The upper mantle section of the Cuobuzha ophiolite in the northern subbelt of the Yarlung Zangbo suture zone in southwest Tibet comprises mainly clinopyroxene (cpx)-rich and depleted harzburgites. Spinels in the cpx-harzburgites show lower Cr# values (12.6–15.1) than the spinels in the harzburgites (26.1–34.5), and the cpx-harzburgites display higher heavy rare earth element concentrations than the depleted harzburgites. The harzburgites have subchondritic Os isotopic compositions (0.11624–0.11699), whereas the cpx-harzburgites have suprachondritic 187Os/188Os ratios (0.12831–0.13125) with higher Re concentrations (0.380–0.575 ppb). Although these geochemical and isotopic signatures suggest that both peridotite types in the ophiolite represent mid-oceanic ridge–type upper mantle units, their melt evolution trends reflect different mantle processes. The cpx-harzburgites formed from low-degree partial melting of a primitive mantle source, and they were subsequently modified by melt-rock interactions in a mid-oceanic ridge environment. The depleted harzburgites, however, were produced by remelting of the cpx-harzburgites, which later interacted with mid-oceanic ridge basalt– or island-arc tholeiite–like melts, possibly in a trench–distal backarc spreading center. Our new isotopic and geochemical data from the Cuobuzha peridotites confirm that the Neo-Tethyan upper mantle had highly heterogeneous Os isotopic compositions as a result of multiple melt production and melt extraction events during its seafloor spreading evolution.


Ophiolites are fragments of ancient oceanic lithosphere that have been emplaced onto continental margins, accretionary prisms, or island arcs during collisional and subduction-accretion events (Moores, 1982; Dilek and Robinson, 2003; Dilek and Furnes, 2011, 2014; Yang et al., 2014, 2015). The well-preserved peridotites within these ophiolites provide important information on melt extraction, partial melting, and melt-rock interaction within the upper mantle sections of the paleo-oceanic lithosphere (e.g., Dilek and Thy, 1998, 2009; Shallo and Dilek, 2003; Zhou et al., 2005; Arai et al., 2007; Dilek et al., 2007; Dai et al., 2011; Morishita et al., 2011; Uysal et al., 2012, 2015; Saka et al., 2014; Saccani et al., 2015; Niu et al., 2017). They display critical structural, petrological, and geochemical evidence for the processes that took place during oceanic lithosphere formation in different tectonic settings (e.g., Dilek and Newcomb, 2003; Pagé et al., 2009; Dilek and Furnes, 2014; Saccani et al., 2017).

The Yarlung Zangbo suture zone (YZSZ) in southern Tibet includes the remnants of Neo-Tethyan oceanic lithosphere and marks a major suture between the Indian plate to the south and the Lhasa terrane of Tibet to the north (Dupuis et al., 2005). The ophiolites within the YZSZ have been studied over the past 30 yr (Hébert et al., 2012, and references therein); however, their tectonic setting remains controversial in terms of whether they represent mid-oceanic ridge–generated oceanic lithosphere (Girardeau and Mercier, 1988; Nicolas, 1989) or the remnants of hydrous oceanic lithosphere that formed above an actively subducting slab in suprasubduction zone settings (Hébert et al., 2012). Recent field-based geochronological and geochemical studies along the YZSZ have shown that the internal structure of the suture zone is complex, with several distinct subbelts separated by continental fragments and showing different lithological, geochemical, and structural characteristics (Liu et al., 2015a, 2015b; Xu et al., 2015). In the western part of the YZSZ, the northern and the southern subbelts form two subparallel zones of mafic-ultramafic rock assemblages with overlapping crystallization ages (Wei et al., 2006; Xiong et al., 2011; Hebért et al., 2012; Liu et al., 2015a). A narrow sliver of deformed continental rocks occurs between these two subbelts within the YZSZ. The origin of this continental sliver and the potential genetic relationships between the different ophiolite massifs in these subbelts remain unknown (Liu et al., 2015b; Feng et al., 2015).

