Identification of a new source for the Triassic Langjiexue Group: Evidence from a gabbro-diorite complex in the Gangdese magmatic belt and zircon microstructures from sandstones in the Tethyan Himalaya, southern Tibet

An ongoing continent-continent collisional orogen, the Himalayan-Tibetan orogen, has attracted much attention among the geological community (Fig. 1; Yin and Harrison, 2000; Spencer et al., 2012). The Indo-Asian collision took place at ca. 60–50 Ma, triggering the uplift of the Tibetan Plateau (Ding et al., 2016; Hu et al., 2015; Jin et al., 2018; Sun et al., 2016; Zhu et al., 2015). However, the pre-plateau history of the Lhasa terrane, especially the evolutionary history of the Neotethyan Ocean, remains enigmatic (Li et al., 2010; Zhu et al., 2010). The Gangdese magmatic belt, located in the southern margin of the Lhasa terrane, documents voluminous Middle Triassic to Late Cretaceous subduction-related igneous activity (Ji et al., 2009; Meng et al., 2016a, 2019a; Mo et al., 2005a; Wang et al., 2016a), indicating that the Gangdese magmatic belt experienced a protracted history prior to the Indo-Asian collision. Thus, study of the magmatic rocks in the Gangdese magmatic belt is very important for deciphering the subduction-accretion orogeny of the Gangdese magmatic belt and the framework of the Neotethyan realm before the final collision and formation of the Tibetan Plateau.

During the past decades, progress has been achieved on understanding the formation of the Himalayan-Tibetan orogen. However, many basic questions remain open to debate. This study focuses on the following issues: (1) the timing for initial subduction of the Neotethyan oceanic lithosphere; and (2) the tectonic setting of the Late Triassic Langjiexue Group in the Tethyan Himalaya, in other words, the provenance for the sediments of the Langjiexue Group.

Recent studies have revealed that voluminous calc-alkaline igneous rocks are exposed in the Gangdese magmatic belt, with ages ranging from Middle Triassic to Late Cretaceous (Ma et al., 2018a; Wang et al., 2016a). The Middle Triassic to Jurassic magmatic rocks are ascribed to southward subduction of the Bangong-Nujiang Tethyan oceanic lithosphere beneath the Lhasa terrane (Zhu et al., 2013; Yang et al., 2017), or to the northward subduction of the Neotethyan oceanic slab beneath the Lhasa terrane (Guo et al., 2013b; Kang et al., 2014; Ma et al., 2018a; Wang et al., 2016a). Whether the southward or northward model is correct, these results suggest that these magmatic rocks were generated in a convergent margin setting either as an active continental margin or intra-oceanic arc. About 40% of the modern convergent margin around the globe is interpreted as intra-oceanic subduction zones (Larter and Leat, 2003). This raises the questions of whether an intra-oceanic subduction system developed within the eastern Neotethyan realm, and whether some intra-oceanic arc rocks are preserved in the Gangdese magmatic belt.

The Late Triassic Langjiexue Group, exposed in the Tethyan Himalaya belt, plays a pivotal role in reconstructing the framework of the Neotethyan realm. However, its tectonic affinity has been hotly debated for decades. Models proposed to explain the formation of the Langjiexue Group include basin-fill during the initial rifting between the Indian and Lhasa blocks (Dai et al., 2008; Webb et al., 2012), forearc basin deposition due to the northward subduction of the Neotethyan oceanic lithosphere beneath the Lhasa terrane or an intra-oceanic arc (Li et al., 2010), passive continental margin deposition along the northern or northwestern margin of the Gondwana landmass (Cai et al., 2016; Cao et al., 2018; Fang et al., 2018; Wang et al., 2016b; Meng et al., 2019b), and a multi-source model within the Neotethys (Li et al., 2016; Zhang et al., 2017). Thus this question has been hindering our full understanding of the reconstruction of the Triassic paleogeographic configuration for the eastern Neotethyan realm, as well as the possible source for the Langjiexue Group.

The foregoing issues are closely related to the opening of the Neotethyan Ocean. Based on the paleogeographic reconstruction of the Pangea supercontinent and the Neotethyan realm, the Neotethys has been suggested to have opened in the early Permian (Angiolini et al., 2003; Garzanti et al., 1996; Kroner et al., 2016). However, as remnants of the Neotethyan oceanic lithosphere, the Indus–Yarlung Tsangpo ophiolites, whose formation is attributed to forearc extension, mainly fall into an age range of 130–120 Ma (Wu et al., 2014; Liu et al., 2016; Maffione et al., 2015; Xiong et al., 2017). Furthermore, the opening of the Neotethys has been proposed to have been a byproduct of the southward subduction of the Bangong-Nujiang Tethys during the Late Triassic (Zhu et al., 2013; Yang et al., 2017). These debates necessitate more work to constrain the opening of the Neotethys. This dispute hampers our recognition of the tectonic regime for the eastern Neotethyan realm, especially the Cimmeride and the northern Gondwana landmasses.

In this study, we discuss new results from the ca. 240 Ma gabbro-diorite complex in the Gangdese magmatic belt and microstructures of detrital zircon grains of sandstones from the Late Triassic Langjiexue Group in the Tethyan Himalaya. Based on our combined analyses of regional geology, we propose the existence of another possible source in the Gangdese magmatic belt, which partly provided some source materials for the Langjiexue sandstones.

The Tibetan Plateau was formed through the sequential accretion of terranes to the southern margin of the Asian continent (Yin and Harrison, 2000). From north to south, these terranes include the Songpan-Ganze, Qiangtang, Lhasa, and Tethyan Himalayan terranes separated by the Jinsha, Bangong-Nujiang, and Indus–Yarlung Tsangpo suture zones (Fig. 1; Zhang et al., 2014; Leary et al., 2016). Among these accreted terranes, the Lhasa terrane is considered to have been the last block to aggregate with the Asian continent before the Indo-Asian collision (Yin and Harrison, 2000; Tapponnier et al., 2001). Recent studies suggest that the Lhasa terrane is not intact, and several suture zones are proposed within the Lhasa terrane, such as the Shiquanhe–Yunzhug–Namu Tso and Sumdo ophiolite belts (Pan et al., 2012; Yang et al., 2007; Zeng et al., 2018). The Lhasa terrane is separated from the Tethyan Himalayan terrane by the Indus–Yarlung Tsangpo ophiolite belt, whose formation ages cluster around 130–120 Ma (Liu et al., 2016; Xiong et al., 2017).

The Gangdese magmatic belt occupies the southern margin of the Lhasa terrane. The east-west–trending Gangdese magmatic belt, 2500 km in length and ∼100 km in width, is located immediately north of the Indus–Yarlung Tsangpo suture zone (Fig. 2; Yin and Harrison, 2000). Diabasic gabbros, diorites, and granitoids, as well as their eruptive equivalents, are tightly dispersed within the Gangdese magmatic belt. Temporally, the igneous rocks across the whole belt span a large age range from 237 to 13 Ma (Ji et al., 2009; Wang et al., 2018; Meng et al., 2019a). These voluminous magmatic rocks belong to four major flareups at 237–160 Ma, 100–80 Ma, 65–40 Ma, and 30–13 Ma. The first flareup has been suggested to have subduction-related calc-alkaline affinity, owing to the subduction of the Neotethyan oceanic lithosphere (Lang et al., 2017; Meng et al., 2016a). Two different models have been proposed for the 100–80 Ma magmatic flareup: one is ridge subduction (Guo et al., 2013a; Zhang et al., 2010), and the other is extension due to slab rollback (Ma et al., 2015). The 65–40 Ma magmatic flareup, represented by the Linzizong volcanic rocks and Quxu batholith, resulted from slab rollback and breakoff associated with the Indo-Asian collision (Ding et al., 2003; Mo et al., 2003, 2005b; Lee et al., 2009b). The post-collisional (30–13 Ma) igneous rocks within the Gangdese magmatic belt consist of S-type granites, adakites, basaltic dikes, and ultrapotassic to potassic volcanics (Chung et al., 2005). Post-collisional porphyry-type Cu deposits have been identified in the Gangdese magmatic belt (Hou et al., 2015; Lu et al., 2015; Yang et al., 2016).

