Abstract

As a pivotal junction between the North China and Tarim cratons, the Alxa Terrane provides an ideal window to constrain the final closure of the middle segment of the Paleo-Asian Ocean (PAO). This study carried out petrological, whole-rock geochemical, and zircon U-Pb-Hf isotopic investigations on four major granitic plutonic complexes in the Alxa Terrane. The Bayan Nuru and Yabulai plutonic complexes are I-type granitoids that yield crystallization ages of 281–268 Ma and 277–270 Ma, respectively, with negative zircon εHf(t) values (−11.5 to −3.2), primarily sourced from the Neoproterozoic rocks in the region. The Nuergai granitoids yield crystallization ages of 281–268 Ma and show I-type affinities and positive zircon εHf(t) values (+1.6 to +6.4), indicating an origin related to magma mixing. Emplaced at ca. 241 Ma, the Oliji granitoids display varying zircon εHf(t) values from −0.9 to +11.5, which necessitate a predominant source of mantle-derived materials. A compilation of zircon εHf(t) and whole-rock εNd(t) values of the magmatic rocks in the Alxa Terrane illustrates a decreasing trend from the Late Carboniferous to the Early Permian and an increasing trend during Middle Permian to Triassic time. The marked shift with a large variation of zircon εHf(t) and whole-rock εNd(t) values at 280–265 Ma indicates a tectonic switch from subduction to post-collision tectonic regimes in the Alxa Terrane, marking the final closure of the middle segment of the PAO. Comparable isotope variations are also identified from 260 to 245 Ma magmatic counterparts on the northern margin of the North China Craton, hence suggesting a progressively eastward closure of the PAO.

INTRODUCTION

As one of the most magnificent orogens in Earth’s history, the Central Asian Orogenic Belt (CAOB) formed from the accretion and amalgamation of the Siberia and East Europe cratons to the north and the North China and Tarim cratons to the south, due to multiple episodes of subduction and closure of the Paleo-Asian Ocean (PAO) (Şengör et al., 1993; Khain et al., 2003; Xiao et al., 2003, 2009, 2013, 2015; Jahn, 2004; Windley et al., 2007; Schulmann and Paterson, 2011; Wilhem et al., 2012; Eizenhöfer et al., 2014; Han et al., 2016a, 2016b; Chen et al., 2017). In the past decade, considerable progress has been made in understanding the final closure of the PAO along the Tianshan and Solonker suture zones, which were adjacent to the northern margins of the Tarim and North China cratons (Şengör et al., 1993; Xiao et al., 2003, 2009, 2013, 2015; Windley et al., 2007; Wilhem et al., 2012; Eizenhöfer et al., 2014; Han et al., 2016b). However, controversy still exists regarding the timing of the final closure of the PAO, with different perspectives arguing for distinct closure times that range from the Late Devonian to the Late Triassic (Xiao et al., 2003, 2009, 2013, 2015; Charvet et al., 2007, 2011; Windley et al., 2007; Jian et al., 2008, 2010; Han et al., 2011, 2016b; Wilhem et al., 2012; Xu et al., 2013; Eizenhöfer et al., 2014, 2015a, 2015b; Zhang et al., 2015a, 2015b, 2016a, 2016b).

Located between the North China Craton (NNC) to the east and the Tarim Craton to the west (Fig. 1A), the Alxa Terrane in westernmost Inner Mongolia occupies a crucial place to understand the closure processes in the middle segment of the PAO. The terrane is characterized by large volumes of late Paleozoic to early Mesozoic magmatic rocks, of which the changes of magma sources and isotopic records with time can provide important information for unraveling the history of the consumption of the middle segment of the PAO. In this study, we carried out field-based petrological, zircon U-Pb-Hf isotopic and whole-rock major- and trace-elemental investigations on a variety of Permian to Early Triassic granitoids across the Alxa Terrane. By integrating the new results and previous data, we aim to better constrain the origin of the magmatic rocks, the regional tectonic settings, and the timing of the final closure of the PAO.

GEOLOGICAL BACKGROUND

Located in westernmost Inner Mongolia, the Alxa Terrane is a triangular block separated from the NCC to the east by the Langshan fault (Fig. 1B). The Alxa Terrane borders the North Qilian Orogen to the southwest by the Longshoushan fault and the southernmost CAOB to the north by the Enger Us fault (Fig. 1B). The ca. 300 Ma Enger Us ophiolitic mélange along the Enger Us fault is considered to demarcate the final closure site of the PAO (BGMRIM, 1991; Wang et al., 1994; Wu et al., 1998; Xie et al., 2014; Zheng et al., 2014). Largely covered by the Badain Jaran desert, the Alxa Terrane is subdivided into the Nuru-Langshan and Shalazhashan belts separated by the Badain Jaran fault, along which ca. 275 Ma Quagan Qulu ophiolitic mélange outcrops (Fig. 1B) (BGMRIM, 1991; Wang et al., 1994; Wu et al., 1998; Zheng et al., 2014).

Precambrian basement rocks and Paleozoic to Mesozoic magmatic rocks widely occur in the Nuru-Langshan belt (NLB). The Precambrian basements in the belt are composed mainly of ca. 2.5 Ga tonalite-trondhjemite-granodiorite, 2.3–2.0 Ga orthogneisses reworked by high amphibolite-facies metamorphism at 1.93–1.80 Ga, Paleoproterozoic paragneisses, and 970–800 Ma magmatic and sedimentary rocks (Geng et al., 2002; Li et al., 2004; Peng et al., 2010; Geng and Zhou, 2011; Dan et al., 2012, 2014b; Gong et al., 2012; J.X. Zhang et al., 2013; Hu et al., 2014; Wu et al., 2014; Zhao et al., 2015). Voluminous late Paleozoic to early Mesozoic magmatic rocks across the NLB yielded zircon U-Pb ages of 418–239 Ma, with a main peak at the Permian time (summarized in Liu et al., 2016). In addition, some early Paleozoic dioritic and granitic intrusions with zircon ages of 460–423 Ma have been reported in the eastern and southern parts of the NLB (Dan et al., 2015a; Liu et al., 2016).

