This study documents an early Cenozoic continental high-pressure (HP) metamorphic complex along the Yarlung (India-Asia) suture zone in southern Tibet. The complex is exposed in the Lopu Range, located ∼600 km west of the city of Lhasa. HP rocks in the core of the complex have Indian passive-margin (Tethyan Himalaya Sequence) protoliths and are exposed in the footwall of a top-to-the-north, normal-sense shear zone. Phengite geobarometry, Zr-in-rutile geothermometry, and pseudosection modeling indicate that these rocks reached pressures ≥1.4 GPa at temperatures ≤600 °C. A meta-Tethyan graywacke yielded a garnet Lu-Hf date of 40.4 ± 1.4 Ma, which is interpreted as the age of prograde metamorphism. Five Ar-Ar phengite ages between 39 and 34 Ma are interpreted to record the timing of exhumation to midcrustal depths (∼25 km) and concomitant retrogression. The structural geometry and pressure-temperature-time (P-T-t) history of Lopu Range rocks are similar to the Tso Morari and Kaghan Valley complexes, located >700 km to the northwest along the Indus suture zone. However, peak metamorphism and exhumation occurred ∼6 m.y. later in the Lopu Range, and no ultrahigh-pressure assemblages have been identified. We propose a tectonic model that involves steep subduction of the Tethyan Himalaya continental margin at ca. 40 Ma, initial exhumation of HP metasedimentary rocks at ca. 39 Ma, and subsequent northward underthrusting of Greater Indian lithosphere shutting off Gangdese arc magmatism at ca. 38 Ma.

Continental high-pressure (HP) to ultrahigh-pressure (UHP) metamorphic complexes indicate that continental crust can be subducted to mantle depths and exhumed along intercontinental suture zones during collision. Diverse geological observations from continental HP metamorphic complexes reveal a broad range of behaviors that may be modulated by orogenic stage (e.g., Hacker et al., 2013). Small (<5000 km2) complexes generally form early during collision and are characterized by short, <10 m.y. intervals between prograde metamorphism and midcrustal exhumation (i.e., short duration), whereas large complexes form later over longer time scales (Kylander-Clark et al., 2012). Various geodynamic models have been proposed to explain the array of behaviors that fall between these two end members (Hacker et al., 2013); at least six distinct mechanisms have been proposed to explain their exhumation (Ducea, 2016). Geologic investigation of HP metamorphic complexes provides constraints that guide the development of a unifying model of this process, which may play a key role in the formation and recycling of Earth’s continents. Continental HP metamorphic rocks are also a key record of the tectonic evolution of suture zones within individual orogens.

The Tso Morari and Kaghan Valley complexes in the northwestern Himalaya (Fig. 1) demonstrate that the Indian continental margin was being subducted to mantle depths during the Eocene (e.g., Guillot et al., 2008). The lack of recognized continental HP metamorphic complexes elsewhere along the >2000-km-long Indus-Yarlung (India-Asia) suture, however, makes it tenuous to extrapolate geochronologic, metamorphic petrology, and structural data from the northwestern Himalaya ∼1000 km along strike to southern Tibet—the geographical center of the Himalayan-Tibetan orogenic system (Fig. 1). For example, a lack of analogs in southern Tibet limits the significance of minimum collision ages deduced from the Tso Morari and Kaghan Valley complexes. Differences in structural geometry, fault kinematics, and/or timing of metamorphism between continental HP metamorphic complexes along strike might reflect variations in continental margin geometry, subduction dynamics, and/or timing.

In this study, we report geologic mapping, structural analysis, metamorphic petrology, garnet Lu-Hf geochronology, and phengite Ar-Ar data from the Lopu Range, located ∼700 km southeast of the Tso Morari complex (Fig. 1). Our results document the first-recognized HP metamorphic complex along the Yarlung portion of the India-Asia suture in southern Tibet. Our data indicate that Indian passive-margin rocks were metamorphosed by 40 ± 1 Ma, reached peak-pressure conditions of ≥1.4 GPa at temperatures ≤600 °C, and were exhumed along the Yarlung suture during Middle–Late Eocene time. The size, short duration between HP metamorphism and exhumation, and structural geometry of the Lopu Range are similar to other small, early-formed (during collision) HP metamorphic complexes globally, as well as at Tso Morari and Kaghan Valley. We propose a kinematic model of Indian continental subduction and exhumation in southern Tibet that integrates our observations with regional tectonic data.

