Abstract

Garnet-bearing mafic granulite and amphibolite exposed as lenses, boudins, or interlayers within metasediments in the Qingshigou area, Dunhuang block, southernmost Central Asian orogenic belt, record important information for understanding the tectono-metamorphic evolution of subduction and collision zones in the southern Central Asian orogenic belt during the mid-Paleozoic. Three stages of metamorphic assemblages (M1, M2, M3) are recognized in the high- and medium-pressure mafic granulite and amphibolite. In the high-pressure mafic granulite, the prograde assemblage (M1) is represented by inclusion minerals (hornblende + plagioclase + quartz) preserved in garnet porphyroblasts; the metamorphic peak assemblage (M2) is characterized by garnet porphyroblasts and matrix minerals (garnet + clinopyroxene + plagioclase + quartz ± zircon ± titanite); and the retrograde assemblage (M3) is marked by coronitic symplectite (hornblende + plagioclase + quartz ± magnetite) rimming the garnet porphyroblasts. In the medium-pressure mafic granulite, the prograde assemblage (M1) of hornblende + plagioclase + quartz is included in the garnet porphyroblasts; the peak assemblage (M2) consists of garnet + orthopyroxene + clinopyroxene + plagioclase + quartz ± zircon ± titanite (M2) in the matrix; and the retrograde assemblage (M3) of hornblende + orthopyroxene + plagioclase + quartz (M3) surrounds the garnet porphyroblasts. In the amphibolite, the prograde assemblage (hornblende + plagioclase + quartz + ilmenite) is preserved as inclusions in garnet (M1); the peak assemblage (M2) is composed of garnet + hornblende + plagioclase + quartz ± zircon ± titanite; and the retrograde assemblage (M3), consisting of hornblende + biotite + plagioclase + quartz + epidote + magnetite, rings the garnet porphyroblasts. Geothermobarometric calculations suggest that the metamorphic pressure-temperature paths pass from 568 °C and 8.8 kbar through 607 °C and 10.6 kbar and 861 °C and 16.9 kbar and finally to 598 °C and 4.4 kbar for the high-pressure mafic granulite; from 756 °C and 9.0 kbar through 750–874 °C and 9.3–11.6 kbar to 675 °C and 4.7 kbar for the medium-pressure mafic granulite; and from 686 °C and 7.6 kbar through 715–766 °C and 10.6–11.2 kbar to 671 °C and 5.6 kbar for the amphibolite, and the paths show clockwise pressure-temperature loops typical of an orogenic process. The metamorphic peak of the high-pressure mafic granulite lies in the eclogite facies, which is indicative of a subduction zone environment. High-resolution secondary ion mass spectrometry (SIMS) U-Pb dating of metamorphic zircon indicates that the metamorphism occurred in the Early Silurian (ca. 430 Ma) and lasted for at least 65 m.y. This study reveals a possible southward subduction history of a branch of the Paleo–Asian Ocean, the Liuyuan Ocean, from the Silurian to Late Devonian, which may be an important event in the accretionary history of the Central Asian orogenic belt.

INTRODUCTION

The Central Asian orogenic belt is a huge accretionary orogenic belt (Xiao et al., 2015a) formed via multiple convergence and accretion events of many orogenic components during multiple phases of amalgamation (Xiao et al., 2015b). However, not much work has been done on the southernmost part of this belt. Fortunately, granulites and amphibolites are extensively exposed in the Dunhuang block (the southernmost part of the Central Asian orogenic belt), and they provide a perfect window from which the tectono-metamorphic evolution of the southernmost Central Asian orogenic belt can be deciphered.

Granulites and amphibolites are common metamorphic rocks in orogenic belts, and in many cases, they bear testimony to convergent settings from subduction to collision (Ernst and Liou, 2008; Shen et al., 2014). It is therefore important to uncover the metamorphic evolution of the mafic granulite and amphibolite rather than simply determining the pressure-temperature (P-T) conditions of the rocks (O’Brien and Rötzler, 2003). P-T variations (P-T paths) are usually well documented in the form of metamorphic textures and compositional zoning in metamorphic garnet, pyroxene, and/or plagioclase. Determination of the metamorphic P-T conditions of these rocks and more importantly, linking the absolute timing of metamorphism to the P-T paths are key factors in understanding their tectonic evolution (Harley, 1989; Grujic et al., 2011; Zhao et al., 2010). Deciphering metamorphic textures and pressure-temperature-time (P-T-t) history enables us to better understand the dynamic factors producing the corresponding P-T conditions, the duration of these conditions, and the subsequent exhumation process (e.g., Harley, 1989; O’Brien and Rötzler, 2003; Ambrose et al., 2015; Kuiper et al., 2015; Likhanov et al., 2015).

High-pressure mafic granulite is a high-grade metamorphic product and is known for its distinctive orthopyroxene-free mineral assemblage of garnet + clinopyroxene + plagioclase + quartz (e.g., Green and Ringwood, 1967; O’Brien and Rötzler, 2003; Rushmer, 1993; Yardley, 1989). In contrast, medium-pressure mafic granulite is formed under lower pressures beneath the “orthopyroxene-out” curve (e.g., Pattison, 2003), and it is defined by the diagnostic assemblage of garnet + orthopyroxene + clinopyroxene + plagioclase + quartz (e.g., Green and Ringwood, 1967; Shen et al., 2014; Yardley, 1989). Granulites are often generated via prograde development from a hornblende + plagioclase + quartz ± garnet–bearing precursor amphibolite with “orthopyroxene-in” marking the transition (O’Brien and Rötzler, 2003; Pattison, 2003).