The Cuobuzha ophiolite in the northern subbelt includes an extensive upper mantle peridotite suite intruded by doleritic dikes and overlain by gabbros (Feng et al., 2015). Previous zircon laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) dating of these mafic dikes has yielded a crystallization age of 125.8 Ma (Liu et al., 2015b). This Early Cretaceous age of the mafic dikes provides a minimum time constraint for the formation of the Cuobuzha massif in the northern subbelt, and is compatible with the available crystallization ages from other ophiolite massifs in the southern subbelt, including the Dongbo (128 ± 1 Ma from gabbros; Xiong et al., 2011), Purang (130 ± 3 Ma from gabbros; Liu et al., 2011), and Xiugugabu (126–122 Ma from doleritic dikes; Wei et al., 2006). Although the available ages from both ophiolitic subbelts reveal nearly coeval timing of oceanic crust formation, the mantle melt evolution and the isotopic features of the upper mantle peridotites exposed within the YZSZ are still poorly known.

In this paper we present new mineral chemistry, whole-rock major and trace element geochemistry, and Re-Os isotopic data from the Cuobuzha ophiolite, and discuss its mantle melt evolution within the framework of the Tethyan tectonics of the YZSZ. Our results and interpretations provide new insights and constraints for the heterogeneous nature of the Neo-Tethyan mantle, currently exposed along the >2000-km-long suture zone.


The YZSZ is the southernmost and the youngest of the sutures that divide the Tibetan Plateau into various east-west–oriented terranes (Fig. 1A). This suture extends discontinuously for nearly 2000 km and is divided into the eastern Qushui-Motuo and the central Angren-Renbu sections, and a western section from the west of Saga to the Sino-India boundary (Liu et al., 2015a). The latter is further divided into the northern Dajiweng-Saga and the southern Daba-Xiugugabu ophiolite subbelts, which are separated by the Zhada-Zhongba continental sliver. The southern belt of the western YZSZ contains a series of well-preserved ophiolite massifs, including the Dongbo (∼400 km2) and Purang (∼600 km2) ophiolites (Xiong et al., 2011; Zhou et al., 2014; Liu et al., 2015a, 2015b; Niu et al., 2015). In comparison, the majority of the ophiolites within the northern subbelt are generally highly fragmented and altered, with a notable exception of the better preserved Dajiweng, Baer, Cuobuzha, Jianabeng, and Saga ophiolites (Liu et al., 2015b).

The northwest-southeast–trending Cuobuzha ophiolite crops out in the northern subbelt of the western YZSZ (Fig. 1B). It is predominantly made of upper mantle peridotites, composed of harzburgite and clinopyroxene (cpx)-harzburgite units with minor lenses and veins of dunite, which contain local occurrences of massive and disseminated chromitite. Massive chromitite is more common in these peridotites in comparison to podiform types. All peridotites are crosscut by mafic northwest-southeast–oriented doleritic and gabbroic dikes, parallel to the general trend of the ophiolite (Fig. 1C).


As a tectonic block, the southern margin of the Cuobuzha massif (15 km in length, 0.2–0.5 km in width) was covered by Quaternary sediment, and the northern margin is uncomformably contacted with the accretionary volcanic-sedimentary unit, which consists mainly of shale, alkaline basalt, siliceous rocks, siliceous limestone, and pyroclastic rocks. The hydrothermal metasomatic listwanite is 10–20 m in thickness overlying the Cuobuzha massif. We examined the dunite and harzburgite outcrops carefully in the field in order to document their contact relationships. We then systematically collected 13 harzburgite samples from the Cuobuzha ophiolite. Petrographic examination of thin sections provides detailed mineralogical and textural information on discrimination of the harzburgite and clinopyroxene (cpx)-harzburgite. It is difficult to distinguish these two groups of peridotites in the field.

