Introduction

Suprasubduction zones, commonly referred to as “subduction factories”, play an important role in plate tectonics, melt generation and crustal evolution. The “subduction factory” is a metaphor for the geochemical interaction between four possible components, i.e., the subducted slab, mantle wedge, asthenospheric mantle and overlying crust (Defant and Drummond 1990; Hussain et al. 2004; McCulloch and Gamble 1991; Omrani et al. 2008; Stern and Kilian 1996; Tatsumi and Kogiso 2003). This multi-source complex environment commonly leads to widespread magma mixing in subduction zones, such as that at Mount Shasta in northern California (Streck et al. 2007). However, most workers have focused on specific subduction-related rocks, such as adakite, high-magnesium andesite and arc lavas, when studying magma mixing in subduction zones (Anderson 1976; Debaille et al. 2006; Pal et al. 2007; Streck et al. 2007), ignoring the significance of mafic microgranular enclaves (MMEs) in subduction zone magmatism. MMEs are common in calc-alkaline granitoids and can provide important information on the nature of the magma source and the generation of the granitic melt. They are valuable means for studying mixing of crustal and mantle magmas (Arvin et al. 2004; Castro et al. 2006; Ma et al. 1992; Qin et al. 2010; Yang et al. 2007a), but little work in this respect has been done on subduction-related granitoids and their enclaves (e.g. Parada et al. 1999), and their geochemical characteristics and genesis are still poorly understood.

In the Northern Qinghai-Tibet Plateau, owing to the northeastward subduction of Paleo-Tethyan oceanic lithosphere beneath East Kunlun, voluminous Late Permian-Middle Triassic (260–245 Ma) arc magmatism occurred (Harris et al. 1988; Mock et al. 1999; Yang et al. 1996), which is recorded by calc-alkaline granitoids. These granitoids contain abundant MMEs, making East Kunlun a natural laboratory for the study of magma mixing in subduction zone environments. Although numerous studies have been carried out on the MMEs in East Kunlun, all focused exclusively on those found in syn-collisional (245–220 Ma) rocks (Liu et al. 2002; Liu et al. 2004; Chen et al. 2005). Thus, little is known about the origin of the East Kunlun subduction-related magmatic rocks and their MMEs, and the role that these rocks played in the evolution of Paleo-Tethys in the Qinghai-Tibet Plateau. Here, we report the results of a detailed study of the subduction-related MMEs in the East Kunlun orogenic belt, using zircon U-Pb geochronology, petrography, mineralogy, whole-rock geochemistry and Sr-Nd-Hf isotope geochemistry.

Geological setting

The East Kunlun orogenic belt, located in the Northern Qinghai-Tibet Plateau, is bounded by the Qaidam basin to the north, the Bayan Har-Songpan Ganzi block to the south, the Qinling-Dabie orogenic belt to the east and the NE-trending Altyn Tagh Fault to the west (Fig. 1a). The East Kunlun orogenic belt is part of the Paleo-Tethyan tectonic domain. The A’nyemaqen part of Paleo-Tethys began to subduct beneath the Kunlun terrane in the Late Permian, and the ocean finally closed in the early-middle Triassic, forming the A’nyemaqen Paleo-Tethys suture zone (Fig. 1a). Due to subduction of the Paleo-Tethyan lithosphere and subsequent collision, the East Kunlun orogenic belt was a region of large-scale, late Permian to late Triassic magmatism (Chen and Wang 1996; Konstantinovskaia et al. 2003; Yang et al. 1996; Yang et al. 2009).

