The Migration History of the Nazca Ridge Along the Peruvian Active Margin: a Re-Evaluation

Earth and Planetary Science Letters 203 (2002) 665^679 www.elsevier.com/locate/epsl

The migration history of the Nazca Ridge along the Peruvian active margin: a re-evaluation

Andrea Hampel

GEOMAR Research Center for Marine Geosciences, Wischhofstr. 1^3, 24148 Kiel, Germany

Received 21 March 2002; received in revised form 9 July 2002; accepted 23 July 2002

Abstract

The collision zone of the 200 km wide and 1.5 km high Nazca Ridge and the Peruvian segment of the convergent South American margin between 14‡S and 17‡S is characterized by deformation of the upper plate and several hundred meters of uplift of the forearc. This is evident by a narrowing of the shelf, a westward shift of the coastline and the presence of marine terraces. As the Nazca Ridge is oblique with respect to both trench and convergence direction of the Nazca Plate, it migrates southward along the active plate boundary. For reconstructing the migration history of the Nazca Ridge, this study uses updated plate motion data, resulting from a revision of the geomagnetic time scale. The new model suggests that the ridge crest moved laterally parallel to the margin at a decreasing velocity of V75 mm/a (before 10.8 Ma), V61 mm/a (10.8^4.9 Ma), and V43 mm/a (4.9 Ma to present). Intra-plate deformation associated with mountain building in the Peruvian Andes since the Miocene reduces the relative convergence rate between Nazca Plate and Peruvian forearc. Taking an intra-plate deformation at a rate of V10mm/a, estimated from space-geodetic and geological data, into account, does not significantly reduce these lateral migration velocities. Constraining the length of the original Nazca Ridge by its conjugate feature on the Pacific Plate yields a length of 900 km for the subducted portion of the ridge. Using this constraint, ridge subduction began V11.2 Ma ago at 11‡S. Therefore, the Nazca Ridge did not affect the northern sites of Ocean Drilling Program (ODP) Leg 112 located at 9‡S. This is supported by benthic foraminiferal assemblages in ODP Leg 112 cores, indicating more than 1000 m of subsidence since at least Middle Miocene time, and by continuous shale deposition on the shelf from 18 to 7 Ma, recorded in the Ballena industrial well. At 11.5‡S, the model predicts the passage of the ridge crest V9.5 Ma ago. This agrees with the sedimentary facies and benthic foraminiferal stratigraphy of ODP Leg 112 cores, which argue for deposition on the shelf in the Middle and Late Miocene with subsequent subsidence of a minimum of several hundred meters. Onshore at 12‡S, the sedimentary record shows at least 500 m uplift prior to the end of the Miocene, also in agreement with the model. ß 2002 Elsevier Science B.V. All rights reserved.

Keywords: Nazca Ridge; oblique subduction; plate reconstruction; forearc; Peru

1. Introduction

Seamount chains, submarine ridges and other * Present address: GeoForschungsZentrum Potsdam, Tele- grafenberg, 14473 Potsdam, Germany. bathymetric highs on oceanic plates entering sub- Tel.: +49-331-288-1376; Fax: +49-331-288-1370. duction zones will, in general, laterally migrate E-mail address: [email protected] (A. Hampel). along the active margin, unless they are parallel

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 666 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 to the convergence direction (e.g. [1,2]), and may would result in a variable lateral migration veloc- a¡ect the sedimentological and tectonic evolution ity. of the forearc system signi¢cantly. The lateral mo- The fate of bathymetric highs during subduction of such features can lead to a temporal se- tion to greater depth has long been subject to quence of uplift and subsidence of the forearc, controversy. While some authors note the tempo- frequently accompanied by enhanced surface and rally irregular occurrence and reduced number of tectonic erosion as well as steepening of the inner large earthquakes in the vicinity of such features trench wall and faulting in the upper plate (e.g. (e.g. [10]), others argue that subducting sea- [3^8]). These e¡ects are generally recorded in the mounts and ridges form asperities, at which earth- morphology and sedimentary facies of the forearc quakes may nucleate [11] and increase seismic and in uplifted coastal shorelines. As a conse- coupling [12]. In addition, the buoyancy of sub- quence, models resolving the history of forearc ducted bathymetric highs may decrease the dip of and arc systems must account for these three-di- the subducting slab and thus may terminate the mensional e¡ects and their development through magmatic activity in the overriding plate [7,10,13^ time. 15]. The velocity at which a bathymetric high moves An outstanding example of a subducting bathy- along an active margin is controlled by three pa- metric high migrating along an active plate rameters: the convergence velocity vc and the two boundary is the Nazca Ridge, which has a¡ected angles a and P, de¢ned by the orientation of the the Peruvian portion of the long-lived Andean bathymetric high relative to convergence direction subduction zone. Due to southward migration of and trench, respectively (Fig. 1). The lateral ve- the ridge, the Peruvian margin displays, from locity vlat of a bathymetric high parallel to the south to north, di¡erent stages of its tectonic evo- plate boundary is then: lution during and after ridge passage. Various v sina features in the o¡shore and onshore geology of v ¼ c lat sinP the Peruvian margin, such as uplift and subsidence of forearc basins, tectonic erosion of the lower continental slope and uplift of marine ter- Even if the convergence velocity is constant, a races have been attributed to ridge subduction curvature of the trench line, i.e. a variable angle P, [16^21]. Moreover, the coastal area above the subducting ridge was ruptured by two shallow thrust earthquakes with magnitudes of Mw = 8.1 and Mw = 7.7 in 1942 and 1996, respectively [22]. The downward continuation of the ridge has been related to a zone of reduced intermediate depth seismicity and to the southern boundary of the low-angle subduction segment beneath Southern Peru [23^25], which coincides with the terminus of the Quaternary volcanic arc [14,26]. To correlate these di¡erent observations with the subduction of the Nazca Ridge, it is crucial to constrain both the rate of its lateral movement along the margin and the original length of this feature. The ¢rst part of this study calculates the migration velocity of the Nazca Ridge and yields a sig- Fig. 1. Geometric relations between the lateral migration ve- ni¢cantly slower lateral motion than previously locity v of a bathymetric high parallel to an active plate lat inferred [16,18^21,25,27,28], with the consequence boundary, the plate convergence velocity vc, and the orientation of the bathymetric high relative to convergence direction that ages at which the ridge passed speci¢c sites and trench [9]. increase signi¢cantly. The second part speci¢es