The majority of the peridotite samples appear to have undergone only minor serpentinization. The depleted harzburgite consists mainly of olivine (55%–65%) and orthopyroxene (opx, 25%–35%) with minor amounts of cpx, spinel, and magnetite (Fig. 2A). The cpx-harzburgite contains olivine (55%–60%), opx (25%–35%), cpx (∼3%), and minor amounts of spinel and magnetite (Fig. 2E). The opx grains in these peridotites occur as 8–10-mm-long porphyroblasts that are generally free of alteration (Fig. 2C); some opx porphyroblasts display kink banding and undulose extinction, indicating that they underwent large strain in lithospheric conditions (Fig. 2H). Larger opx crystals generally contain cpx exsolution lamellae and olivine inclusions (Fig. 2C). The spinel in these peridotites is present as an interstitial phase or as euhedral inclusions within olivine and opx grains (Figs. 2B, 2F).


Mineral Chemistry

Major element compositions of the representative minerals were determined using a JXA-8100 electron microprobe at the Institute of Geology, Chinese Academy of Geological Sciences (Beijing, China). An accelerating voltage of 15 kV, a beam current of 1 × 10−8 A, and an electron beam diameter of 1 mm were utilized, and natural standards were used for calibration.

Whole-Rock Major and Trace Element Compositions

Whole-rock major and trace elements were analyzed at the China National Research Center of Geoanalysis, Chinese Academy of Geological Sciences (Beijing, China). Major element concentrations were determined on fused glass beads by X-ray fluorescence spectrometry, yielding an estimated analytical accuracy of 1% relative for SiO2 and 2% relative for all other major elements. Trace elements, including the rare earth elements (REE), were determined by solution ICP-MS, with GB/T14506.28–2010 (GSR3) and GB/T14506.14–2010 (GSR5) certified reference standards and three internal standards also measured simultaneously to ensure the consistency of the analytical results. Analytical uncertainties are estimated to be 10% for trace elements with concentrations of <10 ppm, and ∼5% for elements with concentrations of >10 ppm.

Volatile concentrations were determined by gravimetric techniques whereby samples were heated in a closed container before the resulting water vapor was collected in a separate tube to be condensed and weighed.

Re-Os Isotopic Analysis

Re-Os isotopic measurements were done at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (GIG-CAS) using rock samples that were first split into small chips using a hammer wrapped in paper to avoid contamination. Some 0.5–0.6 g of finely powdered sample was spiked with 190Os and 185Re prior to freezing with 2 mL of concentrated HCl and 6 mL of concentrated HNO3 in a Carius tube. The samples were then sealed and heated to 240 °C for 24 h (Shirey and Walker, 1998). Carbon tetrachloride was subsequently added to the acid mixture before Os was extracted from the aqueous phase (Pearson and Woodland, 2000), with the resulting Os purified by microdistillation. Re was separated and purified by anion exchange using AG1X8 resin (100–200 mesh). Isotope abundances in OsO3 and ReO4 were determined using a Thermo Finnigan TRITON mass spectrometer in negative ion detection mode (Volkening et al., 1991), equipped with an oxygen gas leak valve and an ion counting multiplier. Os was loaded onto a high-purity Pt filament (manufactured by H. Cross Co., Ltd.; 99.999% purity, 1 × 0.025 mm) that had been heated in air for >3 min. The resulting Re and Os isotope compositions were measured using a static multiple Faraday collector and a pulse-counting electron multiplier, respectively. The instrumental mass fractionation of Os was corrected by normalizing measured 192Os/188Os ratios to 3.08271 (Nier, 1937), with Re isotopes measured using a total evaporation technique. The oxygen subtraction for Re and Os used 17O/16O = 0.00037 and 18O/16O = 0.002047 ratios (Nier, 1950). Both Re and Os were blank corrected, with total blank levels of 7.1 ± 0.3 and 2.0 ± 0.4 pg (2 standard deviation, SD, n = 13) for Re and Os, respectively, with a blank 187Os/188Os ratio of 0.298 ± 0.028 (2 SD, n = 13). These data indicate only minor contributions of the blanks to the Os and Re contents (<1%). The multicollector-ICP-MS analytical procedures used at GIG-CAS are similar to those described by Li et al. (2010).



