Lithos 220–223 (2015) 164–178

Contents lists available at ScienceDirect

Lithos

j ourna l homepage: www.e lsev ie r .com/ locate / l i thos

Petrogenesis of nephelinites from the Tarim Large Igneous Province,NW China: Implications for mantle source characteristics andplume–lithosphere interaction

Zhiguo Cheng a, Zhaochong Zhang a,⁎, Tong Hou a, M. Santosh a,b, Dongyang Zhang a, Shan Ke a

a State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences, Beijing 100083, Chinab Division of Interdisciplinary Science, Kochi University, Kochi 780-8520, Japan

⁎ Corresponding author. Tel.: +86 13910168892; fax:E-mail address: zczhang@cugb.edu.cn (Z. Zhang).

http://dx.doi.org/10.1016/j.lithos.2015.02.0020024-4937/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 November 2014Accepted 2 February 2015Available online 11 February 2015

Keywords:Tarim large igneous provinceMantle plumeNepheliniteRutile geochronologyMagnesium isotopePlume-lithosphere interaction

The nephelinite exposed in the Wajilitage area in the northwestern margin of the Tarim large igneous province(TLIP), Xinjiang, NW China display porphyritic textures with clinopyroxene, nepheline and olivine as the majorphenocryst phases, together with minor apatite, sodalite and alkali feldspar. The groundmass typically has cryp-tocrystalline texture and is composed of crystallites of clinopyroxene, nepheline, Fe-Ti oxides, sodalite, apatite,rutile, biotite, amphibole and alkali feldspar. We report rutile SIMS U-Pb age of 268 ± 30 Ma suggesting thatthe nephelinite may represent the last phase of the TLIP magmatism, which is also confirmed by the fieldrelation. The nephelinite shows depleted Sr-Nd isotopic compositions with age-corrected 87Sr/86Sr andεNd(t) values of 0.70348–0.70371 and +3.28 to +3.88 respectively indicating asthenospheric mantle source.Based on the reconstructed primary melt composition, the depth of magma generation is estimated as 115–140 kmand the temperatures ofmantlemelting as 1540–1575 °C. The hotter than normal asthenosphericmantletemperature suggests the involvement ofmantle thermal plume. TheMg isotope values display a limited range ofδ26Mg from −0.35 to −0.55‰, which are lower than the mantle values (−0.25‰). The Mg isotopic composi-tions, combinedwith the Sr-Nd isotopes andmajor and trace element data suggest that theWajilitage nephelin-ite was most likely generated by low-degree partial melting of the hybridized carbonated peridotite/eclogitesource, which we correlate withmetasomatism by subducted carbonateswithin the early-middle Paleozoic con-vergent regime. A plume-lithosphere model is proposed with slight thinning of the lithosphere and variabledepth and degree of melting of the carbonated mantle during the plume-lithosphere interaction. This modelalso accounts for the variation in lithology of the TLIP.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Several important large igneous provinces (LIPs) have been identi-fied in relation to the disruption of the global supercontinent Pangeaincluding the Siberian (~251 Ma), Emeishan (~260 Ma), Tarim (~280–300 Ma) and Panjal (~290 Ma) LIPs (Fig. 1a; Kamo et al., 2003;Shellnutt, 2014; Shellnutt et. al., 2014; Xu et al., 2014; Zhou et al.,2002; Zhang et al., 2013). Many LIPs are produced within a fewmillionyears, e.g. Siberian Traps and Emeishan LIP. However, compared withthe other typical LIPs, previous petrological studies of the TLIP (TarimLIP) have documented a long duration of magmatism (erupted from~300 Ma to ~280 Ma) and a wide range of magmatic rock suites withdistinct geochemical features (Z.L. Li et al., 2011; Li, 2013; Xu et al.,2014; Zhang et al., 2008; Zhang et al., 2013). The longevity ofmagmatism andwide variety in lithology implies a differentmelt gener-ation mechanism, and thus detailed research on the TLIP may provide

+86 82322195.

important insights into the mechanism of formation of large igneousprovinces. Previous studies had suggested that syenites in the TLIPmark the last phase of magmatism (Yang et al., 2006). Our recent de-tailed field investigations in the Wajilitage area located in the north-western margin of the TLIP led to the discovery of minor volumes ofnephelinite carrying xenoliths of syenite. This finding adds a new lithol-ogy to the already complex sequence in the TLIP and opens the possibil-ity that the duration of the magmatism could be even longer thanpreviously thought. Typically, the LIPs are interpreted to be the productsof the interaction of mantle plume with lithospheric mantle (e.g.Campbell and Griffiths, 1990). If this is the case, the relationship be-tween the complex lithology in the TLIP and the mantle plume-lithosphere interaction are of prime importance.

Nephelinites are generally rare on the Earth (less than 1%). Becausethey represent small volume deep mantle-derived magma, and havenot been significantly contaminated by crustal materials due to therapid ascent from the mantle to surface, these rocks provide importantinformation on the nature of the mantle source, especially that of thedeep mantle. Thus the Wajilitage nephelinite can be used to constrain

Fig. 1. (a) Distribution of the Permian floodbasalts in Asia and the location of Siberian, TarimandEmeishan large igneous provinces (modified after Z.L. Li et al., 2011). (b) Sketch geologicalmap of the Tarim Basin showing the distribution of the Permian continental flood basalts (after Tian et al., 2010). Abbreviation: KD= Kuche depression; NTU = northern Tarim uplift;NTD= northern Tarim depression; CTU= central Tarim uplift; SWD= southwestern depression. (c) Simplified geological map of theWajilitage area showing the location of the neph-elinite lavas (modified after XBGMR, 1984).

165Z. Cheng et al. / Lithos 220–223 (2015) 164–178

the duration of the TLIP, and also to trace the mantle source composi-tions and the plume-lithosphere interaction. In this contribution, wereport the rutile U-Pb age, major and trace element geochemistry aswell as Mg-Sr-Nd isotopic data for the Wajilitage nephelinite. Basedon the results, we attempt to characterize the mantle source composi-tions and melting conditions, which in turn have critical implicationsfor the geodynamic evolution.

2. Geological setting

The Tarim Craton (TC) is among the three major cratons in China;the other two being the North China and Yangtze Cratons (Fig. 1a).The TC is predominantly composed of a complex Precambrian crystal-line basement with a thick Phanerozoic sedimentary cover (C.L. Zhanget al., 2010; Cheng et al., 2014; Tian et al., 2010; Xu et al., 2014; Zhanget al., 2013). Silurian to Devonian arc rocks were found along the north-ern margin of the TC which were correlated to the southward subduc-tion of the southern Tianshan ocean (440 to 360 Ma; e.g. Ge et al.,2012). Recent studies have revealed that a major Permian thermalevent occurred in this region (Fig. 1b; e.g., Tian et al., 2010; Yu et al.,2011; Zhang et al., 2013; Zhou et al., 2009). As most of the surface ofthe TC is buried by the TaklamakanDesert, the informationof the Permianigneous units are mainly from drill core and seismic data, together withseveral exposures along the margins of the Tarim Basin, such as Bachu,Keping and Puchang County. Systematic studies indicate that the Tarimigneous rocks are distributed within an area of 250,000–300,000 km2

with an average thickness of 600 m (Fig. 1b; e.g. Xu et al., 2014), termedas Tarim large igneous province (Zhang et al., 2008; C.L. Zhang et al.,

2010). The TLIP is considered as the second major LIP after the EmeishanLIP in China (e.g. Bryan and Ernst, 2008; C.L. Zhang et al., 2010; Tian et al.,2010; Xu et al., 2014).

The volcanic rocks in the TLIP consist mainly of flood basalts, withsmall volume of rhyolite, picrite and tuff (Chen et al., 2009; Li et al.,2012; Tian et al., 2010; Z.L. Li et al., 2011). Along the northwesternpart of the TC, synchronous magmatic suites comprising small volumeof A-type granites and mafic-ultramafic intrusions and dykes swarmshave been also been recognized (C.L. Zhang et al., 2010; Cao et al.,2014; Yu et al., 2011; Zhang et al., 2008). The flood basalts lie over theLate Carboniferous Kangkelin Formation and are in turn covered bythe Late Permian Shajingzi Formation. Hence, based on the stratigraphiccorrelation, the flood basalts are suggested to represent Early Permianmagmatism. Stratigraphically, the Tarim flood basalts are intercalatedwith the Kupukuziman Formation and Kaipaizileike Formation, whichinclude two and six volcanic-sedimentary cycles respectively (Xuet al., 2014; Zhou et al., 2009). Based on geochronological data, Xuet al. (2014) proposed that the Tarim magmatism can be divided intothree major phases: 1) small volume of ~300 Ma kimberlitic rocks,which have been interpreted to mark the onset of plume-inducedmagmatism in the TLIP (Zhang et al., 2013); 2) ~290 Ma voluminousflood basalts and small volume of rhyolites with a bimodal nature, andmainly occurring in the interior of the Tarim Basin (e.g. Jiang et al.,2004; Tian et al., 2010; Zhou et al., 2009); and 3) ~280Ma small volumeof A-type granites and mafic-ultramafic intrusions and dykes swarms(Cao et al., 2014; Y.T. Zhang et al., 2010; Yu et al., 2011; Zhang andZou, 2013a,b; Zhang et al., 2008). Furthermore, from the early pulse offlood basalts to the later emplacement of the intrusions, the Sr-Nd-Hf

166 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

isotopes studies suggest that the source ranged from an enrichedmantle (87Sr/86Sr N 0.705, εNd(t) b +1, εHf(t) b +2.5) to depletedmantle (87Sr/86Sr b 0.705, εNd(t) N +1, εHf(t) N +2.5; Li et al., 2012;Xu et al., 2014).

TheWajilitage region is located in the northwestern part of the TLIP(Fig. 1b). Here, within a relatively small region (b10 km2), diverse rocktypes are exposed including kimberlitic rocks, mafic-ultramafic intru-sions, diabase dikes, lamprophyre, nepheline syenite, nephelinite andcarbonatite (Fig. 1c; Li et al., 2012; Zhang et al., 2008; Zhang et al.,2013). These rocks intrude weakly metamorphosed Devonian clasticsequences, which are composed of the Keziletag Formation andYimugangawu Formation frombottomupwards. According to field inves-tigations and available geochronological data, the sequence of formationof the igneous units in the Wajilitage has been identified as: kimberliticrocks → mafic-ultramafic intrusions → carbonatites → diabasedikes → lamprophyre and nepheline syenites (Yu, 2009; Zhang et al.,2013). Our recent field studies led to the identification of small volumesof nephelinite lavas in Wajilitage, which occupy an area of ~1 km2 withan average thickness of ~5 m (Fig. 1c). The nephelinite has erupted ontop of Devonian strata and the mafic-ultramafic intrusions (gabbro;Fig. 2a). Minor nepheline syenite xenoliths have been recognized in thenephelinite (Fig. 2c–d), which are composed of the orthoclase (55–75%), nepheline (5–10%) andminor clinopyroxene. The nepheline syenitexenoliths obviously indicate that the nephelinite was erupted followingthe syenite emplacement, and hence may represent the last phase ofmagmatism of the TLIP.

