Abstract

The Spongtang ophiolite (Ladakh, NW India) constrains the nature of oceanic lithosphere before Indo-Asia collision and key stages in the development of the Himalayas. We report whole-rock 40Ar/39Ar and in situ zircon 238U–206Pb ages from its crustal and upper and lower mantle sequences. Major and trace elements from harzburgite minerals suggest that the ophiolite formed at a mid-ocean ridge-type spreading centre, whereas published spinel compositions from Spongtang dunites are consistent with a suprasubduction-zone setting. Rare earth element-in-two-pyroxene thermometry for the harzburgite yields 1058 ± 13°C whereas temperature from solvus-based two-pyroxene and olivine–spinel thermometry is lower (to 656°C). The distribution suggests that the mantle section of the ophiolite cooled at rates of 100° Ma−1 or slower. Based on ages, major and trace element geochemistry, and geospeedometric estimates, we model the origin of the Spongtang ophiolite as forming within a mid-ocean ridge-type spreading centre with a spreading rate >2 cm a−1 in the Neotethyan Ocean, possibly from the Late Triassic to Jurassic. By the Early Cretaceous, the ridge experienced increasing influence of subduction beneath the Spongtang oceanic lithosphere owing to a subduction polarity reversal. Based on 238U–206Pb ages of the youngest Cenozoic zircon grain, latest obduction occurred between 64.3 ± 0.8 and 42.4 ± 0.5 Ma, in accordance with 56.7 ± 5.2 Ma whole-rock 40Ar/39Ar ages.

Supplementary material: Excel files with details of electron microprobe and inductively coupled plasma mass spectrometry (ICP-MS) analyses, argon isotopic whole-rock and secondary ion mass spectrometry (SIMS) analyses, and the TREE calculations, including an inversion diagram showing regression through measured REE distributions in cpx and opx (from Liang et al. 2013), are available at https://doi.org/10.6084/m9.figshare.c.4261856

During the mid-Mesozoic to Eocene, the Indian subcontinent moved over 60° latitude north towards Asia, closing the ancient Neotethyan Ocean (e.g. Burg 2011). This ocean plays a central role in reconstructing a number of orogenic systems extending from Europe to Asia (e.g. Şengör & Atayman 2009) (Fig. 1). However, debated palaeotectonic issues remain because exposures of the former ocean and its sub-basins are affected by plate collision, arc and terrane accretion, and later deformation. Problems that plague Neotethyan reconstructions include complicated structural relationships between sutures and amalgamated crustal fragments, unknown subduction polarity of oceanic arcs, unknown origin or tectonic setting of the oceanic fragments present in the collision zone, and the potential existence and locations of smaller, marginal ocean basins within the Neotethys itself (e.g. Şengör et al. 1984; Ustaömer & Robertson 1997; Robertson & Collins 2002; Aitchison et al. 2007; Şengör & Atayman 2009). Exposed remnants of Neotethyan ophiolites are present across the Himalayas and have the potential to provide considerable insight into the timing and processes involved during Indo-Asia collision and the nature of Neotethyan ocean crust and upper mantle (e.g. Hébert et al. 2012; Hu et al. 2016).

The Spongtang ophiolite (Figs 1 and 2) was described as early as the late 19th century (Lydekker 1880, 1883; La Touche 1888; MacMohan 1901) and has been the focus of several detailed structural, geochemical and geochronological studies since the 1980s (Bassoullet et al. 1980; Andrews-Speed & Brookfield 1982; Reibel & Reuber 1982; Kelemen & Sonnenfeld 1983; Reuber 1986a, 107b; Reuber & Colchen 1986; Reuber et al. 1987; Niorthe & Reuber 1988; Searle et al. 1997; Corfield et al. 1999, 2001, 2005; Corfield & Searle 2000; Pedersen et al. 2001; Mahéo et al. 2004; Baxter et al. 2010). The ophiolite is one of several isolated basic–ultrabasic massifs exposed across the collision zone (Fig. 1) (Brookfield 1977; Reuber 1986a, 107b; Reuber et al. 1983, 1987, 2005; Hébert et al. 2012; Xu et al. 2015). A comparison of the Spongtang ophiolite complex with others along the Himalayan arc and elsewhere along the site of Neotethyan ocean closure is challenged by limited and problematic age data and alteration owing to the influence of multiple tectonic events (Ghose et al. 2014).

In this study, ophiolite samples collected by Ingrid Reuber in 1986 (Table 1; Fig. 2) were petrographically and geochemically analysed and dated. Ages were obtained from various units of the ophiolite section via in situ (in thin section) secondary ion mass spectrometry (SIMS) (U–Pb zircon) and whole-rock noble gas spectrometry 40Ar/39Ar geochronology to determine its crystallization and deformation history. We apply advances in the interpretation of pyroxene rare earth element (REE) distributions and conventional major-element based geothermometry to discern a cooling rate over high-temperature cooling intervals, an approach that may allow interpretations to be made regarding its geological evolution (Liang et al. 2013; Dygert & Liang 2015).

Geological background

The Spongtang ophiolite is a tectonic thrust slice that overlies Permian to Paleocene–Eocene sediments of the northern Indian passive margin (Andrews-Speed & Brookfield 1982; Ahmad et al. 2008; Green et al. 2008; van Hinsbergen et al. 2012; Clift et al. 2014; Gibbons et al. 2015) (Figs 1 and 2). It is characterized by a complete ophiolite stratigraphy, dominated by peridotites (Reuber et al. 1989; Corfield et al. 2001; Ghose et al. 2014), and is divided by the Photang valley into a western area with mantle rocks and an eastern region of exposed crustal assemblages. Northwest of the Photang valley, the lowermost Photang thrust sheet consists of distinct alkaline volcanic units, overlain by Permian to Cretaceous limestones, which most probably were seamounts in the Neotethyan Ocean (Corfield et al. 1999). Above are two ultrabasic thrust sheets of suboceanic mantle rocks whose contact Reuber (1986a) considered an intra-oceanic thrust. South of the Photang valley, a thin mélange with limestone exotic rocks underlies an upper ophiolitic thrust sheet with gabbros, sheared dykes and pillow lavas overlain by Late Cretaceous andesites interbedded with latest Valanginian to early to mid-Aptian radiolarian cherts (Corfield et al. 1999; Baxter et al. 2010). The Spongtang massif can thus be divided into two distinct suites of related affinities. The older suite is the Jurassic Spongtang ophiolite, comprising two (stratigraphically lower and upper) mantle units as well as crustal rocks, which was deformed in the early Cretaceous. The younger suite is the overlying Spong arc, a c. 500 m thick, volcano-sedimentary package that developed in an island-arc environment (Reuber et al. 1989; Corfield et al. 2001). The arc is similar in age to the Dras arc along the Yarlung Zangbo Suture Zone to the north (Fig. 1) (see review by Hébert et al. 2012).

