Abstract

The Precordillera thrust belt of western Argentina is anomalously close, both horizontally and vertically, to the coeval subduction zone of the Nazca plate. The thin-skinned part of the belt has an unusually deep décollement that is well defined by industry seismic reflection and recent broadband experiments. New area and line-length balanced cross sections show that the central Precordillera has accrued ∼90 ± 21 km of shortening since 13 Ma; much of that shortening occurred between 12 and 9 Ma. Fault-slip data generally show shortening approximately west-northwest–east-southeast, orthogonal to the traces of the thrust and folds in the Precordillera and oblique to the mean vector of local global positioning system (GPS) data. The GPS strain rate is –63 ± 9 × 10–9/yr, whereas strain rate in the thrust belt, averaged over 13 m.y., is –56 ± 4 × 10–9/yr. Although the décollement of the Precordillera cannot cut into Paleozoic Cuyania(?) terrane basement east of the crest of the high Andes, broadband receiver function data show that significant crustal thickening must occur beneath and even east of the thrust belt. We suggest that top-to-the-west shear and thickening of the lower crust due to flat subduction explains the distribution of crustal thickening.

INTRODUCTION

Most retroarc foreland thrust belts, such as the Bolivian Subandean belt or the Mesozoic–early Cenozoic thrust belt of western North America, are 600 km or more inland from the coeval trench and 400–600 km above the subducted plate. This inboard position raises important scientific questions, including the nature of the driving mechanism of a belt so distant from the trench, but it has a useful practical benefit: the broad swath of hinterland allows the luxury of constructing balanced cross sections without worrying about exactly where lower crustal shortening commences, how the transition from upper crustal to whole crust shortening occurs, or even how the thrust plates will restore relative to the trench.

The Precordillera thrust belt of western Argentina, in contrast, is located 350 km from the Chile Trench and just 100 km above the subducted Nazca plate. Thus the amount of crust to work with when attempting to balance the shortening in the thrust belt is significantly less than elsewhere, raising questions about how and where the shortening observed at the surface is accommodated at depth. The first balanced section in the Precordillera (Allmendinger et al., 1990) attempted to address this problem. Since that time, a great deal of new field and geophysical data (e.g., Jordan et al., 1993, 2001; Zapata and Allmendinger, 1996a; Pardo et al., 2002; Brooks et al., 2003; Gans et al., 2011; Judge, 2012) for the region have become available and new structural algorithms allow us to specify the uncertainties inherent in balanced sections (Judge and Allmendinger, 2011; Allmendinger and Judge, 2013). With a new generation of Precordillera studies underway (e.g., Fosdick and Carrapa, 2012), it is timely to reexamine the question of shortening in the Precordillera.

In this paper we present new field data and balanced cross sections of the Precordillera between lat 30°S and 30.5°S. Our fault-slip data demonstrate that the Miocene to Holocene history of this part of the Precordillera is characterized by thrust faulting and shortening that deviates by as much as 40° from the mean vector of global positioning system (GPS) geodetic data. The shortening values from balanced sections, similar to those previously determined, yield a yearly average strain rate that is indistinguishable from the GPS strain rate. The observation that crustal thickening extends east of the deformation front of the Precordillera thin-skinned belt requires top-to-the-west simple shear and thickening in the lower crust, much as described by Bird (1988) for the Laramide Rocky Mountain foreland of the western United States.

TECTONIC AND GEOLOGIC SETTING

The Argentine Precordillera overlies a region of flat subduction of the Nazca plate (Cahill and Isacks, 1992; Gans et al., 2011) located at the southern end of the Central Andes (Fig. 1). This region of flat subduction has been linked to the subduction of the Juan Fernández Ridge, which, because of a dogleg in its now subducted trace, swept southward along the South American margin from 22 to 10 Ma. However, since 10 Ma the segment entering the trench is nearly parallel to the convergence direction, resulting in a stable configuration since then (Yáñez et al., 2002). The link between ridge subduction and the flat geometry appears to be supported by anomalously high frequency of seismicity in the subducted plate aligned with the ridge (Pardo et al., 2002; Gans et al., 2011). Progressive enrichment of arc magmatic rocks indicates that the main phase of shallowing of the subducted plate occurred between 10 and 5 Ma (Kay and Abbruzzi, 1996), in broad agreement with the history of subduction of the Juan Fernández Ridge.

Study of seafloor magnetic lineations, global plate circuits, and GPS geodesy has shown that the convergence rate at the plate boundary has decreased by a factor of 2 in the past 15 m.y. (Pardo-Casas and Molnar, 1987; Somoza, 1998; Angermann et al., 1999; Kendrick et al., 2003). Currently, convergence is ∼63 mm/yr in a direction 079.5° at the latitude of the Precordillera. This convergence produces GPS measurable displacements of ∼10 mm/yr with respect to stable South America in the central Precordillera, which is thought to be due to elastic deformation from a combination of locking of the interplate subduction zone and locking of the Precordillera décollement (Brooks et al., 2003). We discuss the relationship between long-term and short-term upper crustal shortening toward the end of this paper.

