Late Cenozoic tectonism, collapse caldera and plateau formation in the central Andes

Late Cenozoic tectonism, collapse caldera and plateauformation in the central Andes

Ulrich Riller a;*, Ivan Petrinovic b, Juliane Ramelow c, Manfred Strecker d,Onno Oncken a

a GeoForschungsZentrum Potsdam, Telegrafenberg C223, 14473 Potsdam, Germanyb CONICET and Universidad Nacional de Salta, Buenos Aires 177, 4400 Salta, Argentina

c Freie Universita«t Berlin, MalteserstraMe 74-100, 12249 Berlin, Germanyd Institut fu«r Geowissenschaften, Universita«t Potsdam, 14415 Potsdam, Germany

Received 8 January 2001; accepted 2 April 2001

Abstract

The evolution of Andean volcanism including the formation of late Miocene to Recent collapse calderas on the Punaplateau is generally interpreted in terms of the kinematic framework of the Nazca and South American Plates. Wepresent evidence that caldera dynamics and associated ignimbrite volcanism are genetically linked to the activity of first-order NW^SE-striking zones of left-lateral transtension on the local and regional scales. Consequently, ages of collapsecalderas indicate activity of these fault zones which initiated at about 10 Ma on the Puna plateau. The onset of suchfaulting points to a change in the deformation regime from dominantly vertical thickening to orogen-parallel stretchingupon reaching maximum crustal thickness and critical surface elevation. Horizontal magma sheets that formed at mid-crustal level possibly due to heat advection by volume increase of asthenospheric mantle below thickened crust weretapped by sub-vertical faults. This accounts well for the observed tectono-magmatic phenomena at surface. It followsthat formation of collapse calderas and eruption of voluminous ignimbrites appear to be related to the mechanicalevolution of the Andean plateau rather than to changes in the geometry of the Wadati^Benioff zone or plate boundarykinematics. ß 2001 Elsevier Science B.V. All rights reserved.

Keywords: strike^slip faults; calderas; ignimbrite; volcanism; plateaus; Central Andes

1. Introduction

Much of the Altiplano^Puna plateau in the cen-tral Andes (Fig. 1) is covered by late Miocene toRecent felsic, crystal-rich ash £ow deposits [1^15].

The genetic relationship between this volcanismand plateau formation remains to be elucidated[4]. The onset of ignimbrite volcanism at about10 Ma in the northern Puna plateau has beenattributed to partial melting of tectonically thick-ened crust as a consequence of heating by en-hanced in£ux of asthenospheric material. Suchin£ux may have been accomplished either by anincrease in the rate of convergence between theNazca and South American Plates [4] or by thick-

0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 3 3 3 - 8

* Corresponding author. Tel. : +49-331-288-1313;Fax: +49-331-288-1370;E-mail: [email protected]

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

Earth and Planetary Science Letters 188 (2001) 299^311

www.elsevier.com/locate/epsl

ening of the asthenospheric wedge due to steepen-ing of the Nazca Plate or both [16,17]. Partialmelts apparently ascended diapirically to shallowcrustal reservoirs from which they were dis-charged as ash £ows, the eruptive centers ofwhich have been identi¢ed mostly as collapse cal-deras. Thermodynamic consideration suggests,however, that diapiric ascent of highly viscousfelsic magmas within the upper crust is implausi-ble [18]. Moreover, the low water content in un-degassed melt inclusions and the high crystallinityof the Purico and Atana ignimbrites in northChile (Fig. 1) indicate that volatile saturationalone may be in many cases insu¤cient to triggervolcanic eruption [19]. This calls for an alternativemechanism driving magma ascent within theupper crust, caldera formation and deposition offelsic ash £ows in the central Andes.

Regardless of the overall geodynamic setting ofother ignimbrite provinces on Earth, detailedstructural studies of collapse calderas of the prov-inces point to the signi¢cance of wrench faultingin triggering collapse [20,21]. In fact, most calde-ras in the southern central Andes are associatedwith NW^SE-striking ¢rst-order fault systemssuch as the Lipez, Olacapato^El Toro, Archibarcaand Culampaja fault zones [22] and de¢ne respec-tively four major transverse volcanic zones (Fig.1). The spatial coincidence of transverse volcaniczones with these fault zones points to a geneticrelationship between faulting and caldera forma-tion [19,23]. However, clear structural ¢eld evi-dence for a kinematic link between prominentfaults and volcano-tectonic structures is sparseas most of the calderas' substrates, which wouldpotentially host such structures, are obscured byerupted material. Furthermore, apart from thecurvature of structures at the Olacapato^ElToro fault zone indicating a component of left-lateral displacement on this fault zone [24,25],Neogene kinematics of the fault zones associatedwith the transverse volcanic zones are unknown.Consequently, it is uncertain whether NW-strik-ing faults in the central Andes merely served asmagma conduits or whether activity of thesefaults actually triggered incremental caldera col-lapse and magma ascent within the upper crust.

Horizontal crustal shortening is regarded as the

Fig. 1. Simpli¢ed structural map of the southern central An-des showing dominant morpho-tectonic units and distributionof late Cenozoic collapse calderas. NMC denotes the NegraMuerta Caldera. Numbers in boxes indicate calderas andtheir ages in Ma, respectively. 1: Cerro Pastos Grandes [1],2: Cerro Panizos [2], 3: Coruto [3], 4: Vilama [3], 5: Pairi-que [3], 6: Cerro Guacha [4], 7: Purico [4], 8: La Pacana[5,6], 9: Coranzuli [7], 10: Ramadas [8], 11: Aguas Calientes[9], 12: Negra Muerta [10], 13: Galan [11], 14: Wheelright[12], 15: Laguna Escondida [12], 16: Cerro Bayo [13], 17:Mulas Muertas [12], 18: San Franzisco [12], 19: FarallonNegro [14], 20: Incapillo [15]. Arrows on some collapse cal-deras indicate the direction of maximum upper crustal exten-sion based on the orientation of principal collapse-relatednormal faults [4,14].

