Diachronous subduction of the Proto-Tethys Ocean along the northern margin of East Gondwana: Insights from SHRIMP and LA-ICP-MS zircon geochronology in the West Kunlun Orogenic Belt, Northwestern China

ABSTRACT The geological evolution of the Proto-Tethys Ocean remains vague. The Tianshuihai Terrane (TSHT), a subterrane of the West Kunlun, distributed to the south of the Proto-Tethys Ocean during the Early Palaeozoic, records abundant information on the geological evolution of the Proto-Tethys Ocean. In this study, we reported SHRIMP and LA-ICP-MS zircon U–Pb ages, whole-rock major and trace element composition data, and Sr-Nd-Hf isotopes of a suite of monzogranites and the monzonitic-syenitic enclaves in Zankan and syenogranites in Laobing regions of the TSHT, West Kunlun Orogenic Belt. The syenogranites in the Laobing region yielded SHRIMP zircon U-Pb ages of 550.4 ± 6.4 Ma to 547.5 ± 5.3 Ma, which were the earliest age records of subduction-related magmatism in the TSHT during the Late Neoproterozoic-Early Palaeozoic. The host monzogranites in the Zankan area yielded SHRIMP and LA-ICP-MS zircon U-Pb ages of 542.6 ± 8.4 Ma to 540.5 ± 2.8 Ma, which is coeval with the monzonitic-syenitic enclaves ages of 533.8 ± 3.4 Ma to 534.7 ± 3.0 Ma. We speculated that an active margin developed along the TSHT during the Cambrian and the initial subduction of the Proto-Tethys oceanic slab must have occurred prior to the Early Cambrian (>550 Ma). The TSHT and the Southern Kunlun Terrane were distributed between the northern margin of the East Gondwana continent and the Tarim Block. Additionally, the coexistence of two branches of the Proto-Tethys Ocean represented by the Kangxiwa Fault and Kudi Ophiolite Belt during the Early Palaeozoic. Based on the chronological statistics of micro-continental blocks in the northern margin of the East Gondwana continent, subduction of the Proto-Tethys Ocean could be diachronous, initially originating in the northwestern part of the East Gondwana continent, and gradually propagating to the east of the East Gondwana continent. Graphical Abstract


Introduction
The Proto-Tethys Ocean was the predecessor of the Palaeo-Tethys Ocean, located between the East Asian blocks (e.g. Tarim, Alax, and North China) and Australian and Indian blocks (Mattern and Schneider 2000;ZZhu et al. 2012;Zhao et al. 2016;Li et al. 2018;Zhao et al. 2018;Zhang et al. 2019b). It resulted directly from the break-up of the Rodinia supercontinent at ~750 Ma and finally closed during the Early Palaeozoic (500-420 Ma) (Mattern and Schneider 2000;Zhao et al. 2016Zhao et al. , 2018Li et al. 2018;Zhang et al. 2019b;Wang et al. 2020). Multiple small, complex, and dispersed micro-continental blocks (North Qinling, Central Altyn, Central Qilian, West Kunlun, East Kunlun, etc.) were distributed to the south of the Proto-Tethys Ocean during the Late Neoproterozoic-Early Palaeozoic . These micro-continental blocks have often been neglected by previous reconstructions of the Proto-Tethys Ocean evolution (Meert 2003;Metcalfe 2009). However, these Precambrian micro-continental blocks have witnessed multiple tecto-magmatic events, which reflect the history of the Proto-Tethys Ocean from origin to extinction. A detailed study of tecto-magmatic events in these micro-blocks can not only reveal the properties of these blocks but also shed light on the evolution of the Proto-Tethys Oceans Zhao et al. 2018).
The West Kunlun Orogenic Belt (WKOB) (Figure 1a, b), a major tectonic unit of Central China Orogenic Belt (Figure 1c, d), occupies a fundamental position along the tectonic junction between the Asian and Tethyan domains (Matte et al. 1996;Mattern et al. 1996). The WKOB is considered as important for understanding the Proto-Tethys evolution (Matte et al. 1996;Mattern et al. 1996;Wang 2004;Xiao et al. 2005;ZZhu et al. 2016;Liu et al. 2019;Zhang et al. 2018aZhang et al. , b, 2019aZhang et al. , 2019bYin et al. 2020) during the Early Palaeozoic due to development of the Late Cambrian Kudi ophiolite, which was considered as the residuals of the Proto-Tethys Ocean (Wang et al. 2001(Wang et al. , 2002Xiao et al. 2003b), and the Middle Cambrian to Early Devonian granites which are ascribed to the subduction of the Proto-Tethys Ocean (Xiao et al. 2003b;Yin et al. 2020). However, details regarding the evolution of the Proto-Tethys Ocean in the WKOB and peripheral areas, such as the time of initial subduction of the Proto-Tethys oceanic slab (Zhang et al. 2018b;Liu et al. 2019;Yin et al. 2020), the subduction polarity (northward, southward, or bidirectional) (Yin and Harrison 2000;Yuan et al. 2003;Liu, Jiang, Jia et al. 2014;Zhang et al. 2018bZhang et al. , 2020Yin et al. 2020), and the Early Palaeozoic spatio-temporal configurations of different subterranes of the WKOB are unavailable (Mattern and Schneider 2000;Wang 2004;Xiao et al. 2005;Zhang et al. 2018a).
Recently, Early Cambrian gabbros and granitoids were identified in the WKOB; and these magmatic rocks were sparse, but are important to investigate the data unavailability issues mentioned earlier (Yuan et al. 2003;Liu, Jiang, Jia al. 2014;Zhang et al. 2016Zhang et al. , 2018bZhang et al. , 2020Yin et al. 2020). Meanwhile, the Late Ediacaran-Early Cambrian was an important period for the final assembly of the Gondwana supercontinent and initial subduction along the Peri-Gondwana margin (Meert 2003;Cawood and Buchan 2007;Murphy et al. 2011;Ustaömer et al. 2012;ZZhu et al. 2012ZZhu et al. , 2013Hu et al. 2018;Sajid et al. 2018). Thus, exploration of the Late Ediacaran-Early Palaeozoic magmatic rocks distributed in the WKOB may shed light on the ocean-continent configurations between the microcontinental blocks along the Peri-Gondwana margin. Furthermore, it will provide insights to trace the evolution of the Proto-Tethys Ocean and the Gondwana supercontinent.
In this study, we reported zircon SHRIMP and LA-ICP-MS U-Pb ages, whole-rock major and trace elements, and Sr-Nd-Hf isotopic data of the Late Ediacaran-Early Palaeozoic granitoid and enclaves from the TSHT in the WKOB. The findings of our study supplemented by the results from the previous studies can develop strong constraints regarding the issues mentioned.

