RETRACTED-Late Permian to early Triassic gabbro in North Lhasa, Tibet: evidence for plume – subduction-zone interaction of the Palaeo-Tethys ocean

Abstract

The Palaeo-Mesozoic geodynamic evolution of the Tangjia–Sumdo accretionary complex belt, which separates the North and South Lhasa Terrane, remains controversial. Moreover, the lack of geological records restricts the understanding of the evolution of the Sumdo Palaeo-Tethys Ocean from the middle Permian until the middle Triassic. Here we present zircon U–Pb geochronology, whole-rock geochemistry and Sr–Nd–Hf isotopic compositions of the Yeqing gabbro. Zircon U–Pb geochronology yields ages from 254 ± 1 to 249 ± 1 Ma. In situ Hf isotopic analyses yield ϵ _Hf(t) values of −0.2 to +6.3. These samples have high TiO₂ (3.69 wt %) and P₂O₅ (0.78 wt %) contents, with typical patterns like ocean island basalt (OIB). Besides, they are classified as high-Nb basalts (HNBs) based on the high content of Nb (45.3–113.5 ppm). Whole-rock Sr–Nd isotopic compositions are similar to OIB, with initial ⁸⁷Sr/⁸⁶Sr of 0.7047–0.7054, ¹⁴³Nd/¹⁴⁴Nd of 0.512526–0.512647 and ϵ _Nd(t) of 0.3–2.7. These signatures suggest that the Yeqing gabbro is mainly derived from low-degree melting of the garnet lherzolite mantle. Based on field observations of HNBs intruding into the continental margin and their geochemical characteristics, we infer that the Yeqing gabbro was generated in a subduction environment. Combined with the regional geology of the subduction environment and the evolution of oceanic islands in the Sumdo Palaeo-Tethys Ocean, we propose that the Yeqing gabbro may represent a product of the asthenosphere upwelling through a slab window produced by subduction of seismic ridge in the Sumdo Palaeo-Tethys Ocean, called plume – subduction-zone interaction, during the late Permian to early Triassic.

1. Introduction

The Palaeo-Tethys Ocean was a large ancient ocean located between the supercontinents of Gondwana and Laurasia (Şengor, 1987; Metcalfe, 2013). A series of banded terranes (Qiangtang, Lhasa, Himalaya, Cimmerides, Sibumasu, etc.) derived from the north edge of Gondwana drifted northward and were accreted into Laurasia along with the closure of the Palaeo-Tethys Ocean (Yin & Harrison, 2000; Dilek & Furnes, 2011; Metcalfe, 2013), forming a huge orogenic belt, known as the ‘Eastern Tethys System’, in the northern and southeast edge of the Tibetan Plateau and the central orogenic belt between North China and the Yangtze (Xu et al. 2015 and references therein). The Palaeo-Tethys Ocean in the Eastern Tethys System, which probably opened in the Middle Cambrian and continued to grow throughout the Palaeozoic and closed in the later Triassic, is mainly represented by the Jinshajiang and Longmu Co – Shuanghu suture zone in the northern Tibetan Plateau, and the Changning–Menglian and Ailaoshan suture zone of the Sanjiang Tethys realm in the eastern margin of the Tibetan Plateau (Fig. 1a ; Yin & Harrison, 2000; Li et al. 2006, 2008; Zhai et al. 2010, 2013, 2016; Metcalfe, 2013; Fan et al. 2014, 2015, 2017; M Wang et al. 2014, 2019; Zhang et al. 2017; Xie et al. 2017 a, b).

Fig. 1. (a) Tectonic framework of the Tibetan Plateau (modified after Li et al. 2006 and Zhu et al. 2010). (b) Geological sketch map of the Tanga–Sumdo area; published age data are after Chen et al. (2009), Yang et al. (2009), Cheng et al. (2012, 2014), Weller et al. (2016), Cao et al. (2017), Duan et al. (2019, 2022), B Wang et al. (2019) and Song et al. (2022).

The discovery of the late Palaeozoic Sumdo high/ultra-high-pressure (HP/UHP) metamorphic belt in Lhasa terrane reveals that there are records of an oceanic subduction zone, which may represent the southernmost branch of the Palaeo-Tethys Ocean (Fig. 1a; Yang et al. 2006, 2009; Liu et al. 2009; Xu et al. 2015). The fact that Sumdo HP/UHP metamorphic belt is located between Indus – Yarlung Zangbo (Neo-Tethys Ocean) and Bangong Co – Nujiang (Meso-Tethys Ocean) Suture Zone is incongruent with the common view that the Tethys Ocean Suture Zone becomes gradually younger from north to south (Xu et al. 2015). In recent years, further evidence for the evolution of the Sumdo Palaeo-Tethys Ocean (SPTO) has been established in Sumdo and adjacent regions, with examples such as ophiolites (Fig. 1b; Chen et al. 2010; Duan et al. 2019; Wang et al. 2021), oceanic islands (Fig. 1b; B Wang et al. 2019; Zhong et al. 2021; Duan et al. 2022), eclogite and blueschist (Fig. 1b ; Yang et al. 2006, 2009; Liu et al. 2009), arc magmatism (Fig. 1b ; Geng et al. 2009; Zhu et al. 2010; B Wang et al. 2020, 2022; C Wang, 2022) and flysch-like sedimentary strata (Fig. 1b ; Xie et al. 2019, 2021). Thus, this belt is also called the Tangjia–Sumdo accretionary complex belt (TSACB) (Fig. 1a; B Wang et al. 2020, 2022; Xie et al. 2021). Previous research suggests that the SPTO may have opened before the early Carboniferous and subducted initially before the early Permian, then soon after the late Triassic at latest (Cheng et al. 2015; Duan et al. 2019, 2022; B Wang et al. 2020, 2021, 2022; Liu et al. 2022). Based on the study of eclogite and arc magma, we know that the SPTO subducted during the early–middle Permian and middle–late Triassic periods (Cheng et al. 2012, 2015; Zhang et al. 2018 a; B Wang et al. 2020, 2022; Song et al. 2022; C Wang et al. 2022). However, many aspects of the subduction evolution of the SPTO remain unclear, especially during the late Permian to early Triassic owing to gaps in the geological record (Zhu et al. 2010; Cheng et al. 2012, 2015; Zhang et al. 2018 b; B Wang et al. 2020, 2022; Li et al. 2022; C Wang et al. 2022).