The Shalazhashan belt (SLB) is composed mainly of late Paleozoic to early Mesozoic magmatic rocks, with subordinate late Paleozoic sedimentary rocks and minor early Paleozoic magmatic rocks and Precambrian basement rocks (Fig. 1B). The late Paleozoic to early Mesozoic magmatic rocks in the SLB include predominant 301–247 Ma granitoids and subordinate 264–249 Ma diorites and gabbros (Ran et al., 2012; W. Zhang et al., 2013; Shi et al., 2014a, 2014b; Yang et al., 2014). The late Paleozoic sedimentary successions are represented by the Amushan Formation, which has Late Carboniferous to Early Permian depositional ages. The lower and middle sequences of the Amushan Formation consist mainly of clastic rocks interbedded with mafic to felsic volcanic rocks that yielded zircon U-Pb ages of 320–302 Ma, and the upper sequence is a molasse composed of silty shale, sandstone, gravel-bearing sandstone, and conglomerate (Lu et al., 2012; W. Zhang et al., 2013).

SAMPLING

In this study, granitic intrusions in the Alxa Terrane were widely sampled, with 17 samples from the eastern to western parts of the NLB and nine samples from the middle part of the SLB. Due to the paucity of the outcrops in the Alxa Terrane, the contact relationship between different rock types in the same plutonic complex can hardly be recognized in the field. Detailed locations, lithology, mineral assemblages, and analytical results of samples are summarized in Table 1.

The Bayan Nuru Plutonic Complex

Samples 14LQ01A–14LQ01C, 14LQ14A, and 14LQ14B were collected near the Bayan Nuru area in the eastern part of the NLB (Fig. 2A). Intruding the Precambrian basement rocks and intruded by Early Triassic granitoids, the Bayan Nuru granitic plutonic complex contains ultramafic to intermediate intrusive enclosures occurring as small elliptical or irregular bodies (Fig. 2A). These intrusions show crystallization ages varying from 440 Ma to 270 Ma (Fig. 2A; Liu et al., 2016).

Samples 14LQ01A–14LQ01C are monzogranite with a porphyritic texture (Fig. 3A), of which alkali feldspar grains are coarse-grained phenocrysts, and quartz grains mostly appear as fine-grained aggregates showing late recrystallization. Samples 14LQ14A and 14LQ14B are syenogranite, containing more abundance of alkali feldspar with 1% anhedral muscovite and minor minerals of epidote-zoisite and fluorite.

The Yabulai Plutonic Complex

In the middle part of the NLB, four samples (14LQ34A, 14LQ34B, 14LQ35A, and 14LQ35B) were collected from the Yabulai granodioritic plutonic complex, which is intruded by Triassic granitoids in the region (Figs. 2B and 3B). The common minerals are hornblende, biotite, quartz (38%–45%), alkali feldspar (15%–17%), and plagioclase (35%–45%) (Fig. 3C), with minor epidote-zoisite and sphene. Hornblende and biotite grains show preferred orientations yielding a lineation and a gneissosity, respectively (Figs. 3B and 3C). Quartz grains occasionally tend to be tiny grains with undulatory extinction due to ductile shear deformation (Fig. 3C).

The Nuergai Plutonic Complex

Approximately 80 km northwest to Alxa Youqi County (Fig. 1B), the Nuergai granitic plutonic complex in the western part of the NLB is characterized by a varied rock assemblage of quartz diorite, granodiorite, and syenogranite (Fig. 2C). The relationship of these rock suites is unclear due to the paucity of outcrops and intense weathering.

Samples 14LQ26A–14LQ26C are weakly deformed quartz diorite with major mineral assemblages of hornblende (∼20%), biotite (∼10%–20%), plagioclase (50%–55%), and quartz (10%–20%) (Fig. 3D), of which hornblende is locally altered to epidote-zoisite. The syenogranite (samples 14LQ30A, 14LQ30B, and 14LQ31A) contains coarse-grained alkali feldspar of >50% (Fig. 3E) and biotite showing chlorite alteration, whereas the granodiorite (samples 14LQ33A and 14LQ33B) comprises more hornblende (10%–15%), biotite (∼5%) and plagioclase (∼40%). Plagioclase grains show local sericite alteration (Fig. 3F).

The Oliji Plutonic Complex

The Oliji plutonic complex in the middle part of the SLB intrudes the Amushan Formation and is crosscut by a Jurassic granitic porphyry (Fig. 2D). This plutonic complex is characterized by medium- to coarse-grained granodiorite (Fig. 3G) and monzogranite (Fig. 3H).

The granodiorite (samples 14LQ37A and 14LQ37B, 14LQ38A and 14LQ38B, and 14LQ39A) is composed mainly of quartz (∼30%), alkali feldspar (10%–20%), plagioclase (40%–50%), hornblende (3%–5%), and biotite (5%–10%), with minor epidote-zoisite, apatite, and sphene. Alkali feldspar and plagioclase grains show partially clay and sericite alteration, respectively. Hornblende and biotite grains occasionally show secondary alteration of chlorite and epidote-zoisite. Comparatively, four monzogranite samples (14LQ44A, 14LQ44B, 14LQ45A, and 14LQ45B) are characterized by more alkali feldspar (>30%) and quartz (35%–40%) with much less biotite. Occasional anhedral muscovite appears due to late alteration of biotite.