Fig. 1
figure 1

a Tectonic outline of the Tibetan Plateau showing the study area (after Roger et al. 2008); b Simplified geological map of Naomuhun pluton (after Wang et al. 2003)

Full size image

In this study, we focus on the Naomuhun (NMH) pluton, located in the eastern section of East Kunlun (Fig. 1a), about 10 km south of Balong Township. The NMH pluton intruded four basement complexes: (1) the Paleoproterozoic Baishahe Group, consisting of banded amphibolite, mica plagioclase gneiss, granulite, banded marble, mica schist and mica quartz schist; (2) the Middle Proterozoic Xiaomiao Group, composed mainly of quartzite, mica quartz schist, biotite plagioclase gneiss and marble; (3) the Neoproterozoic Wanbaogou Group, which contains various combinations of clastic rocks, volcanic rocks and carbonates; and (4) the Lower Paleozoic Nachitai Group, which consists of volcanic lava, tuff, tuffaceous phyllite, slate, sandstone, schist, and carbonates. The pluton crops out discontinuously for about 50 km in an E-W direction over an area of ~364 km2 (Fig. 1b). It consists mainly of biotite granodiorite with abundant MMEs.

Sampling and petrography

One sample each of the granodiorite (BL13) and the mafic enclaves (BL13-1) were collected for zircon dating at N35°49.665′, E97°24.639′ (Fig. 1b).

The medium-grained granodiorite consists of plagioclase (40–45 vol.%), quartz (20–25 vol.%), potassium feldspar (10–15 vol.%), biotite (5 vol.%) and amphibole (5 vol.%). The accessory minerals are titanite, apatite, zircon, epidote and opaque minerals.

The MMEs form dense swarms and range from submillimeter up to a meter in size and are they angular to ovoid in shape (Fig. 2a). Most of the MMEs are texturally and mineralogically similar to their host rocks, consisting of plagioclase (45–50 vol.%), biotite (15–20 vol.%), euhedral amphibole (5–10 vol.%), anhedral quartz (10–15 vol.%) and <5% accessory minerals, including acicular apatite, zircon, and sphene. Many of the MMEs contain small, subhedral feldspar xenocrysts (Fig. 2b).

Fig. 2
figure 2

Field and thin section photographs of various disequilibrium textures in the NMH mafic microgranular enclaves (MMEs). a Enclave swarms in the NMH pluton; b Plagioclase xenocryst in the MME; (c) Oscillatory zoned plagioclase; d Plagioclase xenocryst rimmed by biotite and hornblende; e subhedral hornblende with inclusions of plagioclase and biotite; f subhedral hornblende with inclusions of plagioclase

Full size image

The MMEs contain disequilibrium textures, such as plagioclase with oscillatory zoning (Fig. 2c), poikilitic plagioclase rimmed by mafic minerals (Fig. 2d), and euhedral hornblende with inclusions of plagioclase and biotite (Fig. 2e and f).

Analytical methods

The samples for dating were collected from fresh outcrops, and the zircons were separated by heavy liquids and using a Frantz magnetic separator. Before analysis, the zircon grains were photographed with an optical microscope, and the internal structures of the polished grains were observed in cathodoluminescence (CL).

The U-Pb dating of zircon was done by LA-ICP-MS at the State Key Laboratory of Geological Processes and Mineral Resources (GPMR), China University of Geosciences, Wuhan. Laser sampling was performed using a GeoLas 2005 System with a spot size of 32 μm. An Agilent 7500a ICP-MS instrument was used to acquire ion-signal intensities. Nitrogen was added into the central gas flow (Ar + He) of the Ar plasma to increase sensitivity and improve precision (Hu et al. 2008). Off-line selection and integration of background and analyte signals, and time-drift correction and quantitative calibration for U-Pb dating were performed by the software ICPMSDataCal (Liu et al. 2008). Concordia diagrams and weighted mean calculations were made using Isoplot/Ex_ver3 (Ludwig 2003). Besides, the data less than 95% concordance were completely rejected. Detailed operating conditions for the laser ablation system and the ICP-MS instrument were the same as described by Zong et al. (2010).