South of the Indus–Yarlung Tsangpo suture zone, the Himalayan orogen is composed of, from south to north, the Lesser Himalaya, Greater Himalaya, and Tethyan Himalaya (Yin, 2006; Guillot et al., 2008; Xu et al., 2013; Leary et al., 2017). The Greater Himalayan rocks are considered to be Indian basement that has been exhumed along the Main Central thrust (DeCelles et al., 2000). The Tethyan Himalaya is separated from the Greater Himalaya to the south by the Southern Tibet detachment system (Yin and Harrison, 2000). The Tethyan Himalaya comprises southern and northern subzones, divided by the Gyirong-Kangmar thrust. The Mesozoic–Cenozoic strata in the Tethyan Himalaya can be divided into the Langjiexue Group (Upper Triassic) and Lure (Lower Jurassic), Zhela (mid-Jurassic), Sangxiu (Upper Jurassic), Lakang (Lower Cretaceous), and Zongzhuo (Upper Cretaceous) formations (Ao et al., 2018). The southern zone is characterized by shallow-water carbonates and clastic sediments, whereas the northern zone is dominated by deep-water turbidites, shales, and cherts (Wang et al., 2016b). In the northern zone, the Late Triassic Langjiexue Group sediments dominate and occupy a huge area (Fig. 2).

The Langjiexue Group was tectonically displaced to the south of the eastern Indus–Yarlung Tsangpo suture zone (Fig. 2). It was thrust to the south over the Nieru Formation and Jurassic–Cretaceous strata of Tethyan Himalaya along the nearly east-west–trending Lazi-Qiongduojiang-Zara thrust and was thrust northward over the Indus–Yarlung Tsangpo ophiolites along the east-west–trending Greater Counter thrust in the north (Fang et al., 2018; Yin, 2006). This group is mainly composed of feldspar and/or lithic sandstone, siltstone, slate, and shale, deposited in a bathyal-abyssal submarine fan environment (Zhang et al., 2015a). The Langjiexue Group was subjected to intense deformation, characterized by the south-north convergence resulting in southward and northward thrusts and east-west–trending axial planes of folds (Fig. 3B; Li et al., 2010). Although strongly deformed, the Langjiexue Group only experienced low- to medium-grade greenschist facies metamorphism (Li et al., 2016).

Traditionally, the Langjiexue Group consists of the Jiangxiong, Jiedexiu, Zhangcun, and Songre Formations. However, an increasing number of studies suggest that the coeval Nieru Formation in the Kangma region, situated at the southern margin of the Langjiexue Group, also belongs to this group. This formation shares similar stratal assemblages and lithological features as well as fossil types (Cai et al., 2016; Li et al., 2010; Li et al., 2016). The Langjiexue Group is characterized by extensive ca. 130 Ma diabase dikes or intrusions (Fig. 3D), showing close affinity to oceanic island basalt (OIB)–type rocks (Zhu et al., 2009). These OIB dikes were first proposed to have been generated by the Kerguelen plume in the Indian plate due to the rifting of the Gondwana landmass (Zhu et al., 2009). Recently, Ao et al. (2018) proposed that these dikes were probably formed in the Neotethyan Ocean and hosted in the Langjiexue Group strata during accretion in the Tethyan Himalayan prism.

Zircon U-Pb ages presented in this paper were obtained in two laboratories. Zircon U-Pb ages of the gabbro-diorite complex were measured by using an Agilent 7500a inductively coupled plasma mass spectrometer (ICP-MS) attached to a Coherent 193 nm laser ablation system at the State Key Laboratory for Mineral Deposits Research, Nanjing University (Nanjing, China). The laser light beam had a diameter of ∼32 μm with a repetition rate of 5 Hz under a 70% energy condition. Isotope mass fractionation was normalized through external standard GEMOC GJ-1 with ²⁰⁷Pb/²⁰⁶Pb age = 608.5 ± 1.5 Ma (Jackson et al., 2004). The analytical accuracy was monitored through the Mud Tank zircon standard, which has an intercept age of 732 ± 5 Ma (Black and Gulson, 1978). Zircon analyses were carried out in runs of 15 analyses including five zircon standards and up to 10 sample spots. The U-Pb dating results were calculated through the online software package GLITTER (ver. 4.4; http://www.glitter-gemoc.com/). Zircon U and Th concentrations were calculated by comparing the relative signal intensity between the standard zircon GJ-1 (U = 330 ppm, Th = 8 ppm) and the zircon samples using the Microsoft Excel program Data Templatev2b from the Australian Research Council National Key Centre for Geochemical Evolution and Metallogeny of Continents (GEMOC). Each rock sample was subject to one or several runs of 15 analyses. The analytical results are presented in Table S1 of the Supplemental Materials¹.

Detrital zircon laser ablation (LA) ICP-MS U-Pb geochronology and trace elements from sandstones, as well as trace elements for the magmatic zircon grains of the gabbro-diorite complex, were obtained at the Key Laboratory of Mineral Resources Evaluation in Northeast Asia, Ministry of Land and Resources of China, Jilin University (Changchun, China). The instrument coupled a quadrupole ICP-MS (Agilent 7900) and 193 nm ArF excimer laser (COMPexPro 102, Coherent) with an automatic positioning system. For the present work, laser spot size was set to 32 μm, laser energy density at 10 J/cm², and repetition rate at 8 Hz. Laser sampling used a 30 s blank, 30 s sampling ablation, and 2 min sample-chamber flushing after the ablation. The ablated material was carried into the ICP-MS by a high-purity helium gas stream with a flux of 1.15 L/min. The whole laser path was fluxed with Ar (600 mL/min) in order to increase energy stability. The counting time was 20 ms for ²⁰⁴Pb, ²⁰⁶Pb, ²⁰⁷Pb, and ²⁰⁸Pb; 15 ms for ²³²Th and ²³⁸U; 20 ms for ⁴⁹Ti; and 6 ms for other elements. Calibrations for the zircon analyses were carried out using NIST 610 glass as an external standard and Si as an internal standard. U-Pb isotope fractionation effects were corrected using zircon 91500 (Wiedenbeck et al., 1995) as an external standard. Zircon standard Plesovice (337 Ma) was also used as a secondary standard to supervise the deviation of age measurement (Sláma et al., 2008). The analytical results of detrital zircon U-Pb ages are presented in Table S2 (^{footnote 1}), whereas the trace elemental results of magmatic zircon grains from the plutonic rocks in the Gangdese belt and the detrital zircon grains from the sandstones of the Langjiexue Group in the Tethyan Himalaya are presented in Table S3 (^{footnote 1}).

For all analyses, isotopic ratios and element concentrations of zircon grains were calculated using GLITTER. Concordia ages and diagrams were obtained using Isoplot/Ex (ver. 3.0) (Ludwig, 2003). A common-Pb correction was applied using LA-ICPMS Common Lead Correction (ver. 3.15, http://gemoc.mq.edu.au/TerraneChron/CommonPb.html), following the method of Andersen (2002). The analytical data are presented on U-Pb concordia diagrams with 2σ errors. The mean ages are weighted means at the 95% confidence level (Ludwig, 2003). More detailed analytical parameters of zircon U-Pb dating can be seen in Text S1 (^{footnote 1}).

