Hydrous melting of metasomatized mantle wedge and crustal growth in the post-collisional stage: Evidence from Late Triassic monzodiorite and its mafic enclaves in the south Qinling (central China)

Triassic collision between the Yangtze and North China blocks is a key aspect of the evolution of the Paleo-Tethys in East Asia. This paper reports age and geochemistry for Late Triassic monzodiorite and its mafic enclaves from the south Qinling (central China). The host monzodiorite and mafic enclaves have identical zircon U-Pb ages of 227 ± 3 Ma and 221 ± 3 Ma, respectively. The host monzodiorite displays moderate SiO2 (62.44–63.85 wt%) and MgO (3.44–4.47 wt%) contents and high Cr (157–244 ppm) and Ni (78–108 ppm) contents. It has evolved whole-rock Sr-Nd isotopic compositions (εNd(t) = –6.2 to –5.0) and slightly positive zircon εHf(t) values (up to +6.4). Given these characteristics in combination with high Th/Nb and Nb/Yb ratios, the host monzodiorite is considered to have been derived from hydrous melting of metasomatized mantle lithosphere. Its moderate Sr/Y and low Yb/Lu ratios indicate the fractional crystallization of hornblende. The mafic enclaves have lower SiO2 (52.85–58.53 wt%) and higher MgO (8.26–9.45 wt%) contents. Most zircons in the mafic enclaves display positive εHf(t) values of +0.9 to +16.5. These features indicate that the pristine mafic melt was derived from depleted mantle lithosphere. Minor grains in the mafic enclaves display lower εHf(t) values (−4 to 0) than the zircons in the host monzodiorite, suggesting that the mafic melt had incorporated some evolved crustal component before it intruded into the host monzodiorite chamber. In summary, in the circumstance of slab break-off and asthenosphere upwelling, hydrous melting of mantle lithosphere has contributed greatly to crustal-derived granites in the Qinling orogenic belts. These results have the following implications for the Triassic granites in the Qinling orogenic belts: (1) hydrous melting of metasomatized mantle wedge greatly contributes to crustal growth in orogenic processes; and (2) mantle-derived hydrous mafic melts induced the extensive melting of crust and led to the voluminous Triassic granites in the Qinling orogen. LITHOSPHERE; v. 11; no. 1; p. 3–20; GSA Data Repository Item 2018389 | Published online 19 November 2018 https:// doi .org /10 .1130 /L1006 .1


INTRODUCTION
The Triassic Qinling-Dabie-Sulu orogenic belt is the most prominent tectonic feature in central China, and resulted from the collision between the North China block and Yangtze block (Meng and Zhang, 2000;Ratschbacher et al., 2003;Zheng et al., 2011;Dong et al., 2015). It also plays a key role in understanding the tectonic evolution of the Paleo-Tethys and eastern Asia continents (Ernst et al., 2007). There is a striking difference between the western part (the Qinling) and the eastern part (the Dabie and Sulu): in the eastern part, the exposure of high-pressure (HP) and ultrahigh-pressure (UHP) metamorphic rocks indicates deep continental subduction (Zheng et al., 2011), whereas widespread high-K calc-alkaline granites and associated mafic rocks occur in the western part (Sun et al., 2002;Qin et al., 2008aQin et al., , 2009Qin et al., , 2010aQin et al., , 2010bQin et al., , 2013Wang et al., 2013).
What is the cause of the striking contrast between the Qinling and Dabie-Sulu parts of the orogenic belt? What geodynamic model can account for the genesis of the Triassic granitoids and related mafic rocks in the Qinling area: delamination of lower crust; a slab break-off model (Sun et al., 2002;Qin et al., 2008b); melting of subducted Yangtze continental lithosphere during the exhumation process (Qin et al., 2010b(Qin et al., , 2013; or partial melting of subducted Paleo-Tethys oceanic crust (Jiang et al., 2010)? Understanding these issues has great significance for investigating the complete tectonic evolution of the Triassic Qinling-Dabie orogenic belt.
In this paper, we present new mineral chemistry, major-and traceelement compositions, Sr-Nd-Pb isotopic compositions, zircon U-Pb dating, and Lu-Hf isotopic compositions for the Late Triassic Gaoqiao monzodiorite and its mafic enclaves from the south Qinling. We use the new data to explore the following two issues: (1) origin of the host monzodiorite and its mafic enclaves; and (2) the genetic link between the melting of mantle lithosphere and formation of Triassic granites in the Qinling orogenic belt.