In order to check the data quality, two reference materials, i.e., zircons 91500 and GJ-1, were analysed and yielded weighted mean 206Pb/238U ages of 1062 ± 3 Ma (n = 24, 2σ) and 599 ± 4 Ma (n = 8, 2σ), respectively, which are in good agreement with the recommended ages of 1065 Ma (Wiedenbeck et al. 1995) and 600 Ma (Jackson et al. 2004). Reported uncertainties were propagated by quadratic addition of the external reproducibility obtained from the standard zircon GJ-1 during individual analytical sessions and the within-run precision of each analysis.

Zircon Hf isotopic analysis was performed on the dated zircon grains using an excimer (193 nm wavelength) laser ablation inductively coupled plasma mass spectrometer (LA-MC-ICP-MS) system at the State Key Laboratory of Continental Dynamics, Northwest University, Xi’an, China. A beam diameter of 44 μm was used in this study with a laser pulse frequency of 10 Hz, and the analysis was placed close to the U-Pb dating spot, either overlapping the dating spot or into the same zone (Figs. 3a and 4a). The 176Lu/175Lu ratio of 0.02669 and 176Lu/172Yb ratio of 0.5886 were used to calculate 176Lu/177Hf and 176Hf/177Hf (Bievre and Taylor 1993; Chu et al. 2002). Zircon 91500, GJ-1 and MON-1 were used as the reference for calibration and controlling the condition of analytical instrumentation, which yielded weighted mean 176Hf/177Hf ratios of 0.282299 ± 0.000021 (n = 9, 2σ), 0.282012 ± 0.000029 (n = 9, 2σ) and 0.282742 ± 0.000007 (n = 9, 2σ), respectively. These results are in good agreement with the recommended 176Hf/177Hf ratios of 0.282296, 0.282015 and 0.282739 for the three reference materials (Elhlou et al. 2006; Woodhead and Hergt 2005). The detailed analytical technique for this method was described by Yuan et al. (2008).

Fig. 3
figure 3

a Representative CL images of zircons (solid circles and dashed circles represent the points for U-Pb and Lu-Hf isotope analysis, respectively) and b zircon U-Pb concordia diagram from NMH granodiorite (sample BL13)

Full size image
Fig. 4
figure 4

a Representative CL images of zircons (the solid circles and dashed circles represent the points for U-Pb and Lu-Hf isotope analysis, respectively) and b zircon U-Pb concordia diagram from NMH MMEs (sample BL13-1)

Full size image

Mineral compositions were analyzed using a JEOL-JXA-8100 electron microprobe at GPMR, China University of Geosciences, Wuhan, using the analytical procedures described by Zheng et al. (2009). Whole-rock major element analyses were carried out using a Rigaku 3080E1-type XRF spectrometer at the Bureau of Geology and Mineral Resources, Hubei Province, China, with analytical precision better than 5%. Trace element, including REE and Sr-Nd isotope analyses were conducted by LA-ICP-MS and Finnigan Triton thermo-ion mass isotope spectrometer at the GPMR Laboratory, respectively. USGS standards AGV-2, BHVO-2 and BCR-2 were used for trace elements calibration (Liu et al. 2008). The preferred values of trace element concentrations for the USGS reference glasses are from the GeoReM database (http://georem.mpch-mainz.gwdg.de/). The analytical accuracies were better than 5% for Co, Ni, Ga, Rb, Y, Zr, Nb, Hf, Ta and light REE, and 5–12% for the remaining elements. The mass fractionation corrections for Sr and Nd isotopic ratios were based on 86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219, respectively. The average 143Nd/144Nd ratio of the La Jolla standard measured during the sample runs is 0.511862 ± 5 (2σ), whereas the NBS-987 standard gave 87Sr/86Sr as 0.710236 ± 16 (2σ). The analytical instruments and detailed operating conditions for the trace element and Sr-Nd isotope analyses are the same as those described by Gao et al. (2004).

Results

Zircon U-Pb ages

The zircons selected for analysis from the granodiorite and its MMEs are transparent colorless to pale yellow, euhedral columnar crystals without inclusions. The grains are 80 to 120 μm long, with width/length ratios of about 1:2–5. All of the zircons show typical oscillatory zoning in the CL images (Fig. 3a and 4a), indicating that they are of magmatic origin (e.g. Corfu et al. 2003; Hoskin and Schaltegger 2003; Wu and Zheng 2004).