The Yeqing gabbro is discovered near the north edge of the TSACB, whose zircon U–Pb ages vary from 254 ± 1 to 249 ± 1 Ma in this study, coincident with the active period of the SPTO. Here, we present zircon U–Pb geochronology, whole-rock geochemistry, as well as zircon Hf and whole-rock Sr–Nd isotopic data, which are significant in resolving the diagenetic age and petrogenesis of the Yeqing gabbro. We also discuss the tectonic setting of the intrusion and draw implications regarding the subduction evolution of the SPTO during the late Permian to early Triassic.

2. Geological background

The Tibetan Plateau is located in the eastern part of the Tethys tectonic domain. The closure of the Tethys oceans created four major suture zones associated with the Tibetan Plateau (Fig. 1a;Yin & Harrison, 2000; Zhu et al. 2010; Torsvik & Cocks, 2013; Zhai et al. 2016). These suture zones divide the Tibetan Plateau, from north to south, into the Songpan–Ganzi, Northern Qiangtang, Southern Qiangtang, Lhasa and Himalayan terranes (Li et al. 2006, 2008; Zhai et al. 2010, 2013; Fan et al. 2014, 2015, 2017; M Wang et al. 2014, 2019; Xie et al. 2017 a, b; Zhang et al. 2017; Hu et al. 2018, 2019). The Lhasa terrane is further divided into North and South Lhasa Terrane by the SPTO, marked by the TSACB (Yang et al. 2009).

The TSACB, correlated spatially with the Luobadui–Milashan Fault (Fig. 1a; Zhu et al. 2010), is dominated by scattered fragments of the SPTO remnants (Fig. 1b). In the study area, the Nuoco Formation, formed in a continental margin environment, mainly consists of sandstone or metasedimentary rock on the north of the TSACB (Zhu et al. 2013). The strata exposed in the TSACB is mainly Sumdo Formation, which is a set of low-grade-metamorphosed terrigenous clastic sandstones and mudstones formed in an initial fore-arc basin environment (Xie et al. 2019, 2021). The Luobadui Formation exposes a little in the study area and is mainly composed of limestone, terrigenous sediments and arc-type volcanic rocks (Geng et al. 2009; Zhu et al. 2010; C Wang et al. 2022).

The SPTO remnants are abundant in the Sumdo Formation as slices, including Permian–Triassic eclogites (Li et al. 2009; Yang et al. 2009; Zeng et al. 2009), late Carboniferous – middle Triassic ophiolites (Chen et al. 2010; Duan et al. 2019; Wang et al. 2021) and early Carboniferous – middle Permian oceanic islands (B Wang et al. 2019; Zhong et al. 2021; Duan et al. 2022). Previous studies have suggested that there are at least two types of eclogites, with ages of the metamorphic peak in the middle Permian (274–261 Ma) and middle–late Triassic (238–227 Ma) (Li et al. 2009; Yang et al. 2009; Zeng et al. 2009; Chen et al., 2010; Cheng et al. 2012, 2015; Weller et al. 2016; Cao et al. 2017; Zhang et al. 2018 a, b). There is also plenty of Permian arc magmatism found in the south edge of the North Lhasa Terrane (Fig. 1a; Geng et al. 2009; Zhu et al. 2010) and a part in the TSACB (Fig. 1b, 278–262 Ma; B Wang et al. 2020, 2022; Mai et al. 2021; Li et al. 2022; C Wang et al. 2022). In addition, the middle–late Triassic (230–200 Ma) granite with typical arc magmatism characteristics may be related to the northward subduction of the SPTO (Li et al. 2020; Song et al. 2022).

3. Field observations and petrology

The study area is located between Tangjia and Sumdo (Fig. 1b). The Yeqing gabbro intrusions expose as near east–west-trending dike in the Nuoco Formation (Fig. 1b). The larger intrusion is c. 8 km long and 50 m thick, while the smaller intrusion is c. 3 km long and 20 m thick. The host rocks of the Yeqing gabbro contains quartzite and sandstone of the Nuoco Formation (Fig. 2a). We can observe the obvious chilled margin between gabbro and quartzite (Fig. 2b).

Fig. 2. Photographs of Yeqing gabbro. (a) Macro outcrop photo of Yeqing gabbro intruding into Sumdo Formation. (b) Close-up photo of the boundary between Yeqing gabbro and Sumdo Formation. (c, d) Micrograph of Yeqing gabbro, Plagioclase is replaced by sericite, and pyroxene is replaced by hornblende, in part of our samples. Pl – plagioclase; Px – pyroxene; hor – hornblende; Sre – sericite; Mag – magnetite.