3. Petrography

The nephelinite is black in hand specimen and exhibits porphyritictexture with varying contents of phenocrysts (Figs. 2b and 3a–b). Thephenocrysts are generally dominated by clinopyroxene (30–50%),nepheline (20–40%) and olivine (5–10%) together with minor apatite,sodalite and alkali feldspar. The clinopyroxenes occur as euhedral tabu-lar and prismatic phenocrysts, varying from 0.4 × 1 to 1 × 4 mm in di-ameter (Fig. 3a and b). Some clinopyroxene grains are zoned or havereaction rims with interstitial amphibole and magnetite (Fig. 3c and

Fig. 2. (a) Photograph showing the nephelinite lavas overlying the upper Devonian Keziletag Foxenoliths in the nephelinite.

d). Our petrographic observations show that the clinopyroxene isenclosed by the magnetite, nepheline and alkali-feldspars. Nephelinephenocrysts occur as square or tabular grains and vary from 0.5 to1 mm in diameter (Fig. 3a and b). Olivine is present as subhedral toeuhedral phenocrysts ranging in size from 0.5 to 1.5 mm (Fig. 3a), andare sometimes enclosed by clinopyroxene. The sodalite and apatite arecommonly subhedral to euhedral grains, ~0.5 mm across (Fig. 3f andg). Occasionally, apatite is enclosed by nepheline. The alkali feldsparsoccur as phenocrysts with 0.1 × 0.2 to 0.5 × 1 mm in diameter. Petro-graphic and textural observations suggest that olivine, pyroxene,magne-tite, apatite, nepheline and alkali-feldspar crystallized consecutively.

Minor (1%) mineral aggregates occurring in the groundmass of thenephelinite vary from 1 to 2mm in size, andmost have ellipsoidal or ir-regular shapes. They are composed predominantly of alkali pyroxene(30%), Fe-Ti oxide (10%) and alkali feldspars (60%). The alkali pyroxenesare greenish aegirine-augite. They are present as subhedral grains vary-ing from 0.1 to 1mm in diameter (Fig. 3i). The Fe-Ti oxides occur as dis-crete grains in the aggregate. Alkali feldspars are anhedral and occur asan interstitial phase in the aggregates. The groundmass of the nephelin-ite is typically composed of crystallite of clinopyroxene, nepheline, Fe-Tioxides, sodalite, apatite, rutile, biotite, amphibole and alkali feldspar(Fig. 3c, e–f and h–i). The clinopyroxene in the groundmass occurs ascrystallite, 0.2 mm across. Minor amounts of small nepheline grainsalso occur in the groundmass. The Fe-Ti oxides (0.02–0.5 mm) occuras discrete grains in the groundmass (Fig. 3b–c and f). Biotite and am-phibole are very small in size (~0.02 mm) and typically occur in thegroundmass (Fig. 3e and f). Rutile grains are mostly subhedral or frag-mental and are ~0.1 mm across (Fig. 3h). Some alkali feldspars areanhedral and occur as an interstitial phase in the groundmass (Fig. 3i).

Xenoliths of nepheline syenite and carbonate-bearing rocks occurwith the nephelinite lavas (Figs. 2d and 4). The proportions of the xeno-liths are less than 1%. The syenite xenoliths vary in size from 0.5 mm to2 cm and have sharply angular shapes with clear reaction rims (Fig. 4aand b). The syenite xenoliths are typically holocrystalline and medium-grained, and are dominantly composed of orthoclase (55–75%), nephe-line (5–10%) and clinopyroxene (15–20%), with minor biotite (5%) andmagnetite (1%). They resemble the syenite intrusion surrounding the

rmation. (b) Field photograph showing the nephelinite lavas. (c) and (d) Nepheline syenite

Fig. 3. Photomicrographs of the nephelinite. (a) Porphyritic texture, crossed-polarized light. (b) Plesiophyric texture, crossed-polarized light. (c) BSE image of the zoned clinopyroxene.(d) BSE of the image Clinopyroxene with the reaction rim of amphibole and magnetite. (e) Euhedral biotite; plane-polarized light. (f) Amphibole and clinopyroxene phenocrystals,plane-polarized light. (g) Ehhedral sodalite phenocrystal, plane-polarized light. (h) Rutile, clinopyroxene and nepheline in the groundmass of the nephelinite, plane-polarized light.(i) The aggregates composed predominantly of alkali feldspars, Fe-Ti oxide and clinopyroxene, plane-polarized light. Abbreviations: Ap = apatite, Cpx = clinopyroxene, Bt = Biotite,Ol = olivine, Am= amphibole, Kfs = alkali feldspar, Sdl = sodalite, Mt = magnetite, Ne = nepheline, Ab = albite, Rt = rutile, Agt = aegirine-augite.

167Z. Cheng et al. / Lithos 220–223 (2015) 164–178

nephelinite lavas (Figs. 1c and 4c–d). The carbonate-bearing xenolithsvary from 2 × 3 cm−1 × 3.2 cm in diameter (Fig. 4e) and show angularor rounded shapes and sharp contact with the host rocks. Thecarbonate-bearing xenoliths show cryptocrystalline texture and com-prise dominantly of albite crystallites (45%), clinopyroxene (30%), calcite(15%) and apatite (10%; Fig. 4f).

4. Analytical methods

To avoid the influence of the syenite xenoliths, the samples werecrushed to 80 mesh grains by a jaw crusher initially. Any visible frag-ments of the syenite xenoliths were removed under a binocular micro-scope and the pure fraction was powdered for the major and traceelement and isotopic analysis. Major element compositions of the min-eralswere analyzed on the polished carbon-coated thin-sections using aJXA-8230 electron microprobe with the wavelength dispersive (WDS)technique at the Institute of Mineral Resources, Chinese Academy ofGeological Sciences, Beijing.Whole-rockmajor and trace elements anal-ysis were performed at the National Research Centre for Geoanalysis,Beijing. Major element analyses were done by X-ray fluorescence spec-trometer following Norrish and Chappell (1977). The FeO content wasanalyzed using conventional wet chemical methods. Trace elements

were performed by ICP-MS with an X-series instrument, using the pro-cedure described by Qi et al. (2000). Secondary ion mass spectroscopy(SIMS) rutile U-Pb analyses were determined using the Cameca IMS-1280 SIMS at the Institute of Geology and Geophysics, Chinese Academyof Sciences. Details of the instrument setting and analytical procedureare described in Q.L. Li et al. (2011). Isotope ratios of Sr and Nd ofwhole rocks were measured using a VG354 mass spectrometer at theCenter of Materials Analysis, Nanjing University (Zhang et al., 2012).Mg isotopic analysis was carried out using MC-ICP-MS at the State KeyLaboratory of Geological Processes andMineral Resources, ChinaUniver-sity of Geosciences, Beijing, following Teng et al. (2007). During the anal-ysis, the δ26Mg values of standards BCR and BHVO are −0.22 ± 0.02‰and −0.25 ± 0.04‰, which are in good agreement with the standardvalues (−0.20± 0.06‰ and 0.26 ± 0.06‰). The δ26Mg values of repeatsample of the nephelinite (DW-18-14) are−0.37 ± 0.04 and−0.39 ±0.03‰ respectively, which coincide with each other well within the an-alytical uncertainties. Moreover, the Mg content of the blank sample isonly 7.56 ng below the laboratory average standard value (10 ng). In ad-dition, the Mg isotopes of the kimberlitic rocks in the Wajilitage werealso analyzed for comparison. General descriptions of the kimberliticrocks and the sample preparation procedures are presented by Zhanget al. (2013) and Cheng et al. (2014) respectively. Only the fresh

Fig. 4. (a) and (b) Photomicrographs of the nepheline syenite xenoliths in the nephelinite lava. The boundary between the xenoliths and nephelinite lava ismarked by dotted line; plane-polarized light (a) and cross- polarized light (b). (c) and (d) Photomicrographs of the nepheline syenite intrusion in theWajilitage area. plane-polarized light (c) and cross-polarized light(d). (e) The carbonate-bearing xenoliths in the nephelinite; plane-polarized light. (f) BSE image of the carbonate-bearing xenoliths, composed of the albite, clinopyroxene, calcite andapatite. Abbreviations: Cc = calcite, Ap = apatite, Cpx = clinopyroxene, Kfs = alkali feldspar, Mt = magnetite, Ne = nepheline, Ab = albite.

168 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

samples were selected and the xenoliths were entirely removed. Theδ26Mg values of repeat sample of the kimberlitic rocks (K-1) are−0.52 ± 0.03 and−0.48 ± 0.03‰ respectively.

5. Results

5.1. Mineral chemistry

5.1.1. ClinopyroxeneClinopyroxene occurs as phenocrysts, and as crystallites in the

groundmass and as major minerals in the aggregates. Clinopyroxenephenocrysts in the nephelinite exhibit restricted compositions ofWo45.7–48.8En34.6–39.5Fs9.52–13.8 (Supplementary Table A.1 in the Elec-tronic Supplementary File; Fig. 5a) and have Mg# [100 × molarMg2+/(Mg2++Fe2+)] ranges from 73.2 to 81.0 with an averagevalue of 76.6. TiO2 and Al2O3 contents are relatively high and varyfrom 1.84 to 4.26% and 2.09 to 6.80% respectively. Na2O contentsvary from 0.53 to 1.39%. The clinopyroxene has higher Mg#, TiO2,Al2O3 and Na2O contents compared with those in the Tarim flood ba-salts (Li et al., 2008; Zhou et al., 2009). Some phenocrysts are zonedwith cores of Wo46.1–48.8En38.0.3–39.5Fs9.52–12.1 and rims ofWo43.7–45.2En38.3–42.6Fs10.3–12.4 (Fig. 5a). The Mg# in the rims (76.2–80.6) are lower than that in the cores (76.5–81.0). According to the

pyroxene thermometer and barometer proposed by Putirka et al.(2003), the estimated temperature and pressure for crystallizationare 1502–1563 °C and 18–22 kbar. Compositions of the clinopyroxenegrains in the groundmass are Wo43.4–48.3En33.5–37.6Fs10.7–16.5 (Fig. 5a).They have relatively lower Mg# (67.7–78.3) and higher Na2O contents(0.91–1.66 wt.%) than the phenocrysts.

The results show that the pyroxenes in the aggregates are alkalipyroxene and have high Na2O (6.98–12.55 wt.%) and TiO2 (1.71–5.49 wt.%) contents. Compositionally, they are aegirine augite withappreciable aegirine component (with 28.2 to 51.6 wt.% aegirine endmember).

5.1.2. NephelineThe nepheline phenocrysts have 9.18–15.62 wt.% Na2O, 32.05–

36.39 wt% Al2O3, 5.46–7.00 wt.% K2O with minor 0.58–1.90 wt.% FeO(Supplementary Table A.2). The nephelines in the groundmass havehigher Na2O contents (15.17–16.42 wt.%), and lower Al2O3 (30.76–33.81 wt.%) and K2O (3.89–6.79 wt.%).

5.1.3. OlivineOlivine grains exhibit a wide compositional variation with the Fo

(Fo = Mg2+/(Mg2++Fe2+total)) ranging from 67.8 to 85.7 (Supple-mentary Table A.3). The cores of olivine phenocrysts are usually Mg-

Fig. 5. (a) En-Wo-Fs diagram of clinopyroxene (modified after Bao et al., 2009). The data of the clinopyroxene in the Tarim flood basalts are from Li et al. (2008) and Zhou et al. (2009).(b) Classification diagram of feldspar. (c) Classification diagram of biotite (after Foster, 1960). (d) Classification diagram of amphibole (after Leake et al., 1997).

169Z. Cheng et al. / Lithos 220–223 (2015) 164–178

rich (Fo79.5–85.7), whereas the forsterite contents decrease to Fo78–79.4and Fo67.8–79.3 in the rims and crystals in the groundmass, respectively.CaO contents range from 0.08 to 0.31 (most of the values N0.2%) suggest-ing themagmatic nature of the olivine (Thompson andGibson, 2000). Ac-cording to the olivine thermometer proposed by Putirka et al. (2003), theestimated temperature for crystallization are1502–1535 °C, close to thoseestimated from clinopyroxene as above.

5.1.4. Alkali feldsparsAlkali feldspars occur as minor phenocrysts and as an interstitial

phase in the aggregate. The phenocrysts are mainly orthoclase andNa-orthoclase with compositions of Ab10.79–39.83An0.00–3.20Or56.97–89.21.The alkali feldspars in the aggregates are albite and exhibit composi-tions of Ab95.43–98.72An0.51–2.15Or0.71–3.16 (Supplementary Table A.4;Fig. 5b).

