Episodic continental extension in eastern Gondwana during the mid-late mesozoic: insights from geochronology and geochemistry of mafic rocks in the Tethyan Himalaya

ABSTRACT Mesozoic extension and rifting processes of the Gondwana continent are critical for understanding the opening and formation of the Indian Ocean. Here, we report petrological, geochemical, zircon U–Pb age, and Lu–Hf isotopic data of mafic dikes in the central Tethyan Himalaya to reveal the timing and mechanism of the eastern Gondwana rifting. These mafic rocks exhibit two groups in terms of TiO2 and MgO contents (Group I with TiO2 < 2.0 wt.%, whereas Group II with TiO2 > 2.0 wt.%), but (1) they have similar trace elements patterns and exhibit enriched Light Rare Earth Element (LREE) patterns without Eu anomalies and (2) they have a wide range of zircon εHf(t) values. Both the above geochemical characteristics are similar to those of the Oceanic Island Basalt (OIB). However, they also show typical features of continental crust input. Compared to Group II, Group I has higher MgO, Cr, and Ni abundances (more primitive) and more depleted Nb-Ta contents, which are similar to those of lower TiO2 Dala mafic rocks, suggesting that the continental crust signature was mainly inherited from the Greater India subcontinental lithospheric mantle (GI-SCLM). Therefore, the mafic rocks here were mainly derived from a hybrid source from both OIB-like enriched mantle and GI-SCLM. Furthermore, two groups of zircon U–Pb ages have been identified: the first group with a weight mean age of 139.9 ± 0.2 Ma, whereas the second group with a weight mean age of 163.2 ± 0.9 Ma. The Middle Jurassic zircons show similar characteristics to those of intermediate to acidic igneous rocks in the continental setting, and thus they might record a stage of magmatism associated with a tectonic extension event in the Indian passive continental margin. The Early Cretaceous ages represent the crystallization timing of the mafic dikes, which are coeval with most mafic rocks in the Tethyan Himalaya. Based on these observations and the literature data in the Tethyan Himalaya, we proposed that the Kerguelen plume was incubating underneath the Tethyan Himalaya at ca. 140 Ma, but it was not located in the triple junction among the Antarctic, Australian and Indian plates. Therefore, the Kerguelen plume might play a synergistic role in the break-up of eastern Gondwana, which had experienced at least two stages of tectonic extension during the Mid-Late Mesozoic.


Introduction
Continental extension and rifting processes play a major role in the continental rupture and break-up and thus are critical for understanding the opening of oceans in the early stage of the Wilson cycle (e.g. Ziegler and Cloetingh 2004;Frizon de Lamotte et al. 2015). These processes are commonly multi-stage and are often associated with magmatic, tectonic, and sedimentary events (Tollo and Aleinikoff 1996;Sciunnach and Garzanti 2012;Mata et al. 2015). As one of the youngest supercontinents, the Gondwana had undergone episodic extension and rifting events, which resulted in the opening of Neo-Tethyan, Atlantic, and Indian oceans, respectively (e.g. Frizon de Lamotte et al. 2015). Of which, the break-up of eastern Gondwana mainly composed of Antarctic, Australian, and Indian plates (Heine and Müller 2005) directly led to the birth of Indian Ocean. As the passive continental margin of the Indian plate, the Tethyan Himalaya is an excellent place to reveal the extension and rifting processes during the eastern Gondwana break-up (Figure 1).