The Dunhuang block is situated in the southernmost Central Asian orogenic belt at the eastern end of the Tarim craton, northwest China (Fig. 1). Rocks with various ages and metamorphic grades are exposed in discrete blocks of the Dunhuang block (He et al., 2014; Long et al., 2014; Meng et al., 2011; Peng et al., 2014; Wang et al., 2013a, 2013b, 2014; Yu et al., 2014; Zhang et al., 2012, 2013; Zong et al., 2012, 2013). Circa 1.85 Ga and ca. 0.44 Ga high-pressure mafic granulites have been found in the Shuixiakou block (Zhang et al., 2012) and Mogutai block (Zong et al., 2012), respectively (Fig. 2A). However, most of the other blocks (Qingshigou, Mt. Dahongshan, Mt. Mingshashan, Mt. Dongbatu) remain unexplored.

Here, we report new findings concerning the metamorphism of the high- and medium-pressure mafic granulites and amphibolites in the Qingshigou area (Fig. 2B) of southwestern Dunhuang. Detailed examinations of metamorphic reaction textures and mineral assemblages of the metabasites were performed to identify metamorphic stages. Thermobarometric evaluations and secondary ion mass spectrometry (SIMS) geochronologic dating were then conducted to determine the P-T conditions of each stage and establish the metamorphic P-T-t paths, which together provide the key factors to understanding the tectono-metamorphic evolution.

GEOLOGICAL BACKGROUND

The Dunhuang block, located between the Beishan orogen to the north and Altyn Tagh fault to the south, is a main terrane in which the metamorphic basement of the Tarim craton is exposed (Fig. 1; Lu et al., 2008). The Dunhuang metamorphic complex consists of a series of medium- to high-grade metamorphosed supracrustal rocks (Dunhuang Group) and subordinate granitoid gneiss (Lu et al., 2008; Mei et al., 1997). The metamorphosed supracrustal rocks are dominated by sillimanite/kyanite-bearing metapelite, mafic granulite, amphibolite, and marble (Lu et al., 2008; Zhang et al., 2013; Zhao and Cawood, 2012; Zong et al., 2012). The granitoid gneisses include tonalite-trondhjemite-granodiorite (TTG) gneiss, monzogranitic gneiss, and K-feldspar-granitic gneiss (Zhao and Cawood, 2012). The complex is distributed linearly and discretely in a NE-SW–trending belt (Fig. 2A). It is not feasible to conduct geological mapping here or distinguish regional lithologic contact relationships due to strongly mixed macroscopic disorder stacking and extensive coverage by Cenozoic sediments. Nevertheless, the discrete TTG gneiss bodies, garnet-bearing mafic granulites, and amphibolites document vital clues to the tectono-metamorphic history. Mesoarchean–Neoarchean TTG gneisses emplaced from 3.06 to 2.50 Ga (Long et al., 2014; Lu et al., 2008; Mei et al. 1998; Zhang et al., 2013; Zhao et al., 2013, 2015; Zong et al., 2013) have been recently shown to be the oldest lithologic unit in the Dunhuang block, and high-pressure mafic granulites have been described (He et al., 2014; Zhang et al., 2012, 2013; Zong et al., 2012). High-pressure granulites in the Shuixiakou block (Fig. 2A) have been reported as lenses or boudins within TTG gneisses, which were metamorphosed at ca. 1.85 Ga and record clockwise P-T paths with isothermal decompression indicative of a possible Paleoproterozoic orogenic event (Zhang et al., 2012, 2013). Other exposures in the Mogutai block (Fig. 2A) occur as enclaves within kyanite-bearing metapelites or marbles (He et al., 2014; Zong et al., 2012). By contrast, a tight clockwise P-T trajectory has been retrieved from these high-pressure granulites, and their metamorphic age (ca. 0.43 Ga) is much younger, implying a Paleozoic tectono-metamorphic imprint in the Dunhuang block. Multiple episodes of magmatic activity have also been recognized, among which mafic magmatism at ca. 1.8 Ga and ca. 1.6 Ga was characterized by island-arc basalt and ocean-island basalt, respectively. The Paleoproterozoic ages are considered as the formation ages of the garnet-bearing metabasites in the Dunhuang block (Wang et al., 2014). The recently revealed later thermal activities are represented by a series of Paleozoic granitic-granodioritic intrusions (Wang et al., 2014; Zhang et al., 2009; Zhu et al., 2014), supporting a Paleozoic tectono-thermal event possibly associated with an orogenic process. Due to the prolonged and complex metamorphic-thermal history, the metamorphic era, tectonic architecture, and affiliation of the Dunhuang block are hotly debated (He et al., 2014; Long et al., 2014; Zhang et al., 2012, 2013; Zong et al., 2012, 2013).

The mafic granulite and amphibolite in the Qingshigou area mostly occur as lenses, boudins, or interlayers preserved in highly deformed pelitic gneisses or marbles. The high-pressure mafic granulite (G104) outcrops as an 80 m × 60 m mass intruded by a garnet-bearing granite body (Figs. 3A and 3B) and exhibits gneissic, locally massive structure (Fig. 3C). It is also characterized by a “white-eye socket” texture (Ma and Wang, 1994), showing garnet porphyroblasts rimmed in narrow white coronas (Fig. 3C). In addition, clinopyroxenes are observed (Fig. 3D). The medium-pressure mafic granulite (WH65) occurs as lenses or boudins within the metapelite, i.e., in garnet-sillimanite gneiss or garnet–mica schist (Figs. 3E and 3F). The amphibolites (WH77, G98) are thin layers intercalated with the metapelites (Figs. 3G and 3H), making it difficult to draw a sharp line between the two. High-pressure mafic granulite sample G104, medium-pressure mafic granulite sample WH65, and amphibolite sample G98 were collected in Qingshigou, whereas the amphibolite sample WH77 was collected in the Wutonggou valley (Fig. 2B). All the samples were collected from the Qingshigou area. A few granitoid plutons (Fig. 2B) intrude the Qingshigou area, amongst which one dioritic body has been dated at 335 ± 2 Ma (Zhu et al., 2014). Some later mafic dikes are locally observed cutting across the amphibolite (Fig. 3I).