The timing of formation and emplacement of the ophiolite and its oceanic island arc cover (Spong arc) and its eventual collision with northern India are controversial but essential for deciphering the history of plate convergence during the closing of the Neotethyan ocean (Rolland et al. 2002; Clift et al. 2014; Hu et al. 2016). Existing age constraints are based on direct dating of rocks associated with the ophiolite, as well as dating of overlying and underlying sediments (Corfield et al. 2001; Baxter et al. 2010; Clift et al. 2014). We summarize available radiometric age constraints in Figure 3. Ages from minerals extracted from the ophiolite sequences, in general, yield large uncertainties reflecting both analytical issues (i.e. excess argon in K–Ar age and complex 40Ar/39Ar gas release patterns) and the unit's crystallization and deformation history.

Reuber et al. (1989, 1990) reported K–Ar amphibole ages from some of the same rocks analysed here that range from 231 ± 23 to 124 ± 75 Ma. All have large uncertainty (the latter age is not plotted in Fig. 3) and are attributed to oceanic deformation and accretion, with some affected by excess argon. Sample SPO26, an amphibolite from the Spongtang metamorphic sole (Fig. 2), yields 135 ± 4 Ma, suggested to time intra-oceanic thrusting (Reuber et al. 1990). The most commonly reported crystallization age for the Spongtang ophiolite is 177 ± 1 Ma, based on three fractions of zircons separated from a diorite intruding high-level gabbros of the ophiolite (Pedersen et al. 2001) (see Fig. 2 for sample location). The timing of activity in the overlying Spong arc is based on a series of zircon ages extracted from andesite that average 88 ± 5 Ma, although zircons from other andesite samples are older (Pedersen et al. 2001) (Fig. 3). The Early Jurassic zircon crystallization age makes the Spongtang ophiolite the oldest known Neotethyan ophiolite (Ghose et al. 2014).

Metamorphic amphiboles from Spongtang diorite dykes intruding lower peridotites yield disturbed 40Ar/39Ar age spectra interpreted to record the initiation of subduction magmatism at c. 130 Ma (Mahéo et al. 2004). None yield plateau ages as defined by Ludwig (2008) and the results demonstrate the problems associated with obtaining reliable 40Ar/39Ar data from the amphibole separates from these rocks, as they are affected by excess argon. The results also suggest the potential presence of alteration minerals recorded by Ca/K as the grains degassed during the step-heating process. Our new argon geochronology data show similar problems and are discussed in detail in the results section below.

Regarding its tectonic affinity, the Spongtang ophiolite has been proposed to have formed at a back-arc basin (Reuber et al. 1983), a slow spreading centre near a transform fault (e.g. Reuber 1990) or a mid-ocean ridge-like setting with an normal mid-ocean ridge basalt (N-MORB) mantle source previously enriched in large ion lithophile elements (LILE) by a metasomatic event that occurred in a suprasubduction zone (SSZ) (Pedersen et al. 2001; Mahéo et al. 2004). Several models have been proposed regarding its emplacement history. Assuming a tectonic scenario similar to that for the Samail ophiolite in Oman (Brookfield 1977; Searle et al. 1988), the Spongtang ophiolite may have been initially emplaced in the Late Cretaceous onto the northern Indian passive margin and deformed in the Late Eocene during the final closure of the Neotethyan Ocean (Brookfield & Reynolds 1981; Corfield et al. 1999; model B of Clift et al. 2014). Alternatively, obduction may have been entirely a late Eocene event (Colchen et al. 1987; Mahéo et al. 2004; Fuchs 2005; model A of Clift et al. 2014). Ophiolitic detritus in underlying sedimentary units suggests that the Spongtang ophiolite was situated on or near the northern Indian margin by the Paleocene (Garzanti et al. 1987; Searle et al. 1990). The detritus is also found in the Paleocene deep-water sediments of the north Indian margin in southern Tibet (Ding et al. 2005). A significant presence of Mesozoic detrital zircons (129 ± 7 Ma) unrelated to the unit indicates that if obduction occurred on the Indian margin during the collision, the Spongtang ophiolite was submarine or un-eroded until later re-thrusting, or emplacement occurred after c. 61 Ma (Clift et al. 2014). This paper presents additional geochronological and geochemical data from selected units of the Spongtang ophiolite to evaluate proposed episodes of its tectonic history and place constraints on the timing and dynamics of its crystallization and deformation.

Materials and methods

Samples and petrography

This study focuses on the analysis of 11 rocks collected from the Spongtang ophiolite (Table 1; Fig. 2). All samples are deformed and altered diabase or diorite dykes, except a plagiogranite (SPO10), harzburgite (SPO25), amphibolite (SPO26) and two subvolcanic assemblages (SPO17 and SPO19). Although no metamorphic sole is purported to exist below the Spongtang ophiolite (Pedersen et al. 2001), we consider sample SPO26 to be a component of the sole based on its structural location, petrography and mineralogy. All rocks were examined initially with a petrographic microscope to document mineralogy and textures. As indicated in Table 1, selected samples were subjected to geochemical (mineral major and trace element) and geochronological (40Ar/39Ar whole-rock and U–Pb zircon SIMS) analysis.

Mineral chemistry of samples SPO25 and SPO26

Major elements in clinopyroxene (n = 3 grains, 19 spots), orthopyroxene (n = 3 grains, 31 spots), spinel (n = 2 grains, 19 spots) and olivine (n = 2 grains, 16 spots) in harzburgite SPO25 and amphibole in amphibolite sample SPO26 (n = 3 grains, n = 9 spots) were determined using the JEOL JXA-8200 electron microprobe at the University of Texas at Austin. Detailed results are presented in the supplementary material. Analyses were conducted during the same analytical session using standard conditions (a focused beam, 20 kV accelerating voltage, c. 20 nA beam current) and a 10-element procedure (SiO2, TiO2, Al2O3, Cr2O3, FeO, MnO, MgO, Na2O, K2O, CaO). Natural and synthetic standards were processed using a JEOL ZAF matrix correction procedure, and results were evaluated to ensure they fit mineral stoichiometry.

Li, Na, Mg, Al, V, Sr, Pb, Y, high field strength elements (HFSE) and REE were analysed in orthopyroxene (n = 10 spots) and clinopyroxene (n = 8 spots) in sample SPO25 and amphibole in SPO26 (n = 6 spots) by laser ablation inductively coupled plasma mass spectrometry (LA-ICP) using a New Wave UP193FX laser with a dual-volume sample chamber coupled to an Agilent 7500ce quadrupole ICP-MS system. We applied a laser fluence of c. 3 J cm−2 in one analytical session at a frequency of 10 Hz with 60 s ablations that penetrated through the samples completely. The process ensured the introduction of measurable quantities of REE into the ICP-MS system. Synthetic NIST glasses (610, 612 and 614) (Jochum et al. 2011) and a natural basalt (BCR-2) (Chen et al. 2000) were used as standards. Data were processed in LASY BOY (Sparks 2011) with Si as the internal standard. Spot sizes ranged from 40 to 120 µm, using the largest spots possible while avoiding cracks or alteration zones. Standards were analysed in triplicate before and after the unknowns using the same spot sizes as the unknowns. Detailed results are presented in the supplementary material.