5.1.5. Fe-Ti oxidesThe Fe-Ti oxides occur as crystallites in the groundmass and asmajor

minerals in the aggregates. In the groundmass, the Fe-Ti oxides aretitanomagnetite, containing 70.56–87.36 wt.% FeO, 7.13–17.52 wt.%TiO2, 0.11–5.01 wt.% Al2O3 and 0.52–1.52 wt.% MnO (SupplementaryTable A.5). The oxides in the aggregates commonly have higher TiO2

(20.50 wt.%), and sometimes ilmenite occurs with 49.30 wt.% TiO2,47.82 wt.% FeO, 2.13 wt.% MnO and 0.10 wt.% MgO.

5.1.6. ApatiteApatite shows restricted compositional variation for P2O5 (40.04–

40.23wt.%) and CaO (54.68–54.75wt.%) concentrations (Supplementa-ry Table A.6). All of the apatites are fluorapatiteswith F contents rangingfrom 3.54 to 3.59 wt.% and Cl concentrations b0.16 wt.%. The otherelements (e.g. Si, Fe and Na) are lower than detection limit.

5.1.7. BiotiteBiotites occur as minor constituents in the groundmass. They are

mostly magnesian-biotite (Supplementary Table A.7; Fig. 5c) with

compositions of 3.90–9.15 wt.% TiO2, 12.12–17.49 wt.% FeO, 10.34–16.81 wt.% MgO and 8.22–9.52 wt.% K2O. The Mg# varies from 0.51 to0.70.

5.1.8. AmphiboleAmphiboles occur in the groundmass, and are ferrokaersutite

(Supplementary Table A.8; Fig. 5d) with compositions of 5.58–5.71 wt.%TiO2,11.76–14.27 wt.% FeOt, 10.29–10.97 wt.% MgO, 10.61–14.47 wt.%CaO, 2.38–3.36 wt.% Na2O and 0.73–1.37 wt.% K2O. The Mg# is low andvaries from 0.56 to 0.62.

5.1.9. SodaliteSodalite is an accessory mineral in the nephelinite. The mineral

shows high Cl content of 7.60–7.97 wt.%, 40.29–43.27 wt.% for SiO2,33.64–35.46 wt.% for Al2O3, 13.55–16.36 wt.% for Na2O (SupplementaryTable A.9).

In summary, the nephelinite is mainly composed of clinopyroxene,nepheline and olivine, and is characterized by the presence of alkaliminerals. The amphibole and biotite are Ti-rich, whereas the sodalite,albite and aegirine-augite are Na-rich minerals. Moreover, the mineralsin the groundmass or aggregates commonly have higher Na and Ticontents than the phenocrysts. These compositional characteristicscorrespond to themagma compositions, and suggest that the formationof nephelinite was related to Ti and Na-rich magma.

5.2. Rutile U-Pb geochronology

Rutile grains analyzed in this study are mostly subhedral or frag-mental and range in size from30 to 100 μm(Fig. 6a). Twenty-five grainswere analysed for U-Pb age, and the data are reported in SupplementaryTable A.10. The results show low U concentrations varying from 0.046–0.880 ppm. The Tera-Wasserburg plot yields a lower intercept age at248 ± 39 Ma and an upper intercept with 207Pb/206Pb = 0.80 ± 0.04for the common-Pb content (Fig. 6c). Individual 206Pb/238U agescorrected by the 207Pb-based common-Pb yield a weighted mean

Fig. 6. (a) Representative CL images of rutile grains from the nephelinite. White circles show locations of the analyzed spots with numbers in the circles indicating spot numbers. The cal-culated ages are also given. (b) The weighted average 206Pb/238U age for the nephelinite lavas. (c) Concordia diagram of rutile from the nephelinite lavas. Error bars are 2σ.

170 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

206Pb/238U age of 268 ± 30 Ma (MSWD= 0.89; Fig. 6b), which is closeto that of the other igneous rocks in the Wajilitage area.

5.3. Major and trace element data

Twenty representative samples of the Wajilitage nephelinite wereanalyzed and the major element compositions are presented in Supple-mentary Table A.11. All of the rocks selected for geochemical analysesare petrographically fresh with no significant hydrothermal alteration,as also reflected by their low loss-on-ignition (LOI), ranging from 0.38to 1.73 wt.% (mostly b1.5 wt.%). The nephelinite samples display SiO2

abundances from 38.65 to 40.54 wt.%, 12.10–14.90 wt.% for FeOt, 5.24–9.41 wt.% for MgO, 10.74–14.47 wt.% Al2O3, 9.69–12.19 wt.% for CaO,3.87–5.81 wt.% for TiO2, 2.12–2.85 wt.% for K2O and 3.61–7.56 wt.%for Na2O. The samples are sodic, as suggested by the low K2O/(Na2O + 0.7K2O) ratios of 0.26–0.51. The CaO/Al2O3 ratios are high(0.72–1.13), which indicates a relatively high melting pressure (Baker

Fig. 7. (a) TAS diagrams for the Wajilitage nephelinite (Le Maiitre, 2002

and Stolper, 1994). The MgO contents show a wide range and the Mg#

varies from 43.8–55.3. The nephelinite is strongly SiO2-undersaturatedwith high total alkalis (Na2O+ K2O= 6.46–9.92 wt.%; Fig. 7a). Accord-ing to the IUGS classification (Le Bas, 1989), the samples classify asnephelinite (Fig. 7b). As shown in Fig. 8, Al2O3, Na2O, K2O and P2O5 con-tents increase whereas total FeO, CaO and TiO2 contents decrease withdecreasing MgO. These trends are in accordance with the fractionalcrystallization, such as olivine, clinopyroxene and Fe-Ti oxides as wellas apatite.

Trace element data are listed in Supplementary Table A.12. Likeother peralkaline rocks elsewhere in the world, the rocks in our studyare characterized by extreme enrichment in incompatible elementswith high total rare earth element (REE) content s of 684–1081 ppm.In the chondrite normalized REE diagrams, the samples show lightREE-enriched patterns [(La/Yb)N = 23–30] with no Eu or Ce anomalies(Fig. 9a). In the primitivemantle normalized patterns (Fig. 9b), the sam-ples show pronounced positive anomalies in Nb, Ta and negative

). (b) Nephelinite-melilitite-basanite classification of Le Bas (1989).

Fig. 8.Major element variation diagrams for the Wajilitage nephelinite.

171Z. Cheng et al. / Lithos 220–223 (2015) 164–178

anomalies in K, Pb, P, Ti, Zr andHf. However, they have subchondritic Zr/Hf ratios (38.90–57.99). Both the REE and trace element spidergrams re-semble those of oceanic island basalts (OIB) in terms of enrichment inLREE, positive anomalies in Nb and Ta, and negative anomalies in K, Pand Ti. Compared with the Tarim flood basalts, the incompatible ele-ments in the nephelinite are much higher (Fig. 9a). In the trace elementpatterns, the nephelinite exhibit negative K, Pb and P anomalies, where-as the Tarim basalts have apparently positive Ba, Pb and P anomalies(Fig. 9b).

5.4. Sr and Nd isotope

Sr and Nd isotopic ratios are listed in Supplementary Table A.13 andshown in Fig. 10 where the data are plotted in conjunction withthe those of other silicate rocks from the TLIP for comparison. The(87Sr/86Sr)t (t = 270 Ma) isotopic compositions of the samples varyfrom 0.70348 to 0.70371 and (143Nd/144Nd)t from 0.51245 to 0.51248with εNd(t) values of +3.28 to +3.88, overlapping the Sr-Nd isotopic

Fig. 9. (a) Chondrite-normalized REE patterns of the Wajilitage nephelinite. (b) Primitive manvalues are from Sun and McDonough (1989) and McDonough and Sun (1995), respectively.

compositions of OIB (Fig. 10). In comparison with the Tarim floodbasalts, the Wajilitage nephelinite shows much lower (87Sr/86Sr)t andhigher εNd(t) values (Fig. 10).

5.5. Magnesium isotope

Magnesium isotopic compositions are presented in SupplementaryTable A.14 and shown in Fig. 11. The results show that the nephelinitehas homogenous δ26Mg values ranging from −0.35 to −0.55‰ withan average of −0.41 ± 0.05‰ (2SD, n = 10; Fig. 11a), which arelower than the average δ26Mg values of the mantle (−0.25 ± 0.07‰;Teng et al., 2007, 2010). The δ26Mg values of the kimberlitic rocks ofthe Wajilitage are much lower, from −0.36 to −0.75‰ (2SD, n = 10;Fig. 11b), than those of the nephelinite. A comparison of the resultsshow similarity with continental basalts (b110 Ma) from the NorthChina Craton,with δ26Mg values varying from−0.42 to 0.60‰ (average−0.46 ± 0.1‰).

tle-normalized trace element patterns of the nephelinite. Primitive mantle and chondrite

Fig. 10. 87Sr/86Sr versus εNd(t) diagram. Data of MORB and OIB is from Zindler and Hart(1986); Kimberlitic rocks and their hosted xenoliths from Jiang et al. (2004) and Zhanget al. (2013); Tarim flood basalts from Li et al. (2012), Tian et al. (2010), Yu et al. (2011);C.L. Zhang et al. (2010); Y.T. Zhang et al. (2010) and Zhou et al. (2009); Wajilitage andPuchang layered mafic-ultramafic intrusions in TLIP from Zhang et al. (2014, in revision);LateNeoarcheanandearly Paleoproterozoic basement fromLonget al. (2011) and referencestherein.

172 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

6. Discussion

As all the selected samples for analyses are fresh, their geochemicalfeatures are considered to reflect those of the original magmatic rocks.In this section, we employ these data to discuss the petrogenesis andgeodynamic implication.

6.1. Crustal contamination and fractional crystallization

Although some syenite xenoliths are recognized in the nephelinite,they are rare, and no alteration has been identified in the xenoliths, sug-gesting that the compositions of these xenoliths have not been affectedby theWajilitage nephelinites. In addition, the geochemical characteris-tics of the Wajilitage nephelinite suggest that crustal contamination isinsignificant. For example, the incompatible elemental ratios such asZr/Nb = 3.75–4.49, Ba/Nb = 6.13–6.53 and Ba/La = 7.19–7.84, aremuch lower than those of continental crust (Zr/Nb = 16.2, Ba/Nb =54, Ba/La = 25; Hofmann et al., 1986), but similar to those of HIMU-OIB (Zr/Nb = 3.2–5, Ba/Nb = 4.9–6.9, Ba/La = 6.8–8.7; Weaver,1991). Furthermore, the trace element patterns resemble those of OIB,including the positive Nb and Ta anomalies and negative K and Pbanomalies, but are clearly distinct from those of continental crust (neg-ative Nb and Ta anomalies and positive K and Pb anomalies). Finally, the(87Sr/86Sr)t values (0.70348–0.70371) and εNd(t) (+3.28−+3.88) are

Fig. 11. (a) δ26Mg versus MgO (wt.%) for the Wajilitage nephelinite. (b) δ26Mg versus MgO (wstand for the δ26Mg values of replicate samples. The green symbols stand for the results of thenormal mantle values.

depleted compared with the continental crust and show no observablecorrelation with SiO2 contents.

The low Mg# (43.8–55.3), Ni (29.6–224 ppm) and Cr (25.5–237 ppm) contents of the Wajilitage nephelinite are consistent withthose of evolved magmas that underwent fractional crystallization. Asshown in Fig. 11, the major element contents (e.g. FeOt, Al2O3, CaO/Al2O3) show obvious correlations with MgO, which is consistent withthe fractionation of olivine, clinopyroxene and nepheline. These obser-vations, along with the high amounts of the clinopyroxene, olivineand nepheline phenocrysts indicate a high degree of fractionalcrystallization.