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 667 the onset of ridge subduction, assuming that the ward shift of the trench and the magmatic arc original length of the Nazca Ridge approximates [27]. However, interpretations of seismic data that of its conjugate feature on the Paci¢c Plate and ODP cores, in particular in the Lima Basin [18,25,27,28]. at 11.5‡S, indicate that during some periods, the forearc subsided at a lower rate than during times of prevailing long-term tectonic erosion or has 2. Geodynamic setting even been uplifted [17,37]. Regarding the temporal evolution of the colli- The Nazca Ridge is a more than 1000 km long sion zone between the Nazca Ridge and the Pe- and 200 km wide aseismic submarine ridge, which ruvian margin, current models di¡er in the lateral formed at the Paci¢c^Farallon/Nazca spreading migration velocities, in the ages of ridge passage center in the early Cenozoic [25,29,30] (Fig. 2). assigned to di¡erent latitudes and in the predicted The linear crest of the ridge is elevated 1500 m length of the original Nazca Ridge. The following above the surrounding sea £oor and trends reconstructions cover the migration history of the N42‡E. The average crustal thickness of the ridge Nazca Ridge along the entire Peruvian margin: derived from the analysis of Rayleigh waves is Pilger ([25]; his ¢gure 4) shows that the ridge ¢rst 18 þ 3 km [30]. Where the ridge descends beneath came in contact with the Peruvian trench at 5‡S in the South American Plate, the trench does not the Middle Miocene and later passed 10‡S at V9 show a pronounced deviation from its linear Ma. Other studies [16,18,27], based on plate re- trend, but the water depth along the trench line constructions [28] and the NUVEL-1A conver- shoals from 6500 m south of the ridge to 4000 m gence rate [38], inferred that the Nazca Ridge be- at the ridge crest. In bathymetry and side-scan gan to subduct 8 Ma ago at 8‡S and was located sonar images, features indicating ongoing surface at 9‡S and 11.5‡S at 6^7 Ma and 4^5 Ma, respec- erosion and faulting have been identi¢ed on the tively. Three other reconstructions concentrate on continental slope [32,33]. Landwards, the recent the migration of the ridge from the end of the collision zone is expressed by a narrowing of the Miocene to the present: Based on the plate mo- shelf, a seaward shift of the coastline and the tion data by Pardo-Casas and Molnar [39],Hsu presence of raised marine terraces at the coast [19] infers a lateral migration velocity of V71 between 13.5‡S and 15.6‡S [19,20]. Above the mm/a. Machare¤ and Ortlieb [20] use the plate northern £ank of the subducted ridge, the recent motion data by Pardo-Casas and Molnar [39] to subsidence of the marine terraces, which had been deduce a passage of the ridge crest at 13‡S at uplifted earlier by the ridge until the passage of its 4 Ma, i.e. a lateral velocity of V64 mm/a. Le Roux crest, illustrates its southward movement [19,20]. et al. [21] suggest that the ridge crest was located Further inland, the Abancay De£ection (Fig. 2), at V13.5‡S at 5.3 Ma and thus o¡ Lima (12‡S) which marks the northern boundary of the zone before the end of the Miocene, i.e. laterally mov- of active arc volcanism and separates segments of ing at a velocity of V42 mm/a, derived from con- continental crust di¡ering in geochemical compo- vergence rates given by Stein et al. [40]. These sition, has been related to the continuation of the di¡erences in the inferred migration rates of the Nazca Ridge [34]. Nazca Ridge underline the importance of the re- North of the collision zone, a small accretion- evaluation presented here. ary wedge may have begun to grow in the wake of the ridge [18]. Further north, o¡ Central and North Peru, the absence of a large accretionary 3. Reconstruction of the migration history prism and tectonic erosion as the dominant mass transfer regime have been recognized [35,36]. 3.1. Lateral migration velocity Along this part of the margin, long-term tectonic erosion since at least the Middle Miocene has led Unraveling the migration history of subducting to rapid subsidence of the forearc and to an east- ridges, seamount chains and other submarine

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 668 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679

Fig. 2. Map [31] showing the location of the Nazca Ridge, the spatial distribution of seismicity and active volcanoes (black trian- gles; from the Smithsonian Global Volcanism Program). ODP Leg 112 sites and two industrial wells (Ballena, Del¢n) are marked by white circles. The Peruvian low-angle subduction segment is located between 5‡S and 14‡S. Note the gap in the intermediate depth seismicity (70^300 km) (dotted line) and the presence of deep seismic events (500^650 km) beneath Brazil (dashed line). (Earthquake data from 1973 to 2002; US Geological Survey^National Earthquake Information Center.)

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 669 bathymetric highs requires knowledge of past not speci¢ed [42], but may be of the order of 10% plate motions, which can be obtained by two [44]. Since the errors are likely to be smaller in the types of data sets. Plate motions averaged over latest time interval, as suggested by the errors of the last 3 Ma are provided by the NUVEL-1A the NUVEL-1A convergence rates [38], and may model, based on evaluation of spreading rates, be larger in the earliest time interval, this study transform fault azimuths and earthquake slip vec- assigns uncertainties of 5%, 10% and 15% to the tors [38]. On longer time scales, paleo-plate posi- convergence velocities of the 0^4.9 Ma, 4.9^10.8 tions and motions can be reconstructed by ana- Ma and 10.8^16 Ma time intervals, respectively lyzing the magnetic anomalies of the oceanic (Table 1). Using these error limits, the uncertain- crust. This method yields average velocity vectors ties in the ages of ridge passage with respect to the for di¡erent time intervals (e.g. [28,41]). convergence rates of the three time intervals are This study uses updated Nazca (Farallon)^ given in Table 1. Potential errors of the geomag- South American relative motions [42] which take netic time scale and of the convergence azimuths into account a revision of the global geomagnetic for the di¡erent time intervals have not been tak- time scale [43]. This data set provides constant en into account. convergence velocities and directions for di¡erent An implicit assumption of this reconstruction is time intervals for the last 40Ma at di¡erent lat- that the decreasing relative convergence rate be- itudes, of which the values given at 12‡S are ap- tween the Nazca Plate and stable South America plied (Table 1). The convergence rate of 75 mm/a over the last 15^20Ma, as derived from plate for the last 5 Ma [42] agrees well with the NU- reconstructions, equals the amount of relative mo- VEL-1A prediction [38]. Both estimates are higher tion between the Nazca Plate and the Peruvian than the current convergence rate determined by forearc. This assumption has also been the basis space-geodetic measurements, i.e. 61 þ 3 mm/a at for all previous reconstructions of the Nazca 12‡S [44,45]. Since the convergence rate may be Ridge motion [16,18^21,25,27,28]. However, the slowing with time, the space-geodetic values are presence of the Andean mountain belt east of less relevant for this reconstruction. the forearc demonstrates that, strictly speaking, Using the average convergence velocities and this assumption is not correct, since some of the directions for the three latest time intervals, three relative plate motion is taken up by intra-plate displacement vectors and respective paleo-posi- deformation within the South American Plate. tions of the Nazca Ridge relative to a ¢xed South Obviously, this intra-plate deformation tends to American Plate are constructed (Table 1 and Fig. reduce the relative motion between the Nazca 3a). The resulting time path allows one to deter- Ridge and the Peruvian forearc system. At mine when the ridge crest passed a speci¢c point present, a rigorous assessment of the amount on the trench line, assuming a linear continuation and the direction of shortening accommodated of the ridge towards the trench, as suggested by in the Peruvian Andes is di⁄cult due to the lack the shape of the present ridge, and a paleo-trench of su⁄cient geological data. Nevertheless, space- position similar to the present trench line [18^ geodetic measurements [44] and geological pro¢les 21,25,27,28] (Fig. 3b). ([46^48] and references therein) across the Andes Uncertainties in the convergence velocities are can be used to estimate the present-day and past