Geology of Continental HP Metamorphic Complexes in the Northwestern Himalaya

HP metamorphic rocks at Tso Morari and Kaghan Valley are exposed in the core of antiformal domes and are juxtaposed against suture zone mélange and ophiolites in the footwall of mylonitic shear zones that act as roof faults (Guillot et al., 2008; Beaumont et al., 2009). Roof shear zones juxtapose HP thrust sheets in the footwall beneath lower-pressure mélange, ophiolite, and metasedimentary thrust sheets in the hanging wall. Where the roof fault dips toward the overriding plate, normal-sense kinematics are observed, suggesting synexhumational shearing (Epard and Steck, 2008; Guillot et al., 2008; Beaumont et al., 2009). Early-formed structures are overprinted by approximately N-S–striking, brittle normal faults that accomplished a late stage of upper-crustal exhumation likely related to regional, gravitationally driven, upper-crustal extension (Beaumont et al., 2009; Epard and Steck, 2008).

The HP thrust sheet in the Tso Morari complex (Fig. 1) consists of Neoproterozoic orthogneiss and overlying Lower Paleozoic Indian passive-margin (Tethyan Himalaya Sequence) rocks (de Sigoyer et al., 2004; Epard and Steck, 2008). Coesite-bearing eclogite lenses in granitic orthogneiss record peak-pressure conditions of 4.4–4.8 GPa at 560–760 °C (Wilke et al., 2015), whereas metasedimentary rocks record much lower peak-pressure conditions of ∼2.0 GPa at ∼550 °C (Guillot et al., 1997). Retrogression under blueschist-amphibolite-facies conditions was accompanied by the development of S2 foliation defined by phengite, chlorite, and staurolite (de Sigoyer, 1998). The HP to UHP thrust sheet in the Kaghan Valley metamorphic complex, located another ∼400 km along strike to the northwest (Fig. 1), consists of Proterozoic orthogneiss and overlying metasedimentary rocks including Neoproterozoic–Lower Paleozoic biotite metagraywacke with intercalated garnet and/or kyanite metapelites (Treloar et al., 2003) and Upper Paleozoic—Mesozoic calc-paragneiss and schist with intercalated Permian amphibolite (Spencer et al., 1995). Coesite-bearing eclogite experienced peak conditions of ∼3.0 GPa and 770 °C (O’Brien et al., 2001), whereas coesite-free rocks record peak conditions of ∼2.4 GPa and 610 °C (Lombardo et al., 2000). Retrogression took place under blueschist-amphibolite-facies and later greenschist-facies conditions (Guillot et al., 2008).

Eclogite-facies metamorphism had initiated by 47 Ma at both localities, based on in situ U/Th-Pb petrochronology and statistical evaluation of previously published data (Donaldson et al., 2013). Exhumation and concomitant retrogression occurred between 46.5 and 31 Ma at Kaghan Valley (Wilke et al., 2010) and between 47 and 34 Ma at Tso Morari (de Sigoyer et al., 2000; Schlup et al., 2003).

Geology of the Yarlung Suture Zone in the Lopu Range Region

The NW-SE–trending Lopu Range transects the Yarlung suture zone (Figs. 1 and 2). Near the Lopu Range, the northernmost unit is a belt of Eocene granodiorite within the Gangdese arc, which developed during northward subduction of Neo-Tethyan oceanic lithosphere (Schärer et al., 1984) and subsequent Tethyan Himalaya–Asia collision (Yin and Harrison, 2000). To the south, a narrow (∼5-km-wide), E-W–trending belt of gently south-dipping, nonmarine conglomerates unconformably overlies Gangdese arc rocks. We correlate these strata to the Oligocene–Miocene Kailas Formation (Gangrinboche conglomerates; Aitchison et al., 2002; DeCelles et al., 2011). Farther south, Cretaceous–Paleocene Xigaze forearc basin strata (Wang et al., 2006) are bounded to both the north and south by steeply dipping, top-to-the-north reverse faults that can be traced continuously along strike as strands of the Miocene Great Counter thrust system (Heim and Gansser, 1939; Yin et al., 1999; Sanchez et al., 2013). South of the forearc basin strata, there is a unit of sedimentary-matrix mélange that includes blocks of Neo-Tethys oceanic crust and Tethyan Himalayan strata, which likely formed as the Tethyan margin entered the trench and which is interpreted to be equivalent to the “wildflysch” of Burg and Chen (1984). Tectonic slivers of ophiolitic mélange are locally exposed between sedimentary-matrix mélange and forearc rocks (Figs. 1 and 2). Structurally beneath the sedimentary-matrix mélange, there is an approximately E-W–trending domal exposure of metasedimentary rocks (Figs. 1 and 2). Based on previous age assignments (Wang et al., 2006) and their U-Pb detrital zircon age signatures (Sanchez et al., 2013), the protoliths of these rocks are Paleozoic passive-margin strata of the Tethyan Himalaya Sequence.