The gabbro samples are fine- to medium-grained and consist mainly of pyroxene and plagioclase (Fig. 2b). Petrographic observations under the microscope reveal that the gabbro shows crystalline texture, consisting of dominant mineralogy of pyroxene (35 %), plagioclase (60 %) and magnetite (<5 %), with slight metamorphism (Fig. 2c, d). Some pyroxene has been replaced by chlorite or hornblende, and the plagioclase is weakly altered to sericite, which shows the sericite microcrystalline aggregates under a microscope with orthogonal polarized light (Fig. 2d).

4. Analytical methods

4.a. Whole-rock geochemistry

Whole-rock geochemical analysis was performed at the experimental centre of the Academy of Sciences, China University of Geosciences, Beijing, China. Major-element analysis was performed in the inductively coupled plasma – optical emission spectroscopy (ICP-OES) laboratory using an X-ray fluorescence spectrometer. The precision is better than 1 % for all elements. Trace-element analysis was performed using an Agilent 7500a ICP – mass spectrometry (ICP-MS) instrument. During the analysis, standard samples AGV-2, W2 and BHOV from the United States Geological Survey and rock samples R-1 and R-3 from the China Geological Testing Center were used to monitor the analytical precision, which is better than 5 % for most elements. Further details of the experimental method are presented by Zhai et al. (2013).

4.b. Zircon U–Pb geochronology

Zircon grains were separated at the premises of the Yuheng Mineral Technology Service, Langfang, China, by conventional heavy liquid and magnetic techniques. Cathodoluminescence (CL) images were taken at the Institute of Geology, Chinese Academy of Geological Sciences, Beijing, China. U–Pb isotopic and trace-element analyses of zircon were carried out by laser-ablation – ICP-MS (LA-ICP-MS) at the Key Laboratory of Mineral Resources Evaluation in Northeast Asia, Ministry of Natural Resources, Jilin University, Changchun, China. The laser beam spot diameter was 32 µm, and helium was used as the carrier gas. Details of the procedure are reported by Liu et al. (2010). Fractionation correction of isotopic ratios was performed using zircon 91500 as the external standard. NIST SRM 610 was used as the external standard for correction of elemental abundances, and ²⁹Si was used as the internal standard. Data were processed using Glitter software (version 4.4; Griffin et al. 2008). Details of specific experimental data reduction techniques are given by Chang et al. (2006). The software package Isoplot 4.15 was used to calculate weighted-mean U–Pb ages of the samples and to generate Concordia diagrams (Ludwig, 2003).

4.c. Zircon Hf isotopes

Zircon Hf isotopic analyses were conducted at the facility of Beijing Createch Testing Technology, Beijing, China. Hf isotopic analyses were performed on the same spots or in the same age domains (as identified by CL images) as used for U–Pb dating. An NWR-213 (nm) laser ablation microprobe coupled to a Neptune Plus multi-collector (MC)-ICP-MS instrument was used for isotopic analysis. The laser beam spot was 40 µm in diameter, with an energy density of 10–11 J cm⁻² and a frequency of 10 Hz. The ablated material was carried to the mass spectrometer by high-purity helium gas. Zircon GJ-1 was used as the reference standard during analyses, whose weighted mean ¹⁷⁶Hf/¹⁷⁷Hf ratio (0.282000 ± 32; 2σ; n = 17) is similar to the commonly accepted weighted mean ¹⁷⁶Hf/¹⁷⁷Hf ratio of 0.282013 ± 19 (2σ) reported for in situ analysis by Elhlou et al. (2006). Technical procedures and instrument operational parameters are described by Hu et al. (2012). Data reduction methods followed those presented by Bouvier et al. (2008).

4.d. Whole-rock Sr–Nd isotopic analysis

Whole-rock Sr–Nd isotopic analyses were carried out using a Thermo Fisher Scientific Neptune Plus MC-ICP-MS instrument at the facility of Beijing Createch Testing Technology. ⁸⁷Sr/⁸⁶Sr ratios were corrected for instrumental mass fractionation using an exponential fractionation law and assuming ⁸⁸Sr/⁸⁶Sr = 8.375209. ¹⁴³Nd/¹⁴⁴Nd ratios were corrected for instrumental mass fractionation using an exponential fractionation law and assuming ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219. The Sr isotope international standard NBS 987 was repeatedly tested to monitor accuracy, yielding a mean ⁸⁷Sr/⁸⁶Sr value of 0.710248 ± 9 (2SD, n = 11). Stability assessment for ¹⁴³Nd/¹⁴⁴Nd was conducted with the in-house standard GSB-Nd, yielding a value of 0.512195 ± 6 (2SD, n = 12). Detailed analytical procedures are given by Hu et al. (2018).

5. Results

5.a. U–Pb zircon geochronology

LA-ICP-MS zircon U–Pb data and zircon trace element of the Yeqing gabbro are presented in Supplementary Tables S1 and S2, respectively. Zircon grains separated from gabbro samples (ST30, ST31, ST38 and ST39) are semi-transparent and columnar to granular. The lengths of the zircon crystals are 100–250 μm, with aspect ratios of 1:1–3:1. Some zircon grains exhibit oscillatory zoning in CL imaging (Fig. 3). The range of Th and U contents is relatively wide (Th = 36–1053 ppm, U = 37–450 ppm) and the Th/U ratios are quite high (0.56–2.57), which is consistent with the magmatic origin (Hoskin & Schaltegger, 2003). Positive Ce and negative Eu anomalies are observed in chondrite-normalized rare earth element (REE) patterns of zircon grains from these samples (Fig. 3).

Fig. 3. Cathodoluminescence (CL) images and chondrite-normalized REE patterns diagram of representative zircon grains from Yeqing gabbro. The yellow circle is the location of U–Pb isotope analysis, and the red is Lu–Hf isotope analysis. Values of chondrite are after Sun and McDonough (1989).

