relatednormal fault E accretion occurringtoday along the presubductionsegment is linked to a dramaticincrease in trench sedimentsupply. We suggest that the change in tectonic regime of the continental margin along the presubductionsegment is climate-inducedas it is along the postsubductionsegment. ",-' 0 ß 5 Ma

8.2. Tectonic Reorganization at the Nazca- rift-relatednormal fault ridgejump Antattica Boundary

During the CTJ cruise, no additional complexity was found in the magnetic lines collected along the CRS I transect, as compared to magnetic data reported in previous works [Herton et al., 1981; Tebbenset al., 1997]. The western flank of the CRS 1 recorded an apparently continuous anomaly sequencefrom the centralanomaly to anomaly2A. There is no Figure 27. (a) Chile ridge subduction occurredsometime after evidence in the magnetic data to question the subduction 780 ka. As the Antarctica plate enteredthe subduction,bulge- history of the CRS I as currently accepted:the Chile triple related normal faults developed. (b) Ridge jump to the west junction defined by the CRS 1 and the trench migrated occurred subsequently, resulting in the developmentof the northward from 46ø25'-46ø30'S to its present location at CRS I in an area of the Antarcticplate previouslydeformed by 46ø09'S during the past 200-300 kyr. However, this simple bulge-related normal faults. Then tectonic deformation reconstruction leaves three major points at issue: (1) Along associatedwith spreading activity along the CRS I occurred. A rift-related normal fault network developed resulting in the the Chile ridge subsegment,the lack of sedimentaccumulation reactivation of the older bulge-relatedfault network. Diagonal strikingly characterizesthe CRS 1 axial valley in contrastto pattern indicates Andean continental margin; arrowhead the Chile trench in its southward prolongation and the pattern indicates oceaniccrust; spreadingcenter (SC); not to adjacentoceanic crust west of them. The reduced supply of scale. See text for more detail. detrital material that transits today through the North Taitao canyondoes not properlyexplain this situation. Sincethe rift fault system of the CRS I sharply cuts off the thick accumulation of trench sediment south of the Chile triple accretionarywedge. Since no signature in the magnetic junction and the thick accumulationof sedimentoverlying the recordsexists, the ridge jump had to occur during the last oceaniccrust that lies directly west of the rift valley wall, we Brunhesnormal polarity event, i.e., ChronCln, that is to say suspectthat an age gap existsbetween the axial valley floor after780 ka [Cande and Kent, 1992]. Becausethe ridge- and the adjacentoceanic crust. (2) The Antarctica plate area parallel tectonic fabricof the Antarctic plate was reactivated located both to the west and to the south-west of the CRS 1 not only west of the CRS I but also west of the trench axis in axial valley exhibits a fault network which evolved through its southwardprolongation, we assumethat ridgejump took two extensionaltectonic pulses, not documentedelsewhere in place between the Darwin fracturezone and the North Taitao the surveyed area. Moreover, in the vicinity of the triple Ridge fault. junction (Figure 8), the rift fault systemof the CRS I trending The North Taitao Ridge fault which offsetsthe subduction N I0øW sharplycuts off the nearly north-southtrending fault front right laterally by-25 km shows no clear evidence of network which characterizesthe Chile trench in this area. (3) such a large strike-slip movementalong its seaward and The complexstructure of an accretionarywedge documented landward prolongations and has the characteristics of a along the Taitao ridge transectshows no extensionto the transformfault. Moreover,it also exhibits a significantnormal north. The tectonic history of the Chile continental margin is slip componentwith a downwardoffset of the Taitao Ridge. different on either side of the North Taitao Ridge fault This is in goodagreement with the Antarcticplate subsidence boundingthe Taitao Ridge to the north. south of the flexurealong the seawardprolongation of the A better understanding of these three major points can be North Taitao Ridge fault. Since no major contractionalfeature achieved by considering the CRS I as