Petrogenesis of Early Cretaceous volcanic rocks from the Rena-Co area in the southern Qiangtang Terrane, central Tibet: evidence from zircon U-Pb geochronology, petrochemistry and Sr-Nd-Pb–Hf isotope characteristics

Abstract

The subduction of the Bangong–Nujiang Ocean is important in the geological evolution of the Tibetan Plateau. In this paper, we report new zircon U-Pb age and Lu-Hf isotopic data and whole-rock elemental and Sr-Nd-Pb isotopic data for Early Cretaceous dacites from the Rena-Co area (RCA) in the southern Qiangtang Terrane (QT), central Tibet and use these data to better understand the tectonic evolution of the Bangong–Nujiang suture. LA–ICP-MS dating of zircons yields ages of 109.5 ± 0.6 Ma to 109.6 ± 0.8 Ma for the dacites from the RCA. Geochemically, these dacites are medium-K calc-alkaline and show high SiO₂ contents of 64.79–70.37 wt.%, high Sr contents of 517–598 ppm and low Y contents of 8.45–10.7 ppm, similar to those of typical adakites. Additionally, all the rocks are strongly enriched in light rare earth elements and some large ion lithophile elements (e.g. Rb, U, K and Cs) but significantly depleted in high-field-strength elements (e.g. Nb, Ta and Ti), consistent with the geochemical characteristics of arc-type magmas formed in the subduction zone. Moreover, these adakite-like dacites show whole-rock initial (⁸⁷Sr/⁸⁶Sr)_i ratios of 0.705119 to 0.705491, (²⁰⁶Pb/²⁰⁴Pb)_i ratios of 18.489 to 18.508, (²⁰⁷Pb/²⁰⁴Pb)_i ratios of 15.591 to 15.612, (²⁰⁸Pb/²⁰⁴Pb)_i ratios of 38.599 to 38.686, ϵ_Nd(t) values of −0.28 to +1.25 and single-stage Nd model ages of 642 to 818 Ma, as well as significantly positive zircon ϵ_Hf(t) values of 3.9–13.1, with young Hf-depleted mantle ages of 331 to 923 Ma. These geochemical and isotopic data indicate that they are most likely derived from the juvenile thickened mafic lower continental crust, which contains partial melts of metasomatized peridotite and subduction-related fluids in the magma source region. Based on previous studies and our new data, we propose that the RCA adakite-like dacites are most likely a result of the northwards subduction of the Bangong–Nujiang Ocean lithosphere beneath the southern QT during the Early Cretaceous and that a slab rollback model could explain the formation of the RCA adakite-like dacites.

1. Introduction

The Tibetan Plateau shares a long and complex geological history involving multiple episodes of spreading, subduction and continental collision in the Tethys Ocean basin. The Bangong–Nujiang suture zone (BNSZ), located between the Lhasa terrane to the south and the Qiangtang terrane (QT) to the north, extends more than 2000 km across the central Tibetan Plateau (Fig. 1a). This zone is generally considered the locus of the ancient Bangong–Nujiang Ocean (BNO) (Yin & Harrison, 2000; Kapp et al. 2003, 2005; Pan et al. 2004; Guynn et al. 2006; Zhang et al. 2012; Li et al. 2016; Zhu et al. 2016). The closure of the BNO, followed by the subsequent collision between the Lhasa and Qiangtang terranes, significantly contributed to the initial growth of the Tibetan Plateau before the Indo-Asian collision (Kapp et al. 2007; Shi et al. 2008, 2020; Zhang et al. 2012, 2014; Zhu et al. 2016; Zeng et al. 2021). However, intense controversies remain on the closure of the BNO, and two main versions of the final closure of the BNO have been proposed. The first hypothesis is that the BNO closed between the Late Jurassic and Early Cretaceous (ca. 160-145 Ma), mainly on the basis of the presence of an angular unconformity between the Lower Jurassic oceanic rocks (flysch deposits and ophiolites) and the Upper Jurassic–Lower Cretaceous shallow marine strata (e.g. Xu et al. 1985; Dewey et al. 1988; Yin et al. 1988; Kapp et al. 2003, 2005, 2007; Raterman et al. 2014). The alternative version suggests that the collision between the Lhasa terrane and QT may have occurred after the Early Cretaceous (Ca. 120-100 Ma), according to abundant Early Cretaceous oceanic island basalt (OIB)-type rocks that are overlain by or interbedded with radiolarian-bearing chert and/or limestone, the existence of a mid-Cretaceous magmatic arc in the southern QT and palaeomagnetic data (e.g. Zhu et al. 2006; Bao et al. 2007; Baxter et al. 2009; Zhang et al. 2012, 2014; Xu et al. 2015; Chen et al. 2018; Fan et al. 2018; Luo et al. 2020; Shi et al. 2020).

Figure 1. (a) Tectonic framework of the Tibetan Plateau (modified from Zhu et al. 2009, 2016; Sui et al. 2013); (b) simplified geologic map showing outcrops of magmatic rocks in the RCA, western Qiangtang, central Tibet [modified from Kapp et al. (2005) and Song et al. (2021)]. Age data in this paper are in red font; other literature age data are from Chang et al. (2011), Kapp et al. (2005), Hao et al. (2016 a) and Song et al. (2021).

The Rena-Co area (RCA) is located in the southern margin of the QT to the north of the BNSZ (Fig. 1b). Late Mesozoic volcanic rocks are widely distributed within this area (e.g. Kapp et al. 2005; Chang et al. 2011; Li et al. 2016; Wei et al. 2017), and this area has been a research focus in recent years (Kapp et al. 2005; Chang et al. 2011; Li et al. 2014 a; Hao et al. 2016 a, b; Song et al. 2021). It is well accepted that understanding Cretaceous volcanism in the southern margin of the QT is of great significance for understanding the evolution of the BNO and Qiangtang–Lhasa collisional orogeny (Yin & Harrison, 2000; Kapp et al. 2003, 2005; Zhu et al. 2016). Late Mesozoic volcanic rocks, especially the dacites of the Meiriqiecuo Formation (MF) from the RCA, could play an important role in constraining the tectonic setting at that time. However, the detailed petrogenesis of the Early Cretaceous dacites in the RCA remains unknown.