A considerable amount of geophysical and geological information about the Precordillera and westernmost Sierras Pampeanas is now available (Fig. 1). Campaign-style GPS measurements are available from the Central Andes Project (Brooks et al., 2003). Early local seismology networks focused on the region around San Juan city in the aftermath of the 1977 Caucete earthquake (Kadinsky-Cade et al., 1985; Smalley and Isacks, 1987; Regnier et al., 1992, 1994; Smalley et al., 1993). More recently, the region has seen two significant broadband seismograph deployments, the 2000–2002 CHARGE (Chile-Argentina Geophysical Experiment) (e.g., Alvarado et al., 2005) and the 2007–2009 SIEMBRA (Sierras Pampeanas Experiment Using a Multicomponent Broadband Array) (Gans et al., 2011). Several geological studies have incorporated seismic reflection data from the Yacimientos Petrolíferos Fiscales (YPF; Allmendinger et al., 1990; Beer et al., 1990; Zapata and Allmendinger, 1996a, 1996b; Zapata, 1998).

Previous geological studies concentrated on both the foreland and intermontane basin stratigraphy and structural geology of the belt (Furque, 1979, 1983; Ortíz and Zambrano, 1981; Ramos et al., 1984, 2002; Johnson et al., 1986; Allmendinger et al., 1990; von Gosen, 1992, 1995; Jordan et al., 1993, 2001; Zapata and Allmendinger, 1996a, 1996b; Siame et al., 1997, 2002, 2005; Colombo et al., 2000; Alvarez-Marrón et al., 2006; Vergés et al., 2007; Fosdick and Carrapa, 2012). Perhaps most germane to the current study is the thrust timing in the Precordillera established by Jordan et al. (1993, 2001); they showed that the Precordillera thrust belt east of the Iglesia Basin initiated between 21 and 19 Ma with progressive eastward migration of the thrust front through time and abundant evidence of simultaneous and out-of-sequence thrust motion.

STRUCTURAL GEOLOGY OF THE PRECORDILLERA BETWEEN JÁCHAL AND GUALILÁN

The Precordillera thrust belt is built on a foundation of a Paleozoic terrane, Cuyania, accreted to South America prior to the start of the Jurassic to present Andean orogeny (Ramos et al., 1986, 2002; Ramos, 2008). The central part of the Precordillera has a layered sequence of Cambrian to Permian strata dominated by the Cambrian–Ordovician San Juan Limestone found at the base of many of the thrust plates. The remainder of the succession is composed of siliciclastic rocks. Several low-angle unconformities within the Paleozoic section occur throughout the belt, and pre-Andean deformation becomes increasingly important to the west. In the westernmost part of the Precordillera, the Ordovician changes facies to slope and basinal flysch with pillow basalts and ultramafic rocks that signal the allochthonous boundary between the Cuyania and Chilenia terranes (Ramos et al., 1986). Thrusts within these rocks are impossible to balance due to completely unknown initial thickness of the deposits and significant pre-Andean folding. South of ∼31°S, the Precordillera is broken up by Triassic grabens (Ramos and Kay, 1991), but those structures are not present in our field area.

To the east, the eastern Precordillera is composed of thick-skinned, west-verging structures more reminiscent of the neighboring Sierras Pampeanas (Ortíz and Zambrano, 1981; Zapata and Allmendinger, 1996b). Deep crustal seismicity beneath the eastern Precordillera confirms its thick-skinned nature (Smalley et al., 1993). To the west, the Iglesia Basin (Fig. 1) has several structures with significant strike-slip components, including the well-known El Tigre fault (Bastías and Bastías, 1987; Siame et al., 1997, 2002) and local features visible on seismic reflection data that resemble flower structures (Alvarez-Marrón et al., 2006). As shown in the following, strike-slip faulting is not significant farther east. Thus, our study focuses primarily on the thrust plates of the central Precordillera, though we also present fault slip data and discuss GPS data from the neighboring areas.

Major Structures of the Central Precordillera

The main thin-skinned thrusts of the Precordillera between the Río Jáchal and the Cienega de Gualilán are, from east to west (foreland to hinterland), Niquivil, San Roque, Blanquitos, Blanco, Caracol East, Caracol West, and Tranca (Figs. 2 and 3). We first describe the surface geology and structures of these thrust plates, then address the upper crustal geometry and shortening.

Niquivil Plate

The Niquivil thrust constitutes the leading edge of the thin-skinned belt at these latitudes and has been active for about the past 5 m.y. (Jordan et al., 1993, 2001). The thrust’s ongoing activity is demonstrated by a 10–15-m-high fault scarp where its frontal trace is crossed by the Río Jáchal at the village of Niquivil (Fig. 4). The fault plane is not exposed anywhere, but the probable surface trace of the Niquivil thrust is ∼50 km in length. To the north, the thrust dies out into the Cuesta de Huaco fault-propagation fold and to the south it terminates at the Río Francia tear fault–transfer zone. In Zapata and Allmendinger (1996b) it was reported that the Niquivil thrust plate is cut and deformed by the thick-skinned fault coring the Niquivil anticline that is just to the east, locally reversing the vergence of the Cuesta de Huaco anticline.

