Basement rocks exposed in the Acatlán Complex of the Mixteca terrane in southern Mexico record two tectonothermal events: (1) a Devonian–Mississippian (ca. 365–318 Ma) event, recording extrusion and exhumation of high-pressure rocks; and (2) an Early to Middle Permian (ca. 289–263 Ma) event, involving N-S dextral shearing, transtensional deformation, and local S-vergent thrusting in a magmatic arc environment. We document an additional, regionally significant, tectonothermal event during the Middle to Late Triassic recorded by 40Ar/39Ar step-heating laser-probe ages ranging from ca. 239 and 219 Ma (≡ cooling from ca. 525 °C to 300 °C) for amphibole, muscovite, and biotite from: (1) the Carboniferous Amarillo unit, consisting of medium-grade, metasedimentary rocks intruded by mafic dikes; and (2) the Pennsylvanian–Middle Permian, low-grade, clastic-calcareous, arc-related Tecomate Formation. U-Pb laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) data yield an age of 339 ± 6 Ma for the youngest population of detrital zircon grains in the Amarillo unit. Lithogeochemical and Sm-Nd isotopic data for the Amarillo unit dikes are very similar to those of other Carboniferous meta-igneous rocks in the eastern and southwestern part of the Acatlán Complex, displaying affinities transitional between mid-ocean-ridge basalt (MORB) and continental tholeiites, and initial ɛNd (t = 339 Ma) values from -6.6 to +6.4, indicating both depleted and enriched mantle sources, as well as variable contamination by continental crust or by subduction-derived fluids. The 40Ar/39Ar cooling ages coincide with an apparent hiatus in magmatic activity in southern Mexico, which is inferred to record a change from steep to flat subduction.

Basement rocks exposed in the Acatlán Complex of the Mixteca terrane in southern Mexico record a polyphase deformational and metamorphic history and provide an excellent opportunity to investigate the Paleozoic to Mesozoic tectonic evolution of the North American Cordillera. Two dominant tectonothermal events have been recognized in rocks of the Acatlán Complex (Keppie et al., 2012, and references therein): (1) a Devonian–Mississippian (ca. 365–318 Ma) event, recording extrusion and exhumation of high-pressure rocks (Middleton et al., 2007; Elías-Herrera et al., 2007; Ramos-Arias et al., 2008, 2012; Vega-Granillo et al., 2009; Keppie et al., 2010, 2012); and (2) an Early to Middle Permian (ca. 289–263 Ma) event, involving N-S dextral shearing, transtensional deformation, and local S-vergent thrusting (Elías-Herrera et al., 2005; Morales-Gámez et al., 2009a; Kirsch et al., 2013). In addition, Vega-Granillo et al. (2009) reported two early Paleozoic tectonothermal events, although their existence is controversial (Keppie, J.D., et al., 2009).

Reconnaissance studies suggest a Middle to Late Triassic tectonothermal event north of Petlalcingo, bracketed between a single U/Pb detrital zircon age of 239 ± 4 Ma (Keppie et al., 2006a) and a 224 ± 2 Ma muscovite cooling age (Keppie et al., 2004a). In this paper, we provide further evidence for this tectonothermal event by presenting lithogeochemical and Sm-Nd isotopic, and laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) U-Pb detrital zircon data, as well as 40Ar/39Ar step-heating laser-probe data from amphibole, muscovite, and biotite from late Paleozoic low- to medium-grade units in the northeastern part of the Acatlán Complex. We show that this tectonothermal event is regional in extent, affecting Carboniferous–Permian rocks, and we investigate its tectonic significance in the context of subduction processes at the North American Cordilleran margin.

The Ordovician to Middle Permian Acatlán Complex forms the basement of the Mixteca terrane and is the largest inlier of Paleozoic rocks in Mexico (Ortega-Gutiérrez, 1978; Campa and Coney, 1983; Sedlock et al., 1993; Keppie, 2004). The Acatlán Complex is bounded to the east by the Permian, N-S, dextral Caltepec fault zone, which separates it from the ca. 1.0 Ga granulite-facies gneisses of the Oaxacan Complex (Figs. 1A and 1B; Elías-Herrera and Ortega-Gutiérrez, 2002) and to the south by the Cenozoic La Venta–Chacalapa fault zone (Tolson, 2007; Solari et al., 2007), which juxtaposes it against the Mesozoic–Cenozoic plutonic and high-grade metamorphic rocks of the Xolapa Complex (Pérez-Gutiérrez et al., 2009b). To the west, the Acatlán Complex is thrust over Cretaceous carbonates of the Guerrero-Morelos platform, which are exposed between the Acatlán Complex and the obducted Guerrero composite terrane (Centeno-García et al., 2008; Ramos-Arias and Keppie, 2011). To the north, the Acatlán Complex is unconformably overlain by continental and marine sedimentary rocks of Upper Permian to Middle Jurassic age (Piedra Hueca and Otlaltepec units: Morán-Zenteno et al., 1993; Matzitzi Formation: Centeno-García et al., 2009), and by Neogene–Holocene volcanic and volcaniclastic rocks of the Trans-Mexican volcanic belt (Ferrari et al., 1999).

The Acatlán Complex records a complex Paleozoic tectonothermal history reflecting the opening and closure of one or more ocean basins and their consequent continental interactions, culminating in the amalgamation of Pangea (e.g., Keppie et al., 2008, 2012, and references therein). These events were accompanied by Carboniferous to Permian subduction beneath southern Mexico (see inset of Fig. 1A). The Middle–Late Triassic of central and southern Mexico marked the development of a passive or rifted margin along the western margin of Oaxaquia, associated with a thick succession of turbiditic siliciclastic rocks deposited in a submarine fan environment (Centeno-García, 2005; Centeno-García et al., 2008; Martini et al., 2009; Helbig et al., 2012a, 2012b), before subduction-related igneous activity was re-established by the Early Jurassic (Fig. 1A; Bartolini et al., 2003).