Argon geochronology

Eight samples (SPO4, SPO6, SPO10, SPO14, SPO15, SPO17, SPO19 and SPO22) were subjected to 40Ar/39Ar whole-rock dating at the Indian Institute of Technology Laboratory at Mumbai. Portions of these rocks were crushed in a tungsten mill, and c. 200 mg powder (120–180 µm size fraction) was cleaned with 1% HCl and deionized water before packing in aluminium capsules for irradiation. The samples were irradiated with a flux monitor (Bern 4 Biotite, 17.3 ± 0.2 Ma, Flisch 1982; 17.21 Ma, Baksi et al. 1996) in the DHRUVA reactor at Bhabha Atomic Research Centre, Mumbai, India in one batch for c. 120 h. Pure CaF2 and K2SO4 salts were also irradiated with the samples to estimate the Ar produced by nuclear reactions on Ca and K isotopes. The following correction factors are used for correcting the reactor produced interfering Ar isotopes: (36Ar/37Ar)Ca = 0.002227, (39Ar/37Ar)Ca = 0.034342 and (40Ar/39Ar)K = 0.0381106. 40Ar/39Ar ages were calculated using the constants recommended by Steiger & Jäger (1977); use of Renne et al. (2010, 2011) constants would increase the age (in the age range of this study) by c. 0.9% (see discussions by Renne et al. 2011 and Schwarz et al. 2011), indistinguishable within their state uncertainties. The relative variation in the neutron fluence within the capsule was monitored by pure Ni wire that was also irradiated with the samples. Argon gas was extracted via step heating in a furnace, starting at 450°C with intervals from 25 to 50°C going up to a maximum of 1350°C, except for sample SPO19, which experienced a maximum of 1050°C. The number of steps varied from 14 to 20. The gas released at each step was cleaned using Ti–Zr getters and analysed for argon isotope ratios in the ARGUS VI multi-collector mass spectrometer (ThermoFisher) at the National Facility at Indian Institute of Technology (IIT), Mumbai, India. Detailed results are presented in the supplementary material.

U–Pb zircon geochronology

All samples were examined using a scanning electron microscope (SEM) for zircon grains for in situ SIMS analysis. Zircons were found only in samples SPO4, SPO6, SPO10, SPO20 and SPO26. The grains ranged from 10–15 μm to c. 100 μm in length. Their locations were documented, and portions of the thin sections containing the grains were cut from the samples using a low-speed saw. The thin section chips were mounted in epoxy with blocks of zircon age standard AS3 (1099.1 ± 0.5 Ma, Schmitz et al. 2003), coated in gold, and analysed using the UCLA CAMECA ims 1270-HR ion microprobe (sample SPO4 only) and CAMECA ims 1280-HR ion microprobe at Heidelberg University (all other samples). In both laboratories, a calibration curve was developed using U and Pb isotopic data sputtered from the age standards. For sample SPO4, 23 spots were placed on standard AS3, which created a calibration curve in UO+/U+ = 5.385 ±0.071 – 0.468 × (Pb/U, relative sensitivity factor), reproducing the standard age to 1093 ± 40 Ma. For all of the other samples, 16 spots were placed on standard AS3, which created a calibration curve in UO2+/U+ = 1.434 ± 0.014 – 0.285 × (Pb/U, relative sensitivity factor), which reproduced the standard age to 1100 ± 37 Ma. After dating, all zircons were imaged using backscattered electrons (BSE), secondary electrons (SE) and cathodoluminescence (CL) to determine zoning characteristics and location of the ion beam.

Results

Petrography

We show images of samples analysed in this study in Figures 46, which illustrate their mineralogy, alteration and deformation textures. All rocks are affected by varying degrees of alteration, expected as they experienced a complex crystallization and deformation history, but most preserve some degree of primary igneous mineralogy and textures. We compare our petrographic observations with those of the same samples by Reuber et al. (1983, 1990, 2005).

Samples from the upper crustal ophiolite sequence include a subvolcanic basalt (SPO17) and pillow lava (SPO19) (Fig. 4). The subvolcanic basalt assemblage (Fig. 4a and b) consists primarily of blocky plagioclase that is variably altered and aligned to define the rock's foliation. Clinopyroxene and opaque minerals fill the interstitial spaces between larger plagioclase grains. This sample has cross-cutting veins filled with similar mineralogy to the rock itself, but with smaller grain sizes. The pillow lava sample (SPO19) is extremely altered with only relict plagioclase laths in a serpentinite–chlorite–sericite matrix (Fig. 4c). The rock is cut by veins with calcite cores surrounded by quartz candles with c-axes that align towards the core. The edges of some veins spread diffusively into the altered rock matrix.

Petrographic images of samples from the upper mantle unit ophiolite sequences are shown in Figure 5. Samples SPO4 (K–Ar amphibole age of 139 ± 8 Ma) and SPO6 (K–Ar amphibole age of 169.0 ± 9.5 Ma; see Reuber et al. 1989, 1990) are from the pegmatite dykes described by Reuber et al. (1989) and consist primarily of large clinopyroxene grains with variable altered plagioclase grains. Zircon grains in SPO4 are rare and are found primarily on the edges of clinopyroxene. These dyke samples have the coarsest grain size and thus differ from samples SPO14 and SPO15, which have similar mineralogy but are composed of smaller grain sizes. Sample SPO15 is more altered than sample SPO14 and is dominated by sericite matrix with quartz and some opaque minerals. We highlight a quartz vein in sample SPO15, which has finer-grained and coarser-grained sections (Fig. 5d). Sample SPO10 is a plagiogranite dominated by quartz–plagioclase mineralogy with minor opaque minerals and sericite owing to the alteration of plagioclase. The majority of the rock has larger quartz grains with lobate grain boundaries and chequerboard textures, but anastomosing veins that cut the sample show much smaller grain sizes. SPO10 also contains minor amounts of frayed biotite that has largely altered to chlorite. These biotite grains are where the majority of zircon in the rock is present.

SPO20 and SPO22 are both deformed diabase dykes, with a clear foliation defined by clinopyroxene, sericite–plagioclase blades and quartz with grain boundary migration textures. Sample SPO20 has quartz veins parallel to the foliation that show undulatory extinction and grain-boundary migration textures. In sample SPO22, we find cross-cutting veins with asymmetrical mineral assemblages, with chlorite–sericite grains opposite coarser plagioclase and quartz. Reuber et al. (1989) suggested that the deformation textures observed in the deformed diabase dykes were produced during intra-oceanic thrusting.