ANALYTICAL METHODS AND RESULTS

LA-ICPMS U-Pb Dating of Zircon

Prior to dating analysis for zircons, cathodoluminescence (CL) images were taken on a MonoCL3 attached to a scanning electron microscope (JSM-6510A, Japan) at Jinyu Technology Co., Ltd., Chongqing, China. Zircon U-Pb dating analysis was carried out using a laser ablation-inductively coupled plasma mass spectrometry (LA-ICPMS) at FocuMS Technology Co., Ltd., Nanjing, China. The system comprises a Photon Machines Excite 193nm laser ablation system and a quadrupole ICPMS (Agilent 7700x). The following were working parameters: 7 Hz repetition rate, 6.71 J/cm2 energy, and 40 μm laser spot diameter. Zircon standard 91500 (Wiedenbeck et al., 1995) was used for U-Pb isotopic ratio correction, and zircon standard GJ-1 (Jackson et al., 2004) was used as unknown. A weighted mean 206Pb/238U age of 601 ± 5 Ma (n = 72) was obtained for the latter, in agreement with the recommended value of 600.4 ± 0.6 Ma (Jackson et al., 2004). Detailed zircon U-Pb isotopic data are listed in GSA Data Repository Table DR11.

Age of the Bayan Nuru Plutonic Complex

Zircons in a monzogranite sample 14LQ01A are euhedral and prismatic. They show oscillatory zoning in CL images (Fig. 4A) and high Th/U ratios of 0.50–0.87, indicating a magmatic origin. Fourteen analyses yield concentrated 206Pb/238U ages, with a weighted mean age of 268 ± 1 Ma (MSWD = 0.8; Fig. 4A), interpreted as the crystallization age of this monzogranite. One grain has an older age of 307 Ma, considered as a xenocrystic zircon in the monzogranite. Zircons from the syenogranite (sample 14LQ14A) are of a magmatic genesis, showing oscillatory zoning (Fig. 4B) and an average Th/U ratio of 0.48. Twenty-three analyses produce a weighted mean 206Pb/238U age of 281 ± 1 Ma (MSWD = 1.0; Fig. 4B). Taken together, the crystallization age of the Bayan Nuru granitic plutonic complex is 281–268 Ma.

Age of the Yabulai Plutonic Complex

Sample 14LQ34A is granodiorite, in which zircons are typical of a magmatic genesis with elongated prismatic forms, oscillatory zoning (Fig. 4C), and high Th/U ratios of >0.4. Nineteen analyses show similar 206Pb/238U ages between 280 Ma and 275 Ma, with a weighted mean age of 277 ± 1 Ma (MSWD = 0.8; Fig. 4C), interpreted as the crystallization age of this granodiorite. Similarly, 21 analyzed zircons from a monzogranite sample 14LQ35A also show oscillatory zoning (Fig. 4D) with high Th/U ratios and give a weighted mean 206Pb/238U age of 270 ± 1 Ma (MSWD = 1.2; Fig. 4D). Thus, the Yabulai plutonic complex yields a crystallization age of 277–270 Ma.

Age of the Nuergai Plutonic Complex

Zircons extracted from a quartz dioritic sample 14LQ26A have stubby prismatic forms (Fig. 4E). They show oscillatory zoning (Fig. 4E) and varying Th/U ratios from 0.50 to 1.17, reflecting a magmatic origin. Twenty-four analyzed grains have concentrated 206Pb/238U ages, with a weighted mean age of 269 ± 1 Ma (MSWD = 0.9; Fig. 4E), interpreted as the crystallization age of the quartz diorite. Samples 14LQ30A and 14LQ31A are monzogranite, in which zircons are commonly elongated in shape with ratio aspects of >3 (Figs. 4F and 4G) and characterized by oscillatory zoning and high Th/U ratios of >0.4. Eighteen zircons in the former and 20 grains in the latter were analyzed and yield a similar weighted mean 206Pb/238U age of 268 ± 1 Ma (MSWD = 0.5; Fig. 4F) and 269 ± 1 Ma (MSWD = 0.6; Fig. 4G), respectively, indicating the crystallization age of the monzogranite. Zircons in another granodiorite sample 14LQ33A are of a magmatic origin with oscillatory zoning (Fig. 4H) and high Th/U ratios, 17 of which yield a weighted mean 206Pb/238U age of 281 ± 2 Ma (MSWD = 1.0; Fig. 4H), relatively older than that of the quartz diorite and monzogranite. Taken together, the crystallization age of the Nuergai plutonic complex is 281–268 Ma.

Age of the Oliji Plutonic Complex

Zircons from samples 14LQ38A and 14LQ44A show Th/U ratios varying from 0.39 to 1.10, and oscillatory zoning (Figs. 4I and 4J), indicative of a magmatic origin. Twenty analyses from the former and 16 from the latter yield an equal weighted mean 206Pb/238U age of 241 ± 1 Ma (MSWD = 0.8; Fig. 4I) and 242 ± 1 Ma (MSWD = 1.4; Fig. 4J), respectively, interpreted as the crystallization age of the Oliji granitoids. In addition, five inherited and/or xenocrystic zircons in sample 14LQ44A show older concordant 206Pb/238U ages between 271 Ma and 251 Ma, coincident with the giant Permian magmatic event (272–250 Ma) in the SLB (Ran et al., 2012; W. Zhang et al., 2013; Shi et al., 2014a, 2014b).

Zircon Hf-Isotope Compositions

Zircon Lu-Hf isotope analysis (presented in GSA Data Repository Table DR2) was conducted using a Neptune Plasma multi-collector ICPMS attached to a New Wave ArF 193 nm COMPex Pro laser ablation system at the State Key Laboratory of Mineral Deposits Research, Nanjing University (NJU), China. The adopted working parameters were: 44 μm ablation pit diameter, 8 Hz repetition rate, and 25 s ablation time. In order to monitor reliability and stability of the instrument, two reference zircon standards 91500 and Mud Tank were analyzed. Mean 176Hf/177Hf ratios of 0.282306 ± 0.000022 (n = 21) and 0.282502 ± 0.000017 (n = 22) were obtained, respectively, equal to suggested values for 91500 (0.282284 ± 0.000003; Wiedenbeck et al., 1995) and Mud Tank (0.282497 ± 0.000018; Hawkesworth and Kemp, 2006) within the error. In order to calculate the zircon initial 176Hf/177Hf ratios, εHf(t) values, depleted mantle model ages and two-stage model ages (T<SUP>C</SUP>DM), the following parameters were adopted. The decay constant is 1.865 × 10−11 per year for 176Lu (Scherer et al., 2001). The 176Hf/177Hf and 176Lu/177Hf ratios for chondrite and depleted mantle at present are 0.282772 and 0.0332 (Blichert-Toft and Albarede, 1997) and 0.283251 and 0.0384 (Vervoort and Blichert-Toft, 1999), respectively. The 176Lu/177Hf ratio for the average continental crust is 0.015 (Griffin et al., 2002).