Dajiling Fault

The oldest exposed structure in the Lopu Range is the Dajiling fault, which is a >2-km-thick mylonitic shear zone that juxtaposes sedimentary-matrix mélange (map unit mlg) in the hanging wall against meta-Tethyan schist and gneiss in the footwall (map unit Pzg; Fig. 2). We interpret the fault as the roof of a structural dome based on the continuity of footwall rocks and consistent kinematics (Fig. 1). Along the northern exposure, the fault dips ∼65° to the north and S-C fabrics, asymmetric boudins, and stretching lineations indicate top-to-the-north, normal-sense shear (Fig. 2). Asymmetric porphyroclasts and S-C fabrics observed in an oriented thin section from the fault zone similarly indicate top-to-the-N shear (Data Repository Fig. DR11). Along its southern exposure, the Dajiling fault dips moderately to the southwest, and stretching lineations and S-C fabrics similarly indicate top-to-the-north shear (Fig. 2). To the south, the low-grade Tethyan thrust sheet (Pp) is in the footwall of a poorly exposed, S-dipping fault with sedimentary-matrix mélange in the hanging wall (Fig. 2). The relationship between this fault and the Dajiling fault is unclear, but meta-Tethyan phyllite is not exposed between sedimentary-matrix mélange and high-grade meta-Tethyan rocks in the north, suggesting that the faults branch at a structurally higher, now-eroded location (Fig. 1).

Great Counter Thrust

Two strands of the steeply S-dipping Great Counter thrust are exposed to the west and east of Lopu Kangri peak (Fig. 2). The northernmost strand juxtaposes Xigaze forearc (KEx) strata in the hanging wall against the Kailas Formation in the footwall. The southern strand juxtaposes sedimentary-matrix mélange in the hanging wall against forearc strata in the footwall (Fig. 2). To the south of Lopu Kangri peak, a steeply S-dipping fault juxtaposes sedimentary-matrix mélange in the hanging wall against Eocene Gangdese granodiorite in the footwall (Fig. 2). We interpret that the two Great Counter thrust strands exposed at higher structural levels away from the Lopu Range merge into a single strand at depth (Fig. 1). The relationship between the Dajiling fault and the Great Counter thrust is not exposed; however, based on the early Miocene age of the Great Counter thrust (Ratschbacher et al., 1994) and geochronologic data presented in this study, we interpret that the Great Counter thrust cuts the Dajiling fault at depth.

Rujiao and Lopukangri Faults

The generally N-S–striking Rujiao (Ding et al., 2005) and Lopukangri (Sanchez et al., 2013) normal faults bound the Lopu Range and crosscut the Dajiling fault and Great Counter thrust (Fig. 2). They juxtapose meta-Tethyan rocks, sedimentary-matrix mélange, and Gangdese arc rocks in the footwall against predominately lower-grade meta-Tethyan rocks and sedimentary-matrix mélange in the hanging wall (Fig. 2; Wang et al., 2006). The faults initiated during middle Miocene time (Sanchez et al., 2013), and triangular facets and fault scarps in Quaternary glacial moraines indicate recent slip. Tectonic exhumation of the Lopu Range footwall across the Lopukangri and Rujiao faults is related to Miocene–recent orogen-parallel extension of the southern Tibetan Plateau.

Electron Microprobe Analytical Procedures

Electron microprobe analyses were conducted using a Cameca SX50 at the University of Arizona utilizing an accelerating voltage of 15 keV, a beam current of 10–20 nA, a beam size of 2 μm, and 10–30 s peak counts for garnet and phengite. Phengite analyses were optimized to avoid beam damage effects, which can yield discrepancies in K content. Zr-in-rutile analyses were conducted using an accelerating voltage of 15 keV, a beam current of 299 nA, and a peak counting time of 240 s to measure Zr, O, and Ti. Typical uncertainties for electron microprobe analyses are 0.5%–1.5% at 1σ for major elements present at concentrations >1 wt%. Uncertainties for Zr-in-rutile calculations incorporated the standard deviation weight percent for each analysis above the detection limit, which was ∼35 ppm. X-ray maps were created for whole microprobe slides from samples 7614AL3 and 61812AL5, which involved measuring backscattered electron (BSE) Z, Ca, Fe, Mg, Mn, Si, Al, K, Mn, Na, and Ti using a 40 μm beam that rastered across the mount in two passes (1024 × 580 resolution).