Petrographic images of lower mantle unit samples SPO25 and SPO26 are shown in Figure 6. Sample SPO25 is a harzburgite and contains olivine (>75%) with lesser amounts of orthopyroxene, clinopyroxene and spinel. The rock is altered by a web-like array of serpentine and chlorite veins that fill grain boundaries and transect the rock. Amphibolite sample SPO26 (K–Ar amphibole, 135 ± 4 Ma, Reuber et al. 1989) originates from the Spongtang ophiolite's metamorphic sole. The rock is composed primarily of hornblende and epidote and is foliated. These rocks were the focus for further geochemical analyses described in the next section.

Mineral chemistry of samples SPO25 and SPO26

Figure 7 shows the locations and compositions across transects of olivine, spinel, orthopyroxene and clinopyroxene in sample SPO25 and amphibole in SPO26. Our analyses of olivine and clinopyroxene grains in sample SPO25 indicate lack of zoning (Table 2). Olivine forsterite contents (Fo, (Mg/(Mg + FeT)) × 100) average 90.2 ± 0.2 (n = 16), and clinopyroxene enstatite compositions (En, Mg/(Mg + Ca + FeT)) average 0.48 ± 0.01 (n = 19). We find that most SPO25 orthopyroxene En contents average 0.892 ± 0.002 (n = 26), with five analyses that are outside this value (from 0.877 to 0.860) where the electron beam excited clinopyroxene exsolution lamellae (Table 3). The compositions of these lamellae grains fit the stoichiometry of pyroxene and have oxide wt% totals of 99–100 wt% (Table 3). The Mg# (Mg/(Mg + FeT)) of spinel in sample SPO25 increases from core to rim (from 0.666 ± 0.001 to 0.687 ± 0.001), whereas spinel Cr# (Cr/(Cr + Al)) decreases (from 0.292 ± 0.003 to 0.248 ± 0.015) (Fig. 7, Table 3). BSE images of amphibole in sample SPO26 suggest that the mineral is zoned, and its brighter cores have slightly higher Fe contents and lower Ca contents compared with rims (Table 2).

In plots of olivine and spinel mineral chemistry, sample SPO25 falls directly within the compositional field defined by abyssal peridotites (Fig. 8a–c). According to methods presented by Hellebrand et al. (2001), the samples appear to have experienced 11.4 ± 0.6% melt extraction (Fig. 8d) assuming near-fractional melting of a ‘fertile MORB mantle’ source. Although data from this sample overlap those for harzburgites and lherzolites reported by Mahéo et al. (2004) for upper and lower levels in the mantle section of the Spongtang ophiolite, spinels in Spongtang dunites have higher Cr contents consistent with interaction with a melt sourced from an SSZ setting.

Trace element data from clinopyroxene, orthopyroxene and amphibole grains are presented in Figure 9. For pyroxenes in general, the light REE (LREE) abundances for the minerals are generally below the instrumental detection limit, but middle REE (MREE) and heavy REE (HREE) abundances could be reliably quantified. Clinopyroxene exhibits REE concentrations 2.5–14 times higher than those for orthopyroxene. Extreme REE depletion characteristic of some SSZ-type settings (e.g. Dygert & Liang 2015) is not observed. Clinopyroxene and orthopyroxene have chondrite-normalized abundances and LREE depletion consistent with abyssal peridotites, suggesting adiabatic decompression melting of a MORB-mantle source beneath a spreading centre (Fig. 10).

Thermometry

Using REE, Y and and major element data, a REE-in-two-pyroxene temperature (TREE) was calculated from mineral compositions averaged from rim and core analyses using the approach described by Liang et al. (2013) and Dygert & Liang (2015) (Table 4). For comparison, temperatures from the two-pyroxene solvus (Wells 1977; Brey & Köhler 1990) olivine–spinel (Fabriès 1979; Ballhaus et al. 1991) and Al-in-orthopyroxene (Witt-Eickschen & Seck 1991) thermometers were also calculated using the averaged mineral compositions for olivine, clinopyroxene, orthopyroxene, and core and rim compositions for spinel (Table 4).

Temperatures range from 656 to 1058°C depending on the thermometer. The olivine–spinel temperature decreases from core to rim by c. 15°C. The pyroxene solvus thermometers yield the lowest temperatures with average Mg orthopyroxene compositions similar to those calculated if the highest values are used. We did not apply the low Mg contents to the barometer, as these are probably exsolution lamellae. The results of the pyroxene solvus thermometry are lower than the results estimated using olivine–spinel thermometry (by c. 17–185°C, depending on calibration). Thermometers based on trivalent element partitioning yield the highest temperatures (983–1058°C).

Figure 10 shows a comparison of the Spongtang TREE and TBKN with temperatures from abyssal peridotites and peridotites from ophiolites with mid-ocean ridge and SSZ tectonic affinities. SPO25 has a TREE and TBKN lower than any abyssal peridotite; SPO25 temperatures are consistent with the coldest ophiolitic peridotites, many of which have been interpreted to have SSZ tectonic affinities (see Dygert & Liang 2015).

40Ar/39Ar whole-rock geochronology

All whole-rock analyses yield disturbed 40Ar/39Ar age spectra and show older ages as step-heating progress (Figs 1113). Groups of steps were linked based on trends identified in the step-heating ages and Ca/K ratios. In some cases steps belong to different groups, as the Ca/K data suggest that they may represent the degassing of potentially the same mineral. None of the samples have plateau ages as defined by Ludwig (2008). Inverse isochron plots show the influence of additional argon and 39Ar/40Ar values typically decrease with increasing step temperature. Sub-groups of these steps show linear correlations between 39Ar/40Ar v. 36Ar/40Ar, which were used to determine some inverse isochron ages (e.g. Korochantseva et al. 2007; Schwarz & Lippolt 2014). We attempted to identify trends in the inverse isochron results to decipher their relationship with plateau ages. However, we do not report results of isochrons that suggest a mixed age in Figures 1113. All data are available in the supplementary material.

Overall, we find the youngest inverse isochron ages in the initial degassing steps (1–4 to 1–10) that range from 1.27 ± 0.17 Ma (±2σ, sample SPO15; Fig. 12) to 3.8 ± 1.9 Ma (SPO6; Fig. 11). Corresponding average initial step ages from most samples are highly uncertain, but isolated dyke samples SPO14 and SPO15 yield 2.0 ± 0.8 Ma and 2.3 ± 0.6 Ma, respectively. The oldest of these initial step ages are Pliocene (5.1 ± 1.9 Ma, sample SPO17; 5.5 ± 3.6 Ma, SPO10). The results are probably related to the degassing of sericite or chlorite, minerals associated with the alteration found in varying amounts in each sample (Figs 46).