Study Area

Amarillo Unit (New Name)

The Amarillo unit occurs on the eastern side of the Totoltepec pluton, north of the road between Santo Domingo Tianguistengo and Santiago de Chazumba (Fig. 2). This unit was previously mapped as the Cosoltepec Formation (Ortega-Gutiérrez et al., 1999), which is now considered to be a composite of both Cambrian–Ordovician and Devonian–Carboniferous units (Talavera-Mendoza et al., 2005; Keppie et al., 2006a, 2008; Morales-Gámez et al., 2008; Ortega-Obregón et al., 2009). The Amarillo unit is thrust upon low-grade metasedimentary rocks of the Tecomate Formation (Fig. 2; Kirsch et al., 2012) and tectonically juxtaposed against the Totoltepec pluton along a normal fault. To the east, the unit is unconformably overlain by Jurassic clastic rocks (Morán-Zenteno et al., 1993). The Amarillo unit is primarily composed of interbedded metapelites and metapsammites that are intruded by mafic dikes. All rocks of the Amarillo unit have undergone intense, polyphase deformation under amphibolite-facies metamorphism, which renders measurement of a type section impossible. The metasedimentary rocks consist of quartz, biotite, muscovite, garnet, and accessory apatite, zircon, and opaque minerals. Secondary minerals include chlorite and calcite. The mafic dikes are made up of amphibole (ferrotschermakite: Table DR11), plagioclase, biotite, quartz, and accessory apatite and ilmenite. Local shearing of the mafic dikes is associated with retrogression to chlorite, epidote, calcite, and apatite. Seven metasedimentary and seven amphibolite samples were collected in the southern part of the Amarillo unit for geochemical, U-Pb, and 40Ar/39Ar geochronological analyses (for locations, see Fig. 2; Tables DR3 and DR4 [see footnote 1]). Furthermore, an ∼3 m thick, undeformed dike of andesitic composition that cuts the metasedimentary rocks in the eastern part of the Amarillo unit was sampled for geochemical and U-Pb geochronological analysis.

Tecomate Formation

The Tecomate Formation is a mildly metamorphosed, but intensely deformed clastic unit consisting of thinly bedded pelitic and psammitic sedimentary rocks, and a few marbles, conglomerates, and volcanic rocks (Keppie et al., 2004b; Sánchez-Zavala et al., 2004; Kirsch et al., 2012). Geochronological and geochemical data indicate that the Tecomate Formation is a composite unit of Pennsylvanian–Middle Permian age, the detritus of which was predominantly derived from a regional continental arc (Kirsch et al., 2012). Rocks of the Tecomate Formation are tectonically juxtaposed against the Carboniferous Salada unit along N-striking, dextral-normal faults and N-dipping shear zones in the western part of the study area (Fig. 2; Morales-Gámez et al., 2008). To the north, they are overlain by red beds of inferred Jurassic age (Malone et al., 2002). In the southern part of the study area, the Tecomate Formation is overthrust by the Totoltepec pluton, and in the northeast, it is overthrust by the Amarillo unit. To the south, the Tecomate Formation is inferred to structurally overlie rocks of the Cosoltepec Formation (Malone et al., 2002). Three metasedimentary rock samples from the Tecomate Formation were collected for 40Ar/39Ar geochronological analysis (for location, see Fig. 2; Table DR3 [see footnote 1]).

U-Pb Geochronology

Analytical Methods

An Amarillo unit metapelite sample (lat 18°15′25.26″N, long 97°46′33.66″W) and an undeformed andesitic dike (lat 18°14′45.59″N, long 97°45′41.87″W) intruding this unit were collected for U-Pb zircon dating and analyzed by LA-ICP-MS at the Laboratorio de Estudios Isotópicos (LEI), Centro de Geociencias, Universidad Nacional Autónoma de México (UNAM), Mexico. Zircons were extracted using standard mineral separation techniques, as described by Solari et al. (2007). For details of the analytical procedures, see Solari et al. (2010) and Kirsch et al. (2012).

In figures, tables, and results, 206Pb/238U ages are quoted for zircons younger than 1.0 Ga, whereas older grains are quoted using their 207Pb/206Pb ages (e.g., Gehrels et al., 2006). The latter ages become increasingly imprecise below 1.0 Ga due to small amounts of 207Pb. Zircon analyses with <10% normal and <5% reverse discordance are considered to be geologically meaningful (e.g., Harris et al., 2004; Dickinson and Gehrels, 2008; Gehrels, 2011) and are used to constrain the crystallization age of igneous rocks, and the maximum depositional age as well as source areas of metasedimentary rocks. Mean ages based on youngest detrital zircons are considered robust if they belong to a cluster of three or more zircons with similar ages (e.g., Gehrels et al., 2006).


The Amarillo unit metapelite TT-80, consisting of quartz, biotite, K-feldspar, and opaque minerals, yielded 89 concordant zircon analyses ranging from 334 ± 4 Ma to 3418 ± 32 Ma (Figs. 3A and 3B; Table DR2A [see footnote 1]). The most prominent population is defined by 20 grains between the ages of ca. 1243 and ca. 989 Ma. The second-largest population consists of 10 zircons of Cambrian age between ca. 544 and 518 Ma. Four smaller populations are defined by ages of ca. 1505–1421 (three grains), ca. 743–727 (four grains), ca. 641–633 (three grains), and ca. 342–335 Ma (three grains), respectively. The weighted mean age (incorporating both internal analytical and external systematic error) of the youngest cluster overlapping in age at 2σ, calculated using the DZ Age Pick program developed at the LaserChron Center of the University of Arizona (, is 339 ± 6 Ma (three grains), which provides an older limit for the time of deposition. A younger limit is defined by a 40Ar/39Ar hornblende age of 239 ± 2 Ma from this unit (TT-665; see following). However, considering that the Amarillo unit is in tectonic contact with the Upper Pennsylvanian–Middle Permian Tecomate Formation, which has very different lithological and metamorphic characteristics and in this area has yielded distinctly younger detrital zircon populations (Keppie et al., 2004b; Kirsch et al., 2012), the depositional age of the Amarillo unit is likely to be Mississippian–Pennsylvanian.

An andesitic dike (MH-91) intrudes rocks of the Amarillo unit. The andesite has a porphyritic texture and is composed of hornblende, plagioclase, quartz, and opaque minerals. Zircon analysis yielded 31 concordant ages between 165 ± 1 Ma and 1560 ± 27 Ma. The TuffZirc (Ludwig and Mundil, 2002) 206Pb/238U age, calculated from a coherent group of 10 analyses, is 177 ± 4 Ma (Figs. 3C and 3D; Table DR2B [see footnote 1]). Significantly older zircons, forming a prominent population at 1220–1095 Ma (nine grains), are likely to represent xenocrysts inherited from the host rock. The TuffZirc age of 177 ± 4 Ma is interpreted as the time of intrusion of the andesite.


