4.1 Major and trace elements

Major and trace element concentrations of fresh granitoid samples from the Tongbai Complex were analysed at the State Key Laboratory for Mineral Deposits Research (SKL-MDR) of Nanjing University using X-ray fluorescence spectrometry (XRF) and inductively coupled plasma-mass spectrometry (ICP-MS) methods, respectively. Based on standards and duplicate sample analysis, the analytical precision is generally better than 2% for major elements and 5–10% for the trace and rare earth element concentrations. More detailed analytical procedures please refer to those described in Jiang, Jiang, Li, Zhao, and Peng (2018) and Niu and Jiang (2020).

4.2 LA-ICP-MS zircon U–Pb isotope analysis

Zircon grains with good internal zonation as revealed by CL images were selected and their U–Pb isotopes were analysed by a Resonetics-S155 laser system coupled with an iCAP Qc ICP-MS at the State Key Laboratory of Geological Processes and Mineral Resources (SKL-GPMR), China University of Geosciences, Wuhan. The instruments include an ArF excimer laser generator to generate 193 nm deep ultraviolet beam, with a 33 μm laser beam spot, a 10 Hz denudation frequency and a 30s denudation time. The laser beam was focused on the surface of zircon grains through the homogenizing optical path. The sample aerosols were transported by high-purity He gas first and then mixed with Ar gas and a small amount of N₂ gas before entering the ICP mass spectrometer. Three standards were used during the analysis seasons, including the NIST SRM 612 glass for signal drift correction, a standard Pleovice as a blind sample to monitor data quality and a zircon standard 91,500 as the external standard. The obtained analytical raw data were processed using ICPMSDataCal software provided by Liu, Gao, et al. (2010); Liu, Hu, et al. (2010). For more detailed analytical procedures, please refer to those described in Jiang et al. (2018) and Zhang, Lentz, Thorne, and Massawe (2020).

4.3 LA-MC-ICP-MS zircon Hf isotope analysis

Lu-Hf isotopic compositions of the same zircon grains used for U–Pb isotopes were further analysed by a Resonetics-S155 laser system coupled with a Nu Plasma II multi collector ICP-MS at SKL-GPMR lab, China University of Geosciences, Wuhan. During this analysis, we used a slightly larger laser beam spot of 50 μm and tried our best to focus on the same domains or zones for the U–Pb isotope analysis. The laser beam has a denudation frequency of 10 Hz and a denudation time of 40s. In order to correct the instrumental mass bias, the Hf isotopic ratios were normalized to a ¹⁷⁹Hf/¹⁷⁷Hf ratio of 0.7325 and then using the equation of β_Yb = 0.8725 × β_Hf, the Yb isotopic ratios were also normalized (Chu et al., 2002; Xu, Wu, Xie, & Yang, 2004). During the analysis, the signal of ¹⁷⁶Hf can be interfered by isotopes ¹⁷⁶Lu and ¹⁷⁶Yb; therefore, a calibration was applied using ¹⁷⁶Yb/¹⁷²Yb = 0.5886 and ¹⁷⁶Lu/¹⁷⁵Lu = 0.02655 (Chu et al., 2002). Along with the sample analysis, a reference material Penglai Zircon was also analysed in regular basis, which yielded a weighted mean ¹⁷⁶Hf/¹⁷⁷Hf ratio of 0.282905 ± 0.000019 (n = 10, 2σ). This value is the same as the reference value (0.282906 ± 0.000010) reported by Li et al. (2010). For more detailed analytical procedures, please refer to those described in Jiang et al. (2018) and Niu and Jiang (2020).