Table 1 Relative plate motion between Nazca and South American plates at 12‡S [42] Time interval Convergence velocitya Convergence direction Length of displacement vector Age uncertainty [Ma] [km/Ma] [km] [Ma] 0^4.9 (chrons 0^3) 75 þ 4 77‡ 368 þ 20 þ 0.3 4.9^10.8 (chrons 3^5) 106 þ 11 82‡ 625 þ 65 þ 0.6 10.8^16 (chrons 5^5C) 123 þ 18 84‡ 640 þ 94 þ 0.8 a Errors are assumed to be 5%, 10% and 15% for the latest, intermediate, and earliest time interval, respectively.

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 670 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679

Fig. 3. (a) Three paleo-positions of the Nazca Ridge and displacement vectors for the present intersection point of ridge and trench. Gray lines represent the assumed linear continuation of the ridge. Inset (b) shows diagram in which the latitudinal position of the linearly continuing ridge crest on the trench line and the migration velocity of the ridge parallel to the plate boundary are plotted versus time. The two black lines are derived by using the relative plate motion data as given in [42]. The black arrow marks the onset of ridge subduction inferred by this study (Section 3.2). The two gray lines refer to a scenario in which a small amount of intra-plate deformation (10mm/a) accommodated in the Peruvian Andes is subtracted from the convergence rates of [42]. shortening rates across the Eastern Cordillera and ening across the Eastern Andes of 5^8 mm/a for the Subandean belt [49]. These data show that the last 25^10Ma and of 10^15mm/a for the last geologic and space-geodetic displacement rates 10Ma [49]. In order to account for the Andean are generally consistent and that directions of intra-plate deformation, the lateral migration ve- shortening in the Eastern Andes are approxi- locity of the Nazca Ridge is also presented for a mately parallel to the Nazca^South America con- scenario in which an average Andean shortening vergence vector. The data have been interpreted in rate of 10mm/a for the last 16 Ma is subtracted terms of a two-stage model with rates of short- from the relative convergence velocity between

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 671

Nazca and South American plates (gray lines in The N70‡W trending, elongated Tuamotu Pla- Fig. 3b). Considering the intra-plate deformation teau is a composite feature consisting of island tends to slightly increase the ages of ridge passage chains and oceanic plateaus with volcanic edi¢ces assigned to speci¢c latitudes, in other words, the that once were subaerial and today form atolls lateral migration velocity of the ridge slightly de- [54], whereas the Nazca Ridge is characterized creases. However, the geological implications of by smaller, but similar submarine volcanic fea- the model (see below) remain valid, even if the tures [32]. Despite these di¡erences in their topog- intra-plate deformation is taken into account. raphy, both ridges have an overall linear trend. To allow a straightforward comparison of the Therefore, the 4000 m water depth contour line model with previous reconstructions of the Nazca of the Tuamotu Plateau has been used to approx- Ridge motion, the following discussion uses the imate the outline and total length of the original model-curve neglecting intra-plate deformation Nazca Ridge [18,25,27,28]. To estimate the length (black lines in Fig. 3b). Once more detailed infor- of the subducted part of the Nazca Ridge, how- mation on Andean shortening rates and directions ever, it has to be taken into account that the in Peru becomes available, it should be incorpo- northwesternmost part of the Tuamotu Plateau rated into the model. formed on 10^20 Ma old oceanic crust of the In summary, the ¢rst part of the reconstruction Paci¢c Plate, indicating an origin 600 km o¡ the demonstrates that the ridge moved signi¢cantly spreading center [54]. The hotspot that generated slower parallel to the margin than inferred by the northwesternmost part of the Tuamotu Pla- previous studies [16,18^21,25,27,28]. In particular, teau [55] most likely had no e¡ect on the Nazca a ridge of su⁄cient length would have passed the Plate [54]. For this reason, the northwestern end ODP Leg 112 sites in the Trujillo/Yaquina (9‡S) of the plateau probably does not have a counter- and Lima basins (11.5‡S) at V14.5 Ma and at part on the Nazca Plate. Another assumption V9.5 Ma, respectively. Apart from this migration made to specify the onset of ridge subduction is history, deducing the onset of ridge subduction the use of the present trench line as the paleo- requires an estimate of the length of the original trench position [18^21,25,27,28]. ridge. To estimate the length of the original Nazca Ridge, a mirror image of the Tuamotu Plateau 3.2. Original length of the Nazca Ridge and onset is created using its 4000 m contour line. To ¢nd of ridge subduction the correct position of the mirror image on the Nazca Ridge, magnetic anomaly lineations of The preservation of oceanic ridges and plateaus the surrounding sea £oor are ¢tted, using a global in the Southeastern Paci¢c o¡ers the possibility to data set [51,52] together with speci¢c data for the constrain the shape of already subducted parts of Tuamotu Plateau region [53]. Chrons 15^20are bathymetric highs on the Nazca Plate by their the oldest magnetic anomalies common to the mirror images on the Paci¢c Plate (Fig. 4). As sea £oor close to both features (Fig. 4b). these pairs of conjugate highs have formed simul- To ¢t these chrons north and south of the Tua- taneously at the Paci¢c^Farallon/Nazca spreading motu Plateau to the ones on the Nazca Plate, no center (e.g. [29,52]), they are thought to have a scaling of the mirror image is needed, which in- similar length and shape assuming symmetric dicates symmetric sea £oor spreading. On the spreading [25,52]. Nazca Plate, the trends of the chrons are better The Nazca Ridge has a common origin with the constrained north than south of the ridge and Tuamotu Plateau at the Paci¢c^Farallon/Nazca appear to be roughly parallel to each other (Fig. spreading center [25,29,30] and the pre-condition 4b). In contrast, the same magnetic lineations are of symmetric spreading seems to be met, since the at an angle with each other north and south of the respective segments of the Nazca and Paci¢c Tuamotu Plateau. As a consequence, ¢tting the plates between chrons 13 and 23 have similar chrons leads to two endmember positions (Fig. widths (see Figs. 4b and 5). 5). Matching chrons 19 and 20, located south of