In this paper, we focus on the MF dacites from the RCA in the southern QT, central Tibet. Here, we present new zircon U-Pb ages, whole-rock major and trace element compositions, and whole-rock Sr-Nd-Pb and in situ zircon Hf isotope data for the RCA dacites. Based on our new data, we provide a new tectono-magmatic model to explain the Early Cretaceous magmatism in the southern QT and constrain the petrogenetic processes of the RCA dacites and the time of the closure of the BNO.

2. Geological background and petrographic observations

The Tibetan Plateau is composed of four continental terranes from south to north: the Himalaya, Lhasa, Qiangtang, and Songpan–Ganzi. These terranes are separated by the Indus–Yarlung Zangbo suture zone (IYZSZ), Bangong–Nujiang suture zone (BNSZ), and Jinshajiang suture zone (JSSZ) (Yin & Harrison, 2000; Shi et al. 2008; Zhu et al. 2016). The BNSZ is 20–95 km wide in a N-S direction and >2200 km long in an E-W direction and represents a geologically significant suture zone that stretches from Bangongco to the west via Gerze, Nyima, Dongqiao, Amdo and Dêngqên to the east and then turns south into the Mogok area of Myanmar (Yin & Harrison, 2000; Kapp et al. 2003, 2005). As the boundary of the LT and the QT, the BNSZ contains abundant Jurassic-Cretaceous flysch, mélange and scattered ophiolitic fragments (Sengör, 1979; Girardeau et al. 1984; Yin et al. 1988; Kapp et al. 2003). To the north of the BNSZ, the southern QT is composed of Carboniferous–Permian shelf strata along with Jurassic marine rocks exposed along its northern and southern margins (Girardeau et al. 1984; Yin & Harrison, 2000; Pan et al. 2004; Zhang et al. 2012). Moreover, many late Mesozoic intermediate–felsic magmatic rocks with ages ranging from 183 to 101 Ma are widely distributed in the southern QT (e.g. Kapp et al. 2005; Chang et al. 2011; Li et al. 2014 a, b, 2016; Hao et al. 2016 a, b; Zhu et al. 2016; Wei et al. 2017, 2018; Song et al. 2021).

The study area in this paper, the Rena-Co area (RCA), is located approximately 50 km northeast of Gerze County, Tibet, on the southern margin of the QT (Fig. 1b). The main strata exposed in the RCA are the middle Permian Longge Formation (P₂ l), Upper Triassic Riganpeicuo Formation (T₃ r), Lower-Middle Jurassic Sewa Formation (J_1-2 s), Lower Cretaceous MF (K₁ m) and Quaternary units. P₂ l is mainly composed of microcrystalline limestone and fine-grained limestone. T₃ r consists mainly of micritic limestone and bioclastic limestone. J_1-2 s is a complex unit of thinly bedded sandstone, medium-bedded quartzose sandstone, quartzofeldspathic sandstone and thinly bedded argillaceous slate. Notably, K₁ m within the area of interest contains MF volcanic rocks, which are mainly dacite with minor andesite and rhyolite. In addition, this area contains several groups of faults and inferred faults across the basin, including NE–SW-trending, E–W-trending and nearly N–S-trending faults and/or inferred faults. The RCA contains several Mesozoic intermediate–silicic intrusions, including granodiorites, diorites and granodiorite porphyries (Fig. 1b).

The K₁ m volcanic rocks are exposed mainly in the southern RCA. K₁ m overlies J_1-2 s and T₃ r across a buried angular unconformity. Our dacite samples analysed in this study were collected from K₁ m west of the Rena-Co area in the southern QT, approximately 35 km north of the BNSZ (Fig. 1b). Petrographically, although our analysed volcanic rocks in the RCA are extensively altered, the original textures appear to have been preserved. The analysed dacites in this study have a porphyritic texture, containing 20–35 vol.% phenocrysts of plagioclase (15–20 vol.%), amphibole (5–10 vol.%) and pyroxene (3–5 vol.%). The matrix is mainly oriented or interleaved plagioclase microlites, cryptocrystalline, Fe–Ti oxides and minor quartz (Fig. 2).

Figure 2. Photomicrographs showing the porphyric texture of the MF volcanic rocks collected from the RNA in the western QT, central Tibet. Note that the groundmass minerals mainly consist of plagioclase microlites, cryptocrystalline minerals and minor magnetite.

3. Results

Analytical methods, whole-rock major and trace elements, Sr-Nd-Pb isotopes and LA–ICP–MS zircon U-Pb and Lu-Hf isotope data for the RCA volcanic rocks are given in Supporting Information I. Herein, we selected the least altered samples for geochemical and isotopic analyses, with their results presented in Supporting Information II.

3.1. Major and trace elements

Whole-rock major and trace element data for the RCA volcanic rocks are presented in Table A of Supporting Information II. Based on field and petrographic observations and loss on ignition (LOI) geochemical data in the analysed samples (1.62–4.44%), with an average of 3.03 wt.%, the analysed dacite samples are relatively fresh. When normalized on a volatile-free basis, the analysed dacite samples in this study are characterized by low TiO₂ contents (0.50–0.59 wt.%), low MgO contents (0.68–0.88 wt.%), low TFe₂O₃ contents (3.97–4.65 wt.%), low Mg^# values (23.83–29.65), as well as relatively high SiO₂ (64.79–70.37 wt.%), Na₂O (3.79–4.70 wt.%), K₂O (1.54–2.46 wt.%) and Al₂O₃ (13.81–16.13 wt.%) contents. The RCA dacite samples plot in the subalkaline dacite field in the SiO₂ vs. Na₂O + K₂O diagram (Fig. 3a) and in the medium-K calc-alkaline series in the K₂O vs. SiO₂ diagram (Fig. 3b), with total alkalis and K₂O/Na₂O ratios ranging from 5.71 to 6.91 and 0.36 to 0.56, respectively. Moreover, these samples share metaluminous characteristics, with A/CNK values (molar Al₂O₃/[CaO+K₂O+Na₂O]) of 0.78–0.94 (Fig. 3c).

