An important objective of volcanic research is to establish a cause-and-effect relationship between the age of fault kinematics and volcanic arc evolution based on structural and stratigraphic evidence. The predominant hyperarid climate in the Central Andes since Miocene times makes it a world-class area for investigating the evolution of a volcanic arc. This region records the complete development of the late Cenozoic Andean volcanic arc. This study focuses on the interpretation of volcanism in the context of recently dated tectonic structures along the southern Central Andes Volcanic Zone between 24.5° and 27° S. This segment of the arc has had a complex evolution and consists of hundreds of volcanoes, including constructional (monogenetic and polygenetic) and caldera volcanoes. By reviewing and reevaluating the geological maps in the literature, we are able to better constrain the temporal evolution of the central Chilean volcanic arc, including timing and kinematics of regional faults. Recognition of 15 Oligocene to Pleistocene ignimbrites and their sources has allowed us to define 11 caldera systems contemporaneous with effusive constructional volcanoes. The extent of the ignimbrite deposits allows them to be correlated between isolated outcrops that preserve different stratigraphic sequences, enabling the construction of a more complete and accurate volcanic stratigraphy. Two main NE-SW– and N-S–oriented thrust systems dominate the structural architecture of this segment of the arc. The first, located in the Precordillera, was active between 25 and 14 Ma and extends over 200 km to the northeast through the Pedernales-Arizaro thrust fault. Parallel to this thrust, the east-vergent Antofalla thrust fault system developed during Oligocene–Miocene times. The second system, located within the volcanic arc, includes sinuous N-S contractional structures that developed in pulses between the middle and late Miocene. There appears to be a cause-and-effect relationship between tectonic pulses and the development of volcanism, whereby changes in the upper crustal stress field lead to the generation of extensional domains. These conditions favor magma storage at upper crustal levels, thus promoting a suction-pump effect. The coexistence of both dominantly effusive constructional volcanism and explosive caldera volcanism results from the same tectonic conditions that produced shortening, as a consequence of the maximum compressive stress and conjugated extensions. In this paper, we suggest a new model that integrates the coexistence and contemporaneity of compressive structures and the widespread development of effusive constructional volcanism and explosive caldera volcanism along the Andean Oligo-Miocene volcanic arc.


The southern Central Andes Volcanic Zone (SCVZ; e.g., Stern, 2004) between 25° and 27° S (Fig. 1) records a complex evolution of the volcanic arc. It consist of hundreds of volcanoes and multiple volcano types, from large calderas (∼45 km across) and associated ignimbrites to lava fields. Hyperarid conditions have prevailed since early to middle Miocene times (Alpers and Brimhall, 1988). Although this has resulted in excellent preservation of volcanoes and caldera structures and their products, it has also restricted erosion, meaning that vertical exposures are scarce. The lack of vertical cuts is one of the most significant obstacles to studies in the region, as such cuts enable clear stratigraphic relationships to be identified and deposits to be well characterized, both of which are important when reconstructing regional stratigraphy.

There have been a number of prior studies of the SCVZ but commonly at low resolution or without detailed correlations. For example, Naranjo and Cornejo (1992) collected important regional and K-Ar geochronological data along the volcanic arc of the SCVZ, however on a large scale (1:250,000). The volcano and ignimbrite units were thus mainly defined by K-Ar geochronology and rough facies mapping. The regional volcanic geology was partially extended to the retro-arc in Argentina by Seggiaro et al. (2006, 2007). Other volcano and ignimbrite studies (e.g., Trumbull et al., 1999; Siebel et al., 2001; Schnurr et al., 2007; Kay et al., 2013; Freymuth et al., 2015; Guzmán et al., 2014) have been focused on sampling all possible specimens primarily for geochemical and isotopic study, but without detailed correlation of units and facies, although Guzmán et al. (2014) presented partially compiled maps and age data. Thus, most ignimbrite sources have remained generally unknown, and many fundamental stratigraphic details of the volcanic fields still need to be understood.

The area comprising the Lastarria, Cordón del Azufre, and Bayo volcanic complexes (Fig. 1) has been an important focus of study during the past decades due to the discovery of an extended (>50 km diameter) uplift pattern at a decimeter scale that began in A.D. 1997–1998 and has been named “Lazufre” (Pritchard and Simons, 2002). The investigation of its possible causes is one of the main research topics of the international project PLUTONS, the main theme of this issue.

During the last five years, the Chilean Geological Survey (Sernageomin) has developed a 1:100,000-scale mapping program that included the southern Central Andes volcanic arc, with a focus on the region between 24.5° and 27° S. Through this work, detailed geological maps of the region and numerous radiometric dates (K-Ar, Ar/Ar, and U-Pb) of this Andean segment are now available (Naranjo et al., 2013a, 2013b, 2016). Thus, many new effusive volcanic geochronological units, volcanic complexes, calderas, and ignimbrites have been identified, and the timing and kinematics of several other tectonic structures have been better constrained. These new findings allow us to further understand key processes involved in the late Cenozoic development of the Andean orogen along the SCVZ.

