Introduction

The Altaids or the southern Central Asian Orogenic Belt, one of the largest accretionary orogens in the world (Fig. 1a), accreted around, and grew southward from, the Siberian craton by the successive accretion of many arcs, accretionary complexes and micro-continents (Bazhenov et al. 2003; Buchan et al. 2002; Coleman 1989; Dobretsov et al. 1995; Şengör and Natal'in 1996; Şengör et al. 1993; Wilhem et al. 2012; Windley et al. 2007; Xiao et al. 2013b, 2018), but there are several contrasting and controversial tectonic models to explain that growth (Bazhenov et al. 2003; Şengör and Natal'in 1996; Windley et al. 2007; Xiao et al. 2013, 2015). Also controversial is the time of final accretion and amalgamation, which varies from the Carboniferous (Gao and Klemd 2003; Han and Zhao 2018; Qin et al. 2011; Zhou et al. 2004) to the Permo-Triassic (Huang et al. 2018; Mao et al. 2014b; Windley et al. 2007; Xiao et al. 2010; Xiao et al. 2004). Accordingly, the Eastern Tianshan, being located in the southernmost Altaids, provides critical information about the final time of amalgamation.

Fig. 1
figure 1

a Schematic tectonic map of Central Asia (Şengör et al. 1993; Xiao et al. 2018) showing the position of the Eastern Tianshan in b. b The geological map of the Eastern Tianshan showing the distribution of the main tectonic unit (modified after Xiao et al. 2004). The locations of the samples and of Fig. 2 are marked. ① Kawabulake-Xingxingxia fault, ② Aqikekuduke-Shaquanzi fault, ③ Yamansu-Kushui fault, ④ Kanguer fault

Full size image

The Eastern Tianshan is the graveyard of many island arcs and accretionary wedges (Windley et al. 1990; Xiao et al. 2010, 2004). The youngest age of the matrix of an accretionary wedge is close to the final amalgamation time, a detrital zircon U–Pb age is a useful constraint on sediment provenance (Benyon et al. 2014; Fedo et al. 2003; Matthews et al. 2017; Sharman et al. 2015), the youngest detrital zircon can constrain the MDA (Coutts et al. 2019; Fedo et al. 2003; Nelson 2001), and the lag time between zircon crystallization and deposition is commonly very short (Cawood et al. 2012). Therefore, the MDA of detrital zircons is a useful method to constrain the time of deposition of sediments in arc-adjacent basins and accretionary complexes, particularly where biostratigraphic and magnetostratigraphic ages are unavailable (Coutts et al. 2019; Sickmann et al. 2018).

The aim of this publication is to report the results of new field-based lithological mapping and structural analysis of the Kanguer mélange in the Eastern Tianshan, which has a block-in-matrix structure. We present new isotopic data of the Maximum Deposition Age (MDA) of the sandstone matrix, and new geochemical data of basaltic blocks in the mélange. The MDA of three samples is consistent, ranging from 234 ± 14 Ma to 242.5 ± 1.3 Ma, and the basalts have an N-MORB chemical affinity. With these data we demonstrate that the Kangure Ocean, a branch of the Paleo-Asian Ocean, had a far younger subduction record than previously envisaged. Therefore, the time of final amalgamation of the Altaids in the Tianshan was most likely in the late Middle Triassic (Ladinian).

Geological background

The Eastern Tianshan, situated between the Siberian craton to the north and the Tarim craton to the south, occupies the bulk of the southmost Altaids (Fig. 1a), and comprises E/W-trending accretionary belts, which include continental margin arcs, microcontinents, island arcs, ophiolites, and accretionary wedges (Qin et al. 2011; Xiao et al. 2004, 2018). The internal tectonic units of the orogen are from north to south (Fig. 1b): the Dannanhu arc, the Kanguer mélange, the Yamansu arc, the Central Tianshan block, and the Southern Tianshan accretionary complex.