4.4 Whole-rock Sr–Nd–Pb isotope analysis by TIMS and MC-ICP-MS

The whole-rock samples were treated the same way as the trace element analysis in the same SKL-MDR lab at Nanjing University. After digestion of the samples, the column chemistry was applied to purify and separate the Sr, Nd, and Pb from the matrix elements, and then isotope analysis was conducted either on a Triton Ti thermal ionization mass spectrometer (TIMS) for Sr and Nd isotopes or a Neptune Plus MC-ICP-MS for Pb isotopes. More detailed analytical procedures can be referred to those described in Pu, Gao, Zhao, Ling, and Jiang (2005) and Xu, Jiang, Luo, Zhao, and Ma (2016). In order to calculate the ⁸⁷Rb/⁸⁶Sr and ¹⁴⁷Sm/¹⁴⁴Nd ratios of the samples, we used the measured concentrations of Rb, Sr, Sm and Nd during trace element analysis by the ICP-MS method.

5.1 Major, trace and rare earth element contents

Ten samples, including five samples for gneissic granites (ZK801-8, ZK701-151, ZK1901-47, ZK1001-88 and ZK1001-82) and five quartz monzonite samples (D1-D5) from the Tongbai Complex, were analysed, and their major, trace and rare earth element results are listed in Tables 2 and 3, respectively. The results show a very low (0.22–0.76%) loss on ignition (LOI) for each sample, suggesting these samples have not been undergone post-magmatic alteration or weathering. The quartz monzonite samples show low SiO₂ (63.00–65.09%) contents but high Al₂O₃ (16.45–17.08%), CaO (2.14–2.34%), TFe₂O₃ (3.42–3.97%), MgO (0.98–1.16%), and TiO₂ (0.65–0.79%) contents. The gneissic granite samples have higher SiO₂ (69.31–72.51%) contents, slightly lower Na₂O (3.98–4.48%), and K₂O (4.46–4.92%) contents and distinct lower Al₂O₃ (13.92–14.88%), CaO (1.22–1.60%), MgO (0.32–0.61%), and TiO₂ (0.21–0.53%) contents relative to the quartz monzonite samples. Nevertheless, both the quartz monzonite and gneissic granite have comparable MnO contents, Na₂O/K₂O ratios and analogous A/NK and A/CNK ratios. The quartz monzonite samples belong to shoshonite series, while the gneissic granite samples spread over on both sides of the boundary between shoshonite series and high-K series (Figure 4a). On the A/NK versus A/CNK diagram, all the samples approach to the border line of meta-aluminous and peraluminous, but the majority pertain to meta-aluminous (Figure 4b). The Al₂O₃, TiO₂, TFe₂O₃, CaO, MgO, and P₂O₅ contents of all the samples show negative correlations with SiO₂ (Figure 5a–f), but the linear correlations between Na₂O and K₂O versus SiO₂ are very weak (Figure 5g,h). Although the Sr/Y and (La/Yb)_N ratios are generally lower in the gneissic granite samples than those in the quartz monzonite samples, their linear correlations are also very weak (Figure 5i,j). The Th and Y contents are positively correlated with respect to SiO₂ and are consistent with I-type granitic trend (Figure 5k,l).

Analogously, some similar features of trace element compositions between gneissic granite and monzonite include: (1) enrichments of Rb, Ba, Th, U, K, and Sr and depletions of Nb, P, and Ti (Figure 6a,c); (2) similar Rb/Y + Nb characteristics, indicating a post-collision environment (Figure 4c); (3) LREE enrichments, HREE depletions and negligible MREE depletions (quartz monzonite: (La/Yb)_N = 74.29–85.31; (La/Sm)_N = 5.51–5.95; (Gd/Yb)_N = 5.46–6.60) and gneissic granite: (La/Yb)_N = 26.45–46.59; (La/Sm)_N = 5.47–8.13; (Gd/Yb)_N = 2.13–3.38) (Table 3; Figure 6b,d); (4) slightly negative or no Eu anomalies and no Ce anomalies (quartz monzonite: Eu/Eu* = 0.86–0.93; Ce/Ce* = 0.97–0.98; gneissic granite: Eu/Eu* = 0.78–0.96; Ce/Ce* = 0.92–0.99) (Table 3; Figure 6b,d). The gneissic granite samples exhibit more deficit Nb, P and Ti than the quartz monzonite samples (Figure 6a,c). Nevertheless, the quartz monzonite samples show more significant LREE/HREE fractionation than the gneissic granite (Figure 6b,d).