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311300

prime mechanism that leads to plateau formationin the central Andes in Neogene time [26,27].Magnitudes of horizontal contraction inferredfrom cross-section balancing, however, are notuniform but decrease systematically away fromthe central bend region [28]. This informationcan be used to construct a ¢nite horizontal strainellipse for the southern central Andes from whichthe kinematics of variably oriented discontinuitiescan be deduced. If crustal shortening and ignim-brite volcanism are genetically related, the kine-matics of NW^SE-striking fault systems predictedby the ¢nite strain geometry should correspond tothat of volcano-tectonic structures of eruptivecenters located on the respective fault system.

The recently identi¢ed 12 km by 7 km NegraMuerta Caldera [29] is located on the Olacapato^El Toro fault zone, a major fault zone transectingthe southern Puna plateau (Fig. 1). The caldera ispart of the Negra Muerta Volcanic Complex [30]and straddles the eastern margin of the plateau(Fig. 1). Here, the southward-draining Calchaqu|River has its source and, along with its tributaries,incises the caldera parallel to its long axis byabout 1300 m. The exposure of basement rocks,volcano-tectonic structures and subvolcanic rocksin its center attests to partial erosion of the cal-dera. This o¡ers an excellent opportunity to studythe structural and magmatic history of the NegraMuerta Caldera. Moreover, the strategic positionof the caldera on the Olacapato^El Toro faultzone makes it an ideal target to elucidate the ge-netic relationship between faulting and origin oftransverse volcanic zones in the southern centralAndes (Fig. 1). We show that components of localhorizontal dilation associated with the activity ofthe Olacapato^El Toro fault zone kinematicallydrove asymmetric collapse. Fault-controlled cal-dera formation bears important geodynamic con-sequences on the late Cenozoic evolution of thePuna plateau.

2. Kinematics of NW^SE-striking fault zones inthe southern central Andes

Gephard [31] noted the high degree of bilateralsymmetry of both topography in the central An-

des and angle of the subducted Nazca Plate aswell as the colinearity of the poles to the resultingsymmetry planes with the Tertiary Euler pole ofthe Nazca and South American Plates. Based onthese facts, he concluded that horizontal contrac-tion was controlled by the dynamics of subduc-tion and achieved mostly parallel to the symmetryplane of the central Andes (Fig. 2). Magnitudes ofhorizontal shortening deduced from balancedcross-sections of the central portion of the orogenand its eastern foreland [28] decrease systemati-cally from about 200 km at 21³S to about 20km at 27³S (Fig. 2). Interestingly, south of 28³S,this gradient in transverse shortening decreases tozero and corresponds to the southern termini ofthe transverse volcanic zones and the central An-dean plateau (Fig. 1). This points indirectly to agenetic link between plateau formation by trans-verse shortening and origin of eruptive centersde¢ning the transverse volcanic zones.

Di¡erential transverse shortening has importantconsequences for the deformation regime, partic-ularly the orientation and kinematic evolution offaults associated with the transverse volcaniczones. The gradient in transverse shortening im-posed a horizontal shear strain (whereby i = 15³)parallel to the trace of the central Andean sym-metry plane (Fig. 2). The applied shear strain de-termines the orientations and relative magnitudesof the maximum and minimum strain axes of thehorizontal sectional strain ellipse, e1 and e2, re-spectively, and thus the orientation and diameterof the ellipse for the southern central Andes.Transformation of this ellipse to the unit circlerelates the present azimuth of the line elementcorresponding to the strike of fault zones associ-ated with ignimbrite activity (128³) to the direc-tion of this element prior to the onset of contrac-tion (lines aP and a, respectively, in Fig. 2). Thisallows one to infer the kinematic history of faultsassociated with the transverse volcanic zones to a¢rst degree. From such analysis it is evident thatthe fault zones have rotated less than 10³ clock-wise upon the onset of horizontal contraction.Clockwise rotation of the respective line elementimposed a small component (6 2³) of left-lateraltangential shear, the angle between line aP and thenormal to the tangent at its intersection with the

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311 301

strain ellipse circumference (A), on the element(Fig. 2). More importantly, however, the initialand ¢nal orientations of these line elements areat high angles to the maximum ¢nite extension(e1). Maximum extension normal to, and left-lat-eral strike shear on, the line elements points toleft-lateral transtension on NW^SE-striking faultzones. This deformation regime provides a logicalexplanation as to why collapse calderas are spa-tially associated with these fault zones and thusde¢ne the transverse volcanic zones in the south-ern central Andes.

The orientation of major normal faults associ-ated with the formation of eruptive centers can beused to constrain the importance of NE^SW di-rected maximum extension (e1) predicted by the¢nite strain ellipse. In fact, major normal faultsand structural lineaments of the La Pacana, CerroGuacha, Pastos Grandes and Coruto Calderas,forming part of the Altiplano^Puna VolcanicComplex, strike NW^SE [4]. Moreover, the align-ment of stratovolcanoes and other normal faultswith this direction [4] corroborates the e¡ect ofupper crustal dilation associated with the Olaca-pato^El Toro and Lipez fault zones (Fig. 2) onthe formation of eruptive centers in this area.

