<<

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 93, NO. B4, PAGES 3211-3231, APRIL 10, 1988

Uplift of the CentralAndean and Bending of theBolivian

BRYAN L. ISACKS

INSTOCCornell Project, Department ofGeological Sciences, Cornell University, Ithaca, New York

Thetopography of thecentral Andes can be consideredthe primary tectonic "signal" of lateCenozoic mountainbuilding in anarid where the effects of upliftand magrnatism arelitfie obscured by denudation. Thespatial coverage ofthe topographic signal is more complete than that for sparsely sampled geological and geophysicaldata. A color-codedimage of digitizedtopography between 12øS and 37øS highlights the -Puna,oneof theworld's most remarkable , and reveals important physiographic clues about theformation ofthat major feature. The topographic data combined with information onstructure, magmatism, seismicity,and palcomagnetism support a simplekinematical model for thelate Cenozoic evolution of the centralAndes. The model does not require collisional effects or enormousvolumes of intrusiveadditions to the crustbut instead calls upon plausible amounts of crustalshortening and lithospheric thinning. The model interrelatesAndean uplift, a changinggeometry ofthe subdueted , and a changingoutline (in map view)of theleading edge of theSouth American plate. Crustal shortening has accommodated convergence betweenthe Chilean-Peruvian forearc and the South American foreland. The Altiplano-Puna plateau can be constructedby a combinationof crustalshortening and thickening and lithospheric thinning above a shallow dipping(20o-30 ø) subductedplate. The seawardlyconcave bend of westernSouth America, the "Bolivian orocline,"was enhanced but not completelyproduced by an along-strikevariation in the amountof late Cenozoicshortening. Maximum shortening in both produced the widest part of theplateau and increasedthe seaward concavity of theBolivian orocline. The along-strike variations of shorteningare hypothesizedto resultfrom correspondingalong-strike variations in ihe widthof a weakenedzone in the overridingplate. Weakeningoccurs above the wedge of asthenospherelocated between the subductedand overridingplates; hence the width of the zone of weakening depends upon the dip of the subducted plate. Two typesof shorteningare recognized:(1) a widespread,basin-and-range, Laramide-like shortening that characterizesmodem activity in the Sierras Pampeanas andlate deformation ofthe Altiplano-Puna and (2) on theeastern side of thecordilleras and plateau, an eastverging foreland fold-thrust belt in whichthe underthrustforeland compresses andthickens the ductile lower crust and produces a plateau uplift of the upper crust.The second type of shorteningcan be applied to Plio-Quatemarydeformations throughout the central Andesbut with a substantialnarrowing ofthe region of plateau uplift in and south of 28øS. A proposed monoclinalflexure of the upper crust on the western side of the plateau uplift explains the remarkably simple andregular morphology ofthe main westem slope of the central Andes. The monocline islocated above the tip of theasthenospheric wedge between the converging plates; it ispostulated to occur above the western limit of lower crustalthickening. In the regionsof horizontalsubduction the monoclinecan be associatedwith a late Mioceneasthenospheric wedge tip.

INTRODUCTION evidenceabout uplift and crustal deformation to supporta model The Alfiplano of southernPeru, Bolivia, and northernChile for the late Cenozoicevolution of the centralAndes. The model connectsalong-strike changes in the amountof shorteningin the andthe Puna of northwesternArgentina together form one of the upperplate with the shapeof the subductedNazca plate and with world'sgreat plateaus. Average elevations near 4 km occurover changesin the map view outline of the leadingedge of the an areaabout 300 km wide and2000 km long. Althoughthe overridingcontinental plate. The key ideais thatan along-strike plateauseems clearly associated with the subductionof theNazca variationin the amountof shorteningimplies the possibility of a plate beneathwestern , its origin remains changein theshape of themap view outline of thelead'rag edge of controversial.While many studieshave assumedthat modern the upperplate. If late Cenozoicshortening of theupper plate Andeantopography is dueto crustalthickening produced by wereto be a maximumin the regionof the seawardlyconcave magmaticadditions of materialfrom the mantle [e.g., Thorpe et cornernear the Peru- border (near 18øS), then the map view al., 1981], an increasingamount of evidencepoints to the curvatureof the forearcmust have increased.This providesa importanceof compressionalcrustal shortening in theformation simple mechanismfor the developmentof Carey's [1958] themodem central Andes [e.g., Dalmayrac et al., 1980;Burchfiel "BolivianOrocline," a conceptthat has recently received support et al., 1981; Jordan et al., 1983a;Chinn and Isacks, 1983; frompaleomagnetic studies [e.g., Kono et al., 1985]. The orøCline Allmendingeret al., 1983; Sudrezet al., 1983;Mdgard, 1984; can be viewedas a bendingof a narrowforearc strip that is Lyon-Caenet al., 1985;Sheffels et al., 1986]. accommodatedby a variableamount of shorteningin a mobilebelt In thispaper I explorethe hypothesis that Andean topography between the forearc and the cratonic foreland. is largelya resultof crustalthickening produced by structural The startingpoint of themodel is theassumption that thermal shorteningof thecrust and of upliftdue to thermalthinning of the weakeningis causedby convectiveand magmatic processes in the lithosphere.I use topographic data in combinationwith published wedgeof asthenospherelocated between the upper and subducted plates.Late Cenozoic magmatism, rather than supplying the large Copyright1988 by the American Geophysic• Union. volumeof crustalmaterial needed to supportthe elevated terrain, canbe takenas a signof theprocess of thermalweakening that Papernumber 7B6065. allowedthe compressionalfailure of the overridingSouth 0148-0227/88/007B-60655-5.00 American plate to happen. The along-strikevariation in

3211 3212 Isnc•: shorteningresults from variationsin thehorizontal width of the THEAL•O-PUNA PLATEAU zone of thermal weakeningin the overridingcontinental plate. Sincethe width of the wedge ofasthenosphere isdetermined by The Plateau the dip of the subductedplate, the along-strikevariation in As shownin Plate 1 andFigure 1, a singlebroad plateau is the shorteningis therebyrelated to the geometryof the subducted maincontinental-scale feature of the centralAndes between about plate. Althoughoversimplified and only kinematic,the proposed 15øSand 27øS. North and southof the plateauthe mountain model is an evolutionary,three-dimensional one. Its success system narrows considerably but continues with average emphasizesthe futility of trying to understandthe subductionelevations greater than 3 km alongthe high Andean cordilleras of processwith traditionalsteady state, two-dimensionalcross- Peru and alongthe Chile-Argentineborder (between 27øS and sectionalmodels. about34øS). For convenience,I distinguish these two segmentsof In the arid and largely unglaciatedcentral Andes where the Andeanmountain system located northwest and south of the denudationis minimizedand the tectonicprocesses are youthful plateauas the "Peruviansegment" (5ø-15øS) and the "Pampean andongoing, the topographyis a direct,first-order expression of segment"(27 ø- 34øS),respectively. theendogenetic processes forming the mountainbelt. Early work The hypsometriccurve for the plateau(Figure 2) hasa sharp such as Bowman [1909a, b, 1916] clearly recognizedthe break in slopeat an elevationof 3.65 kin, approximatelythe youthfulnessof central Andean uplift and its plateaulike character. elevation of the extensiveand internallydrained basins of the However,regional physiography has received very little attention Altiplanoand Puna. Muchof theremaining area of theplateau is in the dominantly geochemical, structural and geophysical characterizedby moderaterelief with elevationsin the rangeof approachesthat characterizemost modemstudy of the region about3.5-4.7 kin. As recognizedby Bowman[1909b] many (Coney[1971] is an outstandingexception). I presenthere a new years ago, peneplanation is the dominating fact in the viewof thecentral Andean physiography based on digitizationof physiographyof theregion. The otherdominating factor is the l:l,000,000-scaletopographic maps now available for mostof the pervasivevolcanism. Many of the highestpeaks are volcanic region.The computer-generatedcolor image of thisdigital data conslxuctionallandforms, although areas of highsu'uctural relief set is shownin Plate 1. The imagedepicts the regional-scale arealso found along the eastern edges of thenorthern and southern physiographyin a particularlyclear and accuratefashion. It partsof theplateau. preservesthe detail lost in a similarlyscaled conventional map in Theexu'usive material largely sits on top of theplateau rather which topographywould be shownby only a limited set of than formingthe volumeof the plateauitself. The upper subjectivelysmoothed contour lines. Miocene-Recentexu'usives form only a thincover on a surfaceof I first summarizerelevant information about the plateau,then beveledolder structures,as is clearly shownby the studiesof describea two-dimensionalmodel for shortening,and finally Baker andFrancis [1978] andBaker [1981],by numerousdetailed describe a three-dimensionalmodel for the late Cenozoic reportsof theareal geology of specificregions of theplateau, and kinematicaldevelopment of thecentral Andean zone quiteclearly by LandsatTM imagerythat I haveexamined for and mountainbelt. The digital topographicdata supportthe muchof theplateau. Large areas are covered by onlya thinlayer modelsqualitatively by highlightingimportant physiographic of extrusives,while the high stratovolcanos,although quite cluesto the natureof the plateauand by showingthe very close numerous,contribute relatively little to the total volume of spatialrelationships of the plateau morphology •o thestrucu•e of elevatedterrain. In detailedstudies of twoparticular areas of the the subductionzone. In addition,the topographicdata are used southernAltiplano of Boliviaand Chile that are largely covered by quantitativelyto constrainmodels of along-strikevariations in upperMiocene to Recentvolcanic rocks, Baker and Francis crustalshortening and thermaluplift. Additionalsources of [1978]and Kussmaul et al. [1977]estimate volumes that amount informationinclude published studies of geology,seismicity, and to an averagethickness of extrusivesof onlyabout 0.3 kin. With gravityand the now extensiveexaminations that I havemade of thedigital elevation data and the assumption that elevations above LandsatThematic Mapper (TM) falsecolor images of muchof the the plateaulevel of 3.65 km all representextrusives, an average centralAndes. The analysisof TM datais ongoingand will be thicknessof 0.5 km is obtainedfor theentire area of theplateau. reportedin subsequentpapers but is mentionedin thispaper in Thisestimate is probablytoo high since some of thehighland area severalinstances where particularly clear support exists for aspectsof theplateau is notvolcanic. of theproposed model. The plateauincludes a largearea of internaldrainage outlined It is importantto emphasizethat this paper focuses only on the by the watershedsshown in Figure 1. The internallydrained late Cenozoic developmentof the Andes with a model of region includesthe large Altiplano basinof southernPeru and deformationthat addsincrementally to the integratedeffects of Bolivia andthe more fragmented basins of theArgentine Puna and previousAndean orogenic episodes that began in theJurassic. It the ChileanAltiplano and Salarde Atacama. In manyof the is widelyrecognized, however, that the was a periodof basins,compressionally deformed upper Tertiary strata are relative tectonic and magmatic quiescenceand of probable exposed,and the bounding highlands were uplifted by basinward reductionof relief formedin earlierTertiary times. The Mesozoic vergingthrust faulting [e.g., Turner, 1960, 1961;Schwab, 1970; and early Tertiary intrusivesthat are now found west of the Everndenet al., 1977;Martinez, 1980; Jordan et al., 1983a;and modemAndes were presumablyformed in a regionof crustal Jordanand Alonso, 1987]. Much of the characteristicbasin and thickeningand uplift but arenow locatedin an areaof relatively rangemorphology of the ArgentinePuna is a manifestionof this low elevation and relief. Although the amount of crustal typeof compressionaldeformation, as is thethe areaaround Lake thickeningremaining from these earlier deformational episodes is Titicaca [Newell, 1949;Martinez, 1980]. The unconformity unknown,it is likely thata largefraction of themodem relief and between the generally flat and little deformed uppermost crustalthickness did in fact form duringthe Neogene.As a first Miocene-Recentextrusives and the older folded and reverse- approximation,I ignore the effects of pre-Neogenedeformation. faultedstrata is seenthroughout the plateauand gives evidence ISAC•: C• A•U:•A.•l•'raAu • BouvIA.•O,•oc•u,• 3213

ELEV TIO KM

25-

- 7 I I I i I

Plate1. Computerdisplay ofdigitized topography ofwestern South America. Subaerial dataare digitized aspoint elevations ona grid spacing of0.05 ø oflatitude and longitude. The data are mainly from 1:1,000,000-scale Aeronautical Charts published bythe U.S. Defense Mapping Agency; afew small areasmissing from those charts are filled in using available Bolivian and Peruvian topographic maps. An offshore strip from the coastline tothe outer slopeofthe Peru-Chile trench was digitized with the same grid spacing asthe land data from the l:l,000,000-scale bathymetric charts ofPrince etal. [1980].Remaining oceanic areas west of the Prince et al. map were taken from the NOAA SYNBAPS digital bathymetry f'de, resampled, andthen "spliced"tothe continental andtrench data with International Imaging Systems model 5'/5 image processing software. Each elevation sample isdisplayed asa pixel colored according tothe scale shown onthe left-hand side. No spatial averaging orsmoothing wasemployed. 3214 L•ACKS:CENTRAL ANDEAN PLATEAU AND BOt.rV•A•

12

s

14

16

2O , TRENCH AXIS , 3 KM OCEAN DEPTH W-B ZONE DEPTH, KM 22 ..... DRAINAGE DIVIDES ß.:!iiii!iii"iii:.. AVGELEV > 3 KM o VOLCANIC CENTERS

24 •}

26

28

3O

32

3480 W 78 76 74 72 70 68 66 64 62 Fig. 1. (a) Theshape of thesubducted Nazca plate, volcanic centers, drainage divides, areas with average elevations above 3 kin,and the axis of thePeru- ChileTrench. The contours of depthto thecentral part of theWadati-Benioff seismic zone depict the shapeof thesubducted plate (T. CahRland B. L. Isacks,unpublished manuscript, 1987). The Neogene volcanic centers were identified on Landsat MSS black and white imagery at a scaleof 1:1,000,000 andrepresent volcanos mainly younger than about 10 m.y. Thedrainage divides were traced and digitized from the 1:1,000,000-scale topographic charts usedfor the digitaltopography. In theshaded areas, average elevations are greater than 3 Ion, as averagedwithin a "square"moving window with dimensions0.25 ø in latitudeand longitude and a 0.25ø step.The axis of thetrench was traced and digitized from the 1:1,000,000-scalecharts of Princeet al. [1980]. Co)The coastline, bordersbetween countries, and certain geographic and tectonic provinces discussed in thetext, together with elevations above3 km andthe trenchaxis taken from Figurel a. "F-P CORD" indicatesthe Frontaland Principal Cordilleras located along the Chile- border,and "AB" is the AtacamaBasin (Salar de Atacarna)of Chile.

