Abundant mid-Neoproterozoic magmatic rocks are exposed in the western Yangtze block, a part of the South China block. Currently, based on contrasting interpretations of their origin, two competing reconstruction models, the internal model and the external model, have been formulated to constrain the paleogeographic position of the South China block in Rodinia. In this study, we examined 17 representative samples from three intrusions (Kuchahe, Leidashu, and Qizanmi) within the Ailao Shan–Red River belt, located in the southwestern Yangtze block. Laser-ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) zircon U-Pb dating suggests that they were emplaced during the mid-Neoproterozoic (ca. 750 Ma). The Kuchahe and Leidashu intrusions share geochemical characteristics of S-type and A2-type rocks, respectively. Elemental and isotopic data suggest that the Kuchahe intrusion was derived by biotite-dehydration melting of a metapelitic source, whereas the Leidashu intrusion was produced by partial melting of earlier Neoproterozoic granodiorites in the middle crust (∼15 km), with mixing of basaltic magmas. The Qizanmi intrusion is composed of hornblende gabbro and was likely derived by partial melting of a shallow and enriched lithospheric mantle source modified by slab-released fluids, with accumulation of plagioclase and fractionation of mafic minerals. The three intrusions were emplaced in a back-arc setting, in response to the roll-back of a subducted slab rather than mantle plume activity. The mid-Neoproterozoic (ca. 750 Ma) subduction process argues against the South China block occupying an internal setting in Rodinia, which was finally assembled before 0.9 Ga.

In the Proterozoic, the supercontinent Rodinia was assembled during the global Grenvillian orogenesis (e.g., Z.X. Li et al., 2008; Cawood et al., 2017). However, the paleogeographic position of the South China block in Rodinia is still debated. The South China block consists of the Yangtze block and Cathaysia block. The collision between the two blocks gave rise to the Jiangnan orogen (e.g., Z.X. Li et al., 2002b, 2008; Zhao, 2015; Cawood et al., 2017). Abundant Neoproterozoic magmatic rocks occur across the Yangtze block as the result of ancient oceanic subduction, mantle plume activity, continent-continent collision, and/or intracontinental rifting (e.g., Z.X. Li et al., 1995, 2002b, 2003b, 2008; X.H. Li et al., 2002a, 2003a, 2005, 2009; X.L. Wang et al., 2006, 2008, 2013c, 2014; Dong et al., 2012; Wang and Zhou, 2012; W. Wang et al., 2013b; Cai et al., 2015; G. Zhao, 2015; Cawood et al., 2017; J.Y. Li et al., 2018; J.H. Zhao et al., 2008a). Based on competing explanations for these rocks in the South China block, two contrasting models have been proposed (Fig. 1): an internal setting within Rodinia (e.g., Z.X. Li et al., 1995, 2002b, 2003b, 2008; X.H. Li et al., 2002a, 2003a, 2005, 2009) versus an external location along the supercontinental margin (e.g., Zhou et al., 2002; J.H. Zhao et al., 2008a; X.L. Wang et al., 2006, 2008, 2013c, 2014; Wang and Zhou, 2012; Zhang et al., 2012; Cawood et al., 2013, 2017; F. Wang et al., 2013a; Zhao, 2015; J.Y. Li et al., 2018). In the internal model, the amalgamation between the two blocks is considered to have occurred from ca. 1.0 Ga to 0.90 Ga (e.g., Z.X. Li et al., 2002b, 2007; X.H. Li et al., 2009; see Figs. 1A and 1B herein). Subsequent drifting of the South China block from Rodinia triggered by a mantle plume is proposed to have produced intensive magmatism between ca. 850 and 745 Ma and rift-related sedimentary sequences (e.g., X.H. Li et al., 2002a, 2003a, 2005; Z.X. Li et al., 2002b, 2003b, 2008). On the contrary, final assembly between the two blocks indicated by the external model (Fig. 1C) did not occur until ca. 830–810 Ma, and arc magmatism along the western Yangtze block continued to ca. 730 Ma (e.g., Zhou et al., 2002, 2006a, 2006b; Zhao, 2015). Thus, further investigation to determine the precise origin of the Neoproterozoic magmatic rocks of the western Yangtze block is important for constraining the location of the South China block within Rodinia.

In this paper, we investigated Neoproterozoic granitoids (Leidashu and Kuchahe) and mafic rocks (Qizanmi) from the Ailao Shan–Red River belt, located at the southwestern Yangtze block. New laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) zircon U-Pb dating, whole-rock geochemical and isotopic data, and in situ zircon Hf isotopic data were obtained to investigate the petrogenesis and associated geodynamic mechanisms of these rocks. Our approach was based on the understanding that different magmatic assemblages occur in response to continental breakup and oceanic subduction in the Wilson cycle. The main objectives of this work were therefore to constrain: (1) the petrogenesis of the studied Neoproterozoic magmatic rocks, (2) the tectonic setting where they were emplaced, and (3) the paleogeographic location of the South China block in Rodinia. All the new data, combined with published data, suggest that the western Yangtze block was located along a convergent continental margin during the mid-Neoproterozoic, rather than representing a continental rift triggered by a mantle plume.