The ca. 268 Ma zircons from the Bayan Nuru plutonic complex show homogeneous and negative εHf(t) values from −6.1 to −3.2, and two-stage model ages (T<SUP>C</SUP>DM) of 1.67–1.49 Ga, whereas the ca. 281 Ma zircons in the Bayan Nuru plutonic complex yield relatively more negative εHf(t) values and older T<SUP>C</SUP>DM ages (Fig. 5A). The 277–270 Ma zircons from the Yabulai plutonic complex have more clustered and negative εHf(t) values (−7.4 to −3.0) and T<SUP>C</SUP>DM ages of 1.75–1.49 Ga (Fig. 5A). In contrast, different granitic samples from the Nuergai plutonic complex display positive εHf(t) values from +1.6 to +6.4 and corresponding younger T<SUP>C</SUP>DM ages of 1.20–0.89 Ga (Fig. 5A).

The ca. 241 Ma zircons in sample 14LQ38A from the Oliji plutonic complex display fairly positive εHf(t) values from +8.2 to +11.5 and young T<SUP>C</SUP>DM ages of 0.75–0.54 Ga, whereas those in sample 14LQ44A have more heterogeneous and enriched Hf-isotope compositions with εHf(t) values varying from −0.8 to +7.4, and older T<SUP>C</SUP>DM ages of 1.32–0.80 Ga (Fig. 5A). In addition, older zircons (256–271 Ma) in sample 14LQ44A show positive εHf(t) values of +5.0 to +7.7 (Fig. 5A).

Whole-Rock Major- and Trace-Element Compositions

Whole-rock major-element compositions were determined using an X-ray fluorescence (XRF) spectrometer at the NJU, and whole-rock trace-elemental analysis was carried out using a quadrupole ICPMS at the State Key Laboratory of Ore Deposit Geochemistry, Guiyang Institute of Geochemistry (GYIG), Chinese Academy of Sciences, China. Generally, analytical precisions for major- and trace-elements are higher than 2% and 5%, respectively. Analytical results are given in GSA Data Repository Table DR3.

Whole-Rock Major-Element Compositions

The Permian granitoids in the NLB display a large variation in major-element compositions with SiO2 contents varying from 55.0% to 76.8% (Fig. 6). The 281–268 Ma Bayan Nuru granitoids and the 277–270 Ma Yabulai granodiorites show calc-alkaline to high-K calc-alkaline affinities (Fig. 6). They plot into the field of weakly peraluminous to metaluminous (Fig. 7A) and magnesian series (Fig. 7B), with molar Al2O3/(CaO+Na2O+K2O) (A/CNK) values of 0.97–1.02 and low FeOt/(MgO + FeOt) ratios of 0.47–0.82, more akin to I-type granitoids (Whalen et al., 1987). Comparatively, the 281–268 Ma Nuergai granitoids show a much larger variation in major-element compositions. They exhibit roughly lower Al2O3, Na2O, K2O, and P2O5 contents (Fig. 6) and lower A/CNK values (0.87–0.98) and Al2O3/TiO2 ratios at similar silica contents than those of the Bayan Nuru and Yabulai counterparts, indicative of typical I-type affinities (Whalen et al., 1987).

In comparison, the ca. 241 Ma Oliji granitoids in the SLB have lower Fe2O3t and K2O contents with lower FeOt/(MgO + FeOt) and Al2O3/TiO2 ratios at the same SiO2 levels than those of the Permian granitoids from the NLB (Figs. 6 and 7). These Early Triassic granitoids show I-type features (Whalen et al., 1987), belonging to calc-alkaline to high-K calc-alkaline (Fig. 6), weakly peraluminous to metaluminous with A/CNK values of 0.92–1.04 (Fig. 7A), and magnesian series (Fig. 7B).

Whole-Rock Trace-Element Compositions

The Permian granitoids in the NLB and the Early Triassic granitoids in the SLB also show a broad range of trace-element compositions, with the former containing relatively lower transition elements (e.g., Sc, V, Cr, Co, and Ni) and Sr contents, higher Rb, Y, rare-earth elements (REEs) and Nb components, and higher Rb/Sr ratios.

Distinct REEs patterns are illustrated in Figures 8A and 8B. The Permian granitoids in the NLB exhibit varying REEs contents ranging from 81 ppm to 249 ppm, weak to medium REEs fractionation with light REEs (LREEs) / heavy REEs (HREEs) ratios of 5.80–9.49 and (La/Yb)N ratios of 5.29–10.30, and strongly to weakly negative Eu anomalies (Eu/Eu* = 0.11–0.87) (Fig. 8A). In contrast, except two monzogranite samples with fairly high silica contents, the Early Triassic granitoids in the SLB mostly display lower REEs contents of 46–135 ppm, stronger REEs fractionation with higher LREEs/HREEs and (La/Yb)N ratios, and less negative Eu anomalies (Eu/Eu* = 0.80–0.98) (Fig. 8B).

In the mid-ocean ridge basalt (MORB)–normalized, incompatible–trace-element spidergrams (Figs. 8C and 8D), almost all the Permian and Early Triassic granitoids in this study demonstrate relative depletions in high field strength elements (HFSEs, e.g., Nb, Ta, and Ti) and P and relative enrichments in K, Rb, and Ce.