that the compressionaldeformations of much of the plateau Pampeansegments north and south of theplateau [e.g., Mdgard, occurredduring lat,e Miocene time. These deformationsare 1984; Jordan and Allmendinger, 1986]. Together, the late approximatelycoeval with importantcompressional deformational Miocenedeformations are for conveniencereferred to here as the phasesof the main cordillerasin the adjacentPeruvian and "Quechua"phase, although studies in different areasreveal ISAC•: C-• Ah'D•A.•PLA•AU A_h'U BOI.•A.• O•oo..u,,m{ 3215

12

S ...... PERU

14

BOLIVIA

16

18

20

22

o NAZCA PLATE z

24

26

ARGENTINA 28

30 PAMPEAN RANGES 32 b 3U/•NFF__p,t4ANDEZ •DGE. 3480 W78 76 74 72 70 68 66 64 62

Fig. 1. (continued) possiblysignificant regional variations intiming (e.g., compare mainly along the eastern edges of theplateau where drainage Jordanand Alonso [1987] with Mdgard [1984]. removesmaterial from the area and thereby accentuates relief. Exceptfor theconstructional volcanic edifices, much of the Notableareas of highrelief that can be seenin Plate1 are plateaunow has a lowto moderate relief. The reduction of associatedwiththe Mesozic and Cenozoic intrusions located along structuralrelief that occurred during and following the Quechua the northeastern sideof the Altiplano and with basement involved deformationsproceeded bothby ofuplifted blocks and by thruststructures of the Eastern Cordillera located along the fillingof adjacentclosed depressions, a kind of "cutand fill" southeasternsideof the Argentine Puna (see Figure 1). processthattends tolevel the terrain ina situation where mass is Althoughsome Pliocene-early compressional nottransported outof the internally drained system. This process deformation isdocumented forparts of the plateau, themain locus createdthenumerous thick sections oflate Cenozoic continental of the surface manifestation of post-Quechua compressional depositsthat characterize much of theplateau [e.g., Jordan and deformationappears to haveshifted eastward to theeastern Alonso,1987]. The nonvolcanic areas of high relief are located cordilleras andsub-Andean belts [Turner, 1966, 1969; Martinez, 3216 ISAC•: C'•• Am•A• PLAmAU Am) BocntIA• Ol•oO•m'•

WIDTH, KM 1980;Jordan and Alonso,1987]. An eastwardshift alsoappears 250 500 to have occurredin the adjacentPeruvian and Pampean segments [Mdgard, 1984; Jordan and Allmendinger, 1986]. Late Quaternaryextensional and strike-slipdeformations developed withinthe plateau, atleast in its southern andnorthern parts [Lavenu,1978; Sdbrier et al. 1985; Streckeret al. 1985; Allmendinger,1986]. Theseyoungest deformations appear to be smallinmagnitude compared tothe compressional deformations, althoughGreen and Wernicke [1986] claim otherwise for a partof the northernAltiplano of southernPeru.

The existenceof a broad,internally drained area of low relief, 0 0.5 1.0 1.5 0.5 1.0 AREA WITH ELEVATIONS ß H, AREA PER KM ELEVATION, now at elevations everywhere greater that 3.6 kin, can be MILLION SQ KM8 MILLION SQ KIdS interpretedas an effect of youthfuluplift, youngenough that a Fig. 2. Cumulativeand differentialhypsometric curves for the digitized 250-300Ion width of elevated terrain has not yet been attacked by elevations shown inPlate I for the region between latitudes 13øSand headwarderosion from the flankingdrainages. Although the 29øS,the region of the Altiplano-Puna plateau. In theleft-hand plot the drainagesystems along the edgesof the plateauoperate over a variableploued on the horizontal axis is the plan view area of landsurface nearly3-4 kmelevation differential, therates of headward erosion at elevations greater than the elevation plotted on the vertical axi•. The horizontalscale at the top of the figure showsthe effective"width" of a maybe small along much of the eastern and the western sides rectangular representation ofthe plateau asdiscussed inthe text. In the wherethe climates are quite arid. There are few constraints onthe right-handplotthe variable plotted onthe horizontal axisis the plan view detailedchronology of uplift,but nearly all workersagree on a late areaof landsurface per unit elevation taken as a functionof elevation Cenozoictime scale. A numberof publishedstudies report an plottedalong the vertical axis; the curve was computed asarea per 0.03 Inn accelerationinthe late Miocene [e.g., Segerstrom, 1963;Turner, and nonnaliz• toarea per I Inn. 1966; Galli-Olivier, 1967; Rutlandet al., 1965;Hollingsworth and Rutland, 1968; Guest, 1969; Mortimer, 1975; Laubaucher, 1978;Martinez, 1980; Crough, 1983; Tosdal etal., 1984; Jordan of the Pampean segment; if added toJordan and Allmendinger's andAlonso, 1987; Allrnendinger, 1986;Benjamin etal., 1987]. [1986] estimates, thiswould produce anet shortening ofabout 100 This timing would imply significantuplift duringand after the km for the Pampeansegment. Althoughsparsely distributed, pervasiveQuechua deformations. theseestimates of shortening allgive strong support forlarge amountsof late Cenozoicshortening in the centralAndes of the EasternThrust Belt orderof 100km and more, amounts that must be accounted forin modelsof crustalthickening. Theeastern side of theplateau is formedby themajor thrust Alongthe eastern side of muchof theAltiplano-Puna plateau beltsof theeastern cordilleras and sub-Andean zones. These thethrust systems verge eastward with the plateau overthrusting mainlyPliocene-Recent thrust zones can be tracedalong strike theforeland. In thoseregions the east facing topographic slope intothe Peruvian and Pampean segments. In all threesegments appears to coincidewith the thrust belt. Althoughthe western theeastern thrust belts are seismically active. In fact, it is now drainagedivide is largely maintained by thevolcanic constructions clearthat crustal seismicity in the central Andes is concentratedthat build up and "dam" the western edge of theplateau, it is only withinthe eastern thrust belts and that the Altiplano-Puna in limitedareas along the eastern watershed (such as the Frailes arerelatively aseismic [Chinn and Isacks, 1983; Suardz et al., ignimbritefield near latitudes 19øS-20øS) that extrusives appear 1983;Froidevaux and Isacks, 1984]. This result is in accordwith toexercise control on the position of theeastern drainage divide. thestudies mentioned in the previous section that conclude that Hencefor much of theeastern side of theplateau the divide may shallowcrustal deformation hasalso become concentrated in the becontrolled byuplift of the hanging wall of the thrust system. In easternthrust belts. thecross section near 22øS published by Allmendingeret al. Severalestimates of theamounts of shorteningin the thrust [1983]the high eastern edge of theplateau is locatedabove the beltsare now available. Allmendinger et al. [1983]report at least inferredposition of the footwallramp where the d6collement 60 km of shorteningbased on a palinspasticreconstruction of rampsdown to deeperlevels of thecrust. This result implies a Mingrarnmet al.'s[ 1979]section through the sub-Andean fold- directrelationship of the thrusting and topography. The inferred thrustbelt near22øS. Sheffelset al. [1986] find 25-36 % rampingof thed6collement beneath the plateau implies that the shorteningof the fold-thrustbelt near18øS where the belt is shorteningmanifested in theeastern thrust belt is likelyto be widest.This yields shortening values of about150-225 kin. For absorbedatdepth beneath the main plateau uplift. I will usethis the sameregion, Lyon-Caen et al. [1985] estimateshortening concept in theproposed model of shortening. amountsbased on analysisof gravitydata greater than 100 km and South of about 22øS the easternmargin of the plateauis mention possible values as high as 400 km. Jordan and increasingly complicated by the appearanceof structures Allmendinger [1986] estimate10-20 km shorteningacross the characteristicof the Pampeansegment, that is, by thick-skinned Pampeanprovince of westernArgentina near 30ø-3 løS which, deformationswith botheast and westvergerice that are foundin when addedto about 50 km shorteningacross the Precordillera the SantaBarbara system, the ArgentineCordillera Oriental, and fold-thrust belt [Fielding and Jordan, 1987; Allmendinger, thenorthern Sierras Pampeanas (Figure lb). The morphologyof personalcommunication, 1987], yields about70 km for a section the plateau edge becomes more irregular (Plate 1) and is acrossthe Pampeansegment. characterizedby large relief associatedwith Pampean-style, Noneof theseestimates is tighdyconstrained, and none include reversefaulted basement blocks. Allrnendinger[1986] showsthat the older Quechuadeformations located west of the thrustbelts. the foreland-vergingthrust system that boundsthe southeastern Mdgard [1984] estimates115 km for a part of the Peruvian marginof the Punaconsists of shorten echelonthrust segments segment that does attempt to include all late Cenozoic whichtogether accommodate shortening amounts of only tensof deformation.Ramos [1985] estimatesabout 30 km of Quechua kilometers. The shorteningaccommodated by the othermajor shorteningin themain cordillera at about33øS in the southernpart structuresin thiscomplex region is unknown. IsAc•: (•,,•• PLATEAU • BOLrVL,U• O•oo,,u,a{ 3217

Still farther south in the Pampean segment itself, the Chile andsouthem Peru may be relatedto an increasingmount of Precordillera,a thin-skinnedfold-thrust belt, forms the eastern glaciation at high elevations in the Western Cordillera. I slopeof themain cordillera (the Cordilleras Frontal and Principal; hypothesizethat the monoclinal structure is a continuousfeature seeFigure lb). In Perua similarrelationship exists between the throughall threeof the segments.Between latitudes 18øS and Cordillera Oriental and sub-Andeanzones with respectto the about26øS along the central plateau segment, transverse drainage Peruvian "Puna"and Western Cordillera. In both the Pampean to the Pacific is virtually nonexistant[Mortimer, 1981]. Thus the and Peruviansegments, d6collements beneath the easternfold- westernmonoclinal edge of the Andes is there preservedand thrustbelts are inferredto ramp down to the west into the crust particularlyclear. beneaththe easternparts of the high centralcordilleras [M•gard, 1984; Jordan and Allmendinger,1986]. As proposedfor the Relationto SulgtuctedPlate Geometry centralplateau segment, theeastern thrust belts inthe Pampean Western slopeand asthenospheric wedge.Plate 1and Figure 2 expressionsandPeruvian of shorteningsegments thatare is also accommodated likely to atbe depththe beneathsurface demonstrate thatthe great westernslopeofthe Andesveryclosely parallelsthe trenchaxis and the trendof the inner slopeof the the main cordilleranuplifts. An addedelement in the Pampean trench, that is, parallelsthe outline of the leadingedge of the segment is the significant amount of distributed foreland . Both featuresshow distinctchanges in deformation east of, and probably beneath, part of the trend near latitudes 17øS, 27øS, and 33øS coincident with Precordilleranfold-thrust belt [Fieldingand Jordan, 1987]. boundariesbetween major segmentsof the central Andean