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 672 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679

Fig. 4. (a) Bathymetric map [31] of the South Paci¢c showing the Paci¢c^Nazca spreading center and the conjugate features Naz- ca Ridge and Tuamotu Plateau. (b) Outlines of the Nazca Ridge and the Tuamotu Plateau are shown by their 4000 m water depth contour lines. The global age grid [50] of the oceanic crust interpolated from magnetic anomalies is shown by color code. Selected magnetic anomaly lineations are represented by black [51], blue [52] and red [53] lines. the ridges, with chrons 18^21 being parallel, leads phases to be picked as chron 18. Since the details to an abrupt bend of the original Nazca Ridge of the picking procedures are not available for all which results in a N16‡W trend and a length of publications ([51^53] and references therein), this about 1000 km corresponding to an onset of sub- study uses the locations of chron 18 as shown in duction V10.0 Ma ago at 8.5‡S (Fig. 5). Adjust- the published maps. Given that the preferred re- ing chrons 16 and 18, located north of the bathy- construction is additionally constrained by chrons metric highs, with chrons 15^20being parallel, 15 and 16, the possible non-unique identi¢cation leads to the position of the mirror image preferred of chron 18 by di¡erent authors is considered to by this study, because in that case the Nazca have only a minor e¡ect on the reconstructed Ridge continues linearly beneath South America length of the Nazca Ridge. for 1100 km, suggesting that the ¢rst contact of The values of 1000 km and 1100 km for the ridge and trench occurred V12.5 Ma ago at a original length of the Nazca Ridge, as inferred latitude of 10‡S (Fig. 5). Regarding the location above, are maximum values. Taking into account of chron 18, it should be noted that the spatial that the V200 km long northwesternmost part of extent of its magnetic signal allows di¡erent the Tuamotu Plateau most likely does not have a

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 673

Fig. 5. Migration history of the Nazca Ridge on the assumption that the mirrored Tuamotu Plateau resembles the subducted part of the Nazca Ridge. The magnetic anomalies on the Nazca Plate are marked in black and blue. The magnetic lineations north and south of the Tuamotu Plateau have red and green colors, respectively (see inset). At the present collision zone, two endmember models for the continuation of the Nazca Ridge are shown: Adjusting chrons 15^20, located north of both features, yields the red mirror image of the plateau. Fitting chrons 18^21, located south of both ridges, leads to a position of the mirrored Tuamotu Plateau shown as the green mirror image. Both mirror images are plotted without consideration of the variable dip of the subducting plate. For both mirror images, the lighter colors at their northeasternmost ends mark the V200 km long part of the Tuamotu Plateau which most likely does not have a counterpart on the Nazca Plate (see text for details). Thus, the red mirror image with a linear continuation of V900 km is the preferred scenario of this study. Note the coincidence of the preferred red mirror image with the reduced intermediate depth seismicity (dotted line) and with the presence of deep seismic events beneath Brazil (dashed line). For the onset of ridge subduction, three di¡erent scenarios are presented: Using the preferred con¢gu- ration, the original Nazca Ridge entered the trench V11.2 Ma ago at 11‡S (red). If the original Nazca Ridge continues for 1100 km, its subduction began V12.5 Ma ago at 10‡S (light red). The northernmost possible contact of ridge and trench at 8.5‡S cor- responds to a mirror image adjusted to chrons 19^21 (green). counterpart on the Nazca Plate yields the pre- evolution of the Nazca Ridge predicts, for both ferred scenario of this study, in which the original endmember positions described above (Fig. 5), a ridge continues for V900 km beneath South lateral migration history that di¡ers signi¢cantly America and entered the trench V11.2 Ma ago from previous studies [16,18^21,25,27,28]. With at 11‡S (Fig. 5). respect to the two possible positions of the mirror image of the Tuamotu Plateau, this study prefers ¢tting the magnetic anomaly lineations 16 and 18 4. Discussion north of the plateau instead of chrons 19 and 20 south of it for the following reasons: First, with The new model presented for the kinematic this ¢t, the straight Nazca Ridge continues with-

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 674 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 675