DISCUSSION

Permian to Triassic Magmatic Episodes in the Alxa Terrane

LA–ICPMS zircon U-Pb dating on ten samples from four granitic plutonic complexes indicates a Permian to Triassic magmatic event in the Alxa Terrane. New results reveal that the crystallization age of the Bayan Nuru and Nuergai plutonic complexes is 281–268 Ma, roughly overlapping the age of the Yabulai plutonic complex at 277–270 Ma. Therefore, an episode of magmatism between 280 and 268 Ma took place in the Alxa Terrane, characterized by varied granitoids including quartz diorite, granodiorite, monzogranite, and syenogranite. This is consistent with previous studies that suggested a Permian magmatic event in the Alxa Terrane (e.g., Dan et al., 2014a, 2015b; Liu et al., 2016). Apart from the predominant granitoids, some 280–264 Ma ultramafic to intermediate magmatic rocks were also unveiled in the eastern part of the NLB and the middle part of the SLB (Feng et al., 2013; L. Zhang et al., 2013; Shi et al., 2014a; Liu et al., 2017).

This study also unveils that the Oliji granitoids were emplaced at ca. 241 Ma, slightly younger than the previous reported zircon U-Pb ages of 254–250 Ma for the nearby monzogranite and granodiorite from the Oliji plutonic complex (Fig. 2D) (W. Zhang et al., 2013; Shi et al., 2014a). Thus, the Oliji granitic plutonic complex most likely evolved in a period from the Latest Permian to the Early Triassic. In addition, coeval 255–239 Ma ultramafic to felsic plutonic rocks have also been identified in the SLB and the eastern part of the NLB (Zhang et al., 2010; Ran et al., 2012; Shi et al., 2014a, 2014b; Wang et al., 2015b).

Petrogenesis

Fractional Crystallization

The Bayan Nuru and Nuergai granitoids in this study contain a large variety of rock assemblages with crystallization ages ranging from 281 Ma to 268 Ma. Fractional crystallization can be the potential cause for such a large age span of different granitoids. If this is the case, the younger granitoids should have more felsic compositions. However, the ca. 268 Ma Bayan Nuru monzogranites yield lower SiO2 contents than those of the ca. 281 Ma syenogranites, while the younger Nuergai quartz diorites show much less felsic compositions (Table DR3 [see footnote 1]). Such a long period of ca. 13 Ma is also unreasonable for a single fractional crystallization process. Therefore, fractional crystallization alone cannot explain the distinct geochemical compositions for the Bayan Nuru and Nuergai plutonic complexes.

The Yabulai and Oliji granitic plutonic complexes exhibit obvious linear correlations for major elements on the Harker diagrams (Fig. 6), suggesting a significant role of fractional crystallization. Negative correlations of TiO2, Al2O3, Fe2O3t, MgO, CaO, and P2O5 with respect to SiO2 (Fig. 6) indicate fractional crystallization of mafic minerals (e.g., biotite and/or hornblende) and plagioclase with minor apatite and ilmenite. The positive correlation between K2O and SiO2 (Fig. 6) rules out the possibility of fractionation of alkali feldspar or biotite. Accordingly, the Yabulai and Oliji granitoids show fractional crystallization of hornblende + plagioclase + ilmenite + apatite. However, fractional crystallization alone cannot explain the distinct geochemical compositions of these granitoids, especially in terms of their distinct REE patterns and zircon Hf-isotope compositions.

To conclude, heterogeneous sources were probably a more important factor in controlling the petrogenesis of the Permian and Early Triassic granitoids in this study.

Sources

The 281–268 Ma Bayan Nuru granitoids and the 277–270 Ma Yabulai granitoids show homogenous and enriched zircon Hf-isotope compositions with negative εHf(t) values and old T<SUP>C</SUP>DM ages of 2.01–1.49 Ga, suggesting the involvement of the Precambrian basement rocks as source materials. The Neoarchean to Paleoproterozoic basements in the Alxa Terrane are excluded because they have fairly older T<SUP>C</SUP>DM ages of >2.5 Ga (Fig. 5B). The Neoproterozoic rocks in the region can potentially be source materials, since the zircon Hf-isotope evolutionary array of these rocks covers the zircon εHf(t) values of the Bayan Nuru and Yabulai granitic plutonic complexes, despite different rock types (Fig. 5B). These Neoproterozoic rocks are sporadically exposed in the eastern and southwestern parts of the NLB and consist mainly of predominant 970–840 Ma gneissic granitoids and metasedimentary sequences yielding youngest detrital zircon ages of 910–810 Ma and minor 830–800 Ma ultramafic intrusions and felsic volcanic rocks (Geng et al., 2002; Li et al., 2004; Geng and Zhou, 2010; Peng et al., 2010; Dan et al., 2014b; Hu et al., 2014; Tang et al., 2014). Among these rock units, the Neoproterozoic granitoids and felsic volcanic rocks show high silica contents of mostly >70% (Peng et al., 2010; Geng and Zhou, 2011; Dan et al., 2014b), unlikely to generate the various granitoids from the Bayan Nuru and Yabulai plutonic complexes. In addition, the metasedimentary rocks cannot produce these I-type granitic magmas, and the ultramafic intrusions are also excluded because it is difficult for granites to be generated by progressive fractional crystallization of mafic magmas (e.g., Clemens et al., 2011; Clemens and Stevens, 2012; Nandedkar et al., 2014). Alternatively, we suggest that the mixing of quartzofeldspathic rocks (i.e., granitoids and/or metasedimentary rocks) and ultramafic rocks might have been the potential candidates for source materials of these I-type granitoids, as evidenced by a large amount of geochemical modeling works (e.g., Gray, 1984; Patiño Douce, 1995, 1999; Collins, 1998; Moyen et al., 2001). The various inputs of mixing members can contribute to the distinct geochemical compositions for the Bayan Nuru plutonic complex.