In summary, the Wajilitage nephelinites have not experiencedsignificant crustal contamination, but have been considerably affectedby fractional crystallization.

6.2. Original melt composition and melting conditions

Based on geochemistry and the available data from experimentalstudies, the formation of the nephelinite magma is attributed to the fol-lowing three factors (Ali et al., 2013; Vapnik et al., 2007): (1) lowdegreepartial melting; (2) high pressure; (3) CO2-bearing or metasomatizedmantle source. However, whether the natural samples could representthe primarymagma remains controversial. Based on the highMgO con-centrations (N8%), some researchers suggested that the nephelinite andthe associated alkaline basalts could represent the primary melts (Aliet al., 2013; Zeng et al., 2010). Furthermore, these studies also attributedthe varied magma compositions to different degrees of mantle melting.With increasing degree of melting, the melts vary from carbonatite,melilitite, nephelinite to alkaline basalts (e.g. Sisson et al., 2009;Vapnik et al., 2007). However, some workers do not agree with theviewpoint that nephelinite represents primary melts, but considerthem as the representative of evolved magmas by fractional crystalliza-tion (Barker et al., 2012; Klaudius and Keller, 2006) or a result of liquidimmiscibility with carbonatite (Mourão et al., 2010) in a CO2-rich sili-catemelt. However, most of the studies do not specify the compositionsof the CO2-rich silicate melt, with the exception of Klaudius and Keller(2006) and Keller et al. (2006)who proposed that the olivinemelilititesat Oldoinyo Lengai, Tanzania, may be candidates for primary melt com-positions of the nephelinite.

Thus, the Wajilitage nephelinite magma may not represent theprimary magma and might be a product of fractional crystallization(Tatsumi and Saknyama, 1983). In order to discuss the depths andtemperatures of mantle melting, the primary magma must be recon-structed (Herzberg et al., 2007). Previous studies reconstructed theprimary magmas by addition or subtraction of olivine, such as olivine-melt solid–liquid equilibrium (Zhang et al., 2006), addition of olivine(Herzberg and Asimow, 2008), and Ni content in olivine (Korenaga

t.%) for the Wajilitage kimberlitic rocks. Error bars are 2SD uncertainties. The red symbolsnephelinite and the kimberlitic rocks respectively. The grey field stands for the δ26Mg of

173Z. Cheng et al. / Lithos 220–223 (2015) 164–178

and Kelemen, 2000). However, these procedures cannot be directlyadopted in the casewhere themagmaunderwent high degree of pyrox-ene fractionation. In the Wajilitage nephelinite, the most magnesianolivine is Fo85.8. According to Zhang andWang (2003), MgO (melt) cor-relates with FeO (melt) when the Fo (olivine) content is in equilibriumwith the melt, according to the following formula:

ω MgO; meltð Þ ¼ 0:56095 Kd � Fo� 1‐Foð Þ � ω FeO; meltð Þ;

Where Kd is the partition coefficient of the Fe2+-Mg between theolivine and themelt, ranging from 0.30 to 0.33with the increasing pres-sure (Falloon and Danyushevsky, 2000). We attempt to estimate theprimary compositions of the Wajilitage nephelinite in equilibriumwith olivine of Fo85.8, i.e., magmaswithMg# of 68 through the followingprocedure: (1) by adding equilibrium olivine, (2) by simultaneouslyadding olivine and clinopyroxene and (3) by modeling using theMELTS software. In the scheme 2, we evaluated the effect of fraction-ation of olivine and clinopyroxene in a fixed proportion (3:1). Thecorrected compositions are listed in Supplementary Table A.11 andthe results show similar compositions with the only exception of TiO2

content (ranging from 3.3–6.72 wt.%). The melts from scheme 1 andscheme 2 were tested by running MELTS programme, and both ofthem displayed the crystallization of major minerals (e.g. Ol, Cpx, NeandMt). The crystallization sequence bymodelingwith theMELTS soft-ware is also consistent with the petrographic observation. The comput-ed primary melt shows a restricted range of SiO2 (38.15–39.31 wt.%),low Al2O3 (8.44–9.50 wt.%) and CaO (8.53–10.03 wt.%), high FeOt

(13.89–14.58 wt.%) and TiO2 (3.3–6.72 wt.%) contents. MgO contentsvary from 15.81–16.37 wt.% and the Mg# is 67. We should cautionthat the reconstructed melts may not represent the exact primarymelt composition, because the composition and proportion of the lostpyroxene remains uncertain. However, our approach provides neces-sary corrections to make the obtained melts in equilibrium with themost magnesian olivine, so that the reconstructed melts may beregarded as near-primary melts. Furthermore, the evaluated primarymelts plot in the region of the melitiite according to the classificationdiagram of Le Bas (1989) and resemble the compositions of the olivinemelilitites from the Oldoinyo Lengai.

Based on previous studies, the generation of the nephelinite magmais attributed to small-degree partial melting of the CO2-bearing ormetasomatized mantle source under high pressure (Ali et al., 2013;Vapnik et al., 2007). We attempted to use the reconstructed primarymelts to evaluate the conditions under which the melts were generated.Based on Albarede (1992)’s empirical equation T(°C) = 2000[MgO/(SiO2 + MgO)] +969 and ln[10P](Gpa) = 0.00252 T-0.12SiO2 + 5.027,the source potential temperature and pressure can be estimated to be1542-1547°Cand 4.1-4.3Gpa, corresponding to 135-140 km in depth. Ac-cording to the adiabatic P-T paths model for primary ultramafic magmasproposed by Herzberg and O’Hara (2002), an initial melting temperatureof 1550–1575 °C and pressure of 3.5–3.7Gpa is obtained, in accordancewith 115–120 km in depth. Walter (1998) suggested that magma pro-duced under relatively high pressures is characterized by high FeOt andCaO/Al2O3 ratios and low SiO2 and Al2O3 contents, which is consistentwith the primary melt of the Wajilitage nephelinite. The evaluated highpressure (3.5–3.7Gpa) is consistent with a depth of 115–140Km, withinthe garnet stability field. The estimated source temperature (1540–1575 °C) is ~200 °C higher than the normal asthenosphere beneathMORB (Putirka, 2005; Putirka et al., 2007).

6.3. Nature of mantle source

Previous studies have suggested several source types for nephelinite,such as garnet pyroxenite, eclogite, hornblendite, carbonated peridotiteandmixed sources (Zeng et al., 2010). The experimentalmelts producedin high pressure experiments do not match with all the featuresof the Wajilitage nephelinite (Dasgupta et al., 2007; Hirose, 1997;

Hirschmann et al., 2003; Kogiso and Hirschmann, 2006; Kogiso et al.,2003; Pilet et al., 2008). For example, compared to theWajilitage neph-elinite, the experimental melts from the garnet pyroxenite have higherSiO2 and Al2O3 and lower FeOt and TiO2 contents (Hirschmann et al.,2003; Kogiso et al., 2003). Melts from the eclogite have higher FeOt

and TiO2 and lower MgO contents (Dasgupta et al., 2006; Kogiso andHirschmann, 2006). Melts from the hornblendite have higher TiO2 andAl2O3 and lower FeOt contents (Pilet et al., 2008).

The carbonated peridotite is commonly interpreted as the mainsource of nephelinite (e.g. Dasgupta et al., 2007; Hirose, 1997; Sissonet al., 2009; Zeng et al., 2010). Melts from the carbonated peridotiteare characterized by higher MgO (N15 wt.%) and enrichment in the in-compatible element. In addition, carbonatitic liquids also account for thenegative anomalies of Zr and Hf and high Zr/Hf values as shown by themost carbonatites worldwide, because there is no mineral phase incabonatite that incorporates these elements. Moreover, the bulk parti-tion coefficient of Hf is higher than Zr in the carbonatitic system(DHf N DZr) (Dasgupta et al., 2009). Hence, melts derived from the car-bonated mantle are expected to be depleted in Zr and Hf and havehigh Zr/Hf ratios. Such melts are similar to the Wajilitage nephelinitein both major and trace element compositions. Notably, the light Mgisotopic compositions (δ26Mg value ranging from −0.35 to −0.55‰)are commonly interpreted as signatures of carbonated mantle source(Brenot et al., 2008; Higgins and Schrag, 2010; Yang et al., 2012). More-over, the carbonatitic xenoliths recognized in the nephenilite resemblethe carbonate-bearing cumulate xenoliths in Tertiary alkali basaltsfrom southern Slovakia, which were interpreted to have accumulatedfrom volatile-rich magma linked to a low-degree partial melting of car-bonated mantle (Hurai et al., 2007). Consequently, we suggest carbon-ated mantle as the most likely source for the Wajilitage nephelinite.However, the Wajilitage nephelinite is also characterized by extremelyhigh FeOt and TiO2 contents which cannot be explained by the partialmelting of carbonated peridotite source. Therefore, we seek a hybrid-ized source for the Wajilitage nephelinite. Both hornblendite andeclogite could account for the high FeOt and TiO2 abundances. However,the hornblendite source does not seem to be reasonable, because thehornblende cannot be stable beyond 90 km depth (Tatsumi, 1989).Dasgupta et al. (2006) and Kogiso and Hirschmann (2006) proposedthat melting of eclogite can contribute high FeOt and TiO2 contents si-multaneously. In addition, it is well known that rutile is a commonmin-eral phase in eclogites and rutile is the major reservoir of the HFSEs,particularly for the Nb, Ta and Ti. Melting of rutile in the source athigh temperatures induced by plume would produce elevated Nb, Taand Ti abundances in the melts (Klemme et al., 2002). Such a scenariois well in accordance with the Nb and Ta positive anomalies and highTiO2 contents observed in the Wajilitage nephelinite. Therefore, itis possible that some eclogitic components are involved in the car-bonated mantle source. In the SiO2 + TiO2 + Al2O3-FeO + MgO-CaO + Na2O + K2O diagram based on experimental results, theWajilitage samples are plotted in the overlapping field of carbonatedperidotite and eclogite (Fig. 12a). This hypothesis is also consistentwith the studies of Lassiter et al. (2000) and Prytulak and Elliott(2007), who proposed that some eclogite or pyroxenite was involvedin the carbonated peridotite source of the Hawaiian lavas based onthe Re-Os isotopic studies.

As stated above, the nephelinite samples are characterized bymarked enrichment in the incompatible elements and have high La/Ybratios, which suggest that the rocks were produced by low degrees ofpartial melting. The high La/Yb ratios also indicate an important rolefor the residual garnet during melting. As suggested by plots in the(La/Yb)N-(Dy/Yb)N diagram (Fig. 12b), the Wajilitage nephelinite wasgenerated by small degrees of melting (1%–1.5%) of carbonated mantleperidotite.

The Mg isotopic compositions and the major and trace elements ofthe nephelinite indicate that the subcontinental lithospheric mantle(SCLM) had been metasomatized by carbonatitic melt. The carbonatitic

Fig. 12. (a) SiO2 + TiO2 + Al2O3-FeO + MgO-CaO + Na2O + K2O diagram based on the experimental results. Data of hornblendite from Pilet et al. (2008); Garnet pyroxenite fromHirschmann et al. (2003) and Kogiso et al. (2003); Eclogite fromDasgupta et al. (2006) and Kogiso and Hirschmann (2006); Carbonated peridotite fromDasgupta et al. (2007) and Hirose(1997). (b) (La/Yb)N − (Dy/Yb)N diagram. The hypothetical mantle source is assumed to be composed of the 70% carbonated mantle peridotite and 30% eclogite. The carbonated mantleperidotite is assumed to be 38% olivine, 10% orthopyroxene, 33% clinopyroxene, 17% garnet and 2% calcite. The eclogite is assumed to be 60% clinopyroxene and 30% garnet. Partition co-efficients are following Salters and Stracke (2004).

