You are on page 1of 9
Plate reconstructions, aseismic ridges, and low-angle subduction beneath the Andes REX H, PILGER, JR. ABSTRACT Low-angle—subduction segments be- neath Peru and Chile are believed to repre sent interaction ofthe subduction zone with the aseismic Nazca and Juan Fernandez Ridges. On the basis of symmetric sea- floor—spreading models and plate-horspot reconstructions, predicted continuations of the relatively buoyant ridges correspond well with the limits of the low-angle— subduction segments, Further, the history of interaction of the aseismic ridges with the subduction zone, as predicted by plate re- constructions, is not inconsistent with available information on volcanic episodes of the Andes, if the contemporary correla tion of voleanic gaps and low-angle sub- duction is also applicable to the rest of the late Cenozoic ‘The occurrence of deep seismicity be- ‘neath the eastern flanks of the Andes is also believed to reflect the effects of low-angle subduction. The deep-seismic segments are adjacent to segments which are presently, ‘or were recently, experiencing low-angle subduction. Thus, the deep seismicity is in- terpreted as representative of remnants of “normal” oceanic lithosphere subducted before interaction of aseismic ridges with the subduction zone. ‘Other geologic phenomena are associated ‘with low-angle subduction beneath South America, including landward shift in loci of tectonism, surficial subsidence, and Lara- mide-style deformation. The aseismic- ridge—buoyancy hypothesis appears to be a valuable predictive tool for the interpreta- tion of such geologic effects INTRODUCTION A considerable body of data now sup- ports the hypothesis that the inclined seis mic zone beneath Peru and Chile consists of four major segments with contrasting ang les of dip (Fig 1) [sacks and Molnar, 1971; ‘Geological Soc of Amerie Balti, Pat, 92 ‘ws Figure 1, Peru-Chile subduction zone: ‘contours (in kilometres), on top of inclined seismic zone and active volcanic centers, from Isacks and Barazangi (1977); motion vectors from Chase (1978). Srauder, 1973, 1975; Barazangi and Isacks, 1976, 1979; Isacks and Barazangi, 197; Hasegawa and Sacks, 1979). Two of the segments, each dipping at angles of ahout 30", exist beneath the volcanic central ‘Andes of northern Chile and southernmost Peru and the volcanically active southern Andes of southern Chile. The other two Segments dip at very low angles beneath volcanically inactive regions of northern and central Peru and central Chile. The rec- ognition of very low angle subduction and its cortespondence with the absence of ac 1-486, 6 fas 1 table, July 198 Department of Geology, Louisiana State University, Baton Rowge, Louisiana 70803 tive voleanism has significant implications for models of subduction-related mag- rmatism and tectonics and, in all probability, for the driving mechanisms of plate This pape is devoted to a discussion of cone major hypothesis for the origin of low angle subduction — that the relative buoyancy of aseismic ridges and island seamount chains involved in subduction is responsible for the law subduction angle The aseismic Nazca Ridge and the Juan Femander island~seamount chain each in- tersect the Peru-Chile trench near the souchern boundaries of low-angle—sub dlucsion segments beneath westezn South America and may be the controlling infivences on the inclination of the sub- duced slab (Fig 1). Vogt (1973), Vogt and ‘others (1976), and Kelleher and MeCann (1976, 197) have empirically documented anomalous effects associated with subduc: ton of aseismic ridges and other anomalous anc eras, with emphasis onthe western Pacific, which they attribute tothe buoyan- cy of the ridges. These effects include Physiogeaphic discontinuities, anomalous Seismicity, and retarded back-arc spreading, which aze largely surficial or restricted 10 Shallow depths (~50 km or less. The low- angle subduction segments beneath the Andes extend to depths of about 150 km (Isacks and Barszangi, 1977), implying deeper influence of aseismic ridges and island-seamount chains than in the western Paci, ifthe hypothesis is correct. “This paper, an extension of earlier studies (Pilger, 1977; Aleman-Ramirez and Pilger, 1978), elaborates and explores the ascis mi-ridge—low-angle-subduction hypothe- sis by attempting to predict the trend ofthe possibly subducted portions of the Nazca Ridge and Juan Fernandez Island-seamount chain using plate reconstructions. It also