Garnet Lu-Hf Analytical Procedures

Disaggregation was accomplished using traditional crushing and grinding methods; garnets were isolated by handpicking, and magnetic and density separation techniques. Garnet aliquots (∼200 mg) were weighed into Savillex beakers for dissolution in ∼10 mL of 10:1 acid mixtures of 29 M HF and 14 M HNO3. Two whole-rock fractions were digested and analyzed alongside garnet separates, the first of which was accomplished using high-pressure Teflon bombs (WRB1), ensuring digestion of refractory mineral phases such as zircon. The second whole-rock fraction (WRS1) was digested in Savillex beakers to avoid dissolution of refractory minerals. Following digestion, both whole-rock and garnet fractions were equilibrated with 176Lu-180Hf and 149Sm-150Nd tracer spikes on a hot plate. Sm, Nd, Lu, and Hf were separated from final solutions using several different types of cation exchange resins. Details of these separation techniques are described in Cheng et al. (2008). Isotopic data were analyzed using a ThermoFinnigan NEPTUNE inductively coupled plasma–mass spectrometer (ICP-MS) at Washington State University following the procedures outlined in Vervoort et al. (2004).

Ar-Ar Analytical Procedures

Bulk separates of white mica grains for 40Ar/39Ar analysis were irradiated at the U.S. Geological Survey (USGS) TRIGA Reactor in Denver, Colorado, along with flux monitors to calculate J-factors and K2SO4 and CaF2 salts to calculate correction factors. Following a cooling period, samples were loaded into a resistance-heated furnace and heated to 120 °C at the same time that the entire extraction line was baked for 48 h at 220 °C; both were independently degassed. Samples were loaded into the furnace, where Ar was extracted using a step-heating routine with ±20 °C accuracy. Each heating step lasted 12 min and was followed by a cool down to 500 °C prior to advancing the gas into two successive gettering stages for purification. Ar was then admitted into a VG 5400 mass spectrometer, where it was ionized and detected by a VG electron multiplier and digitized with a Keithley 617 Electrometer. Decay constants used were those recommended by Renne et al. (2010). Baseline values were subtracted, and the isotopic measurements then were regressed to time zero using linear regression. Additional corrections and associated uncertainties were applied to account for blanks, machine discrimination, atmospheric contribution, and interfering isotopes produced in the reactor from Ca, K, and Cl present in the samples.

High-grade meta-Tethyan rocks (Pzg) consist of four typical lithologies: phengite-chlorite-albite-quartz-tourmaline schist, chlorite-phengite-calcite-quartz-albite-tourmaline ± staurolite schist, phengite-chlorite-quartz-albite ± ilmenite ± staurolite metagraywacke, and quartzite with minor albite, phengite, staurolite, and biotite (photomicrographs in Data Repository Fig. DR2). Phengite, which is an intermediate species along the solid solution between muscovite and celadonite, is generally considered a good indicator of HP conditions (Schertl and O’Brien, 2013). Most samples contained rutile, apatite, zircon, and allanite as accessory phases. One metagraywacke sample contained garnet (7614AL3). Microtextural observations revealed two generations of ductile fabrics: (1) primary foliation locally preserved as macroscopic crenulation cleavage and as asymmetric, often sigmoidal inclusion trails composed of phengite and tourmaline in albite porphyroblasts, and (2) secondary foliation defined by intergrown phengite, chlorite ± staurolite ± biotite in the matrix at high angles to inclusion trails and, locally, crenulation cleavage. Garnets in sample 7614AL3 appear to have formed alongside primary phengite and partially reacted to form a matrix of secondary phengite + chlorite ± staurolite (Data Repository Fig. DR2).