Western Monocline subductionsystem. Along thecentral plateau segment the western edge of the zone of volcanic centers(Figure la), the volcanic As shownby the unsmootheddigital terrain model of Plate 1, "front,"is alignedalong the westernplateau edge and tracks the the westernslope of theplateau is a strikinglyregular feature that main westernslope except in the regionof 23ø-24øSwhere the is continuouslytraceable north and southalong the flanking slopebifurcates to formthe Atacarna Basin. Peruvianand Pampean segments. The smooth morphology of the In a studyof thegeometry of theChilean interplate boundary westernslope contrasts with the complex and variable morphology between about latitudes 20øS and 40øS, Kadinsky-Cade [ 1985] of the thrustbelts along the eastern side of theplateau. Although showsthat the main westernslope of the plateau(or of the high extrusivesbuild up, spillover, and thus cover large areas of the cordillerato the southof the plateau)is locatedapproximately westernplateau edge, there are "windows" where an underlying above the downdipend of the interplateboundary. This is subvolcanicbasement can be seen.In thesewindows the edge and supportedby theparallelism of theslope and the contour of 75 km slopeis formedby deformedpre-Neogene rocks that have been depth to the Wadati-Benioffzone shownin Figure la. The beveledto a relativelysmooth surface, a surfacethat is now tilted western slope would thus track the axis of the trench if the to form the great westernslope Andes. Much of the slopeis interplateboundary were to haveboth a constantdip anddowndip coveredby a thin layerof extrusivesand by erosionaldebris widthas functions of along-strikedistance. The geometry of the forminga giantalluvial drape. These features can be clearly seen Chileaninterplate boundary does not appear to change appreciably onLandsat TM imageryand on Chilean regional geological maps northof at least33øS. andhave led several investigators to suggestthat the uplift of the However,a morerevealing association may be betweenthe plateauwas accommodated by a tiltingor flexureof thecrust [e.g., locationof the westernslope and the tip of the wedgeof mantle Mufioz,1956; Mortimer, 1973]. I carrythis forward by proposingthat exists between the inclined subducted plate and the horizontal that the westernedge of the plateauis simply a crustal-scaleoverriding plate. The wedge is presumablyasthenospheric monocline. materialthat is relativelyhot andweak. The tip of the wedgewill Althoughevidence for lateMiocene-Pliocene compressive or be locatednear or just downdip of theinterplate boundary. Nearly extensionaldeformations have been reported for someareas of the all modelsof the generationof subduction-relatedmagmatism westernslope [e.g., Thomas,1970; Labsen, 1982], the amount of involveinteractions between subducted material with the mantle lateCenozoic deformations associated with these structures appear materialin the wedge;thus the tip of thewedge probably controls toosmall and spatially discontinous to explainthe large elevation thelocation of themagmatic "front." The shapeof thewedge and differentialacross the slopeand the continuityand regularity of the movement of the subductedplate determine a special morphologyalong-strike. Important Cretaceous-early Tertiary convectivesystem, as shown by many publishedtheoretical deformationshave been recognized along parts of theslope [e.g., studies.It is thusquite reasonable to supposethat convection in Vicente et al., 1979]. Someof thesestructures may have been the wedgepromotes major modifications of theupper plate in the reactivatedin the late Cenozoic,but most of the deformation form of magmaticintrusions, thermal thinning, and consequent occurredtoo longago to be associatedwith theuplift thatformed mechanicalweakening. I thuspropose that the parallelismof the themodem plateau and cordilleras. westerntopographic slope and the trench is a resultof thefact that In contrastto the northernChilean region, the westernslope thewestern slope tracks the tip of the asthenosphericwedge. along the Pampeansegment is considerablymore incisedby In areasof nearly fiat subduction,as in the Peruvianand transversedrainage into the . This is dueprimarily Pampeansegments, a wedge-shaped region of manfieis foundfar to the southwardlyincreasing amount of precipitationfalling on inlandof the trench,nearly 800 km in eachof thesegments, where thewestern side of theAndes. Nevertheless,the clear along-strike the subductedplate finally bends downward and developsa continuityof the westernslope (demonstrated in the color image significantinclination. However,there is goodevidence that in of topography)suggests that the monoclinalflexure continues bothsegments the present subducted plate shape has evolved from throughthe Pampean segment. The sameargument applies to the a more steeplydipping, late Miocene geometrythat probably Peruviansegment, where transverse drainage is alsodeveloped. existedduring much of thedevelopment of themain cordilleras in Examination of TM images suggeststhat the northward thetwo segments.In bothsegments the westernslopes are closely developmentof the spectaculartransverse canyons of northern associatedwith thewestern fronts of thenow extinctMio-Pliocene 3218 Isnc!r_s:L-M•RAL AND!•AN PLATEAUAND BOLr•IAN ORcaa.m-• magmaticzones. Well developedmagmatic arcs existed prior to trendof the intermediatedepth Wadati-Benioff zone (compare about7-10 m.y., indicatingthat astheno spheric wedges existed Figure l a and the topographicimage). The upper slope and beneath the cordilleras then. The near extinction of the magmafieare both swingback towardthe trenchnear 24øS, and magmatismin late Miocenetimes is takento datethe flatteningof farther south,a single westernslope againparallels the trench. the subduetedplates in bothsegments [Noble and McKee, 1977; The intermediate-depthcontours (125 km and deeper)of the Coira et al., 1982;Jordan et al., 1983b;Jordan and Gardeweg, seismiczone do not swingback, however,but continuealong a 1987; Kay et al., 1987]. I thus hypothesizethat in both the SSE trend and thus divergefrom the shallowpart of the plate Peruvianand Pampean segments the main cordilleras were located boundary.This divergenceproduces a fiat part of the subdueted over late Miocene asthenosphericwedges and that the western slabthat progressively widens southward. topographicslopes mark the tipsof thoseprevious wedges. A clear implicationof the southwardflattening is that the In fact, the topographicslopes are both now locatedwhere the curvature of the subduetedplate downdip of the interplate subduetedplates bend back from the 200-30ø inclinationsof the boundarymust change from convexupward where the plate dip plate boundariesto more nearly horizontalinclinations. The continuesto steepenwith depthto concaveupward where the plate geometryof the plateboundary itself is quitesimilar throughout bends back to nearlyhorizontal. This canbe seenby comparing the threesegments, and it is only below depthsof about75-100 the sectionsin Figure 3 (see alsoHasegawa and Sacks[1981]. km that the striking variations in dip of the subductedplate Although the seismicity data still do not determine the exact develop. If in the regionsof horizontalsubduction the subduetedlocation of thetransition in curvature,it doesappear to be southof plate were straightenedout to maintainthe nearly30 ø dip of the the doglegjog in the trendof the magmaticfront near 24øS. The interplateboundary, the wedgetips so definedwould track the transitionmay well be locatednear the regionof 270-28ø where westernslopes. thereis a subtlebut clear changein the trendsof both the trench Along-strikelimits of the plateau. As shownin Plate 1 and axis and of the westernslope (see Plate 1 and Figure 2). The Figure 1, the along-strikeextent of the plateaucoincides with the developmentof upwardconcave curvature of the subduetedplate active Plio-Quaternarymagmafie are. In southernPeru the could thus be associatedwith the westwardprotrusion of the northern limits of the plateau and volcanic are quite closely westerntopographic front that developssouth of 27øS. This coincidewith the sharpflexure (near 15øS)between the steeply wouldplace the transitionadjacent to thenarrowing of theplateau dipping and nearly horizontalsegments of the subductedplate. andnear the southernend of themagmafie are. The flexure is developeddowndip of the region where the map view shape of the leading edge of the overriding plate has a MODELFOR PLATEAU UPLIFF changeinseaward curvature fromconvex (Peru) toconcave Altiplano section (southern Peru-northern Chile). The strike of the flexure is approximatelyaligned along the directionof relative motion As shownin the precedingsection, a major featureof the betweenthe converging plates [Bevis and lsacks, 1984; Grange et centralAndes is the east-westasymmetry of themountain belt in al., 1984]. Theseresults suggests a close connection between the termsof lateCenozoic structures and morphology: on thewestern developmentof theflexure and the shape of theoverriding plate in side a monoclinelikecrustal-scale flexure parallelsthe plate mapview. boundary,whereas a major thrustbelt formsthe easternside. The However, the correlationof plateau and subductedplate shorteningaccommodated by the easternthrust belt is roughly geometryfarther south is not assimple as indicated by Barazangi coevalwith a plateaulikeuplift of theAltiplano-Puna. The cross- andlsacks [1976] or by Jordanet al. [ 1983a].Near 27øS the high sectionalmodel shown in Figure4 attemptsto accountfor these plateaunarrows, and a singlecontinental drainage divide replaces features for a representativesection through the Bolivian Altiplano thetwo that delineate the Puna-. Latitude 28øS is (The modelapproximates a cross section in the regionof 21- approximatelythe southernend of the zone of Plio-Quaternary 22øS). magmatism.In contrastto theserather sharply defined southern In thefirst stage (Figure 4a) theupper plate is thinnedduring boundariesof the plateauand magmatic arc, the Wadati-Benioff anepisode of relativelylow-angle (but not horizontal) subduction. zoneonly graduallyfiattens between latitudes of about21øS and The heatingand thinning of thelithosphere may have started 25- 32.5øS [Bevis and lsacks, 1984; T. Cahill and B. L. Isacks, 30 m.y. ago, when widespreadmagmatism resumed after an unpublishedmanuscript, 1987]. Significantflattening is already Oligocenehiatus [e.g., Baker and Francis, 1978;Coira et al., apparentat latitudesof 27ø-28øS. 1982;Jordan and Gardeweg,1987], and whenthe directionof The developmentof the flatteningbetween 21øS and 24øS is plate convergencebecame more nearly normal to the plate associatedwith a marked anomalyin the morphologyof the boundary[Pilger, 1984;Cande, 1986; Pardo-Casas and Molnar, westernAndes in the regionof the AtacamaBasin. Outsideof this 1987]. region,topographic profiles from the trench axis across the forearc The modelstarts with the assumptionthat processesin the typicallyshow a benchlikeshape formed by twosteeper parts, the wedgeof astheno sphere located between the subducted and upper innertrench wall and the main western slope of theplateau or high platesare effective in thinningthe upper plate. Suchprocesses cordilleras,separated by a lesssteep part that includesthe mayinclude convection in the wedge driven by themotion of the coastline.In theAtacama Basin region the western slope of the subductingplate, intrusion of magmasinto the upper plate, and plateaubifurcates in mapview, and a secondbench (occupied by erosionof thelower part of theupper plate by lithosphericstoping the Atacama Basin) is formed at an elevation intermediate or delamination.I hypothesizethat theseprocesses operate within between that of the coastal bench and the high plateau. The a swath whose cross-sectionalhorizontal width is the distance bifurcationactually begins near latitude21øS to form the upper betweenthe wedgetip and the point at whichthe wedgereaches basin of the Loa River. The lower part of the western slope somespecific vertical thickness. This thicknesshas to do with the continuesto parallel the trenchaxis, while the upper slopeand as yet poorlyunderstood convective processes in the wedgeand magmaticarc follow a morenearly SSE trend that is parallelto the canonly be guessed.An estimatefor the thicknessof 250 km is ISAC•: CZ• Al•mmA•PLAmZAU A• BOUVIA•Oaoo. u• 3219 K+•r- CHILE • •EE=10'1- I __ f WESTERN ALTIPLANO EASTERN -5• -IVi•N•LiNE •L,,•c• CORDILLERASUBANDES

MOHO NAZCAPLATE ...... ßß ß ß ß SOUTH AMERICAN 100 ß ß • LITHOSPHERE KM • • = 1200 ø _

200 A

300 200KM 4(X) 6(X) 800 0 1000

5

-5 •_ • CORDILLERASA•F_R•_, ,N,T_A_L'•A PRECORDILLERA SIERRAS PAMPEANAS

100 KM 'NAZCAPLATE mm m 1200 ø .

B

300 200KM 400 600 800 0 1000

Fig.3. Twocross sections showing topog•y andhypothesized deep structure. The upper section crosses the wide part of thecentral Altiplano, while thelower section crosses the Pampean fiat slabsegment. The western endpoints and the azimuths of thesections are 20.5øS, 72.0øW, and 083 ø for the uppersection and 31.0øS, 72.0øW, and 090 ø for the lower one, respectively. Each topographic pmfde (upper plot in each section) shows digital elevations averagedalong a swath30 krnperpendicular to, and 10 krn along, the section. The plothas a verticalexaggeration of 10:1. Theposition of the subductedNazca plate is fixed by the location of theWadati-Benioff zone (T. Cabilland B. L. Isacks,unpublished manuscript, 1987). The wiggly line areasindicate shortened and thickened ductile lower crust. The upper crust overthrusts the foreland along a simplifiedsingle with motion indicated bythe half arrow. Note the implied monoclinal structure ofthe upper crust forming the western topogra•ic slope of the Andes. Other Pampean faults in thelower section are taken from Jordan and Allrnendinger [1986]. The 1200øC isotherm indicates lithospheric thinning.

plausiblein view of seismicstudies of modemasthenospheric dip, the widthincreases to a maximumand then decreases to zero wedges[e.g., Barazangi and lsacks,1971; Utsu, 1971; asthe dip approaches horizontal. A 20ø dip is a plausibleguess Barazangiet al., 1975;Hirahara and Milcumo, 1980], but this forthe dip for maximum width; that is, at that dip angle there may choiceremains a somewhat ad hoc starting point for the model. still be enoughvolume in the asthenosphericwedge for the With a fixed verticalthickness the horizontalwidth of the lithosphericthinning processes to operate.With a dipof 20ø and thermallyaffected swath will increasewith decreasingangle of dip the characteristicthickness of 250 kin, the horizontalwidth of the of thesubducted plate. However,with a dip angleof zero,litfie or wedgewould be 500 km asshown in Figure4a. no asthenospherewould be locatedbetween the two plates. It is Magmatism,implying an efficientadvective transfer of heat thusreasonable to supposethat as the dip decreasesfrom vertical intothe upper plate, would likely be associatedwith thethermal to horizontalin a plot of widthof thermaleffect versus angle of thinningof the lithosphere.Thus both volcanismand thermal 3220 Isncats:• •n• PLAI•AUAW'D B(3LiVIA• O•oo.m•