GEOLOGICAL BACKGROUND AND FIELD GEOLOGY
The geological background of the Triassic Qinling-Dabie-Sulu orogenic belt has been described by many previous workers (Meng and The zircon grains were separated using heavy liquid and magnetic techniques. Representative zircon grains were handpicked and mounted in epoxy resin discs, which were then polished and coated with carbon. Internal morphology was examined using cathodoluminescence (CL) prior to U-Pb and Lu-Hf isotopic analyses. Laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) zircon U-Pb analyses were conducted on an Agilent 7500a ICP-MS equipped with a 193 nm laser, following the method of Yuan et al. (2008). The U.S. National Institute of Standards and Technology (NIST) 610 standard silicate glass was used to optimize the instrument to obtain the maximum signal intensity ( 238 U signal intensity of >2000 cps/g g −1 at a beam diameter of 30 μm and a laser frequency of 6 Hz) and low oxide production (ThO/Th <1%). The ion signal intensity ratio measured for both 238 U and 232 Th (NIST SRM 610) ( 238 U and 232 Th ≈1) was used as an indicator of complete vaporization (Günther and Hattendorf, 2005). The 207 Pb/ 206 Pb and 206 Pb/ 238 U ratios were calculated using the GLITTER program (version 4.4., Macquarie University). Common Pb contents were then evaluated using the method described by Andersen (2002). Age calculations and plotting of concordia diagrams were made using Isoplot (version 3.0) (Ludwig, 2003). Concentrations of U, Th, Pb, and trace elements were calibrated using 29 Si as an internal standard and NIST SRM 610 as an external standard. The two standard zircons yielded weighted mean 206 Pb/ 238 U ages of 1064.2 ± 3.1 Ma (n = 14, 2σ) and 603.1 ± 3.4 Ma (n = 12, 2σ), respectively, which are in good agreement with the recommended isotope dilution-thermal ionization mass spectrometry (ID-TIMS) ages (Wiedenbeck et al., 1995).
In situ zircon Hf isotopic analyses were conducted using a Neptune multiconductor ICP-MS (MC-ICP-MS) equipped with a 193 nm laser. During analyses, a laser repetition rate of 10 Hz at 100 mJ was used and spot sizes were 44 μm. The 176 Yb/ 172 Yb value of 0.5887 and mean Yb value obtained during Hf analysis on the same spot were applied for the interference correction of 176 Yb on 176 Hf (Iizuka and Hirata, 2005). The detailed analytical technique is described by Yuan et al. (2008). During analyses, the 176 Hf/ 177 Hf ratio of the standard zircon (91500) was 0.282294 ± 15 (2σ, n = 20), similar to the commonly accepted 176 Hf/ 177 Hf ratios of 0.282302 ± 8 and 0.282306 ± 8 (2σ) measured using the solution method (Goolaerts et al., 2004;Woodhead et al., 2004); the 176 Lu/ 177 Hf ratio of the standard zircon was 0.00031.

Major-and Trace-Element Analysis
For major-and trace-element analysis, fresh chips of whole-rock samples were powdered to 80 µm using a tungsten carbide ball mill. Major and trace elements were analyzed using X-ray fluorescence (XRF) (Rikagu RIX 2100) and ICP-MS (Agilent 7500a), respectively. Analyses of U.S. Geological Survey and Chinese national rock standards (BCR-2, GSR-1, and GSR-3) indicate that both analytical precision and accuracy for major elements are generally better than 5%. For trace-element analysis, sample powders were digested using an HF + HNO 3 mixture in high-pressure Teflon bombs at 190 °C for 48 h. Analytical precision is better than 10% for most of the trace elements.

Whole-Rock Sr-Nd-Pb Isotope Analysis
Whole-rock Sr-Nd-Pb isotopic data were obtained using a Nu Plasma HR multicollector mass spectrometer; Sr and Nd isotopic fractionation was corrected to 87 Sr/ 86 Sr = 0.1194 and 146 Nd/ 144 Nd = 0.7219, respectively. Standard NIST SRM 987 yielded an average value of 87 Sr/ 86 Sr = 0.710250 ± 12 (2σ, n = 15), and the La Jolla standard gave an average of 143 Nd/ 144 Nd = 0.511859 ± 6 (2σ, n = 20). Whole-rock Pb was separated by anion exchange in HCl-Br columns; Pb isotopic fractionation was corrected to 205 Tl/ 203 Tl = 2.3875.   Table DR1: Analytical results of representative minerals from the host monzodiorite and mafic enclaves from the Gaoqiao pluton, south Qinling; Table DR2: Zircon LA-ICP-MS U-Pb results for the host monzodiorite and enclaves from the Gaoqiao pluton; Table DR3: Zircon Lu-Hf isotopic compositions of the host monzodiorite and enclaves from the Gaoqiao pluton, south Qinling; and Table DR4: FCA results, is available at http://www.geosociety .org /datarepository /2018 or on request from editing@geosociety.org.