The South China block consists of the Yangtze block and Cathaysia block. The collision between the two blocks occurred at ca. 1000 Ma (e.g., Z.X. Li et al., 2002b) or between 830 and 800 Ma (e.g., X.L. Wang et al., 2008, 2013c; Zhao, 2015), producing the Jiangnan orogen. Sensitive high-resolution ion microprobe (SHRIMP) zircon U-Pb ages (ca. 968 Ma) of the Xiwan plagiogranites within the NE Jiangxi ophiolites suggest that the Jiang orogen was a part of the Grenvillian orogenic belt (Li et al., 1994). However, new data from magmatic rocks and sedimentary strata imply that the final collision between the two blocks occurred at ca. 830–800 Ma (e.g., Wang et al., 2008; W. Wang et al., 2013b), i.e., substantially younger than the Grenvillian period. LA-ICP-MS U-Pb ages of detrital and magmatic zircons from the volcanic-sedimentary strata in the Jiangnan orogen demonstrate that its folded basement sequences were produced during the period 860–825 Ma (Wang et al., 2014). Wang et al. (2014) considered that the final amalgamation between the two blocks occurred subsequent to ca. 825 Ma.

The Yangtze block features a poorly exposed Archean to Paleoproterozoic basement that is unconformably overlain by variably deformed and metamorphosed Neoproterozoic to Mesozoic volcanic-sedimentary successions (Zhao and Cawood, 2012). The oldest rocks in the Yangtze block are the ca. 3.45 Ga protoliths of the Kongling tonalite-trondhjemite-granodiorite (TTG) gneisses (Guo et al., 2014). The Ailao Shan–Red River belt is located between the South China block and the Indochina block (Fig. 2). It has similar mid-Neoproterozoic magmatism, sedimentary records, and basement as the Yangtze block (Chen et al., 2017). Thus, Chen et al. (2017) argued that the belt should belong to the South China block rather than the Indochina block. The Ailao Shan–Red River belt includes four metamorphic massifs, designated as: Xuelong Shan, Diancang Shan, Ailao Shan, and Phan Si Pan. The basements in the four massifs are Archean to Paleoproterozoic high-grade metamorphic rocks (F. Wang et al., 2013a; W. Wang, 2016a). Paleozoic to Mesozoic rocks lie to the west of the basement rocks. These rocks were generally mylonitized by Cenozoic Ailao Shan–Red River shearing activity (Tapponnier et al., 1990). The Ailao Shan massif has high-grade metamorphic basement (amphibolite facies) with a Mesoproterozoic protolith age. It consists of marbles, calc-silicates, schists, gneisses, amphibolites, and granulites. Paleozoic–Mesozoic low-grade metamorphic strata occur along the Ailao Shan–Red River belt to the southwest (Fig. 3A). They are composed of phyllites, mica schists, and garnet–mica schists.

In this contribution, we investigated the Kuchahe, Leidashu, and Qizanmi intrusions, which were emplaced into the high-grade metamorphic rocks (Fig. 3B). All three intrusions exhibit significant mylonitization. The Kuchahe intrusion consists of gray-white, medium- to fine-grained two-mica granite (Fig. 4) with a composition of quartz (25%–35%), K-feldspar (30%–40%), plagioclase (10%–20%), muscovite (5%–10%), and biotite (∼3%), as well accessory minerals such as zircon and apatite. The Leidashu intrusion is composed mainly of gray quartz monzonite porphyry (Fig. 4). The phenocrysts (15%–20%) are composed of K-feldspar and plagioclase, whereas the groundmass (80%–85%) consists of quartz (10%–15%), plagioclase (30%–55%), K-feldspar (10%–40%), biotite (3%–15%), and amphibole (∼5%). Zircon, sphene, and apatite are common accessory minerals. The Qizanmi intrusion contains medium- to fine-grained hornblende gabbro (Fig. 4) and consists of hornblende (40%–50%) and plagioclase (50%–60%), with minor amounts of sphene and apatite as common accessory minerals. It is worth noting that some plagioclase is locally altered to sericite and zoisite. Seventeen representative samples from the three intrusions were sampled for analysis, and we present detailed sampling locations in Figure 3B.

All samples were subjected to bulk geochemical and Sr-Nd isotopic analyses. Four samples (10AL15-1, 11AL12-1, 13AL03-1, and 13AL01-1) were selected for zircon U-Pb dating and in situ Hf isotopic analysis. All analytical methods are described in Appendix A in the GSA Data Repository Item1, and the results from these samples are summarized in Tables 1 and 2 and in Data Repository Table DR1.

LA-ICP-MS Zircon U-Pb Dating

Zircons separated from our samples were 50–200 µm in length and exhibited regular or irregular oscillatory zoning with no apparent core-rim structure (Fig. 5). Their Th/U ratios were in the range of 0.1–1.2 (Data Repository Table DR1), consistent with those of magmatic zircons (Williams et al., 1996). U-Pb dating of the four selected zircon samples all yielded concordant results with weighted mean 206Pb/238U ages as follows: 748 ± 4 Ma (1σ, mean square of weighted deviates [MSWD] = 0.4; n = 22) for 11AL15–1 (22°53.90′N, 103°12.12′E), 750 ± 4 Ma (1σ, MSWD = 0.6; n = 22) for 11AL12–1 (22°50.65′N, 103°12.18′E), 750 ± 4.2 Ma (1σ, MSWD = 1.9; n = 21) for 13AL03–1 (22°49.54′N, 103°11.29′E), and 751 ± 5 Ma (1σ, MSWD = 1.9; n = 19) for 13AL01–1 (22°50.95′N, 103°11.48′E; Fig. 5). These data suggest that the studied felsic and mafic intrusions have uniform mid-Neoproterozoic emplacement ages.