HEATEDAND WEAKENED ZONE 27-14 MY o KM •REARc ...... '...... FORELAND

lOO

A

2oo

, ,I 200I I 400I I 600 800I

B FORARC FORELAND

100KM. c QUECHUA

L 130KM, D

v v v v v v v v v v v v v v v v,.ff-•a.- I POST-QUECHUA

PRESENT 0v[- TRENCH ALTIPLANO-PUNA KM FOREARC FORELAND NAZCA PLATE

100 ASTHENOSPHERE•

1200 ø

2OO 0 200 400 KM 600 800

Fig.4. Simpldiedmodel of lateCenozoic evolution of a sectionnear 21 ø--22øS. Faults and ductile lower crest are shown as in Figure3. Theinitial and œma].stages are depicted in the(a) upperand (a) lowersections, while the three middle sections show a veryschematic (but balanced) block model of the cmstaldeformation. (b) is a simplifiedinitial cmstal section. (c) Thecross hatching indicates pervasive shortening of theUpl•er britde cmst during the "Quechua"phaseof deformation. (d) Forclarity the ove•ktrusting of the upper crest onto the foreland is shownwithout any isostatic flexural adjustment; (e) the isostaticadjustment is shown. upliftare implied in theearly-middle Miocene phase depicted in dippingsubducted plates and the contrast between this mode of Figure4a, andboth occur above a wide asthenosphericwedge. plateconvergence and the onecommonly found in the western The thermalthinning would increase the temperatureof the Pacific,where the sinking of theoceanic plate "draws" the island uppermostmantle and thus weakenthat strongestpart of the arc seaward,are well recognizedand are the subjectof several continentallithosphere. If theplate is subjectto horizontalstress papers[e.g., Wilson and Burke, 1972; Molnar and Atwater, 1978; producedby theglobal convective system, the weakening would Uyedaand Kanamori, 1979; Dewey, 1980]. The increased rates of localize and concentratedeformation in the thermallyaffected convergencein late Miocenetimes may have helpedtrigger the zoneand thus closely tie the spatialextent of the deformationto failure[Pardo-Casas and Molnar, 1987]. theposition of theasthenospheric wedge. The compressivefailure of the weakenedswath of lithosphere I assumethat the combinationof globalplate motions and is postulatedto haveoccurred in two phases.A widepreadand mantleconvection led to a situationwhere the continentalplate pervasivehorizontal shortening and vertical thickening of thecrust overrides the oceanic plate and produces a pervasive (Figure4c)corresponding to the Quechuaphaseof deformation compressionalstress in theSouth American plate. The association was followed by a concentrationof deformation along the eastern of compressivestresses in theupper plate with relatively shallow sideof theorogen where the upper crust thrusted over the foreland Islet. s: • A•• Pl•u • BOL• Oao• 3221 along an eastwardlyverging thrust belt (Figure 4d). The latter Cretaceousand lower Tertiary plumns found in thecoastal regions phaseproduce• the basic east-we•t asymmetry of the structureof of Peru and Chile, the samebatholiths that serveas modelsfor the plateauand accountsfor the evidencethat the widespread crustalthickening by magmaticintrusion, are now locatedat low Quechua deformation was followed by a concentrationof elevationsand are not part of the presentelevated mountain mass compressivedeformation along the eastern thrust bell of the Andes. With the 100-150 m.y. time scale,it seemslikely In thelatter phase, deformation of the brittle upper crust was thaterosional remov•l of mass would be able to keep up with the concentrated along an eastern thrust belt, although the slowrate of additionof crustalmaterial by magmatism. underthrustingforeland continued to compressthe ductilelower As a first approximation,I thereforeneglect the magmatic crustwest of the thrustbelt. The consequentthickening of the contribution,partly to seeto what extentcrustal shortening and lower crustproduced a plateaulikeuplift west of the thrustbelt lithosphericthinning alone can accountfor the plateauand partly where the rising upper crust suffered little additional post- because the intrusive volumes are quite uncertain. When Quechuadeformation. "Cut-and-fill"processes localized by the information about crustal shortening,thickness, and density internaldrainage system of the plateaureduced the relief of the becomessufficiently accurateto constrainmodels, then the Quechuastructures on the upliftingplateau and thereby produced discrepenciesbetween observed and calculatedparameters might the characteristicphysiography of the Altiplano-Puna. The be usedto estimatethe intrusivecomponent. conceptfor the secondstage of compressivefailure [lsacks, 1985] issimilar (but independently derived)to that proposed byZhao Application tothe Flat Slab Segments and Morgan [1985] for the uplift of the . In the central Andesthe "hydraulicram" of Zhao and Morgan is the The modelfor the uplift of the plateaushown in Figure4 can underthrustingSouth American foreland. also be applied to the adjacentalong-strike segments of the The modelshown in Figure4 impliesthat on the westernside convergentzone. The "plateau"between the monocline and thrust of theplateau a monoclinalflexure accommodates the differential zone is taken to narrow into the high cordillerasof the two uplift of theupper crust at the westernlimit of ductilethickening segments,becoming the CordillerasPrincipal and Frontalin the of the lower crust. The monoclinalflexure actually appears to Pampeansegment and the CordillerasOccidental and Oriental in form the western slope beneath the extrusivecover and is the Peruviansegment In fact, the Peruvian"Puna" [McLaughlin, particularlywell preservedin the hyper-ariddesert of the central 1924; Coney, 1971], a narrowhigh plateauregion with mature partof theplateau. The westernlimit of thickeningof the lower relief characteristicssimilar to the Altiplano,is locatedbetween crust is located at the westernboundary of the heated and theCordilleras Oriental and Occidental in Peruand is a northward weakenedzone in theu• plate,and this in mrnis locatedabove continuationof theAltiplano. Coneyshowed a hypsometriccurve the tip of the asthenosphericwedge separatingthe converging for elevationswithin a small area of the centralPeruvian Andes plates. The positionof the westerntopographic slope is thereby thatis quitesimilar in form to thatshown in Figure2. The widths tied closelyto thelocation of the magmaticfront and the geometry of thepostulated shortened and thickened lower crusts beneath the of thesubducted plate. cordilleran"plateaus" in the Peruvianand Pampean segments are The modelshown in Figure4, althoughhighly simplified, is smallerand involve less shortening for thesame amount of uplift scaled accuratelyto representrealistic amountsof crustal The wedgesof asthenospherehypothesized to be responsible shorteningand thickeningwith preservationof cross-sectionalfor the weakeningof the swathsare no longerlocated beneath the area; it is a "balanced"cross section. The plateau'spresent Peruvianor Pampeancordilleras, as alreadydiscussed, but are elevationis supportedisostatically by a combinationof crustal postulatedto have migrated inland. The "flattening"of the thickeningand thermal expansion (or lithosphericthinning in the subductedplate is partly the effect of the continentalplate sensediscussed by Turcotte and MacAdoo [1979]). Measure- overridingthe downwardbend of the Nazca plate and thereby mentsof crustalthickness and densitiesin the regionare still too movingwestward relative to theasthenospheric wedge. The tip of crudeto provideuseful constraints on the detailsof the model. In the wedgethus appears to move eastwardor inlandrelative to the the f'malstage (Figure 4e), a 4-km-highplateau is compensatedby upperplate. The tip is now locatedat the easternedges of the a thickeningof the crustfrom 40 to 65 km and by a thermal deformedforelands of the two flat slab regionsand may be thinningof thelithosphere from 140to 70 km. Thisis a workable responsiblefor minor Neogenevolcanism in these regions but nonuniquemodel. If the shorteningrequired to producethe [Stewart,1971; Kay et al., 1987]. thickeningis averagedover 15 m.y., the convergencebetween the In the middle Miocene before the overriding, a relatively forelandand forearc would be about1 crn/yr. Thisis a reasonablenarrow asthenosphericwedge is implied by the narrow late numberfor forelandfold-thrust belts [Allmendinger et al., 1985]. Miocenemagmatic arcs in the Peruvianand Parnpean segments. The model doesnot includeadditions of magmaticmaterial The narrowwidth of the wedgeimplies that the subductedplates from the mantle. As pointed out above, the volume of late thendipped more steeply beneath these segments than beneath the Cenozoicextrusives contribute little to the overall volume of centralBolivian segment. This increaseddip is supportedby Kay elevatedcrust. The intrusivevolume is unknown,but Kay and et al. [1987]. The narrowthermally weakened swaths also failed Kay [1985]estimate a ratioof intrusiveto extrusivevolumes of underthe pervasivecompressional stress in the upperplate and aboutonly 1:1 for the Aleutianisland arc. Baker and Francis thereby producedthe crustal thickeningbeneath the high [1978] discussratios near 10:1that would be requiredto buildthe cordillerasof thetwo segments. Andes. This would lead to an extremelyhigh rate of magmatic As the continentalplate passed over the asthenosphericwedge additionif the late Cenozoictime scalefor the uplift is assumed in the Pampeansegment, it wassomewhat thinned and weakened, [Isackset al., 1986]. Baker and Francis,in fact, proposethat but the overriddenand still nearlyhorizontal Nazca plate quickly intrusiveshave been thickeningthe crust since Jurassictimes. replacedthe lostcontinental lithsosphere section to form a kind of This time scale,however, seemsat odds with the considerable doublelithospheric structure. A thin, weak interfacebetween the evidencementioned in the precedingsection that much of the subductedand upper plates would still effectivelycouple the two uplift occurredin the late Cenozoic. In addition,the massive platesin termsof verticalisostatic adjustment. The low regional 3222 lsncr, s: CB• AND•A• PLnT•AU ANDBot• OaOOJNB elevationof the SierrasPampeanas east of the Precordillera(see deformation.In mapview theboundaries of the forearcwould be Figure3) couldthus reflect the isostaticeffect of the combined the axis of the trenchand the westerntopographic slope. A lithosphericthicknesses, while thedeformations manifested by the movementof the forearcrelative to the SouthAmerican SierrasPampeanas could result from the decoupling of thethinned mustbe takenup in an interveningzone of deformation.The (and weakened)upper plate from the lower plate in respectto movementof the forearccould consist of a rotationof the forearc horizontallithospheric stresses. The stresssystems in the as a wholecombined with a changein its mapview shape.It is overridingplate and in the Nazca plate may be coupledmainly importantto point out that becausethe forearcis very long and alongthe interplateboundary located west of the mountainbelt narrow, a rather substantialchange in shapecould occur with and transmittedonly throughthe upperplate. If the overriding relativelylittle deformationin theforearc itself. plateis fifinned,then the horizontal compressive stress is increased The spatialrelationship between the presentAndean forearc in thethinned section combed to thethicker craton to theeast by shapeand the geometryof thesubducted Nazca plate can be taken theratio of respectivelithospheric thicknesses. to suggestthat the forearc did, in fact,change shape during the late An alternativeview is thatthe foreland deformations in regions Cenozoic. The divergencesouth of 22øSbetween the strikesof of flat slab subductionare producedby the effect of shearing thesubducted Nazca plate at depthsgreater than about 150 km and stressesaccumulated along the large areaof contactbetween the the modernplate boundarysuggests that the shapeof the deeper two plates[e.g., Dickinson and Snyder,1978]. The problemwith partof the subductedNazca plate reflects an olderplate boundary thishypothesis is thatit doesnot explain why deformationsdo not configurationthat is lessindented than the present one. Figurel a extendeast of the zone of plate contactwhere the accumulatedshows that the plan view shapeof the 150-kmcontour is distinctly stressesmight be expectedto be largest.It seemslikely thatshear less embayedthan the trace of the modem trench axis. The heatingwithin theplate boundary interface zone (located beneath implicationis thatthe westwardconcavity of the forearcincreased the Chilean forearc) would reduce the strengthof the zone duringthe subductionof the seismicallyvisible part of theNazca downdipof the interplateboundary [e.g., Yuen et al., 1978]. This plate,i.e., duringthe past12 m.y. [Wortel,1984; T. Cahill andB. is supportedby the apparentmaximum depths of interplateL. Isacks,unpublished manuscript, 1987]. earthquakesof about 60 km [e.g., Chinn and backs, 1983; Along-strikevariations in theamount of shorteningin the zone Kadinsky-Cade,1985] andthe lackof evidenceof giantinterplate of accommodationbetween the forearc and South American craton eventsthat would involveseismogenic slippage along the nearly couldimply a changeof shapeof the forearc. Sincethe Puna- horizontalparts of the interfacezone. Thus thecoupling between Altiplano is locatedinland of the indentationof the Peru-Chile the plates in terms of horizontal compressivestress may be coastline,it is reasonableto supposethat the plateauand the concentratedin theforearc interplate boundary. indentationare related. In particular,the indentationcould be Bird [1984] arguesthat Laramidedeformations of the western produced,or at least enhanced,if the along-strikevariation in United Statesresulted from a shearingof the lower crust and shorteningwere a maximumin Bolivia adjacentto theindentation. lithosphereby the eastwardmoving, horizontalFarallon plate The fact that this is where the Andes reach maximum width beneath western . The rotational shearing encouragesthis point of view. transportedlower crustalmaterial eastward and thereby thickened Data to constrainthe movementof the forearcrelative to the the forelandcrust. The Pampeandeformations are at an early cratoninclude estimates of the amountsof shorteningacross the stagecompared to the durationof the the Laramideorogeny, so Andesat variouslatitudes and paleomagnetic determinations of the effectsof shearingof the lower Pampeancrust would not yet rotationof theforearc relative to stableSouth America. These two be so large as in the Laramidecase. The topographicprofile linesof evidenceare developedin the followingsection. Direct shownin Figure3 is stillprobably dominated by theeffects of the suucturalevidence of deformationsassociated with bending of the Miocene episodeof steep subductionand associatedcrustal forearcis sofar generallylacking, but, asmentioned above, these shortening and thickening beneath the high cordilleras and deformationsare likely to be smallrelative to thosein theback arc Precordilleraand by the isostaticeffects of the overridenNazca that absorbthe relative motion between the forearc and South plate. American craton. It is interestingthat althoughforeland basement deformation of the Peruviansegment is reported[e.g., Sudrez et al., 1983; Chinn Along-StrikeVariations of Cross-SectionalArea and lsac•, 1983], thesedeformations appear quite reducedin extent and amount comparedto the Pampeansegment. This Area measurements. The amount of shorteningis not contrastsuggests that flat subductiondoes not necessarilyproduce accuratelyconstrained in any single crosssection through the Laramidelike structm'es. centralAndes, although rough estimates are availablefor several sections.I attemptto add to theseestimates by consideringthe

MODEL FOR TH• BOLIVIAN OROCLINE cross-sectionalarea of the uplifted terrainas an effect of crustal shorteningand thickening. The digitized topographyis easily The terms"back arc" and "forearc"were originally introduced usedfor thispurpose. to describewestern Pacific island arcs. Karig and Moore [1975] The cross-sectionalarea is measuredfor eachof the22 profiles discussthe forearc as a deformablebeam (in map view), and shownin Figure5. The elevationsare averaged within a moving Karig et al. [1978] arguethat the Marianasforearc deformed window100 km wide (perpendicularto the crosssection)by 10 somewhatin accommodatingthe openingof the Marianasbasin km long,and the resulting curve integrated to givethe total cross- (see also McCabe [1984], and Faure and Lalavde [1987]. sectionalarea above a baselevel of 0.3 km. The baselevel is an Althoughthe concept of a narrowforearc platelet is mostobvious estimateof the elevationof the plainseast of the the Andesthat whenapplied to westernPacific Island arcs with largeamounts of are unaffectedby the uplift. The areasso calculatedare plotted back arc extension,it can also be usefully applied to South againstthe along-strikeposition of the profilesas measuredby America. The Andean"forearc" is a long narrowplate located distancealong a line parallelingthe trench axis but located300 km betweenthe interplateplate boundaryand the zone of "backarc" to the east of it. The distanceof 300 km is measuredfrom the Isnc•s: Camm• Al,n•n.n l:•nu • Bouv•n.• O•oo•u,u• 3223

10 probablybegan earlier when widespread magmatism started about s 25-30 Ma. The trenchaxis, digitized from the 1'1,000,000 bathymetric chartsof Princeet al. [ 1980], is an easilydefinable and convenient delineationof the shapeof the forearc. Although a line in the centerof the forearcwould be more accurate,I approximatethe deformationof the forearcby the deformationof a line delineating its westernborder. A problem with this simplificationis that accretionof sedimentsduring the time period consid• wouldbe