174 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

melts are believed to be formed by partial melting of the subductedplate or derived from the asthenosphere (D’Orazio et al., 2007;Schiano et al., 1994). However, it is difficult to distinguish the two ori-gins by conventional isotopic system and element geochemistry. Morerecently, magnesium isotope has been employed as a potential tracerfor recycled carbonate. Previous studies reported limited δ26Mg valuesfor the normal mantle (−0.25 ± 0.07‰; Teng et al., 2007, 2010;Zhang and Li, 2012), but for a larger range for carbonate(−1 ~ −5.5‰; Brenot et al., 2008; Higgins and Schrag, 2010). It hasbeen suggested that sedimentary carbonates can bewell preserved dur-ing subduction (e.g. Kerrick and Connolly, 2001). As the main reservoirof the light-δ26Mg on the earth, the recycled carbonates are the mostlikely factor that contributes to the range of mantle Mg isotope compo-sitions. Wang et al. (2014) noted that crustal dolomite can preservetheir initial light δ26Mg values during subduction, whereas the δ26Mgvalues of limestone will be elevated after carbonate-silicate interaction.Notably, although the calcite-rich carbonates appear to become slightlyheavier through isotopic exchange between silicate and carbonate com-ponents during subduction, the residual carbonates after interaction arestill lighter than the normal mantle. On the other hand, the carbonatedsilicates (e.g. carbonated eclogite) would shift to lighter δ26Mg values(Wang et al., 2014). Consequently, the recycled carbonates and silicatesmay cause the mantle Mg isotopic heterogeneity and yield light-δ26Mgcomponents in themantle source. Thus, the lightMg isotopic signaturesof some mantle-derived rocks (δ26Mg b −0.25‰) are commonlyinterpreted to reflect the incorporation of subducted carbonate intheir mantle sources, such as in the case of the Mesozoic-Cenozoic ba-salts in the eastern part of China (Yang et al., 2012). The Mg isotope sig-nature of the Wajilitage nephelinite is lower than that of present-daynormal mantle, and with less possibility of older (Proterozoic) recycledcarbonates, we consider the Paleozoic subduction event as the mostplausible mechanism that resulted in the incorporation of the carbon-ates in the mantle source. Otherwise they may be uniformed to normalmantle values again. This interpretation is also supported by the region-al geological setting of northern margin of the TC. Several ophiolite mé-langes (600–418 Ma) and arc-type magmatic records (430–420 Ma)have been reported from the northern margin of the TC, and have beeninterpreted as a response of the southward subduction of the SouthernTianshan oceanic slab (Ge et al., 2012; Huang et al., 2013; Zhao et al.,2014). Moreover, the geochemical data of the nephelinite are character-ized by HIMU features, such as high U/Pb (0.60–0.83), Th/Pb (0.60–0.83), Nb/U (32.14–55.18) and Ce/Pb (38.88–68.64) ratios, which arealso commonly interpreted as the signature of recycled subducted oceaniccrust (Hofmann, 1997). Hence, the mantle beneath the Wajilitage was

most probably modified by recycled carbonates during the early-middlePaleozoic subduction event.

6.4. Relation to mantle plume

Large igneous provinces are generally considered to be related tomantle plume. However, the hypothesis of mantle plume has beenquestioned recently because of the absence of critical geologic evidencein some LIPs, such as hotspots trails, kilometer scale crustal uplift, giantmafic dyke swarms and high-temperature picrites or komatiites(Condie, 2001; Ernst and Buchan, 2001). Alternately, they proposedseveral alternative models to explain these large volume of magmas,for example, large scale delamination of lithosphere and the break-offof the subducted plates (Keskin, 2003), rift-related decompressionmelting (King and Anderson, 1995) and meteorite impact (Jone,2005). However, these models failed to match all the characteristicsfrom LIPs, and in comparison, the plume model is more favorable.Although the TLIP has been suggested to be the consequence of thePermian mantle plume, critical evidence still remains scarce. Severallines of evidence have been proposed to support the mantle plume in-volvement, such as the kimberlitic rocks in Bachu area (Zhang et al.,2013), pre-magmatism crustal doming (Li et al., 2014) and the occur-rence of picrites (Tian et al., 2010). In this contribution, additionalevidence for the mantle plume model for the Wajilitage nephelinite isdiscussed below.

Nephelinite generally shows spatial and temporal distributions associ-ated with rifts (East African Rift), oceanic islands (Cape Verde, Hawaiian)and flood basalt provinces (Deccan LIPs; Karoo LIPs), which have beencorrelated to plume events. One of the robust evidence for relating neph-elinite to mantle plume is the similar isotopic compositions betweennephelinite andOIB,which is also displayed by theWajilitage nephelinite.Moreover, the higher thermal anomaly relative to the normal astheno-sphere is thought as one of the major plume expressions. The magnitudeof thermal anomaly could extend to 350 °C in the plume axis areawhere-as it may be just ~200 °C in the periphery of plume head or in the latestage (Campbell, 2007). The occurrence of picrite in the ELIP has beensuggested as evidence for high mantle potential temperatures of 1630–1690 °C with at least 280 °C thermal anomaly (Zhang et al., 2006). Thethermal anomaly for the mantle source of the Wajilitage nephelinite ap-pears to be slightly lower than that of a plume-axis origin but is compara-ble with the thermal anomaly obtained for the periphery or late-stage ofthe mantle plume. This scenario is in good agreement with our inferencethat the nephelinite magma was produced in the later stage of TLIP as aresult of the plume-lithosphere interaction.

175Z. Cheng et al. / Lithos 220–223 (2015) 164–178

As described previously, the Wajilitage nephelinite lies over theKeziletag Formation of Devonian strata. The absence of the Carbonifer-ous strata between late Devonian and early Permian indicates a crustaldoming uplift and denudation of the Carboniferous strata prior to thenephelinitic volcanism. The Carboniferous strata, as noted by regionalstratigraphic correlation, were mainly carbonate platforms sedimentswith a stable cratonic basin setting and could be up to 971m in thickness(e.g. Jia et al., 2004; Li et al., 2014). Therefore, the extent of vertical upliftis expected to be in kilometers scale which is commonly interpreted asan evidence for the involvement of mantle plume (Saunders et al.,2007). This scenario conforms with the observation of Li et al. (2014),who reported two regional connecting-well profiles and two seismicprofiles across the Tarim Basin, and recognized a regional unconformitybetween Carboniferous and Permian. The Carboniferous strata of thenorthern and western part of Tarim basin has been denuded to thehighest degree and gradually weakens to the south. They concludedthat the erosion of the Carboniferous strata imply a crustal uplift eventat ~300Mawith vertical extent of more than 887m, suggesting a possi-ble genetic link to mantle plume.

6.5. Implications for the TLIP

6.5.1. Eruption age of the Wajilitage nepheliniteAs a potential carrier of uranium, rutile has been used for U-Pb dat-

ing in several studies (e.g. Bracciali et al., 2013; Q.L. Li et al., 2011). Thetechnique is more commonly applied for Precambrian rocks because ofthe low U content in rutile (Clark et al., 2000; Ireland and Williams,2003). Recently, relatively young rutile U-Pb ages were reported bySIMS method such as the U-Pb age of 318 ± 7Ma for the southwesternChinese Tianshan eclogites (Q.L. Li et al., 2011), and other young rocks(Li et al., 2003). In the present case, the data yield a weighted average206Pb/238U age of 268 ± 30 Ma (MSWD = 0.89; Fig. 6b). Althoughmore precise age is still required for the nephelinite, the data fromnepheline syenite xenoliths also indicate that the lavas erupted afterthe emplacement of nepheline syenite. The nepheline syenite hasbeen considered to represent the youngest magmatic event of the TLIPwith the intrusive age of 277 Ma by SHRIMP U-Pb zircon age dating(Yang et al., 1996). If this is the case, the nephelinite is apparentlyyounger than the syenite and the duration of the TLIP is longer than pre-viously considered.

It is generally accepted that the LIPs are generated in just a fewmil-lion years, such as the Emeishan LIP (~10 Ma), Siberian Traps (~6 Ma)and Deccan (~3 Ma; Bryan and Ernst, 2008 and the references therein).Compared with the other typical LIPs, the TLIP appears to have a muchlonger duration. Indeed, some LIPs with lifespan of N15 Ma have alsobeen reported, e.g. the Kerguelen, North Atlantic and Ontong Java pla-teaus (e.g. Frey et al., 2003), and the overall age duration of the LIPs ap-pears to be maximum up to ~50 Ma (Ernst and Buchan, 2001). Theshort-lived LIPs are typically characterized by single pulse of magmaticevent with a high eruption rate, whereas the LIPs of long duration arelikely to be pulsatory (Bryan and Ernst, 2008). For example, in theNorth Atlantic Igneous Province (62–53 Ma), the first pulse from 62 to58 Ma corresponded to the eruption of the terrestrial continentalflood basalt sequences, whereas the second syn-rift pulse at 56–52 Mawas represented by the bulk of the volcanic sequences along the conti-nental shelves which formed the ‘seaward-dipping reflector series’(Storey et al., 2007). It is worthy to note that although some LIPscould be long-lived, the N90% of the total volume may have beenemplaced within the shortest period (Condie, 2001), and this scenariois also in agreement with the TLIP. The geochronological dating fromthe Tarim flood basaltic volcanism, which were considered as themajor phase of TLIP, begin at ca. 292 Ma and ended around 287 Ma(Xu et al., 2014 and the references therein). It is apparent that thelithosphere (N150 km) played a critical role for the melt generation,and accounts for the long time span and a relatively low eruption rate(e.g. Kent et al., 1992; Turner et al., 1996).

6.5.2. Geodynamic modelThe ~280 Ma igneous rocks of the TLIP display variable

εNd(t) isotope compositions. For example, the mafic-ultramafic layeredintrusions exhibit εNd(t) values from −1.05 to +3.01 (Zhang et al.,2014, in revision), the dykes show εNd(t) ranging from −2.22 to +4.8(Wei et al., 2014; Y.T. Zhang et al., 2010), and the nephelinite has nota-bly uniform εNd(t) values from+3.28 to+3.88,which tend tobehigherwith time. This scenario indicates that the low εNd(t) material wanedwith time, and might represent the gradual depletion of the easilymelted lithospheric wall-rock in the magmatic conduit system, or themelting out of a more fusible component in the mantle source itself.More low-εNd(t) material in the source should normally begin meltingat greater depths than the more refractory components (e.g. Ito andMahoney, 2005). Somewhat similar secular variations have been ob-served in a portion of the ~260 Ma Emeishan flood basalt pile, and arealso interpreted in terms of cooling of a plume and variation in litho-spheric thickness (Zhang et al., 2006). In the Maymecha River basin ofthe Siberian Traps, alkalic, high-Ti picritic basalts and meimechites lieabove a sequence of low-Ti tholeiitic lavas. This late high-Ti group isinterpreted to have formed by small amounts of partial melting ofgarnet-facies mantle at even greater depths (N200 km, Arndt et al.,1998) than those we estimate for the Wajilitage nephelinite lavas. Xuet al. (2014) proposed a plume incubation model to account for themain characteristics of the TLIP, in which the ~280 Ma magmas weresuggested to be derived from the convecting mantle in the weak sec-tions at the margins of the TC whereas the ~290 Ma flood basaltswere likely formed as a result of mixing of plume-derived melts withSCLM-derived melts (e.g., lamproitic melt) as they rose through theSCLM. Although this model can be applied to interpret the variation ofεNd(t) values in thewhole TLIP, it fails to account for the variable Nd iso-topic compositions of the ~280Ma magmatic event. This is because theεNd(t) values of all ~280 Ma igneous rocks should be similar, if thismodel is valid.Moreover, if the plumeflows to theweak zones or thinnedspots (~80 km) at the margins from the internal TC (~150 km) and in-duced decompression melting subsequently, the ~280 Ma magmatismis expected to be voluminous, which is not the case. Thus, the broadly de-pleted isotopic signatures of ~280 Ma igneous rocks do not necessarilymean that they were derived from the convective mantle. Instead,these features can be explained by a plume-lithosphere interactionmodel, withinwhich progressive increasing inmelting depth and the im-pingement of the plume materials are expected with the thinning litho-sphere (Fig. 13).

