Early Triassic roll-back of subducted Paleo-Tethys oceanic lithosphere: Insights from A₂-type silicic igneous rocks in the Pingxiang area, southwest China Open Access

The Triassic closure of the eastern branches of the Paleo-Tethys Ocean and associated backarc basins resulted in a vast composite continent in eastern and southeastern Asia, which comprises the North China Craton, South China Block (SCB), Indochina Block, Baoshan (Sibumasu) Block, and Qiangtang Block (Fig. 1; Metcalfe, 2011). The tectonic development of South China during this time involved large-scale intracontinental deformation and associated magmatism. The Triassic tectonic history of the southwestern SCB was affected by the Indosinian orogeny (Fromaget, 1941) and also records the amalgamation of the Indochina Block and SCB during the late Permian to Triassic due to closure of the eastern branch of the Paleo-Tethys Ocean (Cao et al., 2015; Halpin et al., 2016). However, subsequent studies have advocated that subduction of the Paleo-Pacificplate also occurred beneath the SCB in the late Permian and proposed a flat-slab subduction model followed by slab foundering to account for both the Indosinian orogeny and broad spatial distribution of the late Mesozoic magmatism (e.g., Li and Li, 2007; Li et al., 2012a). Although the origin and tectonic implications of the Mesozoic igneous rocks in South China have been extensively studied, the link between the Permian–Triassic igneous rocks in the southwestern SCB and a specific subduction zone is still controversial (Wang et al., 2021). The Mesozoic tectonomagmatism in the SCB is thought to be related to subduction in the Paleo-Tethys Ocean (Jiao et al., 2015; Li et al., 2016; Xu et al., 2018; Wang et al., 2020) or the Paleo-Pacific Ocean (Li et al., 2006, 2012b; Li and Li, 2007).

There are two Paleo-Tethys sutures, the Song Ma and Song Chay suture zones, between the Indochina Block and the SCB (Fig. 1). The Dian-Qiong suture (DQS) was identified by Cai and Zhang (2009). The western DQS approximately follows the China-Vietnam border, and the eastern DQS crosses Hainan Island (Fig. 1). The western DQS extends from Jianshui close to the Red River, through Babu and Napo, to Pingxiang in theChina-Vietnam border area (Fig. 2A; Dong and Zhu, 1999; Wu et al., 1999, 2000, 2002). The DQS contains high- to ultrahigh-pressure metamorphic belts, early Carboniferous to early Permian ophiolites and mid-oceanridge basalts, late Permian to early Triassic island-arc tholeiites and arc-related igneous rocks, and lower-middle Triassic sedimentary rocks (Li and Li, 2007; Cai et al., 2008; Cai and Zhang, 2009; Hu et al., 2012; Qiu et al., 2017; Li et al., 2019; Xiang et al., 2021). Analysis of these rocks has constrained the timing of opening and closure of the Paleo-Tethys Ocean and the subsequent accretional and collisional tectonic events. Widespread middle to late Triassic silicic igneous rocks along the suture can constrain the tectonic evolution of the continental crust and its magmatic history (Qi et al., 2007; Qin et al., 2018; Li et al., 2019). Gao et al. (2017) suggested that the ca. 240 Ma silicic magmatism in southwestern South China occurred during a tectonic transition from compression to extension but was mainly caused by extension. However, controversy remains regarding the timing of the evolution of the Paleo-Tethys, due to the variable ages and tectonic settings of the igneous rocks (Sun and Jian, 2004; Peng, 2006; Hennig et al., 2009; Zi et al., 2012a). Previous studies have proposed that subduction in the Paleo-Tethys Ocean started in the early Permian (Li, 2014; Wang et al., 2018a) and that the ocean closed during the middle Triassic (Zi et al., 2012a; Wang et al., 2018a). Other evidence from volcanic rocks suggests that subduction along the suture zone occurred during the early Permian–middle Triassic, followed by collision (Xin, 2018; Xin et al., 2018). Therefore, the early Triassic igneous rocks along the suture can provide clues to the subduction and collisional processes prior to closure of the Paleo-Tethys Ocean.

Many of the Permian–earliest Triassic igneous rocks along the Ailaoshan–Song Ma suture zones have arc geochemical signatures with marked negative Nb-Ta anomalies, which indicate a subduction setting related to convergence of the SCB and Indochina Block (Zi et al., 2012b; Lai et al., 2014; Liu et al., 2015; Wang et al., 2018b). The Permian–Triassic igneous rocks along the DQS are only exposed in the southwestern Youjiang Basin where they occur in the lower-middle Triassic Beisi and Banba formations (Fig. 2; e.g., Wang and Deng, 2003; Qin et al., 2011, 2012; Li et al., 2019; Gan et al., 2021). The age and petrogenesis of these igneous rocks have only been constrained in limited areas of this region (Wang and Deng, 2003; Qin et al., 2011, 2012; Li et al., 2019; Gan et al., 2021). The basalts, rhyolites, dacites, and granites in the Ningming, Pingxiang, and Chongzuo areas along the southwestern margin of the Youjiang Basin formed from 256 Ma to 241 Ma during the early to middle Triassic and exhibit subduction-related arc-like geochemical features that are thought to be related to subduction in the Paleo-Tethys Ocean or syn- or post-collisional processes between the SCB and Indochina Block (Wang and Deng, 2003; Qin et al., 2011, 2012; Li et al., 2019; Gan et al., 2021). As such, the age and petrogenesis of the Permian–Triassic silicic igneous rocks in the Youjiang Basin need to be further constrained and may provide new insights into the tectonic evolution of the Paleo-Tethys Ocean between the SCB and Indochina Block (Wang and Deng, 2003; Qin et al., 2011; Gan et al., 2021).