presents a feamework for more complete test ofthe hypothesis when more detailed knowledge of the volcano-tectonic history ‘ofthe Andes is available. Further the impli- cations of low-angle subduction for the tectonic evolution of Cordillera mountain systems in general, and the Andes in par ticular, and for the deep seismicity of the Peru-Chile subduction zone are considered. ORIGIN OF LOW-ANGLE SUBDUCTION [Among the factors, other than interaction with aseismic ridges, which may be respon- sible for variations in the angle of dip of the descending oceanic slab in subduction ones, ate vations in (1) convergence rate, (2) rate of “absolute motion” of the upper plate, and (3) age of the descending slab, which also may control the style of surface teetonism (Cross and Pilger, 1978a5, ‘Molnar and Atwater, 1978). That subduc tion beneath all of Peru and Chile in the depth range of 50 to 350 km occurs at ang. les of dip of 30° oF less is probably a reflection of a combination of these factors: the age of the descending slab is apparently less than 50 my. (Herron, 1972), and the rate of Nazca—South American con- vergence is quite high (Minster and Jordan, 1978). Further, recent “absolute-motion”™ ‘models imply a significant teenchward component of motion of the upper South ‘American plate (Chase, 1978; Minster and Jordan, 1978). However, the contrast be- ‘tween the very low-angle— and moderate angle~subduction segments cannot be as- bed to any of the three controls. The boundaries between the subduction seg. ments do not correspond with any trans verse plate boundaries in ether the upper or lower plate, so that contrasts in the con- vergence or “absolute-motion” rates are in- adequate as factors controlling the semen: tation of the subduction zone Carr and others (1974) have suggested that boundaries between subduction seg. ments may reflect fractures developed along fossil transform faults, although they did not elaborate on the causes of the variable dips of the segments. In any case, the boun- aries between segments. beneath South ‘America do not correspond with recognized fracture zones (compare with Mammerickx and others, 1975; Mammerickx and Smith, 1979; Herron, 1972; Handschumacher, 1976}, Jischke (1975) developed a hypothesis for segmented subduction based on inferred hydrodynamic forces developed in the vicinity of the descending slab. Unfortu- nately, itis difficult to make testable pre- dictions based on Jischke’s model, due to LOW-ANGLE SUBDUCTION BENEATH ANDES uncertainties in the rheologic properties of the descending slab and surrounding as- thenosphere and a lack of knowledge of as thenospheric and mantle-flow config trations. For these reasons, it seems prefer- able to first consider alternative models, Buoyancy of subducted anomalous crust is the preferred model explored here. Kel- leher and McCann (1976, 1977) developed this hypothesis on the basis of crude calew lations of the celative buoyancy effect of the aseismic ridge, With available estimates of the average density of oceanic crust, litho- spheric mantle, and the asthenosphere, they showed that the gravitational instability of “normal” oceanic crust may be significantly reduced by the increased thickness of the ‘oceanic crust beneath aseismic ridges. Un- certainty in density estimates of the litho- sphere and asthenosphere is 00 great 10 provide more detailed calculations than those made by Kelleher and McCann. Nevertheless, the significantly increased thickness of oceanic crust beneath various aseismic ridges, including the Naaca Ridge, is well documented (Kogan, 1979; Detrick and Watts, 1979; Cutler, 1977). ‘seismic ridges such 26 the Nazca Ridge appear to form synchronously with adja- ‘cent oceanic crust at mid-ocean spreading ‘centers, as indicated by age dating (Sclater and Fisher, 1974), subsidence curves (Det rick and others, 1977), and isostatic geavity anomalies (Kogan, 1979; Detrick and ‘Watts, 1979), so that there is likely to be little thermal contrast berween the ridges and adjacent crust. Any buoyancy effect ddue to the ridge is probably ascribable to the well-documented increase in crustal thickness beneath aseismic ridges (Cutler, 197; Kogan, 1979), although the suberus- tal mantle may be’ significantly different from normal suboceanic mantle, as indi- cated by anomalous seismic-refraction vel- cities (Cutler, 1977) In contrast with aseismic ridges, island seamount chains such as the Juan Fernan: dee chain appear to form