The results of U-Pb dating of zircon from the granodiorite (sample BL13) and MMEs (sample BL13-1) are listed in Supplementary Table 1. The zircons have high concentrations of Th and U (average, 456 and 511 ppm, respectively) and Th/U ratios (0.5–1.5). A total of 19 concordant U-Pb analyses were obtained from BL13, with 206Pb/238U ages ranging from 253 ± 3 to 266 ± 3 Ma and yielding a weighted mean 206Pb/238U age of 261 ± 2 Ma with an MSWD of 1.4 (confidence intervals are always reported at 95% confidence level; Fig. 3b). We interpret the weighted mean age as the emplacement age of the NMH granodiorites, which formed close to the Middle-Upper Permian boundary.

Zircon ages of the MME (BL13-1) are concordant within analytical uncertainties, with 206Pb/238U ages ranging from 256 ± 3 to 268 ± 3 Ma (1σ, Supplementary Table 1). Nineteen analyses provided a weighted average age of 263 ± 2 Ma (MSWD = 1.9, 95% confidence limit; Fig. 4b), which is within analytical uncertainty identical to the age of the granodiorite. This suggests that the granodiorites and MMEs formed at the same time, and that they are petrogenetically linked.

Mineral chemistry

A careful selection of plagioclase, K-feldspar, biotite and amphibole was done under optical microscope prior the electron microprobe analysis, in which the Fe2+ and Fe3+ values of hornblende and biotite were calibrated after Lin and Peng (1994).

Feldspar

Compositions of plagioclase and K-feldspar from the granodiorites and MMEs are given in Supplementary Table 2. Plagioclase is of andesine composition with An30-An49 in the MMEs and An30-An44 in the host granodiorites (Fig. 5; Supplementary Table 2). Likewise, the K-feldspar in both the MMEs and granodiorites is orthoclase, with Or component between 91–95% (Supplementary Table 2). Zoned-plagioclase in the MMEs is slightly more anorthitic in the core than in the rim, whereas in the granodiorites the rims are typically richer in anorthite than the cores (Supplementary Table 2). This reverse zoning of plagioclase in the granodiorite could be interpreted as evidence of mixing between silicic and mafic magmas (Ginibre et al. 2007; Pietranik and Koepke 2009; Slaby and Götze 2004).

Fig. 5
figure 5

Chemical compositions of feldspars from the host rocks and MMEs in NMH pluton

Full size image

Amphibole

Representative analyses of amphiboles from both rock types are given in Supplementary Table 3. Most of the amphiboles can be classified as calcic magnesiohornblende (Leake et al. 1997), with a few actinolitic hornblendes in the MMEs (Fig. 6). The amphiboles in the MMEs and host rocks are very similar, but those in the MMEs have slightly higher Mg# [(Mg# = atomic ratios Mg/(Mg + Fe), where Fe is total iron)] (0.62–0.68) than those in the granodiorites (0.53–0.56). Few calcic amphibole crystals are zoned with a magnesiohornblende core and an actinolite rim (Supplementary Table 3; Fig. 6), which are interpreted as marking progressive oxidation in the magmatic environment (Martin 2007). The amphiboles have medium SiO2 contents, and low TiO2 and Na2O + K2O, similar to those from calc-alkaline igneous rocks in subduction environments (Coltorti et al. 2007; Martin 2007).

Fig. 6
figure 6

Composition and classification (Leake et al. 1997) of amphiboles from the host rock and MMEs in East Kunlun. Same symbols as in Fig. 5

Full size image

Biotite

All biotite crystals are uniform in composition, magnesian in character and plot within the calc-alkaline field in the MgO-FeOT-Al2O3 diagram (Supplementary Table 4; Fig. 7a and b). The FeOT/MgO ratios of the biotites in the MMEs (1.66–1.91) are identical to those in the host granodiorites (1.78–2.03), suggesting complete equilibration between the respective biotites. Furthermore, the studied biotite is similar in composition to that commonly found in calc-alkaline (mostly subduction-related), I-type granitoids (Abdel-Rahman 1994).