Bulk Geochemistry

Kuchahe Intrusion

The Kuchahe granite has high SiO2 contents of 73.91–75.18 wt% (Table 1). All the samples are strongly peraluminous and are enriched in alkali contents (K2O + Na2O = 7.66–8.26 wt%), with alumina saturation index A/CNK [= molar Al2O3/(CaO + Na2O + K2O)] > 1.10 (Fig. 6A). These samples exhibit similar Mg# [= atomic Mg/(Mg + FeT)] (0.33–0.39) to experimental partial melts of metasedimentary rocks under continental pressure-temperature (P-T) conditions (Fig. 6B). In chondrite-normalized rare earth element (REE) patterns, these rocks are enriched in light (L) REEs relative to heavy (H) REEs, with negative Eu anomalies (Fig. 7A). In primitive mantle (PM)–normalized spidergrams, they exhibit negative anomalies in Nb, Ta, La, Ce, Sr, Ba, and Ti (Fig. 7B).

The Kuchahe samples exhibit slightly negative εNd(t) values of –0.8 to –0.2 (Table 2). Given that these sample have high Rb/Sr ratios (≥ 1), their initial Sr isotopic ratios were not analyzed.

Leidashu Intrusion

The Leidashu intrusion has intermediate compositions, with SiO2 contents ranging from 60.76 to 63.89 wt% (Table 1). They plot in the syenite, monzonite, and quartz monzonite fields in the diagram of SiO2 versus K2O + Na2O (Fig. 4). Most samples are metaluminous to weakly peraluminous and show similar Mg# values to the Kuchahe intrusion (Fig. 6). The Leidashu intrusion has similar REE and trace-element patterns to the Kuchahe intrusion (Fig. 7A). Compared to the Kuchahe intrusion, the Leidashu intrusion exhibits higher REE and high field strength element (HFSE) contents and lower Ba and Pb contents (Fig. 7B). All samples are characterized by high alkali contents (K2O + Na2O = 7.05–9.44 wt%), with high 10,000*Ga/Al ratios (mostly greater than 2.6) as well as Zr + Nb + Y + Ce greater than 350 ppm (Table 1).

The Leidashu intrusion exhibits negative εNd(t) values of –4.1 to –3.5 (Table 2; Figs. 8A and 8B). Its 87Sr/86Sri values range from 0.7047 to 0.7061 (Table 2; Fig. 8A).

Qizanmi Intrusion

The Qizanmi intrusion is mafic, with low SiO2 contents of 48.2–49.7 wt% (Table 1). It has low TiO2 (0.21–0.56 wt%) and FeOT (4.0–9.4 wt%) contents and relatively high MgO contents (6.3–9.6 wt%) and Mg# values (Fig. 6B). All the samples are enriched in LREEs relative to HREEs, with positive Eu anomalies (Fig. 7A). In PM-normalized spidergrams, all the samples exhibit notably negative Nb-Ta, La, Ce, Pr, Zr-Hf, and Ti anomalies and positive Ba, K, Pb, Sr, and Eu anomalies (Fig. 7B). Compared to mid-ocean-ridge basalt (MORB), the Qizanmi intrusion is more depleted in REEs (except for La and Ce) and HFSEs, but more enriched in large ion lithophile elements (LILEs; Fig. 7B).

These mafic rocks have low initial 87Sr/86Sr values of 0.7040–0.7051 and positive εNd(t) values of +0.4 to +1.7 (Table 2; Fig. 8A).

Zircon In Situ Hf Isotope

We carried out in situ Hf isotopic analyses on those zircons that were analyzed for the LA-ICP-MS zircon U-Pb dating. The results are shown in Data Repository Table DR1 (see footnote 1). The zircons from the Kuchahe intrusion have uniform initial εHf values of +3.7 to +4.8 (Data Repository Table DR1; Fig. 8C). The Leidashu intrusion also has uniform initial εHf values, ranging from +0.6 to –1.5 (Data Repository Table DR1; Fig. 8C). The zircons from the Qizanmi hornblende gabbro have positive initial εHf values of +4.5 to +6.0 (Data Repository Table DR1; Figs. 8B and 8C).


Kuchahe Intrusion

The Kuchahe intrusion consists mainly of leucogranite and contains a small amount of biotite (<5%), with low FeOt + MgO + TiO2 contents (<2 wt%; Fig. 9). It is strongly peraluminous with high A/CNK values (>1.1; Fig. 6A). All samples contain muscovite, which is highly aluminous. The muscovites exhibit subhedral shapes. They are not enclosed by or enclosing K-feldspar or cordierite, which can break down to form muscovite through subsolidus alteration. All the petrographic criteria indicate that the muscovite in the Kuchahe intrusion exists as a primary magmatic phase (Barbarin, 1996). The relatively low FeOT/MgO ratios indicate that the Kuchahe intrusion is not a highly fractionated granite (Fig. 10). Taken together, the geochemical signatures suggest that the Kuchahe intrusion can be classified as an S-type granite (Barbarin, 1996; Sylvester, 1998; Jiang and Zhu, 2017). Several models have been formulated with regard to the origin of peraluminous S-type magmas:

  • (1) Some peraluminous granites were produced by fractional crystallization (FC) of metaluminous magmas in closed systems, with fractionation of amphibole and clinopyroxene (Cawthorn and Brown, 1976). They are a minor component of granitoid suites that are dominantly metaluminous. For the Kuchahe intrusion, the associated metaluminous rocks are not exposed in the studied area. The significant fractionation of amphibole would produce positive Eu anomalies and result in decreasing Dy/Yb and increasing La/Yb with increasing SiO2. The presence of concave REE patterns is diagnostic of prominent amphibole fractionation. We do not observe these geochemical signatures (Fig. 7A). Thus, the Kuchahe intrusion was not derived by the FC model.