within plates, in association with a thermal anomaly of un- certain origin, as indicated by their youth relative to adjacent, older oceanic crust. As a consequence of ther intraplate origin any increase in ceustal thickness beneath such chains probably represents only the vol- ccanic edifice built on the sea floor (Detrick and Crough, 1978). However, the thermal anomaly responsible for the chains aso re sults ina decrease in density andlor thick- ress of the lithosphere beneath young chains, as manifested in elevation of the sea floor adjacent to individual edifices (Detrick 449 and Crough, 1978). Thus, both aseismic ridges and island-seamount chains repre- sent density anomalies in oceanic plates the former due to thickened oceanic crust, the Tatter due to elevated temperatures and cor- sequent thermal expansion of the litho- sphere. ‘A major objection 10 the buoyancy hypothesis comes from theoretical and ex- perimental evidence which suggests that oceanic erust should transform to much denser eclogite in the depth range of 40 t0 30 km (Ahrens and Schubert, 1975) Such a transformation should reinforce the gravi- tational instability of the descending plate, instead of maintaining relative buoyancy. Clearly, examination ofthe apparent effects of subduction of aseismic ridges is appro- priate, in light of the possible eclogite trans- formation alone. EMPIRICAL EFFECTS OF ASEISMIC RIDGE SUBDUCTION In addition to the Nazca and Juan Fer nandez Ridges, at least two other aseismic ridges and two seamount chains appear to be interacting with subduction zones. Such inceractions have well-documented effects fon the inclined seismic zone. A distinctive kink in the inclined seismic zone is de- veloped down dip from the intersection of the Louisville Ridge with the Tonga- Kermadec Ridge (Vogt and others, 1976; Isacks and Barazangi, 1977}; the dip of the inclined seismic zone appears to be slightly reduced along the kink, The arc-trench width of the Alaskan Peninsula—Aleutian Arc appears 10 increase from south to north in the vicinity of the intersection of the Kodiak-Bowie seamount chain with the Aleutian trench, even when the effects of sedimentary accretion are taken into ac count (Jacobs and otbers, 1977) The intersection of the Middle American trench and the Cocos Ridge corresponds with a voleanic gap and a distiner decrease jn the angle of subduction of the Cocos plate beneath Costa Rica (Cross and Pilger, 1979; and unpub. data). The intersection of the Camegie Ridge with the PerwChile trench appears to correspond with a pecul- jar orientation of the inclined seismic zone beneath northern Peru and Ecuador (Pen: ington, 1975). In each of the above-cited cases, a buoyancy effect probably exists; in any case, evidence for enhanced gravitational instability is clearly lacking. These observa- tions definitely present problems for the eclogite-transformation hypothesis, The 450 thickened oceanic crust beneath aseismic ridges, particularly if transformed to eclo- gite, should result in an increase in dip of. the inclined seismic zone, rather than a de- crease, as the crust presumably makes up a larger portion of the lithosphere beneath, aseismic ridges. ‘The Nazca Ridge, an aseismic ridge formed at a spreading center (Cutler, 19775 Pilger and Handschumacher, 1981), inter sects the Peru-Chile trench within the lower half of the Peruvian segment of low-angle subduction (Fig. 1). As the dimensions of the segment are much larger than the width fof the ridge, itis difficult to see how the subducted portion could be responsible for buoyanely maintaining the shallow dip of the entire segment, if the subducted ridge has the same northeasterly trend as the as yet unaffected portion. It is believed that stronger support for the buoyancy hypoth: cesis would result if bend in the subducted part of the Nazca Ridge could be demon- strated, and that the northern part of the Tow-angle—subduction segment is com- posed of largely aseismic ridge material Isacks and Barazangi (1977) noted the correspondence of the Juan Fernandez island-seamount chain with the southern oundary of the central-Chile segment of ow-angle subduction and postulated that the chain might be a hinge-faul effect. As with the Nazca Ridge, independent docu- ‘mentation of an anomalous (or normal") ‘nature of the lithosphere within the gently inclined slab is desirable as a test of the buoyancy hypothesis, ‘THE SUBDUCTED PORTION OF ‘THE NAZCA AND JUAN FERNANDEZ RIDGES