METAMORPHIC REACTIONS AND METAMORPHIC STAGES

High-Pressure Mafic Granulite

The term “metamorphic stage” (e.g., M1, M2, and M3) refers to the sequential metamorphic episodes of a single metamorphic event. Micropetrographic study of the high-pressure mafic granulite sample G104 identified three generations of mineral assemblages, i.e., the prograde stage (M1), the peak stage (M2), and the retrograde stage (M3). Detailed petrographic features, metamorphic stages, and reactions are depicted as follows, with mineral abbreviations after Whitney and Evans (2010).

Prograde Metamorphic Stage (M1)

The earliest mineral assemblage (M1) in the high-pressure mafic granulite occurs as fine-grained, randomly distributed inclusion minerals in the garnet porphyroblasts. They consist of hornblende (Hbl1) + plagioclase (Pl1) + quartz (Qz1) (Fig. 4A) as an amphibolite facies precursor. Coarse or irregular mineral grains randomly growing along cracks in garnet were excluded because these are probably later minerals formed by retrogression or fluid alteration.

Peak Metamorphic Stage (M2)

The coarse-grained garnet porphyroblasts and matrix minerals, i.e., ∼15% garnet porphyroblast (Grt2) + ∼20% clinopyroxene (Cpx2) + ∼25% plagioclase (Pl2) + ∼20% quartz (Qz2), as well as minor zircon and titanite, constitute the peak metamorphic mineral assemblage (M2) of the high-pressure mafic granulite (Figs. 4A–4C). “Matrix-like” hornblendes make up ∼20% of the mode, most of which are retrogression products (see next section). Titanite, zircon, rutile, and apatite exist as accessory minerals. Possible metamorphic reactions involved in the transformation of the M1 to M2 assemblages are: 
graphic

Retrograde Metamorphic Stage (M3)

The retrograde, gray-white symplectic assemblages (M3) form fine-grained, worm-like mineral intergrowths rimming the garnet porphyroblasts (Fig. 4A). Such symplectic mineral assemblages consist of hornblende (Hbl3) + plagioclase (Pl3) + quartz (Qz3) ± magnetite (Mag3). Essentially, the coronitic symplectite was produced in garnet rims and the adjacent matrix minerals and highlights a decompression process. Such coronitic symplectite around garnet was given the vivid title “white-eye socket” texture (Ma and Wang, 1994) in the Chinese literature. Another retrogression phenomenon is the replacement of clinopyroxene by hornblende at the edge of clinopyroxene (Fig. 4B). Extensive retrogression also produced abundant coarse-grained hornblendes. Possible metamorphic reactions dominating the retrogressions are thought to be as follows: 
graphic

Medium-Pressure Mafic Granulite

In the representative medium-pressure mafic granulite sample WH65, three generations of mineral assemblages are likewise recognized, ascribed to the prograde stage (M1), the peak stage (M2), and the retrograde stage (M3), respectively.

Prograde Metamorphic Stage (M1)

Sparse inclusion minerals, mainly composed of hornblende (Hbl1) + plagioclase (Pl1) + quartz (Qz1) within the garnet porphyroblasts (Fig. 4D), are the earliest assemblage (M1) recognized. The scarcity of the M1 minerals probably reflects a long duration of the peak metamorphism leading to thorough exhaustion of the precursor minerals.

Peak Metamorphic Stage (M2)

The assemblage consisting of ∼10% porphyroblastic garnet and sizable matrix minerals, ∼20% orthopyroxene (Opx2) + ∼20% clinopyroxene (Cpx2) + ∼15% plagioclase (Pl2) + ∼25% hornblende (Hbl2) + ∼10% quartz (Qz2), as well as minor zircon and titanite, defines the peak metamorphic assemblage (M2) of the medium-pressure mafic granulite (Figs. 4D–4F). Hornblende, and/or plagioclase, clinopyroxene, and orthopyroxene in the matrix display ubiquitous ∼120° triple conjunction (Figs. 4E and 4F), suggesting that thermodynamic equilibrium had been approached at the metamorphic peak stage. Ilmenite and zircon are present as accessory minerals. Possible metamorphic reactions responsible for the mineral transformation from the M1 to M2 stages are thought to be as follows: 
graphic

Retrograde Metamorphic Stage (M3)

Similarly, fine-grained and worm-like intergrowths in the “white-eye socket” texture (Fig. 4D) make up the retrograde mineral assemblage (M3). The symplectic assemblage is mainly composed of hornblende (Hbl3) + plagioclase (Pl3) + orthopyroxene (Opx3) + quartz (Qz3). Perhaps as a result of metamorphic fluid, garnet pseudomorphs occasionally formed after thorough decomposition of the porphyroblast to fine-grained symplectites (Fig. 4G). Metamorphic reactions leading to the retrogression are possibly as follows: 
graphic

Garnet-Bearing Amphibolite

Both of the amphibolite samples WH77 and G98 share similar mineral associations and exhibit gneissic structure.

In WH77, three generations of mineral assemblages are recognized (Fig. 4H), i.e., the prograde (M1), the peak (M2), and the retrograde (M3) metamorphic stages. The M1 mineral assemblage consists of plagioclase (Pl1) + hornblende (Hbl1) + quartz (Qz1) + ilmenite (Ilm1) as inclusion minerals within garnet. The M2 mineral assemblage is made up of ∼15% garnet porphyroblast (Grt2) and ∼25% hornblende (Hbl2) + ∼20% biotite (Bt2) + ∼25% plagioclase (Pl2) + ∼15% quartz (Qz2) and minor zircon and titanite as matrix minerals. Magnetite, ilmenite, and allanite are present as accessory minerals. The metamorphic reaction from the M1 to M2 mineral assemblages is speculated to be: 
graphic
The M3 mineral assemblage is characterized by “white-eye socket” symplectite (Fig. 4H) composed of fine-grained hornblende (Hbl3) + biotite (Bt3) + plagioclase (Pl3) + quartz (Qz3) + epidote (Ep3) + magnetite (Mag3). Possible metamorphic reactions forming the “white-eye socket” texture are: 
graphic

In sample G98, only two generations of mineral assemblages are observed, i.e., the prograde (M1) and the peak (M2) metamorphic stages. The M1 mineral assemblage consists of plagioclase (Pl1) + quartz (Qz1) + ilmenite (Ilm1) as inclusion minerals within garnet. The M2 mineral assemblage is composed of ∼15% garnet porphyroblast (Grt2) and the matrix minerals ∼20% hornblende (Hbl2) + ∼25% biotite (Bt2) + ∼20% plagioclase (Pl2) + ∼15% quartz (Qz2) + ∼5% magnetite (Mag2).