The base of the Niquivil thrust plate contains the thickest exposures of the Ordovician San Juan Limestone of any plate in this segment of the Precordillera. The thickness of the remainder of the Paleozoic section is highly variable: at Cuesta de Huaco, no Silurian or Devonian strata are present and the Carboniferous and Permian are unconformable on the Ordovician. Due east of the town of Jáchal, thin remnants of Silurian Los Espejos and Devonian Talacasto Formations appear beneath the late Paleozoic unconformity. At the south end of the Niquivil plate, just north of the Río Francia, Silurian strata are present but the Devonian section is missing (Fig. 3; Plate 1). As described in the following, the Niquivil thrust is the only one crossed by YPF seismic lines (Allmendinger et al., 1990; Zapata and Allmendinger, 1996b). Although the fault plane is not exposed anywhere, interpretation of the YPF seismic lines indicates that the 35° west-dipping thrust places Ordovician limestone over Miocene sandstone, with a stratigraphic throw of 12–15 km.

San Roque Plate

The San Roque thrust trace is at least 120 km long; it probably extends from the Guandacol area in the north (outside our map area) to the Cienega de Gualilán in the south. It was active from 10 or 9 Ma to 3 or 2 Ma (Jordan et al., 1993, 2001). The fault has a significant lateral ramp just west of the village of Niquivil: to the north, the fault is within the San Juan Limestone but to the south it steps upsection to within the Silurian Los Espejos Formation. Both north and south of the hanging-wall lateral ramp, the footwall strata are Miocene sandstones. There is a duplex at the base of the thrust plate, 10 km south of the lateral ramp, and a thin sliver of Ordovician limestone is present at the base of the plate. The San Roque fault and its hanging wall go through an 80° bend and appear to have ∼200 m of separation across the Río Francia tear fault, which dies out into the San Roque plate. This suggests that at least some movement on the Niquivil thrust postdates the San Roque thrust.

The San Roque plate contains a thinner section of San Juan Limestone than the Niquivil plate. North of the Río Jáchal, the rest of the lower Paleozoic section is completely missing and the upper Paleozoic directly overlies the limestone. To the south, a thicker Silurian and substantial Devonian section overlie the limestone, with a thin upper Paleozoic sequence unconformably overlying the Devonian. On the western flank of the San Roque range proper, a series of small thrusts and tight folds thicken the Silurian and Devonian section with minor involvement of the Carboniferous. Significant folding of the Devonian section is also prevalent farther south, complicating the task of determining original stratigraphic thickness.

Blanquitos Plate

The Blanquitos plate, the shortest and most enigmatic in this segment of the Precordillera, underwent a brief period of activity between 11.5 and 9.5 Ma (Jordan et al., 1993, 2001). Its surface trace is <50 km long and along most of that length Devonian strata are at the base of the upper plate. At 30.37°S, there is a lateral ramp and the thrust cuts downsection northward to include a thin sliver of San Juan Limestone at the base of the upper plate, with Miocene sandstone in the lower plate. The Cerro Bayo anticline (Fig. 3; Plate 1) generated by this lateral ramp propagates across the entire upper plate and appears to plunge beneath the Blanco thrust to the west. Less than 10 km north from where the limestone first appears at the base of the upper plate, it disappears in a north-plunging anticline. Due to extensive Quaternary cover north of this point, it is unclear whether the anticline represents a north-plunging tip-line fold or if the fault continues northward and the fold represents a hanging-wall lateral ramp where the fault once again steps upsection northward into the Silurian. There are excellent exposures of the main thrust plane where the limestone overlies the Tertiary on the backlimb of the San Roque plate. The fault in these exposures has a south to south-southwest strike and dip that varies between 30°W and 55°W.

South of the lateral ramp, 1800–2700 m of Devonian Punta Negra Formation are overlain directly by a thin sequence of Cenozoic foreland basin strata, including a distinctive eolian cross-bedded sandstone, which overlies a redbed with a tuff dated as 21.6 ± 0.8 Ma, found through much of the region (Jordan et al., 1993; Milana, 1993). To the north of the lateral ramp and Cerro Bayo anticline, the Silurian and Devonian form a broad expanse of strata folded at multiple wavelengths, including spectacular outcrop-scale kink folds that can be seen on the main road between Jáchal and the village of Rodeo in the Iglesia Basin west of the Precordillera.