U–Pb data from all four samples are concordant (Fig. 4). The data yield weighted-mean ages of 249 ± 1 Ma for ST30 (n = 25, MSWD = 0.34, 1σ), 251 ± 1 Ma for ST31 (n = 25, MSWD = 0.75, 1σ), 252 ± 1 Ma for ST39 (n = 28, MSWD = 0.51, 1σ) and 254 ± 1 Ma for ST38 (n = 23, MSWD = 0.92, 1σ).

Fig. 4. U–Pb zircon Concordia of representative zircon grains from Yeqing gabbro. MSWD = mean squared weighted deviation.

5.b. Whole-rock geochemistry

Whole-rock major- and trace-element geochemical data are listed in Supplementary Table S3. These samples contain variational content of SiO₂ (41.91–50.12 wt %) and MgO (4.15–8.56 wt %). The contents of Fe₂O₃ ^T (Fe₂O₃ total) and Al₂O₃ are relatively concentrated, with means of 13.34 wt % and 15.39 wt % respectively. In particular, the contents of TiO₂, P₂O₅ and (Na₂O + K₂O) are comparatively high, with means of 3.69 wt %, 0.78 wt % and 4.36 wt % respectively. These samples plot mostly in the alkaline basalt field in the SiO₂ vs Nb/Y diagram (Fig. 5a).

Fig. 5. (a) SiO₂ vs Nb/Y (Winchester & Floyd, 1977) plot. (b) Nb/La vs MgO (Kepezhinskas et al. 1996) plot. (c) Nb*2 vs Zr/4 vs Y figure (Meschede, 1986). Within-plate alkali basalts – AI, AII within-plate tholeiites – AII, C; plume-type MORB- B; N-type MORB- D; volcanic arc basalts – C, D. (d) TiO₂ vs MnO₂*10 vs P₂O₅*10 figure (Mullen, 1983). MORB – mid-ocean ridge basalt; IAT – island arc tholeiite; CAB – calc-alkaline basalt; OIT – ocean island tholeiite; OIA – ocean island alkaline. Data of the NEBs, HNBs from Baja, Nicaragua, Renso, Dubuzha, Tuotuohe and Gerze are after Storey et al. (1989), Luhr et al. (1995), Benoit et al. (2002), Wang et al. (2007), Gazel et al. (2011), Li et al. (2016) and Hao et al. (2018).

The REE contents of these samples are relatively higher (168.54–559.62 ppm) than mid-ocean ridge basalt (MORB) (39.11 ppm; Sun & McDonough, 1989), with light REE (LREE) enrichment (La_N/Yb_N = 11.51–27.10) in chondrite-normalized REE patterns (Fig. 6a). Positive Nb and Ta anomalies and negative Th, Zr and Hf anomalies are observed in primitive-mantle-normalized trace-element spider diagrams, which is similar to the typical OIB and the NEBs and HNBs from Baja (California), Nicaragua, Renso and Duobuzha (Tibet), and different from those from Tuotuohe and Gerze (Tibet) (Fig. 6b, d). The higher Nb contents (45.3–113.5 ppm) of the Yeqing gabbro are similar to those of Nb-enrichment basalts (NEBs; 20 > Nb > 5 ppm) and high-Nb basalts (HNBs; Nb > 20 ppm) which are enriched in LILEs, LREEs and HFSEs and have weakly negative or positive primitive-mantle-normalized Nb and Ta anomalies (Castillo et al. 2007; Hastie et al. 2011); these samples plot in the NEBs and HNBs field in the Nb/La vs MgO plot (Fig. 5b). In addition, the Yeqing gabbro shows characteristics of intraplate alkaline basalts in the tectonic discrimination diagrams (Fig. 5c, d).

Fig. 6. Chondrite-normalized REE patterns and primitive-mantle-normalized spider diagrams. Values of chondrite, primitive mantle, OIB and E-MORB are after Sun and McDonough (1989).

5.c. Zircon Hf and whole-rock Sr–Nd isotopic analysis

Zircon Hf and whole-rock Sr–Nd isotope data are listed in Supplementary Tables S4 and S5, respectively. The Hf isotope analyses of the zircon grains from the gabbro samples show low ¹⁷⁶Lu/¹⁷⁷Hf ratios of 0.0003–0.0021 and ¹⁷⁶Hf/¹⁷⁷Hf ratios of 0.282608–0.282794. Calculated small negative or positive ϵ _Hf(t) values range from −0.2 to +6.3. Initial Sr isotope ratios and ϵ _Nd(t) values were calculated using ages of c. 254 Ma and c. 251 Ma reported in this study. These samples have a narrow range of initial (⁸⁷Sr/⁸⁶Sr) ratios of 0.7047–0.7054, ⁸⁷Sr/⁸⁶Sr ratios of 0.7047–0.7054 and ¹⁴³Nd/¹⁴⁴Nd ratios of 0.512526–0.512647. Calculated small positive ϵ _Nd(t) values range from 0.3 to 2.7.

6. Discussion

6.a. Petrogenesis

Low loss-on-ignition (LOI) values of 1.14–1.77 wt % and metamorphic minerals under the microscope indicate that the samples have undergone low-level metamorphism or alteration. This process may have modified the contents of mobile elements (e.g. Na, K, Rb, Ba and Sr), whereas the REE and high-field-strength elements (HFSE; e.g. Th, Zr, Hf, Nb, Ta, Ti and Y) should preserve primary magma compositions (Barnes et al. 1985; Jochum et al. 1991). In fact, most mobile elements and HFSE display good correlations with MgO (Fig. 8, further below), indicating there is almost no significant disturbance by metamorphism or alteration on most elements.