Fig. 7
figure 7

a Mg-(AlVI + Fe3++Ti)-(Fe2++Mn) diagram (Forster 1960) and b MgO-FeO-Al2O3 diagram (Abdel-Rahman 1994) for biotite from host rocks and MMEs in East Kunlun. Same symbols as in Fig. 5

Full size image

P-T conditions of crystallization

Various aluminum-in-hornblende geobarometers (Supplementary Table 5) were used to estimate the pressures under which the MMEs and their host granitoids formed, because all the rocks contain the required phase assemblage (quartz + plagioclase + K-feldspar + hornblende + biotite + apatite + zircon + titanite). In addition, we use the hornblende-plagioclase thermometer to calculate the crystallization temperatures of the studied rocks. Based on these calculations, the MMEs and their host granodiorites crystallized under the same P-T conditions (Supplementary Table 5). The pressures (3.2–3.4 kbar) calculated from the geobarometer of Schmidt (1992) correspond to shallow depths of ~12 km. Using the thermometers of Blundy and Holland (1990), the calculated equilibrium temperature of the co-existing hornblende and plagioclase in the MMEs (706–771°C, average = 737°C) is essentially identical to that in the host rocks (729–761°C, average = 742°C; Supplementary Table 5).

Whole-rock geochemistry

Whole-rock geochemical data are presented in Supplementary Table 6. Most of the granodiorite samples plot consistently in the granodiorite field on the classification diagram of Middlemost (1994), whereas the MMEs plot in the monzodiorite field (not shown). The MMEs and granodiorites both are high-K, calc-alkaline and metaluminous to slightly peraluminous rocks with A/CNK ranging from 0.75 to 1.05 (Supplementary Table 6), similar to I-type granite (Fig. 8a and b).

Fig. 8
figure 8

Chemical classifications of the NMH granodiorites and MMEs. a K2O vs SiO2 diagram (Peccerillo and Taylor 1976); b A/NK vs A/CNK diagram (Chappel and White 1974; Maniar and Piccoli 1989). Same symbols as in Fig. 5

Full size image

Chondrite-normalized REE patterns of the MMEs and granitoid rocks are invariably enriched in light REE (LREE) with respect to the HREE [(La/Yb)N = 4.22–4.47 and 9.84–17.02 for MMEs and granodiorites], and all samples display moderately negative Eu anomalies (δEu = 0.50–0.87; Supplementary Table 6; Fig. 9a). In the primitive mantle-normalized spidergram (Fig. 9b), all of the samples are enriched in large ion lithophile elements (LILE), such as Rb, Ba, Tu, U, K, and are depleted in high field strength elements (HFSE) Nb, Ta, P, Ti, typical of subduction-related magmatic rocks (Brown et al. 1984; Hussain et al. 2004; Omrani et al. 2008; Wade et al. 2006).

Fig. 9
figure 9

a Chondrite-normalized REE patterns and b Primitive mantle-normalized element spider diagrams for MMEs and granodiorites. Chondrite REE abundances are after Taylor and Mclennan (1985), Trace element abundances for the primitive mantle are after Sun and Mcdonough (1989). Same symbols as in Fig. 5

Full size image

Sr-Nd-Hf isotope geochemistry

Whole-rock Sr-Nd isotopic compositions for the MMEs and their host granodiorites are given in Supplementary Table 6 with initial ratios calculated at an age of 262 Ma. The MMEs and granodiorites have relatively homogeneous isotopic compositions, with high initial Sr ratios [(87Sr/86Sr)i = 0.70850–0.70863] and low εNd(t) values (εNd(t) = −6.1 to −4.6). The MMEs and granodiorites have similar Nd model ages (TDM2), ranging between 1.40 and 1.52 Ga.