The Wajilitage nephelinite occurs only in very small volume withinb1 km2 area, in agreement with the conclusion that these rocks formedby small amounts of partial melting largely in a carbonated garnet peri-dotite mantle source. Thus, the plume-lithosphere model is the mostplausible one, if melting occurred beneath a thick lithospheric lid. Thepresent-day lithospheric thickness beneath the Tarim region rangesfrom 150 to 200 km, from the margin to interior, which is comparablewith the thicknesses (~150 km) during early Permian (Lei and Zhao,2007; Xu et al., 2002; Zhang et al., 2013). However, compared with~150 km thickness in the early stage (Zhang et al., 2013), the litho-sphere had been slightly thinned in the late stage (b115–140 km), assuggested by the nephelinite in this study. It is not clear why plume-related stress on the overlying lithosphere had not led to major litho-spheric thinning like in the case of the other typical LIPs, and furtherstudies are required to address this enigma.

In combination with regional geology, we interpret the variation inlithology to be a result of the different stages of the plume-lithosphereinteraction (Fig. 13): (1) In the early-middle Paleozoic, the SouthTianshan oceanic plate subducted southward to the TC (Charvet et al.,2011; Ge et al., 2012; Wang et al., 2011), and both the asthenosphericand lithospheric mantle of the TC was metasomatized by subductioncomponents. As a result, themantle beneath the TCwas enriched in car-bonate or other components, e.g. phlogopite and amphibole (Fig. 13a;Foley, 1992). (2) ~300 Ma, the mantle plume reached the base of the

Fig. 13.Geodynamicmodel showing themajor geological process in the TC (modified after Xu et al., 2014; Zhang et al., 2014). (a) In the early-middle Paleozoic, the South Tianshan oceanicplate subducted southward beneath the TC, and the SCLMof the TCwasmetasomatized. (b) ~300Ma, themantle plume reached the base of the lithosphereof the TC, and triggeredmeltingof the enriched constituents, generating the kimberlitic rocks. (c) ~290 Ma, the massive flood basalts were produced by incompatible-enriched lithospheric mantle in response to theplume-lithosphere interaction. (d) ~280Ma and later, the lithospherewas slightly thinned by plume thermal erosion and the SCLM continuedmelting to produce the ~280Ma intrusions.Later, as a result of the lithosphere thinning, upwelling of the carbonated asthenospheric mantle occurred and the resultant thermal anomaly causedmelting to generate the nephelinitemagma.

176 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

lithosphere of the TC, and triggered melting of the fusible constituentssuch as the carbonate and phlogopite veins (Zhang et al., 2013) to pro-duce the kimberlitic rocks (Fig. 13b). (3) ~290 Ma, the massive floodbasalts were produced by partial melting of incompatible element-enriched lithospheric mantle in response to the plume-lithosphere in-teraction (Fig. 13c). (4) ~280 Ma and later, the lithosphere was thinnedto some extent by plume thermal erosion and the deeper SCLM contin-ued melting to produce the ~280 Ma intrusions. Later, as a result of theslight lithosphere thinning, the upwelling of carbonated asthenosphericmantle occurred and generated the nephelinite magma (Fig. 13d). Con-sequently, themajormagmatic events of the TLIPwere derived from theSCLM, but not the convective mantle, except for the Wajilitage nephe-linite which stands for the decompression melting of the upwellingmantle plume. Thewide variety of rock types in the TLIP can be attributedto the depth and degree of melting of the different mantle sources, asillustrated in our geodynamic model.

7. Conclusions

The olivine phenocrysts in nephelinite ofWajilitage area crystallizedfrom liquids with Mg-number as high as 85.8. We estimate that thenephelinitic lavas were generated from primary melt compositionswith 15.8 wt.% to 16.4 wt.% MgO. The source potential temperature iscalculated to have been as high as 1540–1575 °C,~200 °C hotter thanthe normal asthenosphere beneath MORB. The thermal anomaly forthe source of Wajilitage nephelinite and pre-volcanic crustal uplift inthe Bachu region appear to confirm the mantle plume involvement.Both the field relations and SIMS rutile U-Pb age data (~268 ± 30 Ma)

shows that the Wajilitage nephelinite erupted during late Permian,representing the last phase of the TLIP magmatism.

Based on the Sr-Nd isotopes and themajor and trace element data, inconjunction withMg isotopic data, theWajilitage nephelinite is consid-ered to have been generated by low-degree partial melting of the a hy-bridized carbonated peridotite/eclogite source with heat input frommantle plume. Moreover, the asthenospheric mantle beneath the TC isinferred to be carbonated, which is suggested to be metasomatized bysubducted carbonate during early-middle Paleozoic. Isotopic character-istics of the three magmatic events of TLIP is consistent with a plume-lithosphere interaction model. Progressive increasing in melting depthand the impingement of the plume materials are expected to interpretthe variety of rock types and their distinct petrological features duringthe evolution of the TLIP. We suggested that major igneous rocks ofthe TLIP are derived from the SCLM, except for the nephelinite whichis considered as a result of decompression melting of the upwellingmantle plume.

Acknowledgements

We thank Ting Gao for technical support of Mg isotope analyses andDr. He Huang and Jinchao Ye for field research. Financial support for thiswork was supported by the National Nature Science Foundation ofChina (41472060 and 41230209), 305 Project of the State Science andTechnology Program of China (2011BAB06B0204), 973 Program(2012CB416806), 111 Project (B07011) and the Fundamental ResearchFunds for the Central Universities.

177Z. Cheng et al. / Lithos 220–223 (2015) 164–178

Appendix A. Supplementary material

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.lithos.2015.02.002.

References

Albarede, F., 1992. Howdeep do commonbasalticmagmas formand differentiate? Journalof Geophysical Research 97, 10997–11009.

Ali, S., Ntaflos, T., Upton, B.G.J., 2013. Petrogenesis and mantle source characteristics ofQuaternary alkaline mafic lavas in the western Carpathian–Pannonian Region, Styria,Austria. Chemical Geology 337–338, 99–113.

Arndt, N.T., Chauvel, C., Fedorenko, V., Czamanske, G., 1998. Two mantle sources, twoplumbing systems: tholeiitic and alkaline magmatism of the Maymecha River basin,Siberian flood volcanic province. Contributions to Mineralogy and Petrology 133,297–313.

Baker, M.B., Stolper, E.M., 1994. Determining the composition of high-pressure mantlemelts using diamond aggregates. Geochimica et Cosmochimica Acta 58, 2811–2827.

Bao, P.S., Su, L., Zhai, Q.G., Xiao, X.C., 2009. Compositions of the kimberlitic brecciatedperidotite in the Bachu area, Xinjiang and its ore-bearing potentialities. Acta GeologicaSinica 83, 1276–1301 (in Chinese with English abstract).

Barker, A.K., Troll, V.R., Ellam, R.M., Hansteen, T.H., Stillman, C.J., Andersson, A., 2012.Magmatic evolution of the Cadamosto Seamount, Cape Verde: beyond the spatialextent of EM1. Contributions to Mineralogy and Petrology 163, 949–965.

Bracciali, L., Parrish, R.R., Horstwood, M.S.A., Condon, D.J., Najman, Y., 2013. U-PbLA-(MC)-ICP-MS daing of rutile: New reference materials and applications to sedi-mentary provenance. Chemical Geology 347, 82–101.

Brenot, A., Cloquet, C., Vigier, N., Carignan, J., France-Lanord, C., 2008. Magnesium isotopesystematics of the lithologically varied Moselle river basin, France. Geochimica etCosmochimica Acta 72 (20), 5070–5089.

Bryan, S.E., Ernst, R.E., 2008. Revised definition of large igneous provinces (LIPs). EarthScience Reviews 86, 175–202.

Campbell, I.H., 2007. Testing the plume theory. Chemical Geology 241, 153–176.Campbell, I.H., Griffiths, R.W., 1990. Implications ofmantle plume structure for the evolution

of flood basalts. Earth and Planetary Science Letters 99, 79–93.Cao, J., Wang, C.Y., Xing, C.M., Xu, Y.G., 2014. Origin of the early PermianWajilitag igneous

complex and associated Fe-Ti oxidemineralization in the Tarim large igneous province,NW China. Journal of Asian Earth Sciences 84, 51–68.

Charvet, J., Shu, L.S., Laurent-Charvet, S., Wang, B., Faure, M., Cluzel, D., Chen, Y., De Jong,K., 2011. Palaeozoic tectonic evolution of the Tianshan Belt, NW China. Science ChinaEarth Science 54, 166–184.

Chen, H.L., Yang, S.F., Li, Z.L., Yu, X., Luo, J.C., He, G.Y., Lin, X.B., Wang, Q.H., 2009. Spatialand temporal characteristics of Permian large Igneous Province in Tarim Basin.Xinjiang Petroleum Geology 30 (2), 179–182 (in Chinese with English abstract).

Cheng, Z.G., Zhang, Z.C., Santosh, M., Hou, T., Zhang, D.Y., 2014. Carbonate- and silicate-rich globules in the kimberlitic rocks of northwestern Tarim Large Igneous Province,NW China: Evidence for carbonated mantle source. Journal of Asian Earth Sciences.http://dx.doi.org/10.1016/j.jseaes.2014.04.018.

Clark, D.J., Hensen, B.J., Kinny, P.D., 2000. Geochronological constraints for a two-stage historyof the Albany–Fraser Orogen, Western Australia. Precambrian Research 102, 155–183.

Condie, K.C., 2001. Mantle plumes and their record in Earth history. In: Condie, K.C. (Ed.),Mantle plume generation and melting. Cambridge University Press, Cambridge,pp. 115–143.

Dasgupta, R., Hirschmann,M.M., Stalker, K., 2006. Immiscible transition from carbonaterichto silicate-richmelts in the 3 GP a melting interval of eclogite + CO2 and genesis ofsilica-undersaturated ocean island lavas. Journal of Petrology 47, 647–671.

Dasgupta, R., Hirschmann, M.M., Smith, N.D., 2007. Partial melting experiments ofperidotite+CO2 at 3 GPa and genesis of alkalic ocean island basalts. Journal of Petrology48, 2093–2124.

Dasgupta, R., Hirschmann, M.M., McDonough, W.F., Spiegelman, M., Withers, A.C., 2009.Trace element partitioning between garnet lherzolite and carbonatite at 6.6 and8.6 GPa with applications to the geochemistry of the mantle and of mantle-derivedmelts. Chemical Geology 262, 57–77.

D'Orazio,M., Innocenti, F., Tonarini, S., Doglioni, C., 2007. Carbonatites in a subduction system:The Pleistocene alvikites fromMt. Vulture (southern Italy). Lithos 98 (1–4), 313–334.

Ernst, R.E., Buchan, K.L., 2001. Large mafic magmatic events through time and links tomantle-plume heads. In: Ernst, R.E., Buchan, K.L. (Eds.), Mantle plumes: Their identifi-cation through time. Geological Society of America Special Paper vol. 352, pp. 483–575.

Falloon, T.J., Danyushevsky, L.V., 2000. Melting of refractory mantle at 1.5, 2 and 2.5 Gpaunder anhydrous and H2O-underasurated conditions: implications for the petrogen-esis of high-Ca boninites and the influence of subduction components on mantlemelting. Journal of Petrology 41, 257–283.