Similarly, NE^SW horizontal extension has alsobeen inferred from the geometry of normal faultsin the Farallon Negro Complex [14] and is con-sistent with dilation on the Culampaja fault pre-dicted by the horizontal sectional strain ellipse(Fig. 2). In spite of this evidence in favor offault-controlled volcanic activity, only examina-tion of the kinematic history of volcano-tectonicfaults of eruptive centers will decide on the sig-ni¢cance of left-lateral transtension, in particularits component of strike shear, on the evolution ofthe transverse volcanic zones. Field evidence forleft-lateral strike shear on the Olacapato^El Torofault zone implies that in addition to dilation suchshear should be also apparent from volcano-tec-tonic structures of the Negra Muerta Caldera.

3. General characteristics of the Negra MuertaCaldera

Near the Negra Muerta Caldera, Cretaceous toearly Tertiary sedimentary rocks rest unconform-ably on the Precambrian to Cambrian Puncovis-cana Formation (Fig. 3). Collectively, these rocksform the basement to the extrusive volcanic rocks,

Fig. 2. Bilateral symmetry of the central Andes [31]. The shaded area is higher than 3 km a.s.l. NW^SE-striking lines delineatethe traces of major fault zones associated with the transverse volcanic zones. The horizontal sectional strain ellipse is deducedfrom the gradient in transverse shortening [28]. For explanation of strain ellipse parameters see text.

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311302

exposed chie£y NW of the collapse caldera (Fig.4). Depending on the paleotopography of thebasement, the volcanic sequence varies in thick-ness but thins generally toward the NW. The se-quence formed mainly during two eruptive epi-sodes, depositing K-rich, calc^alkaline volcanicrocks (Fig. 3). Dacitic and andesitic ash £owsare characteristic for the ¢rst episode. Near thenorthern crater wall, these £ows are several tensof meters thick and made up of ¢amme-rich,welded ignimbrites. The second episode encom-passes two eruptive pulses producing rhyolitic toandesitic volcanic rocks, respectively. In particu-lar, a very distinctive rhyolitic ash £ow is coveredin places by a 10^20 m thick vesicle-rich, rhyoda-citic lava £ow. The rhyolitic ash £ow is charac-terized by vesicular, quartz and biotite-rich pum-ice and lithic fragments of Puncoviscanametapelite and subvolcanic rhyolite set in a whitematrix devoid of welding. We traced this ash £owto, and thus correlated it with the 7.4 Ma `Toba 1'ignimbrite, exposed about 20 km north of thecaldera [10], whose eruptive center had remainedto be found. During the second eruptive pulse, theToba 1 ignimbrite was covered by monotonousandesitic lava £ows which are up to 300 m thickand hydrothermally altered at the caldera margin.

Due to partial erosion of the caldera by £uvialand Pleistocene glacial activity, di¡erent morpho-logical features delineate its asymmetry. To thenorth and west of the depression, the crater mar-gin is up to 5300 m above sea level (a.s.l.) andmade up of extrusive volcanic rocks and dikes(Fig. 3). The pronounced curvature of the south-ern margin is well apparent in the TM image (Fig.4). It is about 600 m lower than the northernmargin but composed dominantly of basementrocks topped sporadically by patches of andesite.To the east, the caldera is con¢ned by a N^S-trending crest of basement rocks (s 6000 ma.s.l.) which is covered in places by andesite onits western £ank (Fig. 3). Intracaldera volcanicrocks include dikes, domes and vents of mostlyrhyolitic to andesitic composition. A prominentNNW-trending body of subvolcanic rhyolite inthe caldera center forms a morphological ridgefrom which rhyolite dikes emanate toward theSW. These dikes are cut by dacitic and andesitic

ones (Fig. 3) and corroborate the two-stage, calc^alkaline magmatic evolution of the second vol-canic episode inferred from the stratigraphic suc-cession of equivalent extrusive rocks.

4. Tectono-magmatic history of the Negra MuertaCaldera

Two fault sets, comprising NW-dipping normaland ENE-striking left-lateral oblique-normalfaults (Fig. 5), are evident in basement rocks ofthe caldera center. Displacement magnitudes onmost faults from both sets are on the order oftens of meters. However, a magnitude of at least150 m is indicated by the juxtaposition of theearly Tertiary Mealla Formation against rocksof the Cretaceous Pirgua Subgroup on a steeplysouth-dipping normal fault (Fig. 3). The horizon-tal strike separation of lithologic contacts on amajor oblique-slip fault at the eastern margin ofthe caldera amounts to 700 m. Normal faults, letalone of this magnitude, are unknown from thePuna plateau and must, therefore, be of local im-portance. A volcano-tectonic origin of the faults issuggested by (1) abundant striations on fault sur-faces and up to 1 m thick breccia zones pointingto catastrophic failure in the brittle realm, (2)strong local hydrothermal alteration of damagezones on either side of the surfaces, typical forpost-magmatic activity and restricted spatially tothe caldera center as well as (3) concordance ofdikes, elongate vents and domes with the faultspointing to a genetic relationship between defor-mation and magmatism.