The Dannanhu arc is composed predominantly of Ordovician to Permian tholeiitic basalts, calc-alkaline andesites, pyroclastic rocks, and tuffs (Du et al. 2018a, b, 2019; Qin et al. 2011; Wang et al. 2018). The Yamansu arc consists of Devonian-Carboniferous volcanic and intrusive rocks (Du et al. 2019; Luo et al. 2016), which (Chen et al. 2019) interpreted as a continental arc built on the margin of the Central Tianshan Arc. The Central Tianshan block comprises calc-alkaline basaltic andesites, volcaniclastic rocks, minor I-type granites, and a Precambrian basement (Long et al. 2020; Ma et al. 2014; Mao et al. 2018) and the South Tianshan accretionary complex contains discontinuous slices of Mid-Devonian to Early Carboniferous ophiolites (Ao et al. 2020; Gao and Klemd 2003; Sang et al. 2020).

The Kanguer mélange is situated between the Kanguer fault in the north and the Yamansu-Kushui fault in the south (Fig. 1c); it was previously mapped as the Carboniferous Wutongwozi Formation (XBGMR 1993). It is thrust-imbricated and dismembered, and thus contains blocks of serpentinite, gabbro, basalt, limestone, and chert in a meta-sandstone matrix. The Kanguer mélange has been variously interpreted as an inter-arc basin between the Dannanhu and Yamansu arcs (Han and Zhao 2018; Xiao et al. 2004), as a component in the final suture zone that separates the Siberian craton to the north and the Tarim craton to the south (Ma et al. 1997), and as an accretionary wedge in a short-lived minor ocean that formed in the Carboniferous (Wang et al. 2019).

Field observations and sampling

In order to understand the composition and structure of the Kanguer mélange, four areas from west to east were mapped in detail: the Gas pipeline station (GPS), the Haluo highway (HLH), the Yamansu West (YMW), and the Huangshan South (HSS) (Figs. 1, 2). Our resultant maps included previous map data (XBGMR 1993). We describe below the main lithologies and structures that are representative of the whole Kanguer mélange.

Fig. 2
figure 2

a A geological map of the GPS area in the Kanguer mélange showing the block-in-matrix structure. Sample locations and cross-section are marked. b Stereographic projection of cleavages from the matrix. All stereonet plots are equal-angle lower-hemisphere projections, utilizing Stereonet 11 of (Cardozo and Allmendinger 2013). c The cross-section illustrates the thrust-imbrication of blocks in the matrix of sandstone and chlorite schist

Full size image

The GPS area

The Kanguer mélange has a characteristic block-in-matrix structure (Fig. 2a). The matrix is mainly composed of chlorite-phyllite schist and meta-sandstone, which have a strong slaty cleavage that strikes almost E-W and dips moderately to sub-vertical (Fig. 2b). The primary depositional features of the matrix sediment are difficult to discern, because they have been heavily overprinted by the strong cleavage. The blocks consist of limestone, chert, sandstone, basalt, diabase, and gabbro that have been thrust-imbricated (Fig. 2c). Granodiorite intrusions crop out as elongate E/W-trending blocks (see upper left corner of Fig. 2a). The limestone and diabase blocks generally form ribbons that are several kilometers long. The cherts mostly occur in lenses that are several meters to several hundred meters long that have aspect ratios of 1:3 to 1:10, and their long axes are parallel to enclosing cleavage and to other lithologies. The gabbros and basalts form elongate blocks whose long axes are aligned sub-parallel to the main EW-trending faults. We consider the blocks of basalt, diabase, gabbro, limestone, and chert to be fragments of ocean plate stratigraphy, which were mixed and accreted in the original accretionary wedge. Samples from eleven basaltic blocks (SQ05-2/3, SQ06-1/3/4/5, SQ07-2/3, and SQ09-1/2/3) were collected (Fig. 2a) and analyzed to constrain their tectonic setting.

The HLH area

In the HLH area the Kanguer mélange has a block-in-matrix structure in which blocks of gabbro, massive basalt, pillow basalts, diabase dykes, chert, limestone, and sandstone/siltstone are embedded and imbricated in a matrix of chlorite-phyllite schist and cleaved sandstone (Fig. 3a). The matrix sandstones and chlorite-phyllites are highly cleaved and were intruded by large quartz veins indicating the movement of hydrothermal fluids during the shearing (Fig. 3a). The EW-trending, subvertical cleavage in the matrix in the northern (Fig. 3b) and central (Fig. 3c) parts of the HLH area is penetrative and has overprinted the bedding so strongly that the primary depositional structures are mostly difficult to observe.