Geological setting
The WKOB, a tecto-magmatic terrane in the northwest region of the Tibetan Plateau, borders the Tarim Basin to the north, offsets from the East Kunlun Orogenic Belt and Songpan-Ganzi Terrane by the Altyn Fault to the east, and is adjacent to the Pamir Plateau to the west (Mattern and Schneider 2000;Yin and Harrison 2000) ( Figure 1a). Tectonically, the WKOB is subdivided from north to south into three main units, including the Northern Kunlun Terrane (NKT, Figure 1b (Mattern and Schneider 2000;Xiao et al. 2005).
The SKT was considered to have a common Precambrian basement (Wang 2004;Wang et al. 2020). The Precambrian rocks include the Kulangnagu Group, Saitula Group, and the Sangzhutage GrouThe main lithologies of the Kulangnagu Group are greenschistamphibolite facies and marble (Xu 2017). A zircon LA-ICP-MS age of 2025 ± 13 Ma was acquired from the basalts of this rock group . The Saitula Group mainly consists of greenschist facies and gneiss (Xu 2017), which was primarily intruded by granites of 815 ± 57 Ma (Zhang et al. 2003). The Sangzhutage Group is composed primarily of schist, marble, and metasandstone intercalated with nearby amphibolite layers. Mesoproterozoic-Neoproterozoic stromatolites, such as Jacutophyton, Conophyton, Baicalia, and Paraconophyton are also found in the Sangzhutage Group . The SKT was considered to be an integral part of the Tarim Block (Mattern et al. 1996) or an exotic block aggregated to the NKT during the Early Palaeozoic (Yuan et al. 2002;Xiao et al. 2003b). In contrast, Zhang et al. (2019a) argued that the basement of the SKT, locally known as the Saitula Group, represented a massive Early Palaeozoic accretionary wedge.
The TSHT is located between the Mazha-Kangxiwa Fault to the north and the Qianertianshan-Hongshanhu Fault to the south, forming an NW-SE trending belt (Zhang et al. 2019a). The nature of the TSHT is under controversy. Previously, it was considered as a huge accretionary wedge associated with a Late Palaeozoic-Early Mesozoic orogenic process (Xiao et al. 2003a. However, recent studies have reported Palaeoproterozoic metavolcanic rocks and confirmed the existence of the Precambrian basement of the TSHT (Ji et al. 2011). The Precambrian members in TSHT are termed as the Bulunkuole Group (primarily composed of Precambrian gneisses, schists, and marble) (Ji et al. 2011) and Tianshuihai Group (composed majorly of meta-greywackes and limestone) (Zhang et al. 2019a). Late Palaeozoic to Early Mesozoic accretionary wedge overlies the Precambrian basement in the fault unconformably e.g. the widespread Silurian Wenquangou Group, which consists mainly of quartz sandstone, marble, and quartzite.
The Early Palaeozoic magmatic rocks developed in the WKOB were considered to be associated with the evolution of the Proto-Tethys Ocean (Yuan et al. 2002(Yuan et al. , 2003 . ; Liu, Jiang, Jia et al. 2014;Zhu et al. 2016;Yin et al. 2020;Zhang et al. 2020). Previous studies suggest that the Early Palaeozoic magmatic rocks in the WKOB were distributed mainly in the SKT (including the 510-450 Ma I-type granites and the 430-400 Ma high Ba-Sr granites) (Yuan et al. 2003;Ye et al. 2008;Yin et al. 2020) and subordinately in the NKT. Recently, multiple studies revealed that the Early Palaeozoic magmatic rocks, such as the Kelule, Nanpingxueshan, and Ayilixi plutons are ubiquitously distributed in the TSHT ZZhu et al. 2016;Liu et al. 2019;Yin et al. 2020;Zhang et al. 2020).

Sample description
In this study, we focused on the Early Palaeozoic granitic intrusions in the TSHT to reveal the Early Palaeozoic evolution of the WKOB. Igneous samples, including granitoid host rocks and mafic-dioritic enclaves in the TSHT, were collected from the Zankan and the Laobing regions ( Figure 2).

SHRIMP zircon U-Pb dating
Zircon grains from samples of 13LB05, 13LB06, 13ZK06, and 13ZK08 were subjected to image and SHRIMP U-Pb Academy of Geological Sciences. Photomicrographs of the zircon grains in transmitted and reflected light, as along with cathodoluminescent (CL) scanning electron microscope images were developed to determine the internal microstructures of the grains and to estimate the best locations for U-Pb analyses. Zircon U-Pb dating was conducted using SHRIMP II, with the analytical procedures and conditions similar to those described by (Williams 1998). The 206 Pb/ 238 U ages are highly reliable for concordant Phanerozoic zircon analyses (Compston et al. 1992). Standard zircons TEMORA 1 (age = 417 Black et al. 2003) were used for calibration of the 206 Pb/ 238 U ratio. Common Pb correction was based on the measured 204 Pb abundances. Data processing was conducted using Squid and Isoplot programs (Ludwig 2003). Uncertainties indicated in Table S1 were quoted at ± 1σ, whereas the weighted mean ages were quoted at ±2σ, at a 95% confidence level.