Zircon LA-ICP-MS U-Pb Dating
Samples of host monzodiorite (sample GQ-1) and mafic enclaves (sample GQB-1) were selected for zircon U-Pb isotope analysis. Zircon CL images and U-Pb concordant diagrams are presented in Figure 4, and the U-Th-Pb isotope data are listed in Table DR2.

Major-and Trace-Element Chemistry
Major-and trace-element analysis results for the monzodiorite and enclaves are listed in Table 1.
The host monzodiorite displays tonalitic to dioritic composition (

Whole-Rock Sr-Nd-Pb Isotopic Compositions
Whole-rock Sr-Nd-Pb isotopic compositions are given in Tables 2  and 3. Initial isotopic values were calculated according to the zircon U-Pb age. Whole-rock Nd model ages were calculated using the model of DePaolo (1981).
The host monzodiorite has ( (Figs. 9A, 9B), the host monzodiorite and its enclaves plot in the transitional zone between the compositions of the North China and Yangtze blocks (Zhang et al., 1997). They also display a similar Pb isotopic composition to that of the Neoproterozoic Bikou (Yan et al., 2004) and Yaolinghe basalts (Xia et al., 2008).

Zircon Lu-Hf Isotope and Trace-Element Chemistry
Zircons from the host monzodiorite and mafic enclaves that were dated by U-Pb were also selected for Lu-Hf analysis in the same domain. The results are listed in Table DR3. Initial 176 Hf/ 177 Hf ratios and ε Hf(t) values of the magmatic zircons were calculated according to their U-Pb ages. Figure 10A shows ε Hf(t) values versus crystallization ages.
Twenty-four (24) out of 36 spots in the host monzodiorite (sample GQ-1) were collected for Lu-Hf isotope analysis. Spots 1, 5, 6, 10, 15, 16, and 22 display discordant U-Pb ages; their Hf isotopic compositions have no geological significance. The other 17 grains display variable Hf isotopic compositions, ε Hf(t) = −0.4 to +6.4, with corresponding two-stage Hf model ages of 738-1053 Ma. Twenty-four (24) out of 36 spots in the mafic enclaves (sample GQB-1) were selected for Lu-Hf analysis. Spots 4 and 10 display discordant U-Pb ages. The other 22 spots have ε Hf(t) = −3.5 to +16.5. Five out of 22 spots display negative ε Hf(t) values of −3.5 to −0.7, with two-stage Hf model ages of 1104-1239 Ma; the remaining 17 spots display positive ε Hf(t) values of +0.9 to +16.5, with two-stage Hf model ages of 208-1018 Ma. As shown in Figure 10B, zircons from the mafic enclaves can be subdivided into two groups: the first group has ε Hf(t) values concentrating around +6 to +8, and the second group displays lower ε Hf(t) values than those of the zircons from the host monzodiorite.
Zircons from both host monzodiorite and mafic enclaves display significant positive Ce anomalies and negative Eu anomalies; these features indicate that plagioclase had begun crystallization by the time magmatic zircon was being formed (Figs. 11A, 11B). Zircons from host monzodiorite display lower ΣREE contents and Dy/Yb ratios than those from the mafic enclaves (Figs. 11C, 11D), suggesting different crystallization conditions.