Zircon Hf isotopes were analyzed using a 193 nm laser attached to a Neptune multi-collector (MC) ICP-MS (Thermo Finnigan, Bremen, Germany) at Key Laboratory of Deep-Earth Dynamics, Institute of Geology, Chinese Academy of Geological Sciences (CAGS; Beijing, China). MC-LA-ICP-MS analyses were carried out with a beam size of ∼32 μm. A decay constant for ¹⁷⁶Lu of 1.867 × 10⁻¹¹ yr^–1 (Söderlund et al., 2004) was adopted to analyze initial ¹⁷⁶Lu/¹⁷⁷Hf ratios, while chondritic values of ¹⁷⁶Lu/¹⁷⁷Hf = 0.0336 and ¹⁷⁶Hf/¹⁷⁷Hf = 0.282785 were chosen to obtain the ε_Hf(t) values (Bouvier et al., 2008). The single-stage model age (T_DM) was calculated relative to the depleted mantle with a present-day ¹⁷⁶Hf/¹⁷⁷Hf = 0.28325 and ¹⁷⁶Lu/¹⁷⁷Hf = 0.0384. A two-stage continental model age (T_DM^C) was calculated by projecting the initial ¹⁷⁶Hf/¹⁷⁷Hf of zircon back to the depleted-mantle growth curve using ¹⁷⁶Lu/¹⁷⁷Hf = 0.015 for the average continental crust (Griffin et al., 2000). The analytical results are presented in Table S4 (^{footnote 1}).

The whole-rock elemental results were determined at ALS Laboratory (ALS Mineral–ALS Chemex; Guangzhou, China). Oxide abundances were measured by a PANalytical Axios X-ray fluorescence spectrometer with analytical precision of better than 5%. Trace and rare earth elements (REEs) were measured using ICP-MS; relative standard deviations are <10% for trace elements and <7% for REEs. Analytical procedures are described by Qi et al. (2000), and the results are presented in Table S5 (^{footnote 1}).

For whole-rock Sr-Nd isotopic analysis, rock powders were decomposed using a high-pressure polytetrafluoroethylene (PTFE) bomb. Strontium and neodymium were purified from the same digestion solution by two-step column chemistry. In the first exchange column, Bio-Rad AG 50W-X8 and Eichrom Sr-Spec resins were used to separate Sr and REEs from the sample matrix. Neodymium was separated from other REEs on the second column with Teflon powder coated in Eichrom Ln-Spec resin. The Sr- and Nd-bearing elutions were dried and redissolved in 1.0 ml 2 wt% HNO₃. Small aliquots of each were measured using an Agilent Technologies 7700x quadrupole ICP-MS to determine the contents of Sr and Nd. Diluted solutions (50 ppb Sr, 50 ppb Nd, doped with 10 ppb Tl for both) were introduced into a Nu Instruments Nu Plasma II MC-ICP-MS through a Teledyne Cetac Technologies Aridus II desolvating nebulizer system. Sr-Nd isotope analyses were carried out at Nanjing FocuMS Technology (Nanjing, China). Detailed measurement and calculation procedures are presented in Zhu et al. (2017). The analytical results are presented in Table S6 (^{footnote 1}).

Representative hornblende and plagioclase minerals were selected from thin sections for electron microprobe analyses (EMPA). Mineral compositions were measured using a JEOL JXA-8230 electron microprobe (20 nA beam current, 5 μm beam spot, 15.0 kV accelerating voltage) at the Institute of Mineral Resources, CAGS (Beijing, China). The analytical results are presented in Table S7 (^{footnote 1}).

Hornblende and hornblende-plagioclase geothermometry are used based on the values of Si and Al cations in the tetrahedral positions (Blundy and Holland, 1990; Holland and Blundy, 1994; Ridolfi and Renzulli, 2012). Blundy and Holland (1990) proposed an empirical hornblende-plagioclase thermometer based on the edenite-tremolite reaction for quartz-bearing intermediate to felsic igneous rocks. Later on, Holland and Blundy (1994) reported thermometer A (tremolite-edenite reaction) for quartz-bearing metabasites and thermometer B (richterite-edenite reaction) for quartz-free igneous rocks. The hornblende thermometer of Ridolfi and Renzulli (2012) was also used to obtain the crystallization temperature of the pluton. Detailed calculation procedures can be referenced in Ma et al. (2018b). The analytical results are presented in Table S7 (^{footnote 1}).

Approximately 150 zircon grains of each sample were mounted on double-sided tape on a glass disk. After carbon coating, these detrital zircon grains were analyzed for surface microstructures (roundness and surface textures) at the Institute of Geology, CAGS (Beijing, China) using a FEI NOVA NANOSEM 450 scanning electron microscope (SEM). Magnification scales ranged from ∼100× to 460×. The SEM worked in secondary electron mode under high vacuum with 10 or 15 kV voltage and a working distance of ∼5 mm. After the SEM analyses, cathodoluminescence (CL) images were obtained. The detailed analytical method for CL images will not be presented here.

Detrital zircon morphology (roundness) features were categorized using parameters of Gärtner et al. (2013). The I–X roundness scale reveals the smoothing of crystal edges and abrasion of crystal faces: a roundness of I emphasizes that the crystal is completely unrounded, whereas a roundness of X implies that the crystal is completely rounded. Crystal surface textures were classified using the standards of Finzel (2017) and Vos et al. (2014). The analytical results of zircon CL and SEM images are presented in Text S2 (Fig. S1 [^{footnote 1}]).

The gabbro-diorite complex, located in the southern margin of the Gangdese magmatic belt, shows varied petrographic structures such as hornblende cumulate layers, huge phenocrysts, magmatic flow banding, and equigranular features (Fig. 4). In all structures, hornblende dominates as the main mafic mineral, with plagioclase occupying the interstitial space between hornblende phenocrysts (Fig. 5). In addition, plagioclase occurs as small laths, and crosscuts and postdates the hornblende. The main body of the pluton is equigranular, with rock composition akin to gabbro, diorite, and granodiorite, but still marked by abundant hornblendes. The rock type of gabbro to granodiorite combined with the mineral structures (phenocrysts to equigranular) and the hornblende cumulate layers and plagioclase layers reveal that the pluton was most likely formed through a fractional crystallization process. Abundant hornblendes suggest that the magma source of the pluton was wet (Murphy, 2013). Two later dikes that intruded the pluton have also been observed (Fig. 3C).

The Langjiexue Group in the Renbu region (Fig. 2) mainly consists of interbedded dark mudstones and sandstones formed in a submarine fan environment (Zhang et al., 2015a). These strata have been strongly deformed, showing fold and thrust structures (Fig. 3B) with muscovites oriented along cleavage planes (Figs. 5E–5F). According to our detailed field work, we find that these north-verging folds with east-west–trending axial planes experienced a second structural event (east-west–verging folding), evidenced by varying hinge orientations (Fig. 3E). The second deformation event was probably associated with the east-west–trending orogen-parallel extrusion of the Himalayan terranes in the late Oligocene and Miocene (Xu et al., 2013). The extrusion was probably triggered by the ∼20° clockwise vertical-axis rotation during the early Miocene (Antolín et al., 2011). The diabase dikes hosted within the sedimentary rocks are cut by conjugate quartz veins (Fig. 3D) whose formation could be closely associated with the later deformation event. The orientations of these conjugate quartz veins indicate the same direction of shortening as the second folds. In spite of intense deformation (Fig. 6), the graded sandstones of the Langjiexue Group exhibit clear Bouma sequences. The sandstones have subangular plagioclases and quartz, and abundant in lithic volcanic fragments.