Fig. 3
figure 3

a A geological map of the HLH area in the Kanguer mélange showing the block-in-matrix structure. The locations of Figs. 6 and 7 are marked. b, c Stereographic projections of cleavages in the mélange matrix in the northern and middle areas. d Stereographic projection of bedding planes in a sandstone block in the southern part of a

Full size image

In the central-northern parts of the HLH area, the blocks of gabbro, diabase, and basalt crop out in discontinuous lenses or ribbons ranging in length from several meters to several hundred meters. The blocks of gabbro and basalt are elongate with their long axes aligned sub-parallel to the main bounding faults of the mélange. The cherts generally crop out as lenses ranging in length from several meters to a kilometer with aspect ratios from 1:1.5 to 1:7 with their long axes parallel to the adjacent cleavage. The contacts between the blocks and matrix are thrusts. The blown sandstones, red cherts, and massive green basalts have a clear imbricate structure indicating they have been thrust to the south (Fig. 6a, b). Both the basalt and chert blocks are more strongly cleaved close to their bordering thrust planes (Fig. 7a, b).

Sandstone blocks in the south of the HLH area are huge—several kilometers wide and tens of kilometers long. The limestone blocks form ribbons or discontinuous lenses ranging in length from several meters to tens of kilometers. The sandstones and limestones retain their primary sedimentary bedding, and appear to be locally coherent in the Kanguer mélange. The thick sandstone blocks are intruded by elongate granodiorite dykes that are folded but only weakly cleaved; the long axes of the granodiorite intrusions are parallel to the regional cleavage. Folded bedding planes indicate tight to isoclinal sub-vertical folds in sandstone (Fig. 3d), the axes of which are parallel to the east–west cleavage of the matrix. Overall, these structures indicate a regional N–S compression.

The thrusts and cleavages in the entire HLH area of the Kanguer mélange are displaced by late NW/SE-trending dextral strike-slip faults. The blocks of sandstone and basalt are thrust slices, which belong to top-to-the-south thrust duplexes (Figs. 6, 7). All these structures indicate that the blocks of basalt, diabase, gabbro, limestone, sandstones and chert probably formed in different environments, and were inter-mixed during addition to the accretionary complex. Three sandstone samples (08K01, 02, and 03) were collected in the HLH area and their detrital zircons were analyzed for U–Pb dating, Hf isotopes and whole-rock major and trace elements. Thirteen basalt samples (HL01-1/3/4, HL05-1/2, and HL07-14) were collected (Fig. 3a) for whole-rock geochemical analyses to constraint their tectonic setting.

The YMW area

In the YMW area the Kanguer mélange has a block-in-matrix structure and has been intruded by granodiorite plutons (Fig. 4). The sandstone/pelite matrix has an E-W cleavage that is so pervasive that primary depositional structures are rarely preserved. The blocks are mainly basalts that range from hundreds of meters to several kilometers in size, and a few ribbon lenses of limestone and chert. In the north of the YMW area the basaltic blocks are elongate with their long axes aligned sub-parallel to the main thrust faults, but in the south the basalt blocks are juxtaposed and separated only by thrusts that are marked by ribbons of foliated limestone and chert. The southern part of the YMW area possibly contains dismembered remnants of oceanic plate stratigraphy, and the whole YMW area likely came from a high-strain section of an accretionary complex. Eighteen basalts were sampled (YM16-1/2/4/5/6, YM17, et al.) for geochemical analysis to constrain the tectonic setting of the Kanguer mélange.