On account of the absence of inclusion minerals and retrogression texture, it is not feasible to speculate on the prograde or retrograde metamorphic reactions. The latter also indicates that the amphibolite underwent little or no retrogression.

ANALYTICAL TECHNIQUES

Electron Microprobe Analysis

Compositional analyses of representative minerals, backscattered electron (BSE) imaging, as well as X-ray compositional mapping of typical garnets were conducted on the JOEL SEM JXA 8230 wavelength dispersive scanning electron microprobe (SEM) at the School of Resources and Environmental Engineering, Hefei University of Technology, China. The analytical conditions were 15 kV accelerating voltage, 20 nA beam current, 3–5 µm electron beam diameter, and 10–20 s counting time. Natural and synthetic minerals were utilized as standards, and the ZAF program was employed for matrix corrections. At least three grains of each mineral and three to 60 spots of each grain were probed during the analysis. The representative mean mineral compositions used to estimate P-T conditions are listed in Tables DR1–DR5.1 The ferric iron content of garnet and pyroxene was evaluated by stoichiometric and charge balance criteria (Droop, 1987), while the ferric iron content of hornblende was determined by the method of Holland and Blundy (1994).

SIMS U-Pb Dating of Zircon

Zircons were separated by standard heavy-mineral separation processes, handpicked for final purity, mounted, and polished with zircon standards Plešovice (Sláma et al., 2008) and Qinghu (Li et al., 2009) for SIMS analysis. Transmitted and reflected light micrographs as well as cathodoluminescence (CL) images of all zircons were collected to reveal their internal structures and then select potential analysis spots. The mount was vacuum-coated with high-purity gold prior to SIMS analysis.

Measurements of U, Th, and Pb isotopes were conducted using a CAMECA IMS-1280 SIMS at the Institute of Geology and Geophysics, Chinese Academy of Sciences in Beijing. Detailed instrument description and analytical procedure can be found in Li et al. (2009) and Q.L. Li et al. (2010), and only a brief summary is provided here. The size of the primary O2 ion beam spot was set at ∼20 × 30 µm. Zircon standard Plešovice was tested to calibrate Pb/U ratios, and zircon standard 91500 (Wiedenbeck et al., 1995) was employed to calibrate U and Th concentrations. Measured compositions were corrected for common Pb by nonradiogenic 204Pb. An average of present-day crustal composition (Stacey and Kramers, 1975) was adopted for common Pb, assuming that the common Pb is largely related to surface contamination introduced during sample preparations. The Isoplot/Ex v. 3.75 program (Ludwig, 2012) was applied for data reduction. Uncertainties in individual analyses are reported at the 1s level, and concordia U/Pb (Pb/Pb) mean ages are quoted with 95% confidence. The “207-corrected” ages were calculated by projecting the uncorrected analysis onto concordia from the assumed common 207Pb/206Pb composition. In most cases, however, the amount of common Pb, revealed by monitoring 204Pb, was relatively small and had little influence on the calculated age.

In order to monitor the external uncertainties of SIMS U-Pb dating, zircon standard Qinghu was alternately analyzed as an unknown with other sample zircons in this study. Nine measurements conducted on Qinghu zircons yielded a concordia age of 160 ± 1.6 Ma (Fig. DR1; Table DR6), which is identical to the recommended value of 159.5 ± 0.2 Ma (Li et al., 2013).

MINERAL CHEMISTRY

Garnet

X-ray compositional mapping analyses were carried out for three representative samples (G104, WH65, and WH77; Fig. 5). Rim-core-rim compositional profile analyses (red lines with arrow in Figs. 4A, 4D, 4H, and 4I) were performed on garnet porphyroblasts for all samples. All the garnets are dominated by almandine, grossular with minor pyrope, and spessartine components (Fig. 6; Table DR1). In sample G104, X-ray mapping and chemical profiling of all garnets show typical growth zoning (Figs. 5A and 6A). A distinct fracture divides the whole garnet into a small part and a large one (Fig. 4A). Traverse chemical analyses through the two parts exhibit characteristic prograde and retrograde zoning (Fig. 6A). XMn and Fe# [Fe/(Fe + Mg)] decrease from the core (XMn = 0.062, Fe# = 0.907) to the inner rim (XMn = 0.001, Fe# = 0.716), representing a “bell-shaped” zoning profile formed during the prograde growth of garnet. At the outermost rim, XMn increases to 0.015 and Fe# to 0.767, showing a thin “kick-up” as a consequence of resorption and Fe-Mg re-exchange (diffusion) during the retrograde stage (Spear, 2014; Spear and Florence, 1992). Chemical profile analyses (Fig. 6A) thus show that the small part is essentially a survivor from later extensive retrogression. In other words, it is the small surviving part that actually records the peak metamorphic compositions, implying that the other part (dashed lines in Figs. 4A and 6A) has been consumed completely. Therefore, in sample G104, at least ∼1 mm of the garnet (∼6 mm in original diameter) was decomposed into symplectites during the retrograde metamorphic stage. This phenomenon may partly account for frequent underestimation of the peak metamorphic conditions. Meanwhile, the XFe, XMg, and XCa data display corresponding chemical zoning (Fig. 6A).