19 equivalentto a movementof thetrench axis seaward relative to the 18 craton. However,this effect is not likely to be appreciableexcept 17 16 south of 33øS where large volumes of accretedsediment are 15 present[Schweller et al., 1981]. 14 In developing the model the present geometry was taken backward in time. The presenttrench axis was first rotated eastwardtogether with the South America plate accordingto 12 Chase [1978]. Then, to model the Andeanshortening, the axis was movedrelative to the cratonby a westwardtranslation and mapview deformationto (1) producea trendsouth of 20øSthat is morenearly parallel to thestrike of thedeeper parts of theWadati- Benioff Zone (which would havebeen subducting near the surface then), and (2) produceshortening amounts that are in agreement 3o with the available constraints. These constraints, discussed in a previoussection, are chosen(with amountsadded for shortening not calculated in the referenced studies) as follows: 90 km at latitude31øS [Jordan and Allmendinger, 1986], 120 km near22øS [Allmendingeret al., 1983],250 km near 18øS[Lyon-Caen et al., 1985;Sheffels et al., 1986], and 100 km in the Peruviansegment 35 northof theAltiplano near 10øS[Mdgard, 1984]. In Figure7 the sequencedescribed above is reversedto showa 80 W 75 70 65 forwardevolution in time since12 Ma. The oceanicplate is held Fig.5. Map showing 22cross sections alongwhich the cross-sectional fixedrelative tothe WSW movement of the overriding South areameasurements plottedin Figure 6 are determined. Thedashed line is Americanplate. Thehypothesized trench axis 12 m.y.ago is located300 km from trench (measured along the section lines); the along- rigidly rotatedwith the SouthAmerican plate and thentranslated strikedistance plotted inFigure 6 is measured from south to north along and deformed eastward to itspresent shape. The two motions of this3 kin,line. and Thedrainage axis ofdivides the trench, are taken areasfrom with Figure average 1. elevations The mapgreateris extendedthan the forearc are apAficially separated inFigure 7 only to isolate the to 10øSto show locations ofpaleomagnetic dataplotted inFigure 6, motionof theforearc relative to thecraton and not to imply a althoughthe digital topography andaveraged elevations end at 12øS.The sequentialchronology. lettersidentifying the sites of paleomagneticstudies are referenced in The actualtrajectories of themotions of pointsin the forearc Figure6. relativeto thecraton are not known; one possibility is that the trajectoriesare more or lessparallel to the directionof relative motionsbetween the Nazca and SouthAmerican plates, as shown trenchaxis along each of the 22 sectionlines. This "strike"line in Figure 7. The motionscan then be decomposedinto two passesthrough the main uplifted area and is usedonly as a components,one perpendicular and one parallel to thelocal strike convenientlinear measureof along-strikedistance within the of the forearc. Thesemotions are assumedto be takenup in the mountainbelt. The 22 greatcircle projection lines define a rough deformingbelt east of the forearc by shorteningand vertical curvilinearcoordinate system fixed to the curvedshape of the thickeningfor the perpendicularcomponent and by horizontal orogen,with a secondset of approximatelyorthogonal lines rotationalshearing for the parallelcomponent. The modelof (parallelto strike) locatedat fixed distancesalong the section shorteningshown by theshaded area in Figure7a is calculatedas lines. the distance between the deformed and undeformed trench axes The calculatedcross-sectional areas are plotted for eachsection measuredperpendicular to the local strikeof the presenttrench in Figure6. Comparisonwith Figure5 showsthat the areais a axis. Thisis of courseonly a veryrough first approximation to the maximum for section 16 acrossthe wide part of the Bolivian heterogeneous,three-dimensional finite strain field within the Altiplanoand Eastern Cordillera. The maximumalso coincides upperplate. with the bend in the coastlinenear the Peru-Chileborder. This The along-strike"location" of a calculationof shorteningis r9sultindicates a maximum of shorteningwhere the forearc has pinnedto thelocations of the22 sectionsshown in Figure5 in the moved farthestback toward the craton, in agreementwith the following way. The model shorteningis measured at the hypothesisof oroclinalbending. intersectionsof the22 sectionlines with thepresent trench axis. Model of shortening.I examinethe orocline hypothesis with Thesepositions are thenprojected along the sectionlines to the thevery simplemodel illustrated in Figure7. The 12 m.y. time line located300 km eastof the trenchaxis, the "strike"line intervalshown in Figure7 is chosento includethe primary phases throughthe main mountain belt shownin Figure5, to determine of compressionaldeformation and uplift. The actualtiming is not the appropriatealong-strike distance corresponding to the well constrained.Thermal uplift and weakening of theupper plate calculatedshortening. The valuesof shorteningobtained in this 3224 ISACga: • Ah'•nS PLA'mAU AND BOUVIAN O•ocLm•

c PALEOMAGNETIC DETERMINATIONS OF o lO .o.i FORARC ROTATION

•- -10

0 Oi

I . 3O

25oo - 438 ,- ', SHORTENING MEASUREDAREA , , MODEL 2

2000 - 350 ß 14 ß ß 12 ,, :0 1500 262 18 SHORTENING •2 1000 MODEL 1 175

500 THERMAL EFFECT - 88

. I . I . I o o 10 20 DISTANCE ALONG STRIKE, DEGREES

Fig. 6. The solidcircles in the lower sectionshow cross-sectional area measured along the numberedsections in Figure5. The preferredshortening model 1 is showntogether with anothermodel 2. Model2 hasa maximumshortening of 425 Ion insteadof 250 Ion. The combinedshortening and thermal effect is shownonly for the preferredmodel, and only model 1 is illustratedin Figure 7. The shorteningvalues used in the modelsare given on the right-handvertical axis. The upperpanel plots paleomagneticallydetermined rotations of the forearcat the siteslabeled by lowercaseletters in Figure5. In bothpanels the horizontalaxis is distancemeasured along the dashed"strike" line in Figure5. The paleomagneticdata include a, Becket al. [1986]; b, Palmer et al. [1980a] andgoao et al. [1985]; c, Turner et al. [1984]; d, gono et al. [1985]; e, He/d et al. [1985]; f, Heki et al. [1983] and Palmeret al. [1980b];g, goa• et al. [1985]; h, He/d et al. [1985];and i, May aad Butler [1985]. The rotations corroding to thetwo shorteningmodels are shownby solid(model 1) anddashed lines (model 2) as indicated.

way areshown for model1 on theright-hand ordinate of Figure6. use a simple Airy isostaticmodel with local compensation. Cross-sectionalarea calculatedfrom shorteningmodel. Elevatedmaterial with density Pt is compensatedby a rootwith Shorteningproduces uplift by theisostatic effect of a thickeneddensity Dr anda mantlewith densityPro' The initialcrustal crust.Assuming preservation of two-dimensional cross-sectional thickness is H O, with an initial elevation hO. Aftershortening, the area,then the sum of thecross-sectional areaof elevatedterrain sumof theadditional elevation, h - hO, and the added root r are andthe depressed crustal root will equalthe product of theoriginal integratedover the width affected by theshortening to givethe crustalthickness times the amount of shortening.An estimateof addedcross-sectional area of crust,A c. If cross-sectionalareais the elevationresulting from thermalexpansion can be calculated conserved, fora givenamount oflithospheric thinning and added tothe Ac = SH 0 effectsof shortening.Magmatic additions are not includedin this model.I alsoneglect the loss of materialby erosion. In fact,for whereS is theamount of shortening.The measured cross- muchof the central and southem parts of the plateau the erosional sectional area of materialelevated above the base level h o is A r. lossof material isprobably quite minimal owing to the large areas With simple Airy isostasy wehave of internaldrainage and the aridityof the climate. The effectsof extrusive additions and erosional losses can be considered as perturbationssuperimposed on the effects of shorteningand A•=Ac +a heating. where The shorteningmodel depictedin Figure 7 can be used to calculate the cross-sectional area of elevated material if the initial h-h0 Pm-Pt crustal thickness and crusud and mantle densities are assumed. I r p• Isncgs: CRh'•RALANt•AN PLAT•nU AN• BOLWI• O•OCLm!• 3225

The cross-sectionalareas determined from the shortening

10 modelcan thenbe comparedto the measuredareas as in Figure6 with assumedvalues for ct andH O. Wollard[1966] estimated an ct of 1/7.5 basedon empiricalrelationships between elevation and crustal thickness. For the crude model under consideration, we will use this value and assume an initial crustal thickness of 40 km. The valuesof shortening,S, are takenfrom the modelshown in Figure8A (thevalues are indicated on theright hand ordinate of Figure7).

20 The resulting calculatedcurve mimics the main latitudinal s variationof measuredcross-sectional area, except that near 27øS the measuredarea ramps up to largervalues. This occursat the southernend of the Puna plateau. These higher values can be accountedfor by a thermaleffect, as discussedbelow. Another SHORTENING modelwith the shorteningvalues increased in the Bolivianregion relativeto the flanksis alsoshown in Figure6 (labeledmodel 2). The maximumshortening for thismodel is about425 km. This is near the upperlimit of the estimatesof shorteningdiscussed by 3O Lyon-Caen et al. [1985]. This curve is everywhereabove the measuredshortening values, even without any thermal effects i 12 MY BP added. Model 1 seemsto give a better fit, especiallyif thermal effects are added.

70 W 60 Thermaleffect. I hyIx)t•size thatthe excess elevation north of latitude27øS for model 1 is mainly an effectof thermalexpansion or lithosphericthinning. The thermaluplift for a 70-km thinning of the lithosphereis shownin Figure6. This wouldcorrespond, 10 for example,to a changeof lithosphericthickness from 140 km, an arguablevalue for a stablecontinental lithosphere [Sclater et al., 1981], to 70 km. A simple linear approximationto a lithospheric conduction gradient is assumed [Turcotte and McAdoo, 1979] with a coefficientof thermalexpansion equal to 3 x 10-s cgs units. An effective width of 400 km is then used to obtain the cross- sectionalarea that is producedby 'thumingof the lithosphere in the 2O regionof the Altiplano-Puna.The effectivewidth is estimatedon S thebasis of thehypsometric curve of Figure2. This curvecan be viewedas a kind of averaged,one-dimensional cross section. The abscissa is area of land above the altitude shown on the ordinate. If the plateauis simplifiedinto a stackof rectangularprisms, the area shownis the productof the length of the plateautimes a width within which elevationseverwhere exceed the particular elevation.The lengthof the Puna-Altiplanoregion between about 3O 13øSand 27øS,measured along strike, is about2000 kin. With thislength fixed, the abscissaaxis can then be labeledin termsof 12 MY BP the widthvariable as shownalong the upper part of Figure2. The hypsometriccurve shows that the averaged"side" of the plateau hasa nearlylinear slope between elevations of theplateau "edge" 80 70 W 60 of 3.65 km and an elevation of about 1.3 km and an upward concaveprofile at lower elevations.I assumethat the effective Fig. 7. Hypothesizedevolution of (a) subductedplate and Co) upper plate since12 Ma. The Nazcaplate is heldfixed with respectto thepresent map widthis determinedby a pointhalfway down along the linear part coordinatesystem while the South American plate and forearc are r•ated. of theslope. The width of theplateau from this analysis is about Figure7a showsdepths to thesubducted plate in km for thepresent and for 400 kin. 12 Ma. For the sametimes, Figure 7b showsthe hytx•esized weakened The agreementof the estimatedthermal effect and the zones(darkest shading) in the upperplate and the bendingof the forearc. In Figure7a the thicklines show the trench axis at 12 Ma, a rigid rotation magnitudeof the discrepancybetween observed area and area of that axisfor the 12 rn.y. interval,and the presentdeformed axis. The calculatedfrom the shorteningis quitegood. Note thatthe along- shadedarea between the last two positionsof the trenchaxis represents the strike extentof the thermaleffect is simply fixed to producethe lossof surfacearea of the upperplate by the convergenceof the forelancl addedarea at 27øS. The interpretionof the 27øSjump in areais towards cratonic South America and is the basis for the preferred that north of this latitude the overridingand subductedplates are shorteningrnodel 1 shownin Figure6. In Figure 7b the trenchaxis coincideswith the westernboundary of the foreland at 12 rn.y. and at decoupledwith enough intervening asthenosphereto permit present,while the thick line is the 12-rn.y.rigid rotation of thetrench axis thermal isostasy to operate, while to the south the nearly (bef• bendingof the forearc) as also shown in Figure 7a. Thetrajectories horizontal Nazca plate is coupledor "stuck"to theoverriding plate offorearc movement toward theSouth American craum are abitrarily taken and effectively holds it down.This transition inplate coupling is parallelto the directionsof convergenceof the two large plates. The lightershading onthe left side (present time)of Figure 7bshows areas of superimposedona northwardly increasing amount ofshortening theupper plate weakened bypassing over the asthenospheric wedgeas the to producethe observed northward increase in cross-sectional plateoverrides the Nazca plate in theflat slabsegments. area. 3226 Isncr, s: CitYreAL AW'•A• Pl•mm{nuAlvi) BOLlVIAW O•oo.,m'•