In this paper, we report whole-rock major- and trace-element, whole-rock Nd and zircon Hf isotope, and zircon U-Pb age data for newly discovered peraluminous A₂-type rhyolites and biotite granites from the Pingxiang region, Guangxi Province, southwest China. We use these data to constrain the petrogenesis and tectonic setting of these rocks, which provide new insights into the tectonic evolution of the Paleo-Tethys Ocean during the Triassic.

South China is a continental terrane that is bounded to the north by the Qinling-Dabie collisional orogenic belt, to the east by the Pacific plate, and to the south by the Indochina Block (Fig. 1; Li et al., 2017). It is a collage consisting of the Yangtze Block in the northwest and Cathaysia Block in the southeast, which were amalgamated during the Neoproterozoic (860–820 Ma) along the Jiangshan-Shaoxing fault (Fig. 1; Li et al., 2009; Zhao, 2015; Xia et al., 2018; Yao et al., 2019). The Youjiang Basin is located at the southwestern margin of the SCB (Fig. 1) and is generally considered to have developed between the early Devonian and late Triassic, associated with the evolution of the Paleo-Tethys Ocean (Du et al., 2013; Hu et al., 2017). The Pingxiang area is located at the southwestern margin of the Youjiang Basin in the western DQS (Fig. 1). In this area, the combined effects of the coastal Pacific and Tethys tectonic domains resulted in frequent marine transgressions and regressions, intense magmatism, and a complex geological history.

Carboniferous to Paleogene strata are exposed in the Pingxiang area, but the Triassic rocks are most widely exposed (Fig. 2B). Based on the paleontology, contact relationships, and lithological features, the Triassic strata can be divided into a lower and middle series. The middle Triassic Banna Formation consists of sandstone. The lower Triassic Beisi Formation consists mainly of dolomite and volcanic rocks (Fig. 2C). There were two stages of magmatism in the area: Variscan (i.e., late Paleozoic) mafic volcanic rocks in the Naxiao-Shangshi area and Indosinian (i.e., early Mesozoic) large-scale silicic volcanic and intrusive rocks. During subduction in the Tethys oceans, a high heat flow has led to partial melting of the upper mantle. Subsequently, the upwelling of mantle-derived mafic magmas caused partial melting of the crust, which generated a series of intermediate-silicicmagmas in an active continental margin tectonic setting.

The Pingxiang region contains large areas of Triassic volcanic rocks and a granitic pluton that intruded Triassic strata in the Fuboshan area, along with scarce mafic dikes in the Permian and Triassic strata (Fig. 2B). Triassic sedimentary rocks and dacites in the Pingxiang area (Qin et al., 2011; Song et al., 2014; Li et al., 2019) formed in an island arc–extinct backarc basin system. In this study, we investigated granites of the Fuboshan pluton, which occur as a NE-SW–trending oval-shaped batholith with a length of 14 km and width of 3–8 km. The pluton is mainly exposed in Shangshi town and Banwang village of Pingxiang city, where porphyritic biotite granite and granite porphyry crop out. Rhyolites in this area crop out near the Fuboshan pluton and are mainly exposed in Madong village and Xiashi town of Pingxiang city.

Representative samples of igneous rocks were collected from the Pingxiang area (Figs. 3A, 3D, and 3G), including biotite granite and rhyolite. The petrography of the samples is described below.

The biotite granite (sample 19-px-4) is gray, medium grained (Figs. 3A and 3B), and contains 25–30 vol% plagioclase, 5–10 vol% K-feldspar, 35–40 vol% quartz, 3–5 vol% biotite, and accessory minerals (<5 vol%) (Fig.3C).

The rhyolites (samples 19-px-9 and -10) are gray and porphyritic (Figs. 3E and 3H). The phenocrysts are mainly quartz (20–30 vol%) and plagioclase (10–15 vol%), along with accessory minerals such as epidote (<3 vol%). The plagioclase is subhedral and exhibits polysynthetic twinning. The quartz grains are generally rounded and have embayed margins (Figs. 3F and 3I).

All analyses were carried out at the Guangxi Key Laboratory of Hidden Metallic Ore Deposit Exploration, Guilin University of Technology, Guilin, China (Liu et al., 2020).