In samples WH65, WH77, and G98, the garnets exhibit no meaningful growth zonation (Figs. 6B–6D), suggesting possible homogenization during the peak metamorphic episode, while two garnets (in WH65 and G98) show resorption and Fe-Mg re-exchange during retrogression.

Plagioclase

Plagioclase is present as inclusion (Pl1), matrix (Pl2), and symplectic (Pl3) minerals. Pl1 is chemically heterogeneous, ranging from An18 to An29 in mafic high-pressure granulite (G104), and averaging An81 in mafic medium-pressure granulite (WH65) and An37 in amphibolite (WH77) (Table DR2). BSE images of matrix plagioclases (Pl2) in granulites (G104 and WH65) show heterogeneity in composition (Figs. 4C and 4F). Rim-core-rim traverse compositional analyses (red lines with arrow in Figs. 4C and 4F) on the plagioclases show compositional zoning with CaO increasing and Na2O decreasing from core to rim (Fig. 7). Only the core compositions of Pl2, as An32 in G104 and An53 in WH65, were used to obtain the peak metamorphic conditions of the granulites. No chemical zonation was found in samples WH77 (An29) and G98 (An49). The compositions of symplectic plagioclases (Pl3) also differ in each sample, being An47 in G104, An79 in WH65, and An31 in WH77. A systematic CaO increase from Pl2 to Pl3 suggests that the symplectic plagioclase (Pl3) formed from the breakdown of garnet rim and matrix minerals.

Clinopyroxene

Clinopyroxenes (Cpx2) in the high-pressure (G104) and medium-pressure (WH65) granulites show chemical homogenization. The clinopyroxenes contain low jadeite components (Na2O = 0.26 wt% in the high-pressure granulite and 0.55 wt% in the medium-pressure granulite; Table DR3) and are diagnosed as diopside based on the classification of Morimoto et al. (1988), although a difference exists in Al2O3, FeO, and MgO (respectively, 0.76 wt%, 10.85 wt%, and 11.31 wt% in the high-pressure granulite and 3.63 wt%, 6.77 wt%, and 13.93 wt% in the medium-pressure granulite).

Orthopyroxene

Orthopyroxenes are only observed in the medium-pressure granulite (WH65) as a subeuhedral matrix mineral (Opx2) or an intergrowth mineral in the “white-eye socket” symplectite (Opx3; Figs. 4D–4G). The matrix orthopyroxene (hypersthene, with XMg of 0.68) is slightly richer in FeO and poorer in MgO than the symplectic orthopyroxene (bronzite, with XMg of 0.71; Table DR3).

Hornblende

Hornblendes exist as inclusion (Hbl1), matrix (Hbl2), and symplectite (Hbl3) minerals in the mafic medium-pressure (WH65) granulite and amphibolite (WH77), as inclusion (Hbl1) and symplectite (Hbl3) minerals in mafic high-pressure granulite (G104), and only as matrix minerals (Hbl2) in amphibolite (G98). In samples WH65 and WH77, the FeO contents increase from Hbl1 (8.8–17.23 wt%) to Hbl2 (10.59–22.51 wt%) and then slightly decrease to Hbl3 (9.01–22.49 wt%). All hornblendes contain similar CaO (∼11–12 wt% in Table DR4) and are ascribed to the calcic amphibole group (Leake et al., 1997).

Biotite

Biotites are found as a matrix mineral (Bt2) or as a symplectic mineral in the “white-eye socket” (Bt3) in sample WH77, and only as a matrix mineral in sample G98. Their compositions are listed in Table DR5. In sample WH77, the matrix biotite possesses slightly higher TiO2 (3.0 wt%) and FeO (23.04 wt%) than the symplectite biotite (2.55 wt% and 22.84 wt%, respectively).

METAMORPHIC P-T PATHS

Geothermobarometry

For each of the mineral assemblages of the different metamorphic stages described here, corresponding accurate and precise geothermometers and geobarometers have been applied to determine their metamorphic conditions, and then reconstruct the P-T paths of the metabasites in the Qingshigou area.

To determine the peak metamorphic conditions, (1) the garnet-clinopyroxene (GC) Fe-Mg exchange geothermometer (Ravna, 2000) coupled with the garnet-clinopyroxene-plagioclase-quartz (GCPQ) geobarometer (Eckert et al., 1991) were used for the high-pressure mafic granulite (G104); (2) the GC geothermometer (Powell, 1985; Ravna, 2000) coupled with the GCPQ geobarometer (Eckert et al., 1991), the plagioclase-hornblende (PH) geothermometer (Holland and Blundy, 1994) coupled with the garnet-hornblende-plagioclase-quartz (GHPQ) geobarometer (Dale et al., 2000), and the garnet-orthopyroxene-plagioclase-quartz (GOPQ) geothermobarometers (Bhattacharya et al., 1991; Lal, 1993) were used for the medium-pressure mafic granulite (WH65); and (3) the plagioclase-hornblende geothermometer (Holland and Blundy, 1994) coupled with the GHPQ geobarometer (Kohn and Spear, 1990) and the garnet-biotite (GB) geothermometer (Holdaway, 2000) coupled with the garnet-biotite-plagioclase-quartz (GBPQ) geobarometer (Wu et al., 2004) were applied to the amphibolites (WH77 and G98). In the computation, the inner rim compositions of garnets with the lowest XMn and Fe# values (Table DR1) and the core compositions of plagioclases with the lowest XAn values (Table DR2) were used to estimate the peak metamorphic P-T conditions.

Since the M1 and M3 minerals could hardly be at equilibrium with the garnets (Wu et al., 2014), garnet-free geothermobarometers were applied to these two assemblages for all samples. Accordingly, the PH geothermometer (Holland and Blundy, 1994) combined with the HPQ geobarometer (Bhadra and Bhattacharya, 2007) were used to quantify the P-T conditions of the M1 and M3 assemblages.

