Introduction

The Bangong–Nujiang Suture Zone (BNSZ) on the Tibetan Plateau is an important tectonic boundary and metallogenic belt in the eastern section of the Tethys tectonic domain. However, the nature and tectonic evolution of the Bangong–Nujiang Tethyan Ocean (BNTO), represented by the BNSZ, are debated. For example, some scholars have proposed that the BNTO was a long-lived ocean basin that developed continuously from the Paleozoic to the Mesozoic, and that it represents the main basin of the Paleo-Tethys Ocean on the Tibetan Plateau (Pan et al. 1997, 2006, 2012; Wang et al. 2002; Zhu et al. 2013). Others inferred that the BNTO developed mainly during the Mesozoic, opening at first during the Late Permian–Early Triassic, and finally closing during the Cretaceous. In this model, the ocean either represented the northern arm of the Neo-Tethys Ocean (Chang and Zhen 1973; Chang 1978; Smith and Xu 1988; Yin 1997; Ren and Xiao 2004; Shi et al. 2008; Huang et al. 2012; Liu et al. 2014; Fan et al. 2014b, 2015b), or it represented the Meso-Tethys Ocean (Girardeau et al. 1984; Zhou et al. 1997; Wang 2000; Yong and Jia 2000; Wang et al. 2008). Others suggested that the BNTO was a short-lived ocean basin that existed for no more than 100 Myr from the Late Triassic to the Late Jurassic (Kapp et al. 2003, 2007; Qiu et al. 2004; Zhang 2007; Xia et al. 2008; Qu et al. 2010). In this model, it is inferred that the ocean either represented a back-arc ocean basin of the Jinshajiang Paleo-Tethys Ocean or of the Indus–Yarlung Zangbo Neo-Tethys Ocean (Wang et al. 1987). Some scholars have even proposed that the BNTO was not an ocean in the strict sense (Lu et al. 2003; Zhao et al. 2004).

The fundamental problem with the above arguments is that the oldest reliable ages of oceanic BNSZ rocks (e.g., ophiolites and ocean islands) are Late Triassic (Fan et al. 2017) and Early–Middle Jurassic (Qiu et al. 2004; Xia et al. 2008; Shi et al. 2008; Qu et al. 2010; Zhang et al. 2014; Wang et al. 2008, 2013, 2016), and there is a general absence of earlier oceanic remnants, which means little information is available on how and when the early BNTO developed. It is therefore important to find and analyze oceanic remnants of the BNTO in the BNSZ that pre-date the Late Triassic.

In this paper, we describe for the first time the newly discovered Middle Triassic Nare ocean island fragment (NaOI) in the middle segment of the BNSZ (Fig. 1a). We present the results of detailed geochemical, geochronological, and Lu–Hf isotopic analyses of the basalts and phonolites of the NaOI, and we use the data to discuss the genesis and evolution of these rocks. Based on these data and the results of research on modern ocean islands, we also discuss the Middle Triassic evolution of the BNTO.

Fig. 1
figure 1

a Tectonic framework of the Tibetan Plateau and b geological map of the study area (modified after Fan et al. 2014b). Cz Cenozoic, K 2 j Late Cretaceous Jingzhushan Formation, K 1 l Early Cretaceous Langshan Formation, K 1 z Early Cretaceous Zhaga Formation, J 2 jb Middle Jurassic Jiebuqu Formation, J 1-2 s Early to Middle Jurassic Sewa Formation, JM Late Triassic–Jurassic Mugagangri Group, T 3 w Late Triassic Wuga Formation, Pz late Carboniferous to early Permian Zhanjin Formation, Oph (D) Dong Co Ophiolite, Oph (K) Kangqiong Ophiolite, ZhOI Zhonggang ocean island fragment, NOI Nadong ocean island fragment, NaOI Nare ocean island fragment

Full size image

Geological background and petrological observations

The Tibetan Plateau is located in the eastern section of the Alpine–Himalayan tectonic domain, and is divided from south to north into the Himalayan, Lhasa, Southern Qiangtang, Northern Qiangtang, and Bayan Har–Garze terranes (Fig. 1a; Allègre et al. 1984; Dewey et al. 1988; Pan et al. 1997, 2012; Yin and Harrison 2000; Li et al. 2006; Wang et al. 2013). These terranes are separated by the Indus–Yarlung Zangbo Suture Zone (IYZSZ), the Bangong–Nujiang Suture Zone (BNSZ), the Longmuco–Shuanghu–Lancangjiang Suture Zone (LSLSZ), and the Jinshajiang Suture Zone (JSSZ; Fig. 1a), respectively.

The BNSZ represents the remains of the BNTO and extends eastwards for 2000 km through the areas of Bangong Co, Gerze, Dongqiao, Dingqing, and Jiyuqiao, and into Burma, Thailand, and Malaysia (Fig. 1a; Pan et al. 1997, 2012). The BNSZ in China is divided into three segments, which from west to east are the Bangong Co–Gerze, Gerze–Dingqing, and Dingqing–Nujiang segments (Bureau of Geology and Mineral Exploration of Tibet Province 1993; Pan et al. 1997; Qiu et al. 2004).