Foley, S.F., 1992. Vein-plus-wall-rock melting mechanisms in the lithosphere and theorigin of potassic alkaline magmas. Lithos 28 (3–6), 187–204.

Foster, M.D., 1960. Interpretation of the composition of trioctahedralmicas. U S GeologicalSurvey Professional Paper 354 (B), 11–49.

Frey, F.A., Coffin, M.F., Wallace, P.J., Weis, D., 2003. Leg 183 synthesis: Kerguelen Plateau-Broken Ridge-a large igneous province. In: Frey, F.A., Coffin, M.F., Wallace, P.J., Quilty,P.G. (Eds.), Proceedings of the ODP. Scientific Results vol. 183, pp. 1–48.

Ge, R.F., Zhu, W.B., Wu, H.L., Zheng, B.H., Zhu, X.Q., He, J.W., 2012. The Paleozoic northernmargin of the Tarim Craton: passive or active? Lithos 142–143, 1–15.

Herzberg, C., Asimow, P.D., 2008. Petrology of some oceanic island basalts: PRIMELT2.XLSsoftware for primary magma calculation. Geochemistry, Geophysics, Geosystems.http://dx.doi.org/10.1029/2008GC002057.

Herzberg, C., O’Hara, M.J., 2002. Plume-associated ultramafic magmas of Phanerozoic age.Journal of Petrology 43, 1857–1883.

Herzberg, C., Asimow, P.D., Arndt, N., Niu, Y., Lesher, C.M., Fitton, J.G., Cheadle, M.J.,Saunders, A.D., 2007. Temperatures in ambient mantle and plumes: Constraintsfrom basalts, picrites, and komatiites. Geochemistry, Geophysics, Geosystems.http://dx.doi.org/10.1029/2006GC001390.

Higgins, J.A., Schrag, D.P., 2010. Constraining magnesium cycling in marine sedimentsusing magnesium isotopes. Geochimica et Cosmochimica Acta 74 (17), 5039–5053.

Hirose, K., 1997. Partial melt compositions of carbonated peridotite at 3 GPa and role ofCO2 in alkali–basalt magma generation. Geophysical Research Letters 24, 2837–2840.

Hirschmann, M.M., Kogiso, T., Baker, M.B., Stolper, E.M., 2003. Alkalic magmas generatedby partial melting of garnet pyroxenite. Geology 31, 481–484.

Hofmann, A., 1997. Mantle geochemistry: The message from oceanic volcanism. Nature385 (6613), 219–229.

Hofmann, A.W., Jochum, K.P., Seufert, M., White, W.M., 1986. Nb and Pb in oceanicbasalts: New constraints on mantle evolution. Earth and Planetary Science Letters79 (1–2), 33–45.

Huang, H., Zhang, Z.C., Santosh, M., Zhang, D.Y., Zhao, Z.D., Liu, J.L., 2013. Early PaleozoicTectonic Evolution of the South Tianshan Collisional Belt: Evidence from Geochemistryand Zircon U-Pb Geochronology of the Tie’reke Monzonite Pluton, Northwest China.The Journal of Geology 121 (4), 401–424.

Hurai, V., Huraiova, M., Konecny, P., Thomas, R., 2007. Mineral-melt-fluid composition ofcarbonate-bearing cumulate xenoliths in Tertiary alkali basalts of southern Slovakia.Mineralogical Magazine 71 (1), 63–79.

Ireland, T.R., Williams, I.S., 2003. Considerations in zircon geochronology by SIMS. In:Hanchar, J.M., Hoskin, P.W.O. (Eds.), Zircon: Rev. Miner. Geochem 53, pp. 215–241.

Ito, G., Mahoney, J.J., 2005. Flow and melting of a heterogeneous mantle: 1. Importance tothe geochemistry of ocean island and midocean ridge basalts. Earth and PlanetaryScience Letters 230, 29–46.

Jia, C.Z., Zhang, S.B., Wu, S.Z., 2004. Stratigraphy of the Tarim Basin and Adjacent Areas.Science Press, Beijing, pp. 190–289 (In Chinese).

Jiang, C.Y., Zhang, P.B., Lu, D.R., Bai, K.Y., 2004. Petrogenesis and magma source of theultramafic rocks at Wajilitage region, western Tarim Plate in Xinjiang. Acta PetrologicaSinica 20, 1433–1444 (in Chinese with English abstract).

Jone, A.P., 2005. Meteorite impacts as triggers to large igneous provinces. Elements 1,277–281.

Kamo, S.L., Czamanske, G.K., Amelin, Y., Fedorenko, V.A., Davis, D.W., Trofimov, V.R., 2003.Rapid eruption of Siberian flood-volcanic rocks and evidence for coincidence withthe Permian-Triassic boundary and mass extinction at 251 Ma. Earth and PlanetaryScience Letters 214, 75–91.

Keller, J., Zaitsev, A., Wiedenmann, D., 2006. Primarymagmas at Oldoinyo Lengai: The roleof olivine melilitites. Lithos 91, 150–172.

Kent, R.W., Storey, M., Saunders, A.D., 1992. Large igneous provinces: Sites of plume impactor plume incubation? Geology 20, 89–894.

Kerrick, D.M., Connolly, J.A.D., 2001. Metamorphic devolatilization of subducted marinesediments and the transport of volatiles into the Earth's mantle. Nature 411 (6835),293–296.

Keskin, M., 2003. Magma generation by slab steepening and breakoff beneath asubduction-accretion complex: An alternative model for collision-related volcanismin Eastern Anatolia, Turkey. Geophysical Research Letter 30, 8046–8049.

King, S.D., Anderson, D.L., 1995. An alternativemechanism of flood basalt formation. Earthand Planetary Science Letters 136, 269–279.

Klaudius, J., Keller, J., 2006. Peralkaline silicate lavas at Oldoinyo Lengai, Tanzania. Lithos91, 173–190.

Klemme, S., Blundy, J.D., Wood, B.J., 2002. Experimental constraints on major and traceelement partitioning during partial melting of eclogite. Geochimica et CosmochimicaActa 66 (17), 3109–3123.

Kogiso, T., Hirschmann, M.M., 2006. Partial melting experiments of bimineralic eclogiteand the role of recycled mafic oceanic crust in the genesis of ocean island basalts.Earth and Planetary Science Letters 249, 188–199.

Kogiso, T., Hirschmann, M.M., Frost, D.J., 2003. High-pressure partial melting of garnetpyroxenite: possible mafic lithologies in the source of ocean island basalts. Earthand Planetary Science Letters 216, 603–617.

Korenaga, J., Kelemen, P.B., 2000. Major element heterogeneity in the mantle source of theNorth Atlantic igneous province. Earth and Planetary Science Letters 184, 251–268.

Lassiter, J.C., Hauri, E.H., Reiners, P.W., Garcia, M.O., 2000. Generation of Hawaiian post-erosional lavas bymelting of amixed lherzolite/pyroxenite source. Earth andPlanetaryScience Letters 178, 269–284.

Le Bas, M.J., 1989. Nephelinitic and basanitic rocks. Journal of Petrology 30, 1299–1312.Igneous rocks: A classification and glossary of terms. In: Le Maiitre, R.W. (Ed.), Cambridge

University Press, Cambridge, U.K, p. 252.Leake, B.E., Woolley, A.R., Arps, C.E.S., 1997. Nomenclature of amphiboles: Report of the

Subcommittee on Amphiboles of the International Miner-alogical Association, Com-mission on NewMineral and Mineral Names. American Mineralogist 82, 1019–1037.

Lei, J.S., Zhao, D.P., 2007. Teleseismic P-wave tomography and the upper mantle structureof the central Tien Shan orogenic belt. Physics of the Earth and Planetary Interiors 162,165–185.

Li, Q.L., Li, S.G., Zheng, Y.F., Li, H.M., Massonne, H.J., Wang, Q.C., 2003. A high precision U-Pbage of metamorphic rutile in coesite-bearing eclogite from the Dabie Mountains in cen-tral China: a new constraint on the cooling history. Chemical Geology 200, 255–265.

Li, Z.L., Yang, S.F., Chen, H.L., Langmuir, C.H., Yu, X., Lin, X.B., Li, Y.Q., 2008. Chronology andgeochemistry of Taxinan basalts from the Tarim basin: evidence for Permian plumemagmatism. Acta Petrologica Sinica 24 (5), 950–970 (in Chinese with English abstract).

Li, Q.L., Lin, W., Su, W., Li, X.H., Shi, Y.H., Liu, Y., Tang, G.Q., 2011. SIMS U-Pb rutile age oflow-temperature eclogites from southwestern Chinese Tianshan, NW China. Lithos122, 76–86.

178 Z. Cheng et al. / Lithos 220–223 (2015) 164–178

Li, Z.L., Chen, H.L., Song, B., Li, Y.Q., Yang, S.F., Yu, X., 2011. Temporal evolution of thePermian large igneous province in Tarim Basin, Northwest China. Journal of AsianEarth Sciences 42, 917–927.

Li, Z.L., Li, Y., Chen, H.L., Santosh, M., Yang, S.F., Xu, Y.G., Langmuir, C.H., Chen, Z.X., Yu, X.,Zou, S.Y., 2012. Hf isotopic characteristics of the Tarim Permian large igneous prov-ince rocks of NW China: Implication for the magmatic source and evolution. Journalof Asian Earth Sciences 49, 191–202.

Li, Y.Q., 2013. Study onMagmaDynamics and Ore Potential of the Early Permian Tarim LargeIgneous Province (Ph.D thesis), Zhejiang University (in Chinese with English abstract).

Li, D.X., Yang, S.F., Chen, H.L., Cheng, X.G., Li, K., Jin, X.L., Li, Z.L., Li, Y.Q., Zou, S.Y., 2014. LateCarboniferous crustal uplift of the Tarim plate and its constraints on the evolution ofthe Early Permian Tarim Large Igneous Province. Lithos 204, 36–46.

Long, X.P., Yuan, C., Sun, M., Zhao, G.C., Xiao, W.J., Wang, Y.J., Yang, Y.H., Hu, A.Q., 2011.Archean crustal evolution of the northern Tarim craton, NW China: Zircon U-Pband Hf isotopic constraints. Precambrian Research 180, 272–284.

McDonough, W.F., Sun, S.S., 1995. The composition of the Earth. Chemical Geology 120,2523–2553.

Mourão, C., Mata, J., Doucelance, R., Madira, J., Brum da Silveira, A., Silva, L.C., Moreira, M.,2010. Quaternary extrusive calciocarbonatite volcanism on Brava Island (CapeVerde): A nephelinite-carbonatite immiscibility product. Journal of African EarthScience 56, 59–74.

Norrish, K., Chappell, B.W., 1977. X-ray fluorescence spectrometry. In: Zussman, J. (Ed.),Physical methods in determinative mineralogy, 2nd edition Academic Press, NewYork, pp. 201–272.

Pilet, S., Baker, M.B., Stolper, E.M., 2008. Metasomatized lithosphere and the origin ofalkaline Lavas. Science 320, 916–919.

Prytulak, J., Elliott, T., 2007. TiO2 enrichment in ocean island basalts. Earth and PlanetaryScience Letters 263, 388–403.

Putirka, K.D., 2005. Mantle potential temperatures at Hawaii, Iceland, and the mid-oceanridge system, as inferred from olicine phenocrysts: evidence for thermally drivenmantle plumes. Geochemistry, Geophysics, Geosystems. http://dx.doi.org/10.1029/2005GC000915.

Putirka, K.D., Mikaelian, H., Ryerson, F., Shaw, H., 2003. New clinopyroxene-liquidthermobarometers for mafic, evolved, and volatile-bearing lava compositions, withapplications to lavas from Tibet and the Snake River Plain, Idaho. American Mineral-ogist 88 (10), 1542–1554.