Figure 3. Selected plots of the MF volcanic rocks in the western QT, central Tibet. (a) Total alkalis (wt.%) vs. SiO₂ (wt.%) diagram (Middlemost, 1994) for classification, (b) K₂O (wt.%) vs. SiO₂ (wt.%) diagram (Peccerillo & Taylor, 1976), (c) A/NK vs. A/CNK diagram (Maniar & Piccoli, 1989). Analytical data of 33 other Duolong volcanic rocks samples collected from (Wang et al. 2015; Sun, 2015; Li et al. 2016; Wei et al. 2017; Shi et al. 2019) are shown for debate. See text for details.

These analysed rocks show a remarkably right deviation in the chondrite-normalized rare earth element (REE) patterns (Fig. 4a), with very low total REE contents (ΣREE = 98.8–123.1 ppm) and varying enriched light rare earth elements (La_N/Yb_N = 18.48–20.23, where N denotes values normalized to the chondrite composition of Sun & McDonough, 1989). They share slightly negative to negligible Ce anomalies [δCe = 0.89–1.01, δCe = 2×Ce_N/(La_N+Pr_N)] and weakly negative Eu anomalies [δEu = 0.84–0.89, δEu = 2×Eu_N/(Sm_N+Gd_N)]. They exhibit high Sr (517-598 ppm) and low Y (8.45-10.7 ppm) contents, coupled with limited Sr/Y ratios (11.0–13.7). In addition, they generally show strong enrichments in large ion lithophile elements (LILEs: K, Rb, U and Pb) relative to high-field-strength elements (HFSEs: Nb, Ta, and Ti) in primitive-mantle-normalized multielement variation patterns (Fig. 4b), which are similar to the geochemical characteristics of arc-type magmas worldwide (Pearce & Peate, 1995; Wilson, 1989; Hawkesworth et al. 1993).

Figure 4. Chondrite-normalized REE (a, c, e) and primitive-mantle-normalized trace element patterns (b, d, f) for the MF volcanic rocks in the western QT, central Tibet. The date sources are for the MF volcanic rocks the same as in Figure 3. N-MORB and OIB (Sun & McDonough, 1989). Data for normalization and plotting are from Sun & McDonough (1989). Note that the MF volcanic rocks show significantly different patterns from those of N-MORB and OIB.

3.2. Zircon U-Pb ages

The zircon U-Pb dating results are listed in Table B of Supporting Information II. In cathodoluminescence (CL) images, zircons from the studied samples are mostly euhedral to subhedral in morphology. Zircons from sample RNC1610 have long axes of 50 μm to 220 μm and length-to-width ratios of 2:1 to 3:1 (Fig. 5a), while zircons from sample RNC1613 have long axes of 50 to 150 μm as well as length-to-width ratios of 1:1 to 3:1 (Fig. 5b). The CL images show that the zircon grains from these two samples have wide magmatic oscillatory zoning. The Th/U ratios of zircons range from 0.50 to 1.17 in sample RNC1610 and from 0.47 to 1.18 in sample RNC1613. Their generally high Th/U ratios, more than 0.1, indicate a magmatic origin (Corfu et al. 2003; Hoskin & Schaltegger, 2003). Thus, zircon U-Pb ages could be interpreted as representing the eruptive age of host rocks.

Figure 5. Cathodoluminescence (CL) images of representative zircon grains and concordia plots of the MF volcanic rocks in the RCA. Solid and dashed circles indicate the locations of LA-ICP-MS U-Pb dating and Hf analyses, respectively. The scale bar length in CL image is 150μm.

After rejecting discordant ages, the 21 analyses from sample RNC1610 are generally concordant and share zircon ²⁰⁶Pb/²³⁸Pb ages ranging from 108.8 Ma to 110.6 Ma (Fig. 5a), with a weighted mean age of 109.5±0.6 Ma (MSWD=0.02). Similarly, the 12 analyses from sample RNC1613 are also concordant and share zircon ²⁰⁶Pb/²³⁸Pb ages ranging from 108.3 to 110.8 Ma (Fig. 5b), with a weighted mean age of 109.6±0.8 Ma (MSWD=0.01). Therefore, the RCA volcanic rocks were generated in the Early Cretaceous (ca. 109.5-109.6 Ma), similar to the Duolong MF volcanic rocks (105.7-118.5 Ma; Wang et al. 2015; Sun, 2015; Li et al. 2016; Wei et al. 2017; Shi et al. 2019) and the northern Gerze andesites (∼124 Ma; Liu et al. 2012).

3.3. Whole-rock Sr-Nd-Pb isotopic data

Whole-rock Sr-Nd-Pb isotopic data are listed in Table C of Supporting Information II. All samples show homogeneous Sr-Nd-Pb isotopic compositions (Fig. 6). Initial Sr, Nd and Pb isotopic ratios were corrected by the ages of 109.5 Ma and 109.6 Ma. These dacite samples have low ⁸⁷Sr/⁸⁶Sr_(i) values (0.705119–0.705491) and exhibit a restricted range of ϵ _Nd(t) values ranging from −0.28 to +1.25 (corrected to related ²⁰⁶Pb/²³⁸U ages), with young single-stage Nd-depleted mantle ages of TDM1 = 642 to 818 Ma. In addition, these rocks exhibit variable (²⁰⁶Pb/²⁰⁴Pb)_i values of 18.489 to 18.508, (²⁰⁷Pb/²⁰⁴Pb)_i values of 15.591 to 15.612 and (²⁰⁸Pb/²⁰⁴Pb)_i values of 38.599 to 38.686.