5.2 Zircon U–Pb ages

Two samples (HJG-D1 and ZK801-8) were selected for LA-ICP-MS zircon U–Pb dating, and the results are listed in Table 4, and their corresponding concordia and weighted mean ages, together with the representative CL images, are shown in Figure 7.

These zircon grains are mostly euhedral, transparent and colourless with typical magmatic oscillatory zoning. Thirteen analysed spots on 13 zircon grains from sample HJG-D1 exhibit U contents of 137–635 ppm and Th contents of 25–1,514 ppm with Th/U ratios of 0.06–2.38. These data yielded a weighted mean age of 130.2 ± 1.7 Ma (MSWD = 1.7) (Figure 7a). Seventeen spots on 17 zircon grains from the sample ZK801-8 demonstrate high U/Th ratios of 0.31–0.97 and a weighted mean age of 144.2 ± 1.3 Ma (Figure 7b).

5.3 Zircon Hf isotopic compositions

The Hf isotopic compositions for zircon grains from quartz monzonite and gneissic granite are presented in Table 5 and Figure 8. The monzonite zircons yield ¹⁷⁶Hf/¹⁷⁷Hf ratios of 0.282098–0.282383, ¹⁷⁶Yb/¹⁷⁷Hf ratios of 0.001944–0.030197 and ¹⁷⁶Lu/¹⁷⁷Hf ratios of 0.000107–0.000842, and the corresponding ε_Hf(t) values are between −21.0 and −11.0 (Figure 8b). The two-stage Hf model ages (T_DM2) are 1.88–2.51 Ga, belonging to Palaeoproterozoic (Figure 8c). The ¹⁷⁶Hf/¹⁷⁷Hf, ¹⁷⁶Yb/¹⁷⁷Hf, and ¹⁷⁶Lu/¹⁷⁷Hf ratios of the gneissic granite zircons range from 0.282106 to 0.282172, from 0.012072 to 0.048787 and from 0.000521 to 0.001921, respectively, and the relevant ε_Hf(t) values are between −20.8 and −18.5 (Figure 8b), with two-stage Hf model ages (T_DM2) from 2.35 to 2.49 Ga (Figure 8c).

5.4 Whole-rock Sr–Nd–Pb isotopic compositions

Five gneissic granite samples and four quartz monzonite samples were selected for whole-rock Sr–Nd–Pb isotope analyses, and the results are listed in Tables 6 and 7 and plotted in Figure 9. The values of (⁸⁷Sr/⁸⁶Sr)_i, ε_Nd(t) and Pb_i have been calculated based on zircon U–Pb ages. Gneissic granite and quartz monzonite samples show (⁸⁷Sr/⁸⁶Sr)i values of 0.70731–0.70786 and 0.70730–0.70757, respectively (Table 6; Figure 9a). Gneissic granitic samples show ε_Nd(t) values of −14.1 to −13.4 (Figure 9b) with Nd model ages between 1.62 and 1.74 Ga (Table 5). The ε_Nd(t) values of quartz monzonitic samples are from −11.8 to −11.4 (Figure 9b), and the corresponding Nd model ages range from 1.4 to 1.5 Ga (Table 6).

The gneissic granites show initial Pb isotope ratios of ²⁰⁶Pb/²⁰⁴Pb 17.123–17.436, ²⁰⁷Pb/²⁰⁴Pb 15.388–15.474 and ²⁰⁸Pb/²⁰⁴Pb 37.811–37.986, respectively (Table 7; Figure 9c,d). The quartz monzonite samples show slightly lower initial Pb isotope ratios of ²⁰⁶Pb/²⁰⁴Pb (17.112–17.141) but similar ²⁰⁷Pb/²⁰⁴Pb (15.386–15.393) and ²⁰⁸Pb/²⁰⁴Pb (37.786–37.828) ratios (Table 7; Figure 9c,d).