The Quaternarymagmatic arc ends near 28øS, just southof the track studiessuggest removal of possiblymore than 10 km of southernend of the broadplateau, but thereis no evidenceof an overburdenduring the past 10-15 m.y. [Crough,1983; Benjamin abruptchange of dip of the subductedplate as found at the et al., 1987]. The lossof volumeon the easternslope south of northernend of theplateau near 15øS. However, as suggested in a 18øSmay therefore be onlythe material missing from the youthbY previoussection, the curvature of thesubducted plate just downdip valleysand depositedin the forelandbasin close to the eastern of the plateboundary changes from convexto concaveupwards marginof themountain front. The materialin the forelandbasins near27øS, and this change could be relatedto boththe southward aboveelevations of 0.3 km is includedin the cross-sectionalarea terminationof magmatismand the isostatic coupling between the measurements.Thus it is reasonableto arguethat significant plates. effectsof erosionalremoval of materialare seenonly in sections Effects of denudation. The erosionalsystems vary quite 18 and19. substantiallyaround the edges of theAltiplano-Puna plateau. The Why, then, do the deficienciesin cross-sectionalarea not westernside is extremelyadd with practically no streams reaching extendto sections20-22 aswell, where the erosional vigor of the thePacific between 18 ø and26 ø [Mortimer,1981]. In addition, Amazoniandrainage should perhaps be evengreater than in thevolcanism on the western edge of theplateau adds volume and sections18 and 197 Oneproblem is thatthe essentiallytwo actsagainst headward erosion by dammingthe plateau drainage. dimensional approach of the shorteningmodels becomes Asidefrom transfer of materialdown the giant alluvial drape increasinglyinaccurate in thecomplex re. curved northern end of alongthe western plateau slope, little volume has probably been theplateau. The sections taken there are somewhat oblique to the removedfrom the western side of theplateau south of 18øSduring plateauand thus overestimate the cross-sectional area by an thelate Cenozoic phase of uplift. Headwarderosion of Pacific amountthat accounts for about half of thediscrepency between drainageinto the plateau has developed in theregion north of 18øS theobserved and calculated areas. asmanifested by thespectacular series of transversecanyons in Anotherpossible effect is the erodability of the material. The southern Peru. The still very youthful character of that batholithsof the easternAndes of Bolivia and southernPeru physiography,however, suggests that the mass lost is small,that becomeboth more extensive and older from south to north [Grant is,only that needed to fill thewidely spaced canyons. etal., 1979;Dalmayrac et al., 1980].In theBeni River basin the The easternslopes are characterized by tworegimes also Triassicand late Oligocene-early Miocene plutonic bodies of the separatedapproximately by the latitude 18øS, the Amazon River CordilleraReal form very high but areally small massifs within drainageto thenorth, and the semiarid Paran• River drainage to thesurrounding weakly metamorphosed sedimentary thesouth. Sections 18 and 19 cross into the Amazon system and rocks[e.g., McBride et al., 1983].The Beni River system has thus showthe largest deficiency in cross-sectionalareawith respect to cut deeplyinto the lessresistant rocks and has removed theshortening model (Figure 6). I proposethat this anomaly is, in considerablematerial from the plateau. Farthernorth, the fact,partly due to materialremoved by unusuallyhigh erosion northeastern margin of theplateau isformed by areally extensive ratesin thisregion. Sections 18 and 19 cross into the drainage late Paleozoic and Mesozoic batholiths and associated strongly basinof theBeni River, a tributaryof theAmazon River. Besides metamorphosed sedimentary rocks. These rocks may be have low the vigorouserosion under modem conditions, significant rates of erosioncompared tothose farther south in theareas east of glaciafionof theCordillera Real, the high massif forming the La Pazand thus, with their greater areal extent, may constitute a easternedge of theAltiplano in theregion crossed by sections 18 moreeffective bastion against the headward migration of the and 19, probablyaugmented the erosiveefficiency of this Amazoniandrainage. drainagebasin during the Pleistocene. The mountainous upper Therelatively small anomaly near section 13 is associated with reachesof theBeni basin forms a deeplyeroded segment of the theAtacama basin, the steplike feature of thewestern side of the easternslope of theplateau, as is obvious from inspection of Plate plateaudiscussed in a previoussection. This anomaly does not 1. Thephysiography clearly suggests a large erosive cut into the appearto be explainable asan effect of denudation. side of the plateau. Filling this cut to the presentlevel of the plateau18 and 19.would account forabout half the discrepancy forsections Paleomagnetic Data In contrast,the drainagesystems farther south are far less Severalinvestigators, notably Kono et al. [1985], arguethat vigorous. In fact, the easterndrainage south of about 19øS paleomagneticdata for the Chileanand Peruvianforearc support actuallycarries little materialto the Paran/tRiver, since the rivers the hypothesisof oroclinalbending. Beck et al. [ 1986] pointout flowingout of the easternAndes dry up within 100 km from the possibly severe problems with the dating of remnant mountainfront. This is associatedwith the increasingaridity magnetizationsand with thedeterminations of paleopolepositions southof 18øSand the decreasing amount of Pleistoceneglaciation for cratonicSouth America (see also Valencio et al. [ 1983]) but evenat highelevations. The Andesbetween about latitudes 18øS neverthelessalso show that the published data do seemto indicate and 33øSa• to be in a climaticbelt [e.g., Clapperton,1983] a counterclockwiserotation of the coastal areas of Peru and that is much less affected by the powerfully erosive alpine northernmostChile whichcontrasts with a clockwiserotation of glaciationsof the Quaternarythat so profoundlyaffected the the forearcfarther south. This resultis seenin the summaryof landscapeof othersegments of themountain system. selectedforearc paleomagnetic data plotted in Figure6. The drainagesystem of theeastern slope of theplateau south of Also shownin Figure6 are simplegeometric calculations of 18øSis incisingthe folded and faultedstructures of the Bolivian therotations that occur in themodels of deformationof the forearc EasternCordillera and sub-Andeanthrust belts, but remnantsof relativeto the craton. Again,the change in shapeof the forearcis the high plateausurface are apparentin the topographyand on modeledby the changein shapeof the trenchaxis. Both the Lariat imagery. For example,the Potosfintrusion, now exposed modelcalculations and the selected paleomagnefic data are plotted at elevationsof 3 km (near 19.6øS,65.7øW) is estimatedto have with respect to along-strikepositions determined by the 22 beenemplac• closeto the late Mioceneland surfacewith little sections(Figure 5) as discussedin theprevious section. The plot materialsubsequently lost from abovethe intrusion[Francis et al., is extendednorthward to cover centralPeruvian paleomagnetic 1983]. This contrastswith the Zongointrusion located farther determinations.The rotationscalculated from the shortening northin the upperreaches of the Beni River basin,where fission model are done in a very crudefashion: the net rotationsnear ISACKS:Cl{• ANDI{AN PLATF_.AUAIqD Bo•rv•xs O•oo•u• 3227 sections4-12 and20-21 are first determined,and the intervening Peruvianand Pampeanflanks. This geometryis schematically curve is then roughedin with the assumedconstraint that the indicatedin Figure7a. A somewhatsimilar geometry can now be changefrom clockwiseto counterclockwiserotation occurs near observedin the Izu-Bonin/NorthHonshu/Kurile system of the sections16-17 where the shorteningreaches a maximum. More northwestPacific. The Pacificplate dips gently (30 ø) beneaththe numerousand better constrained palcomagnetic data would justify North Honshusegment but steepensconsiderably in both the a moresophisticated model of forearcbending and its associated flankingsegments, the Izu-Bonin and Kurile arcs. The overall rotations. At this stage,however, the fit of the data certainly form is that of a plunging anticline, a shape that can be encouragesconsideration of the oroclinal hypothesis. The topologicallycompatible with theconcave shape of thetrench axis magnitudesof rotationfor model 1 tendto be too small,while in termsof deformationsof an inextensiblelithospheric shell model2 providesa somewhatbetter fit. However,all of the [Yamaokaand Fukao, 1986, 1987]. palcomagneticdeterminations shown are for Mesozoic In the central,shallow-dipping section the asthenospheric magnetizationsand thus integrate the deformation over the entire wedgewas widest and consequently thinned and weakened the Cenozoic.Hence model 2 may be a beuermodel for the entire broadestswath of theupper plate, while a narrowermagmatic and Cenzoic,while model 1 is preferredfor theNeogene. thermalsystem operated along the flankingregions where the In contrast,a completestraightening out of the Peru-Chile subductedplate had steeper dips (see Figure 7b). At sometime forearc, an effect discussedby Kono et al. [1985], requires afterthe weakeningbegan, the variable-width swath began to fail rotationsthat are muchlarger than the well-determinedrotations under the compressivestress system transmittedacross the shownin Figure6. This impliesthat the embaymentof western convergentboundary. A greateramount of shorteningocctured in SouthAmerica was an inheritedfeature that was enhanced but not the central segmentbecause the weakenedarea there was the causedby Cenozoicshortening. I make use of thisimplication in widest. The assumptionis that for a relatively constant discussingthe late Cenozoicevolution of subductedplate lithosphericstress, shortening will proceeduntil a limiting geometryin thenext section. elevationis reached[e.g., Dalmayrac and Molnar, 1981; $udrez et The blocklikeclockwise rotation of Peru suggestedby the al., 1983;Froidevaux and Isacks, 1984]; thusthe mount of palcomagneticdata as far northas 5øS[Heki et al., 1983;May shorteningwill dependupon the width of theweakened .swath. and Butler, 1985] implie.sa steadilydecreasing amount of The flanking segmentsof forearcconsequently tended to shorteningtoward the northwestwithin the Peruvianorogen. In overridethe oceanicplate relative to the centralsegment and agreementwith this,results of a preliminaryexamination of the therebyhelped flauen the subductedslabs. However, the amount topographicmaps of Perunorth of 12øSshow a clearlydecreasing of movementof the flanking segmentsrelative to the central cross-sectionalarea toward the north. High altitudesstill occuras segmentis not enoughto explainthe flattening.An appealmust sharppeaks in the glaciatedcordilleras, but the total cross- be made to other unconstrainedfactors. Possibilitiesinclude sectionalarea decreases. effects of subductedtopography [e.g., Kelleher and McCann; An apparentproblem with the oroclinalhypothesis is thefact 1976;Pilger, 1981; Nut andBen-Avraham, 1981], subdueted plate that the counterclockwiserotations characteristic of the Peruvian age [e.g., Molnar and Atwater, 1978; Wortel, 1984], relative and "block"are alsofound in theregion just southof the Chile-Peru absoluteplate velocities [e.g., Luyendyk, 1970; Cross and Pilger, border(near 18øS) and therefore south of thepresumed kink or 1982]and mantle flow [e.g.,Hager and O'Connell, 1978; Tovish hingein themodern coastline where the rotations would reverse et al., 1978]. sign. However,the sharpness of the"kink" in thecoastline near The transitionnear 17øSin subductedslab dip is associated the Chile-Peruborder is not a featureof the regional-scalewith the inflectionof mapview curvatureof the trenchaxis and morphology. The topographicimage showsclearly that the westernslope. The sharpnessof thatinflection may have• coastlineis not an accuratedelineation of the more smoothly enhancedby lateCenozoic crustal shortening, as suggested by the curvedtrend of thetrench-mountain system but reflects a second- modelof Figure7. Thusthe localization of thetransverse flexure orderstructure on the topographic"bench" between the inner in thesubdueted plate may be partly a resultof thelate Cenozoic trenchslope and the main western slope of the Andes.Thus in changein theoutline of theleading edge of theoverriding plate. this regionthe shapeof the coastlineis an inaccurateindicator of the shapeof the forearc. The modelshown in Figure7a for QuaternaryDeformation of thePlateau forearcbending has maximum curvature in theregion of the Peru- Evidence that extensional structures developed on the Chile border, but the exact location of the change from Altiplano-Punaplateau during the Quaternarycan be relatedto the counterclockwiserotation to clockwiserotation depends upon the topographic effects of uplift as argued, for example, by detailedtrajectories of materialin thedeforming forearc. The sites Froidevauxand Isacks[1984]; (seealso Dalmayrac and Molnar in northernmostChile, in fact, appear to be slightly on the [1981], but the extensionalstructures have so far been reported northernside of themaximum of inferredshortening, in agreement only for the northernand southern parts of theplateau. According with the oroclinalhypothesis. to the model of deformationshown in Figure 7, and assuming forearctrajectories relative to the forelandthat are approximately Relationshipto a ChangingNazca Plate Geometry parallel to the directionsof motionbetween the forelandand the Why shouldshortening vary along-strike?The following Nazcaplate, shortening in theregion of thenorthram Altiplano is scenariois proposedwith very tentative timing. During the period obliqueto thedeformational zone with a significantcomponent of 25-12 Ma the map view outlineof the leadingedge of the left-lateralshearing along the strike of the orogen. This is overridingplate had a less pronounced(but still distinctly consistentwith the predominantlynorth-south direction of seawardlydirected) concave curvature than now existsalong extensiondescribed by $dbrier et al. [1985] in the northern westernSouth America. Duringthat period the subductedplate Altiplano. It is intriguingthat the topographicimage of Plate 1 dippedat an angle of 20ø-30ø beneaththe centralpan of the showsrather marked linear features along the northwestern side of embaymentbut steepenedto dips of 40ø-60ø in the adjacent the Altiplano with northweststrike. Large yet unrecognized 3228 Isncxs: • A.•l•AS Pt•X•AU A• BOL•V!AS Oaoo•m• strike-slipfaults may existin thisregion. The shorteningin the The modeldepicts an evolutionary,three-dimensional process southernpart of theplateau is alsooblique but with a right-lateral ratherthan steady state, two-dimensional one. Changesin map senseof shearparallel to strike. This is alsoconsistent with the view shapeof the upperplate affectthe dip of the slaband vice orientationsof theyoung extensional and strike-slip deformations versa. The three-dimensionalaspects of the modelpresented in along the southeasternmargin of the plateaureported by thispaper imply a very intricate,tightly coupled, and two-way Allmendinger[1986]. relationshipbetween the configuration of thesubdueted plate and The northernand southern "ends" of theplateau are, however, the plan view outlineof the forearc. The followingsequence is likely to be difficultto model,and field studiesthere are likely to proposedfor thelate Cenozoic evolution of thecentral Andes. An reveal complexpatterns of deformation[e.g., A llmendinger, initial seawardconcavity in themap view shapeof theoverriding 1986]. In thosetwo regionsthe three-dimensionalaspects of the plate led to an along-strikevariation in the dip of the subdueted deformationare important,preexisting structures control the plate, which, by affecting convective processesin the detailsof thedeformations [e.g., Marocco, 1978;Allmendinger et asthenosphericwedge, produced an along-strikevariation in width al., 1983],and along-strike changes in thermaluplift areprobably of a thermally weakened zone in the upper plate. The associatedwith the flaueningof the subductedplates. Thus, for compressionalfailure of the weakenedzone then occurred with an example,in the regionof the southeastmargin of the Argentine along-strike variation in the amount of shorteningand a Puna, a northwardly increasing amount of shortening is consequentchange in the shape of the leading edge of the accommodatedby a northwardlychanging style of deformation(in overridingplate. The changein shapeenhanced, but did not transitionbetween the Laramide-stylethick-skinned deformations produce, the bendingof the Bolivian orocline. The changein of the Pampeansegment and the Bolivianfold-thrust belt), and shapealso involved the Peruvian and Pampean segments moving thisis all overprintedby a thermaluplift thatmay now controlthe seawardrelative to the central segment,an effect that only locationof thesoutheastern plateau margin. partiallyaccounts for the flattenLngof the subductedplates in the two segments.Effects of subduetedplate buoyancy and manfie CONCLUSIONS flow probablyalso are important. The modeldepicted in Figure7 impliesthat the changes in dip In thispaper I have arguedthat the late Cenozoicuplift of the fromrelatively steep to nearlyflat occurin two regionsnear 17øS Andesis a resultof thermalthinning of the lithosphereand crustal and 23ø-27øS. Theseregions include the anomalousstructures thickeningproduced by crustalshortening. Although the model andphysiographies of the "AbancayDeflection" [Marocco, 1978] dependsupon assumptionsabout crustal densitiesand initial andthe Atacama Basin--Southern Puna, respectively. The location conditions,crustal shortening of the orderof 100 km is now well of theseregions above along-strike "hinges" in thesubducted plate establishedand must certainlybe an importantcomponent in any geometrymay help to understandthe tectonicevolution of these modelof the centralAndes. It wouldalso be hardto argueagainst complexregions. a thermal component in the uplift of a plateau that is so As perhapsthe Iz4me exampleof noncollisional,subduction- extensivelycovered by volcanicsas the Altiplano-Puna. The relatedmountain building, the late CenozoicAndes offer a modem model can account for the topography without inclusion of exampleof processesthat may have played a rolein thehistory of magmaticaddition to crustalvolume. more complexmountain belts, such as westernNorth American At the other extreme, a primarily magmaticmodel for the cordillerasor the . The idea that the essentialelement uplift wouldrequire an unusuallylarge rate of extractionof crustal of cordilleranmountain building in the Andesis compressional density from the mantle wedge and would imply failureof a weakenedswath of the overridingcontinental plate, physiographiccharacteristics that are not observed.The volcanic weakenedby thermaland magmaticprocesses operating in the material forms a thin cover on a relatively uniformly elevated asthenosphericwedge between the plates, is reminsicentof "tabletop."Although veryyouthful normal faulting isdeveloped discussions ofthe Late Cre•us-early Tertiary Laramide-Sevier in thenorthern and southern parts of theplateau, the net deformations inthe bySales [1968], Coney deformationissmall compared tothe much larger late Cenozoic [1972], Burchfiel andDavis [1975], andArmstrong [1982]. compressionaldeformations of the crust. Thereis no evidencefor A key featurein explainingthe unusualwidth of the central majorcrustal scale that might result if large amounts of Andes deformational beltis the low angle ofsubduction inthe materialwere injected intothe crust asdikes. The physiographic centralmore steeply dipping segment. I hypothesize a20 ø dip characteristicsof the plateau thus indicate that substantial duringthe Miocene,while the presentdip is about30 ø . Two magmaticcontributions would have to be in the form of sill-like factorsmay also favor low-anglesubduction in additionto the intrusionsor horizontallydistributed underplatings of material concavityin theshape of theoverriding plate. As theEast Pacific ratherthan large vertical dikelikeintrusions. The uniformplateau Riseapproaches the subducfion zone, (1) theage of thesubdueted elevation would indicate a rather even distribution of material platedecreases and its buoyancymay thereforeincrease, and (2) throughoutthe large area of the plateau. There is no evidenceof the mantle flow beneaththe subductedplate may have an topographicbulges that might be expectedabove large intrusive or . increasinglylarge eastwardcomponent. These last two factors underplatedvolumes concentrated beneath the main volcanicareas would also help explain the postulatedlow-angle subduction alongthe westernside of the plateau. Instead,thephysiography andstructure ofthe Altiplano-Puna beneathlate Mesozoic-early Tertiarywestern NorthAmerica. reveal a plateaulikeuplift boundedon one sideby a major thrust beltand on the other by a monoclinalflexure. The topography is a Acknowledgments.I thank R. Allmendinger,M. Barazangi,E. directexpression of thetectonics over much of theplateau because Fielding,T. Jordan,M. Mpodozis,and V. A. Ramo•,for veryhelpful of thelow rates of erosion in anarid climate. Thus, in contrast, for criticalreviews of thispaper. Manydiscussions with Jordan, Allmendinger,S. M. Kay, R. Kay, A. L Bloom, andother members of the example,tothe Central Range ofTaiwan [Suppe, 1981], elevation Cornell Andes Project have contributed substantially tothe viewpoints of thecentral Andes may be controlled ultimately by lithosphericpresented inthis paper. This research was supported by NASA Grant stressesrather than by denudation. NAS5-28767and NSF Grant EAR-85-18402. ISAC•: • Am•a_• Pt•X•AU • Bot,lVla_•O!to•m'l• 3229