Geometrically, normal and oblique-normalfaults form respectively second-order tensionaland Riedel fractures to left-lateral master disloca-tions of the Olacapato^El Toro fault zone evidentimmediately to the north and south of the NegraMuerta Caldera (Figs. 4 and 5). All of these faultswere largely active in Miocene time as they a¡ectthe Eocene/Oligocene Maiz Gordo Formation[24]. The kinematic analysis of the second-orderfaults using the orientation of fault planes andrespective striations indicates NNW^SSE stretch-ing. The angular departure between this stretchingdirection and the one predicted by the horizontal

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311 303

sectional strain ellipse (Fig. 2) can be explained bythe variation in the magnitude of left-lateral strikeshear on the Olacapato^El Toro fault zone.NNW^SSE stretching is compatible with the ori-entation of subvolcanic sheet intrusions in the cal-dera center (Fig. 3) and motion on the ¢rst-orderfaults (Fig. 5). Therefore, geometrical, temporaland kinematic relationships indicate that ¢rst-

and second-order faults are genetically linked.Surfaces of oblique-normal faults are generallydecorated by quartz and are concordant to severalmeters wide fault breccia zones containing frag-ments cemented as well by quartz. These struc-tures truncate chlorite-coated normal faults andsuggest that faulting occurred in two increments.Furthermore, rhyolite dikes are concordant to

Fig. 3. Geologic map of the Negra Muerta collapse caldera displaying major volcano-tectonic structures. See Fig. 1 for locationof the caldera.

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311304

normal faults (Fig. 3) and, unlike andesite dikes,are cut by strike^slip faults. As rhyolite dikes wereemplaced prior to andesite dikes (see above), nor-mal faults must have controlled the ascent anderuption of rhyolitic magma whereas strike^slipfaults became important later in the tectono-mag-matic history. The sum of the above evidence notonly attests to fault-controlled ascent of magmabut also to a two-stage, tectono-magmatic evolu-

tion of the Negra Muerta Caldera that was drivenby left-lateral transtension at the Olacapato^ElToro fault zone.

5. Implications for the evolution of the centralAndean plateau

Crustal thickening by horizontal shortening is

Fig. 4. Satellite TM image (RGB bands: 7, 4, 1) of the Negra Muerta Caldera and prominent faults of the Olacapato^El Torofault zone.

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311 305

regarded as the prime mechanism for plateau for-mation, i.e. surface uplift, in the central Andeswhereby the Altiplano^Puna plateau formed intwo stages [26,27]. On the Puna plateau, the ¢rststage appears to have started in Oligocene timeand resulted in elongate ranges made up of base-ment rocks thrust over deformed sedimentaryrocks of internally drained compressional basins[32^34]. By about 13^11 Ma, the plateau surfacemust have been su¤ciently elevated in order to bethe source area of coarse clastic sediments, suchas the Rio Guanaco and Angastaco Formations[8,35], in the eastern foreland of the plateau. Thesecond stage started at about 10 Ma [36] and ischaracterized by an eastward shift of folding andthrusting from the plateau into the eastern fore-land. It is during this stage that collapse calderasformed on the Puna plateau and therefore consti-tute important geodynamic markers regarding theevolution of the plateau.

Volcano-tectonic faults of the Negra MuertaCaldera cut folded strata of its basement, notablyat the southeastern portion of the depression (Fig.3). Furthermore, undeformed dacitic to andesiticpyroclastic rocks cover a paleotopography whichformed chie£y by large-scale folding of basementrocks around N^S-trending axes [24]. However,folding a¡ected the distinctive Toba 1 ignimbriteto some extent, e.g. 20 km north of the caldera

[37], and occurred contemporaneous with left-lat-eral slip on other NW-striking faults, such as theSaladillo fault (Fig. 4), of the Olacapato^El Torofault zone [24]. This shows that eruption of ignim-brites in this area occurred towards the end of the¢rst stage of plateau formation characterized byvertical thickening and points to a local change inthe deformation regime from dominantly verticalto more orogen-parallel upper crustal extension.The deformation regime associated with the trans-verse volcanic zones suggests that orogen-parallelstretching a¡ected the entire Puna plateau. Con-sequently, activity of fault zones associated withthe transverse volcanic zones should also be ap-parent by the ages of ignimbrites in the southerncentral Andes.

The Aguas Calientes Caldera, located about 20km northwest of the Negra Muerta Caldera (Fig.1), has also been attributed to activity of the Ola-capato^El Toro fault zone [23]. Along with thenearby Ramadas Caldera [37], the three calderasformed within a period of 2.5 Ma at the southernsegment of this fault zone (Fig. 1). By contrast,the La Pacana Caldera and the Purico Complexare located at the northwestern portion of theOlacapato^El Toro fault zone (Fig. 1) and areclearly younger than the Aguas Calientes, NegraMuerta and Ramadas Calderas. The petrologicalcharacteristics of associated ash £ows and asym-

Fig. 5. Lower-hemisphere equal-area projections showing the orientation of second-order faults in the Negra Muerta Caldera(great circles) and the sense of slip of hanging wall blocks on these faults (small arrows). Large arrows indicate the bulk stretch-ing direction. The geometric relationship between these faults and master discontinuities of the Olacapato^El Toro fault zone areshown schematically.

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311306

metry of the La Pacana Caldera strongly suggestan external i.e. tectonic cause driving the forma-tion of these volcanic centers [6]. Based on the ageof thrust faults kinematically linked with the Ola-capato^El Toro fault zone, segments of this faultzone were active during Neogene time [25]. Thesum of the above structural and temporal evi-dence indicates that caldera dynamics and activityof fault segments of prominent NW-striking faultzones in the central Andes are genetically linked.Consequently, ages of ash £ow deposits related tocollapse calderas are expected to portray pulses offault activity on the plateau. Following this con-cept, we compiled published radiometric ages ofash £ows from collapse calderas located close toprominent central Andean strike^slip faults on thePuna plateau. Based on these ages (Fig. 1) andstructural ¢eld evidence for the fault-controlledorigin of the Negra Muerta Caldera, left-lateraltranstension on NW-striking fault zones on thePuna plateau initiated at about 10 Ma but oc-curred episodically on individual segments ofthese fault zones.