6.1 Early Cretaceous granitic magmatism in the Tongbai orogen

The granitic magmatism in the Tongbai orogen mainly concentrates on two periods: Palaeozoic and Mesozoic (Zhang et al., 2013). In terms of exposed outcrop area, the Mesozoic granitoids occupy more than half of the Tongbai orogen, while the Palaeozoic granitoids scatter mostly in the northern segment of the Tongbai orogen. These Mesozoic granitoids were mainly formed during the Early Cretaceous (130–150 Ma), represented by the gneissic granites in the Tongbai Complex, and those un-deformed granites at Jigongshan, Caodian and Qijianfeng (Table 1). There are also some small granitic intrusions such as Laowan and Liangwan (Table 1), which may be connected to the Caodian and Jigongshan below the surface. Recent geochronology studies of gneissic granites in the Tongbai Complex indicate that they formed at 131–145 Ma (e.g., Chen et al., 2017; Cui et al., 2012; Liu et al., 2016; Liu, Gao, et al., 2010; Liu, Hu, et al., 2010; Zhang et al., 2018) rather than previous reported of 746–776 Ma by Kröner, Zhang, and Sun (1993). The new gneissic granite age of 144. 2 ± 1.3 Ma in this study is basically the same as that of previous studies (Table 1; Figure 7b). The Caodian, Jigongshan, Laowan and Liangwan granites have zircon U–Pb ages ranging from 125 to 139 Ma (Jiang et al., 2009; Liu, Jiang, et al., 2008; Zhang et al., 2013, 2018), consistent with their field intrusive relationship (Zhang et al., 2018). The Qijianfeng granitic complex with an age of 133–141 Ma (Table 1) belongs to the undeformed granites, so as the Caodian and Jigongshan granites, which are all later than the gneissic granites in the Tongbai Complex (Chen et al., 2018). Quartz monzonite in the Tongbai Complex reveals a much later age of 130.2 ± 1.7 Ma in this study, indicating that the magmatic events in the Tongbai region lasted up to 10–15 Ma in the Early Cretaceous period. On account of geology and U–Pb ages (Figures 2 and 7c), three stages of granitoid magmatism can be identified in the Tongbai orogen (Table 1), including (1) gneissic granites in the Tongbai Complex concentrated in 138–145 Ma; (2) un-deformed monzogranite situated on both sides of the Tongbai Complex such as the Caodian, Jigongshan and Qijianfeng with emplacement ages of 132–141 Ma; and (3) quartz monzonite in the Tongbai Complex at 130 Ma. The gneissic granites in the Tongbai Complex formed first and then were deformed in the early stage of the Cretaceous tectonic period, afterwards the un-deformed granites (Caodian and Jigongshan) and quartz monzonites emplaced in sequence in the Tongbai region. Nonetheless, the above magmatic rocks of the Early Cretaceous all intruded into the Tongbai area from 145 to 130 Ma, and there was no obvious chronological boundary among them.

Although the granitic magmatism in the Tongbai orogen can be subdivided into three stages based on their age, deformation structure and compositions, all of them formed before 130 Ma and exhibit high Sr/Y ratios (>40) (Niu & Jiang, 2020; Zhang et al., 2013, 2018). In contrast, two stages of granitic magmatism had been defined in the Dabie region (Table 8): (1) >130 Ma HSG: the early high Sr/Y granitoid plutons (130–143 Ma) and (2) <130 Ma NG: the later normal granitoid plutons (117–130 Ma) (He et al., 2013; Xu, Ma, & Ye, 2007). However, so far no granitoid younger than 130 Ma have been found in the Tongbai region, implying that the second stage of granitic magmatism may have not well developed in this region.