In this paper we present an integrated overview of the Oligocene to recent volcanic stratigraphy of this area (Fig. 1) and its implications for the spatio-temporal model of the SCVZ. Furthermore, based on several key examples, including a remarkable record of tectonic and volcanic structures at ca. 14–13 Ma, we suggest that there is a clear relationship between tectonism (the age of fault kinematics) and volcanic evolution. Based on the brittle nature of the uppermost crust, we suggest a new hypothesis that allows better explanation of the coexistence and contemporaneity of inherited compressive structures responsible for the Andean uplift with the generalized development of effusive constructive and explosive caldera volcanism in the Central Andes.


The Andean orogeny formed due to variations in the subduction vector and geometry of the more dense oceanic Nazca plate under the continental South American plate. These changes have driven the structural architecture of western South America, as well as controlled the position of the volcanic arc. In Paleogene times, an oblique subduction vector oriented approximately N60°E (Cande and Leslie, 1986; Pardo-Casas and Molnar, 1987) meant that the SCVZ volcanic arc was located ∼100 km to the west of the present-day arc. The evidence of this earlier arc is documented by the Chile-Alemania Formation (Naranjo and Puig, 1984; Puig et al., 1988; Fig. 1). During the late Cenozoic (ca. 25 Ma), the volcanic arc migrated to the east to its present position (west margin of the back-arc Puna plateau; Acocella et al., 2007). This reflected a shift in the convergence angle to almost orthogonal (N75°) and an increase in the rate of subduction (from 90 to 120 km/m.y.) (Cande and Leslie, 1986). This change in plate kinematics led to the initiation of Andean uplift and extensive deposition of piedmont sediments, known as the Atacama Gravels, in particular to the west of the arc (Fig. 1) between 25 and 10 Ma (Mortimer, 1973; Naranjo and Paskoff, 1985; Nalpas et al., 2008). The tectonic regime of the southern Puna has been dominantly compressive from the late Oligocene to recent times (Allmendinger et al., 1997; Kraemer et al., 1999).

Crustal thicknesses vary throughout the CVZ. Along the southern part of the volcanic arc (SCVZ), thicknesses are 40–65 km, less than those reported in the northern part of Chile up to the Arica bend north of 24° S (62–80 km; Yuan et al., 2002, Zandt et al., 1994; Allmendinger et al., 1997; McGlashan et al., 2008; Bianchi et al., 2013). The reason for this difference is disputed. Some studies propose the segment of the arc between 24.5° and 27° S as a “transition zone” between the normal subduction segment to the north (subduction angle of 25°–30°; north of 24° S) and the flat-slab segment to the south (28°–33° S; Cahill and Isacks, 1992; Graeber and Asch, 1999; Gutscher et al., 2000; Schnurr et al., 2007; Kay et al., 2013).

The volcanic arc in the SCVZ (24.5° to 27° S) is located 250–300 km to the east of the Chile-Perú trench with a width spanning 70–60 km. The locations of both constructional volcanoes and calderas have been associated with NW-SE regional left-lateral strike-slip fault zones (e.g., Olacapato–El Toro, Archibarca, Culampaja fault zones; Riller et al., 1999, 2001; Ramelow et al., 2006). Based on aeromagnetic images of the southern Puna, Chernicoff et al. (2002) argued that the intersection of large fault structures focused magmatism.

Most previous studies in the area have focused on chemical analyses of lavas and ignimbrites in order to identify compositional changes in the volcanic arc related to the geodynamic evolution of the CVZ. Trumbull et al. (1999) concluded that the amount of crustal contamination has varied since the Oligocene. They estimated that magmas erupted between 25 and 8 Ma were only slightly contaminated, following which contamination increased until 5 Ma and remained stable thereafter. The ignimbrite chemistry has remained uniform during the late Cenozoic and shares chemical traits with deposits from constructional volcanoes, where the dominant erupted products are andesites, dacites, and rhyolites (Siebel et al., 2001; Stern, 2004; Schnurr et al., 2007). Crustal contamination of CVZ magmas is thought to occur by intra-crustal assimilation combined with crystallization of these magmas, and/or crustal anatexis (Stern, 2004, and references therein). Some authors have agreed that the extent of crustal contamination is lower within the SCVZ than at the Altiplano-Puna Volcanic Complex (APVC; de Silva, 1989), an ignimbrite province recognized to the north of 24° S (Siebel et al., 2001; Schnurr et al., 2007). New insights into the detailed geology of the SCVZ are summarized below and may help to improve on the previously published partial interpretations of the geological history along some segments of the Central Andes. In particular, the new detailed geological mapping carried out in the SCVZ between 25° and 27° S (Clavero et al., 2012; Naranjo et al., 2013a, 2013b, 2016) has enabled the stratigraphy of both effusive and explosive volcanic units to be reevaluated, and previously uncorrelated scattered ignimbrite outcrops to be correlated.