After these initial steps, all samples trend towards older ages. The most precise Paleocene age is found in steps 10–14 of sample SPO14 (58.7 ± 0.7 Ma), and more precise Paleocene to Late Cretaceous ages are found as in groups of steps, as opposed to inverse isochron ages. These results were generated in the final step-heating schedule and include 73.7 ± 2.2 Ma (sample SPO19, Fig. 13) and 101.7 ± 1.2 Ma (SPO15, Fig. 12). It should be noted that sample SPO17 had precise Cretaceous and Triassic ages for steps 11–14 (98.4 ± 2.3 Ma) and 14–16 (205 ± 3 Ma) (Fig. 13).

U–Pb zircon geochronology

Zircon grains were dated in samples SPO4, SPO6, SPO10 and SPO20 (Figs 14 and 15, Table 5). These grains are small (10–100 μm in length). In sample SPO4, zircons are found at the edge of clinopyroxene grains, whereas in sample SPO10, the grains are found in contact with and along the edges of frayed biotite. The grains are challenging to locate using the optics of the SIMS system and given the fine-grained nature of these rock types. We attempted to place the beam on the majority of the grain, but the ion microprobe spot overlapped the matrix in many cases during the dating process. Thus, we applied an aperture filter to measure isotopic data from only the radiogenic regions in the central region of the spot. After dating, the location of the spot was determined using SE. CL images of the grains provide additional information to assist with interpretations.

Overall, zircon ages from these samples range from Triassic (234 ± 18 Ma, 238U/206Pb age, ±1σ, sample SPO20) to Early Eocene (42.4 ± 0.5 Ma, SPO6). Sample SPO20 yields the only Triassic result, whereas one zircon from sample SPO6 and six from SPO10 are Jurassic. All samples yield Cretaceous age zircons (SPO4 n = 4, SPO6 n = 2, SPO10 n = 2 and SPO20 n = 2). Only sample SPO6 contains one zircon that yielded Paleocene–Eocene ages.

The zircon that yields the Triassic age and the zircon with the second oldest age (187.1 ± 22.5 Ma, SPO6 zircon 2) are discordant, whereas zircon 19b in sample SPO10 is reversely discordant (Fig. 14). All other results are concordant. The spot on Triassic age zircon is located in a CL-dark interior domain of a euhedral grain, whereas the core of the early Jurassic age zircon appears mottled and has a highly luminescent rim (Fig. 15). The reversely discordant grain (SPO10_zircon 19b; Fig. 14b) is probably due to an overcorrection of common Pb. The youngest age is located in the cracked centre of a zircon that contains regions of bright 1 μm sized inclusions. Spots that overlap these brighter inclusions have higher Th/U (Table 5). The least radiogenic grain yields the results with the largest uncertainty (SPO10 zircon 1, 150.2 ± 34.0 Ma); it should be noted that the uncertainty in these SIMS ages largely depends on the amount of common Pb measured in the grain. Multiple spots were placed on larger individual zircons in samples SPO10 and SPO6. Both spots on zircon grain 7 in sample SPO10 overlap its brighter CL rim, but the spot that overlaps more of the darker CL core is c. 20 myr older. The zircon grain with two spots in sample SPO6 shows homogeneous CL brightness, but spots differ by c. 40 myr in age.

Seven zircons are Jurassic (from 187 ± 23 Ma, SPO6, to 148 ± 9 Ma, SPO10), and the average of all Jurassic results is 161 ± 18 Ma (weighted mean age (WMA) is 160 ± 17 Ma) and is consistent with a single population (mean square weighted deviation (MSWD) of 0.8). It should be noted that we include the WMA as it considers uncertainty in its estimation, and the age that is less precise contributes less to the final age, which is not the case for the average age and uncertainty calculation, which considers all of the uncertainties equally. The most precise Jurassic zircon age is 163 ± 8 Ma (SPO10) and has a Th/U >1. Seven zircon grains are early Cretaceous, and range in age from 142 ± 21 Ma (SPO6) to 86.1 ± 2.9 Ma (SPO4). The Cretaceous zircon ages average 124 ± 12 Ma with a WMA of 119 ± 3 Ma (MSWD = 13). The spot that yields the oldest early Cretaceous age is located adjacent to the spot that yields a Jurassic age (SPO6). This zircon shows no zoning in BSE or CL. The most precise Early Cretaceous age is 138 ± 6 Ma (sample SPO4), and the average of Early Cretaceous zircon ages is 133 ± 13 Ma (WMA is 134 ± 4 Ma, MSWD 4.4). Four zircons yield ages that average 141 ± 12 Ma (WMA of 140 ± 6 Ma, MSWD 0.01), and these analyses are of either whole grains or rims. Two zircon grains are Late Cretaceous (97.4 ± 13.4 Ma rim analyses, SPO20, and 86.1 ± 2.9 Ma, SPO4).

Discussion

Tectonic environment

The Spongtang ophiolite has a debated formation history (Reuber et al. 1983, 1990; Reuber 1986a, 107b; Pedersen et al. 2001; Mahéo et al. 2004). The major element geochemical results we obtained for spinel and olivine in harzburgite sample SPO25 suggest that it is a peridotite with mid-ocean ridge affinity, the product of partial melting of a MORB source beneath a spreading centre with no discernible subduction influence (Fig. 8). This scenario is consistent with the REE patterns of Spongtang clinopyroxene (Fig. 9). With their LREE-depleted patterns, they show no evidence of late-stage melt impregnation nor the extremely great extents of partial melting associated with some SSZ settings (see Pirard et al. 2013; Dygert & Liang 2015).

However, an SSZ setting influence on the ophiolite cannot be excluded, on the basis of samples recovered by Mahéo et al. (2004), who reported four Spongtang dunites that have spinel Cr#s implying evident SSZ affinity (Cr# >0.6; Fig. 8). We requested the location of these samples to compare with SPO25, but the data are unavailable. We note that Mahéo et al. (2004) reported data for lherzolites and harzburgites consistent with our results (Cr# <0.4). The Spongtang ophiolite is commonly assumed to be solely a remnant of an SSZ setting (e.g. Aitchison et al. 2000; Baxter et al. 2010; Aitchison & Davis 2004; Clift et al. 2014), but data reported here suggest a multistage origin.

The seemingly inconsistent Cr#s and REE patterns suggest that the ophiolite contains peridotites from multiple tectonic settings, or that it evolved in a setting that experienced increasing subduction influence through time. The occurrence of peridotites with abyssal and SSZ affinity in one locality has been reported elsewhere (e.g. Choi et al. 2008; Dilek et al. 2008; Uysal et al. 2012; Kingson et al. 2017). The record of multiple tectonic settings may be attributed to complex melt–rock reaction (e.g. Elthon 1992; Michibayashi et al. 2009; Dygert et al. 2016) and/or re-melting and re-crystallization processes that affect mineral compositions and modes as melt ascends (e.g. Niu et al. 1997). Such processes may be expected to influence dunites in particular, which are interpreted to represent conduits for mantle melt extraction (e.g. Kelemen et al. 1995, 1997; Suhr 1999) and have been shown to host multiple generations of melt with distinct source compositions (e.g. Dygert et al. 2016).