The study area is located in the Dong Co region, ~ 90 km to the east of Gerze County in the middle segment of the BNSZ (Fig. 1a). The area contains a number of approximately E–W trending regional structures (Fig. 1b), and the rocks are strongly deformed. The study area contains numerous ophiolites and ocean island fragments, as well as sediments of the Zhanjin, Wuga, Mugagangri, Sewa, Jiebuqu, Langshan, and Jingzhushan groups and formations (Fig. 1b). The Late Carboniferous to Early Permian Zhanjin Formation in this area is dominated by metasandstones and phyllite, whereas the Late Triassic Wuga Formation is dominated by sandstone. The Late Triassic–Jurassic Mugagangri Group consists of deep sea to bathypelagic flysch deposits that formed within the BNTO (Bureau of Geology and Mineral Exploration of Tibet Province 1993; Wang and Zhong 2002; Zeng et al. 2014). The Early–Middle Jurassic Sewa Formation and the Middle Jurassic Jiebuqu Formation are both dominated by shallow marine limestone and sandstone that formed within the BNTO. The Early Cretaceous Langshan Formation consists mainly of limestones that include shallow marine carbonate platform deposits. The newly defined Early Cretaceous Zhaga Formation is dominated by graywacke and shale, and shows load casts, convolute bedding, and a–b–c–d Bouma sequences indicating bathyal to abyssal turbidite deposition (Fan et al. 2015a). The conglomerate- and sandstone-dominated Late Cretaceous Jingzhushan Formation consists of post-collisional molasse that formed within the BNSZ.

The ophiolites in the study area include the Kangqiong and Dong Co ophiolites. The Kangqiong Ophiolite is located in the eastern part of the study area (Fig. 1b), and is dominated by peridotite, cumulate gabbro, plagiogranite, gabbro, and basalt. A zircon U–Pb age of 115 Ma indicates this ophiolite formed during the Early Cretaceous (Xu et al. 2015). Additional studies have determined that the Kangqiong Ophiolite is a supra-subduction zone (SSZ)-type ophiolite that formed in an extensional fore-arc basin that resulted from intra-oceanic subduction of the Bangong–Nujiang Tethyan oceanic crust during the Early Cretaceous (Xu et al. 2015).

The Dong Co Ophiolite, located in the center of the study area, was emplaced tectonically into the Late Triassic–Jurassic Mugagangri Group and the Jurassic to Early Cretaceous Zhonggang ocean island fragment (Fig. 1b). The Dong Co Ophiolite represents a section of oceanic crust that is more than 5 km thick and consists of ultramafic rocks, serpentinite, troctolite, gabbro (either massive or exhibiting cumulate textures), and a sheeted dike complex of diabase, pillow basalt, plagiogranite, and minor radiolarian chert (Qiu et al. 2004; Bao et al. 2007; Zhang 2007; Zeng et al. 2010; Li et al. 2013; Wang et al. 2016). In addition, the Dong Co Ophiolite contains the dunite–troctolite–olivine gabbro (DTG) series of rocks (Bao et al. 2007), as well as minor amphibolite facies metamorphic rocks (Wang et al. 2008).

The interpretation that the Dong Co Ophiolite formed during the Jurassic was based mainly on the following two lines of evidence: (1) an Early Jurassic Sm–Nb isotopic age of 191 ± 22 Ma was obtained from the cumulate gabbro of the Dong Co Ophiolite, and Late Jurassic to Early Cretaceous K–Ar ages of 140 ± 4.07 and 152.30 ± 3.60 Ma were obtained from the same cumulate gabbro samples (Qiu et al. 2004); and (2) some Jurassic radiolarians occur in chert in the Dong Co area surrounding the Dong Co Ophiolite (Zeng et al. 2010). However, Bao et al. (2007) recently obtained Early Cretaceous ages of 132 Ma (SHRIMP U–Pb zircon) from the cumulate gabbro, and 137 Ma (whole rock 40Ar–39Ar) and 141 Ma (whole rock 40Ar–39Ar) from the basalt of the Dong Co Ophiolite, indicating that the Dong Co Ophiolite formed during the Early Cretaceous (Bao et al. 2007). In addition, Wang et al. (2016) obtained a Middle Jurassic age of 164 Ma (LA–ICP–MS zircon U–Pb) from the cumulate gabbro (Wang et al. 2016). Given that geochronological studies of the Dong Co Ophiolite have reported Jurassic and Cretaceous ages, we propose that the Dong Co Ophiolite represents a multi-stage tectonic mélange that formed during the Jurassic to Early Cretaceous.

The ocean island fragments in the study area include Zhonggang, Nadong, and Nare (Fig. 1b), and these different fragments represent different ocean islands that formed in the BNTO with different ages. The Zhonggang ocean island fragment is the largest ocean island fragment in the BNSZ, covering an area of > 400 km2 (Zeng et al. 2010; Fan et al. 2014b). Similar to modern ocean islands, the Zhonggang ocean island fragment has a double-layered lithological structure with a mafic basement of basalt and gabbro and an oceanic sedimentary cover sequence of limestone, chert, and conglomerate (Fan et al. 2014b). Early Jurassic and Early Cretaceous U–Pb zircon and 40Ar/39Ar ages indicate the Zhonggang ocean island fragment formed from a long-lived hotspot that was active from the Early Jurassic to the Early Cretaceous (Zeng et al. 2010; Zhang et al. 2014; Fan et al. 2014b, 2015b), and these age data supported that the BNTO was still at least partially open, and developing during the Early Cretaceous. The Nadong ocean island fragment covers an area of > 50 km2, has an E–W extent of > 25 km, and is dominated by basalt, gabbro, limestone, and conglomerate. Jurassic fossils and U–Pb zircon ages have been obtained from this ocean island fragment, indicating formation in the Jurassic (Fan et al. 2014a). The existence of the Jurassic Nadong ocean island fragment indicates that the BNTO was ongoing during this time.