REFERENCES andconstraints on the stateof stress,Earth Planet. Sci. l.•tt., 55, 473- 481, 1981. Allmendinger,R. W., Tectonicdevelopment, southeastern border of the Dalmayrac,B., G. Laubacher,and R. Marocco,Caract•res g6n6raux de PunaPlateau, northwestern Argentine Andes, Bull. Geol. Soc.Am., 97, l'6volutiong6ologique des AndesP6mviennes, G6ologie des Andes 1070-1082, 1986. P6ruviennes,Tray. Doc. ORSTOM, 122, Paris,501 pp., 1980. Allmendinger,R. W., V. A. Ramos,T. E. Jordan,M. Palma,and B. L. Dewey, J. F., Episodicity, sequenceand style at convergentplate Isacks, Paleogeographyand Andean structuralgeometry, northwest boundaries,The ContinentalCrust and It• Mineral Deposits,edited by Argentina,, 2, 1-16, 1983. D. W. Strangway,Special Paper Geol. Assoc.Canada 20, 553-573, Allmendinger,R. W., B. L. Isacks,T. E. Jordan,and R. Marrett,Uplift of 1980. the Altiplano-Punaplateau: The structuralperspective from the SE Dickinson, W. R., and W. S. Snyder, Plate tectonicsof the laramide margin of the Puna, NW Argentina(abstract), Ear Trans. AGU, 66, ,Laramide Folding Associated With BasementBlock Faulting 1088, 1985. in the WesternUnited States,edited by V. Matthews1TI, Mere. Geol. Armstrong,R. L., Cordilleranmetamorphic core complexes-from Arizona Soc.Am., 151,355-366, 1978. to sonthemCanada, Annu. Rev. Earth Planet. Sci., 10, 129-154, 1982. Evernden,J. F., S. J. Kriz, and C. CherroniM., Potassiumargon ages of Baker,M. C. W., The natureand distribution of upperCenozoic ignimbrite someBolivian rocks,Econ. Geol., 72, 1042-1061, 1977. centresin the centralAndes, J. Volcaaol. Geotherm.Res., 11,293-315, Faure,M., andF. Lalav6e,Bent structuraltrends of Japan: Flexural-slip 1981. folding relatedto the Neogeneopening of the Seaof Japan,Geology, Baker,M. C. W., and P. W. Francis,Upper Cenozoic volcanism in the 15, 49-52, 1987. centralAndes--Ages and volumes,Earth Planet. Sci. Left., 41, 175- Fielding, E. J., and T. E. Jordan, Active deformationat the boundary 187, 1978. betweenthe Precordilleraand Sierras Pampeanas,Argentina, and Barazangi,M., and B. L. Isacks, Lateral variationsof seismicwave comparisonwith ancient Rocky Mountain deformation, Rocky attenuationin the uppermantle above the inclinedearthquake zone of MountainForeland and the Cordilleran ThrustBelt, editedby W. J. the Tongaisland arc: Deepanomaly in the uppermande, J. Geophys. Perryand C. J. Schmidt.,Spec. Pap. Geol. Soc. Amer. in press,1988. Res., 76, 8493-8516, 1971. Francis,P. W., C. Halls, and M. C. W. Baker, Relationshipsbetween Barazangi,M., and B. L. Xsacks,Spatial distribution of earthquakesand mineralization and silicic volcanism in the central Andes, J. Volcanol. subductionof theNazca plate beneath South America, Geology, 4, 686- Geotherm.Res., 18, 165-190, 1983. 692, 1976. Froidevaux,C., andB. L. Isacks,The mechanical state of thelithosphere in Barazangi,M., W. Pennington,and B. Isacks,Global studyof seismic the Altiplano-Punasegment of the Andes,Earth Planet.Sci. L•tt., 71, waveattenuation in theupper manfie behind island arcs using pP waves, 305-314, 1984. J. Geophys.Res., 80, 1079-1092, 1975. Galli-Olivicr, C. , Pediplainin northernChile and the Andeanuplift, Beck, M. E., Jr., R. E. Drake, and R. F. Butler, Paleomagnetismof Science,158, 653--655, 1967. Cretaceousvolcanic rocks from centralChile and implicationsfor the Grange, F., D. Hatzfeld, P. Cunningham,P. Molnar, S. W. Roecker,G. tectonicsof the Andes,Geology, 14, 132-136, 1986. Suar6z, A. Rodrigues,and L. Ocala, Tectonic implicationsof the Benjamin,M. T., N. M Johnson,and C. W. Naeser,Recent rapid uplift in microear&quakeseismicity and fault pla• solutionsin southernPem, J. the Bolivian Andes: Geophys.Res., 89, 6139-6152, 1984. Evidencefrom fission-trackdating, Geology, 15, 680-683, 1987. Grant, J. N., C. HaHs, W. A. ,and N.J. Snelling,K-Ar agesof Beyis,M., and B. L. Isacks,Hypocent• trendsurface analysis: Probing igneousrocks and mineralization in partof the Boliviantin belt,Econ. the geometryof Benioffzones, J. Geophys.Res., 89, 6153-6170, 1984. Geol., 74, 838-885, 1979. Bird, P., Laramide crustal thickeningevent in the Rocky Mountain Green,A., and B. Wemicke,Possible large-magnitude Neogene extension forelandand Great Plains, Tectonics, 3, 741-758,1984. on thesouthern Pemvian Altiplano: Implications for thedynamics of Bowman,I., The physiographyof the centralAndes, I, The maritime mountainbuilding (abstract) Eos Trans. AGU, 44, 1241, 1986. Andes,Am.J. Sci., 28, 197-217,1909a. Guest,J. E., UpperTertiary ignimbrites of theAndean Cordillera of partof Bowman,I., The physiographyof thecentral Andes, H, The easternAndes, the AntofagastaProvince, northern Chile, Geol. Soc.Am. Bull., 80, Am. J. Sci.,28, 373-402, 1909b. 337-362, 1969. Bowman, I., The Andes of SouthernPeru, 336 pp., American Hager,B. H., andR. J. O'Connell,Subducfion zone dip anglesand flow GeographicalSociety, New York, 1916. drivenby platemotion, Tectonophysics, 50, 111-133,1978. Burchfiel,B.C., and G. A. Davis,Nature and controlsof Cordilleran Hasegawa,A., andI. S. Sacks,Subduction of theNazca plate beneath Pem orogenesis,western United States: Extension of an earliersynthesis, asdetermined from seismicobservations, J. Geophys. Res., 86, 4971- Am. J. Sci.,275-A, 363-396, 1975. 4980, 1981. Burchfiel,B.C., P. Molnar,and G. Suar6z,Possible thin-skin tectonics in Heki, K. Y., Y. Hamano,and M. Kono, Palcomagnetismof the central the easternAndes of Bolivia andPem (abstract),Ear Trans.AGU, 62, Andes:A preliminaryreport, in AndesSci. 2, 67-76, 1983. 1047, 1981. Heki, K. Y., Y. Hamarm,and M. Kono,Paleomagnetic study of Cretaceous Cande, S., Nazca•th America plate interactions,80 My to present AtajanaFormation and the Arica Dike Swarmnorthernmost Chile, J. (abstract),Ear Trans.AGU, 64, 865, 1986. Geomagn.Geoelectr., $7, 107-117,1985. Carey, S. W., The orecline concept in geotectonics,Proc. R. Soc. Hirahara,K., and T. Mikumo, Three-dimensionalseismic structure of Tasmania, 89, 255-288, 1958. subductedlithospheric plate under the Japanislands, Phys. Earth Chase, C. G., Plate kinematics: The ,east , and the rest of Planet. lnter., 21, 109-119, 1980. the wodd, Earth Planet. Sci. Len., 37, 355-368, 1978. Hollingsworth,S. E., andR. W. R. Rutland,Studies of Andeanuplift, part Chinn, D. S., and B. L. Isacks, Accurate sourcedepths and focal I., Post Cretaceousevolution of the San Bartolo area, northern Chile, mechanismsof shallowearthquakes in westernSouth America and in Geol. J., 6, 49-62, 1968. the New Hebridesisland arc, Tectonics,2, 529-563, 1983. Isacks,B. L., Topographyand tectonics of the centralAndes (abstract), Clapperton,C. M., The glaciationof the Andes,Quat. Sci.Rev., 2, 83- Eos Trans. AGU, 66, 375, 1985. 155, 1983. Isacks,B. L., S. M. Kay, E. J. Fielding,and T. Jordan,Andean volcanism: Coira, B., J. Davidson, C. Mpodozis, and V. Ramos, Tectonic and Icingon thecake (abstract), Ear Trans.AGU, 67, 1073, 1986. magmaticevolution of the Andesof northernArgentina and Chile, Jordan,T. E., and R. W. Allmendinger,The SierrasPampeanas of EarthSci. Rev., 18, 303-332, 1982. Argentina:A modemanalogue of Laramidedeformation, Am. J. Sci., Coney,P. J., Structuralevolution of theCordillera Huayhuash, Andes of 286,737-764, 1986. Pem,Geol. Soc. Am. Bull, 82, 1863-1884,1971. Jordan,T. E.,and R. N. Alonso,Cenozoic stratigraphy and basin tectonics Coney,P. J., Cordillerantectonics and North America Plate motion, Am. J. of the AndesMountains, 200-28 ø Southlatitude, Am. Assoc.Pet. Geol. Sci.,272, 603-628,1972. Bull.,71, 49-64, 1987. Cross,T. A., andR. H. Pilger,Jr., Controlsof subductiongeometry, Jordan,T. E., and M. GardewegP., Tectonicevolution of the late locationof magmaticarcs, and tectonics of arcand back-arc regions, Cenozoiccentral Andes (20ø-33øS), in Cenozoicand Mesozoic Geol.Soc. Am. Bull., 93, 545-562, 1982. Evolutionof the PacO•tcMargin, editedby Z. Ben-Avraham,Oxford Crough,S. T., Apatitefissiou-track dating of erosionin theeastern Andes, UniversityPress, New York, in press,1987. Bolivia,Earth Planet.Sci. Lett., 64, 396-397, 1983. Jordan,T. E., B. L. Isacks,R. W. Allmendinger,J. A. Brewer, V. A. Dalmayrac,B., andP.Molnar, Parallel thrust and normal faulting in Peru Ramos,and C. J. Ando, Andeantectonics related to geometryof 3230 Iascrs: • ANO•AN PLAT•nU • BOLIVIAN Oaoc•n•