evolving through a exists along the North Taitao Ridge fault, we assumethat recent westwardjump to its Presentlocation (Figure 27). This convergencealong the frontaldecollement trends parallel to ridge jump accounts for the age gap which is suspected the fault on eitherside of it. Therefore,there is a discrepancy between the CRS 1 axial valley and the Antarctica adjacent betweenthe regionaland the local convergencesbetween the seafloor and the two extensionaltectonic pulses documented Antarcticand SouthAmerica plates that trendN90øE [DeMets in this specific area. We assumethat the older extensional et al., 1990] and N50øE, respectively.The E-W trending faults developedduring a first phase of ridge subduction in thrust fault systemwhich offsetsthe subduction front in an relation to the developmentof the associatedslab bulge. Then areanorth of point X (Figure 13b) is believedto compensate the westward ridge jump occurred,resulting in the CRS 1 for the northwardcomponent associated with the local N50øE developmentin its present location. Subsequently,the rift- trendingconvergence. North of the North TaitaoRidge fault, relatedtectonic activity of the CRS I inducedthe development in the Taitaocanyon area, the width of the accretionaryprism of a new fault network and the bulge-related older fault decreasesabruptly to only 5-6 km, and the apex of the back reactivation. Moreover, this model allow us to propose a stopis 1000 m deeperas comparedto the TaitaoRidge transect coherentreconstruction of the margin area located between the (AB, Figure 16). We inferthat the tectonichistory of the Chile Darwin and Taitao fracturezones, including the Taitao Ridge continental margin is different on either side of the North 8382 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

,-,-,650 ka A -,,-,600ka B evolution as shown on Figure 28. Sometime after 780 ka (Figure 28a), a first subduction of the Chile ridge segment •AP occurred between the Taitao and the Darwin fracture zones. As the Chile triple junction migratedto the north, an accretionary prism (Figure 28b) was rebuilt southof it. We assumethat the thrust sheetsaccumulated along the Taitao Ridge transectare a remain of this old accretionary prism. When in the ridge-

R trench-trenchconfiguration, the triple junction migrates at a rate of 160 cm/yr. Therefore it took <50 kyr for the ridge segmentto subducttotally. At about 300-250 ka, a section of the Chile ridge located between the North Taitao Ridge fault -,-,300-250 ka C ..,140 ka D and the Darwin fracturezone jumped west of the subduction front (Figure 28c) along the Present CRS 1, in an area of the Antarctic plate younger than 780 ka. This ridge jump created an ephemeralmicroplate (the Chonosmicroplate), as proposed elsewhereby DeLong and Fox [1977]. The Chonosmicroplate boundaries are the CRS 1, the Darwin fracture zone, the Chile trench, and the North Taitao Ridge fault. The subductionof the Chonos microplate was associated with frontal subduction- _NTR• .,.:..•'_•..';,•R erosion that removed most of the previously accumulated accretionaryprism. South of the North Taitao Ridge fault, no ridge jump occurred (Figure 28d) preventing subduction erosion along the Taitao ridge transect (Figure 28e). As the PRESENT E Chile ridge (i.e., the CRS I which is the western boundary of the Chonos microplate)reached the trench again, the triple junction changed from a transform-trench-trenchto a ridge- trench-trenchconfiguration. The sectionof the margin south of the triple junction began to develop an accretionaryprism; meanwhile, the northern section stayed in a subduction- erosionmode, as it is today.

8.3. Topography of the Ocean Floor Sincethe trenchaxis is devoidof sedimentsalong the Chile Figure 28. Reconstructionof ridge subduction history along ridge subsegment,we infer not only that the sourceof detrital the transect located between the Taitao and Darwin fracture material accumulatedalong the presubductionsegment was zones and tectonic model of birth and death of the ephemeral located to the north but also that the slab segmentation Chonos microplate. During the past 780 kyr, two ridge- determined the locus of trench sediment accumulation. The subduction phases separated by a westward ridge-jump occurred. This evolution controlled the tectonic