Cathodoluminescence (CL) images reveal that most of the examined zircon grains of the gabbro-diorite complex have broad, banded zoning, whereas a few show oscillatory zoning (Fig. 7). Most of these zircon grains display euhedral prismatic shapes with aspect ratios of ∼2:1 (∼100–200 µm in length and ∼50–100 µm in width; Fig. 7). No obvious metamorphic overgrowths were present, and all of the zircon grains show high Th/U ratios between 0.8 and 2.7 (with a peak at ∼1.6), typical of magmatic zircon grains. These features suggest that these zircon grains are of magmatic origin (Corfu et al., 2003). Fifteen (15) zircon U-Pb age analyses of samples xm82, xm84, xm85, and xm86 yield weighted mean ages of 244 ± 1.7 (mean square of weighted deviates [MSWD] = 0.68), 242.2 ± 1.6 (MSWD = 0.42), 243.3 ± 1.8 (MSWD = 0.37) and 238.3 ± 1.6 Ma (MSWD = 0.48), respectively (Fig. 7).

Detrital zircon grains extracted from the sandstones of the Langjiexue Group in the Renbu region were analyzed for U-Pb dating. A total of 308 zircon grains have been dated (77 grains from each of four sandstone samples). Generally, discordance filters in literature vary from 1% to 30% depending on the level of interpretation desired and the data processing precision (Spencer et al., 2016, 2017). In the present study, we choose 20% discordance as the filter based on the following reasons: (1) LA-ICP-MS has lower analytical precision than secondary ion mass spectrometry (SIMS) or thermal ionization mass spectrometry (TIMS), thus it should be acceptable within larger discordance; (2) a filter of 20% versus 10% would not have big impact on the statistical analysis for the dated zircon in the present study; and (3) a large number of detrital zircon ages of the Langjiexue Group have been published in previous work, which could be used as a compelling comparison to the results of our present study. Among the 308 zircon grains, 297 passed a concordance filter of 80%–100% concordance. In addition, we have adopted the evaluation rule of discordance in Spencer et al. (2016, 2017) to filter the 297 zircon U-Pb ages (with concordance between 80%–100%). The plots reveal that only 279 zircon grains fall along the ²⁰⁶Pb/²³⁸U versus ²⁰⁷Pb/²³⁵U 1:1 age line, with 18 falling off the line (Fig. S2 in Text S2 [^{footnote 1}]). The remaining 279 ages exhibit a wide range from Mesoarchean to Late Triassic (3190–201 Ma) (Fig. 8A), indicating multiple sources and a variety of rock types in the provenance. Several prominent age populations exist: Mesoarchean (3200–3100 Ma), early Paleoproterozoic (2500–2400 Ma), late Paleoproterozoic–early Mesoproterozoic (1755–1500 Ma), early Neoproterozoic (1000–800 Ma), late Neoproterozoic–early Paleozoic (620–420 Ma), and early Permian–Late Triassic (299–201 Ma) (Fig. S3 in Text S2 [^{footnote 1}]).

In addition, a large number of published detrital zircon U-Pb ages for the Langjiexue Group have been collected and compiled in the Table S2 (^{footnote 1}). The age of the cutoff between ²⁰⁷Pb/²⁰⁶Pb and ²⁰⁶Pb/²³⁸U ages is chosen as 1.5 Ga according to the empirical equation of Spencer et al. (2016). In their study, the best-fit trendlines reveal an empirical crossover point of ²⁰⁷Pb/²⁰⁶Pb and ²⁰⁶Pb/²³⁸U ages at ca. 1.5 Ga. Thus, the ²⁰⁷Pb/²⁰⁶Pb ages are used as the best ages for zircons older than ca. 1.5 Ga, and the ²⁰⁶Pb/²³⁸U ages are used as the best ages for younger ones (Spencer et al., 2016). The distribution patterns of these collected detrital zircon U-Pb ages have been evaluated by kernel density estimation (KDE) plotting according to the method of Vermeesch (2012) and Vermeesch et al. (2016) (Fig. 8B).

All of the analyzed rock samples from the gabbro-dioritic suite plot in the subalkaline field of gabbro to granodiorite with major oxide abundances as follows: SiO₂ = 51.85–59.28 wt%, Al₂O₃ = 15.14–17.52 wt%, Fe₂O₃^T (^T means total iron) = 5.98–10.69 wt%, MgO = 2.7–6.4 wt%, CaO = 4.8–9.41 wt%, TiO₂ = 0.45–0.51 wt%, and total alkali (Na₂O + K₂O) = 4.14–7.49 wt% (Fig. 9). Such oxide abundances resemble those of low-MgO high-alumina basalts or andesites (Sisson and Grove, 1993b). These chemical characteristics, together with FeO*/MgO (asterisk means total) ratios from 1.24 to 2.19 (Fig. 9B), FeO*/MgO versus SiO₂, and TiO₂ versus FeO^T/MgO (^T means total) (Figs. 9B–9C), indicate typical calc-alkaline features. Diagrams of ε_Nd(t) versus SiO₂, Cr versus Mg# (molecular MgO/(MgO + FeO^T)*100), and La/Sm versus La reveal that these studied rocks exhibit a gradual variation trend (Figs. 9D–9F), which, with the variation of mineral structures and rock types in the field and the wide range of SiO₂, indicates that they were formed through a continuous fractional crystallization process.

In chondrite-normalized REE patterns, all samples have similar enrichment in light REEs with (La/Yb)_N (_N stands for chondrite-normalized) = 4.25–8.05 and weak negative Eu anomalies (Fig. 10A). In the primitive mantle–normalized spidergram, these rocks display a strong Pb enrichment and negative Nb, Ta, and Ti anomalies, similar to the western Aleutian high-Al basalt (Fig. 10B). Popular discrimination diagrams (Condie, 1989; Pearce et al., 2005; Plank, 2005) are employed to distinguish the tectonic setting of the studied plutonic rocks. These rocks plot within the intra-oceanic arc fields, as shown in the Th/Nb versus La/Nb, La versus Th, La/Yb versus Th/Yb, and Th/Yb versus Ta/Yb diagrams (Figs. 10C–10F).

Dated zircon grains of the gabbro-dioritic rocks are also employed for Hf isotopic determination. The ε_Hf(t) values range from +10 to +16 (with a peak at ∼+14), corresponding to the depleted mantle modal ages of 500–240 Ma (with a peak at ca. 300 Ma) (Figs. 11A–11B).

Initial isotopic ratios were calculated for the crystallization age of the plutonic rocks. The results show that these samples have a narrow range of ⁸⁷Sr/⁸⁶Sr ratios (0.703714–0.704119) and initial ⁸⁷Sr/⁸⁶Sr ratios of 0.703432–0.703739. In addition, the plutonic rocks are characterized by positive ε_Nd(t) values ranging from +4.5 to +5.5 (Fig. 11C).

The magmatic amphibole (hornblende) and plagioclase pairs in the plutonic rock of sample xm83 were selected for EMPA analysis. According to the nomenclature of Leake et al. (1997), the amphiboles fall into the magnesiohornblende field (Fig. 11D). These hornblendes are characterized by Ca_A (Ca in the A site) <0.50, (Na + K)_A <0.50, and Ca >1.50 on the M₄ sites. Therefore, the selected hornblendes are typical of magmatic calcic amphiboles. The analyzed plagioclases exhibit a narrow range of anorthite (An) and albite (Ab) values. In the calculations of hornblende-plagioclase geothermometry, the mean composition of all of the plagioclases was adopted. The geothermometry yields a ∼720 °C crystallization temperature for the gabbro-dioritic pluton.

Based on the roundness classification of Finzel (2017), all of the analyzed grains from the sandstones of the Langjiexue Group in the Tethyan Himalaya are classified using a I–X roundness scale. It is clear that the Precambrian zircon grains are more rounded, above IV on the roundness scale. In contrast, the zircon grains with ages of 300–200 Ma are less rounded, ranging from I to IV on the roundness scale (Fig. 12). Similarly, the zircon grains of 300–200 Ma have fresh surface microstructures compared to the Precambrian zircon grains. No obvious chemical or mechanical impact features could be seen from these younger zircon grains (Fig. 13). In contrast, the older ones show common chemical (scaling, weathered surfaces) and mechanical (upturned plates, abrasion features, fracture faces, conchoidal fractures) features.