  • (2) Experimental studies demonstrate that partial melting of quartz amphibolite at P > 10 kbar can produce strongly peraluminous granitic magmas (Fig. 6A). Sylvester (1998) suggested that the CaO/Na2O ratio is a better indicator to constrain the sedimentary source component of strongly peraluminous granites. This is because the ratio is dominantly controlled by the plagioclase/clay ratio of the sedimentary source. Compared to strongly peraluminous melts derived from plagioclase-poor, clay-rich sources (pelites), those originating from plagioclase-rich, clay-poor sources (e.g., psammites) have higher CaO/Na2O ratios (> 0.3). In amphibolites, amphibole and plagioclase are dominant phases. They have high CaO and Na2O contents and CaO/Na2O ratios. Thus, we suggest that strongly peraluminous melts derived from amphibolites also have high CaO/Na2O ratios. This is supported by the experimental data on a quartz amphibolite (Patiño Douce and Beard, 1995; Fig. 9A). However, the extremely low CaO/Na2O ratios of our samples preclude amphibolites as their source.

  • (3) The consensus view is that strongly peraluminous granitic magmas are most easily produced by partial melting of metasedimentary rocks (e.g., Barbarin, 1996; Sylvester, 1998; Jiang et al., 2011; Jiang and Zhu, 2017). The initial Nd-Hf isotopic data of the Kuchahe samples also plot in the field of continental sediments (Fig. 8). Experimental studies on metagraywackes and metapelites suggest that the resulting partial melts are dominantly peraluminous (Fig. 6A). The low CaO/Na2O ratios of our samples mirror those of experimental melts of a natural metapelitic rock (Patiño Douce and Johnston, 1991). This implies that the Kuchahe intrusion might have originated from a metapelitic source (Fig. 9A). Previous studies have demonstrated that leucogranitic melts are generally formed by the following melting reactions, including: (1) 9 muscovite (Ms) + 15 plagioclase (Pl) + 7 quartz (Qtz) + xH2O = 31 Melt (fluid-present melting at < 750 °C), (2) 22 Ms + 7 Pl + 8 Qtz = 5 K-feldspar (Kfs) + 5 aluminosilicate (Als) +2 biotite (Bt) + 25 Melt (fluid-absent melting at 680–770 °C), and (3) Bt + Als + Pl + Qtz = garnet (Grt) + Kfs + Melt (fluid-absent melting at higher temperature (760–830 °C; Inger and Harris, 1993; Patiño Douce and Harris, 1998; Vielzeuf and Holloway, 1988; Knesel and Davidson, 2002). In amphibolite-facies metapelites, Pb is dominantly enriched in muscovite and then its breakdown can release large amounts of Pb. The resulting restite assemblage (Qz + Bt + Sil ± Pl ± Grt ± Kfs) can retain minor amounts of Pb but much Ba because both biotite and K-feldspar in the restite contain very high Ba contents (Finger and Schiller, 2012). Finger and Schiller (2012) also proposed that low-T and low-degree anatectic events, such as muscovite-dehydration melting or fluid-present melting of muscovite, could strongly enrich Pb relative to Ba in the resultant partial melts. In contrast, during higher-T and larger-degree partial melting, breakdown of biotite will lead to less enrichment of Pb and less depletion of Ba. The Kuchahe leucogranites have relatively low Pb contents at a given Ba content (Fig. 9C). All samples are similar to the high-T S-type granites from the Lachlan fold belt and the European Variscides, which have been regarded as products of biotite breakdown reactions (Clemens and Wall, 1981; Chappell et al., 2004; Finger and Schiller, 2012). We thus conclude that the Kuchahe intrusion was produced by higher-T and larger-partial melting reactions. This is also reconciled with the fact that the zircon saturation temperatures of the Kuchahe samples are distributed between 759 °C and 782 °C (Fig. 10D).

The samples from the Kuchahe intrusion have similar Mg# as the experimental melts (Fig. 6B), reflecting insignificant mantle contribution. This speculation is supported by their extremely low compatible element contents (Cr = 1.62–2.05 ppm, Ni = 0.78–1.05 ppm, and V = 6.34–6.72 ppm; Table 1). Furthermore, basalt mixing will lead to a marked increase in FeOt + MgO + TiO2 and a marked decrease in SiO2 contents (Fig. 9B). The low FeOt + MgO +TiO2 and high SiO2 contents also rule out a mixing process. These samples have high εNd(t) of -0.8 to -0.2 and positive εHf(t) of +3.7 to +4.8 (Fig. 8). However, as discussed already, these isotopic signatures cannot be attributed to direct input of mantle-derived magmas. The Mesoproterozoic Hf model ages of 1.18–1.12 Ga indicate that the Kuchahe intrusion was associated with late Mesoproterozoic juvenile crust. The enrichment in LREEs and LILEs relative to HFSEs and HREEs is a common feature of arc magmas (Taylor and McLennan, 1995). The arc-like trace-element patterns of the Kuchahe samples were likely inherited from preexisting Mesoproterozoic arc magmatic rocks. The ca. 1066 Ma (based on SHRIMP zircon U-Pb dating) Shimian ophiolite supports the inference that a late Mesoproterozoic arc system had developed along the western margin of the Yangtze block (Hu et al., 2017). The Mesoproterozoic juvenile arc crust is further evidenced by the positive Nd-Hf initial isotopic components in other Neoproterozoic magmatic rocks distributed along the same region (X.H. Li et al., 2005; X.F. Zhao et al., 2008b; Zhao and Zhou, 2008; Lai et al., 2015; Xu et al., 2016). The weathering and erosion of the Mesoproterozoic arc crust could then have generated the juvenile arc–derived sediments that were subsequently buried to form the metapelitic source of the Kuchahe intrusion. Similarly, the Neoproterozoic Xucun (ca. 823 Ma), Shi’ershan (ca. 779 Ma), and Jiuling (ca. 819 Ma) granites from the eastern part of the Jiangnan orogen have positive zircon εHf(t) values, with juvenile Mesoproterozoic Hf model ages (Zheng et al., 2008; X.L. Wang et al., 2013c). They have been interpreted as products of reworking of late Mesoproterozoic juvenile crust (Zheng et al., 2008; X.L. Wang et al., 2013c). To sum up, the Kuchahe intrusion was likely produced by biotite-dehydration melting of a metapelitic source that had been formed by the burial of juvenile arc–derived sediments.