regime of the Chile ridge relief entering the subduction appearsto be adjacent continental margin with two steps of subduction- dammingthe sedimentoriginating from the north. South of the accretionseparated by subduction-erosion.South of the North Darwin fracture zone, no subduction-accretionoccurs because Taitao Ridge fault, no ridge jump occurred, allowing the no sedimentis enteringthe subductiontoday. Conversely, previously constructed accretionaryprism to be preserved northof the Darwin fracturezone, the along-strikedeepening alongthe Taitao Ridge. See text for more detail.Antarctic plate of the trench axis basement allowed the detrital sediment (AP); Chonosmicroplate (CMP); Darwin fracture zone (DFZ); originating from the north to accumulate and therefore Nazca plate (NP); North Taitao Ridge fault (NTRF); South subduction-accretionto be active.The Chile ridge considered America plate (SAP); Taitao fracture zone (TFZ); Taitao as a relief divertingthe sedimentsupply to the trenchplays a peninsula(TP); Taitao ridge (TR); triangles, subductionfront (SF) and back stopfront (BSF); double lines, active spreading major role in the tectonic developmentof the continental ridges; solid arrows, westwardridge jump; dashedpattern and margin.From ODP Leg 141 data,Behrmann et al. [1994] have solid dots, first and second accretionary prism, respectively; shownthat no frontalaccretion of sedimenthas occurred along light shaded areas, oceanic crust accretedbetween 780 and the Chile ridge subsegmentsince the late Pliocene. At O DP -300 ka; dark shaded areas,accreted oceanic crust younger Site 863 (Figure8b), mostof the materialaccreted during the than---300 ka. Pliocene was subsequentlyremoved by subduction-erosion. As along the northern Taitao canyon, the continental basement extends to within few kilometers landward from the trench axis, indicating that considerable volume of forearc TaitaoRidge fault with a well-developedaccretionary prism basementhas been tectonically eroded in an area where the alongthe TaitaoRidge transect but showingno prolongation Chile ridge is supposedto be at depth. This factoradds the north of it. The along-strikedownward shift of the decollement role of the Chile ridge as a thermal-inducedrelief, which of 1000m southof the North Taitao Ridge fault suggeststhat diverts sediment.Subsidence of the margin should also be the Taitao Ridge accretionaryprism extendednorthward and expectedfrom subduction-erosionrelated to, ridge subduction was subsequentlyremoved north of it. We suggesta tectonic at depth. BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8383

At the toe of the Taitao Ridge accretionary prism, active seawardedge of the ridge and a thick midslope thrust sheet subduction of trench sediment is occurring in relation to landward. This accretionary wedge stacked beneath the Chile trench infill compressional deformation. Today, the thick continentalmargin back stop is 25-28 km in width. Along the trench infill is actively underthrust landward below a Taitao Ridge, the back stop apex (BSA, Figure 13a) is located subhorizontal frontal decollement. The frontal decollement 16-17 km seaward from the Taitao peninsula shoreline, to slicesthe three thrust sheets of the Taitao Ridge accretionary 75ø54'W in water depth as shallow as 1500 at As the prism prism horizontally at their base indicating that (1) frontal grew, the wedge beneath the back stop was underplated subduction-erosionalso occurs today and (2) the Taitao ridge causinguplift of the apex. The midslopethrust sheet exhibits a accretionary prism grew during an older phase during a volcanic basement with petrological, geochemical, and different tectonic regime at the subduction front. Since the mineralogical signaturesconsistent with an oceanic origin frontal decollement follows the same 3250 m water depth [Guivel et al., 1999] and a well-stratified sedimentarycover isobath on both side of the South Taitao canyon, we infer that with a seismic signature similar to that of the adjacent active subduction occurs farther to the south beneath a flat Antarctic plate. We propose that the tectonic accretion of a lying decollement, along the Taitao and Tres Montes slice of ocean floor occurredalong the Taitao Ridge transect peninsulatransect. Along the reentrantat 46ø40'S, the