Zircon crystals were extracted from rock samples by conventional crushing, heavy liquid, and magnetic separation techniques, and then handpicking. Cathodoluminescence (CL) images of the crystals were used to examine their internal structures and select sites for U-Pb dating. In situ U-Pb dating was undertaken by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS). Helium was used as a carrier gas to enhance aerosol transport to the ICP source and to minimize aerosol deposition. Argon was used as the makeup gas and was mixed with the carrier gas. Nitrogen was added into the central gas flow (Ar + He) of the Ar plasma to reduce the detection limits and improve the precision (Hu et al., 2008; Liu et al., 2010b). Each analysis comprised 15–20 s of background data acquisition and 50 s of data acquisition during ablation, with a laser spot size and depth of 40 μm and 20–40 μm, respectively. The zircon 91500 (weighted-mean ²⁰⁶Pb/²³⁸U age = 1062 ± 6 Ma) and GJ-1 (weighted-mean ²⁰⁶Pb/²³⁸U age = 600 ± 4 Ma) standards were used to monitor the data quality. A detailed description of the analytical instrumentation and procedures was given by Yuan et al. (2008). In-house software ICPMSDataCal (Liu et al., 2010a) was used for off-line data selection, integration of background and analytical signals, time-drift corrections, and quantitative calibration of the U-Pb dating. Isoplot software was used for plotting the concordia and weighted-mean age diagrams (Ludwig, 2003).

Fresh rock samples were collected, and weathered surfaces were avoided or removed. The samples were first crushed into small chips that were powdered in an alumina ceramic shatter box. After fusion with lithiummetaborate–lithium tetraborate flux and the oxidizing agent lithium nitrate, the products were poured into a platinum mold to generate a fused disk for major-element analysis by X-ray fluorescence spectrometry. Loss-on-ignition (LOI) values of each sample were measured after heating to 1000 °C. Trace-element analyses were undertaken with an Agilent 7900CX ICP-MS instrument after acid dissolution of the samples. About 12 mg of sample powder and standard samples were dissolved in a mixture of HNO3, HCl, and HF. The U.S. Geological Survey standards BHVO, AGV, W-2, and G-2 and Chinese national rock standards (GSR-1, -2, and -3) were analyzed to monitor the data quality. Pure elemental standards were used for external calibration. The sample solutions were placed in clean plastic centrifuge tubes and analyzed by ICP-MS (Liu et al., 2020). The analytical precision was better than ±5% for major elements and better than ±2%–5% for most trace elements.

Whole-rock powders for Nd isotope analysis were dissolved in Savillex Teflon capsules with 2 mL of 22 N HF and 1 mL of 8N HNO3 at 120 °C on a hot plate for 5–7 d. After drying the sample, 3 mL of HNO3 was added for 2 d to dissolve the samples. Neodymium was purified with HDEHP resin. Neodymium isotope ratios were measured with a Neptune Plus multi-collector (MC)-ICP-MS instrument. During the analysis period, theSm-Nd blanks were <100 pg. The isotope ratios were normalized to ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219 to correct for instrumental mass fractionation. During the analysis period, data quality was assessed by analyzing the international standard JNdi-1. The JNdi-1 Nd standard yielded ¹⁴³Nd/¹⁴⁴Nd = 0.512081 ± 0.000008 (n = 40; 2 SD). The U.S. Geological Survey reference material BHVO-2 was analyzed to monitor the data accuracy and yielded ¹⁴³Nd/¹⁴⁴Nd = 0.512965 ± 0.000004 (Liu et al., 2020).

Zircon Lu-Hf isotope analyses were carried out with a Neptune Plus MC-ICP-MS coupled to an ArF excimer LA system. Instrument settings and analytical procedures were described by Wu et al. (2006) and Geng et al. (2011). The analyses were obtained on the same spots or domains as used for U-Pb dating. The analytical conditions were as follows: laser beam diameter = 44 μm, laser repetition rate = 6 Hz, and laser ablation time = 30 s. The zircon standard GJ-1 yielded a weighted-mean ¹⁷⁶Hf/¹⁷⁷Hf ratio of 0.282333 (n = 93). The present-day chondritic ratios of ¹⁷⁶Hf/¹⁷⁷Hf = 0.282772 and ¹⁷⁶Lu/¹⁷⁷Hf = 0.0332 were used to calculate ε_Hf(t) values (Blichert-Toft and Albarède, 1997).

A biotite granite sample (sample 19-PX-4) and two rhyolite samples (19-PX-9 and 19-PX-10) were chosen for zircon U-Pb dating (Table 1). Zircon grains from the biotite granite and rhyolites are colorless or bright in color and transparent to translucent. The zircon grains are mostly euhedral and long-prismatic crystals. The zircon grains are 70–130 μm in size and exhibit well-developed oscillatory zoning in CL images (Fig. 4), indicating a magmatic origin (Hoskin and Schaltegger, 2003).

Twenty-eight analyses of 28 zircons from the biotite granite yielded a weighted-mean ²⁰⁶Pb/²³⁸U age of 249 ± 1 Ma (mean square of weighted deviates [MSWD] = 0.16) (Fig. 4A). Zircon grains from the biotite granite have U contents of 378–1758 ppm and Th contents of 185–1365 ppm, with Th/U ratios of 0.20–0.78. The euhedral crystal morphology and oscillatory zoning suggest that 249 ± 1 Ma represents the crystallization age of the granite.