The U-Pb sample from the Amarillo unit contains a major detrital zircon age population between ca. 989 Ma and 1243 Ma, which is within the range of ages documented from the adjacent Oaxacan Complex (Keppie et al., 2001, 2003; Solari et al., 2003, 2013). Whereas 500–550 Ma (12 analyses), 1400–1600 Ma (three analyses), and 1800–2500 Ma (four analyses) zircons could have been derived from Amazonia, Oaxaquia, and/or Laurentia, those with 700–950 Ma ages (16 zircons) can come only from Amazonia or Oaxaquia (Keppie et al., 2008). The source for the three zircons with ages between 445 and 479 Ma may be the rift-related granitoid plutons within the Acatlán Complex, which have yielded ages between 440 and 480 Ma (Keppie et al., 2008). Seven detrital zircons in the sample from the Amarillo unit are Late Devonian–Mississippian in age. These zircons could be derived from a postulated Devonian–Carboniferous arc on the western margin of the Mixteca terrane, later removed by subduction erosion (Proenza et al., 2004; Keppie et al., 2008, 2010, 2012; Galaz et al., 2013). The measured Th/U ratios (Table DR3A [see footnote 1]), which are >0.01, support a magmatic origin for these zircons (e.g., Rubatto, 2002).

40Ar/39Ar Geochronology

Analytical Methods

Nine samples from rocks of the field area were collected for 40Ar/39Ar analyses. Amphibole was separated from amphibolites of the Amarillo unit (TT-665) and the Cosoltepec Formation (COS-589–1), respectively. Muscovite was extracted from metasedimentary rocks of the Tecomate Formation (TT-6, TT-61a, TT-91) and the Providencia shear zone (COS-1, PET-719-2). Biotite was separated from Amarillo unit metapelites (TT-29, TT-45b). Mineral grains were concentrated using standard techniques, cleaned in an ultrasonic bath, and subsequently handpicked to ensure >99.9% purity. For muscovite and biotite, grain sizes typically ranged from 250 to 420 µm (see Fig. 4), except for mylonitic sample TT-61a and metapelite sample TT-45b (both 177–250 µm). The amphibolites were finer grained and yielded grain sizes between 149 and 200 µm. Mineral separates and flux monitors were irradiated in the McMaster Nuclear Reactor (Hamilton, Ontario), and 40Ar/39Ar analyses were performed at the 40Ar/39Ar Geochronology Research Laboratory at Queen’s University in Kingston, Canada (for details of the analytical procedure, see Cubley et al., 2013). Complete results of the 40Ar/39Ar analyses are presented in Table DR3, with representative microprobe analyses of amphibole listed in Table DR1 (see footnote 1). All data have been corrected for blanks, mass discrimination, and neutron-induced interferences. The dates are referenced to the Hb3Gr hornblende standard at 1072 Ma (Turner et al., 1971; Roddick, 1983). Errors shown in the tables and on the age spectra and inverse isotope correlation diagrams represent the analytical precision at 2σ. This is suitable for comparing within-spectrum variation and determining which steps form a plateau (e.g., McDougall and Harrison, 1988, p. 89). For the purposes of this paper, a plateau is defined as three or more contiguous steps containing >50% of the 39Ar released, with a probability of fit >0.01 and mean square of weighted deviates (MSWD) <2. Plateau ages were calculated using ISOPLOT v. 3.60 (Ludwig, 2008).

Results and Interpretation

Age spectra.Ferrotschermakite (Table DR1A [see footnote 1]) separated from an Amarillo unit amphibolite (TT-665) yields an excellent plateau age of 239 ± 2 Ma (MSWD = 0.34), defined by 12 fractions and representing 94.7% of the total 39Ar released (Fig. 4A). The first three steps are associated with high atmospheric contamination and some contributions of Ar from other minor contaminating phases (Table DR3B [see footnote 1]). Given a grain size of 149–177 µm and a cooling rate of 15 °C/m.y., the plateau age is calculated to date cooling through 509 ± 3 °C (Dodson, 1973).

Tschermakitic hornblende (Table DR1B [see footnote 1]) from an amphibolite sample (COS-589-1) belonging to the Cosoltepec Formation in its type area yields a well-behaved 40Ar/39Ar age spectrum with a plateau age of 241 ± 9 Ma defined by nine fractions, representing 99.5% of the total 39Ar released (Fig. 4B). The first step is associated with high atmospheric contamination (Table DR3C [see footnote 1]). Individual plateau segments have relatively high errors due to the small amount of gas released in these fractions. The sample is calculated to date cooling through 509 ± 3 °C.

Muscovite from Tecomate Formation sample TT-6 yields a good plateau age of 239 ± 1 Ma (MSWD = 0.52), defined by five fractions and representing 66.7% of the total 39Ar released (Fig. 4C). The first four steps are associated with high atmospheric contamination (Table DR3D [see footnote 1]). The closure temperature for muscovite from this sample is calculated to 390 ± 8 °C (Hames and Bowring, 1994). A duplicate of sample TT-6 yields a virtually identical plateau age of 239 ± 1 Ma (MSWD = 0.44), defined by nine fractions and representing 77.1% of the total 39Ar released (Fig. 4C; Table DR3E [see footnote 1]). The slight increase in the apparent age of the last step in the age spectrum of this sample is associated with some mineral contamination.

Whole-rock sample TT-91 from a muscovite-rich metapsammite from the Tecomate Formation yields an excellent plateau age of 233 ± 2 Ma (MSWD = 0.53), defined by 11 steps containing 92.7% of the total 39Ar released (Fig. 4E; Table DR3F [see footnote 1]) and inferred to date cooling through 390 ± 8 °C.

The 40Ar/39Ar age spectrum of foliation-parallel white mica within a mylonite sample (TT-61a) taken from the shear zone between the Totoltepec pluton and the Tecomate Formation displays an apparent Ar-loss profile, in which the apparent ages climb over four steps from ca. 138 Ma to a very good plateau age of 232 ± 1 Ma (MSWD = 0.23), defined by 10 fractions and representing 80.1% of the total 39Ar released (Fig. 4E). The first fraction is associated with high atmospheric contamination. The last four steps have large errors due to the small amount of gas released in these fractions (Table DR3G [see footnote 1]). The plateau age is inferred to date cooling through 376 ± 5 °C.