In comparison, the 281–268 Ma granitoids from the Nuergai plutonic complex in the western part of the NLB show more concentrated and radiogenic zircon Hf-isotope compositions with positive εHf(t) values and relatively younger T<SUP>C</SUP>DM ages of 1.20–0.89 Ga. However, no pre-Permian rocks in the Alxa Terrane have the same Meso- to Neoproterozoic model ages to generate these granitoids. Instead, the Hf-isotope continuum of the Nuergai granitoids lies between a mantle-like component and the package of the Neoproterozoic rocks in the region, possibly due to crust-mantle magma mixing during their genesis. This interpretation is in accordance with I-type affinities of the Nuergai granitoids because such rocks are commonly considered to form in magma mixing processes (e.g., Gray, 1984; Foden et al., 2002; Griffin et al., 2002; Yu et al., 2005; Kemp et al., 2007, 2009). In order to quantify the proportions of crust- and mantle-derived materials, a simplified dual-members mixing model is adopted (Liu and Chen, 1991). In this model, the Neoproterozoic component in the region is assumed to be the evolved end member. This crust-derived magma is assumed to have zircon Hf-isotope compositions of these Neoproterozoic rocks (Geng and Zhou, 2011; Dan et al., 2014b), while an average Hf content of this crust-derived magma is equal to that of the Bayan Nuru and Yabulai granitoids that were sourced from these basement rocks. We adopt a Hf-isotope component of “New Crust” proposed by Dhuime et al. (2011) as that of the juvenile end member, and an average Hf content of this juvenile magma is equal to that of average arc basalts and andesites (Kelemen et al., 2003). Our calculated results show that ∼70% mantle-derived materials should have been added to generate the Nuergai granitoids. If the crust member involves older Archean to Paleoproterozoic basements in the region, the proportion of the juvenile materials would be larger. The geochemical heterogeneity of the Nuergai granitoids can be explained by various proportions of crust- and mantle-derived materials.

Similarly, the ca. 242 Ma monzogranite from the Oliji plutonic complex in the SLB displays a large variation of zircon εHf(t) values ranging from −0.8 to +7.4, also consistent with the crust-mantle magma mixing model. In contrast, the ca. 241 Ma granodiorite in the Oliji plutonic complex shows fairly positive zircon εHf(t) values from +8.2 to +11.5, reflecting primarily mantle-derived materials. Calculated results of the simplified dual members mixing model suggest that ∼76% mafic end member may have been added in producing the monzogranite magma and up to 90% for the granodiorite with such depleted zircon Hf-isotope compositions. Such a large proportion emphasizes a considerably significant role of juvenile materials that are most likely the newly accreted lower crust.

Tectonic Setting

Permian Granitoids

A giant Permian granitic flare-up has been identified in the Alxa Terrane, whose tectonic setting, however, is still controversial, with different models advocating oceanic subduction, post-collisional, or mantle-plume settings (Ran et al., 2012; Shi et al., 2012; L. Zhang et al., 2013; W. Zhang et al., 2013; Dan et al., 2014a; Lin et al., 2014; Hu et al., 2015; Wang et al., 2015b).

The mantle-plume setting seems inadequate to account for the relatively smaller size (∼0.05 Mkm2) of the Permian granitic event in the Alxa Terrane than that of common silicic igneous provinces (>0.10 Mkm2; Bryan, 2007). Furthermore, mantle-plume–associated high-temperature continental flood basalts and komatiitic sequences (e.g., Campbell et al., 1989; Campbell and Griffiths, 1990) are lacking in the Alxa Terrane.

The post-collisional setting is also not preferred. The Permian magmatism was widespread both along the northern margin of the NCC and on the Alxa Terrane (Liu et al., 2016). Such a ∼2000-km-long linear magmatic belt cannot be interpreted in a post-collisional setting. In addition, the Early Permian porphyries in the eastern part of the Alxa Terrane experienced north-vergent deformation (Lin et al., 2014), which is amenable to a compressional south-dipping subduction setting, rather than an extensional post-collisional setting.

We thus favor the subduction setting in interpreting the origin of the Permian magmatic rocks in the Alxa Terrane based on the following lines of evidence:

  • (1) This study has revealed that the 281–268 Ma Nuergai granitoids were sourced from mixing magmas from the Neoproterozoic basements and mantle-derived materials, which is prone to occur in a subduction setting (e.g., Gill, 1981; Blundy and Sparks, 1992; Tepley et al., 2000; Zhou and Li, 2000; Annen et al., 2006; Reubi and Blundy, 2009).

  • (2) The zircon-saturation temperature calculated from whole-rock compositions of granitoids has been a powerful tool to estimate the crystallization temperature of magma at the time when zircon forms (Watson and Harrison, 1983). The calculated results reveal that most of the Permian granitic samples have low crystallization temperatures of 803–670 °C (Table DR3 [see footnote 1]). These low-temperature, I-type granitoids are more inclined to form at the high flux of water in a subduction setting (e.g., Hermann and Green, 2001; Collins et al., 2016), which goes against the mantle-plume or post-collisional models.

  • (3) A database has been compiled concerning the available zircon εHf(t) and whole-rock εNd(t) values in the Late Carboniferous to Triassic magmatic rocks in the Alxa Terrane. The increasingly evolved isotope compositions during Late Carboniferous to Middle Permian time (Figs. 9A and 9B) are in agreement with an advancing compressional subduction setting, instead of an extensional setting in the post-collisional or mantle-plume regimes (e.g., Kemp et al., 2009; Collins et al., 2011; Roberts et al., 2013; Han et al., 2016a).

  • (4) Coeval 280–268 Ma ultramafic to intermediate intrusions showing arc-like geochemical affinities have also been reported in the eastern part of the NLB, in favor of a subduction regime (Feng et al., 2013; L. Zhang et al., 2013; Liu et al., 2017).

Early Triassic Granitoids

The Early Triassic Oliji granitoids in this study are considered to form in a post-collisional setting according to following reasons:

  • (1) These Early Triassic granitic magmas show juvenile isotopic compositions (Fig. 5A), placing a predominant role of mantle-derived materials in magma generation. This is amenable to occurring in an extensional setting where asthenospheric mantle is prone to ascend and partially melt (e.g., Liégeois et al., 1998; Sylvester, 1998; Chen and Arakawa, 2005; Yang et al., 2008).