Fig. 4
figure 4

A geological map of the YMW area in the Kanguer mélange showing two discordant granodiorite intrusions, and the sample locations

Full size image

The HSS area

In the southern HSS area the Kanguer mélange also has a block-in-matrix structure (Fig. 5a). The blocks are mainly of basalt, limestones, chert, and sandstone. The blocks of limestone and basalt are elongated with their long axes sub-parallel to the main enclosing faults. The cherts generally outcrop as lenses that range in length from several meters to a kilometer, and their aspect ratios range from 1:5 to 1:11. The contacts between the blocks and matrix are thrusts. The margins of the blocks are typically fractured and foliated; the foliation is parallel to the long axes of the blocks and to the cleavage of the matrix. The matrix consists of chloritic to phyllitic meta-siltstone/sandstone, which is cleaved and locally folded (Fig. 5b, c). The bedding planes outline tight to isoclinal sub-vertical folds (Fig. 5b), the axes of which are parallel to the E/W-trending cleavage of the matrix. In the northern part of the area the chloritic-phyllitic siltstones/sandstones and the basalts belong to thrust imbricates, which indicate top-to-south compression. The whole Kanguer mélange in the HSS area is transected by several NE/SW-trending dextral strike-slip faults. Five basalts (HS19-2/3/4/6/7) were sampled from the area for geochemical analysis.

Fig. 5
figure 5

a A geological map of the HSS area in the Kanguer mélange showing lens-shaped blocks in the matrix, a discordant granodiorite intrusion and the sample locations. b, c Stereographic projections of the planes of bedding and cleavage of the matrix

Full size image

Geochemical results

Whole-rock analysis

Whole-rock major and trace elements of the basalts are listed in Appendix Table 1.

The major element compositions of eleven whole-rock tholeiitic basalts (Fig. 10b) from the GPS area vary across a relatively narrow range with SiO2 = 46.0–48.5 wt%, Al2O3 = 14.2–15.3 wt%, CaO = 8.8–10.9 wt%, and MgO = 6.8–7.9 wt%. These basalts exhibit relatively high Cr (81–252 ppm) and Ni (62–103 ppm) concentrations. The Na2O concentrations are somewhat variable, possibly reflecting seawater alteration. Rare earth element (REE) patterns of the basalts (Fig. 9a) are comparable to those of N-MORBs with (La/Yb)N contents ranging from 1.7 to 2.0, and with positive Ba and Sr anomalies and a negative Gd anomaly (Fig. 9b).

Thirteen basalt samples from the HLH area have relatively narrow elemental ratios with SiO2 = 43.7–48.6 wt%, Al2O3 = 13.1–16.4 wt%, CaO = 7.3 –11.6 wt%, and MgO = 6.3–8.4 wt%. These basalts exhibit relatively high Cr (184–228 ppm) and Ni (76–111 ppm) concentrations. The samples are altered and thus have variable contents of LOI ranging from 2.2 to 40.5 wt%. Na2O concentrations in the whole rocks are variable from 1.42 to 4.01 wt%, likely reflecting seawater alteration. REE patterns (Fig. 9c) are similar to those of N-MORBs with (La/Yb)N ratios between 0.8 and 2.0; some basalts have positive Ba, Sr, and Zr anomalies and a slightly negative Gd anomaly (Fig. 9d).

Eighteen basalts from the YMW area have contents of SiO2 = 41.9–46.7 wt%, Al2O3 = 10.7–17.8 wt%, CaO = 8.5–14.3 wt%, and MgO = 3.6–11.9 wt%. These basalts exhibit relatively high Cr (293–817 ppm) and Ni (115–894 ppm) concentrations. Na2O concentrations in whole rocks are variable from 0.9 to 4.2 wt%, maybe reflecting seafloor alteration. REE patterns (Fig. 9e) are comparable to those of N-MORBs with (La/Yb)N ratios between 0.5 and 1.7, and they have positive Ba and Sr anomalies and negative Gd and Ti anomalies (Fig. 9f).

The major element compositions of five basalts from the HLH area have variable contents of SiO2 = 42.1–52.9 wt% and CaO = 5.8–9.1 wt%, and relatively narrow ranges of Al2O3 = 12.2–14.1 wt%, and MgO = 2.3–3.8 wt%. These basalts exhibit slightly low Cr (33–40 ppm) and Ni (34–51 ppm) concentrations. REE patterns (Fig. 9g) are like those of N-MORBs, with (La/Yb)N ratios of 0.6–0.7, and they have a positive Ba anomaly and negative Sr and Ti anomalies (Fig. 9h).