At least five phases of magmatic events have been identified in the continental margin of the Tethyan Himalaya (Figure 1(c-d)), which have been recorded by the following different rocks: (1) The Early-Middle 2018; Zhou et al. 2018;Bian et al. 2019;Tian et al. 2019;Zeng et al. 2019); (4) The Middle Cretaceous bimodal intrusive rocks (120-115 Ma) Chen et al. 2018); and (5) The Eocene OIBtype intrusive rocks (~45 Ma) (Ji et al. 2016). Most of the above magmatic events are consistent with sedimentary records: (1) The Early-Middle Permian and the Early Cretaceous magmatic activities were coeval with corresponding stages of uplift recorded by the backstripped subsidence curves in the Tethyan Himalaya (Sciunnach and Garzanti 2012), implying the occurrence of the large igneous province (LIP) which resulted in the continental bulge; (2) The Middle-Late Triassic and the Middle Cretaceous magmatic events were consistent with rapid subsidence stages associated with the tectonic extension in the passive continental margin (Sciunnach and Garzanti 2012). Therefore, the magmatic and sedimentary  (Zhu et al. 2008;Zeng et al. 2012Zeng et al. , 2019Wang et al. 2016Wang et al. , 2018Chen et al. 2018;Huang et al. 2018;Zhou et al. 2018;Tian et al. 2019). (e) Geological map of the central Tethyan Himalaya showing the locations of samples (Modified from Dai et al. 2013). Abbreviations: LIP, Large Igneous Province; T 3 , Upper Triassic; J 2 , Middle Jurassic; K 1 , Lower Cretaceous. records can coordinately reflect the multi-stage extension and rifting events of the Tethyan Himalaya. However, the subsidence curves (Patriat et al. 1982;Boote and Kirk 1989;von Rad et al. 1992) suggested the Middle Jurassic abrupt and widely drowning of the Gondwana continent from the Australia to the Arabia, which indicated the occurrence of large-scale extension, but the coeval magmatism in the Tethyan Himalaya has rarely been reported. In addition, the mechanism of eastern Gondwana break-up during the Cretaceous has been controversial. Some authors proposed that the breakup of eastern Gondwana was a passive process controlled by a lithospheric extension (e.g. Forsyth and Uyeda 1975;Jurdy and Stefanick 1991;Bercovici and Ricard 2014;Olierook et al. 2016;Zeng et al. 2019), while others proposed that the mantle plume played a major role in eastern Gondwana rifting process (e.g. Morgan 1972;Ernst and Buchan 2003;Frisch et al. 2011;Zhu et al. 2008Zhu et al. , 2009Zhou et al. 2018;Tian et al. 2019).
Basically, most previous studies related to the Early Cretaceous rifting are focused on the eastern Tethyan Himalaya near Comei (Figure 1(c); Zhu et al. 2008Zhu et al. , 2009Zhou et al. 2018;Bian et al. 2019;Wang et al. 2018;Tian et al. 2019). Recently, two studies have reported on the magmatic rocks of this stage in the western and central Tethyan Himalaya. Wei et al. (2017) first investigated the geochronological and geochemical data of Zhongba diabase dikes in the western Tethyan Himalaya ( Figure  1(c)), and they proposed that these diabase dikes were related to the Kerguelen LIP. Similarly, Zeng et al. (2019) reported Charong diabase dikes in the central Tethyan Himalaya (Figure 1(c, e)), and they proposed that the Charong diabases were derived from an asthenospheric mantle that had been inherently enriched by recycled oceanic and upper continental crust. However, there are more mafic outcrops in the central Tethyan Himalaya and they only focused on one single igneous rock body (Figure 1(e); Zeng et al. 2019), which might lead to an incomprehensive understanding of the origin of these mafic rocks.
In order to address the mechanism of eastern Gondwana break-up during the Mid-Late Mesozoic, we collected samples from the mafic dikes in the central Tethyan Himalaya (Figure 1(e)). The detailed zircon U-Pb dating results reveal the occurrence of the Middle Jurassic captured zircons and the Early Cretaceous original zircons, both of which might record the extension events of the Tethyan Himalaya. The whole-rock major and trace element, zircon Hf isotope and clinopyroxene major element suggest that the mafic dikes were derived from an OIB-like mantle source with crust contamination. Therefore, our data provide new insights into the rifting processes in the Gondwana continental margin.

Geological setting and sample description
The Tibetan Plateau is composed of at least four major terranes including the Songpan-Ganzi, Qiangtang, Lhasa, and Himalaya. These terranes were separated by four major suture zones such as A'nemaqin-Kunlun Suture Zone (AKSZ), Jinshajiang Suture Zone (JSSZ), Bangong-Nujiang Suture Zone (BNSZ), and Yarlung Zangbo Suture Zone (YZSZ) (Figure 1(b); Yin and Harrison 2000). The YZSZ is an east-west trending belt dominated by ophiolitic fragments with a total length of more than 2,000 km. Generally, the YZSZ represents remnants of the Early Cretaceous Neo-Tethyan oceanic lithosphere (e.g. Hébert et al. 2012;Dai et al. 2013). However, there is a strong controversy about the tectonic setting of the Yarlung Zangbo ophiolites ranging from a forearc to a backarc, even to a slow-spreading mid-oceanic ridge (please see Dai et al. 2021 and reference therein). The Himalayas consist of three major lithological units, which are separated by two major north-dipping faults: the Tethyan Himalayan Sequence (THS, generally referred to as Tethyan Himalaya in the text) is juxtaposed over the Greater Himalayan Crystalline (GHC) through the South Tibet Detachment System (STDS), while the GHC is located above the Lesser Himalayan Sequence (LHS) by the Main Central Thrust (MCT; Figure 1(c)). The northern boundary of the Tethyan Himalaya is the YZSZ.