Cooling history

Geothermometers rely on a temperature-sensitive diffusive exchange of an element or component among phases. Because re-equilibration rates of mineral components and trace elements (e.g. trivalent REE v. divalent major elements) may differ by many orders of magnitude (e.g. Cherniak & Dimanov 2010, and references therein), differences in temperatures given by distinct thermometers for rocks that experienced cooling are expected (see Dygert et al. 2017). For samples that underwent slow, continuous cooling, thermometers based on relatively fast-diffusing major elements (e.g. the olivine–spinel of Fabriès 1979, and two-pyroxene solvus of Wells 1977; Brey & Köhler 1990; Putirka 2008) record lower temperatures than those based on slow-diffusing trivalent elements (e.g. the REE-in-two-pyroxene of Liang et al. 2013, and Al-in-orthopyroxene of Witt-Eickschen & Seck 1991). As a general rule, the larger the difference between a temperature given by a trivalent element-based thermometer and olivine–spinel or two-pyroxene solvus thermometer, the slower the sample cooled. Thus, samples that have TREE ≫ TBKN (which plot far to the left in the TREETBKN space shown in Fig. 10) cooled slowly relative to samples that have TREE ∼ TBKN (which plot along or near the blue 1:1 line in Fig. 10).

Sample SPO25 has relatively high and consistent temperatures from trivalent element-based thermometers (990–1058°C) and relatively low (and somewhat scattered) temperatures from solvus-based two-pyroxene thermometers (656–800°C) and olivine–spinel thermometers (794–841°C) (Table 4). The inconsistency among the two-pyroxene solvus temperatures may reflect calibration issues, as these thermometers are extrapolated to temperatures well beyond their calibration datasets. The disparity notwithstanding, TBKN and TREE are at the low end of the field defined by ophiolitic peridotites (Fig. 10a). The difference between the TREE and TBKN values is as large as measured in any other ophiolitic peridotite and suggests that Spongtang had an exceptionally long and slow cooling history. Inference of a meaningful cooling rate from the temperature data requires that SPO25 was not disturbed by metasomatic or melt–rock reaction events; these are not suggested by the major and trace compositions of Spongtang pyroxenes (Fig. 9, Tables 2 and 3), or the TREE inversion. Petrographic images of the sample SPO25 suggest alteration (Fig. 6), but our analyses are of intact grains that show largely homogeneous compositions (olivine, clinopyroxene; Table 2) or preservation of zoning (spinel, orthopyroxene; Table 3).

An estimate of the cooling rate of the sample can be made using a modified form of the Dodson (1973) closure temperature equation (Ganguly & Tirone 1999). Assuming an orthopyroxene grain radius of 0.5 mm (consistent with measured grain sizes in SPO25), we calculated cooling curves using methods of Dygert & Liang (2015); the cooling curves are plotted as magenta lines in Figure 10b. We assume that TREE can be modelled using REE diffusion rates in orthopyroxene (Cherniak & Liang 2007; Yao & Liang 2015), and that TBKN can be modelled using rates of Fe–Mg interdiffusion in clinopyroxene (Dimanov & Wiedenbeck 2006; Dygert & Liang 2015). Cooling rates in degrees per year are shown above the highest cooling curve; the position of SPO25 in this temperature space suggests that the Spongtang mantle section cooled at a rate of 100°C Ma−1 or slower.

Such slow cooling is consistent with but at the low end of rates expected for conductive cooling of the mantle lithosphere formed beneath an oceanic spreading centre with a ‘normal’ 5 km thick crustal section (Dygert et al. 2017). The rate may alternatively reflect extended cooling that occurred during or after the closure of a Tethyan basin, and/or the tectonic emplacement of the ophiolite. The lower TBKN and TREE of SSZ affinity ophiolites (Fig. 10) may partially reflect lower initial temperatures associated with their evolution in relatively cold, hydrous environments. However, we do not view a cooling history as a diagnostic indicator of tectonic affinity because of the possibility of resetting in the events leading up to obduction.

Chronology

The Spongtang ophiolite has been challenging to date because of its multi-stage history coupled with variable alteration. Here we discuss the U–Pb zircon and 40Ar/39Ar whole-rock ages by grouping results in time and focusing on their possible tectonic history by evaluating information provided by petrography and CL imagery of the dated grains.

A series of Triassic ages is recorded by 40Ar/39Ar data from the last degassing steps in basalt subvolcanic sample SPO17 (205 ± 3 Ma) (Fig. 13) and the darker CL core of a zircon crystal in deformed diabase dyke sample SPO20 of 234 ± 18 Ma (Fig. 15). Sample SPO17 was collected from the upper crust of the Spongtang ophiolite and shows minimal alteration (Fig. 4), whereas sample SPO20 is a highly foliated deformed dyke from the upper mantle unit (Fig. 5). Older Carboniferous to Permian 40Ar/39Ar inverse isochron and step ages are found in samples SPO6, SPO10 and SPO22, but these are highly uncertain. Data from last steps are the closest approximation to the high-temperature retention of Ar isotopes in high melting temperature minerals such as pyroxene and plagioclase. It should be noted that previous researchers have also reported Triassic argon ages from amphibole separates from the area, but these are also highly uncertain (Reuber et al. 1989, 1990; Mahéo et al. 2004) (Fig. 3).

Triassic crystallization ages appear inconsistent with the interpretation that the ophiolite formed during the Early Jurassic based on zircon fractions separated from a diorite pod (177 ± 1 Ma; Pedersen et al. 2001). Triassic 40Ar/39Ar whole-rock ages may represent excess 40Ar, and the disturbed age spectra suggest this as a possibility. Also, the Triassic zircon is discordant (Fig. 14c), and the older core may have been entrained in the rock from a different source, potentially continental crustal fragments that were adjacent to the northern passive continental margin of Greater India (Cao et al. 2018). However, the tectonic setting of the Early Jurassic age diorite pod (Pedersen et al. 2001) suggests that the sample intrudes higher-level gabbros of the ophiolite, and thus it may cut older rocks.

We report four zircon ages that overlap with the reported Jurassic age, including diabase dyke sample SPO6 (187 ± 23 Ma) and plagiogranite SPO10 (172 ± 9, 163 ± 8 and 150 ± 34 Ma) (Table 5, Fig. 15). It should be noted that sample SPO6 has a previously reported K/Ar age of 169.0 ± 9.5 Ma (Reuber et al. 1989, 1990). The average of all Jurassic zircon ages is consistent with a single age population and average 161 ± 18 Ma (WMA of 160 ± 17 Ma, MSWD of 0.8). The most precise Jurassic zircon age is located on a SIMS spot that overlaps a darker core and brighter rim (163 ± 8 Ma, SPO10; Fig. 15). The zircon with the Jurassic age in SPO6 shows flat CL zoning, but an adjacent spot is c. 40 myr younger (142 ± 21 Ma) and is similar to the age of the rim of the other Jurassic-age zircon in SPO10 (141 ± 7 Ma).