Sr–Nd isotopes

Ten whole-rock basalts were analyzed for their Sr and Nd isotopes; the results are listed in Appendix Table 2 and shown in Fig. 12b. Two basalts from the GPS area have high initial (87Sr/86Sr)i ratios of 0.706040 and 0.706876, and εNd (t) values of + 8.30 and + 9.34. Five basalts from the HLH area have an initial (87Sr/86Sr)i ratio that varies from 0.701248 to 0.707631, and εNd (t) values of + 8.33 and + 9.45. Three basalts from the YMW area have an initial (87Sr/86Sr)i ratio that ranges from 0.703681 to 0.705359, and εNd (t) values of + 9.52 and + 9.89. Sr is a mobile element during seawater alteration (Hart and Staudigel 1982), thus most of our samples were likely modified by seawater alteration.

Zircon U–Pb age and Hf isotopes

A total of 379 analyses of zircon grains from four sedimentary samples from the Kanguer mélange (Figs. 3, 6, 7) yielded 304 concordant ages (concordance % > 90% or < 110%). All the U–Pb and Lu–Hf isotopic data are shown in Appendix Tables 3 and 4, respectively. Only concordant ages are described and discussed below.

Fig. 6
figure 6

a Field photo of a long cliff face that gives a structural cross-section showing multiple thrusts and imbrication of chert and basalt layers in the Kanguer mélange. b An illustrative sketch of the cliff section shown in (a). Sample locations and labels are marked. Jeep for scale

Full size image
Fig. 7
figure 7

a Cross-section showing the typical imbricate structure of thrusted layers and lenses of chert and basalt of the Kanguer mélange. b A sketch cross-section explaining the details in a. Sample locations and labels are marked. The legends are same to those of Fig. 6. Jeep for scale

Full size image
Fig. 8
figure 8

Micro-photos of dated samples. a Sample 08K01, showing poor sorting, high proportions of detrital altered matrix. Grains are angular and subrounded. b Sample 08K02 and c Sample 08K03, showing weak-orientated and poor sorting, high proportions of detrital and diagenetically altered matrix and a mix of quartz, feldspar and lithic framework fine grains

Full size image

Zircons from sample 08K01 are 100 μm long, have length/width ratios of 1.3–2.0, prominent zones in CL images, and they have variable contents of Th (28–1623 ppm) and U (69–5065 ppm), and Th/U values of 0.12–1.35. Of one hundred and sixty analyzed zircon grains, one hundred and ten yielded concordant ages with major concordant age peaks at 288.1 ± 0. 56 Ma (60.4% of the total), a secondary peak at 433.3 ± 0. 97 Ma (32.4% of the total), and scattered ages at 1209.2 ± 5.8 Ma (7.2% of the total) (Fig. 11a). The youngest three zircon ages are 209 ± 16 Ma, 228 ± 5 Ma, and 236 ± 2 Ma, which give a mean age of 234 ± 14 Ma (Fig. 11b). This mean age is interpreted as the MDA of the sandstone. Lu–Hf isotopic analyses of the detrital zircons yielded εHf(t) values ranging from − 8.1 to + 13.2 (Fig. 12a).

Fig. 9
figure 9

Chondrite-normalized REE (a, c, e, and f) and Primitive Mantle (PM) normalized multi-element (b, d, f, and g) diagrams of basalts from the Kanguer mélange belt. Chondrite values are from (Boynton 1984). The PM values are from (Sun and McDonough 1989). All the blocks of basalt from the GPS, HLH, YMW, and HSS areas in the Kanguer mélange are similar to those of N-MORBs

Full size image
Fig. 10
figure 10

Diagrams used for discrimination of tectonic settings of basalts from the Kanguer mélange belt. a The Zr/Ti vs. Nb/Y diagram after (Floyd et al. 1998), b AFM diagram after (Irvine and Baragar 1971), c Th/Yb vs. Nb/Yb after (Pearce 2008) and d V vs. Ti/1000 diagrams after (Shervais 1982). These diagrams demonstrate that the basalts in the Kanguer mélange belt were probably generated in a Mid-Ocean-Ridge