As geophysical techniques for resolving the structure and composition of descend- ing slabs are stil in theie infancy, determi- nation of thickness variations in the crustal portions of the sinking plates is not yet possible. However, various models for the origin of aseismic ridges and island- seamount chains have some predictive value, when combined with plate recon structions, so that estimates of the positions fof the subducted portions of the Nazca Ridge and San Fernandez chain are possi- be. ‘The most prominent hypothesis for the ‘origin of anomalous linear volcanic chains is the fixed-horspot hypothesis of Wilson (1963) and Morgan (1972). At earlier stages of the present investigation, atemprs were made to predict the position of the RH, PILGER, JR. subducted portions of the Nazca and Juan Fernandez Ridges (Pilger, 1977; Aleman- Ramirez and Piger, 1978) by applying the fixed-hotspot hypothesis, using various published plate-hotspot-motion parame ters forthe Pacific and Nazca plates (includ- ing those of Francheteau and others, 1970; Clague and Jarrard, 1973b; Morgan, 1972; and Pilger, 1978). However, difficulties ‘with generation of a satisfactory fixed: hotspot-plate motion model became clear, as expanded upon elsewhere (Pilger and ‘Handschumacher, 1981). What was appar- cent, however, was that the Nazca and Tuamotu Ridges show a common origin; oth appear to be aseismic ridges which ‘owe their origins to processes at the Pacifc-Farallon spreading center, along, which the adjacent normal oceanic litho- sphere of the Pacific and Nazca plates formed, Therefore, as *micror images,” the older part of the Tuamoru Ridge can be used to predict the configuration of the cor- responding, subducted Nazca Ridge, with a few simple assumptions. If the older parts of the two aseismic ridges formed in a manner similar to that of the younger parts — symmetric with respect to the paleospreading centers recorded in ‘magnetic lineations — and if no significant change in sea-foor~spreading direction oc- curted during their formation, mirsor- image reflections of the Tuamoru Ridge ‘across identified lineations provide synthe- tic images of the Nazca Ridge. Unfortu- Figuee 2. Predicted continuations of the Nazca Ridge beneath South Ame nately, the trends of lineations older than anomaly 16 are less well defined than younger trends adjacent to either the ‘Tuamotu or Nazca Ridges. The trend of anomaly 19 near the Tuamotu Ridge can be cstimated from the trend of the lineation facther south (Fig. 2, taken from Weissel and others, 1977). Two alternate reflection images are presented in Figure 2, for anomaly 16 {whose trend is well known, Hecron, 1972; and for anomaly 19, as esi mated). As a change in the direction of seafloor spreading is suggested by the anomaly 16 and 19 identifications of Weis: sel and others (1977), the older image is probably preferable. Interestingly, the southern edge of the mirror image for anomaly 16 corresponds ‘well with the southern boundary of the low-angle-subduction segment beneath Peru (Fig, 2), but it does not extend into the northern section of the seismically active part ofthe segment. The anomaly 19 mirror image appears to produce a better corre- spondence. Note thatthe predicted position Of the subducted part of the Nazca Ridge coincides well with the limits of the low- langle—subduction segment, as would seem. to be required if aseismic ridges were indeed responsible for such a phenomenon. The simplicity of this approach to synthesizing the alder part of the Nazca Ridge clearly > lustrates the plausibility of postulating a genetic correspondence between the ridge and the low-angle-subduction segment. a (no dip effects are taken into account), mirror image to the Tuamotu Ridge on the Pacific plate: A. Plane of symmetry is anomaly 16. B. Plane of symmetry is anomaly 19, although trend of anomaly is poorly known. LOW-ANGLE SUBDUCTION BENEATH ANDES 3: do 28 ° Ss 10 g 3 z ; ee : 3 13 ° ° fa esr e Ge ince 2x o \ \ on Figure 3. Globally reconstructed positions of Nazca plate (indicated by Nazea and Juan Fernandez Ridges) relative to South America using parameters of Table 1. Unfortunately, because Juan Fernandez Ridge occurs entirely within the Nazca plate i has no corresponding mirror image fon the Pacific plate; and, because of the in: adequacy of suitable horspot models (Pilger and Handschumacher, 1981), its eastern ‘extent cannot be precisely estimated. How- ever, ifthe ridge owes its origin to processes which reflect motion of the Nazca plate relative to the mesosphere (such as a slowly ‘moving hotspot or intraplate stress due to drag), one might still use 2 hotspot model. 