  • (2) The compiled zircon εHf(t) and whole-rock εNd(t) values of the magmatic rocks in the Alxa Terrane show an increasingly depleted trend from the Middle Permian to the Triassic (Figs. 9A and 9B), which may be attributed to the involvement of mantle-derived magmas in an extensional setting (e.g., Kemp et al., 2009; Collins et al., 2011).

  • (3) The 256–249 Ma high-K calc-alkaline granitoids and 266–250 Ma bimodal intrusive associations with radiogenic isotopic compositions have also been identified in the Alxa Terrane, indicating an extensional setting in a post-collisional regime since the Late Permian (W. Zhang et al., 2013; Lin et al., 2014; Shi et al., 2014a, 2014b).

Constraints on the Timing of the Final Closure of the PAO

The identification of the closure of a major ocean can be sufficiently clued from convergent high-strain zones with large-scale thrust faults and regional high-grade metamorphism, blueschist mélanges, and ophiolite remnants (e.g., Dewey and Şengör, 1979). However, unlike Himalaya-type collision, the middle and eastern segments of the PAO closed as a result of multiple accretionary processes via double-sided subduction, lacking the records of intensive deformation and high-grade metamorphism (e.g., Xiao et al., 2003; Eizenhöfer et al., 2014). In addition, the reservation of blueschist and ophiolite remnants in the Alxa Terrane is also scarce. The ca. 300 Ma Enger Us and ca. 275 Ma Quagan Qulu ophiolitic mélanges sporadically outcrop along two faults in the region, considered as a late Paleozoic ocean-arc system (Zheng et al., 2014), which only implies that the PAO did not completely disappear prior to Early Permian.

Nevertheless, the radiometric isotopic signatures of igneous rocks primarily reflect the characteristics of magma sources; samples show a large spread of zircon εHf(t) values probably attributed to magma mixing processes, whereas samples exhibit concentrated zircon εHf(t) values reflecting single old crustal or juvenile source domain. Numerous recent studies have efficiently correlated the variation of radiometric isotope signatures, especially zircon εHf(t) and whole-rock εNd(t) values, with large-scale switching tectonics (e.g., Jahn et al., 2000; Griffin et al., 2002; Kemp et al., 2009; Collins et al., 2011; Roberts et al., 2013; Smits et al., 2014; Boekhout et al., 2015; Han et al., 2016a; Liu et al., 2016). A major tectonic change (e.g., advancing-retreating subduction transition, oceanic closure, or collision) inevitably affects the magma sources and petrogenetic processes of igneous rocks, and hence changes their isotopic characteristics.

In general, a compressional geodynamic setting sees isotopic excursion to more enriched compositions (e.g., Smits et al., 2014; Boekhout et al., 2015; Han et al., 2016a; Liu et al., 2016), which is the case in the Alxa Terrane during Late Carboniferous to Middle Permian time (Figs. 9A and 9B). In this scenario, the advancing subduction of the PAO produced compression in the overriding Alxa Terrane and resulted in a local increase in crustal thickness and a larger degree of subduction erosion, leading to greater crustal assimilation and more enrichment of the mantle. Accordingly, gradually evolved isotope signatures were recorded in the magmatic arc (Figs. 9A and 9B). In contrast, an extensional setting exhibits excursion to more juvenile isotopes (e.g., Smits et al., 2014; Han et al., 2016a), which is also the case in this study; the Middle Permian to Triassic magmatic rocks in the Alxa Terrane demonstrate increasing εHf(t) and εNd(t) values (Figs. 9A and 9B), and the majority of the Triassic zircons exhibit fairly positive εHf(t) values (>+5; Fig. 9A), suggesting primarily juvenile materials as magma source. This extensional setting most likely corresponds to the closure of oceanic domains with collision or accretion events (e.g., Smits et al., 2014) or a retreating arc in the oceanic subduction regime (e.g., Han et al., 2016a). The former model is preferred on the basis of the 256–249 Ma high-K calc-alkaline granitoids and 266–250 Ma bimodal intrusive associations recognized in the region that have been considered to form in a post-collisional regime (W. Zhang et al., 2013; Lin et al., 2014; Shi et al., 2014a, 2014b). In this case, following the closure of the PAO lithosphere, the overriding Alxa Terrane probably underwent crustal thinning in a post-collisional setting. In this scenario, the asthenospheric mantle was upwelled, partially melted, and underplated, finally resulting in magmas with more juvenile isotope signatures. Notably, in the transitional period of ca. 280–265 Ma, a marked shift with a very large spread of positive to negative εHf(t) and εNd(t) values illustrating a mixed magma source can be identified, corresponding to a tectonic switch from subduction to post-collision due to the closure of the PAO. In this scenario, various crustal and mantle components could be involved during magma generation, such as old crustal materials, enriched lithospheric mantle, and depleted asthenospheric mantle. The involvement of different proportions of crust- and mantle-derived materials possibly produced magmas having heterogeneous isotope compositions with a very large spread of positive to negative zircon εHf(t) and whole-rock εNd(t) values (e.g., Smits et al., 2014). Similar cases in which a marked shift with a large variation of isotopes is related to the final closure of a major ocean also occurred in the adjacent areas and worldwide, including ca. 310–300 Ma closure of the western segment of the PAO along the northern margin of the Tarim Craton (Han et al., 2016b; Zhang et al., 2016a) and ca. 1.2–1.1 Ga assembly of the Proterozoic Australia among the North, West, and South Australian cratons (Smits et al., 2014). In addition, a Hf-isotope gap is also documented in the Alxa Terrane during the transitional period, which might have recorded the heterogeneity of the magma source in different locations (Figs. 9A and 9B).