The Nare ocean island fragment (NaOI) covers an area of > 5 km2 (Fig. 1b). Similar to the Zhonggang and Nadong ocean island fragments, the NaOI has a double-layered lithological structure with a basaltic basement and an oceanic sedimentary cover sequence of limestone and conglomerate with eruptive phonolite at the top of the sequence. Three stages in the evolution of NaOI have been identified: (1) formation of the basaltic basement, (2) deposition of the oceanic sedimentary cover sequence (basaltic conglomerate and limestone), and (3) the eruption of layers of evolved phonolitic magma within the upper part of the sedimentary sequence (Fig. 2). It was initially thought that the contact between the NaOI and the Zhaga Formation was conformable (Fan et al. 2015a), but detailed research as part of the present study has shown a fault contact (Fig. 1b), as indicated by the existence of tectonic deformation (Fig. 3a), tectonic breccias (Fig. 3b), and fault springs along the contact of the NaOI and the Zhaga Formation.

Fig. 2
figure 2

Stratigraphic column for the Nare ocean island fragment

Full size image
Fig. 3
figure 3

a Photograph of tectonic deformation within the Zhaga Formation along the contact of the NaOI and the Zhaga Formation. b Photograph of tectonic breccia within the NaOI along the contact of the NaOI and the Zhaga Formation. c Photograph of NaOI basalt. d Photomicrograph of NaOI basalt. e Photograph of NaOI phonolite. f Photomicrograph of NaOI phonolite

Full size image

The basalts within the NaOI are generally gray–green in color (Fig. 3c), and are present as individual units that generally have thicknesses of > 10 m (Fig. 2). The basalts are aphanitic with an intergranular matrix (Fig. 3d). The basalts contain numerous amygdales (up to 30%) of calcite (Fig. 3d). The basaltic conglomerates within the NaOI contain gravel clasts and matrix, both of which are made up of basaltic debris (Fig. 2), and it is noteworthy that the basaltic conglomerates contain no terrestrial debris. The conglomerates provide evidence of a sedimentary environment with a restricted provenance, proximal deposition in a far-shore type setting, and rapid accumulation in the ocean that surrounded the island. The limestones within the NaOI form thick layers that lack terrestrial debris (Fig. 2), and they record deposition in an isolated, clear water sedimentary environment associated with a carbonate platform. The phonolites, which occur as layers within the upper part of the sequence of NaOI cover sediments (Fig. 2), are present as individual units that have total thicknesses of > 30 m. The phonolites are generally gray–white in color (Fig. 3e), and are aphanitic with an interwoven matrix consisting of abundant alkali feldspar with interstitial plagioclase, feldspathoids, and rare mafic minerals (Fig. 3f).

Analytical methods

Whole–rock major and trace element analyses

Five NaOI basalt samples and fifteen NaOI phonolite samples were collected for whole-rock major and trace element analyses. Samples were trimmed to remove weathered surfaces, cleaned with deionized water, crushed, and powdered in an agate mill. Major element compositions were analyzed by inductively coupled plasma–optical emission spectroscopy (ICP–OES; Leeman Prodigy) with high-dispersion Echelle optics at the China University of Geosciences, Beijing, China. Loss on ignition (LOI) values were determined by heating 1 g of sample in a furnace at 1000 °C for several hours before being cooled in a desiccator and reweighed. Analytical uncertainties for the major elements are generally < 1 wt%. Trace element compositions were analyzed by inductively coupled plasma–mass spectrometry (ICP-MS; Agilent-7500a) at the China University of Geosciences, Beijing, China. Analytical accuracy and precision during trace element analysis were calculated from analyses of two standards: AGV-2 (US Geological Survey standard) and GSR-3 (Chinese national geological reference standard). Analytical accuracy, calculated from the relative difference between measured and recommended values, is < 5 wt% for most elements. Analyses of the AGV-2 and GSR-3 international standards are in good agreement with recommended values (Govindaraju 1994). For details on the analytical procedures, see Zhai et al. (2013).

In situ zircon U–Pb LA–ICP–MS analysis

Two samples from the NaOI phonolite were selected for in situ zircon U–Pb analyses. The zircons were separated by conventional heavy liquid and magnetic techniques at the Special Laboratory of the Geological Team of Hebei Province, Langfang, China. The internal structures of zircon were characterized under cathodoluminescence (CL) at the Institute of Physics, Peking University, Beijing, China to select analytical points for laser ablation–inductively coupled plasma–mass spectroscopy (LA–ICP–MS) analyses. U, Th, and Pb isotopes and trace element zircon analyses were carried out by LA–ICP–MS at the Geological Laboratory Centre, China University of Geosciences, Beijing, China. The spot size was 36 and 25 µm for each sample. Helium was used as a carrier gas. Reference and internal zircon standards 91500 (Wiedenbeck et al. 1995) and NIST610 (29Si), respectively, were used for instrument calibration. The Pb correction method of Anderson (2002) was applied. For further details of the analytical techniques, see Yuan et al. (2004). Reported uncertainties for the age analyses are given as 1σ values with weighted mean ages in the 95% confidence level. Isotopic data were processed using the GLITTER (version 4.4) and Isoplot/Ex (version 3.0) programs (Ludwig 2003).

