Geochemistry and tectonic implications of Early Permian granitic rocks in the Xingxingxia area of Chinese Central Tianshan Arc Terrane

4.1 Zircon UPb geochronology

The zircon UPb dating results are given in Table S2. The zircon grains from all six dating samples show typical characters of magmatic zircon such as euhedral crystal and well-developed oscillatory zoning in the CL images (Figure 4). However, zircon grains from sample X11-88-2 are mantled by a narrow and dark rim (Figure 4f), which are interpreted to be magmatic zircon probably affected by later recrystallization (He, Zhang, Zong, & Dong, 2013).

Zircon grains from quartz diorite sample X12-30-3 and associated dioritic enclave sample X12-30-5 yielded weighted mean ²⁰⁶Pb/²³⁸U ages of 284 ± 1 Ma (n = 23; MSWD = 0.45) and 284 ± 1 Ma (n = 22; MSWD = 0.78), respectively (Figure 5a,b). Twenty-five spots were analysed on 25 zircon grains from amphibole granite (sample X10-15-1), yielding a weighted mean ²⁰⁶Pb/²³⁸U age of 281 ± 2 Ma (MSWD = 0.62; Figure 5c). Zircon grains from biotite granite (sample X10-14-1) gave a weighted mean ²⁰⁶Pb/²³⁸U age of 284 ± 1 Ma (n = 25; MSWD = 0.48; Figure 4d). Zircon grains from dioritic porphyry dyke (sample X10–14-2) gave a weighted mean ²⁰⁶Pb/²³⁸U age of 285 ± 2 Ma (n = 16; MSWD = 0.33, Figure 5e). In addition, 12 spot analyses were conducted on zircon grains from weakly mylonitic granitic dyke (sample X11-88-2), yielding a weighted mean ²⁰⁶Pb/²³⁸U age of 275 ± 2 Ma (MSWD = 1.40; Figure 5f).

4.2 Zircon LuHf isotopic compositions

Zircon LuHf isotopic data are presented in Table S3. Zircon grains from quartz diorite (sample X12-30-3) show a narrow range of ε_Hf(t) values of +4.2 to +6.7 and T_DMC model ages of 0.95 to 0.80 Ga (n = 15). Twenty zircon grains from the dioritic enclave sample X12-30-5 yielded ε_Hf(t) values of +4.1 to +8.4 and T_DMC model ages of 0.95 to 0.70 Ga (Figure 6). Eleven zircon grains from amphibole granite (sample X10-15-1) gave similar ε_Hf(t) values of +5.4 to +7.2 and T_DMC model ages of 0.87 to 0.76 Ga. Twelve Hf isotopic spot analyses on zircon grains from the biotite granite sample X10-14-1 yielded ε_Hf(t) values of +4.6 to +7.4 and T_DMC model ages of 0.92 to 0.76 Ga. Zircon grains from the dioritic porphyry dyke sample X10-14-2 have negative ε_Hf(t) values of −8.0 to −5.7 and T_DMC model ages of 1.64 to 1.51 Ga (n = 15; Figure 6). Thirteen zircon grains from the mylonitic granitic dyke sample X11-88-2 yielded ε_Hf(t) values of +7.7 to +10.9 and T_DMC model ages of 0.73 to 0.55 Ga (Figure 6).

4.3 Zircon trace element geochemistry

The trace element compositions of the analysed zircon grains are given in Table S4. Zircon grains from six samples show variable Th and U contents and Th/U ratios (Figure 7a), indicating they crystallized from different composition of magmas (Grimes et al., 2007; Grimes et al., 2015). However, all of them show relatively high U/Yb and Gd/Yb ratios, which indicates great contributions from crustal sources (Grimes et al., 2007, 2015). Furthermore, they plot in the region of continental arc zircon but are distinct from zircon from mid-ocean ridge (MOR), ocean island, and post-collisional settings (Figure 7b–d).