The transitional period of 280–265 Ma is also in accordance with the sedimentary records in the Alxa Terrane. Xie et al. (2014) identified the late Middle Permian to the early Late Permian deep-water radiolarians fossils, i.e., albaillellarians, in the exotic siliceous rocks from the ca. 300 Ma Enger Us ophiolitic mélange, supporting that the final disappearance of the middle segment of the PAO should have occurred after the Late Permian. Moreover, the sedimentological history of the Late Carboniferous to Early Permian Amushan Formation suggests that the PAO might have begun to close in the Early Permian (Lu et al., 2012; W. Zhang et al., 2013). The lower and middle sequences of this formation contain 320–302 Ma mafic to felsic volcanic rocks, clastic rocks, and carbonates, representing oceanic arc and back-arc basin environments. In contrast, the upper sequence of the formation changes to molasse deposition, including silty shale, sandstone, gravel-bearing sandstone, and conglomerate, which indicates the swallowing basin from the Late Carboniferous to Early Permian.

Furthermore, a structural analysis in the Langshan area in the eastern part of the Alxa Terrane (Lin et al., 2014) reveals north-vergent deformation for Early Permian porphyries (290–280 Ma) and non-deformation for Late Permian dolerites (256 Ma). This implies a tectonic change from south-dipping subduction to post-collision between 280 Ma and 256 Ma, supporting our interpretation that the final closure of the PAO most likely occurred at 280–265 Ma.

Eastward Closing of the PAO during Late Carboniferous to Middle Triassic Time

Available data show that a ∼2000-km-long, ENE-stretching Paleozoic magmatic belt developed on the northern margin of the NCC; this belt is considered to extend westerly to the Alxa Terrane (summarized in Liu et al., 2016). For comparison, the zircon εHf(t) values of Late Carboniferous to Triassic magmatic rocks on the northern margin of the NCC are illustrated in Figure 9C. It is noteworthy that 330–250 Ma zircons display a large spread and roughly decreasing εHf(t) values with time (Fig. 9C). Most importantly, a marked shift with a large variation of positive to negative εHf(t) values appears at 260–245 Ma (Fig. 9C), which is seemingly also present on the whole-rock εNd(t) values versus age diagram (Fig. 9D). Subsequently, the isotopic values in the Triassic exhibit an overall increasing trend (Figs. 9C and 9D). Accordingly, a complexity of a mixing source in the Late Permian to Early Triassic and an increasingly juvenile source during Triassic time can be inferred. Such a marked change of Hf and Nd isotopic arrays and a subsequent increasing trend are similar but posterior to those in the Alxa Terrane and most likely reflect the timing of the main tectonic switch from subduction to post-collision orogenesis on the northern margin of the NCC. Moreover, a majority of previous paleogeographic, geochronologic, geochemical, and structural investigations also support that the final closure of the PAO on the northern margin of the NCC occurred in the Late Permian to Middle Triassic (Xiao et al., 2003, 2015; Zhang et al., 2004, 2007, 2009a, 2009b, 2009c, 2012b; Miao et al., 2007; Windley et al., 2007; Jian et al., 2008; Chen et al., 2009; Wilhem et al., 2012; Cao et al., 2013; Eizenhöfer et al., 2014, 2015a, 2015b; Wang et al., 2015a). Therefore, we conclude that the closure time of the PAO on the northern margin of the NCC, probably between 260 and 245 Ma, is slightly younger than that in the Alxa Terrane at 280–265 Ma.

Previous studies have proposed that the PAO completely closed in the Late Carboniferous on the western margin of the Tarim Craton, and the Latest Carboniferous to Early Permian on its eastern margin (e.g., Gao et al., 2011; Han et al., 2011, 2016b; Klemd et al., 2011; Xiao et al., 2015; Zhang et al., 2015a, 2015b, 2016a, 2016b). Taken together, it can be inferred that a notably eastward closing process of the PAO took place during Late Carboniferous to Early Permian time along the northern margin of the Tarim Craton, at the latest Early Permian to Middle Permian in the Alxa Terrane, and in the Late Permian to Middle Triassic on the northern margin of the NCC.

CONCLUSIONS

  • (1) New LA-ICPMS zircon U-Pb dating results reveal that the emplacement of Bayan Nuru-Nuergai, Yabulai, and Oliji granitoids occurred at 281–268 Ma, 277–270 Ma, and ca. 241 Ma, respectively, suggesting a Permian to Triassic magmatic event in the Alxa Terrane.

  • (2) Whole-rock geochemical and zircon Hf-isotopic data suggest that the Permian Bayan Nuru and Yabulai granitoids were sourced primarily from the Neoproterozoic rocks in the Alxa Terrane, whereas the Permian Yabulai granitoids were derived from magma mixing of crust- and mantle-derived materials, and the Early Triassic Oliji granitoids involved more prominent mantle-derived materials.

  • (3) The marked shift of zircon εHf(t) and whole-rock εNd(t) values at 280–265 Ma demonstrates a robust tectonic switch from subduction to post-collision, corresponding to the timing of the final closure of the PAO in the Alxa Terrane.

  • (4) An eastward closure of the PAO is inferred, i.e., in the Late Carboniferous to Early Permian along the northern margin of the Tarim Craton, at 280–265 Ma in the Alxa Terrane, and at 260–245 Ma on the northern margin of the NCC.

ACKNOWLEDGMENTS

This study was financially supported by the National Natural Science Foundation of China (NSFC) Project (41190075) entitled, “Final Closure of the Paleo-Asian Ocean and Reconstruction of East Asian Blocks in Pangea,” which is the fifth research project in the NSFC Major Program (41190070), “Reconstruction of East Asian Blocks in Pangea,” Hong Kong Research Grants Council (RGC) General Research Fund (GRF) (grants HKU7063/13P and 17301915), NSFC General Projects (grants 41230207 and 41390441), and University of Hong Kong (HKU) Seed Funding Program for Basic Research (grant 201311159126). We are grateful to Gang Zeng, Jianfeng Gao, Liang Li, and Tao Yang for their kind assistance in sample preparation and experimental analyses.

1GSA Data Repository Item 2017237, Table DR1: Zircon U-Pb dating results; Table DR2: Zircon Hf-isotope compositions; and Table DR3: Whole-rock major- (wt%) and trace-element (ppm) compositions, is available at http://www.geosociety.org/datarepository/2017, or on request from editing@geosociety.org.