By combining data and analyses from all prior studies in order to form a complete geological map of the SCVZ and put forward a more robust model of the past geology (Naranjo et al., 2013a, 2013b, 2016), we have been able to extrapolate units to the south, east, and north, thus completing a geological map of the entire volcanic arc between 24.5° and 27° S. Constructional volcanoes (stratovolcanoes, volcanic complexes, and lavas) are grouped primarily on the basis of radiometric ages, subtle differences in erosion, or preservation degrees, supported by lithological characterization. A detailed updated account of available ages is provided in the Supplemental Material1. Although contemporaneous explosive and effusive volcanism has been continuous since ca. 25 Ma, geographical distribution of volcanoes is discontinuous in time, without a defined pattern (Fig. 2). Despite this, new age and facies data from previously dispersed ignimbrite outcrops allow us to make new correlations and produce a complete overview of Oligocene–Holocene explosive volcanism; the large ignimbrite deposits can now be considered distinct marker horizons, and because they are interbedded with deposits of constructional volcanoes (Fig. 3), they enable different stratigraphic sequences to be correlated. No evidence of arc migration over the last 25 m.y. is found, nevertheless the arc width has been variable and the overlap of younger volcanoes growing over older volcanic structures is a characteristic feature of the area.

More than 300 individual volcanoes are recognized in the SCVZ, including a wide variety of constructional volcanic forms (monogenetic and polygenetic volcanoes such as single cones, domes, stratovolcanoes, compound volcanoes, maars, lava-domes, coulées, lava fields, and volcanic complexes) and collapse calderas. Some volcanoes preserve evidence of past instabilities that resulted in volcanic debris-avalanche deposits (Naranjo and Cornejo, 1989; Naranjo and Francis, 1987; Naranjo et al., 2015; Villa and Naranjo, 2015a). Few volcanoes show local evidence of Holocene activity (Naranjo, 2010; Naranjo et al., 2013a, 2013b, 2016). Some internal features of volcanic edifices have been exposed through these erosional processes, showing that epithermal alteration extensively penetrated many pyroclastic layers of volcanic edifices. All lava flows within the SCVZ are blocky lavas exhibiting conspicuous central channels with transversal ridges and well-developed levees. These lava structures have been observed even in such ancient volcanoes as the ca. 16 Ma Chaco volcano. Locally, the Negriales del Lastarria (1.8 km3; Naranjo, 2010) and the so-called Chaito Lava (a 12 km3 individual lava flow) (Naranjo et al., 2013b) are notable examples of lava fields. Several ignimbrites formed extensive plateaus and filled intermontane basins, including the extensive Río Frío ignimbrite which covers >3000 km2 of mapped outcrop (Fig. 2).

24 Ma to 14 Ma Arc Volcanism

Constructional volcanoes 24–14 Ma in age occur along the entire arc. Evidence of this is preserved in the remnants of stratovolcanoes, which crop out along the edges of the arc, where younger volcanism did not develop until after 16 Ma (Fig. 2A). Further, deposits formed during this time are recognized at the base of younger volcanoes. This implies continuous volcanism throughout the entire Oligocene–Miocene period and hence a stationary volcanic arc. These ancient edifices are interpreted as moderately eroded large compound volcanoes. Although mafic volcanism has been recognized at León Muerto volcano (Trumbull et al., 1999; Kay et al., 2013), potassium-rich andesite and dacite are the dominant lava compositions erupted during this period (Naranjo et al., 2013a, 2013b).

Deposits from constructional volcanoes are interfingered with several ignimbrites that extend to the east and west of the volcanic arc. The Río Frío and Pajonales ignimbrites are widely distributed in the northern part of the area (Fig. 2A). Maximum thicknesses of these ignimbrites are recognized in the Aguilar resurgent dome, a prominent 30 × 20 km bulge surrounded by a caldera moat, probably the oldest caldera feature preserved so far in the Central Andes (Table 1; Fig. 2A). This structure, formed by the Aguilar (3325 m above sea level [asl]) and La Isla (3965 m asl) salars (salt lakes), is tectonically tilted to the west. These morphologies correspond to the remnants of a ∼45-km-diameter poorly defined piston-type caldera called the Aguilar caldera, which is thought to have been active between 23 and 14 Ma (Naranjo and Cornejo, 1992; Naranjo et al., 2013a, 2013b). Based on preserved deposits (Fig. 2A), the Río Frío and Pajonales ignimbrites probably reached distances of as much as to 200 and 140 km, respectively, from the resurgent dome. The older (ca. 21–17 Ma; Fig. 4), dark brownish-pink Río Frío ignimbrites (Naranjo and Cornejo, 1992) show a distinctive, welded, homogeneous massive lithology that helps with correlation of isolated outcrops (Table 1 and Fig. 3). Lower ignimbrites in the resurgent dome have a subvertical decimeter-scale jointing and development of “normal tiny joints” (Yamagishi, 1987, p. 70), possibly produced as a consequence of subaqueous emplacement. The younger Pajonales ignimbrites (ca. 16.5 Ma; Naranjo and Cornejo, 1992) show two flow units (Qatatiña and Las Cuevas ignimbrites) separated by ∼1 m of lahar deposits (Fig. 4).

On the western flank and moat of the Aguilar resurgent dome, the 20 × 15 km, northwest-oriented Juan de la Vega phreatomagmatic caldera was the source of the Chixi ignimbrite (14.7 Ma; Figs. 2A, 4 [section A], and 5). It represents the youngest activity of the Aguilar caldera system. Low-aspect-ratio volcanoes and one phreatic diatreme developed on the southern part of Juan de la Vega caldera (Cornejo and Naranjo, 1988).