Muscovite (PET-719-2) was separated from a mylonite in the Providencia shear zone, which is a large, NE-dipping shear zone with thrust kinematics containing various tectonic fragments of different units, and juxtaposing metasedimentary rocks of the Chazumba Lithodeme against rocks of the Cosoltepec Formation (Helbig et al., 2012a, 2012b). The 40Ar/39Ar age spectrum is partly disturbed (Fig. 4F; Table DR3H [see footnote 1]). The initial step and the eighth step are associated with high atmospheric contamination. The plateau age of 219 ± 1 Ma is interpreted to date cooling through 376 ± 5 °C.

Biotite from an Amarillo unit metapelite (TT-29) displays a slightly disturbed age spectrum in which the apparent ages increase over five steps from ca. 80 Ma to a maximum apparent age of ca. 236 Ma. The apparent ages then decrease to a plateau age of 232 ± 1 Ma (MSWD = 1.7), composed of steps 7–14 with 53.5% of the total 39Ar released (Fig. 4G; Table DR3I [see footnote 1]). The plateau age is interpreted to represent cooling through 318 ± 7 °C (Dodson, 1973).

Biotite separate TT-45b from an Amarillo unit metapelite yields a similar age spectrum as TT-29. The apparent ages climb from ca. 93 Ma over six steps to a maximum apparent age of ca. 237 Ma, before dropping to a well-defined plateau age of 234 ± 1 Ma (MSWD = 1.0), defined by eight fractions and representing 71% of the total 39Ar released (Fig. 4H). The first six steps are associated with high atmospheric contamination and some contributions from other minor contaminating phases (Table DR3J [see footnote 1]). Due to the smaller grain size of this sample (177–250 µm), its plateau age is inferred to date cooling through 306 ± 5 °C (Dodson, 1973). Slightly larger biotite grains from the same sample (TT-45b duplicate) yield slightly lower 40Ar volumes (Table DR3K [see footnote 1]) and show an apparent Ar-loss profile (Fig. 4H), in which the apparent ages increase monotonically from ca. 117 Ma over five steps to a plateau age of 227 ± 1 Ma (MSWD = 1.04), composed of steps 6–11 with 60.9% of the total 39Ar released. The first three steps are associated with high atmospheric contamination and some contributions of Ar from other minor contaminating phases. The plateau age is inferred to date cooling through 318 ± 7 °C.

Cooling rates.The 40Ar/39Ar data for the Tecomate Formation yield muscovite and whole-rock ages between 239 ± 1 Ma and 232 ± 1 Ma (Fig. 5). Amphibole and biotite from the adjacent Amarillo unit exhibit a similar 40Ar/39Ar age interval between 239 ± 2 Ma and 227 ± 1 Ma. What is more, Amarillo unit tschermakite TT-665 shows an identical 40Ar/39Ar age as Tecomate Formation muscovite TT-6 and its duplicate. The fact that amphibole and muscovite from two units of contrasting metamorphic grade (i.e., the amphibolite-grade Amarillo unit vs. the greenschist-grade Tecomate Formation) yield the same cooling age may be explained by synchronous exhumation, with the Amarillo unit amphibole passing through the 525 °C isotherm at about the same time as Tecomate Formation muscovite higher up in the crustal stack passed through the 400 °C isotherm. In a simplified scenario assuming a linear temperature-time (T-t) evolution, a cooling rate of ∼15 °C/m.y. can collectively be derived for the two units (Fig. 5). Additional petrological data are needed to better constrain the thermal evolution of the eastern Acatlán rocks.

Two samples from the Providencia shear zone (PET-719-2 and COS-1 muscovite, the latter recalculated from Keppie et al., 2004a; Fig. 4K) exhibit 40Ar/39Ar cooling ages that are similar to those of the Tecomate Formation and the Amarillo unit, which suggests active shearing during the Middle–Late Triassic. Further evidence for active thrusting is provided by the 40Ar/39Ar age of 232 ± 1 Ma for muscovite from the shear zone between the Totoltepec pluton and the Tecomate Formation.


Analytical Procedure

In order to determine the tectonic setting for rocks of the Amarillo unit, seven meta-igneous and seven metasedimentary samples were analyzed for major and selected trace elements by X-ray fluorescence (XRF) at the Regional Geochemical Centre, St. Mary’s University, Canada. Of these, six meta-igneous and four metasedimentary samples were selected for analysis of additional trace elements (rare earth elements [REEs], Y, Zr, Nb, Ba, Hf, Ta, and Th) by ICP-MS. Analytical procedures, uncertainties, and precision of the XRF and ICP-MS analyses are detailed in Dostal et al. (1986) and Longerich et al. (1990), respectively. Furthermore, Sm-Nd isotopic compositions of these 10 samples were determined in order to characterize the source and tectonic history of the Amarillo unit. Details of the analytical procedures for the Sm-Nd isotopic analyses are given in Kerr et al. (1995) and Kirsch et al. (2012). Additional lithogeochemical and Sm-Nd isotopic data for two samples of mafic dikes intruding the Carboniferous Salada unit in the western part of the study area are presented in the subsequent section on comparative geochemistry.

Element Mobility

The samples are affected to varying degrees by secondary processes, including amphibolite-facies metamorphism and deuteric alteration. These secondary processes have modified their primary chemical composition, producing elevated loss on ignition (LOI) values and scattered patterns on diagrams featuring K, Rb, and possibly Sr. Hence, inferences about the petrogenesis and tectonic setting of the rocks are largely based on high field strength elements (Ti, Zr, Hf, Nb, Ta, Th, Y, REEs) and some transition elements (Ni, V, and Cr), which are considered to be relatively “immobile” during alteration processes (e.g., Winchester and Floyd, 1977; Pearce, 1996). The coherence of the geochemical data and their similarity to those of fresh, unmetamorphosed rocks suggest that they reflect original protolith compositions.