Zircons from the MMEs have similar initial 176Hf/177Hf values to those of the host granodiorites (average (176Hf/177Hf)i = 0.282478 in the MMEs and 0.282451 in the granodiorites; Supplementary Table 7). However, the εHf(t) values of granodiorites have a wider spread (Fig. 10). The MMEs have εHf(t) values ranging from −6.83 to −3.15, with a weighted mean of −4.68, while the εHf(t) values of granodiorites range from −9.00 to −3.20, with a weighted mean of −5.63. Two-stage model ages (TDM2) are calculated for the source rock of the magma (Supplementary Table 7) by assuming a mean 176Lu/177Hf value of 0.015 for an average continental crust (Griffin et al. 2002). The granodiorites and MMEs have TDM2 model ages ranging from 1.49 to 1.86 Ga (Supplementary Table 7; Fig. 10b), which is in accordance with their Sr-Nd isotopic compositions.

Fig. 10
figure 10

a Histogram of εHf(t) and b Plots of εHf(t) vs U-Pb ages for zircons from sample BL13-1 and BL13 (the values used for constructing the DM and crust reference evolution lines are after Griffin et al. (2000, 2002))

Full size image

Discussion

Petrogenesis of the mafic microgranular enclaves

The MMEs have igneous mineralogy and textures, with feldspar xenocrysts and disequilibrium features such as oscillatory zoning, resorbed rims, poikilitic plagioclase, and acicular apatite, all of which are interpreted as reflecting chemical and/or thermal changes in the melt during crystal growth (Baxter and Feely 2002; Grogan and Reavy 2002). The enclaves in the NMH pluton commonly occur as swarms, which is an indication of mafic magma injection into felsic magma during magma mixing (Arvin et al. 2004; Blundy and Sparks 1992; Tobisch et al. 1997). Magma mixing is further supported by mineral chemistry, particularly the similarity in anorthite content of plagioclase between the MMEs and host rocks. The cores of plagioclase xenocrysts in the MMEs are identical in composition to the plagioclase in the felsic host rocks (An30), but the rims are more calcic (An49) (Supplementary Table 2). Such a compositional change suggests that the plagioclase xenocrysts were mechanically transferred from felsic magma into the mafic magma where a new and more calcic plagioclase overgrew the cores. In addition, some plagioclase grains from the felsic host rocks have sodium-rich cores (An30), which are overgrown by calcium-rich zones (An40)(Supplementary Table 2). This relationship is interpreted as the result of mixing between mafic and felsic magmas. Preservation of the compositional and textural disequilibrium features in the plagioclase is interpreted to be due to rapid crystallization and incomplete mixing.

The MMEs have P-T conditions of crystallization similar to those of their host rocks (MMEs: average 737°C and 3.15 kbar; host rocks: average 742°C and 3.39 kbar), showing that they crystallized together. In addition, the MMEs and host granodiorites have identical mineral assemblages, similar mineral compositions, similar geochemical characteristics, indistinguishable Sr-Nd isotopic compositions and an almost coeval crystallization age, also compatible with magma mixing. The wide range in zircon Hf isotopic compositions of the MMEs and granodiorites precludes a simple evolution by closed-system fractionation, because such mechanisms cannot produce the variable Hf isotopic compositions. On the other hand, such compositions can be explained by magma mixing because such a process can produce zircons with homogeneous U-Pb ages and variable Hf isotopic compositions (Yang et al. 2006; Yang et al. 2007b).