Blanco Plate

The highest relief in the region, including the highest peak in this segment of the Precordillera at more than 3600 m elevation, occurs in the upper plate of the Blanco thrust. The Blanco thrust was active from 13 to 9 Ma, overlapping activity on the Blanquitos thrust (Jordan et al., 1993, 2001). Although the Blanco thrust dies out within our map area at ∼30.3°S, it is undoubtedly one of most important thrust faults in the central Precordillera and can be traced south to the Río San Juan for a total map length of ∼120 km. Throughout most of that trace length, San Juan Limestone is found at the base of the plate. Between the Cienega de Gualilán and the Río Jáchal, the Blanco plate contains a tight syncline in thick Silurian and Devonian strata. On both limbs of the syncline, dips are steep, ranging from 40° to 80°. This fold, named Mogotes Quebraditas syncline (Fig. 3; Plate 1), is genetically related to the Blanco and the Caracol East thrust located on the west flank of the syncline, and its significance is discussed in the following section.

Just north of the Cienega de Gualilán at the Cerro Portezuelo Blanco (Fig. 3), the Blanco thrust overrides and cuts downsection eastward across the back limb of the Blanquitos plate. Three-point calculations show that the fault plane has a dip of <15°W here, whereas the bedding in the footwall dips 30°–40°W. This geometric relationship indicates that at least the youngest motion on the Blanco thrust is out of sequence and postdates the tilting of the strata in the Blanquitos plate.

At the northern end of the Blanco plate, the Ordovician limestone and overlying Silurian and Devonian sequence are deformed into an overturned fault-propagation fold. The forelimb of this fold is complexly imbricated with the limestone thrust over the Silurian–Devonian, which is in turn thrust over the Cenozoic of the Blanco valley. The Blanco fault ramps into the Devonian in the hanging wall and continues for another 8–10 km northward before dying out completely into an anticline separating two large synclines. The exposures of the Silurian and Devonian along the Río Jáchal belong to the combined upper plate of the Blanquitos and Blanco thrust.

Caracol East Thrust

The next fault to the west is an east-dipping, west-verging structure we refer to as the Caracol East thrust. This structure tracks the western limb of the Mogotes Quebraditas syncline. On the eastern side of the Caracol Valley, thin, highly deformed slivers of San Juan Limestone are present locally along the base of the thrust plate. Farther south in the Cordón del Peñon (Fig. 3) more extensive outcrops of limestone and overlying Silurian rocks are tightly folded into anticlines and synclines with westward vergence. The upper plate of Caracol East is also the upper plate of the Blanco thrust, and the tight syncline between the two thrust traces suggests that Caracol East is a back thrust off the Blanco thrust. We hypothesize that Blanco–Caracol East formed a triangle zone early in the history of the thrust belt before it was subsequently deformed and moved again out of sequence.

Caracol West and Tranca Thrust Plates

A west-dipping thrust occurs on the west side of the Caracol Valley (Fig. 5), and it is initially tempting to interpret the Caracol East and Caracol West faults as a continuous fault forming a fenster into the lower plate (e.g., Allmendinger et al., 1990). However, the rocks of the two upper plates are entirely different; the Caracol East plate is dominated by Silurian and Devonian strata, whereas the Caracol West plate contains exposures of the Ordovician Yerba Loca Formation, a deep-water flysch deposit. Therefore the two faults are unlikely to be the same.

We consider the Caracol West and the Tranca thrusts farther west to be a single en echelon system that carries the western deep-water facies of the Ordovician in its upper plate. The Caracol West thrust dies out southward and the Tranca thrust dies out northward at the Río Jáchal with an overlap of ∼22 km immediately south of the river. Along the Río Jáchal, a few highly deformed slivers of limestone are imbricated in the Yerba Loca, but near the Cuesta del Viento (Fig. 3; Plate 1) the Ordovician strata are tightly deformed into pre-Andean west-verging overturned folds associated with mafic and ultramafic igneous rocks.

The age relations of the Caracol and Tranca thrusts were described by Jordan et al. (1993, 2001). The Tranca thrust definitively had activity prior to 19 Ma, and the Caracol West possibly had activity in the same time frame. Both members of the en echelon system also display younger deformation; in the case of the Tranca thrust, it is a subsidiary structure that placed ca. 21 Ma redbeds over the Chestnut Conglomerate (Jordan et al., 1993, 2001), and for the Caracol West, the outcrops near the Río Jáchal show thrusting over mid-Miocene eolian beds. In summary, initial motion on the Tranca and Caracol West thrusts, which carry Ordovician Yerba Loca Formation in their upper plates, is ∼6–7 m.y. older than the other thrusts in this segment of the Precordillera. The Blanco thrust and the more eastern thrusts are all younger than 13 Ma.

Fault-Slip Data

Siame et al. (2005, 2006) presented fault-slip data for the Jáchal segment of the Precordillera; they suggested that the predominant fault population shows approximately horizontal east-northeast–oriented maximum principal stress, σ1, though a secondary, and in their interpretation, older, population that records northwest-southeast–oriented horizontal σ1. The criteria used to separate their data into older and younger σ1 directions are unclear. A remarkable aspect of their data set is that the primary, younger σ1 is not orthogonal to the strikes of the thrust faults or the trends of the fold axes, but makes an angle of as much as 45° to the structures, even though there is little evidence for strike slip or oblique slip in their data set.