LA-ICP-MS zircon U-Pb dating
Zircon U-Pb dating for samples 15ZK04, 15ZK10, 15ZK11, and 15ZK13 was conducted synchronously by LA-ICP-MS at the Key Laboratory for the study of focused Magmatism and Giant ore Deposits, MNR, Xi'an Center of Geological Survey, China Geological Survey. Laser sampling was performed using a GeoLas Pro. An Agilent 7700x ICP-MS instrument was used to acquire ion signal intensities. Each analysis incorporated a background acquisition of approximately10 s (gas blank), which was followed by 40 s data acquisition from the sample. Agilent Chemstation was utilized for individual analysis. Zircon 91,500 was used as an external standard for U-Pb dating. Off-line selection and integration of background and analyte signals, time-drift correction, quantitative calibration for trace element analyses, and U-Pb dating were performed by Glitter 4.4. Concordia diagrams and weighted mean calculations were carried out using Isoplot 3.0 (Ludwig 2003). Trace element compositions of zircon were calibrated against reference materials (NIST610) combined with Si for internal standardization. The expected values of element concentrations for the NIST reference glasses were extracted from the GeoReM database. Details of the instrument conditions and data acquisition procedures were similar to those described by Li et al. (2015).

Whole-rock major and trace elements
For the whole-rock composition analysis of host rocks and enclaves, weathered rims were removed before grinding the samples. Major element compositions were determined using an X-ray fluorescence spectrometer (XRF) after fusing the sample powder using an Axios mAX (4 kW) instrument at the Xi'an Center of Geological Survey, China Geological Survey. The precision and accuracy of XRF analysis using the reference GBW07103 were evaluated to be less than ± 1%. Trace element compositions were determined after digesting the samples in an HF-HNO 3 mixture in steel-jacketed Teflon 'bombs' at 190°C for 48 h followed by ICP-MS analysis using SX-II ICP-MS. The accuracy of the analysis, which was monitored using references of BHVO-2 and AGV-2 from the United States Geological Survey (USGS), was <5%.

Whole-rock Sr-Nd-Hf Isotopic compositions
The Sr-Nd-Hf isotope analyses were performed on Neptune Plus MC-ICP-MS (Thermo Fisher Scientific) at the Wuhan Sample Solution Analytical Technology Co., Ltd., Wuhan, China. The Sr and Nd isotopic compositions were determined in static mode on the Nu Plasma MC-ICP-MS. Sr and Nd (and other rare earth elements) were separated/concentrated using standard chromatographic columns with AG50W-X8 and HDEHP resins. The measured 143 Nd/ 144 Nd and 87 Sr/ 86 Sr ratios were normalized to 0.7219 and 0.1194, respectively. External reproducibility of the isotopic measurements was estimated by repeated analyses according to the international standards.
The Hf isotopic analyses were done in static mode on the Nu Plasma MC-ICP-MS. The newly designed X skimmer cone and jet sample cone of Neptune Plus were used to improve the signal intensity of Hf, Yb and Lu, and for adding small amounts of nitrogen to the central gas flow (Hu et al. 2008(Hu et al. , 2012. A chromatographic column composed of LN-Spec resins was used for Hf separation and purification following the protocol given by Yuan et al. (2007). JMC-475 was used as a standard to calibrate isotope fractionation, which was analysed twice for every five analyses. Values reported for the samples were adjusted to the accepted value (0.282160) for reference Hf standard JMC-475 (Vervoort and Blichert-Toft 1999).
Nearly all the zircon grains from the samples possessed an inherited core ( Fig S2); these inherited cores were carefully avoided during analyses. The analysed zircon spots had varying U (32-608 ppm) and Th (79--1089 ppm) concentrations with the Th/U ratios ranging from 0.32 to 1.71 (Tables S1 and S2), which were consistent with the magmatic crystallization of zircons (Hoskin and Schaltegger 2003). Thus, the estimated ages were considered to represent the time of zircon crystallization, which in turn represented the time of magma emplacement.