Farther south, other ignimbrites of unknown source constitute conspicuous horizons interbedded within the Andean piedmont deposits (Atacama Gravels; Mortimer, 1973; Naranjo and Paskoff, 1985); these are also found at SCVZ intermontane basins. Among these ignimbrites, the youngest, Vega Helada (ca. 19 Ma) and Juncalito (ca. 16.5 Ma), are restricted to the eastern side of the Pedernales Salar (e.g., Naranjo et al., 2016; Fig. 2A). The Vega Helada and Juncalito ignimbrites are notably more similar lithologically to each other than to the Río Frío and Pajonales units: their age, lithologic facies, and spatial distribution point to a possible genetic relationship. However, while the source and facies variation for the Río Frío and Pajonales ignimbrites are well constrained, the source of Vega Helada and Juncalito ignimbrites remains unclear, as they become thicker and coarser toward the south, away from Aguilar caldera. This may indicate the former existence of a local tectonically obliterated caldera of that age, located to the south of Aguilar caldera. The oldest units between 24 and 20 Ma are extensively distributed, but crop out as isolated sections, reaching distances of 60 to >200 km from the volcanic arc. We propose that they were indeed associated with earlier stages of older calderas, obliterated as a consequence of contemporaneous tectonism and younger volcanic activity. Combined, ignimbrites between 24 and 14 Ma reach a total estimated volume of at least ∼2000–2500 km3 (Table 1), which implies collapse caldera eruptions were abundant during the Oligo-Miocene.

14 Ma to 9 Ma Arc Volcanism

The distribution of constructional volcanoes between 14 and 9 Ma along the SCVZ spans the same width as in the earlier period, but edifice remnants are located predominantly along the eastern and western arc borders. Compound volcanoes characterize central volcanism in the northern part of the SCVZ during this period. The western arc front is dominated by relatively large, well-preserved volcanoes (up to 31 km3; Villa and Naranjo, 2015b), some of which display sector collapses such as De La Pena and Quebrado volcanoes (Fig. 2B). Overlapping younger volcanoes partially obscure the spatial continuity of the volcanic arc during this period (Fig. 2B). Locally, volcanoes exhibit wide craters and scoriaceous pyroclastic layers, probably associated with an abundant phreatic water source. Further south, the Piedra Parada volcano probably constitutes the older remnant of the Sierra Nevada Volcanic Range base, suggesting that there was another NW-SE volcanic chain in this area. Lava compositions were primarily andesite and dacite during this time interval (Naranjo et al., 2013a, 2013b).

The NW-SE–oriented Salar Grande–Barrancas Blancas–Pampa de los Bayos caldera complex developed during the ca. 13–9 Ma time interval immediately to the south of the Aguilar resurgent dome (Fig. 2; Table 1). Located at the northwestern end of this caldera alignment, the Salar Grande caldera (ca. 13–11 Ma) presents as a NW-SE–elongated shape (∼50 × 25 km) with discontinuous topographic margins, within which the proximal ∼175-m-thick Salar Grande ignimbrite sequence deposited (Figs. 4 [section D] and 5). Immediately to the southeast of the Salar Grande caldera, the Barrancas Blancas caldera hosts the homonymous ignimbrite (ca. 9.3 Ma, >10 km3; Table 1) as well as constructional volcanoes <10 Ma. This caldera developed a resurgent dome, which collapsed ∼9 m.y. ago forming the 15-km-wide Pampa de los Bayos nested caldera, which is partially surrounded by ring-shaped reverse faults, which formed along the moat after caldera collapse probably in a mechanism similar to that proposed in the model of Martí et al. (1994, their figure 7). Thus, Barrancas Blancas and Pampa de los Bayos is a multistage collapse caldera system, active between 10 and 9 Ma. Accurate stratigraphic relationships indicate that the latter was the source for the San Andrés ignimbrite (ca. 9 Ma; Figs. 4 [section E] and 5), which is found in the plains and gorges to the west between 26° and 27° S (Clark et al., 1967; Naranjo et al., 2016).

The circular, 30-km-wide Los Colorados caldera was the source of the eponymous 80 km3, ca. 9 Ma ignimbrite (Figs. 4 [section E] and 5) and was formed seventy kilometers northeast of the Salar Grande caldera (Fig. 2B; Table 1; Naranjo et al., 2016). While the northern and eastern caldera margins are partially covered by younger <3.5 Ma lavas, the western and southern scarps are clearly preserved, reaching up to 800 m high. The caldera collapse destroyed volcanoes older than ca. 10 Ma and hosts volcanoes of ca. 7 Ma (Richards et al., 2013).