4st Use of the model should provide a fair esti mate of the trend, if not the length, of the subducted part of the ridge. Using the re- constructions of Pilger and Handschu- macher (1981), the estimated continuation of Juan Fernandez Ridge is presented in Figure 2. Note thatthe predicted continua- tion extends through the center ofthe Chi lean segment of low-angle subduction; the ‘older part is well beyond the easternmost epicenters of the segment (which approxi- mately correspond with the 150-km con tour of Isacks and Barazangi, 1977). GLOBAL NAZCA-SOUTH AMERICA PLATE RECONSTRUCTION “The predicted continuations of the sub- ducted parts of the Nazca and Juan Fer. nandez Ridges can provide predictions of the location of former segments of low. angle subduction, if the aseismic ridge hypothesis is correct, and if reconstructions cof Nazca and South American motion are available. These predictions can then be compared with independent evidence, such as voleanic gaps, which suggests low-angle subduction, Instantaneous plate-motion models (Minster and Jordan, 1978; Chase, 1978) provide approximations of plate motions, such as South American-Nazea plate con: vergence, which are probably representative for the past 5 m.y. For older motions i is TABLE 1. PARAMETERS FOR GLOBAL PLATE RECONSTRUCTIONS An La Long Ange Source An Lat Long “Angle” Source SOAM TO AFRC 'ANTA TO PCC 3 BIN ME 15° Chase (1978) 3 662N 8358-42" Chase (1978) He re tr 38 Ladd (1976) 5 687 797 =89 — Molnarand others (1975) ie cise aso) 76 interpolated 6 720 720-157 Molnar and others (1975) 7 398-350 318 interpoated 7 7s 649 202 interpolated aE 942 350) 104 interpolated 8 738 6.4 214 inerpolated Bsa “350 13S Ladd(1976) ia 27) “57.0 -=279 Molnar and others (1975) AERC TO INDI CFC TO NAZC 3 169 83 28 Chase (1978) 3 509 870 7.0 Chase (1978) sus Sis S44 jncerpolated 5 san “915 143 interpolated eee ta) 333, 108 incerpolaced 6 67 30's 30,1 interpolated 7 Bs 536 139 interpolated ae en =w98 391 Pilger (978) 8 BS 337 147 interpolated 5 639 =925 409 Pilger (1978) ba iss 540 192 interpolated 1B 7 10214 49.7 Pilger (1978) INDITO ANTA NAZC TO SOAM 3 321 Chase (1978) 3 489 864 38 Chase (1978) 5 107 316 Weiseleral(i977) 5593 106.1 86 This paper 6 BB M6 Weiseleral(1977) 6602 =972— -201 This paper 7 142 BS interpolated 7 S97 967 259 This paper e142 32 Weiseletal(i977) 8 «= «609— 100» =26. This paper Bg Baa Weiseleral(I977) 156491165 =32.5_—_ This paper Positive counterclockwise * Imerpolations berween anomaly-3 reconstruction and anomaly-16 reconstruction of Norton and Seater (1979) 482 necessary to combine the finite plate recon- structions derived by fitting corresponding magnetic lineations and fossil transform faults across spreading centers. This proce dure has been followed in deriving the rela tive positions of the Nazca and South ‘American plates for discrete anomaly iden tifiations (Fig. 3). The parameters com- bined to produce the global reconstructions include those of Ladd (1976) in the South Atlantic Ocean, Norton and Sclater (1979) and Weissel and others (1977) inthe Indian ‘Ocean, and Molnar and others (1975) and Pilger (1978) in the Pacific Ocean, as sum- marized in Table 1 Tn addition to the reconsteuctions (Fig. 3), it would also be desirable to produce an estimate of the errors in each reconstruc: tion. While most sources of uncertainties can be qualitatively identified, they are very difficult to quantify. Ideally, Gaussian statistics would be a natural accompani: ‘ment (0 the reconstruction parameters; the nonlinearity of the reconstructions and, more importantly, the lack of a physically meaningful criterion for lineation fitting, conspire to frustrate efforts at error qu: tification ina mode analogous to that of Minster and others (1974). Further, the absence of a consistent choice of anomalies for which reconstruc tions are made in each ocean necessitates in- terpolations in some ofthe oceans; the most straightforward method of interpolation involves constant roration parameters, an assumption which also lacks a physical basis. All of the two-plate reconstructions. involve subjective fiting criteria, and most are derived by trial-and-error methods. The tworplate system for which uncertainties are greatest appears to be India-Africa, evi- denced by an