Whole-rock major and trace element compositions
The whole-rock compositions for magmatic rocks from the Laobing and Zankan regions are listed in Table S3. A total of 14 samples, including nine monzogranites and five monzonitic-syenitic enclave samples, were collected from the Zankan region. The range of SiO 2 , K 2 O, Na 2 O, and CaO content of the monzogranite host rocks varied from 70.0 to 73.0 wt %, 0.2 to 1.7 wt %, 6.2 to 7.7 wt %, and 0.9 to 2.6 wt %, respectively (Table S3).
Monzogranite samples appeared in the metaluminous field in the A/CNK-A/NK (molar Al 2 O 3 /(CaO + Na 2 O + K 2 O)-Al 2 O 3 /(Na 2 O + K 2 O)) diagram (Figure 5a, b), belonging to the calc-alkalic to alkali-calcic series (Figure 5c). The MgO contents of the host rock varied from 1.0 to 1.57 wt % with the Mg # (molar 100*MgO/(MgO+FeO T )) ranging from 35 to 64, belonging to the Magnesian series ( Figure 5d). The total REEs concentration of monzogranite host rocks from the Zankan region ranged from 63 to 224 ppm. The REE distribution patterns of monzogranites were lightly fractionated with (La/Yb) N ranging from 5.0 to 15.1 and negative Eu anomalies (Eu/Eu* = 0.17-0.28) (Figure 6a). In the N-MORBnormalized trace element diagram, the samples were enriched in large ion lithophile elements (LILEs) and depleted in Sr and high field strength elements (HFSEs) (Figure 6b). The monzonitic-syenitic enclave samples contained variable SiO 2 (50.1-61.4 wt %) and K 2 O (0.4-3.6 wt %) contents and showed high Na 2 O (4.2-8.1 wt %) contents. The samples contained high alkalis contents (K 2 O+Na 2 O = 7.8-9.4 wt %) in geochemically alkalic or its periphery fields (Figure 5a, c) and metaluminous field with A/ CNK values ranging from 0.66 to 0.73 (Figure 5b). The MgO contents of the enclaves varied from 2.6 to 4.4 wt %, with the Mg # values of 35 to 65. The enclaves contained high concentrations of total REEs (136 to 278 ppm). The samples were enriched in light REEs (LREEs) and depleted in heavy REEs (HREEs), and showed a rightinclined REE pattern, with (La/Yb) N values ranging from 2.7 to 12.5 and negative Eu anomalies (Eu/ Eu* = 0.18-0.24) (Figure 6c). In the N-MORB-normalized trace element diagram, the samples were enriched in LILEs and depleted in Sr and HFSEs (Figure 6d).
Three syenogranite samples collected from the Laobing region had homogeneous SiO 2 contents ranging from 70.6 to 72.9 wt %, which appeared in a weak peraluminous field in the A/CNK-A/NK diagram with A/ CNK ratios in the range of 1.02-1.04 (Figure 5b). The MgO contents of the syenogranites varied between 0.9 and 1.3 wt %, belonging to the Magnesian series ( Figure  5d). Those samples contained high concentrations of total REE ranging from 101 to 177 ppm, enriched in LREE and depleted in HREE and showed a rightinclined REE pattern with (La/Yb) N ratios ranging from 10.6 to 12.8 with medium to negative Eu anomalies (Eu/ Eu* = 0.19-0.25) (Figure 6e). Additionally, in the N-MORB-normalized trace element diagram, the samples were enriched in U and K and depleted in Sr and HFSEs (Figure 6f).

Whole-rock Sr-Nd-Hf isotope compositions
Two monzogranite samples and four monzonitic-syenitic enclave samples from the Zankan area were collected for the whole-rock Sr-Nd-Hf isotope analysis. The Sr-Nd-Hf isotopic data are presented in Tables S4 and S5. The findings of our study and results from previous literature are presented in Figures 7 and 8. The monzogranites from the Zankan region showed a low ( 87 Sr/ 86 Sr) i value ranging from 0.68181 to 0.70630 with a narrow range of ε Nd (t) from −4.2 to −4.7 and ε Hf (t) from −1.4 to +0.3. The enclave samples exhibited a wide variation in the initial 87 Sr/ 86 Sr (0.68248-0.70420) and had similar ε Nd (t) (−5.2 to −3.6) and ε Hf (t) (0.3 to 1.0) values. Sr is one of the most mobile elements, and the 87 Sr/ 86 Sr system is known to be seriously affected by weathering and alteration processes (Ma et al. 2010). Nearly all the Sr isotopic composition of   patterns (b, d, f). The normalized values are after. Sun and Mcdonough (1989) granites in TSHT was affected by alterations (Figure 7), thus, the Sr isotopic data have not been discussed in the following section.

Geochronology of the Ediacaran-Cambrian magmatism in the TSHT
Recently, Early Cambrian gabbros and granitoids have been identified in the TSHT ranging from the south to north. The 528 ± 3 Ma biotite monzogranite to the south of TSHT, 532 ± 3 Ma metamorphic gabbro in the Dahongliutan area , and the 525 ± -2-533 ± 4 Ma granites from Nanpingxueshan have been discovered (Liu et al. 2019;Yin et al. 2020). The dacitic porphyry with ages ranging from 527 ± 9 to 545 ± 7 Ma was reported by Lin et al. (2015) and Qiao et al. (2015) in the central region of the TSHT near the Zankan region. Zhang et al. (2018b) found 526 ± 3-33 ± 4 Ma aged gabbro sheets near the Tashkorgan County, and zircon with crystallization ages of ca. 530 ± 6 Ma from the granites from the Ayilixi pluton was discovered by ZZhu et al. (2016) (Figure 1b). Moreover, this study not only included LA-ICP-MS zircon U-Pb dating but also highprecision SHRIMP zircon dating to estimate the timing of the Late Ediacaran-Early Cambrian magmatism more accurately. Through SHRIMP analysis, we discovered the earliest subduction-related magmatic record (550.4 ± 6.4 Ma). Thus, the igneous rocks varying from the gabbros to granites were emplaced between ca. 550 and 530 Ma, which indicated an extensively developed intense contemporaneous magmatism than those during the Late Ediacaran-Early Cambrian in the TSHT.