6.2 Petrogenesis

6.2.1 Genetic type

Gneissic granites in the Tongbai Complex show high contents of SiO₂ (69.31–72.51 wt%), Al₂O₃ (13.92–14.88 wt%) and Sr (715–853 ppm) but low contents of MgO (0.32–0.61 wt%), Y (9.2–24.2 ppm), Yb (0.89–2.37 ppm), and low Na₂O/K₂O ratio (0.88–1.00) (Tables 2 and 3). They also show slight negative Eu anomalies (Eu/Eu* = 0.77–0.96), (⁸⁷Sr/⁸⁶Sr)_i values of 0.7073–0.7078 and ε_Nd(t) values of −14.18 to −13.44 (Tables 3 and 6), which are basically the same as the high-K (C-type) adakitic features (SiO₂ > 56 wt%, Al₂O₃ > 15 wt%, MgO < 3 wt%, Na₂O/K₂O < 2, Sr > 400 ppm, Y < 18 ppm, Yb < 1.9 ppm, LREE enrichment, non- or slightly negative Eu anomaly, ⁸⁷Sr/⁸⁶Sr_i > 0.704 and ε_Nd(t) <0) (Table 8) as previously reported (Defant & Drummond, 1990; Xiao & Clemens, 2007; Zhang et al., 2002). Nevertheless, some gneissic granite samples yield higher Y and Yb contents relative to C-type adakite (Table 8). A similar phenomenon can be seen on plots of Sr/Y versus Y (Figure 10a) and (La/Yb)_N versus Yb_N (Figure 10b), hence some sample data did not plot in the adakitic areas. The Eu/Eu* values of some samples are lower than 0.8, indicating possibly a small amount of residual feldspar in the source region (Figure 6b; Table 3). A slight feldspar partial melting trend also supports our hypothesis (Figure 11c). On diagrams of Sr versus SiO₂ and Sr versus CaO (Figure 12a,b), the samples plot along the Dabie high Sr/Y granitoids trend (He et al., 2011). Therefore, we suggest that the gneissic granites in the Tongbai Complex belong to high Sr/Y granitoid that probably derived from lower continent crust (He et al., 2011).

The quartz monzonite samples reveal higher Al₂O₃ contents of 16.45–17.08 wt%, Sr contents of 1,475–1,675 ppm and lower SiO₂ contents of 63.00–65.09 wt%, Y of 12.3–15.2 ppm and Yb of 0.73–0.87 ppm as well as barely no Eu anomalies (Eu/Eu* = 0.86–0.93) with respect to the gneissic granites (Figure 6d; Tables 2 and 3). Furthermore, the higher Sr/Y and La/Yb ratios (Figure 12c) with an incompatible feldspar partial melting trend farther manifest that the quartz monzonites are C-type adakite (Figure 10a,b).

Data for quartz monzonite samples are all located in the region of I- and S-type granites, while the data for gneissic granite samples point into the field of A-type granite, which may be due to the gneissic granite undergoing fractional crystallization of feldspar (Figure 10c–f; Wu, Liu, Ji, Wang, & Yang, 2017). Typical S-type granite with mineral assemblages of Crd ± Ms ± Grt ± Sil ± Ilm and typical aluminous A-type granite consisting of Al-rich minerals such as cordierite or muscovite is inconsistent with that of gneissic granite and quartz monzonite in this study (Sang & Ma, 2012). Visible magnetite and haematite as secondary minerals in gneissic granite and quartz monzonite together with high TFe₂O₃ contents (1.43–3.97%), low TiO₂ contents (0.21–0.79%) and low A/CNK (0.95–1.01) prove that they both pertain to I-type granite (or magnetite series) (Chappell, 1999; Sang & Ma, 2012). Moreover, a negative correlation between P₂O₅ and SiO₂ (Figure 5c) and positive correlations of Th and Y contents with Rb contents (Figure 5k, l) are also defined as the I-type granite features (Chappell & White, 1992). Therefore, the gneissic granite and quartz monzonite both belong to I-type granitoid.