The proximity of penetratively deformed and metamorphosed Tethyan Himalaya Sequence rocks to the Yarlung suture, and the ubiquitous presence of phengite, and local occurrences of garnet and staurolite, suggest that Lopu Range rocks experienced HP metamorphism and subsequent retrogression. To investigate this hypothesis, a representative garnet-bearing metagraywacke (sample 7614AL3) and two metapelitic schist samples (61812AL5 and 62412AL3) were targeted for thermobarometry. Most traditional exchange and net transfer thermobarometers are not applicable to Lopu Range samples, because pyroxene and hornblende were not observed and biotite was not observed in the garnet-bearing sample. In addition, garnet did not occur alongside staurolite in sample 7614AL3, and the garnet was too calcic for reliable application of garnet-phengite Fe-Mg exchange geothermometry. Given these limitations, we identified phengite geobarometry and Zr-in-rutile geothermometry as the most promising techniques. In addition, we conducted equilibrium phase diagram (pseudosection) modeling using a Gibbs free energy minimization approach to evaluate the plausibility of our thermobarometry results, provide insights into mineral paragenesis, and provide context for 40Ar/39Ar phengite and garnet Lu-Hf geochronology data, presented in the Geochronology Results section.

ZrO2 solubility in rutile is strongly dependent on temperature and can be applied as a geothermometer when rutile coexists with zircon and quartz (Zack et al., 2004). In this technique, the maximum Zr concentration is interpreted to record the peak temperature attained while rutile was present in the sample. Maximum temperatures of metamorphism were calculated for the two rutile-bearing samples (61812AL5 and 62412AL3) using the equation of Tomkins et al. (2007), which also accounts for a minor pressure dependence. Zr concentration was measured in 35 rutile grains, yielding six analyses with Zr concentrations exceeding the ∼35 ppm detection limit (Data Repository Table DR1). The maximum Zr concentration was 122 ± 30 ppm (1σ), which corresponds to a maximum temperature of 622 +16/–20 °C at 2 GPa. The pressure dependence of this determination is plotted on the pseudosections in Figure 3 and is assumed to be applicable to both samples.

The Si content of phengitic white mica increases with pressure due to a reverse Tschermak substitution along a solid solution series between muscovite and celadonite (e.g., Massonne and Schreyer, 1987). This relationship depends mainly on pressure and can be modeled to account for a secondary dependence on bulk composition and mineral assemblage. The behavior of the Tschermak substitution, however, is complicated by a secondary pyrophyllitic substitution that might result in pressure overestimations up to 0.5 GPa at low-temperature conditions (Agard et al., 2001), or potentially operate in the opposite fashion (Coggon and Holland, 2002). These inconsistencies and the possibility of further unconstrained substitutions in white mica necessitate caution when estimating pressure conditions with phengite geobarometry alone. However, we minimized the uncertainty in this approach by modeling Si-in-phengite isopleths in a pseudosection using an updated white mica solution model adapted from Coggon and Holland (2002) and Auzanneau et al. (2010) that accounts for the Tschermak and pyrophyllite substitutions and their dependence on bulk chemistry and mineral assemblage. To critically evaluate our pressure estimates, we compared the mineral assemblages preserved in Lopu Range samples with their predicted stability fields from pseudosection modeling.

Local bulk compositions for pseudosections were extracted from whole-section X-ray maps using the MatLab graphic user interface XMapTools (Lanari et al., 2014), and they are provided in Figure 3. Pseudosections were modeled in Perple_X 6.6.6 (Connolly, 2009) using MnO, Na2O, K2O, FeO, MgO, Al2O3, SiO2, CaO, MnO, and TiO2 as thermodynamic components with H2O in excess. Al-free chlorite and Ti-phengite end members were excluded to simplify the calculations, as analyses revealed low Ti in phengite (<0.021 a.p.f.u.; Data Repository Table DR2) and Al in chlorite >2.7 a.p.f.u. In addition, rutile was not stable in either pseudosection when the Ti-phengite end members were considered, conflicting with its common presence in thin section. Phengite Si isopleths were calculated using the Mica+(CHA) solution model, which is a nonreciprocal white mica model after Coggon and Holland (2002) and Auzanneau et al. (2010) that incorporates the Tschermak and pyrophyllite substitutions. Additional solution models for carpholite [Carp(M)], stilpnomelane [Stlp(M)], chlorite [Chl(HP)], chloritoid [Ctd(HP)], clinopyroxene [Cpx(HP)], feldspar [AbFsp(C1); OrFsp(C1); Pl(I1,HP)], garnet [Gt(WPH)], ilmenite [IlGkPy], and staurolite [St(HP)] were considered. Pseudosection modeling and Zr-in-rutile geothermometry results are presented in Figure 3.