out a bend. Second, the outline of the northern ported by the conjugate feature of the Nazca arm of the Tuamotu Plateau resembles the mod- Ridge on the Paci¢c Plate, since the entire Tua- ern Nazca Ridge, in agreement with their prob- motu Plateau is at most 1100 km longer than the able alignment during their common origin [30]. modern Nazca Ridge. Thus, although the trench Apart from that, the uncertainties in the direction has probably been shifted eastward for at least 20 of the magnetic anomalies 16 and 18 are consid- Ma due to tectonic erosion [27], a linearly trend- erably smaller than those of the shorter chrons 19 ing original Nazca Ridge could not have reached and 20, which, like in the Tuamotu Plateau re- the trench north of 10‡S. gion, might not be parallel to chrons 16 and 18. The new reconstruction has signi¢cant implica- Di¡erent orientations of chrons 18 and 19 are tions for models of the tectonic, sedimentological also suggested by the magnetic anomalies of the and geomorphic evolution of the Peruvian forearc Nazca Plate south of the Nazca Fracture Zone and arc systems. In particular, di¡erent seismic (Fig. 4). Another argument is that a linear con- data sets (e.g. [59,60]) and ODP Leg 112 cores tinuation of the Nazca Ridge coincides well with in the Trujillo/Yaquina (9‡S) and Lima basins the zone of reduced intermediate depth seismicity (11.5‡S) have been interpreted [17,18,60,61] in and the southern boundary of the segment of low- light of previous reconstructions suggesting that angle subduction beneath South Peru (Figs. 2, 5). the ridge crest passed these sites V6 Ma and The predicted northeastern end of the ridge cor- V4 Ma ago, respectively [16,18,27,28]. According relates with the cluster of deep seismic events to this study, however, the ridge was not su⁄- beneath Brazil between V8.5‡S and V10.5‡S ciently long to in£uence the region at 9‡S, while (compare Figs. 2 and 5). This agrees well with at 11.5‡S, it already caused maximum uplift V9.5 interpretations of the deep seismicity that propose Ma ago (Fig. 3). The marine and subaerial sedi- an association of the southern earthquake cluster mentological record of the forearc, the onshore with the subducted part of the Nazca Ridge tectonic history, and the temporal and spatial [56,57]. Moreover, a VN42‡E trending ridge co- evolution of volcanism of the Andean magmatic incides with the northern boundary of active vol- arc in Peru support the new model and will be canism and the Abancay De£ection [34,15]. Pilger discussed in some detail. [25], however, argued for the position ¢tted to In marine sediments, uplift of the forearc region chron 19, because the ridge then extends farther can, in general, be derived from a trend to coarser to the north and thus can explain the £at slab deposits, possibly accompanied by an increase in beneath Northern Peru. This northern £at slab, the number of unconformities, and from benthic however, may be caused by the subducted part foraminiferal stratigraphy that gives information of the Carnegie Ridge o¡ North Peru/Ecuador on the water depth at which the sediment was or by another, completely subducted oceanic deposited. At 9‡S, benthic foraminiferal assem- plateau [58]. Taken together, these arguments blages in ODP Leg 112 cores and dredge samples strongly support a linear continuation of the ridge indicate that the continental slope and shelf sub- of V900 km and an onset of ridge subduction sided V1500 m between the Middle Eocene to V11.2 Ma ago at 11‡S. Middle Miocene and experienced further subsi- If the Nazca Ridge, continuing with a linear dence of 1300 m since 12^13 Ma [37,62]. Apart trend, had entered the trench at 8‡S [16,18,27, from that, cores recovered during ODP Leg 112 28], its subduction would have begun V16 Ma and two industrial wells are characterized by the ago and the original ridge would have to be at deposition of ¢ne-grained material, while sandy least 1500 km long. Such a length is not sup- deposits are missing in the Miocene (Fig. 6a)

6 Fig. 6. (a) 9‡S: Lithology of ODP Leg 112 sites [60] and of the Ballena industrial well, located on the shelf, with ages of dated samples (black circles) [63]. (b) 11.5‡S: Lithology of ODP Leg 112 sites that drilled into Miocene strata [60]. Paleo-bathymetry is derived from benthic forminiferal assemblages of site 679, located on the outer shelf [37].

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 676 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679

[60]. Especially in the Ballena industrial well, age of V9.5 Ma derived from the new reconstruc- located above the crest of the outer shelf high, tion for passage of the ridge crest. continuous shale deposition between 18 and The new model is also compatible with the sed- 7Ma[63] argues against the disturbance of the imentological record of the R|¤mac^Chillo¤n rivers deposition milieu due to the passage of a ridge at 12‡S, which eroded deep valleys on the Lima (Fig. 6a). A comparison with the recent collision coastal plain during the Miocene. The alluvial fan zone shows that the shelf area is strongly a¡ected deposited by these rivers experienced uplift of at by the Nazca Ridge. Marine deposits of Eocene least 500 m, which is attributed to the passage of to Upper Pliocene age that correlate with equiv- the Nazca Ridge [21]. Potential sea level changes alent strata in submerged o¡shore forearc basins during the Quaternary and Pliocene are smaller o¡ Central Peru have been raised above sea level than V125 m and have been considered [21]. [19]. The uplift maximum at 12‡S was attained before At 11.5‡S, deposits at ODP Leg 112 sites be- the end of the Miocene [21]. come coarser, with a decrease in mud and an in- Another piece of evidence in support of the crease in silt and sand during the Middle and Late presented model may be inferred from the corre- Miocene (Fig. 6b). At site 679, a layer of con- lation of the Nazca Ridge with the associated seg- glomerates has been deposited before the end of ment of low-angle subduction and the cessation of the Miocene. In cores recovered at ODP site 679, magmatic arc activity. At present, the boundary Middle and Late Miocene benthic foraminiferal between active and ceased volcanism in the south assemblages re£ect deposition on the inner shelf and in the north, respectively, is located in the in shallow water [37]. Following the hiatus at the landward continuation of the ridge, but may end of the Late Miocene, deposition resumed at have gradually propagated southward due to the the outer shelf in the early Pliocene. The next lateral movement of the ridge. O¡shore, volcanic foraminifers-bearing strata are of Quaternary ash layers recovered during ODP Leg 112 have age, with deposition depth £uctuating around been interpreted to show higher activity of the 400 m. At site 682, Middle to Late Miocene fora- Peruvian volcanic arc in the Late Miocene for miniferal assemblages have been deposited at mid- 9‡S than for 12‡S [64]. Onshore geochronological dle bathyal depths (500^1500 m), while the Late data throwing light on a possible southward prop- Pliocene paleo-environment was lower bathyal agating zone, where volcanism has ceased, are, (2000^4000 m) [37]. Site 688 is barren of Late however, rather limited [65,66]. Pulses of Miocene Miocene foraminiferal assemblages, however, be- volcanic activity [65,66] have been interpreted in tween Early Miocene and Quaternary, the paleo- context of the Quechua tectonic phases of the biotope changed from upper middle bathyal (500^ Andean orogeny in Peru during the Middle to 1500 m) to lower bathyal depth (2000^4000 m) Late Miocene [64,67]. The Quechua II (V10 [37]. In addition, earlier investigations based on Ma) and Quechua III (V5 Ma) tectonic phases, dredge samples indicate more than 2000 m subsi- which seem to be related to changes in the relative dence for 6 Ma, since Late Miocene benthic for- plate motion of the Nazca and South American aminifers, living at V500 m depth, were recov- plates [65,39], have been correlated with uncon- ered in the Lima Basin at a water depth of formities in ODP Leg 112 cores at 11.5‡S [16]. more than 2600 m [62]. Based on these initial According to this study, the Nazca Ridge in£u- ODP Leg 112 results, a phase of uplift and ero- enced this region V9.5 Ma ago, which seems to sion at 11.5‡S was derived to begin at 11 Ma and coincide with the Late Miocene Quechua II tec- last until 7 Ma, while 6 Ma ago, a transition from tonic phase. Despite this apparent correlation, it uplift to subsidence occurred [16]. The o¡shore should be noted that the concept of distinct tec- geological record of ODP Leg 112 as summarized tonic phases in Peru has been criticized, as the above shows uplift of the forearc during Middle available temporal constraints argue in favor of and Late Miocene and subsidence since the end of prolonged periods of tectonic activity [68]. Never- the Miocene. This correlates very well with the theless, in the Ecuadorian Andes, subduction of