Effect of magma mixing
The occurrence of enclaves is a ubiquitous phenomenon of calc-alkaline granitic plutons (Barbarin 2005). Four popular models have been proposed to explain the origin of mafic-dioritic enclaves: (1) xenoliths (Maas et al. 1997); (2) restite (Chappell et al. 1987); (3) autoliths (e.g. Donaire et al. 2005;Chen et al. 2016); and (4) incompletely digested and homogenized products of magma mixing or mingling (e.g. Barbarin 2005;Yang et al. 2007). The xenolith model refers to the capture of previously formed wall-rocks during magma ascent (Maas et al. 1997), and the restite model emphasizes the mafic-dioritic enclaves, representing residuals of the source rock (e.g. Chappell et al. 1987). In such cases, the enclaves should ideally exhibit older diagenetic ages than the host rocks. However, similar zircon U-Pb ages and isotopic data between the monzoniticsyenitic enclaves and the monzogranites observed in this study (Figure 4, 8) were inconsistent with those of the xenoliths and restite models. In addition, the enclaves were randomly distributed in the host monzogranites displaying igneous microtextures (Figure 3a, b, d). If the enclaves represent restites or xenoliths, metamorphic or residual sedimentary fabric is expected (White et al. 1999). However, the petrogenesis of enclaves observed in this study cannot be attributed to cumulates of hornblende, plagioclase, or biotite, because of the absence of either a cumulate texture (Barbarin 2005) or signals of cumulate hornblende or plagioclase, as shown in the REE pattern diagram. Cumulates of accessory minerals, such as zircon and apatite can elevate the concentration of Zr and P in the enclaves , which was not observed in our samples. Therefore, we can conclude that the enclaves are the products of magma mixing or mingling of the mantle-derived mafic magma and the crustderived felsic magma.
Notably, the host rocks and the enclaves displayed overall whole-rock ε Nd (t) and ε Hf (t) values, which can be interpreted by either the monzoniticsyenitic enclaves and host monzogranites originating from similar primary magma sources or the isotopic equilibrium during the magma mixing (Lesher 1990;Poli et al. 1996). As discussed in Section 5.2.2, the host granites could have originated from the low-K basaltic source, which were distinctly different from the monzonitic-syenitic enclaves with K 2 O contents varying from 0.4 to 3.6 wt %. Moreover, the enclaves displayed higher total REEs concentration and alkali content (Na 2 O+K 2 O) than the coeval calc-alkalic host monzogranites (Figure 5c). These evidence suggest a distinct primary magma source of the monzogranites and monzonitic-syenitic enclaves. Thus, we conclude that isotopic homogeneity may be due to the isotopic equilibrium during magma mixing. Radiogenic isotopic equilibrium is achieved more rapidly than chemical equilibrium during magma mixing (Lesher 1990;Poli et al. 1996). However, to what extent magma mixing affects the chemical composition of the monzogranites and the monzonitic-syenitic enclaves Is relatively unknown. Magma mixing with digestion produces linear correlations between elements and oxides in Harker plots (Clemens and Stevens 2012); however, the major and trace element compositions of the monzonitic-syenitic enclaves and monzogranites from the Zankan region did not combine in coherent magma mixing arrays (Fig S3), implying that chemical composition exchange, especially of major elements, did not occur. Additionally, the distinct contact relationship between the majority of the enclaves and the host rocks indicated that the mixing was not complete (Figure 3a). Therefore, we analysed the chemical composition of the enclaves to understand whether they could dominantly inherit the mantle-derived melts that were injected into the felsic magma chamber. The enclaves were the products of emplacement-level mingling and did not indicate extensive chemical composition exchange. Since the host monzogranites do not contain significant proportions of monzonitic-syenitic enclaves (less than 5% in volume) i.e. considerable high volumes over the enclaves, the chemical composition of host monzogranites magmas remained relatively unaffected.