Putirka, K.D., Perfit, M., Ryerson, F.J., Jackson, M.G., 2007. Ambient and excess mantle tem-peratures, olivine thermometry, and active vs. passive upwelling. Chemical Geology241, 177–206.

Qi, L., Hu, J., Gregoire, D.C., 2000. Determination of trace elements in granites by induc-tively coupled plasma mass spectrometry. Talanta 51, 507–513.

Salters, V.J.M., Stracke, A., 2004. Composition of the depleted mantle. Geochemistry,Geophysics, Geosystems. http://dx.doi.org/10.1029/2003GC000597.

Saunders, A.D., Jones, S.M., Morgan, L.A., Pierce, K.L., Widdowson, M., Xu, Y.G., 2007.Reginal uplift associated with continental large igneous province: The roles of mantleplumes and the lithosphere. Chemical Geology 241, 282–318.

Schiano, P., Clocchiatti, R., Shimizu, N., Wei, D., Mattielli, N., 1994. Cogenetic silicarbonate-rich and carbonate-rich melts trapped in mantle minerals in Kerguelen ultramaficxenoliths: Implications for metasomatism in the oceanic upper mantle. Earth andPlanetary Science Letters 123, 167–178.

Shellnutt, J.G., 2014. The Emeishan large igneous province: A synthesis. Geoscience Fron-tiers 5, 369–394. Shellnutt, J.G., Bhat, G.M., Wang, K.L., Brookfield, M.E., Jahn, B.M.,Dostal, J., 2014. Petrogenesis of the flood basalts from the Early Permian PanjalTraps, Kashmir, India: Geochemical evidence for shallow melting of the mantle.Lithos. http://dx.doi.org/10.1016/j.lithos.2014.01.008.

Sisson, T.W., Kimura, J.I., Coombs,M.L., 2009. Basanite–nephelinite suite from early Kilauea:carbonated melts of phlogopite–garnet peridotite at Hawaii’s leading magmatic edge.Contributions to Mineralogy and Petrology 158, 803–829.

Storey, M., Duncan, R.A., Tegner, C., 2007. Timing and duration of volcanism in the NorthAtlantic Igneous Province: implications for geodynamics and links to the Icelandhotspot. Chemical Geology 241, 264–281.

Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic systematic of oceanic basalts:Implications for mantle composition and process. In: Saunders, A.D., Norry, M.J.(Eds.), Magmatism in ocean basins. Geol. Soc. London. Spec. Publ. 42, pp. 313–345.

Tatsumi, Y., 1989. Migration of fluid phases and genesis of basalt magmas in subductionzones. Journal of Geophysical Research 94, 4697–4707.

Tatsumi, Y.M., Saknyama, H., 1983. Generation of arc basalt magmas and the thermalstructure of the mantle wedge in subduction zones. Journal of Geophysical Research88 (B7), 5815–5825.

Teng, F.Z., Wadhwa, M., Helz, R.T., 2007. Investigation of magnesium isotope fractionationduring basalt differentiation: implications for a chondritic composition of the terres-trial mantle. Earth and Planetary Science Letters 261 (1–2), 84–92.

Teng, F.Z., Li, W.Y., Ke, S., Marty, B., Dauphas, N., Huang, S., Wu, F.Y., Pourmand, A., 2010.Magnesium isotopic composition of the Earth and chondrites. Geochimica etCosmochimica Acta 74 (14), 4150–4166.

Thompson, R.N., Gibson, S.A., 2000. Transient high temperatures in mantle plume headsinferred from magnesian olivines in Phanerozoic picrites. Nature 407, 502–506.

Tian, W., Campbell, I.H., Allen, C.M., Guan, P., Pan, W.Q., Chen, M.M., Yu, H.J., Zhu, W.P.,2010. The Tarim picrite-basalt-rhyolite suite, a Permian flood basalt from northwestChina with contrasting rhyolites produced by fractional crystallization and anatexis.Contributions to Mineralogy and Petrology 160, 407–425.

Turner, S., Hawkesworth, C., Gallagher, K., Stewart, K., Peate, D., Mantovani, M., 1996.Mantle plumes, flood basalts, and thermal models for melt generation beneath con-tinents: assessment of a conductive heating model and application to the Parana.Journal of Geophysical Research-Solid Earth 101, 11503–11518.

Vapnik, Y., Sharygin, V.V., Samoilov, V., Yudalevich, Z., 2007. The petrogenesis of basic andultrabasic alkaline rocks of western Makhtesh Ramon, Israel: geochemistry, melt andfluid inclusion study. International Journal of Earth Sciences 96, 663–684.

Walter, M.J., 1998. Melting of garnet peridotite and the origin of komatiite and depletedlithosphere. Journal of Petrology 39, 29–60.

Wang, B., Shu, L.S., Faure, M., Jahn, B.M., Cluzel, D., Charvet, J., Chung, S.L., Meffre, S., 2011.Paleozoic tectonics of the southern Chinese Tianshan: insights from structural,chronological and geochemical studies of the Heiyingshan ophiolitic melange (NWChina). Tectonophysics 497, 85–104.

Wang, S.J., Teng, F.Z., Li, S.G., 2014. Tracing carbonate-silicate interaction during subduc-tion using magnesium and oxygen isotopes. Nature Communications. http://dx.doi.org/10.1038/ncomms6328.

Weaver, B.L., 1991. The origin of ocean island basalt end-member compositions: Traceelement and isotopic constraints. Earth and Planetary Science Letters 104, 381–397.

Wei, X., Xu, Y.G., Feng, Y.X., Zhao, J.X., 2014. Plume-lithosphere interaction in the genera-tion of the Tarim large igneous province, NWChina: geochronological and geochemicalconstraints. American Journal of Science 314, 314–356.

XBGMR: Xinjiang Bureau of Geology and Mineral Resources), 1984. Geological Map ofWajilitage area in Bachu County, Xinjiang, Scale1:5000 (in Chinese).

Xu, Y., Liu, F.T., Liu, J.H., Chen, H., 2002. Crust and upper mantle structure beneath westernChina from P wave travel time tomography. Journal of Geophysical Research.http://dx.doi.org/10.1029/2001JB000402.

Xu, Y.G., Wei, X., Luo, Z.Y., Liu, H.Q., Gao, J., 2014. The Early Permain Tarim large igneousprovince: Main characteristics and a plume incubation model. Lithos 204, 20–35.

Yang, S.F., Chen, H.L., Dong, C.W., 1996. The discovery of Permian syenite inside TarimBasin and its geodynamic significance. Geochimica 25 (2), 121–128.

Yang, S.F., Li, Z.L., Chen, H.L., Xiao, W.J., Yu, X., Lin, X.B., Shi, X.G., 2006. Discovery of aPermian quartz syenitic porphyritic dyke from the Tarim basin and its tectonic impli-cations. Acta Petrologica Sinica 22 (5), 1405–1412 (in Chinese with English abstract).

Yang, W., Teng, F.Z., Zhang, H.F., Li, S.G., 2012. Magnesium isotopic systematics of conti-nental basalts from the North China craton: Implications for tracing subductedcarbonate in the mantle. Chemical Geology 328, 185–194.

Yu, X., 2009. Magma evolution and deep geological processes of early Permian Tarim largeigneous province. (Ph.D thesis), Zhejiang University (in Chinese with English abstract).

Yu, X., Yang, S.F., Chen, H.L., Chen, Z.Q., Li, Z.L., Batt, G.E., Li, Y.Q., 2011. Permian flood ba-salts from the Tarim Basin, Northwest China: SHRIMP zircon U-Pb dating and geo-chemical characteristics. Gondwana Research 20, 485–497.

Zeng, G., Chen, L.H., Xu, X.S., Jiang, S.Y., Hofmann, A.W., 2010. Carbonated mantle sourcesfor Cenozoic intra-plate alkaline basalts in Shandong, North China. Chemical Geology273, 35–45.

Zhang, Z.C., Wang, F.S., 2003. A method for identifying Primary magma-examples frompicrate and alkali basalts. Journal of Jilin University (Earth Science Edition) 33 (2),130–134 (in Chinese with English abstract).

Zhang, C.L., Zou, H.B., 2013a. Comparison between the Permian mafic dykes in Tarim andthe western part of Central Asian Orogenic Belt (CAOB), NW China: implications fortwomantle domains of the Permian Tarim Large Igneous Province. Lithos 174, 15–27.

Zhang, C.L., Zou, H.B., 2013b. Permian A-type granites and Tarim and western part of theCentral Asia Orogenic Belt (CAOB): Genetically related to a common Permian mantleplume? Lithos 172–173, 47–60.

Zhang, Z.C., John, J.M., Mao, J.W., Wang, F.S., 2006. Geochemistry of picritic and associatedbasalt flows of the western Emeishan flood basalt province, China. Journal of Petrol-ogy 47 (10), 1997–2019.

Zhang, C.L., Li, X.H., Li, Z.X., Ye, H.M., Li, C.N., 2008. A Permian layered intrusive complex inthe Western Tarim Block, Northwestern China: product of a Ca. 275-Ma mantleplume? Journal of Geology 116, 269–287.

Zhang, C.L., Xu, Y.G., Li, Z.X., Wang, H.Y., Ye, H.M., 2010. Diverse Permian magmatism inthe Tarim Block, NW China: genetically linked to the Permian Tarim mantleplume? Lithos 119, 537–552.

Zhang, Y.T., Liu, J.Q., Guo, Z.F., 2010. Permian basaltic rocks in the Tarim basin, NW China:implications for plume–lithosphere interaction. Gondwana Research 18, 596–610.

Zhang, H.M., Li, S.G., 2012. Deep carbon recycling and isotope tracing: Review and pros-pect. Science China (Earth Sciences) 55 (12), 1929–1941.

Zhang, D.Y., Zhang, Z.C., Encarnación, J., Xue, C.J., Duan, S.G., Zhao, Z.D., Liu, J.L., 2012. Pet-rogenesis of the Kekesai composite intrusion, western Tianshan, NW China: Implica-tions for tectonic evolution during late Paleozoic time. Lithos 146–147, 65–79.

Zhang, D.Y., Zhang, Z.C., Santosh, M., Cheng, Z.G., Huang, H., Kang, J.L., 2013. Perovskiteand baddeleyite from kimberlitic intrusions in the Tarim large igneous provincesignal the onset of an end-Carboniferous mantle plume. Earth and Planetary ScienceLetters 361 (1), 238–248.

Zhang, D.Y., Zhang, Z.C., Huang, H., Kang, J.L., Cheng, Z.G., 2014. Genesis of Fe-Ti OxideOres inFerrobasaltic Systems: Insights from the Wajilitage and Puchang Intrusive Complexes,Northwestern Tarim Large Igneous Province. Journal of Petrology (in revision).

Zhao, Z.Y., Zhang, Z.C., Santosh, M., Huang, H., Cheng, Z.G., Ye, J.C., 2014. Early Paleozoicmagmatic record from the northern margin of the Tarim Craton: Further insightson the evolution of the Central Asian Orogenic Belt. Gondwana Research. http://dx.doi.org/10.1016/j.gr.2014.04.007.

Zhou, M.F., Malpas, J., Song, X., Kennedy, A.K., Robinson, P.T., Sun, M., Lesher, M., Keays, R.R.,2002. A temporal link between the Emeishan large igneous province (SW China) andthe end-Guadalupianmass extinction. Earth and Planetary Science Letters 196, 113–122.

Zhou, M.F., Zhao, J.H., Jiang, C.Y., Jiang, C.Y., Gao, J.F., Wang, W., Yang, S.H., 2009. OIB-like,heterogeneous mantle sources of Permian basaltic magmatism in the western TarimBasin, NW China: Implications for a possible Permian large igneous province. Lithos113, 583–594.

Zindler, A., Hart, S.R., 1986. Chemical geodynamics. Annual Review of Earth and PlanetarySciences 14, 493–571.