trench during the past 780 kyr. However, the midslopethrust sheetis fill shows little tectonic deformation; meanwhile, thrust stacked up beneath a back stop consisting of volcanic rocks faulting of the trench fill existsto the north and to the south.A and sediments[Behrmann et al., 1994; Forsythe et al., 1995] slight change in the dip angle of the oceanic basementfrom similar both in composition and age to those cropping out subhorizontal to a clear landward dipping may explain the onshorein the Taitao peninsula.Therefore a major thrust fault weaknessof active tectonic deformationalong the reentrant off must separate the midslope thrust sheet from the Taitao the Taitao and Tres Montes peninsulas. The scar associated peninsula rock units including the Bahia Barrientos with this major reentrant is in the landward prolongation of ophiolite. Indeed, a major age gap exists between the two the Taitao Fracture Zone ridge (Plate 1). Thus it exhibits the pieces ofophiolite. The Bahia Barrientos ophiolite is older characteristics of an impacting seamount that accelerates than 6-4 Ma (i.e., olderthan the 6-4 Ma intruding Cabo Raper removalof material fromthe front of the margin.We speculate pluton [Mpodozis et al., 1985; Guivel et al., 1999]); that a ridge crest located between the Taitao and Tres Montes meanwhile, the midslope thrust sheet ophiolite along the fracture zones in the landward prolongation of the Taitao Taitao ridge must be younger than 780 ka. We assumethat Fracture Zone ridge underthrustbeneath the Chile continental these two pieces of ophiolite have no connection in both margin. The underthrusting action of such a large edifice not space and time. While the process of emplacementof the only wedges up, deforms,and fracturesthe base of the trench midslope thrust sheet ophiolite along the Taitao ridge is slope [Lallemand and Le Pichon, 1987] but also induces understood, that of the Bahia Barrientos remains unsolved. uplift of the overlying continentalwedge. We suggestthat the However, it is probablyreasonable to think that processesat major uplift of the Taitao and Tres Montes peninsulas the origin of the emplacementof bothpieces are similar. originates from a subducting seamountrather than being thermally induced by the presenceof a Chile ridge segment beneath them or induced by the reconstruction of an 9. Conclusion accretionaryprism [Bourgois et al., 1996]. This is in good Along the specific section of the Andean continental agreementwith the absence of significant frontal accretion margin surveyed during the CTJ cruise, two main factors acting at the toe of this subsegmentof the margintoday. This controlling the tectonic regime have been identified. The hypothesisfits also with the uplift and subsidencehistory as trench accumulationis strongly controlled by climate recorded by the Pliocene and Pleistocene sediments of the variation and the tectonic reorganization at the Nazca- Taitao and Tres Montes peninsulas[Bourgois et al., 1996]. Antarcticaplate boundarywhich involved postsubduction ridgejump associatedwith the evolutionof the Chile triple 8.4. Ophiolite Emplacement junction. Underthrustingof slab positivetopography also A major thematic question of the Chile margin triple locally influenced the tectonic evolution of the continental junction area is that arising fromthe Taitao Ridge (Figure 1) margin. The combination of these three unconnected factors which is supposedto be the seawardprolongation [Bangs et and their fluctuation through time determinethe dominant al., 1992] of the onland Bahia Barrientos ophiolite (Figure 3) tectonicregime of a peculiarsegment of the continentalmargin cropping out in the Taitao peninsula. The previous results at any particular time. published by Kaeding et al. [1990] and Le Moigne et al. Rapidincrease in trenchdeposition caused the marginto [1996] that document an oceanic origin for peridotites and switchfrom subduction-erosion to subduction-accretion(1) gabbros of the Bahia Barrientos ophiolite substantiate this afterthe glacial-interglacialepisode at 130-117 ka alongthe assumption.From drilling material recoveredduring Leg 141 postsubductionsegment and (2) after the last deglaciation at Site 862, Behrmann et al. [1994] subsequentlysuggested alongthe presubductionsegment. Both situationsare