Petrogenesis of the granitoids
The Zankan pluton comprises monzogranite with metaluminous field characterized by molar A/CNK values lower than 1.1 (Figure 5b). Corresponding, hornblende was widespread in the rock, while muscovite was absent. Moreover, the SiO 2 content was negatively correlated with the Al 2 O 3 content but positively correlated with the Th abundances ( Fig S3). These results suggest that the Zankan pluton has an affinity to I-type granites (Chappell and White 2001). Similarly, the majority of the contemporaneous granitoids from other TSHT regions also represented the characteristics of I-type granites (Figure 6, S3).
The I-type granites are usually generated by extensive fractional crystallization of basaltic-andesitic magmas or partial melting of crustal materials induced by underplating of mantle-derived basaltic magma and magma mixing to some extent (e.g. Clemens 2003;Sisson et al. 2004;Kemp et al. 2007). In this study, although the whole-rock ε Hf (t) and ε Nd (t) values of the monzogranites were similar to that of the monzonitic-syenitic enclaves (Figure 8), fractional crystallization of the monzoniticsyenitic enclaves or their parental magmas can be overlooked for two reasons. First, high volumes of basaltic or andesitic magmas are essential for the generation of granites by fractional crystallization of mantle-derived magmas. Practically, only a small proportion of gabbros are exposed and felsic components are dominant in the ɛ Nd (t) ɛ Hf (t) Zankan quartz porphyry  Chahekou Granite Ayilixi Granite  Dahongliutan Granite (Liu et al., 2019) Taaxi Gabbro ( Zankan region; this was inconsistent with the hypothesis of the fractional crystallization process. Second, if the granites were formed by fractional crystallization of melanocratic minerals and plagioclase, the coherent decreasing trends in SiO 2 vs. TiO 2 , Fe 2 O 3 , MgO, and CaO, and increasing trends in Na 2 O and K 2 O would be significant ( Fig S3). However, no direct fractional crystallization trends between the monzogranites and monzonitic-syenitic enclaves or the mafic rocks from the study area were noted in the Harker diagram. Moreover, the monzonitic-syenitic enclaves display higher alkalic content than the coeval granites, suggesting that they originated from distinct magmatic sources. Thus, these observations indicate that the monzogranites in the Zankan region were not generated by the fractional crystallization of mantle-derived magmas.
Although partial melting of chemically immature metasedimentary rocks may yield I-type granites, experimental studies revealed that the partial melts of immature metasedimentary rocks are dominantly peraluminous (Clemens 2018), whereas the Zankan monzogranites were metaluminous (Figure 5b). I-type granites are typically expected to form from sources dominated by igneous rocks; however, the magmas ranged from metaluminous to weakly peraluminous. Thus, the granites in this research probably originated from a metaigneous rock basement. This interpretation is supported by low Al 2 O 3 /(FeO+MgO+TiO 2 ) ratios and high Al 2 O 3 + FeO+MgO+TiO 2 values, which clearly indicate meta-basaltic sources (Fig S4a). Previous experimental studies indicated that dehydration melting of low-K basalts can produce intermediate to felsic melts that are typically low in K 2 O content and high in Na 2 O/K 2 O value (>1) (Rushmer 1991;Rapp and Watson 1995). The granite samples in this study show low K 2 O content and high Na 2 O/K 2 O content (3.5 to 30.1), indicating that these rocks may have originated from a low-K basaltic source. The negative whole-rock ε Nd (t) (−4.2 to −4.7) values of the granites may indicate that these rocks were mostly generated by dehydration melting of the ancient lower crust. The presence of monzonitic-syenitic enclaves in the Zankan area and the higher Mg# (35--64, 50.1 on average) than that of pure crust-derived melts (<40) (Patiño Douce 1999), indicate that these monzogranites were affected by different levels of contributions from mantle-derived components. Flattened HREE patterns and high HREE concentrations of the monzogranites observed in this study ( Figure  6a) illustrate that garnet residues were absent because of their extremely high partition coefficients (Kd) for HREEs (Bea et al. 1994), which could result in strong depletion of HREE in residual melts. This hypothesis of low-pressure dehydration melting was also verified through the CaO/Al 2 O 3 vs. CaO+Al 2 O 3 diagram ( Fig  S4b). Therefore, we proposed that the I-type granites observed in the Zankan region may have been derived from the low-K basaltic lower crust above the garnet stability field with different levels of contributions from mantle-derived magmas.
The Laobing syenogranites with weak peraluminous demonstrated the A/CNK ratios that were less than 1.1. Along with the occurrence of hornblende in the samples, a negative correlation between the P 2 O 5 content and SiO 2 content, and a positive correlation between the Th abundances and SiO 2 contents (Fig S3h) was observed, suggesting an affinity of the Laobing pluton towards I-type granites (Chappell and White 2001). The Laobing syenogranites and the Zankan monzogranites were characterized by a low Al 2 O 3 /(FeO+MgO+TiO 2 ) ratio and high Al 2 O 3 + FeO+MgO+TiO 2 value (Fig S4a). The appearance of the granites in the amphibolite-derived area indicated that they originated from the same crusts. The K 2 O content and Na 2 O/K 2 O ratio of the Laobing granites ranged from 0.1 to 0.2 wt % and 34 to 87, respectively, resembling the Zankan host granites. Thus, we can conclude that these granites originated from the same low-K meta-basaltic source.

Petrogenesis of the monzonitic-syenitic enclaves
As mentioned earlier, the monzonitic-syenitic enclaves from the Zankan region belonged to geochemically alkalic series (Na 2 O+K 2 O = 7.8-9.4 wt %) (Figure 5a). The following points should be considered in regard to the genesis of syenite-like rocks: (1) partial melting of the lower crustal material occurs under the influence of fluid or high-pressure conditions (Huang and Wyllie 1981;Lubala et al. 1994); (2) the product of crust-mantle mixing, i.e. mixing of mantle-derived mafic magma with crust-derived felsic magma is followed by subsequent differentiation of the resultant hybrid melts (Wickham et al. 1995(Wickham et al. , 1996Zhao et al. 1995); (3) low-degree partial melting of the enriched lithospheric mantle or the product of residual melt crystallization occurs after the differentiation of alkaline basaltic magma (Sutcliffe et al. 1990;Lynch et al. 1993).
The opinion of the origination of the monzoniticsyenitic enclaves from melting of the lower crustal material can be eliminated by high Mg# (up to 68.8) of some monzonitic-syenitic enclaves which analogous to those of high Mg andesites derived from the mantle source in the TSHT . Besides, previous studies showed that the magmas produced by the partial melting of mafic rocks are generally granitic rather than syenitic (Rapp and Watson 1995;Litvinovsky et al. 2002). As mentioned previously, the Sr-Nd-Hf isotopic compositions of the monzonitic-syenitic enclaves were similar to the host rocks, suggesting the occurrence of magma mixing (e.g. Lesher 1990;Yang et al. 2007). However, the enclaves presented higher total REEs concentration and alkalic content (Na 2 O+K 2 O) than the coeval calcalkalic host granites. These geochemical characteristics cannot be explained by the magma mixing of mantle-derived mafic magma with crust-derived felsic magma, indicating that the primary magma of the enclaves could have been originated from an enriched mantle source. Notably, the enclaves were enriched with of LREEs and LILEs and depleted with HREEs and HFSEs, which was consistent with the arclike volcanic rocks. Additionally, alkaline arc lavas are known to occur globally in modern subduction zones, such as the circum-Pacific subduction zones (Cruz-Uribe et al. 2018).
Furthermore, we deduced that the enclaves might have been derived from an enriched sub-continental lithospheric mantle (SCLM) above the subduction zone. Similarly, the Cambrian gabbros in the TSHT reported by Liu et al. (2019) were considered to have been derived from the SCLM.