In situ zircon Hf isotope analyses

In situ zircon Hf isotope analyses were performed on 17 previously dated zircon grains from the NaOI phonolite samples. The points used for the Hf analyses were the same as those used during the LA–ICP–MS analyses, which were selected from CL images. The data were collected using a NEPTUNE Plus MC–ICP–MS at the Beijing Createch Testing Technology Co., Ltd, Beijing. A single spot ablation mode with a spot size of 55 μm was used to acquire the data. Each measurement consisted of 20 s of background signal acquisition followed by 50 s of ablation signal acquisition. Detailed information on the operating conditions and analytical methods can be found in Hu et al. (2012). The analyzed 176Hf/177Hf ratios for the zircon standard (91500) were 0.282299 ± 31 (2σ n , n-40), which are similar to the 176Hf/177Hf ratios of 0.282302 ± 8 and 0.282306 ± 8 (2σ) for the standard when determined by the solution method (Goolaerts et al. 2004; Woodhead et al. 2004). Off-line selection, integration of analyte signals, and mass bias calibrations were performed using the ICP–MS DataCal program (Liu et al. 2010).

Results

Whole-rock major and trace element geochemistry

NaOI basalts

Whole-rock major and trace element data for the NaOI basalts are given in Table 1. The NaOI basalts have undergone varying degrees of carbonate alteration (Fig. 3d), resulting in variable values of LOI and changes in the concentrations of mobile elements (e.g., Na, K, Ca, Cs, Rb, Ba, and Sr) compared with protolith values. However, concentrations and ratios of immobile elements (e.g., REE, Nb, Ta, Zr, Hf, Ti, and P) and transition metal elements (e.g., V, Ni, Cr, and Fe) have not been affected by these processes, and can therefore be used to investigate the petrogenesis and tectonic setting of the samples.

Table 1 Major (wt%) and trace element (ppm) data for the basalt in the Nare Ocean Island
Full size table

The five basalt samples have variable contents of SiO2 (45.1–52.1 wt%), high contents of TiO2 (2.90–3.31 wt%), Fe2O3 (9.14–11.5 wt%), MgO (4.41–7.47 wt%), CaO (7.61–12.1 wt%), and Na2O (3.09–3.76 wt%), low contents of K2O (1.23–1.85 wt%), and low Mg# values (53–61). The NaOI basalts have much higher TiO2 contents (2.90–3.31 wt%) than typical island-arc volcanic rocks (0.58–0.85 wt%) and mid-ocean ridge basalts (MORB; 1.0–1.5 wt%), but similar contents to those of OIB (2.20 wt%; Hishashi 1954; Irvine and Baragar 1971; Winchester and Flody 1977; Pearce 1983).

The samples of NaOI basalt have high total REE concentrations (147–180 ppm), and they exhibit light rare earth element (LREE)-enriched chondrite-normalized REE patterns (LaN/YbN = 8.80–10.4; Fig. 4a; Sun and McDough 1989). These samples are enriched in high field strength elements (HFSEs; Nb, Ta, Zr, Hf, and Ti) and exhibit primitive mantle-normalized trace element patterns similar to those of OIB (Fig. 4b; Sun and McDough 1989).

Fig. 4
figure 4

a Chondrite-normalized REE variation diagram for samples of NaOI basalt and phonolite. b Primitive mantle-normalized trace element variation diagram for samples of NaOI basalt and phonolite

Full size image

All the NaOI basalt samples fall in the field of alkali basalt on a Zr/TiO2*0.0001 vs. Nb/Y immobile element classification diagram (Fig. 5a; Winchester and Flody 1977), and these samples plot in the intraplate basalt (ocean island basalt) fields on the V vs. Ti/1000 diagram (Fig. 5b), the Ti/100 vs. Zr vs. (Y × 3) diagram (Fig. 5c; Pearce and Cann 1973), and the (2 × Nb) vs. Zr/4 vs. Y diagram (Fig. 5d; Meschede 1986), indicating they formed from OIB-type magma.

Fig. 5
figure 5

a Immobile element-based classification diagram showing the nature of the basalts and phonolites analyzed during this study. b NaOI basalts plotted on the V vs. Ti/100 diagram. c NaOI basalts plotted on the Ti/100 vs. Zr vs. Y × 3 diagram (A island-arc tholeiite, B mid-ocean ridge basalt (MORB) + island-arc tholeiite and calc-alkali basalt, C calc-alkaline basalt, D intraplate basalt). d NaOI basalts plotted on the 2 × Nb vs. Zr/4 vs. Y diagram (AI intraplate alkali basalt, AII intraplate alkali basalt and tholeiite, B E-type MORB, C intraplate tholeiite and volcanic arc basalt, D N-type MORB and volcanic arc basalt). e Al2O3 vs. SiO2 diagram. f Zr vs. SiO2 diagram

Full size image

NaOI phonolites

Whole-rock major and trace element data for the NaOI phonolites are given in Table 2. Although the NaOI basalts have undergone varying degrees of carbonate alteration, very limited alteration has occurred within the NaOI phonolites, as indicated by the field and petrographic observations (Fig. 3e, f), and supported by the very low LOI values (0.90–1.91 wt%) and the consistent primitive mantle-normalized multi-element variation diagram patterns, including the LILE patterns of these samples. Mobile elements (e.g., Na, K, Ca, Cs, Rb, Ba, and Sr) can therefore be used to investigate the petrogenesis and tectonic setting of the samples.