The intrusive complex consists of a group of coeval plutons and hypabyssal rocks that range from mafic to felsic in composition but are dominated by hornblende gabbro to diorite. The most common mafic mineral in this suite is hornblende, which typically occurs as both large prismatic phenocrysts and small crystals in the matrix. No pyroxene was found in field exposures or in the thin sections of the studied rocks. In addition, plagioclase commonly occurs as small laths between the hornblende phenocrysts, and crosscuts and postdates the hornblende minerals. These observations are congruent with the importance of water in the early paths of magma differentiation; the presence of water suppressed the early crystallization of plagioclase when olivine and pyroxene fractionated (Feig et al., 2006; Grove et al., 2012; Smith et al., 2009; Wan et al., 2013).

These plutonic rocks plot in the calc-alkaline field, rather than the tholeiitic field (Figs. 9B–9C), supporting a dominant role of magmatic water (Zimmer et al., 2010). Once magma is wet, the magmatic water would lead to the crystallization of calcic plagioclase while reducing the total proportion of plagioclase in the mineral assemblage, thereby facilitating the development of the calc-alkaline differentiation trend (Sisson and Grove, 1993a).

Generally, the crystallization temperature of a normal basaltic magma is very high. However, if the magma is water saturated, the crystallization temperature would be significantly reduced (Lee et al., 2009a; Wan et al., 2013). The various geothermometers employed in this study yield similar crystallization temperature ranges for the gabbro-diorite complex (Fig. 14). The Zr–in–whole rock geothermometer yields a weighted mean crystallization temperature of ∼713 °C (Table S8 [^{footnote 1}]). Hornblende and hornblende-plagioclase geothermometers present a weighted mean crystallization temperature of ∼720 °C. The Ti-in-zircon geothermometer calculated ∼720 °C for the gabbro-diorite complex (Table S9). Taken together, we find that the crystallization temperature of the gabbro-diorite complex is sharply lower than that of normal basaltic magma. Furthermore, the high alumina contents (15%–17%) of these calc-alkaline rocks, potentially controlled by the delay of plagioclase nucleation, also support a high water content in the magma (Crawford and Falloon, 1987; Kelley et al., 2010).

The water content in continental crust is <0.001 wt%, and <0.1 wt% in the asthenospheric and lithospheric mantle (Williams and Hemley, 2001). However, a high amount of water (>3 wt%) within the magma is needed for the crystallization of hornblende (Grove et al., 2012; Sisson and Grove, 1993a). Generally, the H₂O of arc magma is generated by the dehydration of hydrous minerals, which are carried down by the subducting oceanic slab. The water is released into the overlying mantle wedge, where melting initiates and ascends as hydrous magma (Grove et al., 2012). Therefore, a subduction zone is the best locus to form these magmas, where additional hydration for the magma can occur from the mantle wedge metasomatized by subduction slab-derived fluid (Grove et al., 2012; Hirschmann, 2006; Murphy, 2013).

Zircon grains of gabbro-dioritic rocks show highly positive ε_Hf(t) values (+10 to +16, with a peak at ∼+14; Fig. 11A). Correspondingly, the zircon Hf T_DM modal ages are centered around ca. 300 Ma (Fig. 11B), slightly older than the U-Pb crystallization ages of ca. 240 Ma, indicating short crustal residence time and significant contributions of juvenile addition. Likewise, the whole-rock ε_Nd(t) and (⁸⁷Sr/⁸⁶Sr)_i (_i stands for initial) cluster around ∼+5 and ∼0.703570, respectively (Fig. 11C), further suggesting that the magma source of the gabbro-diorite complex was derived from ∼20% melting of the depleted mantle wedge (Fig. 11E). The compositions of hornblendes fall into the magnesiohornblende field (Leake et al., 1997; Fig. 11D), typical of magmatic calcic amphiboles in I-type plutons, and therefore suggest a subduction setting (Clemens and Wall, 1984). This conclusion is corroborated by the slightly negative Eu anomaly and the significantly negative anomalies of Nb, Ta, and Ti (Figs. 10A–10B), thereby implying that a hydrated mantle wedge is an ideal source for the intrusive complex.

As shown in Figures 10A–10B, the plutonic rocks show similar REE and trace element patterns to the western Aleutian high-alumina basalts and andesites. These plutonic rocks have relatively high Al₂O₃ and low MgO contents, therefore referred to as low-MgO high-alumina basalts (HABs, with <7 wt% MgO) and basaltic andesites (BAs, with <5 wt% MgO) (Crawford and Falloon, 1987; Sisson and Grove, 1993b). HABs and BAs are the dominant igneous products in some modern intra-oceanic arcs such as the western Aleutians (Kay and Kay, 1985; Schiano et al., 2004), the Barren arc (northeastern Indian Ocean) (Luhr and Haldar, 2006), the Lesser Antilles (Melekhova et al., 2017), the South Sandwich Islands (Crawford and Falloon, 1987), as well as the Late Carboniferous Bogda arc, Chinese Tianshan (Xie et al., 2016). In addition, Nb contents and the Rb/Zr ratios are very useful indicators for identifying the geological setting of the granitoid rocks (Brown et al., 1984). The studied plutonic rocks have low Nb contents (0.5–1.7 ppm) and low Rb/Zr ratios (0.09–0.46) (as shown in Table S5 [^{footnote 1}]), consistent with rocks in primitive island arcs. Furthermore, the Th/La ratios (0.13–0.24; Table S5) are slightly higher than the ∼0.12 values found in intra-oceanic arc suites (Jolly et al., 2001), but are obviously lower than the >0.45 values found in the continental-margin Aeolian arc (Ellam et al., 1988). These Th/La ratios reflect values similar to those of typical of oceanic arcs, but with a little contamination of source material by terrigenous sediments (Jolly et al., 2001). Likewise, with Ce/Yb ratios mainly clustering around 14.3–25.5 (Table S5), the studied plutonic rocks correspond to the low Ce/Yb array of Hawkesworth et al. (1993), typical of modern intra-oceanic arc volcanic rocks. In Figure 11F, all data fall into fields close to those of modern intra-oceanic arcs, but with some exceptions falling into the Andes Southern Volcanic Zone (SVZ) field, revealing a slight involvement of subducted sediments into the magma source. We thus conclude that the intrusive complex most likely represents a component of a former intra-oceanic arc within the Neotethys (Figs. 10C–10F). However, a continental arc affinity for the intrusive complex cannot be completely excluded. More geological investigations are needed in future work.

A group of models has been proposed for the tectonic setting for the Langjiexue Group (Cai et al., 2016; Cao et al., 2018; Dai et al., 2008; Fang et al., 2018; Li et al., 2010, 2016; Wang et al., 2016b; Webb et al., 2012; Zhang et al., 2015a, 2017). Nowadays, the most popular model is a passive continental margin basin along the northern margin of the Gondwanan landmass (Cai et al., 2016; Wang et al., 2016b; Cao et al., 2018; Fang et al., 2018). From our perspective, this model is acceptable and reasonable, to some extent. However, one issue needs to be dealt with: the scarcity of nearby Triassic magmatic belt or arc rocks in the northern Gondwanan landmass as the provenance for the remarkable Triassic zircon grains of the Langjiexue Group. The model of a forearc basin along the active continental margin arc of the Lhasa terrane or along the intra-oceanic arc within the Neotethys has been proposed by Li et al. (2010, 2016) and Zhang et al. (2017). However, due to the paucity of regional geological investigation in the Gangdese belt and Tethyan Himalaya terrane, the location of the intra-oceanic arc rocks is still an open question.