Genetic Link Between the Host Monzodiorite and Its Mafic Enclaves
Mafic enclaves in granitic rocks are usually considered to be formed by mixing or mingling of felsic and mafic melt or to represent wall-rock xenoliths, melting restite (Chappell and Wyborn, 2012), or quenched cumulates (Phillips et al., 1981;Donaire et al., 2005;Pascual et al., 2008). The mafic enclaves and host monzodiorite studied here have identical zircon U-Pb ages of ca. 220 Ma (Fig. 4); this indicates that the mafic enclaves formed by a contemporaneous magmatic event, rather than as xenoliths or melting resitite. Most mafic enclaves in the Gaoqiao pluton display a sharp contact boundary with the host monzodiorite ( Fig. 2A), indicating limited magma mingling. In addition, we selected host mozodiorite (sample GQ-21, which has the highest SiO 2 of 63.85%) and mafic enclaves (sample GQ-B09 with the lowest SiO 2 of 52.85%) as the end members of felsic and mafic melts, respectively. As shown in the variation diagrams (Fig. 6)  monzodiorite displays higher crystallization pressures (1.23-1.41 kbar) than those of the hornblende in the mafic enclaves (0.60-0.96 kbar), suggesting that the hornblende in the mafic enclaves crystallized at relatively shallow depth; this result also indicates that the mafic enclaves represent mafic melts that intruded into the partially crystallized monzodiorite chamber in a relatively late stage.
It is intriguing to consider whether or not the host monzodiorite and mafic enclaves derived from the same source region. Several lines of compelling evidence indicate that the host monzodiorite was not formed by the fractional crystallization of the mafic melt that is represented by the mafic enclaves: (1) there is an obvious compositional gap (Fig. 6) between the mafic enclaves and host monzodiorite, which is contradictory to continuous fractional crystallization; (2) zircons from the mafic enclaves and host monzodiorite display different Lu-Hf compositions (Fig. 10), and simple fractional crystallization in closed system would not change the zircon Lu-Hf isotopic composition (Shaw and Flood, 2009); (3) as mentioned above, hornblende in the mafic enclaves and host monzodiorite display different crystallization pressures (Table DR1), suggesting different crystallization paths.
In summary, we propose that the host monzodiorite and mafic enclaves represent contemporaneous magma that derived from two distinct mantle source regions, with the magma mingling process having limited effect on their geochemistry.
The following evidence indicates that the first three mechanisms are not plausible for the genesis of the host monzodiorite: (1) The host monzodiorite is enriched in large-ion lithophile elements (LILEs) and LREEs, depleted in Nb, Ta, and Ti (Fig. 6), and displays the typical signature of continental or arc crust (Wilson, 1989). Zircons from host monzogranite display slightly depleted Lu-Hf isotopic compositions (Fig. 10A), i.e., ε Hf(t) = -0.4 to +6.4. It also has evolved Sr-Nd isotopic compositions (Table 2). These features suggest that the host monzodiorite was derived from melting of metasomatized enriched mantle lithosphere rather than a depleted source region. Compared with the sanukite of the Setouchi volcanic belt, Japan (Tatsumi, 2006), the monzodiorite displays higher Th/La (0.42-0.51) and Th/Yb (10.3-13.4) ratios (Fig. 12), indicating significant involvement of sediment-derived melts (Tatsumi, 2006). (2) Zircons from the host monzodiorite have a unimodal Lu-Hf distribution (Fig.  10B), inconsistent with the model of primitive mafic melts assimilating significant crustal components. (3) The absence of Triassic extensional tectonics in the Qinling area argues against the delamination model (Dong et al., 2015).
Compared with the host monzodiorite, primitive high-Mg andesitic melts derived from hydrous melting of harzburgite have lower SiO 2 (54.0%-55.2%) and higher MgO (7.2%-11.9%) contents (Wood and Turner, 2009). This feature indicates that the primitive mafic melt underwent fractional crystallization of hornblende and other mafic minerals. The mafic enclaves sample (GQB-09) is used as starting parental melt composition to determine if the host monzodiorite was crystallized from mafic melts (we argue that sample GQB-09 can roughly represent the pristine mafic melt, although it may be derived from a more depleted source region). The FC-AFC-FCA (fractional crystallization-assimilation fractional crystallization-fractional crystallization assimilation) and mixing modeler of Ersoy and Helvacı (2010) is used to assess the role of hornblende fractional crystallization in the formation of the host monzodiorite. The results indicate that 15%-20% fractional crystallization of hornblende (80%) + plagioclase (20%), with 20% assimilation of lower continental crust (Rudnick and Fountain, 1995), can form intermediate melts similar to those that formed the host monzodiorite (Fig. 13). This possibility is supported by the host monzodiorite flat HREE patterns (Fig. 7A), and its low Yb/Lu (6.7-6.8), Dy/Yb (1.90-1.99), and (Ho/Yb) N (1.15-1.18) ratios indicate depletion in middle REEs (MREEs) (Moyen and Martin, 2012).
In summary, we propose that the host monzodiorite represents evolved mafic melts that derived from hydrous melting of metasomatized enriched mantle wedge; fractional crystallization of hornblende can account for its moderate Sr/Y and Yb/Lu ratios (Moyen and Martin, 2012). Its evolved Sr-Nd isotopic compositions and high Th/La and Th/Yb ratios suggest metasomatism by sediment-derived melts.
On the other hand, the mafic enclaves have identical U/Th ratios (0.37-0.41) with depleted mantle (0.33); these ratios are higher than those of the host monzodiorite (0.10-0.17) and continental crust (0.25), suggesting depleted components (Hawkesworth et al., 1997). Most zircons in the mafic enclaves display positive ε Hf(t) values of +0.9 to +16.5, with singlestage Hf model ages of 208-1018 Ma, which should have been inherited from their depleted source region. The mafic enclaves display decoupled whole-rock Nd and zircon Hf isotopic compositions (Table 2; Table DR2). Considering the extremely refractory Lu-Hf system in zircons (Hawkesworth and Kemp, 2006), we argue that the zircon grains with positive ε Hf(t) values were crystallized from pristine depleted mafic melts. Assimilation of evolved crustal rocks may account for the evolved Sr-Nd isotopic compositions ( Table 2). The zircon grains that have negative ε Hf(t) values suggest incorporation from evolved crustal melts before the pristine mafic melt intruded into the host monzodiorite chamber.
In summary, it can be considered that the mafic enclaves represent hydrous mafic melt derived from melting of a relatively depleted mantle source. The occurrence of zircon grains with negative ε Hf(t) values and their evolved Sr-Nd isotopic compositions suggest the assimilation of evolved crustal-derived melts before the mafic enclaves intruded into the host monzodiorite chamber.