6.2.2 Magma source

All gneissic granite and quartz monzonite samples yield low zircon Hf isotopic values and plot in the crustal field (Figure 8a). Quartz monzonites show wider range and higher ε_Hf(t) values between −21.0 and −11.0 relative to the gneissic granites (−20.8 to −18.5), implying more proportion of mantle-derived material may have involved in its petrogenesis. Meanwhile, their Sr–Nd isotopic compositions also hint at this hypothesis in the scatter diagram of ε_Nd(t) and (⁸⁷Sr/⁸⁶Sr)i (Figure 9a). Considering that both gneissic granite and quartz monzonite belong to I-type granites, the relation between ε_Nd(t) and ε_Hf(t) indicates that they are consistent with the composition of the global lower crust (Figure 9b). Analogously, their ²⁰⁸Pb/²⁰⁴Pb and ²⁰⁶Pb/²⁰⁴Pb ratios are all close to the average of lower crust, together with ²⁰⁷Pb/²⁰⁴Pb and ²⁰⁶Pb/²⁰⁴Pb ratios approaching to the mixed fields of lower crust and mantle (Figure 9c,d). Therefore, the lower crust may be a rational source region of magma for both the gneiss granites and quartz monzonites, although mantle-derived materials (such as the mafic rocks in the lower crust source region) likely participate in different proportions. Generally speaking, high Sr/Y granitoids (e.g., adakite) may form by the following ways: (1) magma mixing coupled with crustal assimilation and fractional crystallization involving basaltic magma (Chen et al., 2002); (2) partial melting of thickened lower continental crust under continent subduction/collision orogens (Xu et a1., 2002; Xu, Wang, Wan, Guo, & Pei, 2006) and (3) partial melting of delaminated lower continental crust (He et al., 2011, 2013).

The above discussion has ruled out the hypothesis of magma mixing as a major role. Some previous studies indicate that basaltic magmatic fractional crystallization may produce granitic magmas, but in very small proportions (<10%), or need complementary existing mafic and ultramafic rocks to match them (e.g., Sang & Ma, 2012). As well as the adakites generated from fractional crystallization of parental basaltic magmas, usually are followed by more voluminous other types of rocks (e.g., basalt and andesite) to constitute a lava series (Castillo, Janney, & Solidum, 1999; Rooney, Franceschi, & Hall, 2010). No significant occurrence of mafic-ultramafic or intermediate rocks in the Tongbai Complex rules out conspicuous fractional crystallization involving basaltic magma to gneissic granites, inconsistent with our discussion above. Furthermore, these shallow level processes (e.g., fractional crystallization and contamination) cannot cause their distinctly high Sr/Y and La/Yb ratios (Castillo, 2012). Partial melting experiment results suggest that the high-K melts are mainly controlled by a number of factors, including the source rock K₂O contents, high pressure and dehydration melting (Xiong et al., 2011; Zhang et al., 2002). High K₂O contents of gneissic granite and quartz monzonite manifest that they are inconsistent with the sodium-rich melts by partial melting of basaltic components (Xiao & Clemens, 2007), hence high-K protoliths (e.g., metabasalt, gabbro and eclogite) may act as one of the source rocks (Wang et al., 2007).

If we assume that gneissic granites and quartz monzonites formed by partial melting of a delaminated lower continental crust in the mantle, the delaminated crustal part subsequently interacts with the mantle during upward transport and results in more mantle materials added during their evolution (Xu et al., 2007), then the granitoids formed by this process should all show high MgO, FeO, CaO, Na₂O, Ni, Cr, and Sr contents and relatively low SiO₂ and K₂O contents (e.g., Gao et al., 2004). Both gneissic granites and quartz monzonites, in effect, exhibit lower contents of MgO (0.32–1.16 wt%), CaO (1.22–2.34 wt%), Na₂O (3.98–4.97 wt%), and Cr (<10 ppm) contents (Tables 2 and 3). The MgO (%), Mg^# and Cr (ppm) versus SiO₂ (%) diagrams (Figure 13) reveal that the gneissic granites derived from partial melting of the lower crust while the quartz monzonites originated from eclogite melts.