Several salient age peaks are shown in the detrital zircon U-Pb age spectra for the sandstones of the Late Triassic Langjiexue Group in the Tethyan Himalaya (Fig. 8). There is no obvious difference between our results and published data (Cai et al., 2016; Wang et al., 2016b; Cao et al., 2018; Fang et al., 2018). Before dispersal of the Pangea supercontinent during the late Carboniferous–early Permian, all of the Indian and Australian blocks, as well as the Cimmeride continental blocks (e.g., Lhasa, Iran terranes, etc), were integral constituents of the united Gondwanan landmass. In view of this, all of these continents should have the same or similar pre-Permian detrital zircon grains, as shown by the Grenville (1100–750 Ma) and Pan-African (650–500 Ma) zircon grains in the sandstones from Tethyan Himalaya (Fig. 8). Therefore, the zircon grains with ages >300 Ma may not be useful for identifying a specific provenance in the Pangea supercontinent. We thus focus on zircon grains with ages ranging between 300 and 200 Ma to better constrain the tectonic setting for the Langjiexue Group in the Tethyan Himalaya. The youngest zircon U-Pb age peak of ca. 210 Ma approximates the maximum depositional age for the Langjiexue sandstones (Fig. 8). The salient age peaks of ca. 210 and 252 Ma necessitate penecontemporaneous magmatism in the adjacent region (Li et al., 2010, 2016; Cai et al., 2016; Wang et al., 2016b; Fang et al., 2018; Meng et al., 2019b).

Previous studies have argued against an intra-oceanic arc-derived or Lhasa-derived model due to the lack of identified Early–Middle Triassic magmatic rocks within the Gangdese belt (Cai et al., 2016; Wang et al., 2016b). However, this is no longer the case. Several exposures of Middle–Late Triassic magmatic assemblages have now been identified in the Gangdese belt, southern Tibet (as shown in Fig. 2)—for example, the 237–211 Ma Changguo and Beise volcanic rocks in southern Lhasa (Wang et al., 2016a, 2018), the ca. 220–215 Ma cumulate appinite in the Quxu region (Meng et al., 2016b; Ma et al., 2018b), the 230–225 Ma Daga granite (Meng et al., 2018), the 212–206 Kazi granite (Ma et al., 2017a), and the 203–201 Ma Cuijiu igneous complex (Xu et al., 2019). The Changguo and Beise volcanic rocks, Daga granite, and Cuijiu igneous complex are considered to have been formed in an active continental margin (Wang et al., 2016a, 2018; Meng et al., 2018; Xu et al., 2019), while the Quxu ca. 215 Ma cumulate appinitic suite is proposed to have affinity with rocks in an intra-oceanic arc setting (Ma et al., 2018b). These new findings support a possibility of northern provenance, whether an active continental margin arc or an intra-oceanic arc, which supplied some material to the sandstones of the Langjiexue Group.

Detrital zircon microstructure (grain shape and surface texture) is a useful tool to help decipher polycyclicity and transport processes (Finzel, 2017; Gärtner et al., 2013; Vos et al., 2014). Most of 300–200 Ma detrital zircon grains in the sandstones of the Langjiexue Group in the Tethyan Himalaya fall into the range of I (completely unrounded) to IV (poorly rounded) on the roundness scale (Fig. 12). In contrast, most zircon grains with ages >300 Ma fall in the range of V (fairly rounded) to X (completely rounded), indicating a high degree of rounding. The surface microstructures of these detrital zircon grains are different between these two age populations. For the zircon grains with ages >300 Ma, the most prominent are preweathered and weathered surfaces. The preweathered surfaces are marked by minute pits and etching triggered by acid acting on the zircon surface (Fig. 13). The preweathered surfaces are then overprinted by weathered features (fractures or abrasion). Furthermore, other microstructures, including craters, conchoidal fractures, V-shaped cracks, and fracture faces, are found on the Precambrian zircon grains. These features are probably caused by mechanical grinding, grain-to-grain impacts, or scraping. In contrast, the 300–200 Ma zircon grains uniformly feature fresh surfaces, occasionally with some abrasion features. These microstructural observations further suggest that the zircon grains of 300–200 Ma were probably not subjected to long-distance transportation or polycyclic recycled processes. Therefore, a near source of 300–200 Ma detrital zircon grains was necessary for the formation of the sandstones of the Langjiexue Group in the Tethyan Himalaya. However, some studies doubt the conclusion that there is some positive relationship between the transporting distance and the zircon textural maturity (Garzanti, 2017); different transport systems, even from similar environments, do not have consistent length or duration (Zoleikhaei et al., 2016). Consequently, future investigations beyond the zircon textural maturity are needed to precisely define the transport distance, systems, and provenance of the sandstones in the Langjiexue Group.

The sandstones of the Langjiexue Group are dominated by fine to medium sand grains of feldspar and lithic sandstone or greywackes. These greywackes consist of angular to subangular feldspar and quartz as well as lithic fragments (sedimentary, volcanic, and metamorphic lithics, with lithic volcanic fragments accounting for ∼5%–20%), showing poorly to moderately sorted features of low to moderate compositional and textural maturity (Cai et al., 2016; Wang et al., 2016b; Zhang et al., 2017; Meng et al., 2019b). This low to moderate maturity is likely related to short-distance transportation. In addition, the lithic volcanic fragments and subangular plagioclase confirm the existence of contemporaneous volcanism adjacent to the deposition location. Although there is a consensus that the Langjiexue Group was deposited along or adjacent to the northern passive continental margin of the Gondwanan landmass (Cai et al., 2016; Wang et al., 2016b; Cao et al., 2018; Fang et al., 2018), the detrital zircon U-Pb age histograms exhibit a salient age peak of ca. 210 Ma close to the deposition age of the strata. This feature indicates that the sediments of the Langjiexue Group were probably deposited in a convergent basin, contrasting with a passive continental margin basin (Cawood et al., 2012). Furthermore, whole-rock geochemistry, heavy minerals, and framework petrology consistently imply an orogenic provenance for the Langjiexue sandstones (Meng et al., 2019b). Taking all of these observations together, we interpret that the Langjiexue Group has multiple sources. This conclusion is in good agreement with the observations of Li et al. (2016): numerous Cr-spinels found in the Langjiexue Group exhibit contents in Al₂O₃ of 5%–257%, in TiO₂ of 0.01%–1.0%, in Cr₂O₃ of 44%–100%, and in Cr# (Cr/(Cr + Al)) of 48%–95%, revealing several distinctive parent lithologies.

How do we incorporate these seemingly paradoxical tectonic regimes into one reasonable and feasible configuration? We contend that the Langjiexue Group was deposited adjacent to the passive Gondwanan continental margin (Cai et al., 2016; Wang et al., 2016b), but associated with deep-ocean submarine fans (Li et al., 2016; Zhang et al., 2017). This submarine fan basin was near an intra-oceanic arc (Li et al., 2010, 2016). Taken together, our observations suggest that the Langjiexue Group was deposited not far from volcanic loci in an intra-oceanic arc, such as that represented by the 220–215 Ma (Ma et al., 2018b) and ca. 240 Ma intrusive complex of the present study in the Gangdese belt.