Pseudosection modeling revealed the expected stable mineral assemblages for pressure and temperature fields between 0.5 and 3.0 GPa and 300–800 °C given the local bulk chemistry of sample 7614AL3, a chlorite-phengite-quartz-garnet-albite-ilmenite-allanite-apatite-zircon schist, and sample 61812AL5, a chlorite-phengite-albite-quartz-rutile-tourmaline-apatite-zircon-allanite schist (Fig. 3). Electron microprobe transects across two garnets from sample 7614AL3 revealed a decrease in Mn (Sp 0.56–0.47) and an increase in Fe (Alm 0.32–0.38) from core to rim (Fig. 4), consistent with garnet growth during prograde metamorphism. Spessartine- and almandine-rich garnets that spanned the compositional range of those in sample 7614AL3 (Data Repository Table DR3) are expected to be stable across a relatively narrow temperature range of 420–460 °C at ∼1.5 GPa, and across a broader range below 1.25 GPa (shaded region, Fig. 3). We interpret that Lopu Range samples passed through the broad portion of this stability field during prograde metamorphism at ∼0.5 GPa and 400 °C, within the chlorite-phengite-garnet-sphene ± rutile stability field (Fig. 3). Si content in phengite averaged 3.22 ± 0.05 (1σ) a.p.f.u. in sample 7614AL3 (n = 26), 3.23 ± 0.06 a.p.f.u. in sample 61812AL5 (n = 15), and 3.24 ± 0.04 a.p.f.u. in sample 62412AL3 (n = 11; Data Repository Table DR2). In samples 7614AL3 and 61812AL5, phengite inclusions in garnet and albite had Si content up to 3.36 a.p.f.u.; no phengite inclusions were observed in sample 62412AL3. Based on these data, we interpret that peak-pressure metamorphism took place at ∼1.4 GPa within the chlorite-phengite-garnet-rutile stability field in both samples, suggesting that the maximum-temperature constraint from Zr-in-rutile (∼600 °C at 1.5 GPa) applies. The available data are unable to constrain whether peak-temperature conditions were attained concomitant with or after peak-pressure metamorphism.

Peak-pressure estimates from phengite geobarometry and pseudosection modeling are likely minima, as the presence of albite and absence of garnet in most samples, as well as higher Si content in phengite inclusions in albite and garnet, are indicative of retrogression. Chlorite was not preserved as inclusions in garnet and albite, suggesting that early-formed assemblages were located above the chlorite-in reaction in pressure-temperature (P-T) space, which is >1.5 GPa at T ≤ 600 °C in both samples (Fig. 3). Albite porphyroblasts likely grew at ∼1.0–0.8 GPa during retrogression, accompanying the rutile-ilmenite transition in sample 7614AL3 (Fig. 3). In sample 61812AL5, the appearance of plagioclase is predicted to be accompanied by carpholite, which was not observed, possibly indicating further retrogression at lower metamorphic grade or spurious pseudosection modeling results. The absence of kyanite, pyroxene, amphibole, and biotite in the modeled samples is consistent with our interpreted P-T path (Fig. 3). The lowest Si-in-phengite values in all three samples were ∼3.11–3.15 a.p.f.u., likely recording recrystallization at P <0.5 GPa.

Three aliquots of garnet separate and a powdered whole-rock fraction (WRS1) from metagraywacke sample 7614AL3 were analyzed to construct a Lu-Hf isochron, which yielded an age of 40.4 ± 1.4 Ma with a mean square of weighted deviates (MSWD) of 7.3 and an initial 176Hf/177Hf ratio of 0.28263 ± 0.00035 (Fig. 5). We chose to calculate the isochron with whole-rock fraction WRS1 because it had a lower Hf concentration and a less radiogenic 176Hf/177Hf ratio than whole-rock fraction WRB1—likely the result of increased zircon dissolution in the high-pressure Teflon bombs (WRB1; Data Repository Table DR4). Consideration of both whole-rock fractions produced an age of 40.6 ± 0.8 Ma with a higher MSWD of ∼18, whereas whole-rock fraction WRS1 alone produced an age of 40.9 ± 1.3 Ma with an MSWD of 6.3. Higher-than-optimal MSWD values for each combination might reflect the longer-duration garnet growth than chronologic resolution, mixing with zircon, and/or Lu/Hf zonation. The sample did not possess an adequate spread in Sm/Nd ratios to produce a meaningful Sm-Nd date.