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 677 the Carnegie Ridge since the Middle Miocene passage with regions that have been in£uenced may be responsible for the development of a high- by the ridge, but otherwise share similar boundary er topography, a compressional stress regime, and conditions. Such a comparison may enable a bet- increased crustal cooling and exhumation rates, ter quanti¢cation of the geodynamic in£uence of deduced from ¢ssion track data in the collision the Nazca Ridge on the Peruvian margin in future zone [69]. studies. The case of the Nazca Ridge emphasizes While, in summary, no single observation is that models regarding the geodynamic evolution conclusive about its relation to the subduction of active margins have to take into account the of the Nazca Ridge, the combination of the argu- migration history and three-dimensional e¡ects ments raised above strongly suggests that the new associated with laterally migrating bathymetric model is more compatible with the existing geo- highs. logical and geomorphic data.

Acknowledgements 5. Conclusions Helpful comments and discussions with Nina This new reconstruction of the migration history Kukowski, Onno Oncken, Ulrich Riller and of the Nazca Ridge along the Peruvian margin David Hindle are gratefully acknowledged. Udo suggests that the lateral motion of the ridge has Barckhausen and Garrett Ito are thanked for decelerated through time. Considering that a small their help with the magnetic anomaly data and amount of the relative convergence rate between useful comments. Many thanks to Edmundo the Nazca and South American plates is taken up Norabuena for his helpful comments on the plate by intra-plate deformation in the Andean moun- motion data. The GMT [70] software was used to tain belt results in slower lateral migration of the create Figs. 2^5. I thank the reviewers Emile ridge. However, this has no e¡ect on the geolog- Okal, Tim Dixon and Steven Cande for construc- ical implications of the new model. On the as- tive comments that helped to improve the manu- sumption that the original Nazca Ridge has a script. Funding was provided by the German length similar to its mirror image on the Paci¢c Ministry of Education, Science and Technology Plate, it continues for V900 km beneath South (BMBF) within the GEOPECO project (Grant America. Therefore, the northeastern end of the no. 03G0146A).[AC] Nazca Ridge entered the trench V11.2 Ma ago at 11‡S. As a consequence, the ridge did not have an impact on the region north of 10‡S, where the northern transect of ODP Leg 112 is located. References The region at 11.5‡S o¡ Lima has been a¡ected by V [1] J. Corrigan, P. Mann, J.C. Ingle, Forearc response to ridge subduction 9.5 Ma ago. Support for the subduction of the Cocos Ridge, Panama-Costa Rica, model is provided by the sedimentological and pa- Geol. Soc. Am. Bull. 102 (1990) 628^652. leo-bathymetric record in ODP Leg 112 and indus- [2] T.W. Gardner, D. Verdonck, N.M. Pinter, R. Slingerland, trial well cores. At 9‡S, cores show mostly ¢ne- K.P. Furlong, T.F. Bullard, S.G. Wells, Quaternary uplift grained sediments on the continental slope and, astride the aseismic Cocos Ridge, Paci¢c coast, Costa Rica, Geol. Soc. Am. Bull. 104 (1992) 219^232. on the shelf, continuous shale deposition. At [3] W.Y. Chung, H. Kanamori, A mechanical model for plate 11.5‡S, the predicted age of the new model corre- deformation associated with aseismic ridge subduction in lates well with a Late Miocene period of uplift and the New Hebrides Arc, Tectonophysics 50(1978) 29^40. erosion followed by subsidence since V6Ma. [4] L.V. LeFevre, K. McNally, Stress distribution and sub- In light of this study, seismic and drilling data duction of aseismic ridges in the Middle America subduction zone, J. Geophys. Res. 90(1985) 4495^4510. sets acquired along the Peruvian margin in the [5] W.R. McCann, R.E. Habermann, Morphologic and geo- last decades o¡er the possibility to compare re- logic e¡ects of the subduction of bathymetric highs, Pure gions that have not been a¡ected by the ridge Appl. Geophys. 129 (1989) 41^69.

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart 678 A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679