Leidashu Intrusion

The Leidashu intrusion consists of monzonite, quartz monzonite, and syenite. Petrographically, these samples contain xenomorphic amphibole and interstitial biotite. Geochemically, they are enriched in alkalis (K2O + Na2O = 7.05–7.83 wt%), REEs (except for Eu), and HFSEs, but they are depleted in Ba, Sr, and Eu. All samples exhibit high zircon saturation temperatures (833–850 °C), and elevated 10,000*Ga/Al ratios (mostly >2.6), as well as high Zr + Nb + Y + Ce contents (> 350 ppm), which are indicative of A-type granitoids (Fig. 10). Currently, two main petrogenetic models have been proposed for the formation of A-type granitoids: (1) derivation from transitional to alkaline oceanic-island basalt (OIB)–like mafic and intermediate magmas with or without significant crustal contamination (e.g., Bonin and Giret, 1990; Turner et al., 1992; Eby, 1992; Bonin, 2007), or (2) partial melting of specific crustal sources, such as residual meta-igneous (granulitic) rocks, newly underplated crust, and calc-alkaline granitoids (e.g., Collins et al., 1982; Creaser et al., 1991; Eby, 1992; Shellnutt and Zhou, 2007; Patiño Douce, 1997). Furthermore, Poitrasson et al. (1995) proposed that mantle-derived magmas mixed with a crustal source could also produce A-type magmas.

Eby (1990) suggested that any specific suite of A-type granitoid suites has a relatively constant Y/Nb ratio, and the ratio is insignificantly disturbed by most magmatic processes such as fractional crystallization and partial melting. Generally, A1-type melts, derived from an OIB-like mantle source, exhibit low Y/Nb ratios (<1.2), whereas A2-type melts, coming from a crustal source, have relatively high Y/Nb ratios (>1.2; Eby, 1990, 1992). In this study, all the Leidashu samples have high Y/Nb ratios of 1.5–3.2, consistent with those of A2-type granitoids (Fig. 10C), indicating a crustal source. The crustal source is also inferred from the negative εNd(t) values of −4.1 to -3.5 and negative Nb-Ta-Ti anomalies (Fig. 7). The contrasting Sr-Nd-Hf isotopic components between the Leidashu intrusion and the coeval Qizanmi mafic intrusion also rule out a mantle source (Fig. 8). More importantly, incompatible elements of the Qizanmi intrusion are extremely depleted, especially HFSEs, which are even lower than MORB. Such depleted melts cannot be reconciled with high absolute abundances of HFSEs of A-type melts (Fig. 7B). In addition, differentiation of alkali basalts with high incompatible element contents is more likely to form peralkaline melts rather than metaluminous A-type granites (e.g., Bonin, 1986; Shellnutt and Zhou, 2007). Thus, we prefer the interpretation that the Leidashu intrusion mainly originated from a crustal source.

Most of the Leidashu samples are metaluminous to weakly peraluminous, inconsistent with the strongly peraluminous partial melts of metasedimentary rocks (Fig. 6A). In addition, the Leidashu intrusion has different elemental and isotopic signatures from the Kuchahe intrusion (Figs. 69). Based on these observations, we conclude that the Leidashu intrusion was not derived from a metasedimentary source. Several (meta-)igneous sources have been proposed for A-type granitoids, including: newly underplated crust (Jahn et al., 2000; Shellnutt and Zhou, 2007), residual meta-igneous (granulitic) rocks (Collins et al., 1982), and calc-alkaline granitoids (tonalite and granodiorite; Creaser et al., 1991). The newly underplated crust refers to those igneous rocks formed by magmatic underplating. They have short “crustal residence times” and exhibit PM-like isotopic signatures, with positive εNd(t) and εHf(t) and low 87Sr/86Sri (Arndt, 2013). Partial melts derived from a crustal source will inherit its isotopic component. However, the crust-like isotopic signatures of the Leidashu intrusion suggest that it was unlikely to have resulted from a newly underplated crust (Figs. 8A and 8B). Alternatively, Collins et al. (1982) proposed that A-type granites might originate from a refractory and residual granulitic source that had previously undergone magmatic extraction. Experimental studies have revealed that refractory granulitic residues produced by partial melting of both metasedimentary and meta-igneous rocks are markedly depleted in Na2O + K2O relative to Al2O3 (Patiño Douce and Beard, 1995). In addition, it was argued that remelting of the residues could not yield A-type granitic rocks, as the latter have high (Na2O + K2O)/Al2O3 ratios (Creaser et al., 1991). Meanwhile, experimental and modeling results have indicated that A-type granites could be derived by partial melting of tonalitic to granodioritic igneous rocks at both low pressure (P = 4 kbar) and high temperature (T = 950 °C; Creaser et al., 1991; Patiño Douce, 1997). This model is bolstered by the initial Sr-Nd-Hf isotopic similarity between the Leidashu intrusion and the earlier Neoproterozoic granodiorites (810–753 Ma) from the Ailaoshan tectonic zone (Figs. 8A and 8B). Compared to the A-type melts from experiments (Patiño Douce, 1997), the Leidashu samples have lower SiO2 and higher Mg# and CaO/Na2O, as well as FeOt + MgO + TiO2, indicating mixing of basaltic magmas (Figs. 6 and 9). Furthermore, the high HREE (Yb = 2.97–4.91 ppm) and Y (38.78–54.71 ppm) contents, coupled with relatively flat HREE patterns (Fig. 7A), suggest that garnet was not a residual phase after partial melting. The markedly negative Eu and Sr anomalies support the presence of plagioclase as a residual mineral phase (Fig. 7). Altogether, these geochemical signatures also lend credence to a relatively shallow crustal source. The high zircon saturation temperature (833–850 °C) requires additional heat from the mantle for crustal melting (Fig. 10D). The occurrence of the coeval basic intrusion (Qizanmi) suggests that crustal melting might have been triggered by the intrusion of mantle-derived magmas. Based on all the evidence presented here, the Leidashu intrusion was likely derived through partial melting of earlier Neoproterozoic granodiorites in the middle crust (∼15 km), with mixing of basaltic magmas.