Because the Siame et al. (2005, 2006) data appear to have been collected primarily along the roads and more than 50% of their measurements are in Paleozoic bedrock, we collected a more extensive data set (Judge, 2012) that we know to be completely of late Cenozoic age, as all of our faults are located within Cenozoic strata or on faults that place Paleozoic over Tertiary (see the Supplemental Table1). In addition, we prefer an analysis in terms of infinitesimal strain (Marrett and Allmendinger, 1990) rather than stress. The strain from multiple faults sets is cumulative (additive in the case of infinitesimal deformation and multiplicative for finite strain; Cladouhos and Allmendinger, 1993), whereas stress is instantaneous only, and thus one must assume that all the faults in a data set formed at the same time or that the same homogeneous stress state persisted for the duration of the faulting.

We plot the P-axes and T-axes (pressure and tension) for individual faults associated with each major structure (Fig. 6) and summarize each data set with a composite fault-plane solution calculated using a moment tensor sum where all the faults are assumed to have the same weight (Marrett and Allmendinger, 1990). P-axes and T-axes, despite their names, are simply the infinitesimal strain axes for the faults. In detail, individual data sets show a broad distribution of individual P-axes and T-axes, but taken at a broader scale, two fundamental observations emerge from our fault-slip data.

First, there is very little evidence for significant strike-slip faulting in the Neogene rocks of the Precordillera, except for the Niquivil anticline of the Eastern Precordillera (Fig. 7). Immediately to the west between the Iglesia Basin and the Precordillera, the El Tigre fault (Fig. 3) has documented Quaternary right-lateral displacement (Bastías and Bastías, 1987; Siame et al., 1997), although this fault dies out just south of the Río Jáchal. Thus, there must be a partitioning of displacements very much as described by Siame et al. (2005). Alvarez-Marrón et al. (2006) proposed that the thrust faulting in the central Precordillera was Paleozoic in age and that Neogene deformation is characterized by minor high-angle strike-slip fault dismemberment of the ancient thrust belt. Our fault slip data provide no evidence for this hypothesis; instead they are completely consistent with Neogene horizontal shortening on dip-slip faults. Alvarez-Marrón et al. (2006) also ignored the evidence from seismic reflection data across the frontal thrust fault (Allmendinger et al., 1990; Zapata and Allmendinger, 1996a, 1996b) and seem not to have considered the possibility that, in an imbricate stack of thrusts such as the Precordillera, older thrusts are rotated to a high angle by younger more eastern thrusts. Alvarez-Marrón et al. (2006) are correct that considerable pre-Andean deformation in the Paleozoic rocks of the Precordillera is indicated by numerous unconformities in the Paleozoic section and by tight folding of the Paleozoic rocks that are unconformably beneath the Neogene strata.

The second major observation from our fault-slip data is that shortening everywhere within the belt is mostly orthogonal to the primary map-scale structures (i.e., strikes of thrust faults, trends of fold axes; Fig. 8). For example, data from near or in the Blanquitos fault zone yield a west-dipping nodal plane that strikes 020°, the local strike of the thrust fault. Conjugate strike-slip faults from the Niquivil anticline of the eastern Precordillera (Fig. 7) yield a shortening axis that is orthogonal to the fold axis (Figs. 6 and 8). To the south at the Río Francia tear fault, shortening axes are anomalous with respect to the rest of the belt, but then so is the local strike of the Francia thrust.

Comparison to GPS Geodesy

The orthogonal shortening is important because the strikes and trends of the structures vary by 70° or so, even though GPS vectors in this part of the Precordillera (Brooks et al., 2003) have a consistent mean orientation of 075° (Figs. 1 and 8). To compare GPS data to deformation features of the Earth’s crust, however, one must use the gradient of the GPS velocity field rather than the velocity field (Allmendinger et al., 2009). We do that using the program SSPX by Cardozo and Allmendinger (2009) and the most recently available GPS data for the region (Brooks et al., 2003). The results presented here could change when the same analyses are applied to updated data, which are still being processed (B. Brooks, 2013, personal commun.). There are several ways to calculate strain from GPS data (Allmendinger et al., 2009); we use two, distance weighting and a simpler nearest neighbor approach, which calculates a single best-fit strain rate ellipse to the group of GPS stations closest to our study area.

Regional smoothed two-dimensional (2D) principal horizontal shortening strain-rate axes, using a distance-weighted algorithm that smooths out the strain-rate solution at long wavelengths (Allmendinger et al., 2009), are rotated ∼5° clockwise from the mean GPS vector in the Jáchal Precordillera (Fig. 8). With an orientation of 080°, the angle between GPS shortening and the north-northeast–striking structures in many parts of the belt is only ∼60° (Fig. 1). Because our fault-slip data sets tend to be diverse, we could arbitrarily select subsets of faults that will give shortening parallel to the modern-day GPS vectors or the distance-weighted shortening axes, but we have no objective basis for doing so.