Figure 6. Whole-rock Sr, Nd and Pb isotopic compositions of the MF volcanic rocks in the western QT, central Tibet. Analytical Sr and Nd isotopic data of 12 Duolong volcanic rocks samples collected from previous studies (Sun, 2015; Li et al. 2016; Wei et al. 2017) are shown for debate. In Panel a, data of Bangong–Nujiang Ocean basalt (Liu et al. 2014), MORB forming during 400-179 Ma (Hofmann, 2003), the lower crust (Miller et al. 1999), crustal xenoliths in Cenozoic volcanic rocks (Lai & Qin, 2008), Dexing lower crustal derived adakitic rocks (Wang et al. 2006 b),Cenozoic slab-derived adakites (Defant & Kepezhinskas, 2001), the Gerze volcanic rocks (Liu et al. 2012), Slab-derived adakites in Kitakami area and Metasomatized mantle-derived adakites in Yanji area (Tsuchiya et al. 2005), Early Eocene lower crust-derived adakitic rocks in Gümüşhane area (Karsli et al. 2010), Late Miocene mafic lower crust-derived Tavdağı adakites in the Tortum area (Dokuz et al. 2013) and Late Miocene juvenile lower crust-derived adakites in IKizdere area (Karsli et al. 2019) are shown for comparison. In Panel b, data of Tethyan basalts (150 Ma and 120 Ma; Mahoney et al. 1998), Dazhu-Langceling basalts from the Yarlung Zangbo suture zone (Zhang et al. 2005), Indian Ocean pelagic sediment (V28-343, Ben Othman et al. 1989), Yeba mafic rocks (Zhu et al. 2008), and Linzizong andesites (Mo et al. 2006, 2007) are shown for comparison. The star at the top in Panel b is the average value for Langceling basalts (Zhang et al. 2005), and Indian Ocean pelagic sediment is used as a proxy for oceanic sediment. In Panel c and d, NHRL is from Hart (1984), Geochron (4.55 Ga) and the mantle end-members HIMU, EM1 and EM2 are after Zindler & Hart (1986), and Northern Qiangtang adakitic rocks are from Lai et al. (2007). See text for details.

3.4. Zircon Hf isotopes

A total of 33 dated zircons were analysed by LA–MC–ICP–MS for Hf isotopes in this study, and the results of the initial ¹⁷⁶Hf/¹⁷⁷Hf_(i) ratios and ϵ_Hf(t) values are presented in Table D of Supporting Information II. These two analysed dacite samples share highly similar variations in Hf isotopes in zircons. Zircons from these dacites have uniform Hf isotopic compositions (¹⁷⁶Hf/¹⁷⁷Hf_(i) = 0.282813 to 0.283074), yielding ϵ _Hf(t) values and young Hf-depleted mantle ages (T_DMC) of +3.9 to +13.1 and 331 to 923 Ma (Fig. 7), respectively.

Figure 7. Zircon ϵ _Hf(t) vs. U-Pb ages diagram of the MF volcanic rocks in the western QT, central Tibet. Analytical data of 6 Duolong volcanic rocks samples collected from previous studies (Sun, 2015; Li et al. 2016; Wei et al. 2017) are shown for debate. Inset in the diagram shows histograms of ϵ _Hf(t) zircons of the MF volcanic rocks from the RCA.

4. Discussion

4.1. Tectonic setting

The dacite samples, with ages of ca. 109.5–109.6 Ma, show remarkable enrichments in light REEs (La_N/Yb_N = 18.48–20.23) and LILEs (e.g. Cs, Rb, U, Pb and K) but significant depletions in HFSEs (e.g. Nb, Ta and Ti), consistent with the characteristics of arc-type volcanic rocks formed in a subduction zone environment (Fig. 4a, b; Hawkesworth et al. 1993; Pearce & Peate, 1995). In addition, they share similar geochemical signatures to the Early Cretaceous arc-type rocks related to the northern subduction of the Bangong–Nujiang oceanic lithosphere (BNOL), such as andesites from northern Gerze (ca. ∼124 Ma, Liu et al. 2012), Duolong MF volcanic rocks with ages of ca. 105.7–118.5 Ma (Fig. 4c, d; Wang et al. 2015; Sun, 2015; Li et al. 2016; Wei et al. 2017; Shi et al. 2019) and Duolong ore-bearing and barren porphyry with ages of ca. ∼116–128 Ma (e.g. She et al. 2009; Chen et al. 2013; Li et al. 2013 a, 2016; Sun, 2015; Zhu et al. 2015; Wei et al. 2018). Note that the interval between the RCA and Duolong area measures only approximately 65 kilometres spatially, forming a nearly W-S magmatic belt parallel to the BNSZ (Fig. 1). In particular, the MF volcanic rocks from the two sections yield nearly identical ages, considering their error ranges, and have essentially identical REE and trace element patterns (Fig. 4a, b; Kapp et al. 2005; Chang et al. 2011; Wang et al. 2015; Sun, 2015; Li et al. 2016; Wei et al. 2017; Shi et al. 2019; this study). Taken together, these observations show that these two sections probably share the same tectonic setting.