The average compositions of olivine in the Cuobuzha peridotites are given in Table 1. The forsterite contents [Fo; 100 Mg/(Mg + Fe2+)] of olivine in the harzburgite samples are 90.1–90.9, which are higher than those of olivine grains in the cpx-harzburgites (89.7–90.0). However, the olivine in the depleted harzburgites contains lower concentrations of NiO (0.33–0.35 wt%) in comparison to the olivine in the cpx-harzburgites (0.33–0.39 wt%).


The average compositions of the opx in the Cuobuzha peridotites are given in Table 2. The opx in the depleted harzburgite has enstatite (En) contents of 88.8–90.1, whereas the opx in the cpx-harzburgite has En contents of 87.6–88.8. These data indicate that the En content of the opx grains correlates positively with an increasing depletion of the peridotites. Their Mg# [atomic100 × Mg/ (Mg + Fe)] values range from 90.5 to 91.1 in the harzburgites and from 88.9 to 89.7 in the cpx-harzburgites. Their uniform Al2O3 and Cr2O3 concentrations also range from 2.57 to 4.07 wt% and from 0.25 to 0.63 wt%, respectively.


The average cpx chemical compositions are given in Table 3. The cpx compositions in the harzburgites are dominated by diopside with En, wollastonite (Wo), and ferrosilite (Fs) values of En47.1–49.6Wo46.7–49.3Fs3.4–3.9 and in the cpx-harzburgites by En45.8–46.6Wo49.3–49.8Fs3.8–4.4. The cpx in the harzburgites has Mg# values of 92.5–93.2 compared with the cpx in the cpx-harzburgites having Mg# of 91.2–92.4. In addition, the cpx in the harzburgites contains higher concentrations of TiO2 (0.41–0.53 wt%) and lower concentrations of Al2O3 (3.56–3.95 wt%) than the cpx in the cpx-harzburgite (TiO2 of 0.20–0.25 wt%, Al2O3 of 4.47–4.78 wt%).


The spinel compositions of the Cuobuzha peridotites are given in Table 4. Spinel is widespread in these peridotites but rarely exceed 3% modal abundance. They are relatively Al rich, with Al2O3 concentrations of 37.72–43.76 wt% for the spinel in the harzburgite and 52.55–54.97 wt% for the spinel in the cpx-harzburgite samples. They are classified as spinel and have Mg# values of 65.2–67.9 in the harzburgite and 70.9–74.5 in the cpx-harzburgite samples. Spinels in the Cuobuzho peridotites also have a narrow range in Cr# [atomic 100 × Cr/(Cr + Al)] values (cpx-harzburgites 10.55–17.10; harzburgites 22.06–39.35), with an increase in Cr# values typically correlating with increasing degrees of partial melting in the host peridotites. These spinel features provide strong evidence of mantle depletion (Dick and Bullen, 1984). The spinel in the cpx-harzburgite samples contains lower concentrations of TiO2 (0.02–0.04 wt%) than the spinel in the harzburgite samples (0.10–0.23 wt%). Spinel NiO concentrations systematically decrease from the cpx-harzburgites (0.27–0.30 wt%) to the harzburgites (0.15–0.20 wt%).


Major Oxide and Trace Element Data

The major and trace element compositions of the Cuobuzha peridotites are given in Table 5. All of the samples underwent variable degrees of serpentinization, as evidenced by the loss on ignition values (0.76–6.59 wt%). The influence of serpentinization effects in our samples was removed by recalculating all concentrations to a 100% anhydrous basis, yielding the values discussed in the following.