The best-fit horizontal strain ellipse to the eight stations nearest our studied segment of the Precordillera, assuming that strain is homogeneous, yields a principal horizontal shortening strain rate of –63 ± 9 × 10–9/yr in the direction 093° ± 8° (Fig. 8). The GPS-derived horizontal maximum extension value is more than an order of magnitude lower than the concomitant shortening, suggesting that modern-day deformation is approximately plane strain with no significant active strike-slip faulting in the area covered by the GPS stations. The shortening value obtained in this 2D calculation is essentially the same as that yielded by a 1D transect using the stations between PAGN and AT30, located on the Sierra de Valle Fértil (Fig. 9).

The orientation of the 2D nearest-neighbor GPS shortening, using just the stations closest to our study area, is rotated clockwise by nearly 20° from the mean GPS vector, a result produced by north-south gradients in the GPS velocity vectors. These north-south gradients are produced in part by subduction that deviates by ∼20° from perpendicular (e.g., Bevis et al., 2001) but are probably also due to variations in coupling on both the subduction megathrust and the Precordillera décollement. Nonetheless, the GPS shortening direction is still not orthogonal to many of the eastern structures, including the San Roque and Niquivil thrusts and the folds of the eastern Precordillera in the Sierras de Huaco, which have strikes or trends of 025°, and thus one would expect shortening azimuths to be ∼115° (Fig. 8). At an azimuth of 093° ± 8°, the 2D GPS shortening rate is also not parallel to fault-slip shortening orientations. It is possible that more dense GPS data would better resolve shortening directions.

Assuming that GPS data are representative of the present-day strain rate field, one must postulate that significant vertical axis rotation has occurred since the major structures and the minor faults we have measured formed. Currently available paleomagnetic data were collected in the Sierras de Huaco for magnetic reversal stratigraphy (Johnson et al., 1986; Beer and Jordan, 1989; Beer, 1990) and, because they were based on oriented cubes collected just for polarity, are not very reliable for vertical axis rotations. Thus, we await the paleomagnetic resampling of the Miocene and Pliocene rocks of the region for final resolution of this conundrum.

Shortening Magnitudes from Balanced Sections

The magnitude of shortening in this segment of the Central Andes is of primary interest, given its proximity to the subducted plate. We estimate horizontal shortening in two ways (Figs. 10 and 11): the first is via classic line-length balancing and the second, using area balancing, allows us to estimate the uncertainties of our calculations.

The depth to décollement is a key factor for both types of balancing. On the east side of the belt, we use the value derived from industry seismic reflection data that yields a current depth at the eastern side of the belt of ∼13.7 km below sea level (described in Allmendinger et al., 1990; Zapata and Allmendinger, 1996a, 1996b). This depth puts the décollement close to the top of the very bright positive polarity mid-crustal converter imaged in the receiver function stacks of Gans et al. (2011). Although this converter extends east of the thrust front, we interpret it as the top of Cuyania basement and the de facto décollement (Fig. 12). This converter continues west, uninterrupted, to at least 70°W. The décollement is interpreted on our sections to be at ∼16 km below sea level beneath the Iglesia Basin. The current, unusually deep position of the décollement reflects the magnitude of shortening and foreland basin strata accumulation; at the start of deformation the stratigraphic level of the décollement would have been just 5–8 km deep. The low metamorphic grade of the rocks at the base of each thrust sheet is probably indicative of this initial depth rather than the final depth. In the western Precordillera, the rocks in the thrust sheets are incipient greenschist facies, though that metamorphism could have long predated the Neogene thrusting.

Line-Length Balanced Sections

The two line-length sections (Figs. 10 and 11) have several basic features in common. Based on the seismic reflection data and on the array of back-limb dips, it appears that the initial cut up angle for the belt is steep, between 30° and 40°. This value makes it impossible to preserve bedding thickness on the forelimbs of fault-bend folds (Suppe, 1983), but that is of little concern, because most folds in the belt are better interpreted as fault-propagation folds and outcrops show significant thinning and thickening of fold forelimbs.

Because of pre-Andean deformation, most stratigraphic units change thickness abruptly from one plate to the next and even from north to south within a single plate, as described herein. Thus, the balancing is based on the Cambrian–Ordovician San Juan Limestone, and only includes from the footwall of the Niquivil thrust to the footwall of the Caracol thrust to the west. The Caracol West and Tranca thrust plates cannot be balanced because the thickness of the Yerba Loca Formation is not well known, and these plates contain no markers of known orientation at the start of Andean deformation that can be correlated with the thrust plates farther east. Because the Tranca and probably the Caracol West plates moved several million years before the other plates (according to the chronology described in Jordan et al., 1993, 2001), our shortening values apply to the past 13 m.y. only.