related that the Taitao Ridge is a nascent forearc ophiolite to majorretreat of the Andean ice cap, allowing continental emplacementof which is connected with the Taitao fracture river drainage to feed the trench axis with sediment. zone. It may constitute an ophiolite terranein the processof Conversely, a nonaccretion or a subduction-erosion mode emplacement. characterizedthe presubductionand postsubductionsegments The EM12 data and seismiclines collectedalong the Taitao during glacial maximums.The tectonicregime of the Andean Ridge during the CTJ campaignhave revealed the complex continental margin is climate-dependent.In the area of the three-dimensionalstructure of an accretionarywedge (Figure Chile triple junction the durationof climaticcycles and their 13b) including at least three sheets:two thrust sheetsat the magnitudeand chronology greatly determine the episodesand 8384 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

erosion to subduction-accretion in association with the Chile A triple junction migration. In the presubduction situation (Figure29a), the thin, buoyantlithosphere of the leadingedge of the Chonos microplate is west of the subduction front. Thereforethe subduction channel [Cloos and Shreve, 1996; Charlton, 1988] alongthe decollementis thickening arcward allowing the subducted material to be easily removed landward. As ridge subduction occurs, the buoyant lithosphereof the spreadingcenter moves arcward beneath the overridingblock. The descendingplate actsto moveupward andthe shearzone to thin landward.Conversely, as regarding the presubduction situation, subducted sediment becomes accreted(Figure 29b) becauseof an increasein the pressure gradient acting to resist sedimentsubduction. We think that the subduction channel conceptdeveloped by Shreve and Cloos [1986] and Cloos [1992] for subduction of sediment and seamountasperities closely matchesthe evolution' of the .B tectonicregime at the Chile triple junction area. The westwardridge jump which occurredalong the CRS 1 producedthe ephemeralChonos microplate north of the North Taitao Ridge fault. Slab weaknessand thinning may have been introducedby this processof plate fragmentation.Breakup of ß SCh the young,buoyant slab in the nearbysubduction zone may have two potential consequencesregarding (1) the emplacementof ophiolite slices into the accretionaryprism, i.e., the continental margin, and (2) the contaminationof basalts sampledalong the Chile ridge [Klein and Karsten, 1995]. From anomaloustrace element ratios they proposedthat the back arc affinitiesof Chile ridge lavas eruptedalong the CRS 1 may havebeen produced by mixingof MORB melt with Figure 29. (a) Before ridge subductionthe subductionchannel melts of mantle containing 0.35% sediment and -3% altered (Sch, dotted pattern) is thickening arcward; no accretion of oceanic crust contaminants. We suggest that slab sediment is allowed. (b) After ridge subduction the fragmentationinduced by ridge subductionincreases oceanic subduction channel (Sch, dotted pattern) thins arcward; crust alteration through multiple episodes of faulting and subduction-accretionworks. See text for more detail. Diagonal allows the suboceanic mantle to become contaminated with pattern, Andean continental margin; arrowhead pattern, oceaniccrust; spreading center (SC); thick solid arrow, active material derivedfrom adjacentsubduction. uplift. Acknowledgments.We gratefullyacknowledge the thoughtfuland thoroughreviews of two anonymousreviewers. The CTJ cruise was supportedby IFREMER (France).We thankthe captainand the crew of timingof the Andean convergentmargin development during the R/V L',4talante for their efficient work and the Servicio the past I Myr. Hidrograficoy Oceanografico(SHOA) de la Armada de Chile. This The subductionhistory of the Chile ridge and transformsis work was fundedunder grant 97N51/0353by the InstitutNational des not as simpleas previouslyproposed [Mpodozis et al., 1985; Sciencesde I'Univers (INSU) and additionalsupport from the Centre Leslie, 1986] for the past I Myr. North of the Taitao fracture National de !a RechercheScientifique (CNRS). This work was also zone, the current subductionof the CRS 1 follows a first phase supportedby the ECOS-CONICYT-ANUIESprograms through the of ridge subduction