Magma sources and tectonic implications

Whole-rock geochemical data show that the NMH pluton is composed of metaluminous, high-K, calc-alkaline, I-type granodiorite (Fig. 8). All of the rocks are enriched in LILE and depleted in HFSE, characteristics typical of subduction-related magmatic rocks (Brown et al. 1984). A subduction-related origin for these rocks is consistent with their position in tectonic discrimination diagrams (Fig. 11), and is consistent with the geological evolution of East Kunlun. The A’nyemaqen Paleo-Tethyan ocean opened at least as early as Late Carboniferous (308 Ma), and closed during the late Permian-early Triassic (260–245 Ma), resulting in voluminous subduction-related magmatism in East Kunlun (Bian et al. 2004; Harris et al. 1988; Jiang et al. 1992; Mock et al. 1999; Xu et al. 2007; Yang et al. 1996; Yang et al. 2005).

Fig. 11
figure 11

a Major element plot after Batchelor and Bowden (1985) and b Rb vs Y + Nb plot for the MMEs and granodiorites in East Kunlun (after Pearce et al. 1984). Same symbols as in Fig. 5

Full size image

However, subduction-related, calc-alkaline magmatism may reflect a combination of processes, including melting of continental crust, the subducted slab and the mantle. We suggest that the NMH granodiorites formed by partial melting of continental crust. The granodiorites have relatively low ratios of K2O/Na2O, (Na2O + K2O)/(FeOtot + MgO + TiO2), and Al2O3/(FeOtot + MgO + TiO2) and high ratios of CaO/(MgO + FeO), characteristics, which are consistent with the experimental data (not shown, Gardien et al. 1995; Patiño Douce 1997; Rapp and Watson 1995), suggesting that all of these rocks may be derived from dehydration melting of the lower crust. The depletion of Nb and Ta relative to LILE and LREE reflects retention of Ti-rich minerals (e.g. rutile) or hornblende in the residue because such minerals have high partition coefficients for Nb and Ta (Baier et al. 2008; Green and Pearson 1987; Green 1995; Kalfoun et al. 2002). The granodiorites show enrichment in LREE with respect to the HREE [(La/Yb)N = 9.84–17.02], which suggests equilibrium with amphibole or garnet. Besides, the increasing trend between La/Sm ratio and SiO2 content, together with the absence of a corresponding trend for Sm/Yb vs SiO2 (Fig. 12), support the presence of amphibole as a residual phase in the crustal source of the felsic magmas. Partial melting of mafic lower crust is also consistent with the isotopic data. In fact, the whole-rock Nd model ages (1.40–1.52 Ga) and Hf model ages (TDM2) of zircons from the NMH granodiorites ranging from 1.49 to 1.86 Ga, which are consistent with the ages of the East Kunlun basement rocks, i.e., the Proterozoic Xiaomiao Group (Wang et al. 2003; Xu et al. 2007), suggest their derivation from a major late Paleoproterozoic crustal source (Fig. 10b).

Fig. 12
figure 12

a La/Sm and b Sm/Yb vs. SiO2 for the granodiorites in East Kunlun. Same symbols as in Fig. 5

Full size image

The negative εHf(t) values (average: −5.63 and −4.68 for granodiorites and MMEs, respectively) suggest that old crustal materials or enriched lithospheric mantle played a significant role in the magma generation, consistent with the Sr-Nd isotope data. Importantly, the Sr-Nd-Hf isotopic composition of MMEs is identical to those of granodiorite, which suggest mixing between a crustal magma and a mantle-derived component.

We further suggest that the mafic end-member involved in the magma mixing was derived from enriched lithospheric mantle, rather than asthenospheric mantle, because all the rocks have much lower values of siderophile elements (Co: 5–23 ppm, Ni: 3–50 ppm, Cu: 3–80 ppm) than rocks derived from asthenospheric mantle, and have enriched Sr-Nd-Hf isotopic compositions, which are incompatible with derivation from the asthenospheric mantle. The Sr-Nd isotope data plotted in Fig.13 illustrate that the MMEs are typical of an enriched lithospheric mantle as the major component of the mafic magma sources, which is significantly different from Permian OIB rocks in Northern Qinghai-Tibet Plateau that derived from the asthenospheric mantle (Fig. 13).