The harzburgite and cpx-harzburgite samples have distinct chemical compositions, with the former containing lower concentrations of Al2O3 and CaO (1.19–1.67 and 0.95–1.85 wt%, respectively) than the latter (2.41–2.77 and 2.34–3.05 wt%, respectively), correlating with the cpx abundance in these samples. Most of the peridotites are enriched in the transition elements that are compatible in ultramafic rocks (Ni = 1956–2274 ppm, Cr = 1697–2822 ppm) compared to the primitive mantle (Ni = 1860 ppm, Cr = 2520 ppm; Palme and O’Neill, 2003). Both rock types have similar whole-rock Mg# values (91–92), although the harzburgites contain higher concentrations of MgO (41.03–43.83 wt%) than the cpx-harzburgites (39.29–40.84 wt%). These concentrations can be used as an indicator of the degree of depletion, with increases in MgO correlating with increased depletion and increased olivine abundance. The most effective approach to examining the chemical changes in our peridotite samples is to compare the concentrations of elements of interest with MgO concentrations. Both rock types have negative correlations between SiO2, Al2O3, and CaO with MgO. Figure 3 compares the Cuobuzha peridotites with abyssal (or mid-oceanic ridge type) and suprasubduction zone (SSZ) type peridotites. The Cuobuzha peridotites are more similar to mid-oceanic ridge type than SSZ type, similar to the compositions of peridotites from the Purang (Zhou et al., 2014) and Dongbo (Niu et al., 2015) ophiolites within the southern subbelt and the Dajiweng ophiolite (Lian et al., 2014) in the northern subbelt of the YZSZ.

The upper mantle peridotites within the Cuobuzha ophiolite have total REE (SREE) concentrations of 0.30–1.63 ppm that are well below the primitive mantle concentrations, indicating significant degrees of depletion, presumably as a result of variable degrees of partial melting (Miller et al., 2003). All of these peridotites have REE patterns that are increasingly depleted between Lu and Pr but are slightly enriched in Ce and La (Fig. 4A). The harzburgite samples are more depleted in the heavy (H) REEs than the cpx-harzburgites, and have a shallow slope between the HREEs and the medium (M) REEs. However, all the Cuobuzha peridotites have similar light (L) REE concentrations. In addition, one harzburgite sample (12YC-6) contains lower HREE contents than the other samples, although its MREE and LREE values are similar to those of the other samples. Our peridotite samples also have similar primitive mantle–normalized variation patterns (Fig. 4B), showing enrichment in both the large ion lithophile elements (LILE; Rb and Sr) and the high field strength elements (HFSE; Ta, Hf, and Ti). The more depleted HREE patterns of Sample12YC-6 indicate higher degrees of partial melting than the other samples (Fig. 4).

Os Isotope Data

The Os isotopic compositions of the Cuobuzha peridotites are given in Table 6. The harzburgite and cpx-harzburgite samples contain Re concentrations of 0.080–0.575 and 0.380–0.575 ppb, respectively. The harzburgites have subchondritic Os isotopic compositions (0.11624–0.11699) compared with the carbonaceous chondrite ratio of 0.127 (Shirey and Walker, 1998). In contrast, the cpx-harzburgites have suprachondritic 187Os/188Os ratios (0.12831–0.13125; Fig. 5A) that are similar to the values for peridotites in other ophiolites, including the Purang (also named as Yungbwa) (187Os/188Os ratios of 0.1223–0.1313; Miller et al., 2003; Liu et al., 2012) and Troodos (0.1291–0.1390; Büchl et al., 2002) peridotites.

The Os concentrations of the Cuobuzha peridotites (3.897–7.693 ppb) are higher than the Os concentrations within any possible metasomatic agents (0.100 ppb; Dale et al., 2007), suggesting that the Re-Os system within the Cuobuzha peridotites was not contaminated subsequently. In addition, these peridotites have low Re/Os ratios (0.010–0.136; Table 6) and a negligible age correction for isotopic growth since the time of formation. As such, the Os isotope data presented here most likely reflect the primary Os compositions of the Cuobuzha peridotites.