9 Ma to 3.5 Ma Arc Volcanism

During the 9–3.5 Ma time interval, the western arc front was slightly displaced to the east and central volcanoes covered virtually the entire width of the arc (Fig. 2C). The lavas show a broader compositional range spanning basaltic andesites to dacites with elevated potassium (Naranjo et al., 2013a, 2013b). The main structures are compound volcanoes, which are up to 1200 m high with volumes of 30 km3; these commonly have associated volcanic debris-avalanche deposits (e.g., Agua Amarga volcano; Fig. 2C). Nevertheless, local clusters of small simple cones (<200 m high and <3 km3; Villa and Naranjo, 2015a, 2015b) are exposed north of the La Isla Salar and west of the Lastarria and Cordón del Azufre volcanic complexes. In one of those clusters, an exceptionally voluminous 12 km3 andesitic lava flow (Chaito Lava) has been dated at ca. 6 Ma (Naranjo et al., 2013b). The 9–3.5 Ma volcanic arc is discontinuous in the areas of the Salar Grande–Pampa de los Bayos calderas and Aguilar resurgent dome (Fig. 2).

Explosive volcanism during this time interval is characterized by several ignimbrites and calderas. The ashy, unconsolidated Las Parinas ignimbrite (Table 1; Naranjo and Cornejo, 1992) covers surfaces and fills gorges carved into the Los Colorados ignimbrite and is overlain by the ca. 2–3 Ma, Gemelos volcano (Naranjo et al., 2013a, 2013b). Based on its distribution (Fig. 2C), the most probable source of this ignimbrite is the Alto Parinas caldera, a ∼10-km-diameter depression rimmed by a 100–250-m-high scarp and possible ring-shaped reverse faults (Martí et al., 1994), located 6 km to the north of the Barrancas Blancas caldera (Fig. 2C).

Just to the south of the Barrancas Blancas caldera and 50 km south of the Alto Parinas caldera, an ENE-WSW alignment of five calderas constitutes the most prominent regional feature (Fig. 2C). The ∼19-km-diameter subcircular, multistage Wheelwright caldera (Clavero et al., 2012) is the source of the homonymous ignimbrites (Table 1; Figs. 2C, 4 [section F], and 5). Following the first collapse at ca. 5 Ma (Wheelwright 1 ignimbrite) that produced escarpments between 400 and 700 m high, a second inner escarpment was formed (Wheelwright 2, ca. 4 Ma). Eastward, the Peinado, Laguna Amarga, and Laguna Escondida coalescing caldera cluster is recognized. The Peinado caldera is the 18-km-diameter eastern remnant sector of an obliterated caldera. The younger 30-km-diameter Laguna Amarga caldera mainly preserves its northern and eastern (∼400 m) scarps, as it was affected by the collapse of the Laguna Escondida caldera. The latter 40 × 16 km, E-W–oriented structure has 600-m-high scarps and hosts Pleistocene–Holocene volcanoes. Seggiaro et al. (2006) identified the Laguna Amarga caldera as the source of the ca. 4 Ma Laguna Verde ignimbrite (Table 1; Figs. 2C, 4 [sections E and F], and 5). So far, there are no stratigraphic data of proximal lithofacies to help elucidate in detail the collapse processes of each caldera in this system.

3.5 Ma to 0 Ma Arc Volcanism

Volcanism during the 3.5–0 Ma time interval is discontinuously distributed and forms the present-day active volcanic chains (e.g., Llullaillaco, Lastarria, Bayo, Cordón del Azufre, and Sierra Nevada volcanic complexes; Fig. 2D). Notably, the remnant of one of the oldest volcanoes in the arc (ca. 24 Ma) crops out almost at the foot of the most active volcano in the area (Lastarria volcanic complex). This may indicate the random distribution of units within the effusive volcanic arc. The 35-km-long and NNE-oriented Lastarria-Bayo chain is formed mainly by overlapping volcanic complexes as seen at the Piramide, Atalaya-Chuta, Lastarria, Cordón del Azufre, and Bayo volcanic complexes, some of them with known Holocene activity (Naranjo, 2010; Naranjo et al., 2013b). Approximately 120 km to the south, a 30-km-long, NW-SE–oriented volcanic chain includes the Azufrera Los Cuyanos (<500 ka), Jhuchuy (ca. 1 Ma), and Sierra Nevada (<500 ka) volcanic complexes. Potassium-rich andesites and, to a lesser extent, dacites are the dominant lava compositions produced at these volcanoes (Naranjo et al., 2013a, 2013b).

Evidence of explosive volcanism during this period includes the 1.37 km3 Caletones Cori ignimbrite (Fig. 2D; Table 1), a crystal-rich dacitic tuff, erupted from the Corrida de Cori ridge (Naranjo and Cornejo, 1992; Naranjo, 2010). This eruption is dated to 0.457 ± 0.013 Ma and precedes the formation of the Escorial volcano (Middle Pleistocene age), and overlies the lower Pleistocene Río Grande volcanic lavas (Richards and Villeneuve, 2002; Naranjo et al., 2013b). Further south, the 2.4 km3 Chato Aislado dacitic volcano (Fig. 2D) consists of a 5-km-diameter explosion caldera, an exogenous dome, and associated ignimbrite and pyroclastic fall deposits, and was built on the floor of the Salar Grande caldera at ca. 1.4–1.1 m.y. ago (Naranjo et al., 2013a).