absence of identifications be- ‘ween anomalies $ and 21. Other uncer tainties which apply to these global recon- structions have been discussed by Atwater and Molnar (1973) and Molnar and Awa ter (1973). The reconstructions of Figure 3 are analogous t0 a density-contrast model derived to fit a Bouguer gravity anomaly; they represent one set of a family of so lutions, which can be tested against other, independent data, The reconstructions of Figure 3 suggest that, for about the past 25 m.y., con vergence between the Nazca and South American plates has been rapid, in an es ‘sentially east-west direction, with respect .0 South America, From 35 m.y. ago, or be- fore, until 25 m.y. ago, convergence oc curred in a northeast-southwest direction; with respect to the southern part of the RH. PILGER, JR. Peru-Chile trench (south of latitude 20°), a distinct right-lateral shear component existed prior (0 25 m.y. ago, with near normal convergence since. Conversely, near-normal convergence occurred along the northern Peru-Chile subduction zone (north of latitude 20°) until 25 m.y. ago, followed by convergence with a leftlateral ‘component, extending to the present. The change in convergence direction at 25 m.y. BP. corresponds with the major plate reor- ganization in the east-central Pacific, rec ognized by Handschumacher (1976); this reorganization is shown also in the diver- gent trends of the Nazca Ridge and the Easter-Sala y Gomez island-seamount chain, PLATE RECONSTRUCTION AND. VOLCANIC GAPS ‘The global plate reconstructions, com: bined with the estimates of the downdip continuations of the Nazca and Juan Fer- nandez Ridges (Fig. 2) provide estimates of the loci of intersection ofthe ridges with the Peru-Chile subduction zone (Fig. 4). These loci can then be compared with the avail- able isotopic ages of igneous rocks pre- sumed to have been generated as a result of subduction To test possible correspondence, isotopic ages of volcanic rocks in the Andes younger than 40 m.y. were compiled from published a sources and plotted on sections parallel with the Peru-Chile trench (Fig. 5). Unfor- tunately, the Andes are incompletely sam- pled to provide a readily interpretable his- tory of middle and late Cenozoic igneous activity. It is apparent, though, that very young dates are absent within contempor- ary segments of low-angle subduction; the absence of young dates is not surprising be- cause of the absence of active volcanoes in these segments. Ideally, if the empirical low-angle subduction-avoleanic correlation holds, and the predicted loci are approxi- mately correct, then there should be a cor- respondence between volcanic gaps and the intersection loci ‘The loci of intersection of the more northerly trending Nazca Ridge continua- tion (the anomaly 19 ‘mirror image") ap- 8 €0 corzespond best with the distribur tion of isotopic ages in Peru (Fig. SA), although data in northern Peru are clearly desirable to. provide information on the youngest volcanic rocks and, therefore, a better test of the hypothesis. Interestingly, a close correspondence between a volcanic gap in central and northern Chile and the Jean Fernandez loci is apparent. A lack of isotopic age data could be entirely respon- sible for the apparent gap and correlation, but it is interesting to note that the oldest part of the locus extends close to a “gap” in the most extensively dated segment of the Andes. The part ofthe locus that is between SOUTH AMERICA Ea 30 ae Figure 4. Reconstructed position of Nazca and Juan Fernandez Ridges (and their pre- dicted continuations) using parameters of Table 1. Figure $. Plots of isotopic ages of vol canic rocks 40 m.y. and younger in age, ver sus distance along Andes, and loci of inter section of predicted continuations of Nazca and Juan Femandez Ridges with crest of ‘Andes, fom Figure 4. A. Peruvian segment. B. Chilean segment. Data From: Baker (1977), Baker and Francis (1978), Caelles and others (1971), Clark and Farrar (1973), Clark and others (1967), Drake (1976), Evernden and others (1977), Farrar and others (1970), Grane and others (1979), Halpern (1973), Hormann and others (1973), Mortimer and Farrar (1974), Noble and others (1979), Slitoe (1977), and the compilation of Noble (1977). esewl ote 20s PREDICTED LOCUS, JUAN FERNANDEZ RIDGE-VOLCANIC ARC INTERSECTION © PACIFIC + 30° 454 10 and 15 my. in age extends across isotopic ages plotted in Figure SB, but a gap appears to exist above the predicted locus, prior