The Yeqing gabbro exhibits clear positive Nb and Ta and negative Th, U, Zr and Hf anomalies, with significant differences compared with continental crust (Rudnick & Gao, 2003; Niu, 2009). The Th/Ta ratios of these samples are 0.1–1.29, similar to those of volcanic rocks derived from a primitive mantle source (Th/Ta = 2.3), and much lower than that of the upper crust (Th/Ta > 10) (Thompson et al. 1984; Condie, 1993). The (Th/Ta)_PM (∼0.24) and (La/Nb)_PM (∼0.85) ratios of these samples are both less than 1, indicating that the crustal assimilation is negligible (Peng et al. 1994). Furthermore, these samples plot in the oceanic basalts field without the trend of crustal assimilation in the (Th/Nb)_PM vs (La/Nb)_PM diagram (Fig. 7).

Fig. 7. (La/Nb)_PM vs (Th/Nb)_PM plot (Neal et al. 2002). Middle crust and lower crust data are after Rudnick and Gao (2003). The ‘most oceanic basalts’ data are after Neal et al. (2002 and references therein).

The Mg^# values (=100 × Mg²⁺/(Mg²⁺ + Fe²⁺)) (43.8–61.1) and the Cr (2.9–253.9 ppm) and Ni (2.2–81.9 ppm) contents of the Yeqing gabbro are lower than those of primitive mantle (Mg^# = 68–76, Cr = 300–500 ppm, Ni = 300–400 ppm), indicating that they may have undergone fractional crystallization of olivine, pyroxene and chromite (Wilson, 1989; Jung & Masberg, 1998). In the process of fractional crystallization, Ni preferentially integrates into the olivine phase, and Cr preferentially integrates into the pyroxene phase (Wilson, 1989; Rollinson, 1993). The strongly positive relationships between FeO^T, Ni and MgO suggest that the magma underwent obvious fractional crystallization of olivine (Fig. 8a, h). In addition, the higher content and weak positive relationship with MgO of Cr element illustrate that there is a little or no fractional crystallization of pyroxene (Fig. 8g). The negative relationships between Sr, Na₂O, Al₂O₃ and MgO indicate that these samples underwent almost no fractional crystallization of plagioclase (Fig. 8b, c, i; Fodor & Vetter, 1984; Baker et al. 1997), consistent with the absence of Eu anomaly in the chondrite-normalized REE patterns (Fig. 6a).

Fig. 8 Harker variation diagrams for the Yeqing gabbro. The red arrows represent variation trend of part of major (wt %) and trace (ppm) elements towards the increase of MgO (wt %).

As presented above, the Yeqing gabbro is classified as HNBs with clear LREE enrichment and high Nb content (45.3–113.5 ppm). Two possible mantle sources have been generally proposed to generate HNBs, namely (1) OIB-type mantle source with mixing depleted normal-MORB (N-MORB) type components (Reagan & Gill, 1989; Storey et al. 1989; Luhr et al. 1995; Castillo et al. 2007; Castillo, 2008; Li et al. 2016) and (2) a mantle wedge metasomatized by slab melt (Defant et al. 1992; Kepezhinskas et al. 1996; Sajona et al. 1996; Wang et al. 2007; Hastie et al. 2011; Xu et al. 2017; Hao et al. 2018). In REE patterns and trace-element spider diagrams, the NEBs/HNBs derived from OIB-type mantle source with mixing depleted N-MORB type components show obvious LREE enrichment and positive Nb and Ta anomalies reported in Baja (California), Nicaragua, Renso and Duobuzha (Tibet), called a-type NEBs/HNBs in the following discussion (Fig. 6a, b). In contrast, the NEBs/HNBs derived from mantle wedge with mixing slab melt show an almost flat curve and negative Nb and Ta anomalies reported in Tuotuohe and Gerze (Tibet), called b-type NEBs/HNBs (Fig. 6c, d).

The HNBs that originated from a mantle wedge source metasomatized by slab melt still display some arc geochemical signatures, such as negative Nb–Ta anomalies and enrichment of Th (Fig. 6d; Kepezhinskas et al. 1996; Sajona et al. 1996; Wang et al. 2007; Hao et al. 2018). The Yeqing gabbro samples actually exhibit clearly positive Nb–Ta anomalies and depletion of Th (Fig. 6b, d). In the REE patterns and trace-element spider diagrams, the Yeqing gabbro shows a large slope like the OIB and a-type NEBs/HNBs, as distinct from the b-type NEBs/HNBs (Fig. 6b, d). In the Th/Yb vs Nb/Yb program (Fig. 9a), the gabbro samples plot in the OIB array like the a-type NEBs/HNBs, while the b-type NEBs/HNBs plot in the OIB to enriched MORB (E-MORB) array with a tendency to convert to continental arc, indicating that the magma of Yeqing gabbro generated without any addition of subduction fluids and melts. In the TiO₂/Yb vs Nb/Yb diagram (Fig. 9b), these samples are plotted as OIBs (alkaline) like the a-type NEBs/HNBs, while the b-type NEBs/HNBs show characteristics of shallow melting, indicating that the Yeqing gabbro may be the product of deep melting where garnet is stable. In addition, the Sr and Nd isotopic compositions of the Yeqing gabbro and Sumdo eclogite differ greatly (Fig. 9d), while the metasomatism of the mantle wedge by slab-derived melt would require the formation of Nb-rich magma with Sr and Nd isotopic characteristics similar to oceanic slab (Castillo et al. 2007). Above all, the Yeqing gabbro is notably distinct from the NEBs and HNBs from mantle wedge metasomatized by slab melt with negligible characteristics of arc, precluding the possibility of a mantle wedge source metasomatized by slab melt.