Amarillo Unit Amphibolites

The amphibolites of the Amarillo unit range in SiO2 from 48 to 57 wt% (LOI-free) with variable Mg# (mole MgO/[MgO + FeOt]) between 0.43 and 0.83 (Fig. DR1 [see footnote 1]). On Harker variation diagrams of Mg#, the samples show a negative trend for P, Zr, and REEs (Fig. DR1), indicating enrichment in these elements during fractionation. Ni and Cr display a positive correlation with Mg#, suggesting that these elements were depleted during differentiation (Fig. DR1). One sample (TT-45A) is characterized by low TiO2 (0.4 wt%), Al2O3 (7.8 wt%), and Zr (22 ppm) abundances, and high contents of the transition elements Cr (2517 ppm) and Ni (1136 ppm). Geochemically, the amphibolites correspond to ortho-amphibolites (Fig. 6A), derived from a magmatic parent of subalkaline basaltic to basaltic andesitic composition (Fig. 6B). The FeOt/(FeOt + MgO) versus SiO2 diagram (Miyashiro, 1974; Fig. 6C) illustrates the tholeiitic affinity of the amphibolites. Ratios of Ti/Zr and Cr/Y are variable, but generally similar to mid-ocean-ridge basalt (MORB) values (Figs. 7A and 7B). The chondrite-normalized REE patterns are flat to distinctly enriched in light REEs (LREEs), with (La/Yb)n ranging from 1.2 to 6.2, and (La/Sm)n ranging between 0.9 and 2.7 (Fig. 8A). The heavy REEs (HREEs) are flat, with (Gd/Yb)n varying between 1.3 and 1.7. The two samples with the most LREE-enriched patterns display slight negative Eu anomalies (Eu/Eu* = 0.85–0.78), indicating minor plagioclase fractionation. The trace-element patterns of the amphibolites are similar to enriched (E-)MORB (Figs. 8B), yielding relatively flat patterns, but showing an enrichment in U and Pb as well as large ion lithophile elements (LILEs) such as K, Rb, and Cs. This enrichment is typical for magmas influenced by oceanic-slab–derived fluids in subduction-zone settings (e.g., Bebout, 2014).

The Sm-Nd isotopic signature of the Amarillo unit amphibolites is variable, with initial ɛNd(t) values between -3.9 and +7.6 (t = 339 Ma), 147Sm/144Nd from 0.13 to 0.20, and depleted mantle model ages (TDM; only considered if 147Sm/144Nd < 0.165; see Stern, 2002) between 1.40 Ga and 0.33 Ga (Fig. 9A). Three samples (TT-30, TT-661, and TT-682) have initial ɛNd(t) values that roughly correspond to the contemporary depleted mantle. However, their 147Sm/144Nd values are lower than expected for the depleted mantle (Fig. 9B), suggesting a source that was slightly enriched in Nd over Sm. This interpretation is consistent with the E-MORB affinity of these samples on tectonic discrimination diagrams (e.g., Fig. 10C). Sample TT-47 has a marginally lower initial ɛNd(t) of +5.2, as well as 147Sm/144Nd and Nb/La ratios that are similar to those of mafic rocks interpreted to have been derived from an old, ca. 1.0 Ga subcontinental lithospheric mantle source (e.g., Murphy et al., 2006; Figs. 9B and 9C). The remaining two samples (TT-48a and TT-664) have much lower ɛNd(t) values than expected for juvenile magmas from a depleted mantle source. The latter samples also display higher TDM ages (1.3–1.4 Ga), as well as high Ce/Yb (13.7–18.1; Fig. 10A), Th/Yb (0.61–0.69; Fig. 10B), and Th/Nb (0.11; Fig. 10C), and low Nb/Lacn ratios (0.69–0.82; Fig. 9C), suggesting that these rocks were affected by crustal contamination. However, the fact that they have a basaltic to andesitic composition indicates that crustal contamination was relatively minor, and that these samples, too, were probably derived from an older enriched mantle source.

Amarillo Unit Metasedimentary Rocks

The sampled metasedimentary rocks from the Amarillo unit include five metapelites and two metapsammites. Their major-element abundances correspond to those of typical shales and sandstones, with SiO2 ranging between 50 wt% and 67 wt% and Al2O3 from 16 wt% to 25 wt% (LOI-free basis). SiO2 exhibits negative correlations with TiO2, FeOt, and MnO, as well as with the transition metals Cr, Ni, and Cu (Table DR4B [see footnote 1]), which result from different proportions of clay- and quartz-rich components, as reported from other sedimentary successions (e.g., Bhatia, 1983). The rocks display Th (7–15 ppm), Hf (3–8 ppm), and Zr (61–386 ppm) abundances that correspond to upper continental crust (UCC) values (Fig. 11A; Taylor and McLennan, 1995) and are typical for clastic sedimentary rocks from a felsic source (Feng and Kerrich, 1990). On average, abundances of MgO (1.8–4.4 wt%), Fe2O3 (6–11 wt%), Cr (21–384 ppm), Ni (15–81 ppm), and Co (11–41 ppm) are slightly higher than UCC values, but are within the range for clastic sedimentary rocks derived from a felsic source (Feng and Kerrich, 1990). Further evidence for a silicic provenance includes the ranges in Al2O3/TiO2 (14–28), Cr/Th (1.8–13.9), Th/Co (0.3–1.0), and La/Co (1.4–2.8) displayed by the samples (Cullers, 1994, 2000; Girty et al., 1996). A silicic provenance is consistent with the samples’ chondrite-normalized REE patterns, which display moderate LREE enrichment (La/Ybn = 5.1–9.7), negative Eu anomalies (Eu/Eu* = 0.65–0.77), and ΣREE between 113 and 209 ppm (Fig. 11B). On tectonic discrimination diagrams based on La/Th versus Hf (Floyd and Leveridge, 1987; Fig. 12A) and Al2O3/SiO2 versus (Fe2O3 +MgO) (Bhatia, 1983; Fig. 12B), the samples plot in the magmatic arc field. A subduction signature is also evident in the spidergrams normalized to UCC (Fig. 11A), which display distinctly negative Ta anomalies (Pearce, 1983; Kelemen et al., 2014).