The Tethyan Himalaya was the passive continental margin of the Indian plate in eastern Gondwana (Yu and Wang 1990). The basement of the Tethyan Himalaya is represented by Meso-to Neo-Proterozoic amphibolite-facies metasedimentary rocks (Zhu et al. 2013). The Tethyan Himalaya is mainly composed of marine strata from Cambrian to Eocene (Yin 2006). The Cambrian-Devonian strata might be deposited in a cratonic setting, while Carboniferous-Triassic deposition might be in a continental rifting, and the Middle Jurassic-Early Cretaceous sediments had been strongly influenced by extensional tectonic processes (Brookfield 1993;Liu and Einsele 1994;Garzanti 1999). In the western Tethyan Himalaya, the Early-Middle Permian Panjal Traps are cropped out (Shellnutt 2018). In the eastern Tethyan Himalaya, mafic rocks are exposed, mainly consisting of basaltic lavas, diabase sills and dikes, gabbro intrusions, and a small number of ultramafic rocks (Zhu et al. 2008(Zhu et al. , 2009(Zhu et al. , 2013Liu et al. 2015;Wang et al. 2016Wang et al. , 2018Zhou et al. 2018;Bian et al. 2019;Tian et al. 2019). The intrusive rocks intruded into the Triassic, Jurassic, and Lower Cretaceous strata (Figure 1(e)). However, limited Cretaceous magmatic rocks from the western (Wei et al. 2017) and central Tethyan Himalaya (Chen et al. 2018;Zeng et al. 2019) have been reported.
The study area is located in the central Tethyan Himalaya adjacent to the Xigaze ophiolite in the central YZSZ. The mafic dikes including both diabase and gabbro are distributed in the east-west to northwest directions. The thicknesses of mafic dikes range from tens to hundreds of metres ( Figure 2). The mafic dikes are mainly intruded into the Triassic strata with limited thin-bedded limestone and mainly sandstone (Figure 2(a-g)), termed as the Nieru Formation (Figure 1(e)). Cai etal. (2016) demonstrated that Nieru Formation has an Indian affinity. Six mafic samples were collected from three different sites. Samples DJ2, DJ4, DJ6, and DJ8 at the Naer area and sample DJ9 at the Charong area are exposed in the Upper Triassic strata, while sample DJ10 at the Liuqu area is cropped out in the Jurassic-Cretaceous strata (Figure 1(e)). To minimize the effects of surface weathering on subsequent geochemical analysis, ~1-2-cm-thick surface material was removed before sampling.
The massive gabbros and diabases are in intrusive contact with sandstones and limestones. The gabbros and diabases display massive structures. The gabbros are mainly composed of plagioclase, clinopyroxene, plus minor ilmenite, titanite, amphibole, zircon, and barite. The clinopyroxene crystals are euhedral to a subhedral columnar structure. The plagioclase minerals are subhedral to anhedral columnar (Sample DJ9 in Figure  2(h)). The plagioclase is mainly anorthite. Parts of clinopyroxene and anorthite have been altered to uralite and chlorite, respectively (Samples DJ4 and DJ8 in Figure 2(d, f)). The gabbros showing gabbroic texture ( Figure 2). The Fe-Ti oxides of titanite and ilmenite have been found in the gabbros. The reaction rims of ilmenite have been observed (Samples DJ4 and DJ8 in Figure 2(g,h)). The diabases are greyish-green and mainly consist of plagioclase and clinopyroxene. The light brown clinopyroxene grains are subhedral to anhedral, showing columnar to granular crystal forms. The plagioclase crystals are euhedral to subhedral platy structure. Plagioclase crystals suffered relatively extensive saussuritization, and clinopyroxene has been locally altered to uralite (Sample DJ2 in Figure 2(b)).

Whole-rock major and trace element analysis
Because sample DJ6 has plenty of very fine calcite veins, it was excluded from the whole-rock geochemical analysis. The compositions of major elements were obtained by X-ray fluorescence (XRF) (Primus II, Rigaku, Japan) at Wuhan Sample Solution Analytical Technology Co., Ltd., Wuhan, China. The fresh samples were powdered to less than 200 meshes and placed in an oven at 105°C for drying for 12 hours before analyses. Then, ~1.0 g of dried samples were put into the ceramic crucible and heated in a muffle furnace at 1000°C for two hours. Until cooling to 400°C, the sample was weighted to calculate the loss on ignition (LOI). The 6.0 g cosolvent and 0.3 g oxidant were mixed with 0.6 g of sample and placed in a Pt crucible. The Pt crucible was heated at 1150°C for 14 minutes and then quenched with air for 1 minute to produce flat discs for the XRF analyses. The analytical uncertainties of the major elements are typically <5%. Analytical results are listed in Table S1.