Tectonic setting
The tectonic setting of the THST during the Ediacaran-Cambrian has been under controversy for a long time. Based on stratigraphic and sedimentary characteristics, some researchers had speculated that the tectonic background of THST was subjected to a passive continental marginal tectonic environment during the Early Palaeozoic (Ji et al. 2005). However, the sedimentary strata in the Early Cambrian were absent in the present study area. The Early Cambrian gabbros (ca. 530 Ma) near the Tashkorgan County were suggested to be derived from a metasomatised asthenosphere mantle source in a forearc setting (Zhang et al. 2018b;Liu et al. 2019). Based on the observations of the high Mg andesites (519-513 Ma) in the Mazha area, Zhang et al. (2020) proposed that the northern margin of the TSHT transformed into an active continental margin at 520 Ma. The rhyodacites (527-533 Ma) and iron deposits in the Zankan region were believed to form in the continental marginal arc background . Furthermore, the granite plutons from Ayilixi, Nanxueping, and Maeryang in the TSHT were interpreted to be formed by intense crust-mantle interaction due to the subduction of the Proto-Tethys Ocean (ZZhu et al. 2016;Li et al. 2019;Yin et al. 2020) (Figure 1b).
As discussed above, the development of coeval arc-related mafic rocks, mafic-dioritic enclaves, and high Mg andesites indicated the occurrence of a series of crust-mantle interaction magmatic processes, including partial melting of the mantle, magma mixing and mingling, and partial melting of the crust. Along with the clear zonation of the calcalkaline I-type granitoids (Figure 1b), we infer that an active margin was developed along the TSHT during the Cambrian. Based on the earliest subduction-related magmatic records in the WKOB identified by SHRIMP zircon dating in this study, it is reasonable to assert that the initial subduction of the Proto-Tethys oceanic slab must have occurred prior to the Early Cambrian (>550 Ma).
This was verified, on the tectonic discrimination diagram (Pearce et al. 1984), as the granitoid samples fell in the volcanic arc granite (VGA) (Figure 9a, b), indicating a subduction tectonic setting during the Precambrian-Early Palaeozoic magmatism. Based on the Th/Ta-Yb tectonic discrimination diagram (Gorton and Schandl 2000), the granitoid and enclave samples exhibited Th/Ta ratios in the range of 13.2-17.9; additionally, the data were plotted in the active continental margin field (Figure 9c).