The volcanic arc is flanked by contractional fault systems associated with Andean uplift. Several N-S– and NE-SW–oriented thrust systems also cross and cut arc volcanic edifices; some structures are partially covered and/or obliterated by younger stratovolcanoes and calderas. Thrust systems along the Precordillera are sinuous and NNE-SSW oriented at the Domeyko and Claudio Gay Ranges (Fig. 5). These contractional faults gave rise to the Andean uplift and subsequently produced the Andean piedmont molasses in which ignimbrites dated between 25–16 Ma are interbedded. Moreover, the eastern border of the volcanic arc is limited by the east-vergent Antofalla thrust fault system (Fig. 5), active since the late Oligocene (Kraemer et al., 1999). This NE-SW system is parallel to the east-vergent Pedernales-Arizaro thrust fault (Naranjo et al., 2013a) that extends for >220 km from the Precordillera cutting across the volcanic arc to the northeast; fault scarps associated with this thrust locally exceed 700 m. Based on geochronological and morphostratigraphic evidence, these authors (Naranjo et al. [2013a]) constrained the Pedernales-Arizaro fault activity to the ca. 14–13 Ma period. Notably, all Miocene–Pliocene calderas recognized in the area are located just east of these remarkable structures, suggesting a long-term spatio-temporal relationship between tectonism and mainly explosive volcanism along this SCVZ segment (Fig. 5).

Several thrust faults with opposite vergences accommodate up to 40% of the differential shortening. They affect the Aguilar resurgent dome, which is found in the northern part of the Claudio Gay Range, which is to the east of the Domeyko Range (Fig. 5). The oblique-aligned Salar Grande caldera collapse as well as younger Miocene volcanoes obliterated the connection between these sub-systems. These faults form sinuous, 20–65-km-long, N-S contractional structures developed within the volcanic arc. The activity of these thin-skinned–like faults were also developed in pulses within the middle Miocene (ca. 14–13 Ma), as supported by structures found to the east of the Aguilar caldera resurgent dome (Fig. 6).

The northern Imilac–Salina del Fraile lineament is recognized in the Salina del Fraile pull-apart basin, previously identified as a NNE sinistral strike-slip fault (Reijs and McClay, 2003). Nevertheless, our observations allow us to redefine this basin as a consequence of a NW-SE dextral transtension fault. The activity of this fault occurred before ca. 9 Ma, and its extension to the north controls the volcano alignment since 8 Ma, including Corrida de Cori–Escorial and Llullaillaco volcanoes (Richards and Villeneuve, 2002). Approximately 40 km to the northwest of Salar de Pajonales (Fig. 5), a dextral oblique-slip thrust fault, probably active until ca. 14 Ma, affected the Domeyko Range (Venegas et al., 2013).

We infer that the structures recognized so far in the Precordillera, volcanic arc, and back-arc represent the uppermost crustal architecture of the SCVZ generated during Andean uplift.


Although some authors have proposed volcanic gaps within the evolution of the Miocene–Holocene volcanic arc (Kay et al., 2013), our geochronological evidence reveals that the arc activity covers the entire span between 25 Ma to the present, including localized Holocene activity (Naranjo, 2010; Naranjo et al., 2013a, 2013b, 2016). Along the SCVZ, constructive effusive volcanism has been continuous since the late Oligocene, and apparently, volcanoes are randomly distributed without defined patterns; this likely reflects the complexity of magma pathways through the upper crust and the small edifice size compared to those of the Southern Andes Volcanic Zone (SVZ; Stern, 2004). Recently, Villa and Naranjo (2015b) attributed this difference to greater structural complexity of the upper crust in the SCVZ (in comparison with the SVZ) affecting magma ascent times. The presence of calderas and voluminous lava flows (e.g., the 12 km3, ca. 6 Ma Chaito Lava; Naranjo et al., 2013b) implies growth of large magma reservoirs. Although the actual distribution of older effusive stages of the arc is partially obscured by erosion and overlying volcanoes, it is worth noting that the temporal density of central volcanoes during the ca. 9–3.5 Ma period could represent the most extensive period of effusive volcanism in the SCVZ (Fig. 2). By that time interval, the shallow stress field relaxation along the SCVZ was probably more extensive than in previous stages, as a consequence of the decreased subduction velocity at 10 Ma from ∼120 km/m.y. to ∼90 km/m.y. (Cande and Leslie, 1986). Multiple minor magma reservoirs formed, increasing the extent of crustal contamination in andesitic magmas between 8 and 4 Ma (Trumbull et al., 1999).

Recent reviews of ignimbrite deposits in the SCVZ have been made based on partial compilation of maps and age data. Guzmán et al. (2014) for example proposed that ignimbrites are temporally distributed mainly along N-S to NNE-SSW, NW-SE to WNW-ESE, and NE-SW trends. However, our recent field mapping and thorough stratigraphy supported by detailed geochronology now give a more complete picture. They show that ignimbrites of all ages are distributed according to their caldera source location, which in turn follows tectonic-type patterns. In addition, most ignimbrites extend well beyond the volcanic arc, to both the west and east. However, the distribution of old ignimbrites is commonly hidden by younger lavas, ignimbrites, and Andean piedmont epiclastic deposits. Although the southern Wheelwright–Laguna Amarga–Laguna Escondida caldera system presents much younger and well-preserved structures compared to other caldera systems of similar age (Fig. 2), their ignimbrite stratigraphy is significantly more complex. Nevertheless, the spatial distribution of these calderas and the contemporaneity of their eruptions favor a regional piecemeal plutonic assembly model for their magma reservoirs (see, e.g., Grunder et al., 2008, Kern et al., 2016).