The onset of longitudinal stretching of theupper crust in the southern central Andes hasimportant consequences for constraining the ageof surface uplift in this area. Allmendinger andGubbels [38] made a strong case for thickeningof crust underlying the Puna plateau by distrib-uted, horizontal shortening. An important charac-teristic of this mode of deformation is mechanicalcoupling between the upper and lower crust andthus vertical thickening over the same column ofrock (Fig. 6a). The change in the deformationregime in the southern central Andes from dom-inantly vertical thickening to horizontal, orogen-parallel stretching of upper crust (Fig. 6b) oc-curred while directions of horizontal shorteningand convergence between the Nazca and SouthAmerican Plates remained constant [39]. Suchchange in deformation regime appears to be typ-ical for orogenic belts reaching maximum crustalthickness and critical surface elevation wherebyhorizontal shortening is compensated chie£y byorogen-parallel stretching of isostatically balancedcrust [40]. If coupling between upper and lowercrust underlying the Puna plateau is strong andvertical thickening shifted to the eastern foreland,

horizontal shortening of the entire Puna crust wasaccomplished by longitudinal crustal £ow whilemaximum surface elevation was maintained sinceabout 10 Ma. Evidence for Pliocene to Quater-nary strike^slip faulting in the southern centralAndes [41] indicates that maximum surface eleva-tion has been conserved until today and that thekinematics of upper crustal deformation are de-termined by orogen-parallel stretching.

6. Origin of central Andean felsic ash £ows

The generation of felsic melts in the southerncentral Andes has been attributed to heating byin£ux of hot asthenospheric mantle below thick-ened crust [4]. Such in£ux may have beenachieved by accelerated convergence between theNazca and South American Plates but recent re-calculation of plate motion shows that the rate ofconvergence between these plates actually de-creased over the past 20 Ma [42,43]. Alternatively,steepening of the subducting Nazca Plate to itscurrent dip of about 30³ may have led to en-hanced input of asthenospheric material belowthickened crust underlying the Puna plateau[16,17]. According to Kay et al. [17], ignimbritevolcanism in the central Andes is directly relatedto lower crustal melting by steepening of the Naz-ca Plate. In their model, crustal melts generatedby this mechanism ascended along thrust or re-verse faults in the plateau area largely between 10Ma and 7 Ma whereby an apparent westwardprogression of eruptive centers since 26 Ma is in-terpreted as a direct consequence of slab steepen-ing. However, neither the exact mechanism ofmagma ascent on reverse faults is apparent fromtheir model during this period of crustal thicken-ing nor are eruptive centers located on orogen-parallel thrust faults. Furthermore, ignimbritevolcanism on the Puna plateau clearly lasted untilPleistocene time and a westward migration oferuptive centers is not apparent from the ages ofcollapse calderas (Fig. 1). Thus, the formation ofcollapse calderas and associated felsic volcanismcannot be regarded as immediate consequences ofmelt generation in the lower crust. Speci¢cally,melt generation in the lower crust alone does

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311 307

not provide the physical conditions for eithermagma transport through the crust or upper crus-tal magmatic processes in the central Andes. Thisis supported by the general conclusion that inparticular magma generation and to this end mag-ma segregation and magma transport not onlydi¡er in time and space for a given region butare also strongly dependent on the deformationregime active at the respective crustal level [44].Steepening of the Nazca Plate may well accountfor the generation of magmas in the central An-des. However, we do not regard it as the funda-mental mechanism driving magma ascent in the

upper crust, formation of collapse calderas andassociated ignimbrite eruption.

Volcano-tectonic structures of eruptive centerslocated on the zones of left-lateral transtension(Fig. 1) consist of moderately steep normal faultsand strike^slip faults (e.g. Fig. 5). Sub-verticaldiscontinuities require the lowest magmatic pres-sure to be penetrated by melts and are thus pref-erential pathways for magma ascent [45]. Thisprovides a plausible explanation for preferredmagma ascent within the upper crust at transversevolcanic zones. In our opinion, the variance inignimbrite ages indicates episodic activity of indi-

Fig. 6. Stages of plateau evolution in the southern central Andes. (a) Oligocene to late Miocene: distributed horizontal shorteningand vertical thickening (arrows) are accomplished on individual thrust and reverse faults in the upper crust and ductile deforma-tion in the lower crust underlying the Puna plateau during £at-slab subduction. (b) Late Miocene to Recent: onset of longitudi-nal extension of upper crust at about 10 Ma signi¢es the change in the deformation regime upon reaching maximum crustalthickness and surface elevation while direction of bulk horizontal contraction remains east^west (arrows). Longitudinal extensionis accomplished by upper crustal strike^slip faults tapping mid-crustal magma sheets and controlling the location of collapse cal-deras. Partial melts in the lower and middle crust may have been generated by in£ux of asthenospheric material below thickenedcrust as a consequence of slab steepening. Vertical crustal thickening is transferred to the eastern foreland of the plateau.