Table 2 Major (wt%) and trace element (ppm) data for the phonolite in the Nare Ocean Island
Full size table

Fifteen samples of the NaOI phonolite have high contents of SiO2 (60.3–65.3 wt%), Al2O3 (16.8–19.3 wt%), Na2O (6.12–8.27 wt%), K2O (2.85–5.19 wt%), and Fe2O3 (4.51–5.37 wt%), low contents of TiO2 (0.15–0.20 wt%) and MgO (0.11–0.25 wt%), and very low Mg# values (5–10). Similar to the samples of the NaOI basalt, the samples of the NaOI phonolite have high total REE concentrations (430–619 ppm), and they exhibit LREE-enriched chondrite-normalized REE patterns (LaN/YbN = 10.8–13.3) with negative Eu anomalies (Fig. 4a; Eu/Eu* = 0.41–0.43; Sun and McDough 1989). The NaOI phonolites are enriched in Nb, Ta, Zr, and Hf, and depleted in Ba, U, K, Sr, P, and Ti in the primitive mantle-normalized trace element patterns (Fig. 4b; Sun and McDough 1989). In addition, all the NaOI phonolite samples fall in the phonolite field on a Zr/TiO2*0.0001 vs. Nb/Y immobile element classification diagram (Fig. 5a; Winchester and Flody 1977).

Zircon U–Pb ages

LA–ICP–MS U–Pb data for the zircons in the NaOI phonolites are given in Tables 3 and 4. The zircons are generally hypidiomorphic short prisms (50–200 µm long) with aspect ratios of 1:1 to 2:1. They are typically gray, and exhibit widely magmatic oscillatory zoning in CL images (Fig. 6). Zircon grains from the phonolites of NaOI have ΣREE contents of 448–8104 ppm, with an average of 3099 ppm. All the zircons display fractionated rare earth element (REE) patterns, heavy REE (HREE) enrichment and light REE (HREE) depletion, and clear negative Eu and variable positive Ce anomalies (Fig. 6). Their Th/U ratios are 0.27–2.67, with an average of 1.30. All the above features indicate that the zircons are of magmatic origin (Hoskin and Black 2000). The zircons from the phonolites yielded Middle Triassic weighted mean 206Pb/238U ages of 238.8 ± 2.8 Ma (MSWD = 3.30) and 241.9 ± 1.9 Ma (MSWD = 1.90) (Fig. 6).

Table 3 Zircon LA–ICP–MS U–Pb data of the phonolite in the Nare Ocean Island
Full size table
Table 4 Trace element data of the zircon grains from phonolite samples of the Nare Ocean Island
Full size table
Fig. 6
figure 6

CL images of zircons and concordia diagrams showing the results of LA–ICP–MS analyses of zircons from the NaOI phonolites

Full size image

Because the phonolites occur as layers within the upper part of the sequence of NaOI cover sediments (Fig. 2), the Middle Triassic ages of the NaOI phonolites (239–242 Ma) should represent the age of formation of the NaOI. Therefore, we conclude that the NaOI and the NaOI phonolite formed during the Middle Triassic.

Zircon Lu–Hf isotope data

Some of the analytical spots for U–Pb dating of zircons from the NaOI phonolites were also used for in situ Hf isotope analysis (Fig. 6), and zircon Lu–Hf isotope data for the NaOI phonolites are given in Table 5. The initial 176Hf/177Hf ratios for primary zircons of the NaOI phonolites with ages of ~ 239 Ma vary from 0.282763 to 0.282893, yielding ε Hf(t) values and T C DM model ages (T DM1) of + 4.9 to + 9.5 and 522 to 716 Ma, respectively. The initial 176Hf/177Hf ratios for primary zircons of the NaOI phonolites with ages of ~ 242 Ma vary from 0.282773 to 0.282886, yielding ε Hf(t) values and T C DM model ages (T DM1) of + 5.4 to + 9.4 and 567–693 Ma, respectively. These values indicate that the NaOI phonolites were derived from the partial melting of juvenile crust or by fractional crystallization of the mantle-derived mafic parental magma.