The concentrations of trace elements were performed using an Agilent 7700e ICP-MS at the Wuhan Sample Solution Analytical Technology Co., Ltd., Wuhan, China. The samples were dissolved by HNO 3 + HF in a Teflon bomb and heated to 190°C in an oven for >24 hours. After cooling, the Teflon bomb was heated to incipient dryness and then added HNO 3 to dryness again. HNO 3 + MQ water + internal standard solution of 1 ppm was added, and the Teflon bomb was resealed and placed in the oven at 190°C for >12 hours. The final solution was transferred to a polyethylene bottle and diluted to 100 g by the addition of 2% HNO 3 . Analytical uncertainties of most trace elements are estimated at lower than 10%. Analytical results are listed in Table S1.

Zircon U-Pb and trace element analysis
All six samples were used to separate zircons using the standard density and magnetic separation methods. Zircons U-Pb dating and trace element analysis were accomplished simultaneously by LA-ICP-MS at Wuhan Sample Solution Analytical Technology Co., Ltd., Wuhan, China. The detailed analytical methods of the ICP-MS instrument and data reduction follow Zong et al. (2017). Laser sampling was performed using a GeolasPro laser ablation system, equipped with a 193 nm ArF excimer laser and a MicroLas optical system, using a 5 Hz laser cycle and a 32 μm beam diameter. Zircon U-Pb ages and trace element raw data were corrected offline using ICPMSDataCal 10.9 with zircon 91,500 and glass NIS610 as external standards (Liu et al. 2010). The U-Pb weighted mean ages and concordia diagrams were calculated and plotted using IsoplotR software (Vermeesch 2018). Compositions of zircons trace elements were calibrated by internal standardization and multiple reference materials (Blank, SRM 610, 91,500, and GJ-1). Zircons U-Pb ages and trace elements analytical results are listed in Table S2 and Table S3, respectively.

Zircon Lu-Hf isotope analysis
Hafnium isotopic compositions were determined with a Thermo Finnigan Neptune MC-ICP-MS system coupled to a New Wave UP193 nm laser ablation system at the Laboratory of Isotope Geology, Tianjin Institute of Geology and Mineral Resources, Tianjin, China. In situ zircon Hf isotopic analyses were conducted on the same zircon zones where U-Pb age determinations were made. A laser repetition rate of 11 Hz at 100 mJ was used for ablating zircons, and the laser beam diameter was 50 μm. Raw count rates for 177 Hf, 178 Hf, 179 Hf, 180 Hf, 172 Yb, 173 Yb, 175 Lu, 176 (Hf + Yb + Lu), and 182 W were measured, and the isobaric interference corrections for 176 Lu and 176 Yb on 176 Hf were determined precisely. In order to evaluate the accuracy and precision of the analytical data, standard zircon GJ-1 was analysed at the Laboratory of Isotope Geology, Tianjin Institute of Geology and Mineral Resources, Tianjin, China. During the period of analyses, the 176 Hf/ 177 Hf ratio of 0.281998 ± 0.000021 (2σ, n = 64) for zircon GJ-1 was obtained. This ratio is consistent with the recommended 176 Hf/ 177 Hf ratio of 0.282009 ± 0.000010 for GJ-1 (Zhang and Hu 2020). Helium was used as the carrier gas for the ablated aerosol. Details on the experimental processes and analytical precision were described in Geng et al. (2017). Analytical results are listed in Table S4.

Clinopyroxene electron microprobe analysis
The electron microprobe analyses for clinopyroxene grains in this study were performed using a JEOL JXA-8230 electron probe microanalyzer (EPMA) at the Key Laboratory of Mineralogy and Metallogeny, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences. The analyses were carried out at 15 kV accelerating voltage and 25 nA current with a 1 μm beam in focused mode. Analytical results are listed in Table S5.