to 15 my. BLP. Ifthe length of the pre-anomaly 7 part of the predicted con tinuation of the Juan Fernandez Island- seamount chain were greater, the recon- structed locus of intersection of the sub- ‘ducted part of the chain with the Andean ‘rest would correspond mote closely with the apparent gap in isotopic ages between latitude 22°S and 30°S, from approximately 27 10 16 my. BP. OTHER EFFECTS OF LOW-ANGLE SUBDUCTION, ‘Voleanie gaps represent the most obvious surface geologic effects associated with low-angle—subduction segments beneath the Andes. Several other phenomena may also be related to low-angle subduction, ‘These include lowered surlace elevations relative to the more normal arc, and land- ward shift of tectonic activity. Elsewhere, Cross and Pilger (19786) elaborated on the hypothesis that a transi- tion from steeper-angle subduction to low- angle subduction should be accompanied by a decrease in elevation of the land sur- face behind the former voleanic atc, due to displacement of low-density asthenosphere by higher-density oceanic lithosphere. Iso- static conditions would then requite surface subsidence in order to maintain equilib- rium, Cross and Pilger (1978) cited the RH, PILGER, JR. contrast between the Altiplano of the cen- tral Andes with the lower elevations of the Amazon Basin of central Peru, east of the Andean crest, s support for their hypothe- Barazani and Isacks (1976) postulated that subduction-related igneous activity reflects partial melting along the line of in- tersection of the top of the descending oceanic plate and the asthenosphere. A landward shift in igneous activity might be ‘expected due to a decrease in the subduc~ tion angle, in the context of Barazangi and Isacks’ (1976) model. In association with ‘igneous activity, increase in ductility of the adjacent continental ihosphere should also ‘occur. Indeed, to the east of the Andean crest, within the centeal Chilean segment of Tow-angle subduction, Stoll (1964) has de- scribed a reverse block faulted terrane of late Cenozoic age — the Pampean Ranges of Argentina (Fig. 6). From Stol’s (1964) description, the Pampean Ranges seem to bear a strong, resemblance to the Rocky Mountains of Colorado and Wyoming, which appear to represent compressional failure of western North America in the early Cenozoic (Burchfel and Davis, 19755 Brewer and others, 1980). The Pampean Ranges could then be a reflection of low- angle subduction. Several workers have in- ferred that low-angle subduction beneath North America could also have been re- sponsible for Laramide tectonics in the western United States in the early Cenozoic Burchfiel and Davis, 1975; Cross and Pilger, 1978a). In both low-angle subdue tion Segments beneath the Andes, high levels of crustal seismity and thrust-faulting focal mechanisms are well documented (tauder, 1973, 1975). The east flank of the Altiplano of the cen- tral Andes is characterized by numerous thrust faults of late Cenozoic and older age (Fig. 6), bur thei precise ages are unknown. Ie is unclear whether these faults formed under moderate- oF low-angle subduction conditions. The reconstructions of Figure 4 suggest that the possible northeastern con- tinuation of the Juan Fernandez Ridge may have interacted with the northern Chilean segment of the subduction zone in middle ‘Cenozoic time, which may correspond with the thrusting events of the central Andes. Clearly, a more complete knowledge of the structural history of the region is necessary before this hypothesis can be better tested The deep-focus earthquakes beneath the eastern flanks of the Andes (Stauder, 1973, 1975; Barazangi and Isacks, 1976) may also have an indirect connection to the in- teraction of aseismic ridges with the Peru Chile subduction zone. Two zones of deep earthquakes are recognized: (1) east of the central Peru low-angle~subduction sex- ‘meat (Fig. 1), and (2) east of the northern Chile segment of moderate angle subduc- tion. These segments could represent de- tached remnants of “normal” oceanic ltho- sphere, which was being subducted beneath adjacent parts of the Andes before the Nazea and Juan Fernandez Ridges began inzeracting with those same regions of the subduction zone. Iris significant that the lo- cation of each segment of deep focus earth- ‘quakes is adjacent to that part of the Andes which is, or has been, underlain by a low: angle subduction segment in the past 15 ‘my., according to the reconstructions of Figure 4. The