Table 5 Zircon Hf isotopic composition of the rhyolite samples of the Zhaga Formation
Full size table

Discussion

Petrogenesis of the NaOI basalts and phonolites

Crustal contamination

Before discussing the magma source of the analyzed basalts and phonolites, it is necessary to consider whether they record any crustal contamination. Th and Ta are sensitive indicators of crustal contamination, as contamination results in a marked increase in Th/Ta ratios (Condie 1993). All the NaOI basalts and phonolites have relatively low Th/Ta ratios (1.11–2.54, and 0.93–1.53, respectively), similar to those in volcanic rocks derived from primitive mantle (Th/Ta = 2.3; Condie 1993) and much lower than those of the upper crust (Th/Ta > 10), indicating that none of these phonolites or basalts record crustal contamination. In addition, the NaOI basalts and phonolites have high contents of Nb (23.9–48.0 and 172–256 ppm, respectively) and Ta (1.35–2.96 and 11.8–16.0 ppm, respectively). Furthermore, these rocks do not show Pb enrichment in primitive mantle-normalized trace element patterns (Fig. 4b), further excluding the possibility that they underwent significant crustal contamination, because continental crust (CC) is characterized by depletions in Nb and Ta, and significant enrichment in Pb (Rudnick and Gao 2003; Niu 2009). If the NaOI basalts and phonolites had undergone crustal contamination, a CC signature would have been apparent in the elemental data, but no such signature exists.

Magma source

Various studies have shown that basaltic magmas are commonly derived from the partial melting of mantle peridotite, and that their REE patterns are controlled mainly by the contents of garnet and spinel in their magma source rather than by the contents of olivine, clinopyroxene, or orthopyroxene, or by pressure and temperature (McKenzie and O’Nions 1991; Beattie 1994; Horn et al. 1994; Schwandt and McKay 1998). In general, basalts derived from spinel lherzolite have flat chondrite-normalized REE patterns with inconspicuous fractionation between the LREEs and the heavy REEs (HREEs). However, basalts derived from garnet lherzolite display obvious fractionation between the LREEs and HREEs, and high LaN/YbN and CeN/YbN ratios (Hart and Dunn 1993; Hauri et al. 1994). In addition, partial melting of a spinel lherzolite mantle source does not change the Sm/Yb ratio because both Sm and Yb have similar partition coefficients, whereas it may decrease La/Sm ratios and Sm contents of the melts (Aldanmaz et al. 2000). Therefore, partial melts of a spinel lherzolite source should define melting trends sub-parallel to and nearly coincident with, a mantle array defined by depleted to enriched source compositions (Fig. 7a). On the other hand, garnet has a high partition coefficient for Yb (D garnet/melt = 6.6) relative to Sm (D garnet/melt = 0.25) (Johnson 1994), so that partial melting of garnet lherzolite mantle with residual garnet will produce a more steeply sloping trend on a Sm/Yb vs. La/Sm diagram than will melting of spinel lherzolite (Fig. 7a).

Fig. 7
figure 7

a Sm/Yb vs. La/Sm diagram for the NaOI basalts (modified after Zhao and Zhou 2007). The mantle array (heavy line) is defined by depleted MORB mantle (DM; McKenzie and O’Nions 1991) and primitive mantle (PM; Sun and McDonough 1989). Melting curves for spinel lherzolite and garnet peridotite with both DM and PM compositions are after Aldanmaz et al. (2000). Numbers along lines represent the degrees of partial melting. b–f Variation diagrams for the NaOI basalt samples

Full size image

The NaOI basalts have LREE-enriched chondrite-normalized REE patterns (LaN/YbN = 8.80–10.4; Fig. 4), similar to the patterns of basalts derived from garnet peridotite. That garnet peridotite in the mantle was the magma source of the NaOI basalts, is further supported by (1) the high Dy/Yb (2.96–3.06), Sm/Yb (3.87–4.00), and La/Sm (3.07–3.41) ratios of these basalts; and (2) the fact that the NaOI basalts plot in an area on the Sm/Yb vs. La/Sm diagram that corresponds to 5–10% partial melting of garnet peridotite (Fig. 7a; Sun and McDonough 1989; McKenzie and O’Nions 1991; Aldanmaz et al. 2000; Zhao and Zhou 2007).

Phonolites are volcanic rock types characteristic for intraplate settings at oceanic and continental environments (Ackerman et al. 2015), and two possible hypotheses have been proposed for their genesis: (1) fractional crystallization of mafic parental melts and (2) combined assimilation–fractional crystallization process involving a substantial crustal contamination during the ascent (Barberi et al. 1975; Baker and McBirney 1985; Vieten et al. 1988; Macciotta et al. 1990; Enders et al. 1992; Jung et al. 2013; Ackerman et al. 2015).

In the case of the NaOI phonolites, they have not underwent any crustal contamination, as discussed in Sect. 6.1.1, and this exclude the possibility that they formed at continental environments by the second genesis hypotheses. The inference is further supported by the small and homogeneously positive ε Hf(t) values of the NaOI phonolites (+ 4.9 to + 9.5). If the NaOI phonolites formed at continental environments by the second genesis hypotheses that combined assimilation–fractional crystallization process involving a substantial crustal contamination during the ascent, the ε Hf(t) values would be various and mostly negative. We infer, therefore, that the NaOI phonolites were formed by fractional crystallization of a mafic parental magma. The very low contents of MgO (0.11–0.25 wt%), the very low Mg# values (5–10), and the near-zero contents of Cr (1.27–7.59 ppm), Ni (0.43–7.19 ppm), and Co (0.11–0.38 ppm) indicate that the phonolites underwent substantial fractional crystallization, which in turn is consistent with the inference that they were formed by fractional crystallization of a mafic parental magma. In addition, the NaOI phonolites are enriched in LREE and high-field strength elements (Nb, Ta, Zr, and Hf), and they exhibit chondrite-normalized REE patterns and primitive mantle-normalized trace element patterns that are similar to those of OIB-derived magmas (Fig. 4a, b). We infer, therefore, that the NaOI phonolites were formed by fractional crystallization of an OIB-derived mafic parental magma. However, the absence of a negative correlation on some Harker diagrams (e.g., Al2O3–SiO2 and Zr–SiO2 diagrams; Fig. 5e, f) within the NaOI basalts and phonolites indicate that the NaOI phonolites could not have been produced by fractional crystallization of the basalts analyzed in this paper, and that there are some another associated OIB-type mafic parental magma existed within the primitive Nare Ocean Island.