Fig. 13
figure 13

Correlation diagram between (87Sr/86Sr)i and εNd(t = 260 Ma) for NMH pluton (the data for the basement of East Kunlun is from Yu et al. 2005; data for enriched lithospheric mantle is from Chen et al. 2011; data for arc magmas is from Liu et al. 2003; Chen et al. 2007 and Yuan et al. 2009; data for OIB magmas is from Ma et al. 2007). Same symbols as in Fig. 5

Full size image

Numerous studies of magma mixing in subduction zone environments have focused mainly on mafic rocks, such as basalts in the Andes (Debaille et al. 2006; Hickey et al. 1986; Hickey-Vargas et al. 2002), and concluded that the magma mixing involved melts from the subducted slab, the lithospheric mantle and the asthenospheric mantle. However, although this model may explain the petrogenesis of mafic rocks in subduction zones, it is not applicable to granodioritic or granitic rocks. Thus, MMEs are evidence for magma mixing involving melts from the lithospheric mantle and the overlying lower crust, which is an important mechanism of magma mixing in subduction zones and is applicable to the origin of large scale granitic magmatism, e.g., this study and Carboniferous enclave-rich granitoids of the Andes of central Chile (Parada et al. 1999).

With the A’nemaqen Paleo-tethys oceanic slab being subducted beneath the East Kunlun since ca. 260 Ma (Jiang et al. 1992; Yang et al. 1996), the fluid carried by the slab would induce partial melting of the mantle wedge, which is a major melting mechanism to form mafic magmas in subduction zones (McCulloch and Gamble 1991; Tatsumi and Kogiso 2003). We interpret this to be the reason why the studied rocks show enrichment in LILE and LREE, but are depleted in HFSE and HREE, since LILE and LREE remain are fluid mobile (McCulloch and Gamble 1991). Also, the presence of fluid could contribute to the progressive oxidation for the generation of amphibole crystals with a magnesiohornblende core and an actinolite rim (Supplementary Table 3; Fig. 6; Martin 2007). Therefore, we suggest that the 260-Ma, subduction-related felsic magma (Naomuhun pluton) was derived from partial melting of Proterozoic crustal rocks, i.e., Xiaomiao group, which forms the basement of East Kunlun. This process was triggered by underplated melts derived from an enriched lithospheric mantle that formed MMEs (Fig. 14), just as the scenario in the Andes (Parada et al. 1999). Owing to the LILE- and LREE-rich fluid metasomatism, the mantle wedge underwent partial melting to form the underplated magma. This magma subsequently underplated the lower crust, and produced partial melts of felsic composition, which formed the NMH calc-alkaline granodiorites with mixed mantle component. A small volume of mafic magma derived from the enriched lithospheric mantle mixed with the granodioritic magma to form MMEs (Fig. 14b), which was then emplaced at a depth of ca. 12 km.

Fig. 14
figure 14

a Schematic illustration of magma generation in East Kunlun during slab subduction; b Schematic illustration of magma mixing during magma underplating

Full size image

Conclusions

  1. (1)

    LA-ICPMS zircon U-Pb dating indicates that the NMH granodiorites crystallized at 261 ± 2 Ma and the MMEs crystallized at 263 ± 2 Ma. Thus, this magmatism took place close to the Lower-Upper Permian boundary during subduction of Paleo-Tethyan lithosphere.

  2. (2)

    The MMEs and host granodiorites have identical mineral assemblages, similar mineral compositions, similar geochemical characteristics, indistinguishable Sr-Nd isotopic composition and almost simultaneous crystallization age, suggesting formation by magma mixing.

  3. (3)

    The whole-rock geochemistry and Sr-Nd-Hf isotopic compositions indicate that mafic lower crust is the prime candidate for the source of the NMH granitoid rocks, and that the MMEs are products of magma mixing between lower crustal intermediate melts and mafic melts derived from the lithospheric mantle.

  4. (4)

    The subduction-related calc-alkaline magmatism in the Northern Qinghai-Tibet Plateau, was caused by enriched lithospheric mantle-derived mafic magmas underplating the lower crust.