absence of deep-focus earth: quakes along the southern Andes is believed to reflect a lack of prior subduction of aseismic ridges in that region, as well as the relative youth ofthe lithosphere undergoing, subduction SUMMARY AND CONCLUSIONS The results ofthis study, although temta- tive, are viewed as supportive of the hypothesis that subduction of the Nazca and Juan Femandet Ridges is responsible for the subduction occurring at very low angles. Models for the origin of the ridges are consistent with location of an extension — Figure 6, Tectonic features of the eastern flank of the ‘Andes, with 150-km contour on top of inclined seismic zone. ff the Nazca Ridge throughout the Peeu- vian segment of low-angle subduction. Plate reconstructions and available isotopic dates of volcanic rocks provide further support for the hypothesis. Indicect evidence of low-angle subduction, in the form of apparent voleanic gaps, correspond with crudely predicted loci of interaction between the aseismic ridges and the sub- duction zone. Further, the location of se ‘ments in which deep-focus earthquakes occur are consistent with a reconstructed model of aseismic-ridge—low-angle sub: duction zone interaction. Other geologic phenomena which may be related to low-angle subduction, including isostatic subsidence of the back-arc region, landward shift of tectonic activity, and fore land theust faulting, may provide better constraints on the history of segmented low-angle subduction. Conversely, refine ‘ment of the aseismic-ridge hypothesis, in- corporating better predictions of their down-dip continuations, may provide a powerful tool for tectonic interpretation of Cordilleran mountain belts, such as the Andes of South America. ACKNOWLEDGMENTS Antenor Aleman-Ramirez and Richard Rachel assisted in compiling and digitizing 'soropic age data from the Andes. Com- puter time and facilities were provided by the Department of Geology and the System Network Computer Center of Louisiana State University. Comments of Tom Ander- son, Muawia Barazangi, and Tim Cross on carlier versions of this manuscript were very helpful and appreciated. REFERENCES CITED Alens; T. Ju and Schubert, Gx 1975, Gabbro- ‘eclogite reaction rate and its geophysical Signiicance: Reviews of Geophyice and Space Physics v.13, p.383~400. Aleman-Ramizea, A. M.y and Pilgety R. H Jt 1978, Subducted aseimic ridges beneath South Americ; Evidence from seismicity, plate reconstructions, voleani history, and Focsposs fabs: EOS (American Geophys al Unson Transactions) ¥ 39, p- 383. Aumstiong, RL. Leeman, W. P.,and Malde, HUE, 1975, Kr dating, Quaternary and [Neogene volcanic rocks of the Snake River Plain, Idaho: American Journal of Science, ¥.275,p. 28-251, Arwater, T., and Menard, H. W., 1970, Magne ‘tc lincatons inthe northeast Pace: Earth fand Planetary Science Letters, v. 7, pa45-—a80. Agwater, T, and Molnar, P., 1973, Relative mo: tion of the Pacific and North Amencan plates deduced from seafloor spreading fhe Atlantic, Indian, and. South Pacific Oceans: Stanford University Publications in Geological Sciences, v.13, p. 136148. LOW-ANGLE SUBDUCTION BENEATH ANDES Baker, M.C.W,, 1977, Geochronology of the Upper Tertiary voleanic activity in the Andes of north Chile: Geologivehe Rund Schats, ¥- 66, p. 455~465. Baker, MGW. and Francis, P.W, 1978, Upper Cenozoic voleanim in the central Andes — [Ages and volumes: Earth and Planetary Se: cence Lewes, v.41, p. 175-187 Barazangi, M. and Isic, B. L, 1976, Spatial distribution of earthquakes and subduction ‘of the Nazca plate beneath South America Geology, v4, p 686-692, 1575, Sbducton of the Nazca plate be- neath Peru: Evidence from spatial istib tion of earthquakes: Geophysical Journal of the Royal Astronomics! Society, ¥- 87, p.537-855, Berggren, W. A, McKenna, M. C., Hardenbol J, and Obradovich, J. D, 1978, Revised Paleogene polarity time scale: Journal of Geology, v.86, p. 67-82. Brewer, J. Ay Smithson, R. B., Oliver, J. Ey ‘Kautinan, 8, and Brown, L:D., 1980, The Laramide Orogeny: Evidence from COCORP deep crustal seismic profiles in the Wind River, Mountains, Wyoming. “Teeronophyscs, . 62, p. 165-189. Burchfel, B.C, and Davis, G. A, 1975, Nature and controls of Cordilleran orogeness wwestem United States: Extensions of an flier synthesis: American Journal of Se fences, ve 27504, p 363-396, Caley, J.-C. Clarks A. HL, Fareat, Edward, McBride, 8. Ly and Quiet, Stewart, 1971, Potassiumargan ages of porphyry copper

You might also like