Interpreting the tectonic setting of altered rocks is difficult. The chemistry of some elements, particularly LILEs (e.g. K, Ba, Rb and Sr), has been easily affected by alteration, whereas elements such as Ti, V, Zr, Hf, Nb, Ta, Y, Yb and La are effectively immobile during alteration (Allègre & Minster, 1978; Collins et al. 1982). Therefore, the latter immobile elements should be used in interpreting the tectonic setting of the RCA dacites. When plotted on a discrimination diagram using immobile minor and trace elements, all the MF volcanic rocks from the RCA, together with those from the Duolong area, plot in the arc volcanic setting field on the Hf/3-Th-Nb/16 diagram (Fig. 8a, Wood, 1980); this discrimination diagram could apply to both intermediate–acid rocks and basaltic rocks and is very useful for discriminating arc-related volcanic rocks (Wood et al. 1980; Hugh, 1993). Furthermore, the dacite samples share Ta/Yb ratios of 0.50–0.54, Nb/Yb ratios of 7.35–8.14 and Th/Yb ratios of 7.53–8.59, which plot within or near the active continental margin field rather than the within-plate field (including continental rifts and ocean islands) on the Th/Yb vs. Nb/Yb and Th/Yb vs. Ta/Yb diagrams (Fig. 8b, c; Pearce & Peate, 1995; Agrawal et al. 2008). Note that this is similar to the Andean arc basalts (continental island arc basalts). The Zr/Y ratios (11.02–13.70) of the RCA dacites are higher than the value of 3.0 that was used to separate the oceanic arc (Zr/Y < 3) and continental volcanic rocks (Zr/Y > 3; Pearce, 1982; Pearce et al. 1984; Pearce & Deng, 1988), indicating that the RCA dacites were most likely constructed in a continental island arc setting. Moreover, they share high La/Nb ratios ranging from 3.41 to 3.79 and negative Eu anomalies ranging from 0.84 to 0.89, all of which further reinforce the subduction-related tectonic setting (Condie, 2001). This view is also supported by the fact that these RCA volcanic rocks have high SiO₂ (64.79–70.37 wt.%) and moderate K₂O (1.54–2.46 wt.%) contents, with all the samples plotting in the medium-K calc-alkaline series field (Fig. 3b), given that calc-alkaline volcanic rocks have long been considered the result of subduction-related arc magmatism occurring at convergent plate boundaries (e.g. Li et al. 2013 b, 2015; and references therein).

Figure 8. Selected discrimination diagrams of the MF volcanic rocks in the western QT, central Tibet. (a) Th/Yb-Ta/Yb diagrams (Pearce, 1982); (b) Th/Yb-Nb/Yb diagrams (Pearce & Peate, 1995); (c) Hf/3-Th-Nb/16 diagram (Wood, 1980). The date sources for the MF volcanic rocks are the same as in Figure 3. Note that the MF volcanic rocks were formed in a continental margin setting.

Hence, as presented above, the geochemical features clearly suggest that the MF dacites from the RCA possess an arc volcanic signature that corresponds with subduction, likely suggesting that they formed at a continental margin arc due to the northwards subduction of the BNOL.

4.2. Petrogenesis of the RCA dacites

As outlined above, our MF dacites from the RCA in this study contain plagioclase phenocrysts are characterized by relatively high SiO₂ abundances (64.79–70.37 wt.%), high Al₂O₃ abundances (13.81–16.13 wt.%), high Na₂O abundances (3.79–4.70 wt.%) and high Sr abundances (517-598 ppm), but low Y abundances (8.45-10.7 ppm), low Yb abundances (0.89-1.04 ppm), low Cr abundances (33.3-58.9 ppm) and low Ni abundances (17.6-33.6 ppm), with limited Sr/Y ratios of 11.0–13.7 and La_N/Yb_N ratios of 18.48–20.23, and they plot within the adakite field of Figure 9a. In addition, they yield weakly negative to positive whole-rock ϵ _Nd(t) values (−0.28 to +1.25) and positive zircon ϵ _Hf(t) values (+3.9 to +13.1). Overall, the RCA dacites display adakite-like geochemical characteristics (Fig. 9a, Defant & Drummond, 1990). Thus far, several hypotheses have been proposed to explain the origin of adakitic rocks, including (a) melting of subducted young and hot oceanic slabs (e.g. Stern & Kilian, 1996; Defant & Drummond, 1990; Rapp et al. 1999; Defant et al. 2002); (b) assimilation fractional crystallization (AFC) or fractional crystallization (FC) from parental basaltic magmas (e.g. Feeley & Hacker, 1995; Castillo et al. 1999; Macpherson et al. 2006); (c) magma mixing between felsic and basaltic magmas (e.g. Streck et al. 2007); (d) melting of delaminated lower crust (e.g. Kay & Kay, 1993; Xu et al. 2002; Gao et al. 2004; Wang et al. 2006 a, b); and (e) melting of thickened mafic lower continental crust (e.g. Atherton & Petford, 1993; Hou et al. 2004; Chung et al. 2003, 2005).

Figure 9. Selected discrimination diagrams of the MF volcanic rocks in the western QT, central Tibet. (a) Sr/Y vs. Y (ppm) diagram showing data for adakites and normal calc-alkaline rocks (Defant & Drummond, 1990), (b) Mg^# vs. SiO₂ (wt.%) diagram and (c) Th/La vs. Th (ppm) diagram (Plank, 2005). The date sources for the MF volcanic rocks are the same as in Figure 3. In Panel b, data of metabasaltic and eclogite experimental melts (1-4 Gpa) (Rapp et al. 1999, 2003, and references therein), subducting oceanic crust-derived adakites, delaminated lower crust-derived adakites and thickened lower crust-derived adakites (Wang et al. 2006 a, b, 2008 a, b, and references therein) are shown for debate. In Panel c, data of marine sediments and GLOSS (global subducting sediment) (Plank & Langmuir, 1998), MORB (Niu & Batiza, 1997), subducted oceanic crust-derived adakites (Defant et al. 2002; Martin et al. 2005), subducting continental crust-derived adakites (Wang et al. 2008 a, b), Hohxil adakites (Wang et al. 2005, 2008 a, b) and Duogecuoren adakitic rocks (Wang et al. 2005, 2008 a, b) are shown for debate. See text for details.