Zircons from samples DJ4 and DJ8 can be classified into two groups according to their internal structures. For Group I, zircon grains display distinct broad oscillatory zoning ( Figure 5). Their grain sizes range from 60 to 180 μm in length and from 40 to 70 μm in width. They have various Th/U ratios from 1.67 to 3.33. They yield a weighted mean age of 137.8 ± 0.8 Ma (MSWD = 2.0, DJ4) and 136.6 ± 2.0 Ma (MSWD = 2.4, DJ8). For Group II, zircons present narrow oscillatory zoning with lengths of 60-130 μm and widths of 40-70 μm ( Figure 5). Grains are long-prismatic shapes, and they have a wide range of Th/U ratios from 0.59 to 1.43. The weighted mean age of 162.9 ± 1.1 Ma (MSWD = 3.4) and 162.4 ± 1.4 Ma (MSWD = 2.2) have been obtained from samples DJ4 and DJ8, respectively.
All zircons possess significant HREE enrichment relative to LREE with distinct negative Eu and positive Ce anomalies (Table S3). REE patterns of zircons are almost parallel with restricted ranges of high (Yb/ Gd) N ratios. Eu anomalies (Eu/Eu* = 0.1-0.3) are large to moderate negative, and Ce positive anomalies (Ce/Ce* = 4.8-96.6) are pronounced. Generally, zircons from Group I have higher total REE contents than those from Group II (Table S3). They are also characterized by high Zr and Hf concentrations, which are tens to a hundred thousand times of those in chondrites. Zr/Hf ratios of Group I zircons range from 42.2 to 62.0, whereas those of Group II zircons vary from 41.0 to 46.8, both of which are higher than the chondritic value (chondritic Zr/Hf ratios of 35-40, Weyer et al. 2002), but are comparable to igneous zircons of felsic igneous rocks (Claiborne et al. 2006). Y concentrations and U/Yb ratios of Group I zircons vary from 2076 to 9107 ppm, from 0.3 to 1.9, while those of Group II zircons range from 739 to 2555 ppm, from 0.5 to 1.2, respectively. Ti abundances of zircons from both groups range from 9 to 40 ppm, which are in conformity with the normal abundance of Ti in zircon (≤75 ppm; Hoskin and Schaltegger 2003).

Discussion
In this section, we will discuss the following two factors: (1) petrogenetic processes of the Early Cretaceous mafic rocks using whole-rock major and trace elements, and all the available published data are also compiled and (2) two stages of eastern Gondwana continent extension with mainly constraints from zircon U-Pb ages, compositions and associated whole-rock geochemical compositions.

Petrogenesis of the early Cretaceous mafic rocks in the central Tethyan Himalaya
The effects of metamorphism and alteration should be assessed before the discussion on the petrogenesis of the mafic rocks in the central Tethyan Himalaya. The primary structure of the minerals observed from the petrography did not change significantly. In addition, the low-grade metamorphism (e.g. metamorphic hornblende, epidote, chlorite, and albite) was not observed. The large ranges of LOI values varying from 2.9 to 8.1 wt. % suggest that they have undergone different degrees of hydrothermal alteration, which is consistent with the observations of outcrops and microscopes ( Figure 2). The LILEs and some major and trace elements such as Rb, Ba, Na, and K are generally mobile and sensitive to alteration. Therefore, immobile elements such as REEs, HFSEs, and transitional elements are mainly employed in the following discussion. This is reasonable because they exhibit the parallel trace elements patterns (Figure 4). The mafic rocks in this study show large ranges of MgO and Mg # , implying that fractional crystallization occurred during their formation. The somehow positive correlations between CaO, Cr, Ni, and MgO (Figure S1 (c-e)) suggest the fractionation crystallization of olivine and clinopyroxene, whereas the absence of Eu anomalies ruled out the notable plagioclase fractionation. Interestingly, the compilation of the same kinds of geochemical data of mafic rocks from other locations in the Tethyan Himalaya also shows the same correlations between them ( Figure S1(c-e)).
Both groups of the mafic rocks here display negative Nb-Ta anomalies, but samples from Group I (High-Mg/ low-Ti series) are more depleted in trace elements than those from Group II (Low-Mg/high-Ti series, Figure 4(c, d)). Generally, this characteristic of mafic rocks has been attributed to the input of typical arc rock or continental crustal rock during their formation. The arc rock input can be readily excluded because the mafic rocks were located on the passive continental margin during the Early Cretaceous on the basis of palaeogeographic reconstruction (Zhu et al. 2009). Alternatively, the continental crustal input is highly possible based on the above trace element characteristics. However, the Charong diabase dikes that are close to our samples were proposed to be rarely contaminated by continental crust (Zeng et al. 2019). The most robust evidence for the above conclusion is the lack of older zircons in the Charong mafic rocks (Zeng et al. 2019). As presented above, the Jurassic zircons were identified in these mafic rocks, clearly against the absence of crustal contamination.