which occurredafter 780 ka. The past 1 projectsC96U01 and M96U01. Myr tectonicevolution of the Chile triple ,junctionreveals the sametypical short-terminstabilities and complexitieswhich References have been identified at other triple junctions such as Mexico [Luhr et al., 1985; Bourgoisand Michaud, 1991; Hey, 1998] Atwater, T., Implicationsof plate tectonicsfor the Cenozoictectonic evolutionof westernNorth America, Geol. Soc. Am. Bull.,81, 3513- or Queen Charlotte [Rohr and Furlong, 1995]. North of the 3536, 1970. Taitao fracturezone, the ridge subduction history has deeply Aubouin,J., J. Bourgois,and J. Az•ma,A newtype of activemargin: the influenced the tectonic evolution of the continental margin. convergentextensional margin as exemplified by the Middle Along the North Taitao canyonsubsegment, two episodesof AmericaTrench off Guatemala,Earth Planet. Sci. Lett., 67, 211-218, 1984. accretion separated by an episode of subduction-erosion Bangs,N.L., and C. Cande,Episodic development of a convergent occurredin <780 kyr. Moreover,the northward migration of margininferred from structuresand processesalong the southern the triple junction which occurred twice along this Chile margin,Tectonics, 16, 489-503, 1997. subsegmentintroduced an along-strike variation in the Bangs, N.L., S.C. Cande, S.D. Lewis, and J.J. Miller, Structural magnitudeof accretedmaterial and the duration of each frameworkof the Chilemargin at the Chile Ridgecollision zone, tectonicepisode along a specifictransect. Proc. OceanDrill. Program,Initial Rep.,141, 11-21, 1992. Barazangi,M., andI3. L. Isacks,Spatial distribution of earthquakesand A question to be addressedis the origin of the tectonic subductionof the Nazca platebeneath South America, Geology, 4, regime change of the continental margin from subduction- 686-692, 1976. BOURGOIS ET AL.: CHILE TRIPLE JUNCTION 8385

Behrmann,J. H., et al., Proceedingsof the Ocean Drilling Program, Herve, M., J. Fuenzalida, E. Araya, and A. Solano, Edades Initial Reports.,vol. 141, 708 pp., Ocean Drill. Program,College radiometricasy tectonicaNeogena en el sector costerode Chiloe Station, Tex., 1992a. continental,Xa region, paper presentedat II CongresoGeologico Behrmann, J. H., S.D. Lewis, and ODP Leg 141 Scientific Party, Chileno,Sociedad Geologica de Chile, Arica, 1979. Geologyand tectonics of the Chiletriple junction, Eos Trans.AGU, Hey, R.N., Speculativepropagating rift-subduction zone interactions 73, 404-405, 410, 1992b. with possibleconsequences for continental margin evolution, Behrmann,J. H., S.D. Lewis, S.C. Cande, and ODP Leg 141 Scientific Geology,26, 247-250, 1998. Party,Tectonic and geologyof spreadingridge subductionat the Hollin, J.T., and D.H. Schilling,Late Wisconsin-Weichselianmountain Chile triple junction: A synthesisof resultsfrom Leg 141 of the glaciersand small ice cap. In The Last Great Ice Sheets,edited by OceanDrilling Program,Geol. Rundsch.,83, 832-852, 1994. G.H. Dentonand T.J. Hughes,pp.179-220, John Wiley, New York, Bourgois, J., and F. Michaud, Active fragmentation of the North 1981. AmericaPlate at the triple junctionarea off Manzanilio,Geo-Mar. Hulton,N., D. Sugden,A. Payne,and C. Clapperton,Glacier modeling Lett., 11, 59-65, 1991. and climateof Patagoniaduring the lastglacial maximum, Quat. Res., Bourgois,J., Y. Lagabrielle , R. Maury, J. Le Moigne, P. Vidal, J.M. 42, 1-19, 1994. Cantagrel,and O. Urbina, Geologyof the Taitao peninsula(Chile Hussong,D. M., andL. K. Wipperman,Vertical movement and tectonic margin triple junction area, 46ø-47øS): Miocene to Pleistocene erosion of the continental wall of the Peru-Chile Trench near 11ø30'S obductionof the BahiaBarrientos ophiolite, Eos Trans.AGU, 73(43), latitude, Mem. Geol. Soc. Am., 154, 509-524, 1981. Fall Meet. Suppl.,592, 1992. Kaeding, M., R.D. Forsythe,and E.P. Nelson, Geochemistryof the Bourgois,J., Y. Lagabrielle,J. Le Moigne,O. Urbina,M.C. Janin,and P. Taitao ophioliteand near trench intrusionsfrom the Chile margin Beuzart, Preliminaryresults of a field studyof the Taitao ophiolite triplejunction, J. S. Am. Earth Sci., 3, 161-177, 1990. (southernChile): Implicationsfor the evolutionof the Chile triple Kay, S.M., V.A. Ramos,and M. Marquez, Evidence in Cerro Pampa junction,Ofioliti, 18, 113-129, 1993. volcanic rocks for slab-meltingprior to ridge-trench collision in Bourgois,J., H. Martin, Y. Lagabrielle,J. Le Moigne, andJ. FrutosJara, southernSouth America, d. Geol., 101,703-714, 1993. Subductionerosion related to spreading-ridgesubduction: Taitao Klein, E.M., and J.L. Karsten, Ocean ridge basaltswith convergent peninsula(Chile margintriple junctionarea), Geology,24, 723-726, margingeochemical affinities from the southernChile ridge, Nature, 1996. 