Overall, the Triassic to Late Jurassic ages suggest that the Spongtang ophiolite probably began its origin as part of the oldest portion of the Neotethyan domain (Late Triassic to Late Cretaceous; Sinha-Roy 1982; Şengör & Atayman 2009; Cao et al. 2018). This region is sometimes termed the Ceno-Tethyan Ocean (Metcalfe 1999; Matsuoka et al. 2002; Wakita & Metcalfe 2005). The Yarlung Zangbo Suture Zone itself is mapped as the zone of the closure of the Ceno-Tethyan Ocean (e.g. Metcalfe 1999, 2009, 2013). Based on the thermometry and geochemical data, we model the origin of the Spongtang ophiolite as part of a spreading centre with a spreading rate >2 cm a−1, as spreading centres with slower spreading rates have thin or no crust (e.g. Cann et al. 1997; Dick et al. 2003, 2008; Morris et al. 2009; Craig & Parnell-Turner 2017). The ophiolite cooled rapidly by hydrothermal circulation and tectonic exhumation, recording higher TREE compared with TBKN (Fig. 10; Dygert & Liang 2015).

Both whole-rock 40Ar/39Ar and zircon ages are reported from samples SPO4, SPO6 and SPO10. The expectation is that zircon ages should be older than those obtained using the 40Ar/39Ar approach, with the last degassing argon steps approaching zircon U–Pb results. The larger uncertainty in the zircon ages in samples SPO6 and SPO10 preclude this type of interpretation, although the oldest zircon in sample SPO4 (140 ± 3 Ma) overlaps the last degassing 40Ar/39Ar step-heating results (142.8 ± 16.6 Ma). Reuber et al. (1989, 1990) also reported a 139 ± 8 Ma K/Ar age from amphibole in this sample, suggesting that the c. 140 Ma result should be considered part of the history of the ophiolite. Overall, we report 10 Cretaceous zircon ages from the Spongtang ophiolite, and four of these zircons yield ages that average 141 ± 12 Ma (WMA of 140 ± 6 Ma, MSWD <0.01). Spots on two of these zircons overlap rims in CL, and the textural relationships of the Cretaceous zircons suggest that they could be associated with a tectonic event that facilitated zircon recrystallization after the Jurassic.

The period recorded by the Cretaceous zircons has been ascribed to the development of an intra-oceanic arc represented by the Samail (Oman), Zagros (Iran), Chagai–Raskoh (western Pakistan, southern Iran, eastern Afghanistan), Kandahar and Kohistan–Ladakh, Dazhuqu and Zhongba arcs (Brookfield & Reynolds 1981; Bhutani et al. 2004; Abrajevitch et al. 2005; Dai et al. 2011; Siddiqui et al. 2012; Baxter et al. 2016) (Fig. 1). Reuber et al. (1989, 1990) suggested that a significant intra-oceanic thrusting event affected the Spongtang ophiolite during the Early Cretaceous, and Mahéo et al. (2004) suggested that this was a time when subduction was initiated, based on interpretations of disturbed 40Ar/39Ar amphibole step-heating data. The Early Cretaceous (140–130 Ma) is also a time when the India subcontinent initiated rifting from other continents and fragments associated with Gondwana (e.g. Scotese et al. 1988; Jadoul et al. 1998; Hu et al. 2010).

A Late Cretaceous intra-oceanic arc has been suggested to be present within the Neotethyan Ocean near the palaeo-equator (e.g. Reuber 1986b; Abrajevitch et al. 2005; Metcalfe 2009; Dai et al. 2011; Siddiqui et al. 2012, 2017) or c. 30°N (Zhu et al. 2013). Paleogeographic reconstructions of Lawver et al. (2018) restrict its location to the south of the Lhasa Terrane. It should be noted that some have indicated that the Lhasa Terrane had already accreted onto other Tibetan-related continental fragments by the Late Jurassic–Early Cretaceous (Rolland 2002; Kapp et al. 2003, 2007; Guynn et al. 2006; Zhu et al. 2013), and others have suggested the presence of multiple arcs (e.g. Zyabrev et al. 2008). This observation and alternative palaeogeographical reconstructions (e.g. Ali & Aitchison 2005) do not change our overall observation that the data reported here are consistent with the presence of a Late Cretaceous intra-oceanic Neotethyan arc.

We speculate that the Spongtang ophiolite records a transition from mid-ocean ridge affinity-type spreading centre to a setting with increasing subduction influence, perhaps after a subduction polarity reversal as the Neotethyan basin closed before obduction. This scenario is consistent with the presence of both the abyssal and the SSZ signatures, with the Spongtang ophiolite exposures and the seafloor spreading suggested by the REE and solvus thermometry, and trace element geochemistry. The timing of this transition cannot be specifically constrained with the ages reported here, but we speculate that this occurred sometime after the Jurassic, and is probably recorded by ages of zircon rims (c. 140 Ma).

Late Cretaceous zircon ages may represent continued crystallization as the subduction-zone influence increased, or are the result of the mixing of age domains. It should be noted that these results are similar to the 40Ar/39Ar age of basalt associated with the Indus Suture north of the field area (128.2 ± 2.6 Ma; Bhutani et al. 2004) and therefore are likely to represent continued crystallization and emplacement. Zircon ages from SPO4 (131.2 ± 4.3 Ma) and SPO10 (126.3 ± 10.2 Ma) overlap zircon ages from underlying sediments (129 ± 7 Ma; Clift et al. 2014) (Table 5). The spot that yields the SPO4 age samples multiple zircon clusters and the spot on the zircon in SPO10 overlaps a darker CL rim with veins that intrude its bright core (Fig. 15). These ages thus probably represent mixed age domains. Inverse isochron 40Ar/39Ar Late Cretaceous ages are highly uncertain, but selected step heating ages are more precise (101.7 ± 1.2 Ma, dyke SPO15; 98.4 ± 2.3 Ma, basalt subvolcanic sample SPO17; 73.7 ± 2.2 Ma, pillow lava SPO19) (Figs 12 and 13). These ages overlap the oldest reported age constraints timing the collision of India and the Kohistan–Ladakh island arc during the Late Cretaceous (c. 95–65 Ma) (see review by Khan et al. 2009). The Late Cretaceous has been considered to the time of ophiolite obduction in the region (82 ± 6 Ma; Brookfield & Reynolds 1981) and the crystallization of other adjacent ophiolite sequences that may be correlative (80.2 ± 1.5 Ma, Muslim Bagh; Kakar et al. 2012). Some of the Late Cretaceous ages we report are similar to the oldest ages reported for the Kohistan and Kargil batholiths, part of a granitic system north of the field area that has been used to time Neotethyan subduction (c. 101 Ma; Honegger et al. 1982; Schärer et al. 1984). However, the majority of zircon U–Pb ages reported for those granites are younger (see Weinberg & Dunlap 2000; Khan et al. 2009).