Qizanmi Intrusion

All the samples from the Qizanmi intrusion have basaltic components, with low SiO2 contents that range from 48.2 to 49.7 wt% (Table 1). The intrusion exhibits relatively constant bulk Sr-Nd isotopic ratios, implying that crustal contamination was insignificant. Furthermore, the absence of a negative correlation between incompatible trace-element ratio (Th/Nb) and εNd(t) argues against significant contamination. The insignificant crustal contamination is also indicated by the markedly negative Zr-Hf anomalies in the spider diagrams (Fig. 7B).

The Qizanmi intrusion has variable Mg# (0.59–0.76 wt%), FeOT (4.0–9.4 wt%), MgO (6.3–9.6 wt%), and compatible element contents (Cr = 66–293 ppm; Ni = 71–304 ppm), which suggest differentiation of mafic minerals. Meanwhile, the positive Eu and Sr anomalies provide evidence for accumulation of plagioclase (Fig. 7A). Relative to MORB, the lower REE (except for La and Ce) and HFSE contents of the Qizanmi intrusion can be also attributed to accumulation of plagioclase, because these elements are incompatible in plagioclase.

Incompatible-element ratios and Sr-Nd-Hf isotopic ratios are not generally disturbed by fractionation crystallization and mineral accumulation. Thus, these ratios can be used to constrain the nature of the source. All the samples are evidently enriched in LREEs relative to HREEs (Fig. 7A), with high La/Ybn ratios between 2.16 and 6.90. Also, the 87Sr/86Sri (0.7040–0.7051) ratios and εNd(t) (+0.4 to +1.7) values of these samples are found to differ significantly from those of MORB (Fig. 8A). The low HFSEs contents and Nb-Ta, La-Ce, Zr-Hf, and Ti depletions in the PM-normalized spidergrams are also not reconciled with OIB values (Sun and McDonough, 1989). The relative depletion of these HFSEs has been interpreted as an indicator of a subduction process (Thirlwall et al., 1994). The crust-like geochemical characteristics of our samples, such as their negative Nb-Ta-Ti and Zr-Hf values, as well as positive Pb anomalies (Fig. 7B), indicate a subcontinental lithospheric mantle source that had been modified by subducted slab–released components (fluids or melts). Furthermore, both HFSEs and REEs (e.g., Th, Y, Zr, and Nb) are more immobile in fluids than in melts (Kepezhinskas et al., 1997; Zhao and Zhou, 2007). Consequently, the La-Ce and Zr-Hf depletions in the trace-elements patterns demonstrate the introduction of slab-derived fluids rather than melts into the overlying lithospheric mantle source (Fig. 7B; Zhao and Zhou, 2007). This is also compatible with the presence of abundant hornblende (water-bearing mineral) in the mineral assemblages.

REEs ratios are useful for constraining formation depth of basaltic magmas (McKenzie and O’Nions, 1991). In this study, the low (Dy/Yb)n ratios (1.30–1.43), coupled with flat HREEs patterns (Fig. 7A), suggested that the Qizanmi basaltic rocks likely originated from the spinel stability field (Miller et al., 1999; Liu et al., 2015). As further evidence, the diagram of (La/Sm)n versus (Sm/Yb)n also demonstrates that <15% nonmodal equilibrium melting of spinel lherzolite + minor garnet lherzolite can produce the Qizanmi melts (Fig. 11). This reflects a melting depth located in the field ranging from spinel into the spinel-garnet transition zone (∼60–80 km; Robinson and Wood, 1998). Such a shallow depth is reconciled with the lithospheric source. Taken together, the Qizanmi intrusion was likely generated by partial melting of a shallow enriched lithospheric mantle source that had been modified by slab-released fluids, with accumulation of plagioclase and fractionation of mafic minerals.