Geodynamic interpretation
The Late Neoproterozoic-Early Palaeozoic was an important period during the geological evolution of Gondwana, as the final amalgamation of the Gondwana supercontinent and initial subduction of the Proto-Tethys Ocean along the Peri-Gondwana margins occurred during this period (Meert 2003;Cawood and Buchan 2007;Murphy et al. 2011;ZZhu et al. 2012;Li et al. 2018;Zhao et al. 2018). Many continental/micro-continental blocks along the northern margin of East Gondwana, including Turkey, Iran, Lhasa, Qiangtang, Himalaya, and Sibumasu Terrane suffered the subduction of the Proto-Tethys Ocean Ustaömer et al. 2012;Zhu et al. 2012;Gürsu et al. 2015;Ding et al. 2015;Moghadama et al. 2015Moghadama et al. , 2017. Among them, the TSHT and SKT (subterranes of the WKOB) were micro-continents, closing to the Tarim, west Qiangtang, Iran, and other blocks, which were distributed along the northern margin of the East Gondwana continent .
Spatio-temporal configurations of different subterranes (e.g. Wang 2004;Zhang et al. 2018a;Yin et al. 2020) and the subduction polarity of the WKOB (Yin and Harrison 2000;Xiao et al. 2003a;Wang 2004;Liu, Jiang, Jia et al. 2014;Zhang et al. 2020) during the Ediacaran-Cambrian have been controversial. Based on the study of the high Mg andesites in the TSHT, Zhang et al. (2020) proposed that the northward subduction of the Proto-Tethys Ocean beneath the SKT occurred before 520 Ma, and the bidirectional subduction at 520 Ma transformed the northern margin of the TSHT into an active continental margin. However, this proposed model was contrary to the age of magmatism, which occurred in TSHT before SKT (Figure 1b, 10a, b). Previous studies have proposed the initial subduction of the Proto-Tethys Ocean at 540-530 Ma, and the existence of SKT as a Palaeozoic accretionary wedge (Zhang et al. 2018b;Liu et al. 2019). Based on the discovery of Mesoproterozoic-Neoproterozoic stromatolites such as Jacutophyton, Conophyton, Baicalia, and Paraconophyton in the Sangzhutage Group ) and the 815 Ma gneissic granites in the Kudi ophiolite, the perception SKT being a Palaeozoic accretionary wedge is presently under controversy. Yin et al. (2020) suggested that the Proto-Tethys Ocean subducted southwards beneath the SKT and TSHT as early as the Early Cambrian (>533 Ma), and the oceanic slab roll-back induced arc retreat and northeastward migration of magmatism. This model considered that the TSHT and SKT were combined in the Early Palaeozoic. However, the differences in sedimentary and metamorphic deformation characteristics indicated that the SKT and TSHT remained separate until the late Early Palaeozoic (e.g. Ji et al. 2005, Zhang et al. 2018b. Moreover, the material exchange between the SKT and the TSHT until ca. 431 Ma was inconspicuous, and the initial collision of these two continents occurred at 431-420 Ma . Therefore, the oceanic slab roll-back model cannot well explain the age distribution of magmatism between the SKT and the TSHT. Since the Late Neoproterozoic-Early Palaeozoic stratigraphy and metamorphism in the WKOB were infrequent, a clear understanding of the Cambrian ophiolites and magmatism exposed in the WKOB is fundamental to address these controversies. Conventionally, the Kudi ophiolite in the WKOB has been regarded as the oldest suture zone in the northern Tibetan Plateau (Pan et al. 1994;Yuan et al. 2005) that represented the residual of the Proto-Tethys Ocean (Zhang et al. 2018a, b;Wang et al. 2020;Zhang et al. 2020). Nevertheless, the Kangxiwa-Subashi Suture Zone was inferred as a suture separating in the Late Palaeozoic, and ended in the Triassic, representing the evidence of the Palaeo-Tethys Ocean (Yin and Harrison 2000). The boninites in the Kudi ophiolite were considered to be developed under a supra-subduction zone environment, which was consistent with the predominantly forearc nature, and suggesting the existence of a vast ocean between the SKT and Tarim during the Early Palaeozoic Yin et al. 2020 Linnemann et al. 2008;Ustaömer et al. 2009;Gürsu et al. 2015;Moghadam et al. 2015Moghadam et al. , 2017 (Figure 11a). A continental magmatic arc developed at the periphery of the West African Craton for the Cadomian fragments and a related backarc basin opened during ca. 590 to 570 Ma. Closure of the Cadomian back-arc basin and arc-continent collision occurred at ca. 543 Ma (Linnemann et al. 2008) (Figure 11a). Increasing evidence shows that Late  Table 1.   Meert (2003) and the references therein. The sources of the geochronological data for Andean-type orogen along the proto-Tethyan margin are listed in Table 1. (c) Model showing the Early Late Ediacaran-Early Palaeozoic lithofacies palaeogeographic pattern of Tarim and the WKOB and formation of the Late Ediacaran-Early Palaeozoic magmatic rocks in the Tianshuihai Terrane. The thickness of the crust and the asthenospheric mantle is not to scale. the subduction-related magmatic rocks in the SKT (513-499 Ma) was comparable with the coeval magmatism at the Indian-Australian terranes. However, the age distribution of the subduction-related magmatism in the TSHT (550-499 Ma) showed common chronological characteristics between Arabian terranes and Indian-Australian terranes (Figure 11b). This suggested that SKT might have been located near the Indian-Australian terranes, whereas the TSHT was probably located at the transitional area between the Arabian terranes and Indian-Australian terranes during the Late Ediacaran-Early Cambrian period.
Interestingly, the age of magmatic rocks in the microcontinent terranes (Lhasa Terrane, under debate (Zhang et al. 2012b;ZZhu et al. 2012)), located palaeogeographically in the northern margin of the East Gondwana continent from the west to the east gradually decreased (Figure 10, 11b). This may indicate that the subduction of the Proto-Tethys Ocean could be diachronous, initiating in the western region of the East Gondwana continent, and propagating eastwards of the East Gondwana continent. The initial subduction of the Proto-Tethys Ocean beneath the THST occurred before 550 Ma, which occurred earlier than the previously suggested ca. 540-530 Ma (Zhang et al. 2018b;Liu et al. 2019) or 520 Ma . Thus, the difference in the magmatic age distribution between TSHT and SKT may be due to the different time of the initial subduction of the Proto-Tethys Ocean (Fig 13b). Moreover, based on the SHRIMP zircon U-Pb dating, we speculated that an active continental margin developed along the TSHT during the Late Ediacaran-Early Cambrian and initial subduction of the Proto-Tethys oceanic slab must have occurred prior to the Early Cambrian (>550 Ma) (Figure 11c). Additionally, according to the palaeogeographic reconstruction and the age distribution of magmatism, the TSHT and the SKT were assumed to have been distributed between the northern margin of the East Gondwana continent and the Tarim Block. Thus, two branches of the Proto-Tethys Ocean coexisted, which was represented by the Kangxiwa Fault and Kudi Ophiolite Belt during the Early Palaeozoic. Furthermore, depending on the chronological statistics of microblocks in the northern margin of the East Gondwana continent, the subduction of the Proto-Tethys Ocean could be diachronous, which initiated in the northwestern region of the East Gondwana continent and propagated eastwards.

Conclusions
In this study, we reported zircon SHRIMP and LA-ICP-MS U-Pb ages, whole-rock major and trace elements, and Sr-Nd-Hf isotopic data of Late Ediacaran-Early Palaeozoic granitoid and monzonitic-syenitic enclaves from the TSHT in the WKOB.
(1) The granites in the Laobing region yielded SHRIMP zircon U-Pb ages ranging from 550.4 ± 6.4 Ma to 547.5 ± 5.3 Ma, which were the earliest age records of the Palaeozoic magmatism in the TSHT. The host monzogranites in the Zankan area yielded SHRIMP and LA-ICP-MS zircon U-Pb ages ranging from 542.6 ± 8.4 Ma to 540.5 ± 2.8 Ma, which were coeval with the syenodioritic enclave ages ranging from 533.8 ± 3.4 Ma to 534.7 ± 3.0 Ma.
(2) The monzonitic-syenitic enclaves may have been originated from an enriched SCLM above the subduction zone, and the monzogranites may have been derived from low-K basaltic lower crust above the garnet stability field with varying levels of contributions from mantle-derived magmas.
(3) Furthermore, depending on the SHRIMP zircon U-Pb dating, we speculated that an active continental margin developed along the TSHT during the Late Ediacaran-Early Cambrian and the initial subduction of the Proto-Tethys oceanic slab must have occurred prior to the Early Cambrian (>550 Ma). Moreover, according to the palaeogeographic reconstruction and age distribution of magmatism, the TSHT and the SKT may be distributed between the northern margin of the East Gondwana continent and the Tarim Block. Thus, two branches of the Proto-Tethys Ocean coexisted, which were represented by the Kangxiwa Fault and Kudi Ophiolite Belt during the Early Palaeozoic.
(4) Finally, depending on the chronological statistics of micro-blocks in the northern margin of the East Gondwana continent, the subduction of the Proto-Tethys Ocean could be diachronous, which initiated in the northwestern region of the East Gondwana continent and propagated eastwards.