The continental crustal input is strongly supported by high values of (Th/Nb) PM (the subscript of PM represents that elements have been normalized to the corresponding ones of the primitive mantle from Sun and McDonough 1989). In the (Th/Nb) PM vs. (La/Nb) PM diagram (Figure 7(a), samples from both groups of this study show large dispersion and are plotted between the lower and the middle crust, suggesting an input of continental crust in their generation. However, it remains unclear whether this continental crustal signature resulted from crustal assimilation during magma ascent or it was inherited from a deep source. This problem can be addressed by comparative geochemical characteristics between two groups. Sample DJ4 from Group I exhibits lower TiO 2 but higher MgO, Cr, and Ni abundances that are the characteristics of more primitive mafic magmas. However, this sample has more depleted Nb-Ta and REE compared with these Group II mafic rocks (Figure 4). Wang et al. (2018) proposed that the continental crust signature was mainly inherited from a deep source according to the two similar groups of mafic rocks in the Dala area (Figure 1(c)). Thus, a progressive continental crustal contamination during magma ascent and differentiation was limited. This enriched deep source was likely derived from the Greater India subcontinental lithospheric mantle (GI-SCLM) based on the high MgO contents of Group I for the mafic rocks in the Tethyan Himalaya ). The GI-SCLM might have been enriched by the subduction of continental crust beneath the GI and Western Australia during the amalgamation of the Gondwana (Olierook et al. 2016). How much GI-SCLM input does the source of the mafic rocks in the central Tethyan Himalaya have? It seems that the part of the GI-SCLM might be small on the basis of the following observation: the absence of metamorphic amphibole and phlogopite for our diabase samples, which should be present because the subcontinental lithosphere is normally enriched in volatiles and water (Grant et al. 2007). Our samples have enriched LREEs relative to HREEs (Figure 4) and display a wide range of zircon ε Hf (t) values of −5.3 to 23.4 (Figure 6), which is consistent with those of OIBs, indicating an OIBlike source. In addition, the compositions of clinopyroxenes in our samples also indicate that they were formed within-plate setting (Table S5; Figure 8). Moreover, in the Th/Hf vs. Ta/Hf plot, our samples were plotted into the region of within-plate and rift environments (Figure 7 (c)). Based on the regional geological setting and the geochemical similarities between our samples and other published samples in the Tethyan Himalaya (Zhu et al. 2008(Zhu et al. , 2009Chen et al. 2018;Wang et al. 2018;Zhou et al. 2018;Tian et al. 2019;Zeng et al. 2019), we propose that the mafic rocks in the central Tethyan Himalaya were mainly derived from Kerguelen plume melts. These melts were mixed with the GI-SCLM. Previous studies have estimated the proportion of the GI-SCLM input based on the isotopic ratios ).
According to the end members of the GI-SCLM melts (Srivastava et al. 2005) and the Kerguelen plume melts (Comei LIP from Zhu et al. 2008;Xia et al. 2014; Rajmahal trap from Kent et al. 1997), only a small proportion of the former melt input to the latter melts can yield the Sr-Nd isotopic compositions of most Dala diabase samples . Geochemical characteristics of our samples resemble those of Dala diabases, and therefore, we also proposed that our samples were mainly derived from the Kerguelen mantle plume with a small fraction of the enriched component from the GI-SCLM. In addition to the mantle plume-SCLM mixing end member origin for the Early Cretaceous mafic rocks in the Tethyan Himalaya, the mantle plume end member origin is also recognized (Zhu et al. 2008;Wang et al. 2018). Mafic rocks in the Comei and Cona areas present a typical OIB geochemical signature ( Figure 4) and were proposed to be derived from the Kerguelen plume (Zhu et al. 2008;Tian et al. 2019).