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311308

vidual strike^slip fault segments following thechange in the deformation regime to dominantlyorogen-parallel stretching of the central Andeanplateau. This scenario accounts well for the (1)onset of late Cenozoic ignimbrite volcanism onthe Puna plateau at about 10 Ma, (2) locationof eruptive centers on upper crustal transtensionalfault zones and (3) lack of a uniform westwardmigration of these centers.

Using the teleseismic receiver function method,Yuan et al. [46] imaged a distinctive low-velocityzone at 15^25 km depth below the Puna. Othergeophysical characteristics such as high Vp/Vs ra-tios at this depth suggest that this zone constituteseither a mid-crustal magma sheet or an intercon-nected network of melt channels at the grainscale. A mid-crustal magma sheet from which fel-sic magma rose to shallow crustal reservoirs iscompatible with the strong crustal geochemicalsignature of ash £ows [16] and other petrophysicalevidence [47,48]. Such a melt sheet may haveformed in response to brittle^ductile deformationat [49], or ponding of magma near, sub-horizontallithotectonic boundaries (Fig. 6). Felsic magmabodies in the central Andes most likely formedand resided in the lower and middle crust as aconsequence of advective heat transfer upon in-£ux of hot asthenospheric mantle below thickenedcrust. The change in the deformation regime fromvertical thickening to orogen-parallel horizontalextension resulted in vertical fault zones thattapped felsic melt bodies and facilitated ascentof felsic melt to upper crustal levels (Fig. 6b).This process accounts well for the observed mag-matic characteristics such as emplacement ofdikes concordant to tensile shear fractures in theNegra Muerta Caldera as well as ascent and erup-tion of highly viscous felsic magmas throughoutmuch of the Puna plateau.

7. Conclusions

The orientation and magnitude of the horizon-tal sectional strain ellipse inferred from the gra-dient in transverse shortening in the southern cen-tral Andes indicate left-lateral transtension onNW^SE-striking fault zones. This deformation re-

gime provides a logical explanation for the spatialassociation of collapse calderas with these faultzones. The genetic relationship between calderadynamics and regional left-lateral transtension iscorroborated by a detailed analysis of the tecto-no-magmatic history of the Negra Muerta Cal-dera. Formation of this partially eroded andasymmetric caldera occurred in two incrementsdriven by left-lateral strike shear and fault-normalextension on the prominent Olacapato^El Torofault zone. The intimate relationship between thedynamics of this and other collapse calderas andNW-striking ¢rst-order faults on the Puna plateauis evident due to (1) localized, fault-controlled as-cent of magma, (2) eruption of highly viscous,felsic ash £ows but also (3) clustering of collapsecalderas displaying similar ages near prominentfaults. Consequently, episodes of fault activity ap-pear to be indicated by the age of collapse calderaformation. The onset of activity of ¢rst-orderfault zones at about 10 Ma marks, therefore, thetemporal change from dominantly vertical thick-ening to orogen-parallel stretching of the plateauupon reaching a critical crustal thickness and sur-face elevation. Vertical faults that became activeupon the change in the deformation regimetapped mid-crustal magma sheets and led to thetectono-magmatic characteristics of late Mioceneto Recent ignimbrite volcanism. These character-istics suggest that the formation of collapse calde-ras and associated ignimbrite volcanism are morerelated to the mechanical evolution of the Punaplateau rather than direct consequences ofchanges in the geometry of the Wadati^Benio¡zone and/or in the rate of plate convergence.

Acknowledgements

This project was funded by Grants Ri 916/1-1and Ri 916/1-2 from the German Science Foun-dation (DFG) awarded to U.R. and is part of theproject `Deformation processes of the Andes'(SFB 267). I.P. acknowledges Grant PICT98-03817 from the Agencia Nacional de PromocionCient|¢ca y Tecnologica and Grant 815/3 fromthe CIUNSa. This work bene¢ted much from dis-cussion with D. Hindle and A. Schmitt and re-

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311 309

views of an early version of the manuscript by K.Knesel and S. Kay. Reviews for the Journal byM.A. Gutscher, B. Kennedy and an anonymousreviewer improved the quality of our work signi¢-cantly. We thank Ricardo Alonso, Jose Vira-monte, Fernando Hongn and Ignacio Sabinasfor scienti¢c advice and logistical support.[AC]

References

[1] M.C.W. Baker, P.W. Francis, Upper Cenozoic volcanismin the central Andes ^ ages and volumes, Earth Planet.Sci. Lett. 41 (1978) 175^187.

[2] M.H. Ort, Eruptive processes and caldera formation in anested downsag collapse caldera: Cerro Panizos, centralAndes mountains, J. Volcan. Geotherm. Res. 56 (1993)221^338.

[3] B. Coira, P. Ca¡ee, Volcanismo Cenozoico de PunaNorte, in: G. Gonzalez Bonorino, R. Omarini, J. Vira-monte (Eds.), Geologia del Noroeste Argentino, XIVCongresso Geologico Argentino, Tomo I, 1999, pp.371^386.

[4] S.L. de Silva, Altiplano^Puna volcanic complex of thecentral Andes, Geology 17 (1989) 1102^1106.

[5] P.M. Gardeweg, C.F. Ramirez, The La Pacana Calderaand the Atana ignimbrite: A major ash-£ow and resur-gent caldera complex in the Andes of northern Chile, Bull.Volcan. 49 (1987) 547^566.

[6] J.M. Lindsay, Stratigraphy, age relations and magmaticevolution of large-volume felsic ignimbrites of the La Pa-cana Caldera, central Andes, Chile, Scienti¢c TechnicalReport STR99/16, Geoforschungszentrum Potsdam,1999, 141 pp.