Metamorphic P-T Constraints and Metamorphic P-T Paths

The P-T conditions of the three metamorphic stages of the granulites and amphibolites are listed in Table 1.

High-Pressure Mafic Granulite (G104)

The inner rim compositions of garnet (the lowest XMn) and the core compositions of matrix minerals, when compositional zoning was displayed, were adopted to estimate peak metamorphic (M2) P-T conditions. These proved to be 861 °C and 16.9 kbar based on the GCPQ geothermobarometer. Because plagioclases with different compositions are present as inclusions in garnets, the P-T conditions of the M1 assemblage show prograde variation from 568 °C and 8.8 kbar to 607 °C and 10.6 kbar using the HPQ geothermobarometer (Fig. 8A). The M3 assemblage conditions were 598 °C and 4.4 kbar based on the HPQ geothermobarometer.

Medium-Pressure Mafic Granulite (WH65)

In sample WH65, M1 minerals are rarely preserved as inclusions in garnets, possibly reflecting the long duration of the peak metamorphic stage. From another perspective, where M1 minerals survived, the obtained P-T condition of M1 has to approach the peak metamorphic P-T. Application of the HPQ geothermobarometer indicates that the mineral assemblages of M1 and M3 formed at 756 °C and 9.0 kbar and 675 °C and 4.7 kbar, respectively. For the peak metamorphic assemblage, the GCPQ, GHPQ, and two GOPQ geothermobarometers yielded metamorphic peak P-T conditions of 700–750 °C and 10.0 kbar, 749 °C and 10.3 kbar, 874 °C and 11.6 kbar, and 849 °C and 9.3 kbar in order (Fig. 8B).

Garnet-Bearing Amphibolite (WH77)

The HPQ geothermobarometer demonstrates that the M1 and M3 mineral assemblages formed at P-T conditions of 686 °C and 7.6 kbar and 671 °C and 5.6 kbar, respectively. The GHPQ and GBPQ geothermobarometers yielded P-T conditions for the M2 mineral assemblages of 715 °C and 10.6 kbar and 766 °C and 11.2 kbar (Fig. 8C).

Garnet-Bearing Amphibolite (G98)

In sample G98, only the P-T conditions of the M2 mineral assemblage were estimated as 720 °C and 6.8 kbar and 743 °C and 8.3 kbar using the GHPQ and GBPQ geothermobarometers, respectively. Numerous albite inclusion minerals were observed in garnets, but the prograde P-T condition could not be ascertained for lack of hornblende. However, a prograde process is also suggested here by the consumption of the M1 precursor, giving rise to growth of the incumbent M2 (Fig. 8D).

The peak metamorphic P-T conditions of the high-pressure mafic granulite (G104) lie in the high-pressure granulite facies realm (O’Brien and Rötzler, 2003), approaching the lower limit of the high–P-T facies series (Fig. 8A; Miyashiro, 1961) according to the facies scheme of Spear (1993, p. 19). Meanwhile, the peak metamorphic conditions of the medium-pressure mafic granulite (WH65) and the amphibolite (WH77) plot between the high– and intermediate–P-T facies series (Figs. 8B and 8C). The metamorphic P-T paths established from the three samples form tight clockwise loops (Fig. 8E) similar to the “Franciscan-type” P-T path (Ernst, 1988), which is generally accepted to be the result of subduction and subsequent exhumation processes.

SIMS U-Pb DATING RESULTS OF ZIRCON

Four samples were dated in this study. Results of SIMS analyses are listed in Table DR8. In sample G104, zircons range from 70 to 140 µm along the longest axis and are stubby to ovoid in shape (Fig. 9A), which is a characteristic morphology of high-grade metamorphic zircons (Hoskin and Black, 2000; Hoskin and Schaltegger, 2003). A few grains show dark-bright zoning or “core-rim–like” structure in CL images. However, U-Pb analyses show richer U in dark domains but no apparent distinction in ages (Table DR8). The dark-bright zoning may result from heterogeneous U migration during peak metamorphism. Twenty-two analyzed spots on 20 zircon grains yielded a lower-intercept age of 412.2 ± 6.8 Ma (mean square of weighted deviates [MSWD] = 2.7) in the concordia Tera-Wasserburg diagram (Fig. 10A). Their “207-corrected” ages range from 432.1 ± 8.0 Ma to 388.2 ± 8.4 Ma, with the weighted mean of 409.3 ± 5.8 Ma (MSWD = 2.6; Fig. 10B). Consistent within the error bar, the lower-intercept and “207-corrected” ages place valid constraints on the timing of peak metamorphism.

In sample WH65, zircons display ovoid shapes with lengths of 80–150 µm. Morphologically, the zircons can be divided into two categories: (1) Over 80% of the grains exhibit bright CL with slight diffuse zonation (Fig. 9B) as features of high-grade metamorphic zircons; and (2) ∼20% of the grains, in contrast, show dark CL with relatively homogeneous internal structure (Fig. 9B). Seventeen spot analyses were done on 17 grains. Eleven concordant 206Pb/238U dates from bright grains show a wide range from 374.3 ± 5.7 Ma to 437.8 ± 8.6 Ma, whereas six dates from dark grains show a quite narrow range between 248.0 ± 3.8 Ma and 251.7 ± 3.7 Ma, with a weighted mean of 249.8 ± 3 Ma (MSWD = 0.19; Table DR8; Figs. 10C and 10D). Notably, dark zircon grains are much more U-rich than bright ones. The first age group is interpreted as reflecting the long duration of the regional metamorphism, while the second is viewed as a record of later thermal perturbation (see next section for details).