The model of a forearc basin along the southern margin of the Lhasa terrane cannot be excluded fully for the Langjiexue Group. Regarding this model, however, several issues have to be dealt with. First and foremost, some suture zones should be found between the Langjiexue Group and the Indian plate. So far, no such suture zone has been reported. Secondly, the Langjiexue Group occupies a very large area in the Himalayan region and contains abundant Precambrian zircon grains, which requires enough provenance, especially Precambrian basement. Unfortunately, the southern Lhasa subterrane features juvenile crust without obvious fingerprints of the Precambrian basement (Ji et al., 2009). Furthermore, due to the occurrence of one or two suture zones within the Lhasa terrane, we have to admit that the Lhasa terrane was not intact during the Mesozoic period (Zeng et al., 2018; Zhu et al., 2018). In such a case, the northern Lhasa subterrane could not have provided plentiful material for the formation of the Langjiexue Group. Last but not least, when did the forearc basin rift away from the Lhasa terrane and accrete with the Indian continent? No obvious clues have been indicated by previous research. Thus, the model of an intra-oceanic arc within the Neotethys is more likely to have been another possible source for the Langjiexue Group than a forearc basin along the southern margin of the Lhasa terrane. We tentatively prefer the interpretation of an intra-oceanic arc system within the Neotethys that supplied some material to the Langjiexue Group.

Studies on the formation of the Gangdese magmatic belt play a key role in understanding the Neotethyan history (Lang et al., 2017; Ma et al., 2018a; Wang et al., 2017; Yang et al., 2017; Zhu et al., 2013). One primary question centers on the onset of subduction of the Neotethys, namely: When was an active continental margin initiated along the southern margin of the Lhasa terrane? During the last decade, ages of 237–170 Ma for calc-alkaline magmatism in the Gangdese magmatic belt have indicated that an active continental margin existed along the southern margin of the Lhasa terrane. Examples include Middle Triassic–Jurassic volcanic rocks of the Yeba, Bima, and Xiongcun Formations (Kang et al., 2014; Ma et al., 2017b, 2018a; Wei et al., 2017; Zhang et al., 2012; Zhu et al., 2008) and their coeval plutonic equivalents (Chu et al., 2006; Ji et al., 2009; Lang et al., 2017; Tafti et al., 2014; Wang et al., 2016a, 2017; Xu et al., 2017, 2019). These findings indicate that the southern margin of the Lhasa terrane was subjected to intense Middle Triassic to Jurassic magmatism due to subduction of the Neotethyan oceanic lithosphere (Wu et al., 2010).

Li et al. (2010) proposed that the northward subduction of the Neotethyan oceanic lithosphere beneath the Lhasa terrane triggered intense magmatism in the Lhasa terrane. These arc magmatic rocks probably provided a significant contribution to the Upper Triassic Langjiexue Group flysch in the Tethyan Himalaya. Within the Lhasa terrane, the Paleotethys (Sumdo Ocean), represented by the Sumdo eclogite and peridotite, was considered to have been closed before 220–240 Ma (Li et al., 2011; Yang et al., 2007). This observation precludes the possibility that the Middle Triassic–Jurassic magmatism in the Gangdese belt was induced by southward subduction of the Sumdo oceanic lithosphere. The Bangong-Nujiang eclogites are mainly distributed along the southern margin of the South Qiangtang terrane, immediately north of the Bangong-Nujiang ophiolitic mélange zone, suggesting a northerly subduction polarity (Zhang et al., 2015c). Based on paleomagnetic results of Triassic rocks from the Lhasa terrane, Zhou et al. (2016) proposed that the Bangong-Nujiang Ocean separating the Lhasa and Qiangtang terranes opened up in the Early–Middle Triassic period and continued to expand throughout the Triassic. If correct, this precludes southward subduction of the Bangong-Nujiang oceanic slab. Given the above observations, the likelihood of a Middle Triassic to Jurassic active continental margin in the southern Lhasa terrane remains likely, with the catalyst being the northward subduction of the Neotethyan oceanic slab.

However, the Langjiexue Group has multiple sources, rather than a single source (Li et al., 2016). The evidence for this conclusion comes from the following: (1) bulk-rock Sr-Nd isotopes of sandstones in the Langjiexue Group (epsilon Nd values between −7 and −3) reveal a more depleted source than traditional passive continental margin sediments (Dai et al., 2008); (2) numerous Cr-spinels of the Langjiexue Group exhibit contents in Cr# of 48%–95%, in Cr₂O₃ of 44%–100%, in Al₂O₃ of 5%–257%, and in TiO₂ of 0.01%–1.0%, indicating multiple parent lithologies (Li et al., 2016); (3) a large percentage of older detrital zircon grains (with ages of 3300–300 Ma) imply a probable passive continental source (Cai et al., 2016; Meng et al., 2019b); (4) a remarkable age population of 260–200 Ma denotes an active convergent margin (Wang et al., 2016b; Fang et al., 2018); and (5) detrital zircon grains with ages between 260 and 200 Ma have a large range of zircon epsilon Hf values (−12 to +15) (Li et al., 2010; Wang et al., 2016b). These observations probably reveal the existence of a subduction-related arc as a source of the Langjiexue Group.

Combining published results with those from our study suggests that neither an active continental margin nor a passive continental margin is enough to explain the all sources for the Langjiexue Group during the Triassic (Li et al., 2016). Therefore, it is worth considering whether an intra-oceanic arc system existed within Neotethys.

Globally, intra-oceanic arcs comprise ∼40% of the subduction margins of the Earth (Larter and Leat, 2003). Thus, we wonder: Did the Neotethyan Ocean have scattered intra-oceanic arcs during the Triassic? New findings reveal the possible existence of a Middle Triassic intra-oceanic subduction system in the western section of the Neotethys (Sayit et al., 2015, 2017; Tekin et al., 2016). Furthermore, the 220–215 Ma cumulate appinite in the Quxu region, southern Lhasa, was likely formed in an intra-oceanic arc setting (Ma et al., 2018b). These observations, in conjunction with the voluminous Middle Triassic to Jurassic calc-alkaline rocks hosted within the Gangdese belt, probably support an intra-oceanic subduction system within the Neotethys. The modern western and northern Pacific are characterized by a number of intra-oceanic arcs, such as the Ryukyu-Luzon, Izu-Bonin-Mariana, and western Aleutian arc systems (Mishin et al., 2008; Pearce et al., 2005; Singer et al., 2007; Stern et al., 2003), and may provide analogues to the Triassic Neotethyan framework.

An intra-oceanic arc setting probably plays a pivotal role in the provenance of the Langjiexue sandstones in the Tethyan Himalaya (Li et al., 2016; Zhang et al., 2017). A Late Triassic northward intra-oceanic subduction setting has been proposed by Li et al. (2010), in which the Langjiexue Group was deposited in a forearc basin. Recently, Cao et al. (2018) delineated a Middle Triassic northward intra-oceanic subduction mosaic within the Neotethys. As shown in Ce/Ce* versus Eu/Eu* plots (where Ce and Eu are the chondrite-normalized Ce and Eu concentrations, and Ce* and Eu* are the averages of the chondrite-normalized La and Pr concentrations, and Sm and Gd concentrations, respectively), many of the 300–200 Ma detrital zircon grains of the Langjiexue sandstones have close affinity with those of the arc-type gabbro-diorite complex in the Gangdese belt (Fig. 15A), and a Th-Pb diagram (Fig. 15B) shows affinity to the I-type granitoids, probably indicating a close relationship with the subduction-related arc. Similarly, many younger detrital zircon grains fall into the volcanic arc basalt field in a U-Er plot (Fig. 15C). Most of the young detrital zircon grains have very positive epsilon Hf isotopes (Fig. 15D), revealing the depleted nature of their magma source. This conclusion is consistent with Li et al. (2016)’s observation: Cr-spinels of the Langjiexue sandstones having relatively low TiO₂ (0%–2%) and Al₂O₃ (5%–30%), in concert with arc-basalt or mantle-peridotite sources (Kamenetsky et al., 2001). To some extent, these signatures provide robust constraints for the conclusion that some of the 300–200 Ma detrital zircon grains of the Langjiexue sandstones in the Tethyan Himalaya come from an intra-oceanic arc. These observations necessitate two prerequisites: one is the existence of an intra-oceanic arc, and the other is the free transportation of the intra-oceanic arc material to the Langjiexue Group.