Sixty-five zircon analyses of the rhyolites plot on the concordia (Figs. 4B and 4C). Thirty-five zircons from sample 19PX-9 contain 92–764 ppm Th and 563–2014 ppm U, with Th/U = 0.11–0.54. These values are typical of igneous zircons. The 35 analyses are concordant and yielded a weighted-mean ²⁰⁶Pb/²³⁸U age of 251 ± 1 Ma (Fig. 4B; MSWD = 0.25).

Thirty zircons from sample 19-PX-10 have high Th/U ratios (0.24–0.56), indicative of a magmatic origin. Most of these zircons are characterized by steep and heavy rare-earth element (REE)–enriched patterns with obvious negative Eu and positive Ce anomalies (Figs. 5B, 5D, and 5F), which are also indicative of a magmatic origin (Hoskin and Ireland, 2000; Hoskin and Schaltegger, 2003). Thirty analyses plot on the concordia (Fig.4C) and yielded a weighted-mean ²⁰⁶Pb/²³⁸U age of 250 ± 1 Ma (MSWD = 0.12) (Fig. 4C), which is consistent with the age of sample 19-PX-9 (Table 1; Fig. 4B). As such, the crystallization ages of the rhyolite are ca. 250 Ma.

Whole-rock major- and trace-element data for seven rhyolites and five biotite granites from the Pingxiang area along the southwestern SCB are presented in Table 2 and Figures 5 and 6. The rhyolites have relatively high contents of SiO₂ (71.4–74.2 wt%) and K₂O (4.1–5.5 wt%) and relatively low contents of TiO₂ (0.4–0.5 wt%), MgO (0.4–0.7 wt%), and CaO (0.3–1.6 wt%). The data are plotted in the rhyolite field in Figure 5A, and all samples belong to the high-K calc-alkaline series (Figs. 5B and 5C) and are strongly peraluminous, with A/CNK values (molar Al₂O₃/[CaO + Na₂O + K₂O]) of 1.05–1.36 (Fig. 5D). The rhyolites have low Mg^# values (20–28) and Na₂O/K₂O ratios of 0.34–0.51 (except for a sample with a value of 0.72). The rhyolites are light REE–enriched with (La/Yb)_N (_N indicates normalization to chondrite values) = 4.8–6.1 and have significant negative Eu anomalies (δEu = 0.33–0.49; Fig. 6A). The primitive-mantle–normalized multi-element patterns exhibit enrichments in Rb, Th, U, and Nd and depletions in Nb, Ta, Sr, and Ti (Fig. 6B).

The biotite granites plot in the granite field (Fig. 5A) and have SiO₂ = 73.6–74.8 wt%. They are high-K calc-alkaline rocks (Figs. 5B and 5C) and are weakly peraluminous with A/CNK = 1.08–1.22 (Fig. 5D). The biotite granites have high and variable total REE (ΣREE) concentrations of 256–330 ppm. In chondrite-normalized REE diagrams (Fig. 6C), the samples exhibit light REE enrichment relative to heavy REEs, with highly variable (La/Yb)_N (5.0–6.2), (La/Sm)_N (2.9–3.2), and (Gd/Yb)_N (1.4–1.6) ratios as well as moderately negative Eu anomalies (δEu = 0.42–0.46). The biotite granites have trace-element compositions similar to those ofmiddle–upper continental crust (Rudnick and Gao, 2003), with enrichments in Rb, Th, and U, and depletions in Ba, Sr, Nb, Ta, P, and Ti (Fig. 6D).

Whole-rock Nd isotope data are listed in Table 3. The biotite granites have enriched Nd isotopic compositions, with ε_Nd(t) and T_DM2(Nd) values of –10.6 to –10.4 and ca. 1.9 Ga, respectively. The rhyolites have similar Nd isotopic compositions as the biotite granites, with ε_Nd(t) and T_DM2(Nd) values of –11.5 to –9.7 and 1.8–2.0 Ga, respectively.

Zircon Hf isotope data are listed in Table 4. For the calculation of the ε_Hf(t) and T_DM2(Hf) values, the in situ zircon U-Pb ages were used. The 253–249 Ma rhyolites have enriched zircon Hf isotopic compositions, with ε_Hf(t) and T_DM2(Hf) values of – 12.9 to –6.2 and 1.6–2.9 Ga, respectively. The 250–247 Ma biotite granites have similar zircon Hf isotopic compositions as the rhyolites, with variable ε_Hf(t) (−14.5 to –6.7) and T_DM2(Hf) values (1.7–2.2 Ga).

All the studied samples are fresh, and thin-section observations show that the samples are little affected by alteration. The samples have low LOI values (0.67–1.65 wt%, except for one rhyolite with a value of 3.09 wt%).