The oldest Cenozoic 40Ar/39Ar ages we obtained are from the last degassing steps of samples SPO14 (58.7 ± 0.7 Ma) and SPO10 (56.7 ± 13.6 Ma) (Fig. 12), and the inverse isochron age of the last steps of sample SPO19 (54.76 ± 0.54 Ma) (Fig. 13). These 40Ar/39Ar ages average 56.7 ± 5.2 Ma (WMA of 56.2 ± 0.2 Ma, MSWD 6.6). The oldest Cenozoic zircon age is from a single zircon in sample SPO6 (64.3 ± 0.8 Ma), and the grain has younger age domains of 52.5 ± 1.2 and 42.4 ± 0.5 Ma (Fig. 15, Table 5). BSE images of the dated zircon with bright and cracked zones suggest that the results probably time metamorphism and zircon recrystallization. The ages are consistent with palaeomagnetic data indicating a decrease in speed of the Indian craton at this time (e.g. Klootwijk et al. 1992; Molnar & Stock 2009; Copley et al. 2010; Dupont-Nivet et al. 2010; Cande & Stegman 2011; van Hinsbergen et al. 2011), and the proposed timing of Spongtang ophiolite obduction (70–60 Ma, Searle 1986; 55–50 Ma, Kelemen et al. 1988; Garzanti et al. 2005; Baxter et al. 2016). All Cenozoic ages we obtain are in line with broad estimates of the timing of Indo-Asia collision (e.g. Rowley 1996; Yin & Harrison 2000; Guillot et al. 2003; Aitchison & Davis 2004; Khan et al. 2009; Bhutani et al. 2009; Molnar & Stock 2009; Cai et al. 2011; van Hinsbergen et al. 2011; Hu et al. 2016; Najman et al. 2017; Tong et al. 2017).

Although uncertain, five of the Spongtang 40Ar/39Ar determinations produced Oligocene absolute inverse isochron ages (c. 34 Ma, SPO4, SPO6, SPO22, SPO10 and SPO14; Figs 11 and 12). Others have suggested that a dual-collision scenario occurred in the Himalayas during the Oligocene (e.g. Aitchison et al. 2000, 2007; Aitchison & Davis 2004; Abrajevitch et al. 2005; Ali & Aitchison 2005, 2014). One sample yields a more precise Miocene result (21.2 ± 2.4 Ma; Fig. 13). It should be noted that Oligocene to Miocene ages are commonly reported throughout the Himalayas and may time events related to mountain-building along large-scale thrust systems in the range (see review by Yin 2006). These Oligocene ages are uncertain and could represent the mixing of age domains coupled with excess argon. We choose to discuss the results as a significant number of dated samples seem to have this age signature. The youngest 40Ar/39Ar results (2–5 Ma) recorded by all samples probably represent the degassing of alteration minerals (muscovite, sericite, chlorite).

Conclusions

Here we report geochemical and geochronological data from samples from the Spongtang ophiolite, a piece of obducted oceanic lithosphere located in the NW Himalayas. Major and trace element mineral chemistry and geothermometry suggest that the ophiolite experienced slow cooling after forming at a spreading centre with mid-ocean ridge affinity. The difference between the TREE and TBKN values estimated for harzburgite sample SPO25 is as large as that for any ophiolitic peridotite yet analysed, and suggests an exceptionally long and slow cooling history, assuming the lack of disturbance by metasomatic or melt–rock reaction events. The geochemical and geochronological results reported here, coupled with previous work (Pedersen et al. 2001; Mahéo et al. 2004), suggest that the ophiolite began initial crystallization at a spreading centre with a spreading rate >2 cm a−1 within the oldest parts of the Neotethyan ocean possibly during the latest Triassic to early Jurassic.

The Triassic time frame is constrained by the highest temperature 40Ar/39Ar degassing step of a basalt subvolcanic assemblage (205 ± 3 Ma) and a single 238U–206Pb age of a zircon core in a deformed upper mantle unit dyke (234 ± 18 Ma). The Jurassic timing is the result of multiple 238U–206Pb zircon ages (161 ± 18 Ma, WMA of 160 ± 17 Ma) that are c. 10 myr younger, but still similar to those reported previously (177 ± 1 Ma; Pedersen et al. 2001). Based on the presence of rocks with the chemistry of rocks associated with SSZ settings (Mahéo et al. 2004), we suggest that the ophiolite then experienced increasing subduction influence during continuing deformation that occurred during the Early Cretaceous. The Early Cretaceous timing is constrained by 238U–206Pb zircon data (123.6 ± 12.1 Ma, WMA age of 119.1 ± 2.5 Ma). We model this transition as a subduction polarity reversal along a subduction zone adjacent to the spreading centre that formed the ophiolite owing to closure of the Neotethyan Ocean. This process may have occurred as part of the development of a larger intra-oceanic arc system that has been documented elsewhere across the Neotethyan Ocean. Based on the most precise zircon and whole-rock 40Ar/39Ar ages from several samples, final obduction of the Spongtang ophiolite occurred in the Cenozoic. The 238U–206Pb zircon ages constrain the timing of obduction between 64.3 ± 0.8 and 42.4 ± 0.5 Ma, which is broadly consistent with whole-rock 40Ar/39Ar ages of 56.7 ± 5.2 Ma.

Acknowledgements

No real or perceived financial conflicts of interest exist for any author of this paper. Data supporting the conclusions can be obtained from the supplementary material. Support for obtaining the geochemical data in this paper was provided by funds to E.P. provided by the Jackson School of Geosciences at The University of Texas at Austin. N.D. acknowledges the support of a postdoctoral fellowship provided by the Jackson School of Geosciences. Geochemical data were collected by N.D. using the Electron Microbeam facility and Quadrupole ICP-MS Laboratory in the Jackson School of Geosciences at UT Austin. We acknowledge assistance by N. Miller and D. Zhao. Age data were collected by E.C. with the assistance of the Max Kade Foundation. SIMS facilities at UCLA and Heidelberg University acknowledge support through NSF Instrumentation and Facilities and DFG Scientific Instrumentation and Information Technology programmes, respectively. We appreciate discussions with I. Dalziel.

Scientific editing by Linda Kirstein

This is an Open Access article distributed under the terms of the Creative Commons Attribution 3.0 License (http://creativecommons.org/licenses/by/3.0/)