The Sm-Nd isotopic signature of the Amarillo unit clastic rocks, with ɛNd(t) between -12.2 and -4.5 (t = 339 Ma), 147Sm/144Nd from 0.1159 to 0.1377, and TDM between 1.27 Ga and 1.60 Ga (Fig. 9A), reflects derivation from a mixture of ancient and juvenile crustal sources (e.g., Arndt and Goldstein, 1987; Murphy and Nance, 2002). This interpretation is consistent with detrital zircon data that identify the Oaxacan Complex and a regional Devonian–Carboniferous arc as the main contributing source area for metasedimentary rocks of the Amarillo unit (Fig. 3A).

Comparative Geochemistry

In order to assess their regional significance, we compared rocks of the Amarillo unit with other Carboniferous meta-igneous and metasedimentary rocks in the eastern and southwestern parts of the Acatlán Complex. The comparison includes geochemical and isotopic data from (1) low-grade metasedimentary rocks and intercalated mafic dikes of the Salada unit (Fig. 2, Morales-Gámez et al., 2009b; Dostal and Keppie, 2009; this paper), which have a depositional age bracketed between the youngest detrital zircon of 352 ± 3 Ma and a 323 ± 3 Ma metamorphic muscovite age (Morales-Gámez et al., 2008, 2009a); (2) low-grade metabasalts of the meta-igneous to quartzitic Coatlaco unit, the maximum depositional age of which is constrained by a youngest detrital zircon population of 357 ± 35 Ma (Grodzicki et al., 2008); (3) metabasalts of the Progreso unit (Ortega-Obregón et al., 2010a), which are bracketed by a 403 ± 7 Ma mean of the youngest detrital zircon population and a 335 ± 2 Ma 40Ar/39Ar metamorphic muscovite age (Ortega-Obregón et al., 2009); and (4) sub-greenschist-facies clastic rocks of the Upper Mississippian to Lower Permian Zumpango unit (Fig. 1B, Ortega-Obregón et al., 2009, 2010a).

The meta-igneous rocks of the Amarillo unit and those of the Salada, Coatlaco, and Progreso units show very similar geochemical characteristics. However, amphibolites from the Amarillo unit display a higher compositional variability, e.g., for Mg#, P2O5, Ni, Cr, Zr, and Nb (Fig. DR1 [see footnote 1]). Except for two of the Amarillo unit amphibolite samples that show a more intermediate/alkaline composition (Fig. 6B), Zr/Ti, Nb/Y, and Cr/Y ratios are similar for all meta-igneous suites (Fig. 7). Like the Amarillo unit amphibolites, the trace-element patterns of the Coatlaco and Progreso unit metabasalts bear a greater resemblance to E-MORB than normal (N-)MORB, unlike the Salada unit mafic dikes, the composition of which is more similar to N-MORB (Fig. 8B). The probable derivation from an enriched mantle source for the Coatlaco, Progreso, and Amarillo unit metabasalts is also evident from the Ce/Yb and Ta/Yb ratios and Th-Hf-Nb characteristics (Fig. 10). The Sm-Nd isotopic composition of the Salada unit mafic dikes (our data; Table DR4A [see footnote 1]) is characterized by ɛNd(t) values between +7.6 and +8.5 (t = 352 Ma), and 147Sm/144Nd ratios from 0.2207 and 0.2222 (Table DR4A [see footnote 1]; Fig. 9). These ɛNd(t) values are higher than the contemporary depleted mantle, suggesting an ultradepleted mantle source that had undergone previous melting (Murphy et al., 2011, 2013). Two of the metabasalts from the Progreso unit have a similarly ultradepleted Sm-Nd signature (Ortega-Obregón et al., 2010a; Fig. 9A). Other Progreso unit basalts, with ɛNd(t) values between +6.2 and +6.7 (t = 330 Ma), 147Sm/144Nd values slightly lower compared to the depleted mantle (Fig. 9B), and no signs of crustal contamination (Figs. 9C and 10), suggest derivation from a slightly enriched mantle source, as inferred for three of the Amarillo unit amphibolites. The Salada unit mafic dikes have been interpreted to represent continental tholeiites emplaced in thin crust (Morales-Gámez et al., 2009b). Based on their geochemical and Sm-Nd isotopic signature, both the Progreso unit metabasalts and the Amarillo unit amphibolites may similarly have been emplaced in a continental extensional setting, but they were most likely derived from a more enriched mantle reservoir of different ages.

Compared to the Salada and Zumpango unit metasedimentary rocks, clastic rocks of the Amarillo unit display only subtle compositional differences, such as lower Ta, greater variability in Zr and Hf, and, on average, higher V, Mn, Mg, and Cr, best displayed on the extended spiderplot normalized to UCC (Fig. 11A). This signature is interpreted to reflect a relatively greater influence of arc-derived detritus in samples of the Amarillo unit, which is also evident on the Al2O3/SiO2 versus (Fe2O3 + MgO) discrimination diagram (Bhatia, 1983; Fig. 12B). The detrital zircon analyses of the metapsammite sample from the Salada unit (Morales-Gámez et al., 2008), which yielded only 1/7 the amount of (arc-derived) Devonian–Carboniferous zircons compared to those found in the Amarillo unit metapelite, are consistent with this explanation. The Sm-Nd isotopic compositions of the Zumpango unit clastic rocks are more homogeneous than those of the Amarillo unit (Fig. 9), which may indicate a similar drainage system throughout deposition or source areas with an isotopically uniform composition.

Carboniferous Deposition and Magmatism

The Devonian to Permian tectonothermal history of the Acatlán Complex was predominantly influenced by subduction-related processes along the western margin of present-day southern Mexico (Proenza et al., 2004; Keppie et al., 2008, 2010, 2012; Kirsch et al., 2012; Galaz et al., 2013). Late Devonian–Mississippian detrital zircons of magmatic origin recovered in Amarillo unit metasedimentary rocks, a Sm-Nd isotopic signature indicating derivation from ancient and juvenile crustal sources, and geochemical characteristics typical of a felsic to intermediate arc-related provenance suggest that the Amarillo unit was deposited in an arc environment.