The Early Cretaceous and the Middle Jurassic continental extension events: insights into the Gondwana break-up
As discussed before, at least five stages of magmatic events have been identified in the Tethyan Himalaya. Our samples yielded weighted mean 206 Pb/ 238 U ages of 141.5 ± 0.5 Ma, 141.3 ± 0.5 Ma, 138.9 ± 0.4 Ma,140.4 ± 0.4 Ma, which are consistent with previously reported ages. The earlier studies revealed that their ages are mainly concentrated in 134-130 Ma (Zhu et al. 2008(Zhu et al. , 2009, while more and more studies demonstrated that mafic rocks with ages of 145-137 Ma are sporadically cropped out in a large area in the Tethyan Himalaya (Zhu et al. 2008;Wang et al. 2018;Zhou et al. 2018;Tian et al. 2019;Zeng et al. 2019). Magmatic rocks generated in this stage of the Early Cretaceous are the most widely distributed. They are mainly distributed in the eastern (Comei LIP) and the central of the Tethyan Himalaya (Dala and Charong areas) (Figure 1(c); Zhu et al. 2005;Zhou et al. 2018;Tian et al. 2019). The Comei LIP has an ellipse area of about 40,000 km 2 with lengths of 270 km and widths of 150 km (Zhu et al. 2009). Most mafic rocks in this stage exhibit OIB-like features, such as the absence of Nb-Ta depletion and more enrichment incompatible elements and LREEs and depletion in HREEs (Figure 4(b,d)). In addition, several studies have reported the occurrence of A-type granites, including A-type granites from Comei ), A1-type monzogranites from Cona (Tian et al. 2019), and felsic rocks of the Sangxiu Formation , suggesting that they were formed by partial melting of crust heated by the mantle plume. Another very interesting explanation is that large marine plateaus may be generated by massive linear volcanism at a mid-oceanic spreading ridge, rather than by plume-head eruptions as traditionally thought (e.g. Sager et al. 2019;Whittaker 2019). The three-dimensional finite element model results indicate that the geometry of oceanic ridgeridge-ridge triple junction (TJ) may influence mantle dynamics, which led to excess volcanism close to TJ (Georgen and Lin 2002;Georgen 2008). However, mafic rocks in the Tethyan Himalaya were located on the passive continental margin of Indian plate, far away from the mid-ocean ridge (MOR) of Neo-Tethyan or Indian ocean (Figure 10), which excludes the possibility of being influenced by the MOR. Except for the Comei LIP, the contemporaneous Bunbury basalts located in the western Australia have an area of 3300 km 2 (Olierook et al. 2015). Early studies reveal that the Bunbury Basalts show similar geochemical characteristics and close temporal distributions with the Kerguelen LIPs (Frey et al. 1996), and thus, the Bunbury basalts were associated with the Kerguelen plume (Ingle et al. 2004). However, a recent study indicated that the Bunbury Basalts have incompatible elements and REE patterns similar to average enriched MORBs, implying that they might be derived from the enriched shallow mantle source. Based on newly obtained age constraints, Olierook et al. (2016) inferred that the magmatic products of Bunbury Basalts were not coeval or located close enough to be fed by the plume head. Therefore, Olierook et al. (2016) proposed that the Kerguelen plume was not necessary for the origin of the Bunbury Basalt, whereas the decompression melting of an enriched subcontinental lithospheric mantle source induced by passive rifting is another reasonable model. In addition, Olierook et al. (2019) proposed that the position of eastern Gondwana break-up was probably proximal to the suture zone associated with the Kuunga Orogeny between Indo-Australia and Australia-Antarctica (ca. 550-500 Ma). Episodic extension and rifting events might result in mafic magmatic activities and the break-up of eastern Gondwana lithosphere during the Early Cretaceous (ca. 137-136 Ma, Gibbons et al. 2012;Olierook et al. 2016) without any effect from the Kerguelen plume (Olierook et al. 2019). Currently, however, it is difficult to evaluate the above two different models.
Most mafic rocks that occurred as dikes and lavas in the Tethyan Himalaya have OIB-like affinities (Figure 4), and only a small part of them might be influenced by the GI-SCLM. The palaeomagnetic and geochronological study shows that the original position for the volcanic rocks in the Zhela and Weimei formations of the Tethyan Himalaya was located in the centre of the Kerguelen plume (Bian et al. 2019). Such close spatial and temporal relationships, as well as geochemical connections, make these mafic rocks usually contribute to the Kerguelen plume (e.g. Zhu et al. 2008Zhu et al. , 2009) although the Bunbury Basalts were not related to the Kerguelen plume (Olierook et al. 2016). What is the role of the Kerguelen plume in the rifting of eastern Gondwana? (1) The break-up of eastern Gondwana is estimated to be 137-136 Ma (Gibbons et al. 2012); (2) The timing of the Kerguelen plume activity can be estimated from both mafic rocks here and the association of the bimodal magmatism, which should be around 140-137 Ma (Tian et al. 2019). It is very likely that the Kerguelen plume was incubating underneath the eastern Gondwana continent at least at ca. 140 Ma (Figure 10(b)). Although this timing was earlier than that of the break-up of eastern Gondwana, the Kerguelen plume might play a synergistic role because its location was not at the central of the triple junction among the Antarctic, Australian and Indian plates (Figure 10(b)). Other factors, such as the far-field plate tectonic force, should have synergistically resulted in the break-up of eastern Gondwana during the Early Cretaceous (Buiter and Torsvik 2014). If it is correct, it predicts that other extensional events that had thinned the rifted lithosphere should occur before the Kerguelen plume activity, which would synergistically lead to final continental break-up.