We have no illusions that the geometries shown in sections A-A′ and B-B′ are anything more than plausible. Nonetheless, we note a few key geometric elements. First, the easternmost Niquivil thrust contains the thickest section of the Cambrian–Ordovician limestone and the décollement is at the base of the limestone. Therefore all of the thrusts to the west should have hanging-wall ramps at depth across the unexposed parts of the section producing thickening in the subsurface; this helps to fill the space down to the unusually deep décollement. This geometry was not recognized in Allmendinger et al. (1990) and that error is rectified here. Second, the Blanco and Caracol East thrust are linked; we have interpreted them to have initially formed a triangle zone sometime between 13 and 9 Ma. Third, the most recent movement on the Blanco thrust was out of sequence and younger than that on the Blanquitos thrust, because the former locally cuts downsection across the latter. Taking into account point two, after the triangle zone was formed, the Blanquitos thrust moved and then the Blanco thrust broke through from the blind tip of the triangle zone and cut downsection across the Blanquitos sheet. Chronology that is independent of the geometrical arguments (Jordan et al., 1993, 2001) clearly permits this interpretation.

Both balanced sections yield similar line-length shortening values: The leading edge of the Caracol West plate must restore to 95 km west of its current position on A-A′ and to 99 km west on B-B′; this gives us 72% shortening on A-A′ and 77% on B-B′. These large values are similar to those obtained in Allmendinger et al. (1990) and fundamentally arise because the unusually deep décollement requires a tripling of the stratigraphic section to fill the space.

Area Balance and Uncertainties

To carry out the area balance we use methods that allow us to assess uncertainties (Bird, 1988; Judge and Allmendinger, 2011). The balanced polygons with their error bars are shown in Figures 10 and 11. We use a uniform initial stratigraphic wedge of just Ordovician strata because the stratigraphy of the Precordillera is not sufficiently well known to allow us to define a nonuniform taper. The greatest unknowns are the horizontal position of the western edge of the deformed polygon on the décollement (we somewhat arbitrarily assign ±5 km to that point) and the thickness of the Ordovician Invernada and Yerba Loca Formations on the west side of the initial wedge. We assign an uncertainty of 25% of the thickness, although this is admittedly little more than a guess. We report Gaussian errors, based on the square root of the sum of the squares, in the following analysis; maximum errors are much larger (Taylor, 1997; Bevington and Robinson, 2003).

Determination of shortening magnitude and percent depends on the relationship between the deformed polygon and the initial wedge. To make our area balance as close as possible to our line-length balance, we use the same deformed width. Thus, cross-section A-A′ has a shortening magnitude of 88 ± 22.5 km and a shortening of 71% ± 5.4%, very close to the line-length balance of that section. Section B-B′ yields a shortening magnitude of 84 ± 21 km and a shortening of 73% ± 5%. The line-length shortening for B-B′ is within the error limits from the area balance.

If we assume that the central Precordillera thrusts balanced here were active for the past 13 m.y., then the yearly average strain rate is 56 ± 4 × 10–9/yr; recall that the 2D GPS strain calculation gives 63 ± 9 × 10–9/yr over the geologically instantaneous time span of a decade. Thus, we can say that the rate of strain accumulation today in the Precordillera matches the average rate of strain accumulation over the past 13 m.y. In detail, of course, over geological time the shortening rate (Fig. 13), and thus the strain rate, is not constant. Furthermore, the GPS strain is thought to be due to elastic deformation due to locking of both the subduction zone and locking of the Precordillera décollement (Brooks et al., 2003). Given the 50% decrease in plate convergence rate over the past 15 m.y. (Kendrick et al., 2003), the instantaneous strain rate has probably not been constant. Our result, though, suggests that elastic strain that accumulates in the Precordillera ultimately stays in the Precordillera when it is converted into permanent deformation across numerous earthquake cycles.

CRUSTAL SHORTENING OF THE ANDEAN FORELAND AT 30°S

The cross-section balancing demonstrates that the décollement of the thrust belt continues westward at least ∼115 ± 21 km from the thrust front of the central Precordillera without involving basement, a position beneath or west of the crest of the high Andes (Fig. 12). This raises a space problem because, at the west end of the décollement, the South American crust is already thinning toward the anomalously close plate boundary and no obvious structural geometry exists for transferring the upper crustal shortening into lower crustal thickening. Furthermore, broadband geophysical experiments (Gans et al., 2011) show that the thickest crust in this segment is not underneath the high topography, but instead is beneath the western margin of the Iglesia Basin. The crust is >60 km thick east of the deformation front of the thin-skinned thrust belt. How does the substantial surface shortening in the Argentine Precordillera produce shortening and thickening at deeper levels in the crust?