subductedNazca plate, Geol. Soc. Am. Bull., 94, 341-361, 1983a. subductionof the Nazcaplate beneath South America: Comment, Jordan,T. E., B. L. Isacks,V. A. Ramos,and R. W. Allmendinger, Geology,5, 576--578,1977. Mountainbuilding in thecentral Andes, Episodes, 1983, 20-26, 1983b. Nur, A., and Z. Ben-Avraham,Volcanic gaps and the consmptionof Kadinski-Cade,K., Seismotectonicsof the Chilemargin and the 1977 aseismicridges in SouthAmerica, Nazca Plate: Crustal Formation and Cauceteearthquake of westernArgentina, Ph.D. thesis, 253 pp., Cornell AndeanConvergence, edited by L. D. Kulm,Mem. Geol. Soc. Am., 154, Univ., Ithaca,N. Y., 1985. 729-740, 1981. Kadg,D. E., andG. F. Moore,Tectonic complexities in theBonin island Palmer,H. C., A. Hayatsu,and W. D. MacDonald,Paleomagnetic and K- arcsystem, Tectonophysics, 27, 97-118, 1975. Ar age studiesof a 6 kin-thickCretaceous section from the Chilean Karig,D. E., R. N. Anderson,and L. D. Bibee,Characteristics of back arc Andes,G eophys. J. R. Astron.Soc., 62, 133-153,1980a. spreadingin theMadana Trough, J. Geophys.Res., 83, 1213-1226, Palmer,H. C., A. Hayatsu,and W. D. MacDonald,The Middle Jurassic 1978. CamaracaFormation, Atica, Chile: Paleomagnetism,K-Ar age dating Kay,S. M., andR. W. Kay,Role of crystalcumulates and the oceanic crust andtectonic implications, Geophys. J. R. Astron.Soc., 62, 155-172, in the formationof the lower crustof the Aleutianarc, Geology, 13, 1980b. 461-464, 1985. Pardo-Casas,F., and P. Molnar, Relativemotion of the Nazca (Farallon) Kay, S. M., V. Maksaev,R. Moscoso,C. Mpodozis,and C. Nasi,Probing andSouth American plates since I. ate Cretaceoustime, Tectonics,6, theevolving Andean lithosphere: Mid-late Tertiary magmatism in Chile 233-248,1987. (290-30.5ø S) over the modem zone of subhorizontalsubduction, J. Pflger, R. H., Jr., Plate reconstructions,aseismic ridges and low-angle Geophys.Res., 92, 6173-6190,1987. subductionbeneath the Andes,Geol. Soc. Am. Bull., 92, 448-456, Kelleher,J., and W. McCann,Buoyant zones, great earthquakes, and 1981. unstableboundaries of subduction,J. Geophys.Res., 81, 4885-4896, Pflger,R. H., Cenozoicplate kinematics, subduction and magmatism: 1976. SouthAmerican Andes, J. Geol. Soc.London, 141,793-802, 1984. Kono,M., K. Heki,and Y. Hamano,Paleomagnetic study of thecentral Prince,R. A., W. Schweller,W. Colbourn,G. Shepherd,G. Ness,and A. Andes:Counterclockwise rotation of thePeruvian block, J. Geodyn.,2, Masias,Part I, Bathymetryof the Peru-Chilecontinental margin and 193-209,1985. trench,Geol. Soc. Am. Map ChartSet., MC-34, 1980. Kussmaul,S., P. K. H6rmann,E. Ploskonka,and T. Subieta,Volcanism Ramos, V. A., E1 Mesozoicode la Alta Cordillera de Mendoza: andstructure of southwesternBolivia, J. Volcanol.Geotherm. Res., 2, Reconstrucciontectonica de susfacies-Argentina, Actas, IV Chile 73-111, 1977. CongresoGeologico Chileno, vol. 2, 104-118, Universidaddel Norte Lahsen,A., UpperCenozoic volcanism and tectonism in theAndes of Chile,Antofagasta, Chile, 1985. northernChile, Earth Sci.Rev., 18, 285-302, 1982. Rutland,R. W. R., J. E. Guest,and R. L Grasty,Isotopic ages and Audem I.aubacher, G., G6ologie de la cordillC•eOrientale et del'altiplano au nord uplift,Nature, 208, 677-678, 1965. et nord-ouestdu Lac Titicaca,Tray. Doc. ORSTOM,95,, 217 pp., Sales,J. K., Crustalmechanics of cordilleranforeland deformation: A 1978. regionaland scale-modelapproach, Am. Assoc.Petr. Geol. Bull., 52, Lavenu,A., Ndotectoniquedes s6diments Plio-Quaternaires du nord de 2016-2044,1968. l'AliplanoBolivien (Rdgion de -Ayo-Ayo-Umala),Cab. Schwab,K., Einbeitrag zurjungen Bmchtektonik der Argentinischen Puna ORSTOM,S•r. G•ol. 10(1), 115-126,1978. and ihr Verhaltniszu den angrenzendenAndenabschnitten, Geol. Luyendyk,B. P., Dipsof downgoinglithospheric plates beneath island Rundsch.59, 1064-1087,1970. arcs,Geol. Soc. Am. Bull., 81, 3411-3416, 1970. Schwab,K., Die stratigraphiein der Umgebungdes Salar de Cauchari Lyon-Caen,H., P. Molnar,and G. Su•rez,Gravity anomalies and flexure (N.W.Argentinien), Geotekton. Forsch. 43, 166pp., 1973. of the BrazilianShield beneath the BolivianAndes, Earth Planet.Sci. Schweller,W. J., L. D. Kulm, andR. A. Prince,Tectonics, stmcture and Lett.,75, 81-92, 1985. sedimentaryframework of the Peru-Chiletrench, Nazca Plate: Crustal Marocco,R., Un segmentE-W de la cha•nedes Andes Pdruviennes: I.a Formationand Andean Convergence, edited by L. D. Kulm,Mere. Geol. d6fiexiond'Abancay 6tude g6ologique de la Cordill,reodenude et des Soc.Am., 154, 323-349,1981. hautsplateaux entre Cuzco et SanMiguel sud du Pdrou(12ø30'S i• Sclater,J. G., B. Parsons,and C. Jaupart,Oceans and : 14ø00'S),Tray. Doc. ORSTOM, 94, 195pp., 1978. Similaritiesand differences inthe mechanisms of heat loss, J. Geophys. Martinez,C., Structureet 6volutionde la chatneHercynienne et de la Res.,86, 11,535-11,552,1981. Andinedans le nordde la Cordill,re desAndes de Bolivie,Tray. Doc. Sdbder,M., J. L. Mercier,F. Mdgard,G. Laubacher,and E. Carey- ORSTOM,119, Pads, 352 pp., 1980. Gailhardis,Quaternary normal and reversefaulting and the state of May, S. R., and R. F. Buffer, Paleomagnetismof the Puente Piedra stressin the centralAndes of southPeru, Tectonics, 4,739-780, 1985. formation, centralPeru, Earth Planet. Sci. Lett., 72,205-218, 1985. Segerstrom,K., Maturelandof northernChile and its relationshipto ore McBride, S. L., R. C. R. Robertson,A. H. Clark, andE. Farrar,Magmatic deposits,Geol. Soc. Am. Bull., 74, 513-518, 1963. and metallogeneticepisodes in the northerntin belt, CordilleraReal, Sheffels, B., B.C. Burchfiel, and P. Molnar, Deformationalstyle and Bolivia, Geol. Rundsch.,72,685-713, 1983. crusudshortening in theBolivian Andes (abstract), Eos Trans. AGU, 44, McCabe,R., Implicationsof paleomagneticdata on the collisionrelated 1241, 1986. bendingof islandarcs, Tectonics, 3, 409-428, 1984. Stewart,J. W., Neogeneperalkaline igneous activity in easternPeru, Geol. McLaughlin,D. H., Geologyand physiography of the PeruvianCordillera, Soc. Am. Bull., 82, 2307-2312, 1971. departmentsof Juninand Lima, Geol. Soc.Am. Bull., 335, 591-632, Strecker,M. R., R. Alonso,F. RiveHi, and R. Mon, Quaternarytectonic 1924. movementsin theArgentine Puna, 24 ø to 27ø S latitude,Geol. Soc. Am. M6gard,F., The Andeanorogenic period and its majorstructures in central Abstr.Programs, 17, 729, 1985. andnorthern Peru, J. Geol.Soc. London, 141,893-900, 1984. Sulrez, G., P. Molnar, and B.C. Burchfiel,Seismicity, fault plane Mingramm,A., A. Russo,A. Pozzo,and I• Cazau,Sierras Subandinas, in solutions,dep• of faulting,and active tectonics of theAndes of Peru, SegundoSimposio de Geologia Regional Argentina, vol. 1, pp. 95-138, Ecuador,and southern Colombia, J. Geophys.Res., 88, 10,403-10,429, AcademiaNacional de Ciencias, Cordoba, Argentina, 1979. 1983. Molnar,P., andT. Atwater,Interarc spreading and cordiHeran tectonics as Suppe,J., Mechanicsof mountainbuilding and metataVism in Taiwan, alternatesrelated to the ageof subductedoceanic lithosphere, Earth Mem.Geol. Soc. China, 4, 67-89, 1981. Planet.Sci. Lett., 41,330-340, 1978. ThomasN. A., Beitrag zur tektoniknord , Geol. Rundsch.,59, Mortimer,C., TheCenozoic history of thesouthern , Chile, 1013-1027,1970. J. Geol.Soc. •, 129, 505-526, 1973. Thorpe,R. S., P. W. Francis,and R. S. Harmon,Andean andesites and Mortimer,C., Cenozoicstudies in northermnostChile, Geol. Rundvch., 2, crustalgrowth, Philo,. Trane.R. Soc.London, Ser. A, 301,305-320, 395-420, 1975. 1981. Mortimer,C., Drainageevolution in the Atacamadesert of northernmostTosdal, R. M., A. H. Clark,and E. Farrat,Cenozoic polyphase landscape Chile,Rev. Geol. Chile, II, 3-28, 1981. andtectonic evolution of the CordilleraOccidenud, southe•ost Peru, Mufioz, J., Chile, Handbookof SouthAmedcan Geology, Mem. Geol. Soc. Geol. Soc.Am. Bull. , 95, 1318-1332,1984. Am.,65, 187-214, 1956. Tovish,A., G. Schubert,and B. P. Luyendyk,Mantle flow pressureand Newell, N. D., Geologyof the Lake Titicacaregion, Peru and Bolivia, theangle of subduction:Non-Newtonian comer flows, J. Geophys.Res., Mere.Geol. Soc. Am. 36, 111pp., 1949. 83, 5892-5898,1978. Noble,D.C., andE. H. McKee,Spatial distribution of earthquakesand Turcotte,D. L., andD.C. McAdoo,Geoid anomalies and the thickness of Is•,½xs: C•STRAL ASm•AS PL•,T•,U ASO BouvIAl• 0•o• 3231

thetithosphere, J. Geophys. Res., 84, 2381-2387,1979. Wollard,G. P., Regionalisostatic relationships in the U.S., The Earth Turner,I. C. M., Estratigraflade la Sierrade Santa Victoria y adyacencias, Beneaththe Continents, Geophys. Monogr. Set., vol 10,edited by I. S. Boi.Acad. Nac. Cleric. Argent., 41,163-196, 1960. Steinhartand T. J. Smith,gp. 557-594, AGU, Washington,D.C., 1966. Turner,I. C. M., Estratigraffade1 Nevado de Cachiy sectoral oeste Wond, M. I. R., Spatialand temporal variations in theAndean subduction (Salta),Acta Geol. Lilloana, 3, 191-226, 1961. zone,J. Geol. Soc.London, 141,783-791, 1984. Turner,I. C. M., Perfil transversalde la Puna,latitud 22 ø 15' S Wonel,M. I. R., and N. I. Vlaar,Age-d•dent subductionof oceanic approximada,Actas Segundas Jornadas Geol. Argent., 3, 355-378, tithospherebeneath South America, Phys. Earth Planet. Inter., 17, 201- 19(:g5. 2o8, 1978. Turner,J. C. M., Un perfdtransversal de la PunaAustral, latitud 26 ø S. Yamaoka,K., and Y. Fukao,Spherical shell tectonics: Effectsof RepdblicaArgentina, ,4zoc. Geol. Argent. Rev., 24, 361-366,1969. sphericityand inextensibilityon the geometryof the descending Turner, P., H. Clemmey,and S. Flint, Palcomagneticstudies of a lithosphere,J. Geophys. Res.,24, 27-53, 1986. Cretaceousmolasse sequence in the central Andes (Coloso formation, Yamaoka, K., andY. Fukao,Why do islandarcs form cusps at their northernChile), J. Geol.Soc. Lotion, 141,869-876,1984. junctions?,Geology, 15, 34-36, 1987. Utsu,T., Seismologicalevidence for anomalousstructure of islandarcs Yuen, D. A., L. Fleitout, G. Schubert,and C. Froidevaux,Shear withspecial reference to theJapanese Region, Rev. Geophys., 9, 839- deformationzones along major transform faults and subducting slabs, 890,1971. Geophys.J. R. Astron.Soc., 54, 93-119,1978. Uyeda,S., and H. Kanamori,Back arc openingand the modeof Zhao,W.-L., and W. J.Morgan, Uplift of theTibetan Plateau, Tectonics, 4, subduction,J. Geophys. R•.,84, 1049-1061,1979. 359-369, 1985. Valencio, D. A., J. F. Vilas, and I. G. Pacca,The significanceof the palcomagnetismof Jurassic-Cretaceous rocksfrom South America: B.L. Isacks,INSTOC Cornell Andes Project, Department ofGeological Preddftmovements, hairpins and magnetostratigraphy, Geophys. J.R. Sciences,Snee Hall, Cornell University, Ithaca, NY 14853. A.vtron.Soc., 73, 135-151, 1983. Vicente, J. C., F. Sequciros,M. A. Valdivia, and J. Zavala, E1 sobre- escurrimientode Chincha-Lluta:Elemento del accidentemayor andino (ReceivedJune 25, 1987; al NW de ,Bo/. Stsc.Geol. Peru, 61, 77- 100, 1979. revisedNovember 13, 1987; Wilson,J. T., andK. Burke,Two types of mountainbuilding, Nature, 239, acceptedDecember 5, 1987.) 448-449, 1972.