Based on the nature of their protolith and the pressure-temperature of melting, silicic igneous rocks are generally subdivided into A-, I-, S-, and M-types (Bonin, 2007). Several geochemical approaches have been proposed to discriminate between A-type and other types of granites (Whalen et al., 1987; Eby, 1992), including high Na₂O + K₂O contents, and Ga/Al and molar Fe/Mg ratios, low CaO contents, high-field-strengthelement (HFSE) enrichments, and Sr and Eu depletions (Whalen et al., 1987; Bonin, 2007). On Zr or Nb versus 10,000 × Ga/Al diagrams, the studied rhyo-lites and granites plot in the field for A-type granites (Figs. 7A and 7B; Whalen et al., 1987). Although highly fractionated I- and S-type granites can both have high FeO^T/MgO ratios, the rhyolites and biotite granites plot in the A-type field on a FeO^T/MgO versus (Zr + Nb + Ce + Y) diagram (Fig. 7C), which is generally used to distinguish A-type from highly differentiated I-type granites. It is generally accepted that A-type granites are derived from a relatively high-temperature magma chamber (>800°C) that is usually at shallow crustal levels (Eby, 1990; King et al., 2001; Bonin, 2007). The calculated zircon saturation temperatures of the biotite granites and rhyolites are 835–870 °C (average = 860 °C), which further indicate their A-type nature. The A-type granites can be further subdivided into A₁- and A₂-type granites (Eby, 1992). On Rb/Nb versus Y/Nb, Nb-Y-Ce, and Nb-Y-3 × Ga diagrams (Figs. 7D–7F), all the studied rhyolites and biotite granites plot in the A₂-type field.

The ages of the studied rhyolites (251–250 Ma) and biotite granites (249 Ma) are the same within error, and these rocks have similar geochemical characteristics and isotopic compositions, indicating they were derived from a common magma source. In Harker diagrams (Fig. 8), Al₂O₃, P₂O₅, TiO₂, and Fe₂O₃^T contents exhibit negative correlations with SiO₂, suggestive of fractionation of K-feldspar, plagioclase, biotite, apatite, and Fe-Tioxides. On the basis of the available data, the biotite granites have a petrogenetic link with the coexisting rhyolites, and the biotite granites are more fractionated than the rhyolites (Fig. 8). Therefore, we propose that the biotite granites and rhyolites were cogenetic and formed by the fractional crystallization of the same parental magma, representing crystal cumulates and residual melts, respectively.

In general, A-type granites can be generated by three mechanisms: (1) direct fractionation of mantle-derived basaltic magmas (Turner et al., 1992; Mushkin et al., 2003), (2) hybridization of crust-derived silicic and mantle-derived mafic magmas (Kemp et al., 2005; Yang et al., 2006), and (3) partial melting of a variety of crustal materials at high temperatures and low pressures (Landenberger and Collins, 1996; Dall‘Agnol and de Oliveira, 2007). If the studied A₂-type rhyolites and biotite granites were generated by extreme differentiation of mantle-derived basaltic magmas, the volume of mafic magma intruded into the upper crust should be an order of magnitude greater than that of the rhyolite and granite (Turner et al., 1992; Frost et al., 2001). However, the early Triassic volcanic and intrusive rocks in the Pingxiang area are dominated by rhyolitic and granitic rocks (Fig.2B). The rhyolites and biotite granites are characterized by negative Nb and Ta anomalies (Figs. 6B and 6D), high SiO₂ contents, and low MgO, FeO^T, and MnO contents, which indicate they formed by partial melting of crustal rocks. In addition, the elemental ratios of these rhyolites and biotite granites are more consistent with a crustal than mantle source. For example, the Ba/Rb ratios (3.3–7.5; average = 4.3) are close to the crustal ratio (Ba/Rb = 6.7) of Rudnick and Fountain (1995) and are significantly lower than the mantle ratio (Ba/Rb = 11; Hofmann and White, 1983). Moreover, the studied A₂-type rhyolites and biotite granites do not contain mafic microgranular enclaves and have negative ε_Nd(t) (−11.5 to −9.7; T_MD = 1.8–2.0 Ga) and zircon ε_Hf(t) values (–14.5 to –6.2; T_MD = 1.6–2.9 Ga) (Figs. 9 and 10), as well as high SiO₂ contents and low Mg^# values (Fig. 11A). These features indicate that the rhyolites and biotite granites were derived by partial melting of crustal materials rather than by differentiation of mantle-derived basaltic magmas or magma mixing between crust- and mantle-derived magmas. This inference is further supported by the low Cr (1.6–7.4 ppm) and Ni (0.2–6.0 ppm) contents of the studied rocks.