Garnet can strongly fractionate the HREE from the LREE, distinct positive linear relations of (La/Yb)_N, (Dy/Yb)_N and (Gd/Yb)_N ratios as well as the apparent (Sm)_N depletion in their chondrite-normalized REE patterns reflect garnet as a residual phase during partial melting of both gneissic granitic and monzonitic source rocks (Figure 11f,g) (He et al., 2011). Furthermore, the trends in data points between Nb/Ta and (Dy/Yb)_N suggest not only garnet but also rutile as another residual phase (Figure 11h). Their low Nb/Ta ratios (<15) indicate that more garnet should be involved in the partial melting than rutile, which may be also due to the amphibole in the residual phase (Qian & Hermann, 2013), since Nb is effortlessly incorporated by amphibole over Ta (Foley, Tiepolo, & Vannucci, 2002). As discussed above, plagioclase may not be stable in gneissic granitic and quartz monzonitic source region, so clinopyroxene + rutile + garnet is the most stable mineral assemblage in this situation (e.g., Rapp & Watson, 1995; Rapp, Watson, & MIller, 1991). Moderate low K/Rb (306–389) ratios and the concavity of the middle REE (Figure 6b,d) indicate that residual amphibolite likely existed in their source regions. In summary, rutile ± garnet ± clinopyroxene ± amphibolite may be the residual mineral assemblage of gneissic granites and quartz monzonites. Plagioclase may exist more or less in the gneissic granitic source but none in that of the quartz monzonites.

6.3 Tectonic implications

The gneissic granites in the Tongbai Complex show many similar SiO₂, Al₂O₃, MgO and Sr contents as well as Sr/Y La/Yb ratios but higher Y, Yb, Ni, Cr, and V contents with respect to the Qijianfeng, Caodian and Jigongshan granites (Table 8), indicating more mantle-derived materials may have involved in the gneissic granite petrogenesis, in contrast the Qijianfeng, Caodian, and Jigongshan granites may have originated from an early partial melting of the metasedimentary rocks in the thickened lower crust. Afterward, quartz monzonites generated via partial melting of eclogites (metamorphosed mafic rocks) and yielded the highest Sr contents (1,475–1,675 ppm) relative to other granitoids in the Tongbai region (Table 8). Zhang et al. (2013) divided the granitoids in the Tongbai area into adakites and non-adakites, and Zhang et al. (2018) defined the gneissic granites in the Tongbai Complex as normal calc-alkaline granites. However, it should be noted that the adakite criterion (by Defant & Drummond, 1990) apply only to oceanic subduction rather than continental collision, with which nearly none of the C-type adakite (continental adakite) matches (Moyen, 2009; Xiao & Clemens, 2007). Combined with previous data on granitoids in the Tongbai region (Table 8) (e.g., Niu & Jiang, 2020; Zhang et al., 2013, 2018), we believe that the Early Cretaceous granitoids in the Tongbai area all belong to the C-type adakites and are similar to the high Sr/Y granitoids in the Dabei area (He et al., 2011, 2013). In summary, from gneissic granites through un-deformed granites to the quartz monzonites, the lower crust beneath the Tongbai orogen was gradually thickened, and the lithofacies were transformed from mafic rocks to eclogites.