We argue that the first four genetic hypotheses are not applicable to the RCA dacites based on the following observations. First, adakites from modern arcs that are interpreted as slab melts have mid-ocean-ridge basalt (MORB)-like Sr–Nd isotopic compositions and relatively low K₂O, Th and Th/La values, which originate from the basaltic portion of subducting slabs (Defant & Drummond, 1990; Kelemen et al. 2003; Plank, 2005). In the case of the studied adakite-like dacites from the RCA, they have more evolved Sr-Nd-Pb isotopic compositions (Fig. 6) and much higher K₂O contents (1.54–2.46 wt.%), with K₂O/Na₂O ratios of 0.36–0.56, as well as varying higher Th values (6.9-8.6 ppm) and Th/La ratios (0.27–0.32) than those of the MORB and slab-derived (subducted oceanic crust-derived) adakites (Fig. 9c; e.g. Niu & Batiza, 1997; Martin et al. 2005; Wang et al. 2008 a, b; Tang et al. 2010). They are geochemically distinct from the Hohxil adakites produced by melting of subducted sediments on the northwards-subducting Songpan–Ganzi oceanic slab (Fig. 9c; Wang et al. 2005, 2008 a) and the Mamen adakite-like rocks derived from partial melting of the subducted oceanic slab (MORB + sediment + fluid) (Zhu et al. 2009). On the other hand, if the dacites originated as subducted slab melts, they should have mid-ocean-ridge basalt-like Sr–Nd isotopic compositions (ϵ _Nd(t)≈10; Defant & Drummond, 1990), as shown by the Bangong–Nujiang oceanic basalt (Fig. 6a; Zhu et al. 2006; Liu et al. 2014). As shown in Fig. 6a, the RCA adakite-like dacites show a more radiogenic Sr–Nd isotope signature than MORB that formed at 400-179 Ma (Hofmann, 2003), Cenozoic adakites that formed in a subducting environment (e.g. Kimura et al. 2014), adakites that formed by the partial fusion of modified lithospheric mantle in the Yanji region (Guo et al. 2007) and adakites that originated from slab-derived melt in the Kitakami region (Tsuchiya et al. 2005). In addition, their features of negative Nb–Ti anomalies are not consistent with N-MORB, OIB or oceanic slab sources resulting in melts, as identified by positive Nb–Ti anomalies in the spidergrams normalized to the primitive mantle (Fig. 4a, b; e.g. Sun & McDonough, 1989; Hofmann, 1997), suggesting that their source rocks were not basaltic oceanic crust.

Second, suites of adakitic rocks derived by fractional crystallization of basaltic magmas usually exhibit variable and high Gd/Yb ratios in addition to high La/Yb and Sr/Y ratios (Macpherson et al. 2006; Richards & Kerrich, 2007). However, the RCA dacites exhibit low and relatively low Gd/Yb ratios (2.43–2.67), with low La/Yb ratios of 25.8 to 28.2 and Sr/Y ratios of 53.0 to 70.8. Geochemically, adakites generated by this process generally display distinct compositional trends, in which the Al₂O₃ and La contents decrease with increasing SiO₂ contents, but the La/Y, Dy/Yb and Sr/Y ratios clearly increase with increasing SiO₂ contents (Macpherson et al. 2006). Nevertheless, the RCA dacites obviously exhibit none of these trends, with a relatively narrow range of SiO₂ contents (64.79–70.37 wt.%). In other words, fractional crystallization can also be ruled out because of the lack of a negative linear correlation between SiO₂ concentrations and Mg^# values (Fig. 9b; Castillo et al. 1999), as well as a lack of a positive linear correlation between SiO₂ concentrations and Sr/Y and (Dy/Yb)_N values (not shown; Macpherson et al. 2006), for the RCA adakite-like dacites. Furthermore, these adakite-like dacites are not associated with contemporary basaltic rocks that would correspond to the required mafic end-member in AFC, FC and magma mixing models (e.g. Castillo et al. 1999; Macpherson et al. 2006; Streck et al. 2007), indicating that they could not have been generated directly by crustal assimilation and fractional crystallization from parental basaltic magmas.

Third, it appears unlikely that they formed as a product of magma mixing between felsic and basaltic magmas, as they contain no mafic microgranular enclaves, as also evidenced by the lack of straight linear variations between the concentrations of major elements, such as MgO, K₂O, Al₂O₃ and SiO₂, within these rocks. Moreover, these samples show weakly negative to positive whole-rock ϵ _Nd(t) values (Fig. 6a, b; −0.28 to +1.25) and positive zircon ϵ _Hf(t) values (Fig. 7; +3.9 to +13.1), which are inconsistent with formation from magma mixing between felsic and basaltic magmas.

Fourth, their formation was not related to melting of delaminated lower crust. These RCA adakite-like dacites have whole-rock radiogenic Sr isotope [⁸⁷Sr/⁸⁶Sr_(i) = 0.705119–0.705491], Nd isotope [ϵ _Nd(t) of −0.28 to +1.25] and Pb isotope [(²⁰⁶Pb/²⁰⁴Pb)_i = 18.489–18.508, (²⁰⁷Pb/²⁰⁴Pb)_i = 15.591–15.612, (²⁰⁸Pb/²⁰⁴Pb)_i = 38.599–38.686] compositions (Fig. 6b, c, d), which are remarkably different from the crustal xenoliths in Cenozoic volcanic rocks from the QT, northern Tibet (Fig. 6a; Lai & Qin, 2008). As mentioned above, their geochemical and Sr–Nd isotopic features significantly differ from those of the Cenozoic crust-derived adakites of the Songpan–Ganzi block (Fig. 6c, d; Wang et al. 2008 a), the Mesozoic Anjishan adakites from the Ningzhen area of East China generated by partial melting of delaminated lower continental crust [ϵ _Nd(t) = −6.8 to −9.7, (⁸⁷Sr/⁸⁶Sr)_i = 0.7053 to 0.7066, Xu et al. 2002], the Zougouyouchaco and Dogai Coring adakitic volcanic rocks in the southern and middle Qiangtang region that were most likely derived from partial melting of delaminated lower continental crust [ϵ _Nd(t) = −3.8 to −5.0, (⁸⁷Sr/⁸⁶Sr)_i = 0.706–0.708; Liu et al. 2008], and the Dexing adakites originating from the partial fusion of delaminated lower crust that experienced a melt and mantle peridotite interaction (Wang et al. 2006 b). Moreover, their low Mg^# values (23.83–29.65) and low MgO abundances (0.68–0.88 wt.%) also argue against the origin of melting of delaminated lower crust (Fig. 9b).