Degrees of Partial Melting and Implications

Experimental studies have shown that the equilibrium between olivine and melt remains unchanged by the addition of H2O to the system (Gaetani and Grove, 1998), suggesting that olivine Fo compositions are indicative of the total degree of partial melting undergone by upper mantle peridotites (Uysal et al., 2012). Given that the Cuobuzha peridotites contain olivine with high Fo contents (89.1–91.4), their compositions reflect different degrees of partial melting. In addition, the Cr# and Mg# values of the spinels within these peridotites show a negative correlation, providing more evidence that these rocks were produced by different degrees of partial melting (Fig. 6A). The spinel Cr# values and olivine Fo compositions for these peridotites plot along the olivine-spinel mantle array (Arai, 1994; Fig. 6B), confirming that the Cuobuzha peridotites represent the residues after variable degrees of melt extraction. The cpx-harzburgites represent the residues after low-degree partial melting (∼5%), whereas the depleted harzburgites represent the residues after higher degrees of partial melting (10%–17%). In general, the LREE and the MREE values are moderately affected by mantle metasomatism effects, but HREE abundances are commonly less affected by post-melt extraction metasomatism (Bodinier et al., 1988). Therefore, we have applied modeling of the HREE values in our samples, and obtained similar results, with partial melting degrees increasing from the cpx-harzburgites (5%–8%) to the harzburgites (15%–17%; Fig. 4A).

Oxygen Fugacity and Tectonic Setting

It is generally accepted that mantle wedge peridotites above subduction zones are more oxidized than upper mantle peridotites in other tectonic settings (e.g., Parkinson and Arculus, 1999). Plotting peridotite compositions in a V versus Yb (Fig. 7A) diagram can provide insights into oxygen fugacity (ƒO2) conditions, which are particularly useful because depletion trends are strongly dependent on ƒO2 conditions (Pearce and Parkinson, 1993). The relatively reducing conditions that are characteristic of abyssal peridotite formation (e.g., fayalite-magnetite-quartz, FMQ–1) yield high V3+/(V4+ + V5+) values, consistent with low partition coefficients. The resulting mantle residues are therefore relatively undepleted in V at a given degree of partial melting. In contrast, the relatively oxidizing conditions that are characteristic of the formation of SSZ peridotites (e.g., FMQ+1) are associated with low V3+/(V4+ + V5+) values and high partition coefficients, leading to rapid depletion of V in the mantle residue and a steeper depletion trend in the V versus Yb diagram. All the cpx-harzburgite samples plot between the FMQ and FMQ–1 trends in Figure 7A, whereas most of the harzburgite samples plot close to the FMQ trend (Fig. 7A). All of these samples plot in the same field of the V versus MgO diagram (Fig. 7B), with half of the depleted harzburgite samples plotting between FMQ and FMQ+1 trends. We attribute the wide range of oxygen fugacities but a low Cr# of the spinel compositions in our peridotite samples to fluid-rock interactions (Aldanmaz et al., 2009).

These observations and findings indicate that the Cuobuzha peridotites were the products of low-degree partial melting in a mid-oceanic ridge environment. The depleted harzburgites were most likely derived from remelting of the cpx-harzburgite rocks, and some of the peridotite samples that plot between the FMQ and FMQ+1 trends may have resulted from fluid-rock interactions following the formation of the harzburgites.

Melt-Peridotite Interaction

Conventionally, cpx-harzburgites and harzburgites are thought to represent depleted refractory residues produced by partial melting of the mantle (e.g., Himmelberg and Loney, 1973). However, the Cuobuzha peridotites are also LREE enriched and have high LREE/MREE ratios, indicating that these upper mantle rocks could not just have formed as residual material after partial melting of the primitive mantle (e.g., Frey et al., 1991). They also have LILE (Rb and Sr) enriched primitive mantle–normalized multielement variation patterns (Fig. 4B). They plot away from partial melting curves in an Al2O3 (cpx) versus TiO2 (cpx) diagram (Fig. 8A), as a result of elevated concentrations of TiO2 in their cpx. These observations strongly suggest that the Cuobuzha peridotites were not generated by simple partial melting processes.

One possible explanation for the unusual compositions of these upper mantle peridotites is melt-rock interactions (Sharma and Wasserburg, 1996). However, these peridotites also show HFSE (Ta, Hf, and Ti) enrichments in primitive mantle–normalized multielement variation diagrams, and these elements are commonly considered immobile during low-temperature alteration (Fig. 4B; Dai et al., 2011). These patterns suggest that the Cuobuzha peridotites record variations in their partial melting degrees as well as variable melt-rock interaction effects, as observed in other ophiolitic mantle peridotites (e.g., Zhou et al., 2005; Morishita et al., 2011; Uysal et al., 2012; Wu et al., 2017). Mineralogical and textural evidence for melt-rock interactions is provided by the presence of embayments in opx that are filled with olivine (Fig. 2D) and by the presence of vermicular spinel in our peridotite samples (Fig. 2F).