On the basis of similar age and geochemical characteristics, the metasedimentary rocks of the Amarillo unit may be correlated with those of the Salada unit on the western side of the San Jerónimo fault (Morales-Gámez et al., 2008; Fig. 2) and the Zumpango unit in the southwestern part of the Acatlán Complex (Ortega-Obregón et al., 2009). In agreement with the scenario of a regional Devonian–Permian arc within the Mixteca terrane, the intrusion of the Amarillo unit amphibolites and their potential correlatives, such as the Salada unit mafic dikes (Morales-Gámez et al., 2008), as well as the Coatlaco and Progreso unit metabasalts (Grodzicki et al., 2008; Ortega-Obregón et al., 2009, 2010a), could be associated with intra-arc extension, which, in turn, may have been related to the exhumation of high-pressure rocks (Keppie et al., 2008, 2010, 2012; Ramos-Arias et al., 2008, 2012), and/or transtensional deformation as a consequence of oblique subduction (e.g., Fitch, 1972; Tobisch and Cruden, 1995; Kirsch et al., 2013). The geochemical and isotopic inter- and intrasuite variability displayed by the meta-igneous rocks may reflect local heterogeneities in the composition of the mantle source and differences in the thickness of the crust into which these mafic magmas were emplaced. Alternatively, the contrasting geochemical characteristics may represent a temporal sequence, documenting the stages of an intracontinental rift development, involving the progressive thinning of the crust and the gradual replacement of an old subcontinental lithospheric mantle by a juvenile mantle. Further geochronological work is needed to determine the validity of these respective models.

Mechanism for the Middle–Late Triassic Tectonothermal Event

The 40Ar/39Ar ages of minerals record cooling below their mineral-specific closure temperatures (Dodson, 1973), and the corresponding cooling ages either reflect retrograde closure or overprinting effects (e.g., Armstrong et al., 1966). Due to the absence of Triassic magmatic rocks in the area, the nature of the 40Ar/39Ar age spectra, and the existence of well-developed 40Ar/39Ar age plateaus in most cases, we consider a reheating event, e.g., associated with an igneous intrusion, to be unlikely. Rather, the obtained 40Ar/39Ar ages are entirely consistent with cooling by exhumation due to tectonic uplift and/or denudation. The 40Ar/39Ar samples derived from thrusts between the Totoltepec pluton and the Tecomate Formation (TT-61a), as well as samples derived from the Providencia shear zone (PET-719-2; COS-1), suggest the importance of shearing, compressional tectonics, and crustal thickening to achieve tectonic uplift.

The Carboniferous to Permian tectonothermal history of the Mexican paleo–Pacific margin was characterized by subduction-related activity, which is well documented on the basis of geochronological and geochemical data of igneous rocks and from the detrital zircon record of volcaniclastic sedimentary sequences (Kirsch et al., 2012, and references therein). There is also ample evidence for Early–Middle Jurassic arc activity in Mexico (“Nazas arc”; Fig. 1A; e.g., Barboza-Gudiño et al., 2004, 2008; Campa-Uranga et al., 2004; Fastovsky et al., 2005; Zavala-Monsiváis et al., 2009, 2012; Godínez-Urban et al., 2011). The 177 ± 4 Ma andesitic dike intruding the Amarillo unit metasedimentary rocks is interpreted as a Jurassic Nazas arc representative. In between Permian and Jurassic phases of arc magmatism, i.e., during the Triassic, however, there is little evidence for magmatic arc activity, at least in the south-central Mexican terranes. Although often referred to as a “Permo-Triassic arc” in the literature (e.g., Torres et al., 1999), the temporal continuation of the Carboniferous–Permian arc into the Triassic is to a large part based on K-Ar and Rb-Sr ages (e.g., Damon et al., 1981; Torres et al., 1999; Schaaf et al., 2002), which are known to be less reliable indicators of the time of pluton crystallization than ages obtained by U-Pb zircon geochronology (e.g., Steiner and Walker, 1996). The only Late Permian to Middle Triassic igneous rocks for which U-Pb zircon ages are available include the 258 ± 2 Ma Zanatengo River orthogneiss and the 251 ± 2 Ma Buenavista augen gneiss in the Chiapas Massif (Weber et al., 2005), the 255 ± 1 Ma Etla granite (Ortega-Obregón et al., 2013) and the 254 ± 7 Ma Mixtequita stock (Murillo-Muñetón, 1994) in the Guichicovi Complex, and the calc-alkaline 240 ± 3 Ma Atolotitlán felsite in the Matzitzi Formation (Centeno-García et al., 2009; Elías-Herrera et al., 2011). All of these occurrences are felsic rocks, which are unreliable indicators of arc magmatism (e.g., Kuscu et al., 2010). The age distribution of late Paleozoic to early Mesozoic igneous rocks resembles that of detrital zircons from late Mesozoic, continental clastic rocks in Oaxaquia (Pérez-Gutiérrez et al., 2009a; Mendoza-Rosales et al., 2010; Rubio-Cisneros and Lawton, 2011; Barboza-Gudiño et al., 2012), Guerrero and Mixteca (Talavera-Mendoza et al., 2007, 2013; Ortega-Flores et al., 2013), and Chiapas (Godínez-Urban et al., 2011). Both of these data sets (Fig. 13) show low data density in the period between ca. 235 and 215 Ma (see also Lawton et al., 2010). Assuming both the igneous and sedimentary rocks included in this compilation are arc related, the relative abundance of age data may be used as a proxy for relative magmatic flux rates of the arc (e.g., Laskowski et al., 2013). The observed paucity of Middle–Late Triassic igneous rocks and detrital zircons in central and southern Mexican terranes suggests a relative lull in regional arc magmatism. Moreover, the 40Ar/39Ar cooling ages from the eastern Acatlán Complex roughly correspond to this interval of subdued magmatic activity (Fig. 13), suggesting a possible link between the temporary extinction of arc magmatism and a Middle- to Late Triassic cooling/uplift event of late Paleozoic rocks in the eastern Acatlán Complex.