We suggest that the RCA adakite-like dacite suites were probably generated by melting of juvenile mafic lower continental crust. Notably, their typical geochemical magmatic arc signatures, such as enrichment in LILEs, depletion in HFSEs, especially negative Nb–Ta–Ti anomalies, and light REE (LREE)-enriched chondrite-normalized REE patterns (La_N/Yb_N = 18.48–20.23), with relatively negative Eu anomalies (δEu = 0.84–0.89), are consistent with those of rhyolites in the Rena-Co area (ca. 110 Ma), which are proposed to be derived from juvenile crust (Chang et al. 2011). In particular, they show relatively low MgO and Mg# values and Cr and Ni contents, which are very comparable with those of adakitic rocks originating from the anatexis of the thickened lower crust (Castillo, 2012). Melting experiments of natural hydrous basalts under 0.8–3.2 GPa have revealed that pristine melts of basaltic rocks typically display low MgO and Mg^# values (generally lower than 43; Rapp & Watson, 1995). As mentioned earlier, the RCA adakite-like dacites yield low MgO and Mg^# values, similar to those of the pure experimental melts, which further confirms an origin of juvenile lower crustal melts for their generation (Fig. 9b; e.g. Rapp et al. 1999, 2003; Rapp & Watson, 1995).

They display low initial ⁸⁷Sr/⁸⁶Sr_(i) values of 0.705119 to 0.705491, weak negative to positive whole-rock ϵ _Nd(t) values of −0.28 to +1.25 and positive zircon ϵ _Hf(t) values from +3.9 to +13.1, which are features that might have been inherited from juvenile crustal magmas. As shown in Figure 6a, the adakitic dacites have Sr–Nd isotope signatures identical to those of the late Miocene juvenile lower crust-derived adakitic volcanics in the Ikizdere area (Karsli et al. 2019), the late Miocene mafic lower crust-derived Tavdağı adakitic rhyolite in the Tortum area (Dokuz et al. 2013) and the early Eocene lower crust-derived adakitic volcanics in the Gümüşhane area (Karsli et al. 2010). Furthermore, the adakite-like dacites share relatively young T _DM2 ages ranging from 806 to 931 Ma and Palaeozoic and Neoproterozoic Hf crustal model ages of 332 to 924 Ma, possibly suggesting that they were likely inherited from the melts of juvenile basaltic crust, which in turn was derived from old depleted mantle (van de Zedde & Wortel, 2001). This view is also supported by the fact that these samples have much more enriched whole-rock Sr–Nd isotopic compositions and more depleted zircon Hf isotopic compositions than those of the existing Jurassic intrusive rocks in the southern QT, such as the Larelaxin, Caima, Qingcaoshan, Charkang-Co and Rena-Co plutons, which are considered to be generated by partial melting of ancient lower continental crustal materials or by the mixing of crust-derived and mantle-derived melts (e.g. Li et al. 2014 a, b; Hao et al. 2016 a, b; and references therein).

Note that zircons from the RCA adakite-like dacite samples show large variations in their ϵ _Hf(t) values (up to 7.8 and 9.2 ϵ units for the two samples, respectively), thereby indicating an open magmatic system process (Griffin et al. 2002; Kemp et al. 2007; Chiu et al. 2009). This hypothesis is also supported by highly radiogenic Pb isotopic compositions between mantle and crust xenoliths (Fig. 6b, c, d; Lai & Qin, 2008), likely indicating a mixing trend. The (²⁰⁷Pb/²⁰⁴Pb)_i values of 15.591–15.612 and (²⁰⁸Pb/²⁰⁴Pb)_i values of 38.599–38.686 in the rocks are unusually radiogenic and plot well above the Northern Hemisphere Reference Line (NHRL) (Hart, 1984), shifting to higher (²⁰⁶Pb/²⁰⁴Pb)_i values of 18.489–18.508 than the Geochron (4.55 Ga) (Fig. 6c, d). In addition, their ϵ _Hf(t) values (+3.9 to +13.1; n = 33) are similar to those of the Duolong Early Cretaceous volcanic rocks and intrusive rocks that are considered to be derived from juvenile lower crust (e.g. She et al. 2009; Xin et al. 2009; Chen et al. 2013; Li et al. 2013 a, 2016; Sun 2015; Zhu et al. 2015; Wei et al. 2017, 2018). Notably, these volcanic rocks from the RCA show lower zircon ϵ _Hf(t) values than depleted mantle (Fig. 7), possibly implying at least partial contributions from metasomatized peridotite with partial melts of subduction-related fluids into their magma source (Woodhead et al. 2001; Barry et al. 2006). Such a foregoing interpretation is consistent with the modelling curves defined by the MF adakite-like rocks from the RCA in Sr–Nd and Nd–Pb isotope diagrams (Fig. 6a, b). Consequently, we consider that the RCA volcanic rocks contain a great proportion of approximately 75–80% mantle-derived material contributions, according to their Sr–Nd isotopic compositions (Fig. 6a), as well as melts of ∼3–5% subduction-related fluids, according to their Nd–Pb isotopic compositions (Fig. 6b), given that a small contribution of fluids can sensitively cause a drastic increase in ²⁰⁶Pb/²⁰⁴Pb ratios in subduction-related rocks (Vroon et al. 1995; Rolland et al. 2002). In particular, the same observations occurred for igneous rocks of the Cretaceous Ladakh arc (Rolland et al. 2002) and the Cretaceous Kohistan island arc (Bignold & Treloar, 2003), of which several percent of subduction-related fluids became entrained into the magma source regions, in accordance with the measured Sr-Nd-Pb isotopic compositions.

In summary, these observations lead us to infer that the RCA adakite-like dacites were derived from melting of juvenile thickened mafic lower continental crust that contains melts of metasomatized peridotite and subduction-related fluids in the magma source region, likely as a result of the northern subduction of the BNOL beneath the southern QT.