Fig. 9 (a) Th/Yb vs Nb/Yb plot and (b) TiO₂/Yb vs Nb/Yb plot (Pearce, 2014). (c) La/Sm vs Sm/Yb plot. Mantle array (heavy line) defined by depleted MORB mantle (DMM; McKenzie & O’Nions, 1991) and primitive mantle (PM; Sun & McDonough, 1989); Melting curves for spinel lherzolite and garnet peridotite with both DMM and PM compositions are after Aldanmaz et al. (2000). Numbers along lines represent the degree of partial melting. (d) ϵ _Nd(t) vs (⁸⁷Sr/⁸⁶Sr) _i plot. Mantle arrays are after Zindler and Hart (1986). Bulk Earth is after Depaolo (1988). OIB is after Wilson (1989). The date of Sumdo eclogite is from Li et al. (2009). Dates of Halberstadt, Gerze NEBs and HNBs are after Hastie et al. (2011) and Hao et al. (2018).

Compared to normal arc basalts, the higher TiO₂ (2.38–5.37 wt %), P₂O₅ (0.36–1.42 wt %), Nb (45.3–113.5 ppm) and Nb/Yb (14.6–38.1) suggest a deep mantle origin for the Yeqing gabbro. These samples exhibit LREE-enriched chondrite-normalized REE patterns and high (La/Yb)_PM (11.51–27.12) and (Ce/Yb)_PM (9.91–21.87) ratios, similar to those of basalts derived from garnet lherzolite (Hart & Dunn, 1993; Hauri et al. 1994). These samples also plot in the garnet lherzolite field with a low degree of partial melting in the La/Sm vs Sm/Yb plot near the a-type NEBs/HNBs and far from b-type NEBs/HNBs (Fig. 9c; Aldanmaz et al. 2000), indicating that the magma may be derived from a garnet-stable region (>85 km; Robinson & Wood, 1998) with lesser addition of N-MORB components, where it is generally considered to generate OIB (Niu, 2009). However, according to mantle heterogeneity, the regional evolution and compotation of the study area must be taken into account when considering a ‘true’ OIB or the depleted and enriched mantle model (Hastie et al. 2011). There are several pieces of evidence about oceanic islands of the SPTO found in the TSACB, indicating that a plume had indeed existed in the SPTO (B Wang et al. 2019; Zhong et al. 2021; Duan et al. 2022). Consequently, there exists a true enriched OIB mantle beneath the SPTO crust as the material source of the Yeqing gabbro.

In addition, the Yeqing gabbro has elevated FeO^T (11.54–14.72 wt %) and TiO₂ (2.38–5.37 wt %) like the Fe–Ti basalts which are defined by >12 wt % FeO^T and >2 wt % TiO₂ (Sinton et al. 1983). Most of the Fe–Ti basalts show MORB affinity, such as lower concentrations of MgO, CaO and Al₂O₃. rather than N-MORB (Hollis et al. 2012). The gabbro in this study has a slightly lower MgO (average of 6.06 wt %) and CaO (average of 10.89 wt %) than N-MORB (MgO with average of 7.60 wt %; CaO with average of 11.48 wt %), which may be related to the fractional crystallization of olivine and a little pyroxene. The Al₂O₃ (average of 15.39 wt %) is slightly higher than N-MORB (average of 14.85 wt %). Besides, the Yeqing gabbro shows apparent LREE enrichment and HREE depletion in the REE diagram, and large slope curve in the trace-element diagram, in contrast to the Fe–Ti basalts with flat–modest slope curve in the REE and trace-element diagrams (Sinton et al. 1983; Hollis et al. 2012).

Relative to MORB, the zircon Hf isotopic composition of OIB has a narrow range and a lower ϵ _Hf(t) value, and the mantle typically has a positive ϵ _Hf(t) value (Nowell et al. 1998; Dobosi et al. 2003; Wu et al. 2007). Zircon ϵ _Hf(t) values of these samples are −0.2–6.3 which is lower than those of HNBs from Duobuzha and Rena Tso (2.44–11.64 and 1.9–7.6 respectively; Li et al. 2016) and similar to those of OIBs. The whole-rock Sr–Nd isotopic data of these samples exhibit slightly high (⁸⁷Sr^/86Sr) _i (0.70468–0.70542), low ¹⁴³Nd/¹⁴⁴Nd (0.51253–0.51265) and low ϵ _Nd(t) (0.3–2.7) relative to some other HNBs from mantle wedge (⁸⁷Sr^/86Sr of 0.70341–0.70484; ¹⁴³Nd/¹⁴⁴Nd of 0.51292–0.51307; ϵ _Nd(t) of 2.57–6.80) (Wang et al. 2007; Hastie et al. 2011; Xu et al. 2017; Hao et al. 2018). As we can see, the Yeqing gabbro has a more enriched Sr–Nd isotopic composition than Sumdo eclogite and NEBs/HNBs from Halberstadt in the ϵ_Nd(t) vs (⁸⁷Sr/⁸⁶Sr) _i plot (Fig. 9d). The Halberstadt NEBs and HNBs exhibit relatively depleted features, which were generated from the mantle wedge metasomatized by slab melt (Fig. 9d; Hastie et al. 2011). An analogous Sr–Nd isotopic composition could be found between Yeqing gabbro and Gerze HNBs, which was produced by partial melting of an OIB-type source component involving upwelling asthenosphere mantle (Fig. 9d; Li et al. 2016). In conclusion, the geochemistry and isotopic signatures of the Yeqing gabbro samples can be interpreted as the production of a low-degree partial melting of garnet lherzolite mantle with the negligible contribution of subducted oceanic crust.