[6] S.E. Lallemand, J. Malavieille, S. Calassou, E¡ects of oce- [23] M. Barazangi, B. Isacks, Spatial distribution of earth- anic ridge subduction on accretionary wedges experimen- quakes and subduction of the Nazca Plate beneath South tal modeling and marine observations, Tectonics 11 (1992) America, Geology 4 (1976) 686^692. 1301^1313. [24] A. Hasegawa, I.S. Sacks, Subduction of the Nazca Plate [7] M. Cloos, Lithospheric buoyancy and collisional orogen- beneath Peru as determined from seismic observations, esis: subduction of oceanic plateaus, continental margins, J. Geophys. Res. 86 (1981) 4971^4980. island arcs, spreading ridges and seamounts, Geol. Soc. [25] R.H. Pilger, Plate reconstructions, aseismic ridges, and Am. Bull. 105 (1993) 715^737. low-angle subduction beneath the Andes, Geol. Soc. [8] P. Schnuerle, C.-S. Liu, S.E. Lallemand, D.L. Reed, Am. Bull. 92 (1981) 448^456. Structural insight into the South Ryukyu margin e¡ects [26] A. Nur, Z. Ben-Avraham, Volcanic gaps and the con- of the subducting Gagua Ridge, Tectonophysics 288 sumption of aseismic ridges in South America, in: L.D. (1998) 237^250. Kulm, J. Dymond, E.J. Dasch, D.M. Hussong (Eds.), [9] E.L. Geist, M.A. Fisher, D.W. Scholl, Large-scale defor- Nazca Plate: Crustal Formation and Andean Conver- mation associated with ridge subduction, Geophys. J. Int. gence, Geol. Soc. Am. Mem. 154, 1981, pp. 729^740. 115 (1993) 344^366. [27] R. von Huene, S. Lallemand, Tectonic erosion along the [10] J. Kelleher, W. McCann, Buoyant zones, great earth- Japan and Peru convergent margins, Geol. Soc. Am. Bull. quakes and unstable boundaries of subduction, J. Geo- 102 (1990) 704^720. phys. Res. 81 (1976) 4885^4896. [28] S.C. Cande, Nazca-South American Plate interactions [11] M. Cloos, Thrust-type subduction-zone earthquakes and since 50mybp, in: D.M. Hussong, S.P. Dang, L.D. seamount asperities a physical model for seismic rupture, Kulm, R.W. Couch, T.W.C. Hilde (Eds.), Peru Continen- Geology 20 (7) (1992) 601^604. tal Margin, Marine Sciences International, Woods Hole, [12] C.H. Scholz, C. Small, The e¡ect of seamount subduction MA, 1985. on seismic coupling, Geology 25 (6) (1997) 487^490. [29] R.H. Pilger, D.W. Handschumacher, The ¢xed hotspot [13] P.R. Vogt, A. Lowrie, D.R. Bracey, R.N. Hey, Subduc- hypothesis and origin of the Easter-Salas y Gomez-Nazca tion of aseismic oceanic ridges: e¡ects on shape, seismicity trace, Geol. Soc. Am. Bull. 92 (1981) 437^446. and other characteristics of consuming plate boundaries, [30] M.T. Woods, E.A. Okal, The structure of the Nazca Geol. Soc. Am. Bull. Spec. Paper 172 (1976) 59 pp. Ridge and Sala y Gomez seamount chain from dispersion [14] S. McGeary, A. Nur, Z. Ben-Avraham, Spatial gaps in of Rayleigh waves, Geophys. J. Int. 117 (1994) 205^222. arc volcanism: the e¡ect of collision or subduction of [31] W.H.F. Smith, D.T. Sandwell, Global sea £oor topogra- oceanic plateaus, Tectonophysics 119 (1985) 195^221. phy from satellite altimetry and ship depth soundings, [15] M.-A. Gutscher, W. Spakman, H. Bijwaard, E.R. Eng- Science 277 (1997) 1956^1962. dahl, Geodynamics of £at subduction seismicity and to- [32] R.A. Hagen, R. Moberly, Tectonic e¡ects of a subducting mographic constraints from the Andean margin, Tecton- aseismic ridge: the subduction of the Nazca Ridge at the ics 19 (2000) 814^833. Peru Trench, Mar. Geophys. Res. 16 (1994) 145^161. [16] R. von Huene, E. Suess, Leg 112 shipboard scienti¢c party, [33] C. Li, A.L. Clark, Tectonic e¡ects of the subducting Naz- Ocean Drilling Program Leg 112, Peru continental margin: ca Ridge on the Southern Peru continental margin, Mar. Part 1, tectonic history, Geology 16 (1988) 934^938. Petrol. Geol. 11 (1994) 575^586. [17] R. von Huene, I. Pecher, Vertical tectonics and the origins [34] N. Petford, M.P. Atherton, Crustal segmentation and the of BSRs along the Peru margin, Earth Planet. Sci. Lett. isotopic signi¢cance of the Abancay De£ection: Northern 166 (1999) 47^55. Central Andes (9^20‡S), Rev. Geol. Chile 22 (1995) 235^ [18] R. von Huene, I. Pecher, M.-A. Gutscher, Development 243. of the accretionary prism along Peru and material £ux [35] R.W.R. Rutland, Andean orogeny and ocean £oor after subduction of Nazca Ridge, Tectonics 15 (1996) spreading, Nature 233 (1971) 252^255. 19^33. [36] D.W. Scholl, M.N. Christensen, R. von Huene, M.S. [19] J.T. Hsu, Quaternary uplift of the Peruvian coast related Marlow, Peru-Chile trench sediments and sea £oor to the subduction of the Nazca Ridge: 13.5 to 15.6 de- spreading, Geol. Soc. Am. Bull. 81 (1970) 1339^1360. grees south latitude, Quat. Int. 15/16 (1992) 87^97. [37] J.M. Resig, Benthic foraminiferal stratigraphy and paleo- [20] J. Machare¤, L. Ortlieb, Plio-Quaternary vertical motions environments o¡ Peru, leg 112, in: E. Suess, R. von and the subduction of the Nazca Ridge, central coast of Huene et al. (Eds.) Proc. ODP, Init. Rpts. 112, College Peru, Tectonophysics 205 (1992) 97^108. Station, TX, 1990, pp. 263^296. [21] J.P. le Roux, C. Tavares Correa, F. Alayza, Sedimentol- [38] C. DeMets, R.G. Gordon, D.F. Argus, S. Stein, Current ogy of the R|¤mac-Chillo¤n alluvial fan at Lima, Peru, as plate motions, Geophys. J. Int. 101 (1990) 425^478. related to Plio-Pleistocene sea level changes, glacial cycles [39] F. Pardo-Casas, P. Molnar, Relative motion of the Nazca and tectonics, J. South Am. Earth Sci. 13 (2000) 499^510. (Farallon) and South American Plates since Late Creta- [22] J. Swenson, S. Beck, Source characteristics of the 12 No- ceous time, Tectonics 6 (1987) 233^248. vember 1996 M(w) 7.7 Peru subduction zone earthquake, [40] S. Stein, J.E. Engeln, C. DeMets, R.G. Gordon, D. Pure Appl. Geophys. 154 (1999) 731^751. Woods, P. Lundgren, D. Argus, C. Stein, D.A. Wiens,

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart A. Hampel / Earth and Planetary Science Letters 203 (2002) 665^679 679