Implication for the Hydrous Melting of Metasomatized Mantle Wedge and Crustal Growth in a Post-Collisional Setting
Triassic granites in the Qinling area recorded the most important information about the collision process between the South China and North China blocks. Most workers argued that these granites were caused by the northward subduction of the Mianlue ocean (Jiang et al., 2010;. Qin et al. (2010bQin et al. ( , 2013 proposed that decompression melting of subducted Yangtze continental lithosphere during the exhumation may have produced the Triassic granitoids in the Qinling orogen. The extensive Triassic granitoids across the Qinling orogen (Sun et al., 2002;Jiang et al., 2010;Qin et al., 2010aQin et al., , 2010bQin et al., , 2013 indicate an extensive Triassic crustal melting event in the Qinling orogen, which would have needed a gigantic heat source to trigger the crustal melting in the orogenic process. We have summarized the age and key geochemical features of the Gaoqiao monzodiorite and other crustal-derived granites in the Qinling area (Table 4). It is obvious that the Gaoqiao monzodiorite is synchronous with most Triassic granites in the Qinling orogenic belt, suggesting an extensive melting event. Furthermore, it has the lowest SiO 2 contents, highest MgO and Cr contents, and relatively high positive zircon ε Hf(t) values; these features indicate that the Gaoqiao monzodiorite was formed by evolved mafic melts that derived from melting of enriched metasomatized mantle lithosphere. This indicates that melting of mantle lithosphere and generation of mafic melt provide essential heat for crustal melting. In combination with regional geological background, we propose that slab break-off may be the most plausible model to explain the genesis of the Triassic granites in the Qinling orogen (Fig. 14): (1) In the circumstance    of slab break-off, upwelling of asthenosphere would have caused extension in mantle lithosphere (Davies and von Blankenburg, 1995).
(2) Hydrous melting of metasomatized mantle wedge (which was formed by a previous subduction event) would have produced hydrous mafic melts.
(3) Fractional crystallization of hornblende would account for the moderate Sr/Y and Yb/Lu ratios; the final product would have identical geochemistry with the host monzodiorite. Zircons that crystallized in this evolved mafic magma would have transitional Lu-Hf isotopic compositions. (4) The latestage pristine hydrous mafic melts would have assimilated some crustal rocks before they intruded into the partially crystallized monzodiorite chamber. Limited magma mingling would have formed the mafic enclaves. These results have the following important implications for the genesis of the Triassic granites in the Qinling area: (1) the Triassic collision caused the heterogeneity of the mantle lithosphere in the Qinling orogenic belt; (2) hydrous melting of metasomatized enriched mantle lithosphere produced high-Mg diorites similar to the host monzodiorite; (3) melting of relatively depleted mantle produced pristine mafic melts that had depleted isotopic compositions; these pristine mafic melts incorporated evolved crustal components before they intruded into the host monzodiorite chamber; (4) the mafic melts that derived from melting of mantle lithosphere provided an essential heat source for the crustal-derived granites in the Qinling orogenic belt.
(2) The host monzodiorite was derived from hydrous melting of metasomatized enriched mantle wedge. Subsequent fractional crystallization of hornblende can account for the moderate Sr/Y and Yb/Lu ratios.
(3) The mafic enclaves represent hydrous mafic melt that was derived from melting of relatively depleted mantle source. Their evolved Sr-Nd isotopic compositions were caused by assimilation of some evolved crustal components before they intruded into the host monzodiorite magma chamber.
(4) These results indicate that melting of mantle wedge in the postcollisional stage may greatly contribute to crustal growth and led to the extensive Triassic granites in the Qinling orogenic belt.