6.b. Tectonic setting

HNBs are not intraplate lavas like OIB and other alkaline lavas (Adam & Green, 2010); they are only found in subduction zone environments (Defant et al. 1992), such as Zamboanga Peninsula (Philippines; Sajona et al. 1996), Sulu (southern Philippines; Castillo et al. 2007), Baja (California; Luhr et al. 1995; Aguillón-Robles et al. 2001; Castillo, 2008), Halberstadt (Germany; Hastie et al. 2011), Kamchatka (Russia; Kepezhinskas et al. 1996), Nicaragua (Gazel et al. 2011), Tuotuohe (Tibet; Wang et al. 2007), Rena Tso (Tibet; Li et al. 2016), Duobuzha (Tibet; Li et al. 2016; Xu et al. 2017) and Gerze (Tibet; Hao et al. 2018). Actually, there are multiple hypotheses about the formation of HNBs and NEBs, which are linked with some specific subduction processes, like flat subduction, slab rollback, slab break-off, ridge subduction or plume – subduction-zone interaction (Thorkelson, 1996; Wang et al. 2007; Gazel et al. 2011; Thorkelson et al. 2011; Li et al. 2016; Xu et al. 2017; Hao et al. 2018; Wu et al. 2018). In the case of the HNBs-type gabbro investigated in this study, the hypothesis of plume – subduction-zone interaction is invoked to interpret its generation based on the following evidence.

1. (1) The Yeqing gabbro is a part of the arc-volcanism system of the SPTO, which is supported by its spatial distribution and intrusion time.

In fact, lots of HNBs and NEBs have been reported on the east coast of the Central Pacific as abnormal arc magmatism within the modern oceanic island arc system (Castillo, 2008; Hoernle et al. 2008; Gazel et al. 2011; Fletcher & Wyman, 2015). The subduction polarity of the SPTO is from south to north (ZL Li et al. 2009; Yang et al. 2009; Zhu et al. 2010; Mai et al. 2021; YM Li et al. 2022), proved by the spatial distribution characteristics of the oceanic crust (ophiolites, oceanic islands; Chen et al. 2010; Duan et al. 2019, 2022; B Wang et al. 2019, 2021; Zhong et al. 2021), trench and initial fore-arc basin (Xie et al. 2019, 2021) and arc magmatism (Geng et al. 2009; Zhu et al. 2010; GM Li et al. 2020; B Wang et al. 2020; 2022; Mai et al. 2021; N Li et al. 2022; Song et al. 2022; C Wang et al. 2022) from south to north. The intrusion time of the Yeqing gabbro is coincident with the period of the SPTO subduction in the early–middle Permian and middle–late Triassic by studying the eclogites (Li et al. 2009; Cheng et al. 2012, 2015; Zhang et al. 2018 a, b), which is also proved by the contemporaneous arc magmatism mentioned above (Fig. 10).

1. (2) The flat subduction and slab rollback could scarcely happen in the SPTO.

Fig. 10 Distribution diagram of ages related to the Sumdo Palaeo-Tethys Ocean (data are quoted from Geng et al. 2009; Yang et al. 2009; Zhu et al. 2009, 2010; Cheng et al. 2012, 2014; Zhang et al. 2018; Duan et al. 2019, 2022; Xie et al. 2019, 2021; B Wang et al. 2019, 2020, 2021 , 2022; Li et al. 2020, 2022; Mai et al. 2021; Zhong et al. 2021; Song et al. 2022; C Wang et al. 2022).

The young ocean slab with slighter density subducted with a low angle like flat subduction in the early stage due to greater buoyancy, while the density of the subduction slab increased after the subduction slab dehydrated and then slab rollback occurred (Klein & Langmuir, 1987; Hawkins et al. 1990). However, the SPTO had subducted to a deep mantle in the middle Permian (Li et al. 2009; Cheng et al. 2012, 2015; Zhang et al. 2018 b), indicating that slab rollback and flat subduction is unlikely to have taken place during the late Permian to early Triassic. A magma belt parallel to the subduction belt is commonly required to respond to the asthenosphere upwelling after slab rollback, whereas no analogous magma was discovered in the Sumdo area to constitute a magma belt with the Yeqing gabbro.

1. (3) The slab break-off and ridge subduction are unlikely to lead to formation of the Yeqing gabbro.

Slab break-off generally occurs c. 10 Ma after a continental collision (Benoit et al. 2002; Wu et al. 2018). The closure of the SPTO occurred later at least than the late Triassic, supported by the discovery of middle Triassic ophiolite (232–231 Ma; Duan et al. 2019) in Sumdo and a late Triassic – early Jurassic medium-pressure metamorphic belt (225–192 Ma; Dong et al. 2011, 2015; Lin et al. 2013; Zhang et al. 2014, 2018 b) between the South and North Lhasa terranes accompanied by the coeval magmatism with a geochemical affinity to syn- or post-collisional plutons (227–180 Ma; HF Zhang et al. 2007; Zhu et al. 2011; Li et al. 2012; C Zhang et al. 2018 b). Thus, slab break-off did not occur in the SPTO. The Yeqing gabbro is also unlikely to have formed in ridge subduction, which requires a huge volume of magmatism response, such as A-type granite, adakite, high-Mg andesite and high-temperature metamorphic rocks (Hole et al. 1991; McCrory et al. 2009; Xu et al. 2017), that is missing in the Sumdo area.