In terms of melt-rock interaction processes, the trends in Figure 8B demonstrate clear differences between the harzburgite and cpx-harzburgite samples. The cpx-harzburgite samples plot close to a partial melting trend, whereas the depleted harzburgite samples define a trend between the mid-oceanic ridge basalt (MORB) and island-arc tholeiite (IAT) spinel compositions, such as those documented from the Scotia Sea and the Lau Basin. The incompatible element enrichments observed for the Cuobuzha cpx-harzburgites may have been caused by melt-rock interactions following a partial melting episode in a mid-oceanic ridge setting. In contrast, the Cuobuzha harzburgites most likely formed as a result of interactions between a cpx-harzburgite residual mantle material and MORB melts or melts with MORB to IAT compositions beneath a seafloor spreading axis.

Re-Os Constraints on Provenance and Mantle Processes

Re behaves moderately incompatibly and Os behaves compatibly during partial melting so that partial melting lowers Re/Os ratios and reduces the increase of 187Os/188Os values with time (Büchl et al., 2004). Subchondritic 187Os/188Os ratios are thus most likely the result of ancient episodes of partial melting (Büchl et al., 2004). The Cuobuzha harzburgites have subchondritic Os isotopic compositions, yielding Re-depletion model ages (TRD) ages from 1.8 to 1.7 Ga (Table 6), indicating that the Cubuzha mantle underwent at least one ancient melt extraction event ca. 1.8–1.7 Ga. The Cuobuzha mantle had therefore undergone partial melting long before the Cuobuzha ophiolitic crust was developed. It was discussed that ancient depletion events recorded by Os isotopes can be preserved in the convecting upper mantle for several billion years (Liu et al., 2012).

The cpx-harzburgites within the Cuobuzha ophiolite have chondritic or suprachondritic Os isotopic compositions, which yield future TRD ages. As the elements Re and Al behave in a moderately incompatible fashion during partial melting, the concentrations of these elements in the partial melting residue of the upper mantle are expected to correlate with each other (Zheng et al., 2009; Uysal et al., 2015). However, the Re content of our samples does not show any correlation with the Al2O3 contents, and all three cpx-harzburgite samples have elevated Re concentrations (0.380–0.575 ppb) relative to the primitive mantle (0.28 ppb; Meisel et al., 2001a, 2001b) that cannot be explained by simple melt extraction and depletion. Instead, these findings strongly suggest late-stage addition of Re into the cpx-harzburgites (Uysal et al., 2012). The cpx-harzburgites plot in an Re versus Al2O3 diagram (Fig. 5B) as a result of subsequent addition of Re following the last partial melting event that occurred during mid-oceanic ridge melt evolution processes (Uysal et al., 2015).

Büchl et al. (2002, 2004) argued that primary Os can be removed from mantle peridotites by percolating melts at high melt/rock ratios, causing primary sulfides to be scavenged by these melts, and secondary sulfides to be desegregated from the melt and deposited in the upper mantle peridotites. This process can also produce reactive peridotites with suprachondritic Pd/Ir and Re/Ir ratios (Liu et al., 2012). The two types of Cuobuzha peridotites have similar Ir-group PGE contents (Feng et al., 2016), and the cpx-harzburgites have higher Pd (4.67–8.56 ppb) and Re (0.380–0.575 ppb) values than those of the harzburgites (Pd = 1.32–5.09 ppb; Re = 0.080–0.140 ppb). The much higher Pd/Ir and Re/Ir ratios of the cpx-harzburgites suggest that the suprachondritic Os isotopic ratios within the Cuobuzha cpx-harzburgites may have resulted from melt percolation processes whereby the melts removed sulfides from the mantle rocks but transferred their Os isotopic signature into the Cuobuzha upper mantle.