The main factor that influences the intensity of magmatic activity in continental arcs is the geometry of the subducting plate (e.g., Cross and Pilger, 1982). Additionally, mantle fertility may play an important role (e.g., Ducea, 2001). The most obvious mechanism to explain cooling/uplift during a period of quiescence in arc magmatism is a flattening of the subducting slab. Low-angle subduction will lead to increased interplate coupling, cessation of arc magmatism, and horizontal shortening and uplift of the forearc region (e.g., Espurt et al., 2008). Along the Cordilleran margin, spatial gaps in arc magmatism (e.g., Kay and Coira, 2009) have been explained by flat-slab subduction, associated with changes in plate velocity (Lallemand et al., 2005), resistance to plate bending (e.g., Schellart et al., 2007; Schellart, 2008), subducted bathymetric anomalies (McGeary et al., 1985; Gutscher et al., 2000), or mantle hydration (Billen and Gurnis, 2001; Manea and Gurnis, 2007). Apart from these external controls on subduction-zone geometry, there may be an overarching cyclic process governing the evolution of Cordilleran-type orogenic systems (e.g., DeCelles et al., 2009; D.F. Keppie et al., 2009). It is beyond the scope of this paper to discuss the causes of flat slabs, but it is a viable mechanism to explain the tectonothermal event in the eastern Acatlán Complex associated with a regional gap in arc magmatism.

Paleogeographic Implications

A steadily growing geologic database for the Mixteca and related south-central Mexican terranes shows a temporal variation in arc activity and style (Fig. 13), from a late Paleozoic history dominated by plentiful arc magmatism (Kirsch et al., 2012, and references therein) and local intra-arc extension (Ramos-Arias et al., 2008; Keppie et al., 2012; Kirsch et al., 2013; this paper) to an early Mesozoic transient episode characterized by subdued magmatic arc activity (see references listed earlier) and local compression and uplift (40Ar/39Ar cooling ages; Triassic thrusting along the southern margin of the Totoltepec pluton and along the Providencia shear zone). In the same way that a low relative magma flux in conjunction with upper-plate compression and uplift of rocks may be related to shallow subduction, a steeply dipping slab can be inferred for high magma fluxes combined with local upper-plate extension. The periodic variation in the angle of the subducting slab (e.g., tectonic switching: Collins, 2002) occurs in all modern accretionary orogens, such as the Andes (Gutscher et al., 2000). In the Oaxaquia and Mixteca terranes, Late Triassic to Early Jurassic magmatism and the deposition of siliciclastic rocks with a passive-margin signature (Potosí fan; Silva-Romo et al., 2000; Centeno-García 2005, Centeno-García et al., 2008) could be interpreted as the end of the Middle–Late Triassic flat-slab subduction episode. Polydeformed, amphibolite-facies metasedimentary rocks of the Ayú Complex, an inferred southern correlative of the Potosí fan (Helbig et al., 2012a), contain amphibolites for which geochemistry indicates a backarc setting. Late Triassic to Early Jurassic backarc extension may have been regionally widespread, as Triassic–Jurassic mafic rocks of the Francisco Gneiss (Valencia-Moreno et al., 2001; Keppie et al., 2006b) and the Juchatengo Complex (Grajales-Nishimura et al., 1999) exhibit similar backarc or within-plate geochemical affinities (Helbig et al., 2012a). Middle Jurassic intra-arc basin formation is also inferred for the lower part of the Todos Santos Formation in Chiapas (Godínez-Urban et al., 2011).

In addition to a temporal variation, the geometry of the subducting slab may have varied spatially along the Cordilleran margin. Between ca. 248 Ma to 256 Ma, i.e., before the onset of the shallow subduction that affected the Acatlán Complex, compressional tectonics associated with a medium- to high-grade metamorphic event in the Chiapas Massif has been explained as resulting from an episode of shallow subduction (Weber et al., 2007). In Guatemala, arc-related orthogneisses in the Chuacús metamorphic complex, the early Mesozoic paleogeographical position of which is not well constrained, have yielded LA-ICP-MS U-Pb zircon ages between 226 Ma and 218 Ma (Solari et al., 2011), suggesting arc magmatism during an interval of magmatic quiescence in south-central Mexico. In the southwestern United States and northern Mexico, arc-related igneous rocks with U-Pb thermal ionization mass spectrometry and sensitive high-resolution ion microprobe crystallization ages between ca. 260 Ma and 210 Ma in southern California and western Arizona (Miller et al., 1995; Barth et al., 1997; Walker et al., 2002; Barth and Wooden, 2006), between ca. 284 and 221 Ma in Sonora (Arvizu et al., 2009; Arvizu Gutiérrez, 2012), and between ca. 238 and 220 Ma in Coahuila (Jones et al., 1995; McKee et al., 1999; González-León et al., 2009) suggest that magmatic arc activity here extended well into the Triassic. This conclusion is supported by the detrital zircon record from arc-derived sedimentary sequences in Arizona, Nevada, New Mexico, and Texas (Manuszak et al., 2000; Fox et al., 2005, 2006; Dickinson and Gehrels, 2008; Mackey et al., 2012; Riggs et al., 2012), Sonora (Gehrels and Stewart, 1998; González-León et al., 2005, 2009; Riggs et al., 2009; Mauel et al., 2011; Peryam et al., 2011), and Baja California (Alsleben et al., 2012). These data (Fig. 13) show a distribution of magmatic lulls and peaks that is shifted toward younger ages in comparison to the data for south-central Mexico, suggesting that the variation between steep- and flat-slab subduction was spatially diachronous along the active North American Cordilleran margin. This feature is common in modern arc systems such as the Andes (e.g., Ramos and Folguera, 2008).

We acknowledge the Consejo Nacional de Ciencia y Tecnología (CONACyT; Project CB-2005–1: 24894), the Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica (PAPIIT: IN100108–3), and the Natural Sciences and Engineering Research Council of Canada (NSERC). Discovery grants to Murphy for funds to support the field work and geochemical and isotopic analyses. Carlos Ortega-Obregón and Ofelia Pérez-Arvizu provided technical assistance in the Laboratorio de Estudios Isotópicos, Centro de Geociencias. Kirsch thanks Roberto Molina for helpful discussions, which improved the manuscript. An NSERC Discovery grant to Lee for support of the 40Ar/39Ar Geochronology Research Laboratory at Queen’s University in Kingston, Canada, is also gratefully acknowledged, as is the technical assistance provided by Douglas A. Archibald and Jessica Pickett. Amabel Ortega-Rivera conducted the majority of the 40Ar/39Ar measurements. Uwe Martens and Brian Joy facilitated the microprobe analyses. We thank Lithosphere Science Editor Arlo B. Weil, and three anonymous reviewers for constructive comments on the manuscript. This is a contribution to International Geological Correlation Project 597.

1GSA Data Repository Item 2014109, Figure DR1 and Tables DR1–DR4, is available at, or on request from, Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.