In sample WH77, zircon grains differ in shape from round to ovoid to irregular, ranging from 100 to 300 µm in length (Fig. 9C). Broadly, three types can be observed based on structural morphology: (1) A few grains show bright CL with comparatively homogeneous internal structure or faint zoning as characteristics of metamorphic zircons; (2) other grains contain dark, irregular cores, commonly truncated by bright narrow metamorphic rims, and some cores that show oscillatory zoning indicative of a magmatic origin; and (3) one zircon exhibits a highly irregular shape with the appearance of melting erosion (the last one in Fig. 9C). Twenty spots were analyzed on 18 grains. Thirteen concordant 206Pb/238U dates from the bright homogeneous grains and narrow rims range from 384.4 ± 5.9 Ma to 415.1 ± 6 Ma, with a weighted mean of 400 ± 5.7 Ma (MSWD = 2.4; Table DR8; Figs. 10E and 10F), defining the metamorphic age. Six core U/Pb dates define a discordant curve with an upper intercept at 1782 ± 14 Ma, which is interpreted as the protolith or xenocrystic age, and a lower intercept at 424 ± 16 Ma, which is interpreted as close to the time of metamorphism (Fig. 10E). A single spot analysis on the melting erosion–like zircon produced a concordant 206Pb/238U age of 325.1 ± 4.9 Ma, suggesting a later thermal event (see next section for details).

In sample G98, zircons are irregularly oblong in shape and range from 80 to 200 µm in length (Fig. 9D). Most zircons show bright CL with faint band or sector zoning, indicative of a metamorphic origin. A few zircons contain dark, irregular cores with blurred oscillatory zoning, suggesting a magmatic origin dated at 1566.9 ± 7.9 Ma. Twenty spots were analyzed on 19 grains. Nineteen 206Pb/238U dates range from 355.8 ± 7.5 Ma to 380.9 ± 8 Ma and produce a weighted mean age of 365.1 ± 4.3 Ma (MSWD = 1.2; Table DR8; Figs. 10G and 10H). The latter validly constrains the metamorphic event. Three spot analyses on the cores give three much older and decentralized U/Pb ages (Fig. 10G), showing inheritance properties.

DISCUSSION

Geochronology and Metamorphic Time

The discovery of high-pressure mafic granulites in the Mogutai block (He et al., 2014; Zong et al., 2012) showed that the Dunhuang block had undergone high-pressure metamorphism in the Silurian (ca. 430 Ma). In this study of newly found mafic high-pressure and medium-pressure granulites and amphibolites, most zircon grains or narrow rims show zircon morphologies typical of growth under high-grade metamorphic conditions (Hoskin and Black, 2000; Hoskin and Schaltegger, 2003). The results of high-resolution SIMS dating place firm geochronologic constraints on the Silurian–Devonian evolution of the Dunhuang block. At the same time, certain high-grade metabasites in the Qingshigou area preserve younger geochronological information. The disparate and young 206Pb/238U ages of ca. 249 Ma and 325 Ma, recorded by quite a few zircons in the medium-pressure mafic granulite (WH65) and amphibolite (WH77), have no significance for deciphering metamorphism. An upper-amphibolite to granulite facies metamorphic event is unlikely to be recorded by only one or two pieces of rock in a specific terrane. Indeed, evidence of the Silurian–Devonian metamorphic event is widely indicated by TTG gneiss in the Mogutai block (Zong et al., 2013), amphibolite in the Hongliuxia block (Wang et al., 2014), and metabasite in the Qingshigou area (this study), although the younger 206Pb/238U ages imply local thermal perturbations on mafic granulite or amphibolite in a later event. In fact, intrusion of numerous granitoids and mafic dikes is observed in the Dunhuang metamorphic complex (e.g., Figs. 2, 3A, 3B, and I3I). Zhu et al. (2014) dated a dioritic intrusive in the Qingshigou area (Fig. 2B) with 206Pb/238U ages ranging from 330 ± 5 Ma to 342 ± 4 Ma. The age of 325.1 ± 4.9 Ma recorded by a single zircon in the amphibolite WH77 could possibly record this granitoid event. Our zircon laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) dating results of two representative mafic dikes in the Dunhuang block show the youngest 206Pb/238U ages of 286 ± 9 Ma and 250 ± 8 Ma (Fig. DR2; Table DR7), providing a lower limit of their crystallization ages. The age of 249.8 ± 3 Ma documented by zircons in the medium-pressure mafic granulite WH65 probably relates to this mafic magmatic event.

Metamorphic P-T Paths and Tectonic Implication

Dynamically, P-T paths characteristic of high-grade metamorphic rocks link small-scale petrology to large-scale tectonic events (Spear, 1993, p. 766) and pave the way toward reconstructing the tectonic history (e.g., Bohlen, 1987, 1991; Brown, 1993; Ernst, 1988; England and Thompson, 1984; Harley, 1989; Spear, 1992, 1993; Thompson and England, 1984; Zhao, 2014). The western-Alpine–type and Franciscan-type paths make up the subdivision of clockwise P-T paths (Ernst, 1988). The most striking distinctions in subduction P-T histories lie in the retrograde paths and are attributed to fundamental differences in the convergent tectonics (Spear, 1993, p. 739). The western-Alpine type is characterized by nearly isothermal decompression, suggesting tectonic unroofing or rapid denudation caused by the dynamic changes from subduction of oceanic crust to continental collision (Ernst, 1988). The Franciscan-type retrograde P-T path is characterized by a return path that generally retraces the prograde P-T path (Ernst, 1988), suggesting subduction zone metamorphism. Cloos and Shreve (1988) proposed the concept of a subduction channel to explain the ascent of materials trapped between the upper and lower plates in a subduction zone. In a review of exhumation mechanisms, Teyssier (2011) showed that the transition of particles from downgoing to ascending is essentially triggered by continuous filling of the subduction channel with oceanic and continental margin material, which results in weakening at the roof of the system and facilitates exhumation to the roof.