Step-heating analysis of white mica separates from six samples yielded five 40Ar/39Ar plateau ages between 39.3 ± 1.2 Ma and 34.2 ± 1.2 Ma and one younger plateau age of 27.3 ± 1.2 Ma. Age determinations were calculated from a minimum of 70% of the total gas released, accomplished over 5–9 heating steps, yielding MSWD values between 0.04 and 1.40. Age spectrum plots and isotopic data for all samples are provided in Data Repository Table DR5. Interpreted Ar-Ar plateau ages are shown on the geologic map (Fig. 2).

Preliminary results from phengite geobarometry, Zr-in-rutile geothermometry, and pseudosection modeling indicate that Lopu Range samples likely experienced peak metamorphism at P ≥ 1.4 GPa and T ≤ 600 °C (Fig. 3). Phengite inclusions shielded in garnet and albite suggest higher pressures may have been experienced during prograde to peak metamorphism (possibly up to ∼2.0 GPa). Minimum pressure estimates indicate a depth of metamorphism ≥52 km, assuming a lithostatic gradient of ∼27 MPa/km (37 km/GPa) and the absence of significant tectonic overpressure. Integration of maximum temperature constraints produces a maximum geothermal gradient during peak-pressure metamorphism of 11 °C/km, consistent with subduction-related metamorphism and exhumation prior to thermal relaxation. Peak temperatures ≤600 °C imply thermal equilibration at midcrustal depths (20–30 km), assuming a 20–30 °C/km geothermal gradient, which is also consistent with the pressure conditions indicated by the lowest-Si phengite grains in our samples (Fig. 3; Data Repository Table DR2).

The P-T path recorded by Lopu Range metasedimentary rocks is broadly similar to that of metasedimentary rocks in the Tso Morari complex, which experienced peak-pressure metamorphism at ∼2.0 GPa and ∼550 °C prior to decompression and heating to 630 °C (Guillot et al., 1997). Tso Morari metasedimentary rocks underestimate peak-pressure conditions deduced from coesite-bearing eclogite lenses, which are as high as 4.4–4.8 GPa at 560–760 °C (Wilke et al., 2015). Mafic enclaves in the Tso Morari orthogneiss are the likely protolith of Tso Morari coesite-bearing eclogite (de Sigoyer, 1998). The lack of equivalent rocks within the Lopu Range may reflect a shallower depth of exposure (i.e., the Lopu Range rocks never experienced pressures as high as those at Tso Morari) or the lack of mafic lithologies in the Lopu Range, which are more amenable to preserving higher-pressure mineral assemblages.

The Lopu Range shares many structural characteristics with Tso Morari and Kaghan Valley, including: (1) HP meta-Tethyan rocks exposed at the center of an antiformal dome within the India-Asia suture zone; (2) juxtaposition of thrust sheets with distinct pressure gaps across ductile shear zones that display a mix of normal- and reverse-sense kinematics; (3) juxtaposition of footwall meta-Tethyan rocks against hanging-wall suture zone mélange across a synexhumational, normal-sense shear zone; and (4) overprinting by late normal faults (Figs. 1–2; Guillot et al., 1997; Epard and Steck, 2008; Beaumont et al., 2009). However, the Lopu Range differs from Tso Morari in that the roof fault (Dajiling) displays consistent top-to-the-N kinematics instead of opposing, consistently normal-sense kinematics across the dome (Epard and Steck, 2008; Beaumont et al., 2009). This difference might result from a larger component of pure-shear, vertical thinning at Tso Morari. Structures exposed at Lopu Kangri are consistent with passive, normal-sense shearing along the top of a thrust sheet paired with structurally lower, foreland-directed thrust faulting in the Tethyan Himalaya (Fig. 1) to accommodate bulk crustal shortening.

Prograde metamorphism of Lopu Range rocks was ongoing at 40.4 ± 1.4 Ma, based on our garnet Lu-Hf geochronology. The timing of Lopu Range prograde metamorphism is ∼6 m.y. younger than that of eclogite-facies metamorphism at Kaghan Valley and Tso Morari (47–43 Ma; Donaldson et al., 2013). Lopu Range phengite Ar-Ar ages likely record recrystallization and/or cooling during exhumation and concomitant retrogression. Five of the six ages are between 39 and 34 Ma; a younger age of 27.3 ± 1.2 Ma in one sample might reflect partial resetting. These ages are also younger than the timing of exhumation at Tso Morari and Kaghan Valley, which occurred between 47 and 34 Ma and between 46.5 and 31 Ma, respectively (de Sigoyer et al., 2000; Schlup et al., 2003; Wilke et al., 2010). Despite younger ages of both prograde metamorphism and exhumation in the Lopu Range, the short interval between the two (2–7 m.y.) is consistent with the duration of comparably sized, early-formed continental HP metamorphic complexes globally (2–8 m.y.; Kylander-Clark et al., 2012). The timing discrepancy between Lopu Range and northwest Himalaya HP metamorphism may be a result of, but does not necessarily require, a younger collision age in central Tibet and/or geometric variations along the Eurasian and/or Indian margins along strike. It is possible that meta-Tethyan rocks at both localities were not located along the leading edge of the Indian continental plate—and thus were not immediately subducted—at the onset of collision.