4.3. Geodynamic implications

As one of the ancient orogens in the Tibetan Plateau, the BNSZ extending along central Tibet represents the oceanic remnant of the Bangong–Nujiang Tethyan Ocean (e.g. Allègre et al. 1984; Hsu et al. 1995; Pearce & Deng, 1988; Yin & Harrison, 2000; Kapp et al. 2003, 2005, 2007). Zhu et al. (2016) argued a model in which the Bangong Ocean may have closed through arc–arc ‘soft’ collision driven by divergent double-sided subduction rather than continent–continent ‘hard’ collision. Some studies have argued that the BNO closed between the Late Jurassic and the Early Cretaceous (Yin & Harrison 2000; Kapp et al. 2005; Sui et al. 2013; Wang et al. 2016; Zhu et al. 2016; Hu et al. 2022). However, numerous recent studies that have focused on radiolarians, ophiolitic fragments and mélanges within the BNSZ (Zhu et al. 2006; Bao et al. 2007; Baxter et al. 2009; Fan et al. 2014; Zhang et al. 2014) suggest that the BNO was still open during the Early Cretaceous and that the final Lhasa–Qiangtang amalgamation occurred in the Late Cretaceous (approximately 100 Ma; e.g. Baxter et al. 2009; Fan et al. 2014) rather than in the Late Jurassic–Early Cretaceous (e.g. Xu et al. 1985; Dewey et al. 1988; Yin et al. 1988; Kapp et al. 2003, 2005, 2007; Raterman et al. 2014), as also exemplified by the abundant Early Cretaceous magmatic rocks in the southern QT, which are interpreted as having formed in a continental arc setting (e.g. Kapp et al. 2005, 2007; Chang et al. 2011; Liu et al. 2012; Hao et al. 2016 a, b; Wang et al. 2015; Sun, 2015; Li et al. 2013 a, b, 2015, 2016; Wei et al. 2017, 2018; Shi et al. 2019; this study). Stratigraphic analysis suggests that the QT underwent N–S-directed compression and significant crustal shortening during the Early Cretaceous (Murphy et al. 1997; Kapp et al. 2005, 2007), probably as a result of the collision of an oceanic plateau with the Qiangtang continental margin (Zeng et al. 2021). It has long been recognized that the BNOL subducted northwards beneath the southern QT before the closure of the BNO (e.g. Allègre et al. 1984; Yin & Harrison, 2000; Guynn et al. 2006; Kapp et al. 2003, 2005, 2007; Zhu et al. 2016; Wei et al. 2018). Our data reveal the presence of adakite-like dacites with ages of 109.5–109.6 Ma in the RCA, approximately 30 km to the north of the BNSZ (Fig. 1b). These adakite-like dacites formed from magmas generated by melting of juvenile lower crustal material and have compositions consistent with adakites that formed in a continental marginal arc. Together with previous data on the Early Cretaceous intrusive and volcanic rocks (especially the Duolong Meriqiecuo volcanic rocks) distributed in the southern QT (She et al. 2009; Xin et al. 2009; Liu et al. 2012; Chen et al. 2013; Li et al. 2013 a, 2016; Wang et al. 2015; Sun, 2015; Zhu et al. 2016; Wei et al. 2017, 2018; Shi et al. 2019), we confidently propose that the sources of the RCA adakite-like magmas were most likely associated with crustal thickening, given that crustal thickening is a common geological phenomenon in arc settings. All of this evidence indicates that the timing of the final closure of the BNO might have been later than the Early Cretaceous (109.5.9-109.6 Ma).

Recently, Hao et al. (2016 a, b) also suggested that ancient lower crust was gradually replaced by younger materials between the Late Jurassic and Early Cretaceous and finally resulted in vertical crustal growth by magma underplating for the southern Qiangtang continental arc, based on the tracking of the magma sources of the late Mesozoic intermediate–felsic intrusive rocks in southern Qiangtang. Generally, such magma underplating is considered a result of asthenospheric upwelling, which is related to oceanic subduction with a slab rollback component due to drag and gravity forces (Nakakuki & Mura, 2013). As discussed above, the magma of the RCA adakite-like dacites most likely originated from melts of juvenile thickened lower crust. In this case, we suggest that a slab rollback model explains the magmatic flare-up that occurred in the southern QT during the Early Cretaceous and accounts for the formation of the RCA adakite-like dacites (ca. 109.5-109.6 Ma). This mechanism may also apply to the geological processes in the northern Lhasa and Tethyan Himalayan thrust belts (Raterman et al. 2014; Zhu et al. 2016), both of which represent passive margins on the lower subducting plate of a collision zone.

In this model, the downgoing BNOL carried subducted fluids into the deep mantle during its northwards subduction and then resulted in the initiation of partial melting of mantle peridotite (Elliott et al. 1997). Subsequently, variable degrees of partial melts of mantle peridotite metasomatized by subducted fluids moved upwards, and basaltic underplating gradually replaced the ancient lower crust and accumulated to form juvenile basaltic crust. Thus, as a result of Early Cretaceous slab rollback, this newly underplated juvenile basaltic lower crust was further heated by a flow of upwelling hot asthenospheric materials (van de Zedde & Wortel, 2001; Zhu et al. 2016; Li et al. 2013 b, 2015), finally leading to the generation of primary basaltic magmas for the RCA adakite-like dacites (Fig. 10; Richards, 2003; Zhu et al. 2008).

Figure 10. Schematic illustrations showing the geodynamic evolution of the Bangong–Nujiang oceanic northern subduction beneath QT during the Early Cretaceous (modified from Richards (2003) and Zhu et al. (2008)). Note that a slab rollback model combining Bangong–Nujiang oceanic subduction and underplating of mafic magmas proposed for the generation of the Early Cretaceous Meiriqiecuo magmatism in the southern QT. See text for details.

5. Conclusions

1. (1) Zircon U-Pb dating suggests that the RCA dacites formed at 109.5.9-109.6 Ma, marking Late Cretaceous volcanic magmatism during the northern subduction of the BNOL beneath the southern QT.

2. (2) They exhibit geochemical features with adakite affinity and most likely originated from melts of the juvenile continental lower crust, which contains melts of metasomatized peridotite and subduction-related fluids in the magma source region.

3. (3) Based on previous studies and our new data, we propose that a slab rollback model could explain the formation of the RCA adakite-like dacites, which was most likely associated with crustal thickening.