At least two caldera clusters are spatially (in conjugate alignment) and temporally associated with the main thrust fault systems that have been detected in the southern part of the Central Andes (Fig. 5): (1) The NW-SE–oriented Salar Grande–Barrancas Blancas–Pampa de los Bayos caldera system was formed between 13 and 9 Ma, following the shortening along the NE-SW Pedernales-Arizaro fault (ca. 14 Ma), which obliterated the Aguilar caldera system; and (2) the Wheelwright–Laguna Amarga–Laguna Escondida caldera system farther south developed between ca. 5 and ca. 4 Ma, aligned 60° oblique to the contractional faults at the Claudio Gay Range. We interpret the oblique orientations of the caldera systems to mean that the magma reservoirs of such calderas were likely emplaced at an oblique angle to the main thrust strike, occupying shallow crustal spaces. The activity of these magmatic systems is constrained by the oldest and youngest ignimbrites produced. Thus, different eruptive peaks are recorded within the ca. 24–15 Ma and ca. 14–9 Ma periods for the Aguilar and Salar Grande–Barrancas Blancas–Pampa de los Bayos systems, respectively. On the other hand, the Wheelwright–Laguna Amarga–Laguna Escondida caldera system shows an eruptive peak between ca. 5 and 4 Ma, a notably much shorter interval.

Also noteworthy is a NW-oriented lineament between the Los Colorados caldera and the Lazufre uplift pattern area (Fig. 5) (Pritchard and Simons, 2002), attributable to magma movements since at least A.D. 1998 (e.g., Anderssohn et al., 2009). The Late Pleistocene–Holocene effusive magmatic activity of the volcanic arc is particularly concentrated in that area, and includes the Lastarria, Cordón del Azufre, and Bayo volcanic complexes (Naranjo et al., 2013b). This NW-oriented spatial lineament is also conjugated at ∼30° with the middle Miocene Pedernales-Arizaro thrust fault.


Magma propagation through the upper crust is influenced at deep levels (e.g., by the nature of the magma source) and at shallow levels (e.g., by the stress field, inherited crustal structures) (Le Corvec et al., 2013). Shortening (25–10 Ma) generated a structural architecture that, at shallow brittle crustal levels, consists of major fault networks (Fig. 5) which, in turn, encase minor faults and fracture meshes or domains. This architecture accommodated the long-term deformation in a variety of ways, locally opening and closing fractures at different upper-crustal levels. We interpret the structural and temporal evidence in the SCVZ to implicate very strongly structural control on the emplacement timing of the magmatic systems. Structural control on emplacement of crustal magma has been recognized for decades (Petford et al., 2000). For instance, Pitcher (1997) reviewed how discordant and oblique plutons that form the roots of volcanoes are characteristic of brittle upper crust subject to changes in stress and syn-emplacement seismicity. The association of intrusive bodies with (and oblique to) major faults and the syn-tectonic emplacement of magma was recognized as far back as Anderson (1942). Fractured “permeable” upper crust in conjunction with the “buoyancy” inherent in decompressing magma favor suction and vertical magma transport toward higher, less-dense crustal levels (e.g., de Saint Blanquat et al., 1998). Magmatic pulses due to tectonic forcing and/or pumping or buoyancy instabilities in the deeper crustal source regions (e.g., de Silva et al., 2015) may be accompanied by other mechanisms such as seismic pumping (see Pitcher, 1997; Katz et al., 2006), resulting in accumulation of magmas in the permeable upper crust. We note that Riller et al. (2001) and de Silva et al. (2006) developed similar notions in their suggestion that volcanism in the Altiplano-Puna Volcanic Complex farther to the north is the result of the emplacement of obliquely oriented magma bodies in response to changes in the state of stress in the upper crust there.

A variety of magma ascent mechanisms (diking, diapirism, ascent along faults, and ascent during heterogeneous ductile flow) form final members in a transitional continuum, influenced by both the host rock and magma behavior (Paterson and Miller, 1998; Clemens, 1998, 2012). Vigneresse and Clemens (2000) suggested that fracture-induced magma propagation (diking) within the uppermost crust was not itself sufficient for magma ascent. They argued that deformation through strain partitioning (see Jones and Tanner, 1995; Carreras et al., 2013) was also an essential mechanism for magma ascent. As a possible complementary mechanism, we suggest that variations in the stress field of shallow crustal regions could promote the formation of extensional fracture meshes or domains, locally conjugated to main compressive stresses represented by the major inherited thrust faults (e.g., the ca. 14–13 Ma east-vergent Pedernales-Arizaro thrust fault; Figs. 2B, 5, 7A, and B7B). These domains are heterogeneously distributed in terms of the strain intensity and strain type as a deformation process (strain partitioning). Differences in spacing and size, and magma storage duration, may determine the geometric or structural maturity of a particular magma reservoir and thus explain the coexistence of explosive calderas and more effusive constructional volcanoes. According to Glazner et al. (2004), in places with dense fracture meshes, magma influx could be increased and consequently inhibit freezing. Therefore magma can accumulate at shallower crustal levels, forming larger reservoirs (e.g., the ca. 13–9 Ma Salar Grande–Barrancas Blancas–Pampa de los Bayos caldera system; Figs. 2B, 5, and C7C). In places where the crust is less fractured, magma reservoirs are smaller due to a lower magma input, which prevents large-scale amalgamation of partial melts.