Fractional crystallization

The Cr and Ni concentrations in the NaOI basalts, as well as their Mg# values (Cr = 178–240 ppm, Ni = 74.4–111 ppm, Mg# = 53–61), are generally lower than the values for primary-mantle-derived magmas (Cr = 300–500 ppm, Ni = 300–400 ppm, Mg# = 68–76; Frey et al. 1978; Hess 1992). This indicates that the NaOI basaltic magmas underwent fractional crystallization of olivine, chromite, and pyroxene.

Fractional crystallization of the NaOI basaltic magma is also evident in variation diagrams of Fe2O3–MgO, TiO2–MgO, P2O5–MgO, Cr–MgO, and Ni–MgO (Fig. 7b–f). The existence of negative correlations between Fe2O3 and MgO (Fig. 7b) and TiO2 and MgO (Fig. 7c) indicates the fractionation of Fe–Ti oxides (e.g., ilmenite and rutile), and the existence of a negative correlation between P2O5 and MgO (Fig. 7d) indicates the fractionation of apatite. The Cr contents of the NaOI basalts decrease with decreasing MgO content (Fig. 7e), indicating fractional crystallization of olivine. The Ni contents of the NaOI basalts also decrease with decreasing MgO content (Fig. 7f), indicating fractional crystallization of pyroxene. The absence of negative Eu anomalies in the chondrite-normalized REE patterns (Fig. 4a) indicates the limited fractionation of plagioclase.

As stated above, the NaOI phonolites underwent substantial fractional crystallization. The negative Eu anomalies (Fig. 4a; Eu/Eu* = 0.41–0.43) in the chondrite-normalized REE patterns indicate the fractionation of plagioclase. The remarkable negative anomalies in P and Ti in the primitive mantle-normalized trace element patterns (Fig. 4b) resulted from the extraction of P- and Ti-rich phases such as apatite, Fe–Ti oxides, and other Ti-bearing minerals. The negative Ba and K anomalies in the primitive mantle-normalized trace element patterns (Fig. 4b) indicate the fractionation of feldspar, whereas the negative U anomalies (Fig. 4b) were probably generated by fractional crystallization involving fractionated mineral assemblages with high uranium partition coefficients, such as titanomagnetite, titanite, and apatite.

In summary, the NaOI basalts were derived from the partial melting of garnet peridotite in the mantle. The ascending magmas underwent varying degrees of fractional crystallization but were not contaminated by the crust. The NaOI phonolites were formed by fractional crystallization of the OIB-type mafic parental magma. Similar to the NaOI basalts, the ascending magmas of the NaOI phonolites were not contaminated by the crust.

Tectonic setting: implications for the Middle Triassic evolution of the Bangong–Nujiang Tethyan Ocean

The NaOI retains an ocean island-type double-layered structure with a basaltic basement and an oceanic sedimentary cover sequence (conglomerate and limestone, the latter accompanied by layers of erupted phonolite near the top of the sequence), similar to the ocean island in the modern ocean (e.g., the Brava Ocean Island in the Atlantic Ocean; Mourão et al. 2010). Geochemical and Hf isotopic analyses of the NaOI basalts and phonolites show that the NaOI basalts are OIB-type volcanic rocks, and that the NaOI phonolites were formed by fractional crystallization of the OIB-type mafic parental magma. Neither the OIB-type nor the OIB-derived magma underwent crustal contamination during ascent. All the petrological and geochemical features described in this paper demonstrate that the NaOI was a typical ocean island that formed in the BNTO.