374, 52-57, 1995. Bourgois, J., et al., Bathymetry and seismic reflection imaging of Kulm L.D., et al., Initial Reportsof the Deep Sea Drilling Project, vol. tectonicresponses to ridge-transformsubduction at the Chile triple 18, 1077 pp., U.S. Gov. Print.Off. Washington,D.C., 1973. junctionarea (45-48øS), CTJ cruiseof the R/V L'Atalante,Eos Trans. Lagabrielle, Y., J. Le Moigne, R. Maury, J. Cotten, and J. Bourgois, AGU, 78(46), Fall Meet. suppl.,F- 636, 1997. Volcanic record of the subductionof an active spreadingridge, Cande,S.C., and D.V. Kent, A new geomagneticpolarity time scale for Taitao peninsula(southern Chile), Geology,22, 515-518, 1994. the Late Cretaceousand Cenozoic, d. Geophys.Res., 97, 13,917- Lallemand, S., and X. Le Pichon, Coulombwedge model applied to 13,951, 1992. subductionof seamountsin the Japan trench, Geology, 15, 1065- Cande, S.C., and R.B. Leslie, Late Cenozoic tectonics of the southern 1069, 1987. Chile trench,d. Geophys.Res., 91,471-496, 1986. Le Moigne, J., Y. Lagabrielle, H. Whitechurch, J. Girardeau, J. Cande, S.C., R.B. Leslie, J.C. Parra, and M. Hodbart, Interaction Bourgois,and R. Maury, Petrologyand geochemistryof the ophiolitic between the Chile ridge and Chile trench: Geophysical and and volcanicsuites of the Taitao peninsula,Chile triplejunction area, geothermalevidences, d. Geophys.Res., 92, 495-520, 1987. d. S. Am. Earth Sci., 9, 43-58, 1996. CharItoh,T. R., Tectonicerosion and accretionin steady-statetrenches, Leslie, R. B., Cenozoic tectonicsof southernChile: triple junction Tectonophysics,149, 233-243, 1988. migration,ridge subduction,and forearc evolution,Ph.D. thesis,276 Cloos, M., Thrust-type subduction-zoneearthquakes and seamount pp.,ColumbiaUniv., Ne•v York, 1986. asperities:Physical model for seismicrupture, Geology, 20, 601-604, Lewis, S.D., J.H. Behrmann,R.J. Musgrave,and S.C. Cande(Eds.), 1992. Proceedingsof the Ocean Drilling Program,Scientific Results, vol. Cloos, M., and R. L. Shreve,Shear-zone thickness and the seismicityof 141,498 pp., OceanDrill. Program,College Station, Tex., 1995. Chilean- and Marianas-typesubduction zones, Geology, 24, 107-110, Lowell, T.V., C.J. Heusser, B.G. Andersen, P.I. Moreno, A. Hauser, 1996. L.E. Heusser, C. Schltichter, D.R. Marchant, and G.H. Denton, DeLong, S.E., and P.J. Fox, Geological consequencesof ridge Interhemisphericcorrelation of Late Pleistoceneglacial events, subduction,in IslandArc, DeepSea Trenches,and Back-Arc Basins, Science, 269,1541-1549, 1995. Maurice Ewing Ser., vol. 1, edited by M. Talwani and W.C. Pitman Luhr,J.F., S. Nelson,J.F. Allan, and I.S.E. Carmichael,Active rifting in III, pp. 221-228 AGU, Washington,D.C., 221-228, 1977. southwesternMexico: Manifestationsof an incipient eastward DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein, Current plate spreading-ridgejump, Geology,13, 54-57, 1985. motions,Geophys. d. Int., 101,425-478, 1990. McCarthy, J., and D.W. Scholl, Mechanismsof subductionaccretion Dickinson,W. R., and W.S. Snyder,Geometry of triple junctionsrelated alongthe centralAleutian Trench,Geol. Soc.Am. Bull., 96, 691-701, to San Andreastransform, d. Geophys.Res., 84 , 561-572, 1979. 1985. Forsythe,R.D., and E. Nelson, Geological manifestationsof ridge Miller, A., The climateof Chile,in WorldSurvey of Climatology,vol. 12, collision: Evidence from the Golfo de Penas-Taitao basin, southern Climates of Central Chile and South America, edited by W. Chile, Tectonics,4, 477-495, 1985. Schwerdtferger, pp.113-145, Elsevier, Amsterdam-Oxford-New Forsythe,R. D., E.P. Nelson,M.J. Carr, M.E. Kaeding,M. Herv•, C.M. York, 1976. Mpodozis, M.J. Soffia, and S. Harambour, Pliocene near trench Mpodozis,C. M., M. Herv•, C. Nasi,M.J. Soffia,R.D. Forsythe,and E.P. magmatismin southernChile: A possiblemanifestation of ridge Nelson, El magmatismoPlioceno de peninsulaTres Montesy su collision,Geology, 14, 23-27, 