4.4 Whole-rock major and trace element geochemistry

The whole-rock major and trace elements are presented in Table S5. The quartz diorite and associated dioritic enclave have low SiO₂ contents of 58.08 and 52.48 wt.%, respectively, and show magnesian and metaluminous affinities (Figure 8). They further have high MgO contents (3.47–4.82 wt.%) and Mg^# values (53–56) [Mg^# = Mg/(Mg + Fe²⁺) × 100] and high Cr (52.2–107 ppm), Ni (41.2–80.8 ppm), and Sr (495–517 ppm) and low Y (23–24 ppm) contents, which are similar to high-Mg diorites (Shirey & Hanson, 1984; Smithies & Champion, 2000). Their chondrite-normalized REE patterns show relatively light REE (LREE) enrichment and no significant Eu anomalies (Figure 9a). In the primitive mantle-normalized trace element diagrams (Figure 9b), they display positive anomalies of Rb, K, Pb, Zr, and Hf and negative anomalies of Th, Nb, Ta, and Ti.

The amphibole granite, biotite granite, dioritic porphyry dyke, and mylonitic granitic dyke have high SiO₂ contents (65.78 to 71.80 wt.%) and mainly fall in the fields of magnesian, calc-alkali, and metaluminous to peraluminous granites (ASI [alumina saturation index, Frost et al., 2001] = 0.97 to 1.04; Figure 8). They show LREE enriched REE patterns [(La/Yb)_N = 8.48 to 13.36] and weakly negative Eu anomalies (Eu/Eu* = 0.67 to 0.84; Figure 9a). On the primitive mantle normalized multi-element diagrams, they display positive anomalies for Rb, Th, U, K, and Pb and variable Nb, Ta, P, Sr, and Ti negative anomalies (Figure 9b). All these samples are comparable with the previously published Early Permian granitic rocks from the CTA, and they together plot in the field of Cordilleran Mesozoic batholiths of North America (Figure 8; Frost et al., 2001).

5.1 The multiple magma sources of the Early Permian granitic rocks

The studied quartz diorite and dioritic enclave, amphibole granite, biotite granite, dioritic porphyry dyke, and mylonitic granitic dyke were emplaced between 285 and 275 Ma (Figure 5), which are almost simultaneously when the accuracy of the LA-ICP-MS zircon dating technique is further considered (Horstwood, Kosler, Jackson, Pearson, & Sylvester, 2009; Li et al., 2015). They further exhibit similar magnesian, calc-alkali, and metaluminous to weak peraluminous compositions, which can be compared with the Cordilleran Mesozoic batholiths in a continental arc setting (Figure 8; Frost et al., 2001). Their negative Nb, Ta, and Ti anomalies are typical features for subduction-related magmatic rocks (Figure 9; Keppler, 1996). Their zircon trace element compositions are also consistent with a continental arc setting (Figure 7; Carley et al., 2014; Grimes et al., 2007, 2015). In particular, the majority of their zircon ε_Hf(t) values plot above the crustal basement evolution region of the CTA (see He et al., 2014; He et al., 2015) and extend near the depleted mantle line (Figure 6). Therefore, we suggest that the magmas of these rocks were probably derived by partial melting of a subduction-influenced mantle wedge and were subsequently evolved by fractional crystallization. The ancient crust may still contribute to the magma sources as evidenced by the relatively large zircon ε_Hf(t) variations (from −8 to +10.9; Figure 6). Their relatively weak Sr, Ba, Eu, and P depletion, remarkable REE fractionation [(La/Yb)_N = 8.48–13.36] and roughly flat MREE to HREE [(Dy/Yb)_N = 1.02–1.51] are consistent with amphibole-dominated fractionation. This matches the feature of magmatic evolution in subduction zones that plagioclase crystallization can be delayed due to the water-rich environments, while crystallization of amphibole is favoured (Davidson, Turner, Handley, Macpherson, & Dosseto, 2007; Larocque & Canil, 2010; Sisson & Grove, 1993).

The dioritic porphyry dyke is typically characterized by negative zircon ε_Hf(t) values, which plot in the crustal basement evolution region of the CTA (Figure 6). This suggests that the magma of the dioritic porphyry dyke was probably derived by reworking of the ancient Mesoproterozoic basement in the CTA (see He et al., 2015).