Tectonic Environment

The Qizanmi intrusion is depleted in incompatible elements and exhibits negative Nb-Ta-Ti anomalies, which contradict the geochemical signatures of a plume-related OIB (Fig. 7B). Instead, its mantle source had been strongly modified by subducted slab–released fluids. Similarly, the Panzhihua (738 ± 23 Ma) and Gaojiacun (825 ± 12 Ma) hornblende gabbros in the western Yangtze block have also been identified by Zhao and Zhou (2007) and Zhu et al. (2006). Furthermore, the occurrence of the Neoproterozoic (760–745 Ma) adakites from the Kangding-Panxi area, sourced from an oceanic slab, indicates that modified lithospheric mantle was associated with the coeval subduction rather than ancient subduction (Zhou et al., 2006a, 2006b; Zhao and Zhou, 2007).

The Leidashu intrusion shares geochemical affinities with A-type granitoids. Similarly, the ca. 780 Ma Mianning granite from the western Yangtze block also exhibits A-type geochemical characteristics (Huang et al., 2008). A-type rocks are generally emplaced in extensional tectonic environments such as a continental rift, within-plate setting, postorogenic or back-arc setting, generally postdating calc-alkaline arc magmatism, which occurs mainly at active continental margins (Eby, 1990, 1992). The high zircon saturation temperatures (833–850 °C) coupled with shallow depth of the crustal source also indicate an extensional environment. Given that an arc system was developing along the western Yangtze block during the mid-Neoproterozoic, the Leidashu intrusion was most likely emplaced in a back-arc setting, in response to the oceanic subduction.

The Kuchahe intrusion has strongly peraluminous leucogranites and contains primary muscovite that is highly aluminous. In general, peraluminous granitoids can be divided into two groups based on mineralogy, rock type, and petrogenesis, including muscovite-bearing peraluminous granitoids (MPGs) and biotite-rich, cordierite-bearing peraluminous granitoids (CPGs; Barbarin, 1996). MPGs contain primary muscovite and consist of monzogranites and leucogranites. Evidently, the Kuchahe intrusion belongs to MPGs. MPGs have been postulated to result from “wet” anatexis of crustal rocks, which is associated with major crustal shear or overthrust structures in an orogenic belt (Barbarin, 1996). Transcurrent thrusts and shear zones formed during continental collision can provide not only heat to initiate melting, but also water to enhance it. They are the best places to generate MPGs. Similarly, MPGs from the Qinling orogen (central China) and the Western Kunlun orogen (NW China) were emplaced during a syncollisional orogenic event that temporally followed arc magmatism (Jiang et al., 2010, 2013). Recently, Jiang et al. (2011) and Jiang and Zhu (2017) identified four late Mesozoic MPGs (Ehu, Xinfengjie, Jiangbei, and Dadu intrusions) from the South China block, which were generated in a back-arc or arc setting associated with the paleo-Pacific subduction, indicating that MPGs are not restricted to orogenic belts. As mentioned above, the Kuchahe intrusion was formed by biotite-dehydration melting of a metapelitic source at relatively high temperature (760–830 °C). The high-T anatexis is consistent with a tensional environment in a back-arc basin. The back-arc setting is also supported by the volcano-sedimentary sequence of the Yanbian Group, containing shale, chert, and volcaniclastic sandstone (Zhou et al., 2006a, 2006b; Zhao and Zhou, 2007). Taken together, these results prompted us to conclude that our investigated intrusions were also emplaced in a back-arc setting. Ongoing extension triggered by the slab roll-back would lead to the progressive thinning of lithosphere, and the subsequent asthenospheric upwelling would initiate partial melting of the overlying lithospheric mantle and crustal rocks, leading to the formation of the Qizanmi hornblende gabbros as well as the Kuchahe and Leidashu felsic intrusions.

Location of the South China Block in Rodinia

Currently, two contrasting reconstruction models, the internal model (e.g., Z.X. Li et al., 2002b; Z.X. Li et al., 2008) and the external model (e.g., Zhou et al., 2002; Wang et al., 2006; Cawood et al., 2013, 2017; Zhao, 2015), have been formulated to constrain the paleogeographic position of the South China block in Rodinia. The two models are based on diverging interpretations on the origin of the Neoproterozoic magmatic rocks (<0.9 Ga) in the Yangtze block. In addition, the internal model, unlike the alternative, considers the South China block as a key link between Laurentia and Australia. In contrast, the external model indicates a close link among India, South China, and Australia (e.g., Zhou et al., 2002; J.H. Zhao et al., 2008a; X.L. Wang et al., 2006, 2008, 2013c, 2014; Cawood et al., 2013, 2017; Zhao, 2015).

In this work, detailed geochemical data indicate that the Qizanmi hornblende gabbros were most likely derived from an enriched lithospheric mantle source infiltrated by oceanic slab–released fluids. Furthermore, the association of A-type (Leidashu) and S-type (Kuchahe) intrusions reflects roll-back of the oceanic slab (Fig. 12). All the investigated mafic and felsic intrusions (ca. 750 Ma) were emplaced in a back-arc setting during lithospheric extension. Similarly, a magmatic peak (755–748 Ma) in the Seychelles (western Greater India) has been identified by Tucker et al. (2001), producing granites, quartz diorites, and dolerites. They were also likely emplaced in a back-arc setting, in response to an ancient oceanic subduction event (Tucker et al., 2001). These coeval magmatic activities indicate a setting for South China adjacent to India during the mid-Neoproterozoic. Furthermore, the Neoproterozoic magmatic rocks exposed in the Madagascar, Seychelles, and NW India exhibit similar emplacement ages (850–700 Ma) and geochemical characteristics to the ones from the western and northwestern Yangtze block (e.g., Tucker et al., 2001, 2014; Ashwal et al., 2002; Archibald et al., 2016). All the magmatic rocks are considered to have made up a continuous continental arc (Zhou et al., 2006a, 2006b; Cawood et al., 2017). By analyzing Early Cambrian small shelly fossils from Yangtze, Tarim, Iran, India, Kazakhstan, Australia, West Avalonia, and Siberia, Steiner et al. (2007) recognized close relationships among the Tarim, Iran, India, and Yangtze blocks. Paleomagnetic data also arranged the South China block at the India-Australia edge (Powell and Pisarevsky, 2002; Yang et al., 2004). In addition, latest Neoproterozoic rocks from the Lesser Himalaya (NW India) and Yangtze block exhibit remarkably similar carbonate platform architecture, facies assemblages, and karstic unconformities at equivalent sedimentary sequence positions (Jiang et al., 2003). These similarities indicate that the Yangtze block may have been located close to NW India during the late Neoproterozoic.