The Nazca-South America convergence rate and the re- [56] E.A. Okal, C.R. Bina, The deep earthquakes of 1921-1922 currence of the great 1960Chilean earthquake, Geophy. in Northern Peru, Phys. Earth Planet. Inter. 87 (1994) 33^ Res. Lett. 13 (1986) 713^716. 54. [41] C.L. Mayes, L.A. Lawver, D.T. Sandwell, Tectonic histo- [57] E.A. Okal, C.R. Bina, The deep earthquakes of 1997 in ry and new isochron chart of the South Paci¢c, J. Geo- Western Brazil, Bull. Seismol. Soc. Am. 91 (2001) 161^ phys. Res. 95 (1990) 8543^8567. 164. [42] R. Somoza, Updated Nazca (Farallon)-South America [58] M.-A. Gutscher, J.-L. Olivet, D. Aslanian, J.-P. Eissen, R. relative motions during the last 40My: implications for Maury, The ‘lost Inca Plateau’: cause of £at subduction mountain building in the Central Andean region, J. South beneath Peru?, Earth Planet. Sci. Lett. 171 (1999) 335^ Am. Earth Sci. 11 (1998) 211^215. 341. [43] S.C. Cande, D.V. Kent, Revised calibration of the geo- [59] L.D. Kulm, J. Dymond, E.J. Dasch, D.M. Hussong magnetic timescale for the Late Cretaceous and Cenozoic, (Eds.), Nazca Plate: Crustal Formation and Andean Con- J. Geophys. Res. 100 (1995) 6093^6095. vergence, Geol. Soc. Am. Mem. 154, 1981. [44] E. Norabuena, T.H. Dixon, S. Stein, C.G.A. Harrison, [60] E. Suess, R. von Huene et al., Proc. ODP, Init. Rpts. 112, Decelerating Nazca-South America and Nazca-Paci¢c College Station, TX, 1988. Plate motions, Geophys. Res. Lett. 26 (1999) 3405^3408. [61] E. Suess, R. von Huene et al., Proc. ODP, Sci. Results [45] D. Angermann, J. Klotz, C. Reigber, Space-geodetic esti- 112, College Station, TX, 1990. mation of the Nazca-South America Euler vector, Earth [62] L.D. Kulm, T.M. Thornburg, E. Suess, J. Resig, P. Fryer, Planet. Sci. Lett. 171 (1999) 329^334. Clastic, diagenetic and metamorphic lithologies of a sub- [46] E. Koch, Die Tektonik im Subandin des Mittel-Ucayali- siding continental block: Central Peru forearc, in: E. Su- Gebietes, Ostperu, Geotekton. Forsch. 15 (1962) 1^67. ess, R. von Huene et al. (Eds.), Proc. ODP, Init. Rpts. [47] M. Se¤brier, A. Lavenu, M. Fornari, J.P. Soulas, Tectonics 112, College Station, TX, 1988, pp. 91^108. and uplift in the Central Andes (Peru, Bolivia, and North- [63] H. Schrader, J.C. Castaneda, The Ballena and Del¢n wells ern Chile) from Eocene to present, Ge¤odynam. 3 (1988) o¡ Central Peru: revised ages, in: E. Suess, R. von Huene 85^106. et al. (Eds.), Proc. ODP, Init. Rpts. 112, College Station, [48] J. Kley, C.R. Monaldi, Tectonic shortening and crustal TX, 1990, pp. 209^215. thickness in the Central Andes: How good is the correla- [64] A. Pouclet, H. Cambray, J.-P. Cadet, J. Bourgois, P. De tion?, Geology 26 (1998) 723^726. Wever, Volcanic ash from Leg 112 o¡ Peru, in: E. Suess, [49] D. Hindle, J. Kley, E. Klosko, S. Stein, T. Dixon, E. R. von Huene et al. (Eds.), Proc. ODP, Init. Rpts. 112, Norabuena, Consistency of geologic and geodetic dis- College Station, TX, 1990, pp. 465^480. placements during Andean orogenesis, Geophys. Res. [65] D.C. Noble, E.H. McKee, E. Farrar, U. Petersen, Episo- Lett. 29 (2002) 10.1029/2001GL013757. dic Cenozoic volcanism and tectonism in the Andes of [50] R.D. Mueller, W.R. Roest, J.-Y. Royer, L.M. Gahagan, Peru, Earth Planet. Sci. Lett. 21 (1974) 213^220. J.G. Sclater, Digital isochrons of the world’s ocean £oor, [66] F. Me¤gard, H. Philip, Plio-Quaternary tectono-magmatic J. Geophys. Res. 102 (1997) 3211^3214. zonation and plate tectonics in the Central Andes, Earth [51] S.C. Cande, J.L. LaBreque, R.L. Larson, W.C. Pitman, Planet. Sci. Lett. 33 (1976) 231^238. Magnetic lineations of the world’s ocean basins, Map [67] F. Me¤gard, D.C. Noble, E.H. McKee, E. Bellon, Multiple with text, AAPG, Tulsa, OK, 1989. phases of Neogene compressive deformation in the Aya- [52] S.C. Cande, W.F. Haxby, Eocene propagating rifts in the cucho intermontan basin, Andes of Central Peru, Geol. Southwest Paci¢c and their conjugate features on the Soc. Am. Bull. 95 (1984) 1108^1117. Nazca Plate, J. Geophys. Res. 96 (1991) 19609^19622. [68] C. Noblet, A. Lavenu, R. Marocco, Concept of continu- [53] M. Munschy, C. Antoine, A. Gachon, Evolution tectoni- um as opposed to periodic tectonism in the Andes, Tec- que de la re¤gion des Tuamotu, oce¤an Paci¢que Central, tonophysics 255 (1996) 65^78. C.R. Acad. Sci. Paris 33 (1998) 941^948. [69] R.A. Spikings, W. Winkler, D. Seward, R. Handler, [54] G. Ito, M. McNutt, R.L. Gibson, Crustal structure of the Along-strike variations in the thermal and tectonic re- Tuamotu Plateau, 15‡S, and implications for its origin, sponse of the continental Ecuadorian Andes to the colli- J. Geophys. Res. 100 (1995) 8097^8114. sion with heterogeneous oceanic crust, Earth Planet. Sci. [55] E.A. Okal, A. Cazenave, A model for the plate tectonic Lett. 186 (2001) 57^73. evolution of the East-Central Paci¢c based on SEASAT [70] P. Wessel, W.H.F. Smith, Free software helps map and investigations, Earth Planet. Sci. Lett. 72 (1985) 99^116. display data, EOS Trans. AGU 72 (1991) 441.

EPSL 6378 7-10-02 Cyaan Magenta Geel Zwart