[7] R. Seggiaro, Petrolog|a, Geoqu|mica y mecanismos deerupcion del complejo volcanico Coranzuli, Ph.D. thesis,Universidad Nacional de Salta, 1994, 230 pp.

[8] J.G. Viramonte, J.H. Reynolds, C. Del Papa, A. Disalvo,The Corte Blanco garnitiferous tu¡: A distinctive lateMiocene marker bed in Northwestern Argentina appliedto magnetic polarity stratigraphy in the Rio Yacones,Salta Province, Earth Planet. Sci. Lett. 121 (1994) 519^531.

[9] I.A. Petrinovic, Volcanismo Cenozoico asociado al linea-mento Calama Olacapato^El Toro en el tramo compren-dido entre San Antonio de los Cobres y Olacapato, Pro-vincia de Salta, Ph.D. thesis, Universidad Nacional deSalta, 1994, 174 pp.

[10] I.A. Petrinovic, J. Mitjavila, J.G. Viramonte, J. Marti, R.Becchio, M. Arnosio, Geoqu|mica y geocronolog|a de se-cuencias volca©nicas neogenas de trasarco, en el extremooriental de la cadena volcanica transversal del Quevar,noroeste de Argentina, in: Q. Colombo, I. Queralt, I.A.Petrinovic (Eds.), Geologia de los Andes Centrales Mer-

idionales: El Noroeste Argentino, Acta Geologica His-panica 34, 2000, pp. 255^273.

[11] P.W. Francis, L. O'Gallaghan, G.A. Kretzschmar, R.S.Thorpe, R.S.J. Sparks, R.N. Page, R.E. de Barrio, G.Gillou, O.E. Gonzalez, The Cerro Galan ignimbrite, Na-ture 301 (1983) 51^53.

[12] O. Gonzales-Ferran, P.E. Baker, D.C. Rex, Tectonic-vol-canic discontinuity at latitude 27³S Andean Range asso-ciated with Nasca plate subduction, Tectonophysics 112(1985) 423^441.

[13] M.C.W. Baker, The nature and distribution of upper Ce-nozoic ignimbrite centers in the central Andes, J. Volcan.Geotherm. Res. 11 (1981) 293^315.

[14] A.M. Sasso, A.H. Clark, The Farallon Negro Group,Northwest Argentina: Magmatic, hydrothermal and tec-tonic evolution and implications for Cu^Au metallogenyin the Andean back-arc, Soc. Econ. Geol. Newsl. 34(1989) 7^18.

[15] S. Kay, C. Mpodozis, Southern termination of the Cen-tral Andean Volcanic Zone: a ¢rst look to Incapillo Cal-dera/Bonete Region Magmatism, EOS Trans. Am. Geo-phys. Union 80 (1999) F1051.

[16] B. Coira, S. Mahlburg Kay, J. Viramonte, Upper Ceno-zoic magmatic evolution of the Argentine Puna ^ A modelfor changing subduction geometry, Int. Geol. Rev. 35(1993) 677^720.

[17] S. Kay, C. Mpodozis, B. Coira, Neogene magmatism,tectonism, and mineral deposits of the Central Andes(22³ to 33³S Latitude), in: B.J. Skinner (Ed.), Geologyand Ore Deposits of the Central Andes, Soc. Econ.Geol. Spec. Pub. 7, 1999, pp. 27^59.

[18] J.D. Clemens, C.K. Mawer, Granitic magma transport byfracture propagation, Tectonophysics 204 (1992) 339^360.

[19] A.K. Schmitt, J.M. Lindsay, R. Emmermann, Pre-erup-tive magma storage conditions and evidence for externallycontrolled eruption of large-volume central Andeanignimbrites, EOS Trans. Am. Geophys. Union 80 (1999)F982^F983.

[20] O. Bellier, M. Sebrier, Relationship between tectonismand volcanism along the Great Sumatran Fault Zone de-duced by SPOT image analysis, Tectonophysics 233(1994) 215^231.

[21] I. Moore, P. Kokelaar, Tectonic in£uences in piecemealcaldera collapse at Glencoe Volcano, Scotland, J. Geol.Soc. Lond. 154 (1997) 765^768.

[22] J.A. Sal¢ty, Lineamentos transversales al rumbo andinoen el noroeste argentino, IV Congreso Geologico Chileno,2, 1985, pp. 119^137.

[23] J.G. Viramonte, I.A. Petrinovic, Cryptic and partiallyburied calderas along a strike^slip fault system in theCentral Andes, ISAG Grenoble, 1990, pp. 317^320.

[24] U. Riller, J. Greskowiak, J. Ramelow, M. Strecker, Dom-inant modes of Andean deformation in the Calchaqu|River Valley, NW-Argentina, XIV Congreso GeologicoArgentino, Actas I, 1999, pp. 201^204.

[25] R.W. Allmendinger, V.A. Ramos, T.E. Jordan, M. Pal-

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311310

ma, B.L. Isacks, Paleogeography and Andean structuralgeometry, Northwest Argentina, Tectonics 2 (1983) 1^6.

[26] B. Isacks, Uplift of the central Andean plateau and bend-ing of the Bolivian orocline, J. Geophys. Res. 93 (1988)3211^3231.

[27] R.W. Allmendinger, T.E. Jordan, S.M. Kay, B.L. Isacks,The evolution of the Altiplano^Puna Plateau of the Cen-tral Andes, Ann. Rev. Earth Planet. Sci. 25 (1997) 139^174.