At present, the formation depth of C-type adakite is still highly debated (e.g., Castillo, 2012; Ma, Zheng, Xu, Griffin, & Zheng, 2015). Partial melting experiment results on adakite from subduction demonstrate a pressures of >15 kbar (Rapp & Watson, 1995; Sen & Dunn, 1994). A similar pressure (and depth) is widely assumed for producing adakitic rocks from lower crust mostly in the Tongbai-Dabie orogens (e.g., He et al., 2011; Xu et al., 2007; Zhang et al., 2013, 2018). A series of melting experiments show that partial melting of mafic rocks will generate adakitic melts at a wide range of pressures (10–38 kbar) (e.g., Rapp, Shimizu, & Norman, 2003; Xiong et al., 2009). In practice, only a depth of >50 km (>15 kbar) may be required for producing adakites from MORB, and the condition of 30–40 km and 800–950°C from an average mafic lower crust composition can generate C-adakites (Qian & Hermann, 2013; Rudnick & Gao, 2003). As above discussion, rutile ± garnet ± clinopyroxene ± amphibolite are the common residual mineral assemblages of the gneissic granite and quartz monzonites, and plagioclase is more or less present in the gneissic granitic residual facies but not in that of monzonites. Combined with the melting experiments results (Qian & Hermann, 2013), the pressure–temperature conditions of 10–12.5 bkar and 800–950°C may be suitable to form the melt of the gneissic granites and that of 12.5–15 bkar and 900–1,000°C may be fit for the quartz monzonites. These overwhelming granites (e.g., Caodian, Jigongshan, and Qijianfeng) in the Tongbai area derived from partial melting of the mafic lower crust, as well as only minor quartz monzonites formed by the partial melting of the lower crust of eclogite, suggest that the lower crust in the Tongbai region is closest to 30–40 km. Previous results of the Moho depth have demonstrated ~35 km​-thick present-day crust beneath the Tongbai orogen, whereas there is a thickened lower crust at a depth of well over 40 km in the Dabie orogen (Li & Wang, 1991; Teng, Deng, Badal, & Zhang, 2014).

Many previous researchers suggested that the thickened lower crust of the Tongbai area was delaminated during the Early Cretaceous (e.g., Chen et al., 2017; Cui et al., 2012; Liu, Gao, et al., 2010; Liu, Hu, et al., 2010; Zhang et al., 2013, 2018), and similar situation happened in the Dabie area at 130 Ma (e.g., He et al., 2011, 2013; Wang et al., 2007). The lower crust delamination of the Dabie area can be accurately defined by two different suites of granitoids: namely the >130 Ma high Sr/Y granitoids (HSG) and the <130 Ma normal granitoids. The HSG shows an emplacement age of 143–130 Ma, with low MgO (<1%) contents and high Sr/Y ratios (>40) and without significant negative Eu, Sr, and Ti anomalies. The normal granitoids were emplaced at 130–117 Ma and characterized by low Sr/Y and (La/Yb)_N ratios and strong negative anomalies of Eu, Sr, and Ti (Table 8). Furthermore, they also show many significant differences such as in the diagrams of Sr contents versus SiO₂ and CaO contents (Figure 12a,b) and Sr/Y versus (La/Yb)_N as well as K₂O versus Na₂O contents (Figure 12c,d). The HSG also shows more limited variation of (⁸⁷Sr/⁸⁶Sr)i and ε_Nd(t) values (Figure 9a), but higher (²⁰⁸Pb/²⁰⁴Pb)_i, (²⁰⁷Pb/²⁰⁴Pb)_i and (²⁰⁶Pb/²⁰⁴Pb)_i in comparison with those of NG in the Dabie orogen (Figure 9c,d). The former is believed to be C-type adakites that derived from partial melting of the thickened lower crust, while the latter are the normal granites that stemmed from the partial melting of the lower crust delamination (Huang, Li, Dong, He, & Chen, 2008; Xu et al., 2007). In conclusion, the granitic magmatism in the Tongbai orogen concentrates on 130–145 Ma, in consistent with the HSG in the Dabie area in terms of chemical composition and geochronology (Figures 9 and 12; Table 8). At present, no normal granites formed by partial melting of the delaminated lower crust are found in the Tongbai area, indicating that the Tongbai orogen did not undergo the same large-scale delamination as the Dabie orogen in the Early Cretaceous (Figure 14).