The presence of permeable crust-plumbing systems could also play an important role in country rock assimilation processes (such that assimilation increases with system permeability), leading to a range of structural maturity stages for magma reservoirs (e.g., Zellmer and Annen, 2008). Crustal assimilation in the SCVZ has indeed been detected in past geochemical investigations (Trumbull et al., 1999; Siebel et al., 2001; Schnurr et al., 2007). Mineralogical and geochronological evidence of this assimilation has also been observed in xenocrystic zircons from the ca. 13–11 Ma Salar Grande ignimbrite, which showed new growth ca. 12 Ma on these Carboniferous to Triassic inherited zircons (Fig. 8; for age isotopic data, see sample 060912-10D in Naranjo et al., 2016). This new growth possibly occurred during the formation of the Salar Grande caldera reservoir and would indicate the assimilation of its country rock.


The geology of the southern segment of the Central Andes volcanic arc reveals important links between tectonic and volcanic processes. Hyperarid conditions have dominated the climate since Miocene times in the SCVZ, promoting the preservation of a unique volcanic landscape and fault structures. Contemporaneously with the growth of constructive volcanoes, multiple caldera systems developed, which appear to be related to the main structural features of the arc segment. Potentially, there is a cause-and-effect relationship between tectonic pulses and the development of volcanism through the generation of extensional domains. An outstanding example can be seen in the spatio-temporal relationship between the ca. 14–13 Ma east-vergent Pedernales-Arizaro thrust fault and the conjugate extensional domains occupied by the ca. 13–9 Ma Salar Grande–Barrancas Blancas–Pampa de los Bayos caldera system. As a possible mechanism, we propose that these domains have been capable of storing large volumes of magma by promoting a suction-pump effect in the shallow brittle crust as a result of ongoing subduction-related changes to the stress field. Constructive volcanoes and coexisting calderas are the result of the same tectonic forces that, in turn, produced shortening and conjugated extensions, respectively. In light of the foregoing, we highlight this new model to explain the evolutionary contemporaneity of the compressive structures responsible for the Andean uplift and the generalized development of the volcanic arc with hundreds of constructive effusive volcanoes and large explosive calderas.

Our intensive field mapping and geochronological study in the SCVZ reveal various aspects that require further research. Petrological and geochemical studies of lavas and ignimbrites spatially associated with source calderas are needed, including systematic mineral and isotopic comparisons on the basis of the stratigraphy defined here. Mineral geochemistry has been proven to be a very good tool to complement stratigraphic and geochronological studies in the correlation of pyroclastic deposits (de Silva and Francis, 1989; Lindsay et al., 2001; Breitkreuz et al., 2014; Rawson et al., 2015). In order to detect how magma reservoirs are associated with different structural domains at different depths, it would be interesting to also carry out geobarometry studies along alignments of central volcanoes and calderas and between groups of different structural domains. The tectonic control on magma ascent and storage and its association with differentiation mechanisms have not yet been fully understood. These studies may provide many insights into the workings of magma plumbing systems within the crust in this part of the Andes.


This is a contribution of the Plan Nacional de Geología (PNG) of the Regional Geology Department of the Servicio Nacional de Geología y Minería (Sernageomin). The PNG Project 8011 maps (Carta Geológica de Chile: Áreas Salar de Agua Amarga y Portezuelo del León Muerto, Áreas Salar de Pajonales y Cerro Moño, and Áreas Cerro Panteón de Aliste y Cerro Colorado) are acknowledged. We thank Marcos Lienlaf, who provided important support in the production of digital maps and drawings. We are greatly appreciative of Gonzalo Núñez, Francisco Hevia, José Luis Díaz, Hugo Neira, Roberto Flores, Eduardo Martínez, and René Urbina for invaluable assistance in the field. The authors are grateful for the valuable and important suggestions made by referees Jan Lindsay and Gerhard Wörner. Finally, we greatly appreciate Harriet Rawson and Fernando Henríquez for the final revision of the manuscript and Shan de Silva for editorial handling. This is a contribution to the PLUTONS project (http://plutons.science.oregonstate.edu).

1Supplemental Material. Southern Central Andes Volcanic Zone ages compilation. Please visit http://doi.org/10.1130/GES01350.S1 or the full-text article on www.gsapubs.org to view the Supplemental Material.
Science Editor: Raymond M. Russo
Guest Associate Editor: Shanaka de Silva
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.

Supplementary data