Several sources have been proposed to explain the formation of A-type granites with a crustal origin, including dry residual granulitic lower crust (Collins et al., 1982; Clemens et al., 1986), granulitic metasedimentary rocks (Huang et al., 2011), dehydrated charnockitic middle–lower crust (Zhao et al., 2008; Zhang et al., 2015), and hornblende- and biotite-bearing granitoids (Patiño Douce, 1997). Collins et al. (1982) and Clemens et al. (1986) proposed that A-type granites were derived by partial melting of dry residual granulitic lower crust from which a hydrous felsic melt phase had been extracted. However, Creaser et al. (1991) showed that a residual granulitic source is unlikely to generate A-type granitic melts, based on the major-element compositions of the likely partial melts and proposed source rocks. The studied A₂-type rhyolites and biotite granites are calc-alkaline with (K₂O + Na₂O)/Al₂O₃ ratios of 0.54–0.62, which are inconsistent with a metasedimentary source, because partial melting of crustal metasedimentary rocks typically produces low-melt alkali contents (Karsli et al., 2018). As such, dry residual granulitic lower-crustal and granulitic metasedimentary sources can both be precluded. In addition, the studied rhyolites and biotite granites have negative whole-rock ε_Nd(t) and zirconε_Hf(t) values and old T_DM2 ages, indicating the source was not juvenile middle–lower crust (Figs. 9A and 10).

A-type granites typically represent anhydrous high-temperature magmas (Clemens et al., 1986; Eby, 1992; Dall‘Agnol et al., 2012; Wu et al., 2022), and their origins involve partial melting of crustal igneous rocks of tonalitic to granodioritic composition (Creaser et al., 1991). Based on melting experiments of tonalite and granodiorite at 950 ° melting of hornblende- and biotite-bearing granitoids in the upper continental crust at low pressures (P ≤ 4 kbar; depths ≤ 15 km) have geochemical characteristics similar to A₂-type granites (Figs. 11B and 11C). Moreover, plagioclase and orthopyroxene are the products of shallow (P ≤ 4 kbar) dehydration melting of hornblende-bearing granitoids and become the dominant residual phases. Crystallization of plagioclase and orthopyroxene decreases the CaO and MgO contents of melts with decreasing pressure and explains the low CaO and MgO contents of A-type low CaO (0.28–1.61 wt%) and MgO (0.40–0.84 wt%) contents and large negative Eu and Sr anomalies (Figs. 6B and 6D), which are consistent with a residual low-pressure mineral assemblage of plagioclase + orthopyroxene. Crystallization of plagioclase can also produce melts that are enriched in Ge relative to Al with high Ga/Al ratios (10,000 × Ga/Al = 2.9−4.1). The ε_Nd(t) values and (Th/Nb)_N ratios of our samples are similar to those of granites formed by partial melting of felsic crustal rocks but different from those of granites and rhyolites formed by partial melting of charnockites or juvenile middle to lower crust (Fig. 9B). Furthermore, the magma underwent variable degrees of fractional crystallization and finally formed the biotite granite and rhyolite melt. Therefore, we propose that the A₂-type rhyolites and biotite granites in the Pingxiang area might have been generated by partial melting of felsic crustal rocks at low pressures of ~4 kbar (Figs. 11B and 11C) and high temperatures of ~860 °C (Fig. 11D), and then the magma underwent fractional crystallization that formed the biotite granite and equivalent rhyolite melt.

The tectonic setting of the studied rhyolites and granites can be evaluated with a series of tectonic discrimination diagrams (Fig. 12). On the Rb/30-Hf-Ta × 3 and Nb versus SiO₂ tectonic classification diagrams (Figs. 12A and 12B), all the samples plot in the volcanic arc granite field. On Al₂O₃/SiO₂ versus Fe₂O₃ + MgO and Th/Ta versus Yb diagrams (Figs. 12C and 12D), the studied rocks plot in the active continental margin igneous rock field. Zircon trace-element contents can be used to constrain the tectonic setting of the studied rhyolitic and biotite granitic magmas (Yang et al., 2012). Most zircon grains in the rhyolites and biotite granites haveNb/Hf, Th/U, Th/Nb, Hf/Th, Nb/Yb, and U/Yb ratios similar to those of zircons from arc-related settings and plot in the arc-related (i.e., orogenic) and continental arc fields in plots of Nb/Hf versus Th/U, Hf/Th versusTh/Nb (Figs. 13A and 13B), and log₁₀(U/Yb) versus log₁₀(Nb/Yb) (Fig. 13C). Thus, these zircon grains have a trace-element geochemistry similar to that of zircons formed in an arc setting (Peng et al., 2008; Yang et al., 2012; Grimes et al., 2015).