A tectonic model that can explain the structural and metamorphic evolution of the Lopu Range is shown in Figure 6. HP metamorphism of Tethyan Himalaya rocks now exposed in the Lopu Range likely resulted from steep continental subduction to upper-mantle depths at ca. 40 Ma (Fig. 6B), as evidenced by our garnet Lu-Hf data. The most recent episode of Gangdese arc magmatism persisted from ca. 70 Ma to 38 Ma (Kapp et al., 2007), requiring the presence of a melt-fertile upper-plate mantle wedge during this time interval. We suggest that between 40 and 38 Ma, a transition to shallow northward underthrusting of Greater Indian continental lithosphere occurred, resulting in northward displacement of the Asian mantle wedge and shutoff of Gangdese arc magmatism in southern Tibet (Fig. 6C; Kapp et al., 2007). Shallowing in the dip of the subducting continental plate was also concomitant with the initiation of Lopu Range HP exhumation. The transition from steep to shallow subduction (and Lopu Range rock burial to exhumation) may have been a consequence of tearing off of the leading subducting slab. If so, then slab tearing may have initiated earlier at ca. 45 Ma along strike to the west in the northwestern Himalaya (e.g., Negredo et al., 2007) and to the east at the longitude of Lhasa (Ji et al., 2016) and propagated toward the Lopu Range part of the India-Asia suture zone at ca. 40 Ma. Alternatively, HP rock exhumation and changes in slab dynamics can occur in the absence of slab breakoff (Warren et al., 2008). Future, holistic studies of Himalayan metamorphism, suture-zone structural evolution, and Asian arc magmatism at various longitudes along strike of the India-Asia suture will elucidate any spatiotemporal trends and reveal the relationship between HP exhumation and slab breakoff. Previous tectonic models of India-Asia collision have proposed a similar progression of initial steep subduction, breakoff, and underthrusting (Fig. 6; Chemenda et al., 2000; DeCelles et al., 2011); however, our results indicate that the transition to underthrusting might have occurred later than previously thought (40–38 Ma vs. 45–40 Ma).

Interpretation of continental subduction within the context of subduction angle might explain the broad range of behaviors demonstrated by continental HP metamorphic complexes along collisional plate boundaries (e.g., Hacker et al., 2013; Ducea, 2016). Steep continental subduction might favor the formation of small (<5000 km2) complexes with short durations—such as the Lopu Range, Tso Morari, and Kaghan Valley—whereas shallow subduction might favor the formation of large complexes with long durations (e.g., the Western Gneiss Region, Norway; Kylander-Clark et al., 2012). Variations in subduction dynamics along strike might also explain timing variations within individual orogens. Diachronous slab breakoff and subsequent onset of shallow underthrusting along strike are potential explanations for younger exhumation in southern Tibet (Lopu Range). However, prolonged steep subduction in southern Tibet relative to the northwestern Himalaya (Fig. 1) does not explain the younger age of prograde metamorphism, leaving open the possibility of variations in collision timing or continental margin geometry along strike. Future integration of geochronologic and geologic records in the northwest Himalaya and the Lopu Range may reveal the cause of HP metamorphism timing variations along strike in the Himalayan-Tibetan orogenic system.

We acknowledge B. Hacker, C. Beaumont, C. Mattinson, and F. Mazdab for insightful discussions; K. Domanik and D. Wilford for laboratory assistance; C. Campbell and Xiaohui Liu for help in the field; S. Guillot, J. Lee, and H-.J. Massonne for constructive reviews on an earlier version; and K. Stüwe for editorial handling. This research was supported by grants from the U.S. National Science Foundation Continental Dynamics Program (EAR-1008527), China National Science Foundation (no. 41490610), and Geological Society of America (Student Research Grant).

1GSA Data Repository Item 2016278, containing photomicrographs, electron probe micro-analyzer data, white mica 40Ar/39Ar analytical data, and garnet Lu/Hf analytical data, is available at, or on request from