In contrast, the internal model requires the involvement of the South China block in the assembly of Laurentia and Australia-Antarctica from 1.1 Ga to 0.9 Ga, and it considers the Neoproterozoic magmatic rocks (<0.9 Ga) as products of the activities of a superplume that caused the Rodinia supercontinent to disintegrate (e.g., Z.X. Li et al., 2008). Based on stratigraphic correlations and tectonic analysis, the internal model suggests that the Yangtze block of the South China block could have been a continental fragment between Australian-Antarctica and Laurentia during the late Mesoproterozoic to early Neoproterozoic (from 1.1 Ga to 0.9 Ga) assembly of the Rodinia (Z.X. Li et al., 1995, 2002b). In addition, the Cathaysia block of the South China block has been interpreted as a part of a continental strip adjoining western Laurentia before being attached to the Yangtze block at ca. 1.0–0.9 Ga. The internal model indicates that the Yangtze block had undergone marked oceanic subduction to partake in the final assembly of Rodinia. However, the internal model predicts abundant Grenvillian (>0.9 Ga) arc magmatism in the Yangtze block, which is contradicted by what we observed in the current study (Fig. 13A).

In general, a superplume is composed of a cluster of plumes (Ernst and Buchan, 2002). As demonstrated by Ernst and Buchan (2002, 2003), it is difficult to identify ancient plume clusters due to limited understanding of paleocontinental reconstructions. The most key evidence for recognizing a mantle plume is the presence of a large igneous province, which consists mainly of a radiating system of basaltic sills, dikes, layered intrusions, and small volumes of granitic rocks in the Proterozoic due to the erosion of flood basalts (Ernst and Buchan, 2003). Most of these basaltic rocks exhibit OIB-like geochemical signatures, including positive Nb-Ta anomalies and enrichment in incompatible-element contents. In addition, those granites associated with a mantle plume exhibit generally A-type affinities (Eby, 1990, 1992). However, Neoproterozoic igneous rocks exposed around the Yangtze block are dominated by granitoids with arc-like geochemical affinities. Moreover, these granitoids are dominantly I-type and S-type rocks, and secondarily A-type rocks (Fig. 13B; Table 3). Although some mafic rocks exhibit some similar geochemical signatures (enrichment in both LILEs and HFSEs) to continental flood basalts (X.H. Li et al., 2003a; Zhu et al., 2006, 2008; Zhou et al., 2018), they have markedly negative Nb-Ta anomalies, which are diagnostic of a subduction process. We thus theorize that the mid-Neoproterozoic (ca. 750 Ma) western Yangtze was a convergent continental margin rather than a continental rift triggered by a mantle plume. Thus, it does not seem plausible for the South China block to have been located between Laurentia and Australia during the Neoproterozoic. Based on these findings, we conclude that the South China block was located on the margin of the Rodinia supercontinent, adjoining India.

  • (1) The Kuchahe, Leidashu, and Qizanmi intrusions from the western Yangtze block were emplaced in the mid-Neoproterozoic (ca. 750 Ma).

  • (2) The Kuchahe intrusion is strongly peraluminous leucogranite and was likely produced by biotite-dehydration melting of a metapelitic source at relatively high temperature. The Leidashu A-type intrusion was derived by partial melting of earlier Neoproterozoic granodiorites in the middle crust (∼15 km), with mixing of basaltic magmas. The Qizanmi intrusion is hornblende gabbro and was likely derived by partial melting of a shallow enriched lithospheric mantle source that had been modified by slab-released fluids, with accumulation of plagioclase and fractionation of mafic minerals.

  • (3) All the three intrusions were formed in a back-arc basin in response to subducted slab roll-back.

  • (4) The mid-Neoproterozoic subduction process suggests that the South China block was located on the margin rather than in the interior of the Rodinia supercontinent.

This study was financially supported by the National Natural Science Foundation of China (grant 41703022), Fundamental Research Funds for the Central Universities (lzujbky-2018–52), the Plateau Mountain Ecology and Earth’s Environment Discipline Construction Project (grants C176240107), and the Joint Foundation Project between Yunnan Science and Technology Department and Yunnan University (grants C176240210019). We are grateful to Lithosphere Science Editor Laurent Godin and two anonymous reviewers for their thoughtful reviews and constructive comments.

1GSA Data Repository Item 2019129, Appendix A: descriptions of analytical methods, and Table DR1: LA-ICP-MS zircon U-Pb dating data and in-situ Hf isotopic data of the studied intrusions, is available at http://www.geosociety.org/datarepository/2019, or on request from editing@geosociety.org.
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.