The most relevant extension event is the Middle Jurassic large-scale extension, which was recorded by the subsidence curves (von Rad et al. 1992). However, the coeval magmatic records of this extension event are limited. Currently, the reported Middle Jurassic magmatic activities in the Tethyan Himalaya include (1) captured zircons with ages of ~174 Ma in the Miocene leucogranite  and (2) basalts interbedded with the Middle-Late Jurassic marine clastic rocks (Zhu et al. 2004), but no robust age has been obtained for these basalts. In this study, we have identified the Middle Jurassic zircons with the weight mean ages of 162.9 ± 1.1 Ma and 162.4 ± 1.4 Ma. We tend to think that the Jurassic zircons were captured during the ascent of magma because the mafic rocks originated from a high-temperature mixed mantle source where it was difficult for zircons to be inherited. These Middle Jurassic zircons exhibit clear and narrow bands in the CL images ( Figure 5) have negative ε Hf (t) values varying from −17.7 to −13.1 (Figure 6), and show relatively lower REE concentrations compared with those of the Early Cretaceous zircons (Figure 7(d)). All these observations indicate that the Middle Jurassic zircons might originate from intermediate to felsic igneous rocks. Furthermore, trace elements of these zircons are useful to reveal their tectonic settings. In the diagrams of U/Yb vs. Y and U/Yb vs. Hf (Figure 9(a,b); Grimes et al. 2007), the Middle Jurassic zircons are plotted into the continental field. The above tectonic setting affinities imply that the original hosting magmatic rocks of the Middle Jurassic zircons might be associated with a tectonic extension event in the Indian passive continental margin. This Middle Jurassic tectonic extension of the Indian passive margin might have thinned the lithosphere and finally assisted the subsequent large-scale rifting of eastern Gondwana in the Early Cretaceous ( Figure 10).

Conclusion
Whole-rock major and trace elements and clinopyroxene major elements, zircon U-Pb ages and Hf isotopic compositions of the mafic rocks in the central Tethyan Himalaya, in combination with the literature data and regional geological setting, led us to reveal their source region and magmatic evolution and to constrain the extension and rifting processes of eastern Gondwana. The main conclusions of this study are as follows: (1) The mafic dikes in the central Tethyan Himalaya exhibit two groups in terms of TiO 2 and MgO contents. The more primitive mafic rocks have more depleted Nb-Ta contents, suggesting that the continental crust signature was mainly inherited from a deep source, that is, the Greater India subcontinental lithospheric mantle (GI-SCLM). In addition to Nb-Ta depletion, both groups display similar trace element patterns and zircon ε Hf (t) values (−5.3 to 23.4) to those of the OIB, indicating that they were derived from the Kerguelen mantle plume. The above observations indicate that they had a hybrid source. Previous modelling results indicated that a small fraction of the GI-SCLM melts input to the Kerguelen plume melts can produce the Sr-Nd isotopic compositions of Dala diabase samples in the Tethyan Himalaya . Thus, the mafic rocks here were mainly derived from the Kerguelen mantle plume with a small proportion of the enriched component from the GI-SCLM.
(2) The mafic dikes in the central Tethyan Himalaya were formed during 136.6-141.5 Ma, which are consistent with previous studies. Considering that most mafic rocks in the Tethyan Himalaya might share the same generation, we proposed that the Kerguelen plume was incubating underneath the eastern Gondwana continent at least at ca. 140 Ma. However, the Kerguelen plume might play a synergistic role because its location was not at the triple junction among the Antarctic, Australian and Indian plates (Figure 10(b)), although its timing was earlier than that of break-up.
(3) The Middle Jurassic captured zircons exhibit similar characteristics to those of intermediate to acidic igneous rocks, including the CL images, ε Hf (t) values, and REE concentrations. They also show continental hotspot and evolving rift tectonic setting affinities. These observations imply that the original hosting magmatic rocks of the Middle Jurassic zircons might be associated with a tectonic extension event in the Indian passive continental margin. This stage of tectonic expansion might have thinned the rifted lithosphere and thus is critical for the subsequent rifting of eastern Gondwana in the Early Cretaceous.