[28] J. Kley, C.R. Monaldi, Tectonic shortening and crustalthickness in the central Andes: How good is the correla-tion?, Geology 26 (1998) 723^726.

[29] U. Riller, O. Oncken, I. Petrinovic, M. Strecker, LateCenozoic tectonism and caldera formation in the CentralAndes, EOS Trans. Am. Geophys. Union 80 (1999)F1061.

[30] J.E. Llamb|as, A.M. Sato, J. Tomsic, Geolog|a y carac-ter|sticas qu|micas del stock terciario del Nevado de Acayy vulcanitas asociadas, Assoc. Geol. Arg. Rev. 40 (1985)158^175.

[31] J. Gephard, Topography and subduction geometry in thecentral Andes: Clues to the mechanics of a non-collisionalorogen, J. Geophys. Res. 99 (1994) 12279^12288.

[32] D.S. Vandervoort, T.E. Jordan, P.K. Zeitler, R.N. Alon-so, Chronology of internal drainage and uplift, southernPuna plateau, Argentine central Andes, Geology 23 (1995)145^148.

[33] T.E. Jordan, R.N. Alonso, Cenozoic stratigraphy and ba-sin tectonics of the Andes Mountains, 20³^28³ South Lat-itude, Am. Assoc. Petrol. Geol. Bull. 71 (1987) 49^64.

[34] B. Kraemer, D. Adelmann, M. Alten, W. Schnurr, K.Erpenstein, E. Kiefer, P. van den Bogaard, K. Go«rler,Incorporation of the Paleogene foreland into the NeogenePuna plateau: The Salar de Antofalla area, NW Argenti-na, J. South Am. Earth Sci. 12 (1999) 157^182.

[35] M.E. Grier, J.A. Sal¢ty, R.W. Allmendinger, Andean re-activation of the Cretaceous Salta rift, northwestern Ar-gentina, J. South Am. Earth Sci. 4 (1991) 352^372.

[36] T. Gubbles, B. Isacks, E. Farrar, High-level surfaces, pla-teau uplift, and foreland development, Bolivian centralAndes, Geology 21 (1993) 695^698.

[37] J.G. Viramonte, R.H. Omarini, V. Aran¬a Saavedra, A.Aparicio, L. Garc|a Cacho, P. Parica, Edad, genesis ymecanismos de erupcion de las riolitas granatiferas deSan Antonio de los Cobres, Provincia de Salta, NovenoCongreso Geologico Argentino, Actas III, 1984, pp. 216^233.

[38] R.W. Allmendinger, T. Gubbels, Pure and simple shearplateau uplift, Altiplano^Puna, Argentina and Bolivia,Tectonophysics 259 (1996) 1^14.

[39] F. Pardo-Casas, P. Molnar, Relative motion of the Nazca(Farallon) and South American Plates since late Creta-ceous time, Tectonics 6 (1987) 233^248.

[40] P. Molnar, H. Lyon-Caen, Some simple physical aspectsof the support, structure, and evolution of mountain belts,Geol. Soc. Am. Spec. Pap. 218 (1988) 179^207.

[41] R.A. Marrett, R.W. Allmendinger, R.N. Alonso, R.E.Drake, Late Cenozoic tectonic evolution of the Puna Pla-teau and adjacent foreland, northwestern Argentine An-des, J. South Am. Earth Sci. 7 (1994) 179^207.

[42] R. Somoza, Updated Nazca (Farallon) ^ South Americarelative motions during the last 40 My: implications formountain building in the central Andean region, J. SouthAm. Earth Sci. 11 (1998) 211^215.

[43] E.O. Norabuena, T.H. Dixon, S. Stein, C.G.A. Harrison,Decelerating Nazca^South America and Nazca^Paci¢cPlate Motions, Geophys. Res. Lett. 26 (1999) 3405^3408.

[44] N. Petford, A.R. Cruden, K.J.W. McCa¡rey, J.-L. Vig-neresse, Granite magma formation, transport and em-placement in the Earth's crust, Nature 408 (2000) 669^673.

[45] V. Acocella, F. Salvini, R. Funiciello, C. Faccenna, Therole of transfer structures on volcanic activity at CampiFlegrei (Southern Italy), J. Volcan. Geotherm. Res. 91(1999) 123^139.

[46] X. Yuan, S.V. Sobolev, R. Kind, O. Oncken, G. Bock, G.Asch, B. Schurr, F. Graeber, A. Rudlo¡, W. Hanka, K.Wylegalla, R. Tibi, Ch. Haberland, A. Rietbrock, P. Gie-se, P. Wigger, P. Ro«wer, G. Zandt, S. Beck, T. Wallace,M. Pardo, D. Comte, Subduction and collision processesin the Central Andes constrained by converted seismicphases, Nature 408 (2000) 958^961.

[47] J. Arndt, T. Bartel, E. Scheuber, F. Schilling, Thermaland rheological properties of granodioritic rocks fromthe Central Andes, North Chile, Tectonophysics 271(1997) 75^88.

[48] M. Schmitz, W.-D. Heinsohn, F.R. Schilling, Seismic,gravity, and petrological evidence for the partial melt be-neath the thickened Central Andean crust (21^23³), Tec-tonophysics 270 (1997) 313^326.

[49] C.L. Rosenberg, U. Riller, Partial-melt topology in stat-ically and dynamically recrystallized granite, Geology 28(2000) 7^10.

EPSL 5845 17-5-01 Cyaan Magenta Geel Zwart

U. Riller et al. / Earth and Planetary Science Letters 188 (2001) 299^311 311