The two Cuobuzha peridotite types record different mantle processes. The cpx-harzburgites were first generated by low degrees of partial melting. They subsequently underwent melt-rock interactions during which the percolating melts removed their sulfides and transferred their 187Os/188Os signature to the mantle. The harzburgites are the products of high degrees of partial melting of the cpx-harzburgites. We infer that both partial melting events and the melt-rock interactions might have taken place beneath a trench–distal backarc basin spreading center with little or no subduction influence. This interpretation is also consistent with the recent geodynamic models explaining the tectonic and magmatic evolution of the Tibetan Neo-Tethys and the YZSZ ophiolites in a regional framework (Hébert et al., 2012).

Tectonic Implications

Our results suggest that the cpx-harzburgites underwent Re addition after an initial partial melting episode of the mantle source. Therefore, we have instead used the harzburgites to evaluate the mantle source properties. In the southern subbelt of the western YZSZ, the Dongbo peridotites have 187Os/188Os values of 0.1235–0.1282 and TRD ages of 0.22–0.96 Ga (Niu et al., 2015), whereas the Purang peridotites in the same subbelt have 187Os/188Os values of 0.12228–0.12868 (Liu et al., 2012) and TRD ages of 0.1–1.0 Ga. Combining these extant data with our data from the Cuobuzha harzburgites in the northern subbelt (with lower 187Os/188Os values of 0.11624–0.11699 and older TRD ages of 1.7–1.8 Ga), we find out that the Neo-Tethyan mantle, now exposed along the YZSZ, had highly heterogeneous Os isotopic compositions (Parkinson et al., 1998; Walker et al., 2005). The older than 1.0 Ga depletion events were clearly very important for the evolution of the upper mantle beneath the northern subbelt. The modal ages younger than 1.0 Ga recorded by the peridotites in the Southern subbelt may correspond to a series of depletion events related to the opening of the Proto-Tethys and Neo-Tethys ocean basins (Uysal et al., 2012). More detailed research on the Re-Os isotope systematics of the YZSZ peridotites should reveal critical insights into better understanding of the nature and distribution of the YZSZ mantle heterogeneities beneath this suture zone and the southern Tibetan orogenic belt.


The Cuobuzha upper mantle section in the northern subbelt of the western YZSZ contains cpx-harzburgites and depleted harzburgites that collectively display textural, mineralogical, whole-rock geochemical, and isotopic evidence for two stages of partial melting and refertilization processes during the evolution of the Neo-Tethyan mantle. The cpx-harzburgites resulted from low-degree partial melting (∼5%) prior to their modification by melt percolation processes beneath a seafloor spreading center. The percolating melts removed mantle sulfides and gave these peridotites a melt-like 187Os/188Os signature. In comparison, the depleted harzburgites underwent higher degrees of partial melting. They represent residual peridotites after remelting of the cpx-harzburgites and their interactions with MORB- or IAT-type melts. Comparison of the upper mantle peridotites in the two subbelts of the western YZSZ suggests that the Neo-Tethyan mantle had highly heterogeneous Os isotopic compositions.


This research was funded by grants from the China Geological Survey (DD20160023-01 and DD20160022-01), the Ministry of Science and Technology of China (Sinoprobe-05-02, 201511022), the National Science Foundation of China (41303019, 41373028, 41541017, 41573022, and 41641015), and the Basic Outlay of Scientific Research Work from the Ministry of Science and Technology of China (J1321). We thank Editor Kurt Stuewe for his editorial help and consideration, and Ibrahim Uysal and Jingen Dai for their critical and constructive reviews. We also thank He Rong of the Institute of Geology, Chinese Academy of Geological Sciences, for his assistance with electron microprobe analyses, and Shengmin Lai, Jian Gao, and Jie Li for their assistance with Re-Os isotopic analysis. Dilek’s field work within the Yarlung-Zangbo suture zone in southern Tibet has been funded by research grants from the Chinese Academy of Geological Sciences.

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.