5.2 Implications for the ridge subduction of the South Tianshan Ocean in the Eastern Tianshan

Ridge-subduction accompanied by the formation of slab window and ridge-trench interaction is very common in accretionary orogens, such as the modern Pacific Rim (McCrory & Wilson, 2009; Sisson, Pavlis, Roeske, & Thorkelson, 2003; Windley et al., 2007; Xiao et al., 2015; Xiao et al., 2017). Multiple ridge subduction events have also been proposed from a variety of locations of the CAOB. For example, Jian et al. (2008) suggested that the high-grade metamorphism (451–436 Ma), adakite plutonism (448–438 Ma), and low-P/T metamorphism (440–434 Ma) in the Inner Mongolia were formed by repeated ridge-trench interactions. Sun et al. (2009) and Cai et al. (2012) suggested that ca. 420 Ma zoned mafic-ultramafic intrusions in the Altai and coeval high-T-low-P metamorphisms were related to ridge subduction and a resultant slab window. A suite of ca. 320 Ma charnockites, adakites, high-Mg diorites, and tholeiites in the West Junggar was also proposed to be generated by ridge subduction, which are characterized by depleted ε_Hf(t) and ε_Nd(t) values (Geng et al., 2009; Tang et al., 2010; Tang, Wang, et al., 2012; Yin et al., 2010). Ridge subduction was also reported from the Alxa Block on the southeastern margin of the CAOB (Feng et al., 2013).

As stated above, an Early Permian magmatic “flare-up” occurred in the CTA (Table S1; Figure 10), which could not be generated by “normal” subduction (Sajona et al., 1993; Thorkelson, 1996). Furthermore, the zircon ε_Hf(t) values of the granitic rocks in the CTA increase systematically with emplacement ages from Devonian to Early Permian time (Figure 6), which is consistent with a gradual increase of depleted-mantle inputs and reaching a maximum in Early Permian time. The abundant Early Permian mafic-ultramafic rocks in the CTA also indicate the major contributions of asthenospheric mantle in the magmatisms, such as ca. 284–281 Ma mafic-ultramafic rocks in Baishiquan area (Chai et al., 2007, 2008; Mao et al., 2006) and ca. 290 Ma mafic-ultramafic rocks in Tianyu area (Tang et al., 2009). Both Baishiquan and Tianyu mafic-ultramafic rocks are characterized by zoned complexes and MORB-like geochemical signatures imply that they were generated by asthenosphere decompression melting (Chai et al., 2007; Tang et al., 2009). We propose that the Early Permian magmatic “flare-up” in the CTA and the remarkably juvenile contributions in the granitic rocks can be well explained by a ridge subduction activity that the subduction of spreading centre of the South Tianshan Ocean in the Eastern Tianshan resulted in the formation of a slab window. The asthenospheric mantle will move upward through the triggered slab window, which provide a large amount of heat and result in partial melting and high-temperature metamorphism of the overriding continental crust (e.g., Aguillon-Robles et al., 2001; Brown, 1998; Drummond & Defant, 1990; Geng et al., 2009; Iwamori, 2000; Kelemen, Yogodzinski, & Scholl, 2003; Tang et al., 2010; Yin et al., 2010). The interactions of multiple-derived magmas, including those from the mantle wedge and/or ancient crustal basement, formed the studied Early Permian granitic associations (Figure 11). Furthermore, a ridge subduction activity in the CTA is also supported by other geological evidence in the CTA, as follows.

The Precambrian basement rocks of the CTA experienced two stages (ca. 380 and 300 Ma) of amphibolite- to granulite-facies metamorphisms (He et al., 2014; He et al., 2015). The latter, synchronous with the studied Early Permian granitic associations, was associated with partial melting of Precambrian basement rocks and the generation of leucogranites suggesting a low-pressure/high-temperature metamorphism condition (see He et al., 2015). A recent study of the Late Palaeozoic Yushugou high-temperature (>930 °C) mafic granulite in the Chinese Eastern Tianshan suggested it indicates the subduction of the mid-oceanic ridge (Zhang & Jin, 2016) although the metamorphic age remains speculative (vary from 380 to 278 Ma; Wang et al., 2003; Zhang, Zhang, Xia, & Lü, 2017). In addition, Early Permian A-type granites were also revealed from the Baluntai area of the CTA, reflecting high-temperature crustal melting (Dong et al., 2011). Furthermore, many orogenic gold deposits and CuNi sulphide mineralization in the CTA were also suggested to be associated with ridge subduction, which are similar to the gold mineralization in Southern Alaska (Han et al., 2010). Therefore, our model is rather consistent with the general appreciation of an active continental margin setting of the CTA in Early Permian time and supports that the final closure of the South Tianshan Ocean in the Chinese Eastern Tianshan took place after the Late Permian (Xiao et al., 2009; Xiao et al., 2013; Xiao et al., 2015; Xiao et al., 2017).