It has been suggested that the Permian–Triassic igneous rocks in the southwestern SCB are related to either Paleo-Tethys subduction and closure (Jiao et al., 2015; Li et al., 2016; Xu et al., 2018; Wang et al., 2020; Gan et al., 2021) or Paleo-Pacific subduction (Li et al., 2006, 2012b; Li and Li, 2007). The Paleozoic–Mesozoic ophiolites, high-pressure and low-temperature metamorphic rocks, and arc igneous rocks in the Ailaoshan–SongMa suture zone are indicative of subduction in the Paleo-Tethys Ocean (Xu et al., 2007; Zi et al., 2012b; Hou et al., 2017; Liu et al., 2018). The late Permian to early Triassic tholeiites in the Jianshui, Napo, and Pingxiang areas of the southwestern Youjiang Basin, as well as the brachiopods and other fossils in siliceous shales and chert interbeds, suggest that early Mesozoic magmatism in the Pingxiang region was related to subduction and closure of the eastern Paleo-Tethys Ocean (Dong and Zhu, 1999; Wu et al., 2000, 2002). The Youjiang Basin is close to the Ailaoshan–Song Ma suture zone between the SCB and Indochina Block (Figs. 1 and 2A) but was far from the Pacific Ocean trench during the Triassic (Gan et al., 2021). The late Paleozoic to Triassic detrital zircons and volcanic lithic fragments in the Youjiang Basin indicate a detrital input from an Indosinian orogenic belt (Yang et al., 2012; Hu et al., 2015a, 2015b); this input was triggered by subduction-collision in the Paleo-Tethys Ocean (Cai et al., 2014). Based on the distribution of known paleo-subduction zones in the surrounding regions and the regional sedimentary and igneous rock associations, we conclude that formation of rhyolites and biotite granites in the Pingxiang region was linked to Paleo-Tethys rather than Paleo-Pacificsubduction.

Many authors agree that the tectonic evolution of the Paleo-Tethys Ocean included subduction of an oceanic plate followed by continental collision, crustal thickening, and extensional tectonics (Qin et al., 2013; Yang et al., 2014; Li et al., 2019). The Permian–Triassic Paleo-Tethys evolution can be constrained from magmatism of this age. Late Permian–early Triassic arc granites and volcanic rocks are widespread in southeastern Guangxi and Hainan provinces and are considered to have formed due to northward subduction in the eastern Paleo-Tethys (Jiao et al., 2015; Li et al., 2016; Xu et al., 2018; Wang et al., 2020). The middle–late Triassic volcanic rocks from the nearby Sanjiang orogenic belt record the transition from a continental arc to a syn-collisional volcanic setting (Fan et al., 2014; Zhang et al., 2017; Yang et al., 2020). This was followed by late Triassic mafic and silicic volcanism, with little or no intermediate volcanism. This bimodal volcanism is typical of an extensional tectonic setting (Brewer et al., 2004; Chen et al., 2021). The composite Lincang batholith records the evolution of the Paleo-Tethys Ocean, including subduction in the late Permian–early Triassic, followed by the syn-collisional stage in the middle Triassic and post-collisional stage in the late Triassic (Peng et al., 2013; Deng et al., 2018). Chen et al. (2002, 2008) proposed that Jurassic granites, mafic igneous rocks, and bimodal volcanic rocks in the Nanling region of South China were the products of post-collisional extension of the Indosinian orogeny related to large-scale Tethys tectonism. As such, in the Permian–Triassic Paleo-Tethys Ocean, oceanic subduction occurred in the late Permian–early Triassic, the transition from subduction to syn-collision occurred during the middle–late Triassic, and post-collision and crustal extension occurred in or after the late Triassic.

The studied A₂-type rhyolites yielded weighted-mean ²⁰⁶Pb/²³⁸U ages of 250 ± 1 Ma, revealing that rhyolite eruption occurred in the early Triassic. The new zircon U-Pb age data for the A₂-type biotite granites from the Fuboshan pluton show they were emplaced at 249 ± 1 Ma in the early Triassic. Therefore, the rhyolites and biotite granites in the Pingxiang area were coeval and formed during subduction in the Paleo-Tethys Ocean but before its final closure during the middle Triassic.

The Pingxiang area is located at the southwestern margin of the Youjiang Basin (Fig. 1). Qiu et al. (2017) suggested that the Youjiang Basin was in an arc-related setting in the early Mesozoic, based on a study of clastic sedimentary rocks, whereas continent-continent collisional and post-collisional extensional settings did not exist based on the following geological evidence: (1) from the middle Devonian to early Triassic, complex platform and deep-sea sedimentation occurred in the Youjiang Basin (Cai and Zhang, 2009); (2) crustal deformation in the Youjiang Basin formed a thin-skinned, fold-and-thrust system during collision between the SCB and Indochina Block and the coeval Triassic oceanic subduction (Wang et al., 2020); (3) a detailed investigation of the Lopingian (upper Permian) to middle Triassic sequences of the Youjiang Basin constrained the patterns of basin fill, differential tectonic subsidence, and very high subsidence rates along regional faults, which are indicative of an extensional or transtensional continental backarc setting prior to the middle Triassic (Duan et al., 2020); and (4) at the southern margin of the Youjiang Basin, basalts and andesites formed from 256 Ma to 241 Ma exhibit subduction-related arc-like features (Qin et al., 2011, 2012; Li et al., 2019; Wang et al., 2020). These lines of evidence demonstrate that the Pingxiang region was in a subduction rather than a continent-continent collisional or post-collisional extensional setting during the early Triassic.