Recently, Huang et al. (2018) documented ca. 230 Ma bimodal intrusive rocks (diabase and gabbro and monzonite) that intruded into the Nieru Formation in the southern margin of the Langjiexue Group. This bimodal magmatism was most likely formed in a rifting-related backarc basin, triggered by the southward subduction of the Neotethyan oceanic slab. We thus argue that an intra-oceanic southward subduction system occurred within the Neotethys as early as the Middle Triassic, with an accompanying intra-oceanic arc supplying some sediments to the deposition of the Langjiexue Group under a submarine fan environment (Fig. 16).

The timing of the opening of the Neotethys remains an open question. Based on the late Carboniferous–early Permian large igneous provinces in the Indian and Australian blocks, as well as coeval extensive basalts and mafic dikes in southern Tibet, the Pangea supercontinent was suggested to have broken up during the early Permian, which triggered the initial opening of the Neotethys (Chauvet et al., 2008; Garzanti et al., 1999; Lapierre et al., 2004; Stojanovic et al., 2016; Veevers and Tewari, 1995).

The cause of subduction initiation is still an enigmatic puzzle (Stern and Gerya, 2017). In some models, the cause is spontaneous nucleation due to gravitational instability of oceanic lithosphere at a transform or fracture zone or at a passive margin (Stern and Gerya, 2017). The subduction initiation at a transform or fracture zone was employed to explain the Izu-Bonin-Mariana arc system (Hawkins et al., 1984; Stern and Bloomer, 1992). Subduction is initiated because two lithospheres of different density are juxtaposed across a transform or fracture zone (lithospheric weakness), where old and dense lithosphere would be depressed beneath the level of the asthenosphere (Stern, 2004; Stern and Gerya, 2017). Recent numerical modeling reveals that the denser lithosphere should be at least 30 m.y. older than the younger lithosphere for initiation of an intra-oceanic subduction system (Zhou et al., 2018). Based on our interpretation that the ca. 240 Ma gabbro-diorite complex in the present study formed in an intra-oceanic arc setting, we suggest that the opening of the Neotethys should have been earlier than ca. 270 Ma, in concert with the above-mentioned published model.

The earliest Permian Gondwanan glaciation is documented by diamictites containing palynoflora in Oman, Pakistan, India, Lhasa, Australia, and South America, indicating that all of these regions were in the same paleophytogeographic province (Angiolini et al., 2003; Zhang et al., 2015b). However, the succession at the top of the diamictites is rich in wood logs, indicating glacial retreat and concomitant encroachment of a shallow epicontinental sea. This is in good agreement with the early–middle Permian transition from continental to shallow-marine sediments (Angiolini et al., 2003; Ji et al., 2005), which is characterized by the admixture of warm- and cold-water faunas in the Cimmerian province (including Turkey, Iran, South Pamir, Himalaya, Lhasa, Qiangtang, Baoshan, and Sibumasu terranes) (Ueno, 2003; Zhang et al., 2013).

In the western Tethyan Himalaya, sandstones and conglomerates of lowermost Permian diamictites rich in bryozoans and brachiopods were deposited in a shallow marine setting. These clastics paraconformably overlie lowermost Permian diamictites in the Spiti Valley (northern India), and disconformably overlie lower Carboniferous arenaceous to carbonate sedimentary rocks in the Pin Valley (northern India) (Garzanti et al., 1996). This unconformity continues into the Lahaul and Zanskar regions (northern India). Further east in the Karakoram terrane, pre-Permian strata are unconformably overlain by a lowermost Permian sandstone (Zanchi and Gaetani, 2011). Similarly, upper Carboniferous mudstone and sandstone are unconformably or disconformably overlain by earliest Permian coastal-marine conglomerate and sandstone in the Lhasa terrane (Zhang et al., 2015b; Zhao et al., 2001). This major unconformity probably marks the end of rifting of the Pangea supercontinent, followed by the initial opening of the Neotethys (Garzanti et al., 1996). This conclusion is supported by extensive early–middle Permian rifting-related basalts and diabases in the Tethyan Himalaya (Zeng et al., 2012; Zhu et al., 2010) and recent paleogeographic reconstructions (Kroner et al., 2016; Stampfli and Borel, 2002; Torsvik et al., 2012; Xiao et al., 2015).

Taking into account all of the above observations, we thus emphasize that the Neotethys likely opened during the middle Permian. However, the geodynamic mechanism for the breaking apart of the Lhasa terrane from the Gondwana landmass and the subduction initiation of the Neotethys remain areas for future work.

The Middle Triassic (ca. 240 Ma) gabbro-diorite complex in the Gangdese magmatic belt, southern Tibet, exhibits depleted whole-rock and mineral isotopic signatures. The plutonic rocks are dominated by magmatic hornblende and hornblende phenocrysts with plagioclases occurring as an interstitial phase, which, in conjunction with the lower crystallization temperature (∼720 °C), indicates that the magma source was wet. The geochemical characteristics of the plutonic rocks indicate that the gabbro-diorite complex was probably formed in a subduction-related arc setting, either an active continental margin arc of the Lhasa terrane or an intra-oceanic arc within the Neotethys.

The sandstones of the Langjiexue Group in the Tethyan Himalaya, southern Tibet, yield a wide detrital zircon age range of 3200–200 Ma, with peaks at ca. 2440 Ma, ca. 892 Ma, ca. 579 Ma, ca. 252 Ma, and ca. 210 Ma among others, revealing a multi-source provenance. Zircon microstructures suggest that most of the detrital zircon grains with ages ranging between 300 and 200 Ma show a low level of roundness and no obvious destructive surface microstructure. These observations are consistent with the existence of abundant lithic volcanic fragments in the sandstones. Trace elements of these younger detrital zircon grains show a close relationship to a subduction-related arc setting, with highly positive zircon epsilon Hf values. These observations indicate that some of these younger detrital zircon grains are probably derived from subduction-related arcs, such as an intra-oceanic arc within the Neotethys.

Considering the characteristics of the gabbro-diorite complex and the sandstones together, in conjunction with regional geology, we argue that the Neotethys most likely opened prior to the Middle Triassic, and that multiple sources (including arcs) provided some sediments to the deposition of the Langjiexue Group.

We thank Wenjiao Xiao, Xiumian Hu, and Xianghui Li for sharing ideas and insights into the tectonic framework of the Neotethyan realm. We are indebted to Scott Paterson for his invaluable comments and language polishing on the first draft. Many fruitful discussions with Weiqiang Ji, Guangwei Li, and Jiangang Wang helped to clarify some correlations between the magmatism and sedimentology in southern Tibet. For technical support, we thank Anlin Ma for data analysis, Peixi Zheng and Zheng Wang in zircon dating, Bin Shi in zircon CL and SEM analysis, and Liang Li in whole-rock Sr-Nd analysis. This research was co-supported by the Second Comprehensive Scientific Investigation into Qinghai-Tibet Plateau (SQ2019QZKK2703), National Natural Science Foundation of China (41502198), Research Grant of Chinese Academy of Geological Sciences (J1703), Fund of China Scholarship Council (201809110055), Postdoctoral Scientific Foundation of China (2016T90122), and Geological Survey of China (DD20190060). We are indebted to Christopher Spencer, Ryan Leary, and an anonymous reviewer for their thorough and constructive comments that significantly improved this manuscript. We would like to thank Geosphere Science Editor Shanaka de Silva for guidance and editorial handling of this manuscript and Managing Editor Gina Harlow for kindness and efforts on this paper. Xuxuan Ma would like to thank his wife, Yao Juliana Xu, for her continuing support and encouragement.