1986. relacioncon la evoluciondel puntotriple de Chile austral,Rev. Geol. Forsythe,R. D., J. K. Meen, J. F. Bender, and D. Elthon,Geochemical Chile, 25-26, 13-28, 1985. data on volcanic rocks and glassesrecovered from Site 862: Murdie,R., D. Prior,P. Styles,S. Flint,R. Pearce, and • &gar, Seismic Implicationsfor the origin of the Taitao Ridge, Chile triple Junction responsesto ridge-transformsubduction: Chile , iple junction, region,Proc. OceanDrill. Program,Sci. Results,141, 331-348, 1995. Geology,21, 1095-1098, 1993. Guivel, C., Y. Lagabrielle,J. Bourgois,R. Maury, H. Martin, and S. Nelson,E., R. Forsythe,and I. Arit, Ridge collisiontectonics in terrane Fourcade,New geochemicalconstraints for the origin of ridge- development,d. S. Am. Earth Sci., 7, 271-278, 1994. subduction-relatedplutonic and volcanic suitesat the Chile triple Porter, S.C., C.M. Clapperton,and D.E. Sugden, Chronologyand junction (Taitao peninsula and Site 862, LEG ODP 141), dynamicsof deglaciationalong and near the straitof Magellan,South Tectonophysics,311, 83-111, 1999. America, Sver. Geol. Unders.,81,233-239, 1992. Haeussler,P. J., D. Bradley, R. Goldfarb, L. Snee, and C. Taylor, Link Ramos,V.A., and S.M. Kay, SouthernPatagonian plateau basaltsand between ridge subductionand gold mineralization in southern deformation:backarc testimony of ridge collision,Tectonophysics, Alaska, Geology,23, 995-998, 1995. 205, 261-282, 1992. Herron, E. M., S.C. Cande, and B. Hall, An active spreadingcenter Rohr,K.M.M., andK.P. Furlong,Ephemeral plate tectonics at the Queen collideswith a subductionzone: A geophysicalsurvey of the Chile Charlottetriple junction, Geology, 23, 1035-1038,1995. margintriple junction, Mem. Geol. Soc.Am., 154, 683-701, 1981. Scholl,D.W., Sedimentarysequences in the North Pacific trenches,in 8386 BOURGOIS ET AL.: CHILE TRIPLE JUNCTION

TheGeology of ContinentalMargins, edited by C.A. Burkand C.L. J. Boul•gue,Laboratoire de G6ochimieet M6tallog6nie,Universit6 Drake,pp. 493-504, Springer-Verlag, New York,1974. Pierre et Marie Curie, Tour 26-16, 5•me 6tage, Boite 124, 4 place Scholl,D.W., R. E. yon Huene, T. L. Vallier and D. G. Howell, Jussieu,F-75252 Paris Cedex05, France.(boulegue•cicrp.jussieu.fr) Sedimentarymasses and conceptsabout tectonicprocesses at J. Bourgois,Centre National de la RechercheScientifique, Universit6 underthrustocean margins, Geology, 8, 564-568,1980. Pierre et Marie Curie, Laboratoire de G6odynamique,Tectonique et Shreve,R.L., andM. Cloos,Dynamics of sedimentsubduction, melange Environnement,Tour 16-26, 4•me 6tage, Boite 119, 4 place Jussieu,F- formation,and prism accretion, d. Geophys.Res., 91, 10,229-10,245, 75252 ParisCedex 05, France.(bourgois•ccr.jussieu.fr) 1986. T. Calmus, lnstitutoGeologico, Universidad Autonoma de Mexico, Sisson,V. B., and T.L. Pavlis, Geologicconsequences of plate EstacionRegional del Noroeste,Hermosillo, Sonora, Mexico. reorganization:An examplefrom the Eocenesouthern Alaska (tcalm us•servi dor.unam .mx) forearc,Geology, 21, 913-916, 1993. V. Daux, Laboratoirede G6ologieS6dimentaire, Universit6 Pierre et Tebbens,S.F., S.C. Cande, L. Kovacs,J.C. Parra, J.L. LaBrecque,and Marie Curie, Tour 14-15, 4•me 6tage, Boite 117, 4 place Jussieu,F- H. Vergara,The Chileridge: A tectonicframework, d. Geophys. 75252 ParisCedex 05, France.(daux•ccr.jussieu.fr) Res., 102, 12,035-12059, 1997. C. Guivel, Laboratoirede Plan6tologieet G6odynamique,Universit6 yon Huene,R. E., To accreteor not accrete,that is the question,Geol. de Nantes, 2 rue de la Houssini•re, B.P. 92208, F-44322 Nantes Cedex Rundch., 75, 1- 15, 1986. 3, France.(guivel•chimie.univ-nantes.fr) yon Huene,R. E., and S. Lallemand,Tectonic erosion along the Japan Y. Lagabrielle, Institut de Recherche pour le D6veloppement,B.P. and Peruconvergent margins, Geol. Soc. Am. Bull., 102, 704-720, A5, NoumeaCcdex, Nueva Caledonia.(!agabrielle•noumea. ird.nc) 1990. yonHuene, R. E., andD. W. Scholl,Observations at convergentmargins concerningsediment subduction, subduction erosion, and the growth of continentalcrust, Rev. Geophys., 29, 279-316, 1991. Winograd,l.J., J.M. Landwehr,K.R. Ludwig,T.B. Tyler, and A.C. Riggs,Duration and structure of the pastfour interglaciations,Quat. (ReceivedFebruary 11, 1999;revised August 27, 1999; Res., 48, 141- 154, 1997. acceptedNovember 5, 1999.)