1. (4) The back-arc basin may not be a better position to form the Yeqing gabbro.

Geochemistry characteristics of the Yeqing gabbro show Fe–Ti–P enrichment, and lack of arc- and contamination signature. Fe–Ti-enriched basalts are confined to extensional settings with upwelling of the asthenosphere and have been reported from back-arc basins in the subduction belt (Hollis et al. 2012). However, there have been no reports about the sedimentary rock of a back-arc basin in the Sumdo area. On the other hand, the back-arc basin basalts are characterized by mainly strong arc affinity in the early stage, and clear MORB affinity in the late stage (Klein & Langmuir, 1987; Hawkins et al. 1990). In fact, the arc magmatism in the Sumdo area shows typical island-arc characteristics (Zhu et al. 2010; B Wang et al. 2020, 2022; Mai et al. 2021; Li et al. 2022; C Wang et al. 2022), which also supports the absence of a back-arc basin.

1. (5) The plume – subduction-zone interaction may be a better model to explain the formation of Yeqing gabbro.

As discussed above, the Yeqing gabbro derived from a deep garnet lherzolite mantle with the negligible contribution of subducted oceanic crust. The intrusion from deep garnet lherzolite mantle in the subduction belt is usually related to asthenosphere upwelling in an extension environment, which is caused by a slab window or slab rollback. The slab rollback model has been excluded already. Besides, the Yeqing gabbro samples plot in within-plate alkali basalts and ocean island alkaline field like the a-type NEBs/HNBs that are caused by ridge subduction or plume – subduction-zone interaction, different from the b-type NEBs/HNBs caused by flat subduction or slab rollback (Fig. 5c, d). In the P₂O₅ vs TiO₂ and Nb/Yb vs Nb figure (Fig. 11a, b), most of these samples plot in the slab window field. The slab window is common in slab break-off or ridge subduction, which are not suitable for this study. Moreover, the seismic ridge, a series of seamount island chains formed by oceanic crust moving over fixed hotspot/mantle plume, could bring about a slab window while subducting (Gazel et al. 2011; Fletcher & Wyman, 2015). Actually, researchers have reported multiple ocean islands formed from the latest early Carboniferous to middle Permian (B Wang et al. 2019; Zhong et al. 2021; Duan et al. 2022), implying the presence of seismic ridge in the SPTO. At the same time, the hotspot/mantle plume could provide a perfect mantle source for the Yeqing gabbro and favourable conditions for the possibility of plume – subduction-zone interaction.

Fig. 11 (a) P₂O₅ vs TiO₂ figure (Li et al. 2016). (b) Nb/Yb vs Nb figure (Li et al. 2016). Slab window basalts are from Li et al. (2016 and references therein).

6.c. Geological significance for the subduction evolution of the SPTO

According to the discussion of petrogenesis and tectonic setting above, we propose that the hotspot/mantle plume began to interact with the subduction zone in the SPTO after the middle Permian, and the hotter plume material upwelled through a slab window intruding into the continental edge during the late Permian to early Triassic. Under this framework, the formation of the Yeqing gabbro and subduction evolution of the SPTO can be briefly described as follows combined with the regional geology.

Research into ophiolites and oceanic islands has revealed that the SPTO had already developed an initial ocean basin in the early Carboniferous (Wang et al. 2021; Duan et al. 2022). The SPTO began to subduct not later than the early Permian and subducted to a greater depth beneath the north Lhasa terrane during the middle Permian (Fig. 12a; Yang et al. 2006; Cheng et al. 2012; Weller et al. 2016). Meanwhile, the middle Permian Wenmulang and Ewulang ocean islands imply that the plume under the ocean plate was still active during the middle Permian (Fig. 12a; B Wang et al. 2019; Zhong et al. 2021). Until the late Permian, the seismic ridge in the SPTO, a structurally weak position of oceanic slab, subducted into the trench, and the slab was torn in the frail place forming a slab window (Gazel et al. 2011) (Fig. 12b, c). The detached slab was replaced by a hot and buoyant asthenosphere mantle, which generated the Yeqing gabbro in this study (Fig. 12b, c). This is the hypothesis put forward in this study to explain the generation of the Yeqing gabbro. Nevertheless, it is mainly based on petrological, geochronological and geochemical observations. Further studies will be required to verify this hypothesis.

Fig. 12 Reconstructed palaeogeography and subduction model of the Sumdo Palaeo–Tethys Ocean during the middle Permian (a) and late Permian to early Triassic (b) (modified after Torsvik & Cocks, 2013; Xie et al. 2021). (c) The schematic model of the geological processes of plume – subduction-zone interactions required to explain the formation of Yeqing gabbro (modified after Gazel et al. 2011). GI, Greater India; S, Sibumasu; NL, North Lhasa; SL, South Lhasa; T, Tengchong.

Conclusions

1. (1) LA-ICP-MS U–Pb ages of zircon from Yeqing gabbro are 254–249 Ma, late Permian to early Triassic, which represents the magmatic crystallization age of the Yeqing gabbro.

2. (2) The Yeqing gabbro exhibits positive Nb–Ta anomalies, Fe–Ti–P enrichment, lack of arc- and contamination signature, similar to those of OIB and HNBs, indicating that the Yeqing gabbro may be the product of a low degree of partial melting of garnet lherzolite mantle generated from an extensional environment in the subduction belt.

3. (3) Considering the regional geology of the SPTO, a slab window produced by the plume – subduction-zone interaction is a better explanation for the formation of Yeqing gabbro, proving the SPTO continued to subduct during late Permian to early Triassic.