High-pressure granulites require sufficient tectonic burial, generally to mantle depth, along relatively high dT/dP gradients (O’Brien and Rötzler, 2003). In this study, the peak metamorphic P-T conditions of the high-pressure mafic granulite plotted in the eclogite facies region (O’Brien and Rötzler, 2003; Spear, 1993, p. 19), and the dP/dT gradient was found to be ∼14 °C/km, approaching the high–P-T facies series (Miyashiro, 1961). In most Phanerozoic orogens in China, only high-pressure granulite (occasionally coexisting with eclogite) is found (Shen et al., 2014). However, there are one or two exceptions. For example, several types of granulites, including low- and high-pressure pelitic granulite, medium- to high-pressure mafic granulites, high-pressure felsic granulite, and (ultra)high-temperature pelitic granulite, have been exhumed in the Altay orogen (e.g., Chen et al., 2006; Li et al., 2014; Tong et al., 2014a, 2014b; Wang et al., 2009; Wei et al., 2007), part of the accretionary Central Asian orogenic belt. High-pressure granulites in these Phanerozoic orogens are fundamentally products of oceanic subduction, and not continental collision, because their country rocks are mostly ophiolite suite rocks or ophiolitic mélange (Shen et al., 2014, and references therein). Numerical models (e.g., Gerya et al., 2002; Gerya and Stöckhert, 2006; Z.H. Li et al., 2010) show that differences in subduction depth, duration, and exhumation route could generate diverse types of high-grade metamorphic rocks with different patterns of P-T paths in the subduction channel.

The nearby Beishan orogenic collage (the southern segment of the Central Asian orogenic belt exposed to the north of the Dunhuang block) is believed to have been formed by subduction-accretion of several microcontinents, arcs, and accretionary complexes during the Paleozoic (Şengör et al., 1993; Windley et al., 2007; Xiao et al., 2009). The Liuyuan ophiolitic mélange (Xiao et al., 2010) in the Beishan orogen indicates a precursor of the Liuyuan Ocean (Xiao et al., 2010; Mao et al., 2012a, 2012b) to be the terminal branch of the Paleo–Asian Ocean. Its closure time is constrained to be as late as end-Permian to Early Triassic (Tian et al., 2013, 2015), which is much later than the high-pressure metamorphism in the Dunhuang block. Northward subduction of oceanic crust generated eclogites at ca. 0.47 Ga in the Gubaoquan area (Liu et al., 2011; Qu et al., 2011). The younger Paleozoic metamorphic age reported here from the Dunhuang block is rare in the Beishan orogen, and thus we interpret this metamorphic event to be due to southward subduction (e.g., Liu et al., 2011; Qu et al., 2011; Xiao et al., 2010) of the oceanic crust. Our new zircon ages for the high- and medium-pressure mafic granulites and garnet-bearing amphibolites in the Qingshigou area range from ca. 412 Ma to ca. 365 Ma. Combined with the geochronologic constraints of the high-pressure mafic granulites in the Mogutai block (ca. 430 Ma; He et al., 2014; Zong et al., 2012), we consider that southward subduction of the Liuyuan Ocean began in the Silurian with the following evolution (Fig. 11): (1) Before ca. 430 Ma, the northern margin of the Dunhuang block was a passive continental margin and developed a series of clastic and carbonate sequences; (2) ca. 430–410 Ma, high- and medium-pressure mafic granulites formed at distinct depths due to the southward subduction of the crust of the Liuyuan Ocean, while some basites were successively diving and undergoing prograde metamorphism; (3) ca. 400 Ma, the high-pressure granulites began to back up, while the medium-pressure granulites stayed at depth, possibly due to a lag of exhumation flow, and the amphibolites formed at ∼30 km; (4) ca. 365 Ma, the high-pressure and medium-pressure mafic granulites and the amphibolites were exhumed to a shallower level and underwent retrogression. Meanwhile, some basites underwent short-lived subduction and were stranded at a shallow depth.

CONCLUSIONS

(1) Three generations of metamorphic mineral assemblages are recognized in the high-pressure and medium-pressure mafic granulites and garnet-bearing amphibolite from the Qingshigou area, southwestern Dunhuang block. The metamorphic peak of the high-pressure mafic granulite lies in the eclogite facies region, approaching the high–P-T metamorphic facies series. The derived metamorphic P-T paths show Franciscan-type clockwise patterns, passing through 568 °C and 8.8 kbar, 607 °C and 10.6 kbar, 861 °C and 16.9 kbar, and finally to 598 °C and 4.4 kbar for the high-pressure granulite; from 756 °C and 9.0 kbar through 750–874 °C and 9.3–11.6 kbar to 675 °C and 4.7 kbar for the medium-pressure granulite; and from 686 °C and 7.6 kbar through 715–766 °C and 10.6–11.2 kbar to 671 °C and 5.6 kbar for the garnet-bearing amphibolite.

(2) High-resolution SIMS U-Pb dating of metamorphic zircons from the metabasites in the Dunhuang block indicates a metamorphic event from ca. 430 Ma to ca. 365 Ma, lasting at least ∼65 m.y.

(3) We interpret the high-grade granulite and amphibolite in the Dunhuang metamorphic complex to record the southward subduction of the Liuyuan Ocean in the Silurian–Devonian.

We sincerely thank Professors Xian-Hua Li and Qiu-Li Li for guidance in the secondary ion mass spectrometry experiments. We benefited from discussions with Kai-Jun Zhang and Ling-Ling Xiao. Reviews by the three anonymous referees and the editorial review by R. Damian Nance have greatly improved both the science and the English of the original manuscript. This work was supported by the National Natural Science Foundation of China (41225007, 41372199) and the State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences (Kai201605).

1GSA Data Repository Item 2016254, Tables DR1–DR8 and Figures DR1 and DR2, is available at www.geosociety.org/pubs/ft2016.htm, or on request from editing@geosociety.org.