Allmendinger et al. (1990) noted the same problem: with much poorer data, they attempted some fanciful geometries (triangles zones and duplexes) to solve the problem. Here we do not attempt to provide specific structural geometries, as even today’s data are not sufficient to define the lower crustal deformation style. Instead, the simplified deformation across this segment of the Andes can be visualized as three schematic, horizontal particle velocity profiles depicting the coeval flow of rocks in different parts of the belt (Fig. 14). The weak lower crust is probably advected eastward during flat subduction, much as described by Bird (1988) for the Laramide deformation of the western United States. Conceivably, part of the lower crustal material could be tectonically eroded from the leading edge of South America (Goss et al., 2013).

In all three of our particle velocity profiles, the strong mantle lid is simply translated eastward (Bird, 1988). The eastward flow of the lowermost crust diminishes gradually eastward but extends well into the Sierras Pampeanas. Beneath the Precordillera, the middle crust under the décollement has a velocity close to zero with respect to South America. Beneath the High Andes, the crust near the western end of the décollement probably behaves more like a ductile shear zone than a discontinuity, and the middle crust below has some finite eastward velocity. The horizontal velocity gradient in the middle crust between the high Andes and the Precordillera produces shortening in the middle crust but not at the surface. In the western velocity profile beneath the Miocene forearc (there is no arc in this segment of the Andes today), there is no difference in horizontal velocity between the upper and middle crust; because there is a velocity gradient in the middle crust between western and central velocity profiles, there must be middle crustal shortening there as well.

These schematic horizontal velocity profiles are, of course, oversimplified; they do not address shortening in the Sierras Pampeanas, strike slip along the El Tigre fault, or forearc deformation. Nonetheless they explain, to a first approximation, the patterns of crustal thickening and surface deformation that are observed in the Andes at lat ∼30°S.

The crustal deformation scenario proposed for this segment of the Andes may have broader application. Kley and Monaldi (1998) documented a deficit of surface shortening in the retroarc thrust belts relative to the crustal thickness throughout much of the Central Andes. Translation of lower crustal material eastward, perhaps linked to subduction erosion, could account for at least some of this deficit, although the mechanical model proposed by Bird (1988) is only applicable to flat subduction regimes such as exist beneath the Precordillera.

CONCLUSIONS

Horizontal shortening in the Argentine Precordillera over the past 13 m.y. is ∼86 ± 22 km, or ∼72% ± 5%. The line-length balances of two parallel sections across the belt are also within this range. The average yearly strain rate for this deformation is 56 ± 4 × 10–9/yr, although the temporal history of motion (Jordan et al., 1993, 2001) in this segment of the Precordillera shows that this strain rate must have varied with time. Nonetheless, the average value is indistinguishable from the 1D and 2D strain rates across the Precordillera calculated from available geodetic GPS data (Brooks et al., 2003), which give us 63 ± 8 × 10–9/yr. Although the GPS strain rate is undoubtedly mostly elastic due to loading of the locked décollement and the locked subduction megathrust, it appears that the contemporary strain in the Precordillera will be converted into permanent deformation in the Precordillera by future earthquakes. The nonorthogonality between GPS mean vector orientation and geological shortening is in part related to north-south gradients in the vector field producing a rotation of the infinitesimal strain ellipse, but probably also to some component of permanent vertical axis rotation over geologic time.

The décollement of the Precordillera is well identified by industry seismic reflection data at the eastern limit of the belt, and its westward projection coincides with the high-amplitude middle and/or upper crustal event on the receiver function stacks given by Gans et al. (2011). This event is probably the top of Cuyania basement rocks, and the décollement is at that interface, as there is no outcrop evidence that Cuyania basement is involved in Precordillera thrusting. The broadband seismic data and balanced sections show that the Precordillera décollement projects westward beneath the high Andes (Fig. 12), just 220 km horizontally and 80 km vertically from the plate boundary. This result, combined with the well-imaged Moho, shows that lower crustal thickening also occurs beneath the thin-skinned thrust belt. We suggest that lower crustal flow similar to that proposed by Bird (1988) for the Laramide Rocky Mountains is responsible for the crustal thickening pattern beneath this segment of the Andes.

Judge is grateful to field assistants Jordan Garroway, Rowan Gaffney, Bill Barnhart, and Rachel Valletta for thoughtful field conversations and good company while collecting field data. We thank our colleagues Chris Andronicos, Terry Jordan, Laura Giambiagi, Greg Hoke, Sue Kay, and Manfred Strecker for many fruitful discussions and field visits, and Susan Beck for providing a high-resolution version of the receiver function data for inclusion in Figure 12. We are grateful to Peter DeCelles and Raymond Russo for very helpful critical reviews of the manuscript. This research was supported by National Science Foundation grant EAR-0510785.

1Supplemental Table. A tab-separated, column-oriented text file that contains all of our fault slip data. The Supplemental Table can be displayed in a spreadsheet or opened and manipulated by FaultKin, a freely available program for Macintosh and Windows operating systems that can be downloaded from Allmendinger’s web site: http://www.geo.cornell.edu/geology/faculty/RWA/programs/faultkin.html. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES01062.S1 or the full-text article on www.gsapubs.org to view the Supplemental Table.