Phonolite is a relatively rare rock type in nature, and it makes up only one thousandth of the volume of extrusive volcanic rocks. Phonolite is also rare in modern ocean islands, as it was indicated by that although there are tens of thousands of modern ocean islands, very few have been reported to contain phonolite. For this study, we undertook a survey of modern ocean islands to record those that contain phonolite. Excluding ocean islands that have a controversial genesis (e.g., the Kerguelen archipelago in the Indian Ocean; Kinman et al. 2009; Rao et al. 2014; Frey et al. 2015; Olierook et al. 2016; Watson et al. 2016), we found that the ocean islands containing phonolite are all located in mature ocean basins (i.e., the Pacific and Atlantic oceans) far from a mid-ocean ridge (Giresse and Wiewióra 1999; Marques et al. 1999; Thirlwall et al. 2000; Geldmacher et al. 2001; Rodriguez-Losada and Martinez-Frias 2004; Klügel et al. 2005; Santos and Marques 2007; Reagan et al. 2008; Chan et al. 2009; Takamasa et al. 2009; Mourão et al. 2010; Dyhr and Holm 2010; Ancochea et al. 2012; Kahn et al. 2013; Perlingeiro et al. 2013; Guillou et al. 2014; Bongiolo et al. 2015; Garcia et al. 2016; Weit et al. 2016). Moreover, all these islands formed on a thick oceanic lithosphere (Calmant and Cazenave 1986; Carracedo et al. 1998, 2002; Caplan-Auerbach et al. 2000; Burke 2001; Ingle et al. 2003; Clouard and Bonneville 2004a, b; Humphreys and Niu 2009). For example, the Canary Islands (Andújar et al. 2008; Humphreys and Niu 2009) at the eastern margin of the Atlantic Ocean, the Fernando de Noronha archipelago (Burke 2001; Perlingeiro et al. 2013) at the western margin of the Atlantic Ocean, the Trinidade Islands (Caplan-Auerbach et al. 2000; Gripp and Gordon 2002; Bongiolo et al. 2015) in the South Atlantic Ocean, and the Hawaiian archipelago in the Pacific Ocean formed on lithosphere with a thickness of 90 km, and the lithosphere beneath Tristan da Cunha Island in the Atlantic Ocean is 56 km thick (Humphreys and Niu 2009; Weit et al. 2016), that under Gough Island in the Atlantic Ocean and the Society Islands in the Pacific Ocean 77 km thick (Calmant and Cazenave 1986; Caplan-Auerbach et al. 2000; Clouard and Bonneville 2004a, b; Humphreys and Niu 2009; Kahn et al. 2013), and that under the Austral–Cook archipelago in the Pacific Ocean 46–55 km thick (Clouard and Bonneville 2004b; Takamasa et al. 2009). The Middle Triassic NaOI contains significant amounts of phonolite, so the question is posed: did the NaOI form in a mature ocean on thick oceanic lithosphere?

We believe the answer to this question is yes, because the geochemistry of the NaOI phonolites is consistent with such a proposition. As stated above, the NaOI phonolites have high contents of SiO2 (60.3–65.3 wt%), very low contents of MgO (0.11–0.25 wt%), very low Mg# values (5–10), and near-zero contents of Cr (1.27–7.59 ppm), Ni (0.43–7.19 ppm), and Co (0.11–0.38 ppm), and in the chondrite-normalized REE variation diagram and the primitive mantle-normalized trace element variation diagram (Fig. 4a, b) they exhibit a series of very low negative anomalies (e.g., Eu, Ba, U, K, Sr, P, and Ti), indicating that their parental magma underwent substantial fractional crystallization of plagioclase, apatite, Fe–Ti oxides, and feldspar during ascent. Such a substantial fractional crystallization of the parental NaOI phonolite magma would have been facilitated by a lengthy ascent of the magma together with a long cooling process. In other words, fractional crystallization would have been greatly enhanced if the magma was emplaced into thick oceanic lithosphere.

Based on our survey of modern ocean islands and our geochemical data for the NaOI phonolites, we conclude, therefore, that the Middle Triassic NaOI was formed in a mature ocean basin with a thick underlying oceanic lithosphere.

According to the theory of plate tectonics, oceanic lithosphere initially forms at a mid-ocean ridge and becomes gradually cooler and thicker as it moves away from the ridge (Oxburgh and Parmentier 1977). The thickness of the oceanic lithosphere can be determined from the age of the lithosphere using the half-space lithosphere cooling model (T = 11 × t 1/2 where T is lithosphere thickness in km, and t is the age in Ma). The model is reliable for lithosphere younger than ~ 70 Ma (Parsons and Sclater 1977; Phipps Morgan and Smith 1992; Stein and Stein 1992). Because oceanic lithosphere reaches its full thickness at an age of ~ 70 Myr, we assume a constant thickness of ~ 90 km (i.e., 11 × 701/2 = 92 km) for older lithosphere (Humphreys and Niu 2009). The Middle Triassic NaOI formed on a thick oceanic lithosphere, indicating that at least during the Middle Triassic the BNTO has had a thick oceanic lithosphere. According to the half-space model of lithosphere cooling, the thick oceanic lithosphere of the BNTO during the Middle Triassic indicates that the BNTO had already been open, for a long time before the Middle Triassic. This inference is further supported by the following lines of evidence. (1) Early to Middle Triassic radiolarian cherts and Late Permian–Early Triassic eclogites have been documented in the BNSZ (Pan et al. 2006; Zhang et al. 2016); (2) Detailed geochemical studies for the basalts of the Late Triassic Gufeng ocean island fragment in the western segment of the BNSZ has showed that the BNTO has opened at least during the late Permian (Fan et al. 2017).

In summary, we conclude that the BNTO opened long before the Middle Triassic and that this ocean had developed into a mature ocean with a thick oceanic lithosphere by at least the Middle Triassic.

Conclusions

  1. 1.

    The NaOI basalts were derived from the partial melting of garnet peridotite in the mantle, and the ascending magmas underwent varying degrees of fractional crystallization but were not contaminated by the crust.

  2. 2.

    The NaOI phonolites were formed by fractional crystallization of OIB-type parental magma. In common with the NaOI basalts, the ascending NaOI phonolite magmas were not contaminated by the crust.

  3. 3.

    The Middle Triassic NaOI is a typical ocean island that formed within a mature ocean basin with a thick oceanic lithosphere.

  4. 4.

    The Bangong–Nujiang Tethyan Ocean opened long before the Middle Triassic, and the ocean had developed into a mature ocean by at least the Middle Triassic.