With regards to the subduction polarity, several studies have suggested that the Ailaoshan–Song Ma suture records westward subduction beneath the Indochina Block during the Permian–Triassic (Wei and Shen, 1997; Liu et al., 2011, 2012; Halpin et al., 2016). However, east-dipping subduction beneath the SCB has also been proposed, based on Permian–Triassic sedimentary and paleontological evidence (e.g., the radiolarian assemblage in Qinfang Basin) in the southwest SCB (Hou et al., 2017; Ke et al., 2018; Xu et al., 2019) and arc-type igneous rocks (Fan et al., 2010; Qin et al., 2011, 2012; Wang et al., 2013; Zhang et al., 2014; Wu et al., 2015). For example, the middle Triassic silicic (e.g., dacite and rhyolite) and mafic (e.g., basalt) volcanic rocks in southwest Guangxi Province are geochemically similar to typical subduction-related arc volcanic rocks (Qin et al., 2011, 2012). Regarding the northern Song Ma suture zone, Lepvrier et al. (2004) and Zhang et al. (2014) proposed that eastwards (based on the present-day position but originally northwards) oblique subduction occurred in the Paleo-Tethys beneath the SCB during 254–240 Ma, based on structural and geochronological studies of metamorphic rocks in the Indochina Block. Previous studies of early–middle Triassic volcanic (basaltic andesite)-sedimentary (carbonate-rich conglomerate, pebbly coarse sandstone, and calcareous sandstone) rock assemblages in the Napo-Funing area around Youjiang Basin indicate that the basin was in an active continental margin setting during the Permian–Triassic (Xiang et al., 2021). Therefore, we prefer the hypothesis whereby the middle Triassic Paleo-Tethys Ocean was experiencing bidirectional subduction to both the west and east beneath the Indochina Block and SCB, respectively (Xu et al., 2020). This subduction model can explain the early Triassic A₂-type granitic and rhyolitic magmatism in the Pingxiang area. The studied samples havearc-like geochemical features similar to those of A₂-type rhyolites (Eby, 1992; Eby et al., 1995) and I-type granites that form in extensional settings at convergent margins (e.g., the Simao terrane, Wusu arc, and Youjiang Basin; Figs. 5 and 6; He et al., 2018; Yang et al., 2018; Gan et al., 2021), and thus we suggest they formed in a similar extensional setting.

Data from previous studies and our new data for rhyolites and biotite granites in the Pingxiang area suggest the studied rocks represent magmas generated by subduction in the Paleo-Tethys Ocean along the Ailaoshan–SongMa suture zone prior to its final closure, which was associated with slab roll-back during subduction in the Paleo-Tethys Ocean. Slab roll-back tends to occur when relatively old oceanic crust is being subducted (Niu, 2014), and this may induce upwelling of asthenosphere to compensate for the volume loss in the mantle wedge (Uyeda and Kanamori, 1979). This upwelling results in trenchward migration of arc magmatism and lithospheric extension and crustal melting in the rear-arc regions (Nakakuki and Mura, 2013; Liu et al., 2017; Collins et al., 2019). For example, the A-type granitoids in the Lachlan orogen, southeastern Australia, have been linked to extension associated with episodic slab roll-back (Collins et al., 2019). Arc magmatism at the southwestern margin of the Youjiang Basin during the early Triassic migrated from southeast to southwest, and arc magmatism occurred mainly from 265 Ma to 208 Ma (Qin et al., 2011; Gao et al., 2017; Li et al., 2019; Wei et al., 2020); the arc magmatism exhibits a general younging trend to the southwest. This trend is consistent with roll-back of the subducting Paleo-Tethys oceanic slab. Therefore, a slab roll-back model is proposed to explain the magmatism in the Pingxiang area (Fig. 14). During the early Triassic, the Paleo-Tethys oceanic slab was subducted beneath the SCB, and slab roll-back caused asthenospheric upwelling, which led to the crust and mantle beneath the region becoming progressively thinner. The asthenospheric upwelling heated the subcontinental lithospheric mantle, triggered the partial melting of felsic crustal rocks, and generated the A₂-type rhyolitic and biotite granitic melts that were erupted and intruded in the Youjiang extensional setting.

Zircon U-Pb age and Hf isotope and whole-rock geochemical and Nd isotope data for newly discovered A₂-type rhyolites and biotite granites in the Pingxiang area of western Guangxi Province, southwest China, show that:

1. The A₂-type rhyolites and biotite granites formed in the early Triassic (ca. 251–249 Ma).

2. The rhyolites and biotite granites formed by partial melting of felsic crustal rocks at low pressures and high temperatures.

3. The rhyolites and biotite granites likely formed in an extensional setting induced by slab roll-back during subduction of eastern Paleo-Tethys oceanic lithosphere during the early Triassic.

We thank the GSA editor for handling and three anonymous reviewers for constructive criticism and suggestions that significantly improved this paper. This research was supported by the National Natural Science Foundation of China (grants 92055208 and 42203051), the Guangxi Natural Science Foundation of China (grant 2022GXNSFBA035538), the Guangxi Science Innovation Base Construction Foundation (grant GuikeZY21195031), the Guangxi Natural Science Foundation of China for Young Scholars (grant 2022GXNSFBA035538), and the Fifth Bagui Scholar Innovation Project of Guangxi Province (to Xu Jifeng). This is a contribution to the Guangxi Key Mineral Resources Deep Exploration Talent Highland.