Full size image
Fig. 11
figure 11

a, c, e, g Histograms and normalized kernel density plots of the ages of detrital zircons in four sandstones in the HLH area from the Kanguer mélange (see Fig. 3 for sample locations and supporting information and Appendix Table 3 for data sources). Ages of 206Pb/238U and 1σ, which concordance % > 90% or < 110% are used to density plots by DensityPlotter software Version 8.5 (Vermeesch 2012). b, d, f, h MDAs for each sample. The MDA of each sample to indicate that the overall deposition was at 234 Ma in the late Middle Triassic (Ladinian), which indicates that the Kanguer Ocean still existed at 234 Ma. The youngest three ages of 206Pb/238U and 1σ, which concordance % > 90% or < 110% for each sample are used to calculate the mean age by Isoplot/Ex 2.2 (Ludwig 2001)

Full size image
Fig. 12
figure 12

a Hf isotopic diagram of dated sandstones from the Kanguer mélange showing that the sandstone samples (08K01, 02, and 03) were probably derived from the Dananhu intra-oceanic arc in the north. b εNd(t) vs. (87Sr/86Sr)i diagram of basalts from the Kanguer mélange, showing that the basalts were probably generated in a Mid-Ocean-Ridge, where erupted lavas were influenced by seawater penetration. All initial 87Sr/86Sr values are calculated to t = 290 Ma

Full size image

Zircons from sample 08K02 are dark, irregular, have a grain-size of 120 μm, and clear zones in CL images. The zircons contain variable contents of Th (54–659 ppm) and U (59–2102 ppm), and Th/U values of 0.23–1.28. Of sixty analyzed zircon grains, fifty-nine yielded concordant ages with two major concordant age peaks at 288.32 ± 0. 55 Ma (47.5% of the total) and at 415.23 ± 0.82 Ma (45.8% of the total), and scattered ages at 879.1 ± 4.4 Ma (6.3% of the total) and 2361 ± 22 Ma (1.7% of the total) (Fig. 11c). The youngest three zircon ages are 241 ± 2 Ma, 243 ± 2 Ma, and 243 ± 3 Ma, which yield a mean age of 242.5 ± 1.3 Ma (Fig. 11d), which we interpret as the MDA of the sandstone. Lu–Hf isotopic analyses of the detrital zircons yield εHf(t) values ranging from − 12.6 to + 12.1 (Fig. 12a).

Zircons from sample 08K03 are irregular, are about 100 μm in size, and have sharp zones in CL images. The grains show variable contents of Th (24–1953 ppm) and U (44–3760 ppm), and Th/U values of 0.04–1.95. One hundred and sixty-one zircon grains were analyzed of which one hundred and thirty-five yield concordant ages with major and minor concordant age peaks at 292.18 ± 0. 52 Ma (64.9% of the total) and 436.9 ± 1.4 Ma (26.9% of the total), respectively (Fig. 11e). They also have scattered ages at 1001.6 ± 4.9 Ma (4.4% of the total) and 1510 ± 13 (3% of the total). The youngest three zircon ages are 236 ± 2 Ma, 237 ± 6 Ma, and 240 ± 5 Ma, which yield a mean age of 236.9 ± 2.0 Ma (Fig. 11f), which we interpret as the MDA of the sandstone. Lu–Hf isotopic analyses of the detrital zircons yield εHf(t) values ranging from − 16.5 to + 23.7 (Fig. 12a).

Discussion

Tectonic setting of the Kanguer mélange

The Kanguer mélange is characterized by block-in-matrix structures and has been thrust-imbricated. Blocks in the mélange consist of gabbro, massive basalt, pillow basalt, diabase dykes, chert, limestone, and sandstone, which are enclosed in a matrix of chlorite-phyllite schist and cleaved sandstone (Figs. 2, 3, 4, 5). On a kilometer scale, blocks of different lithologies are repeated several times and exhibit top-to-the-south imbricate structures (Figs. 2b, 6, 7). The sandstone blocks (Fig. 3a, d) and the matrix (Fig. 5) contain sub-vertical isoclinal folds, the axes of which strike east–west. All these structures indicate that the Kanguer mélange underwent intense top-to-the-south thrusting and shearing. The blocks of basalt, diabase, gabbro, limestone, and chert are probably fragments of oceanic plate stratigraphy, which, if intact, would have provided information on the travel history of the oceanic plate from ridge to trench (Kusky et al. 2013); the fragments are inter-mixed in the mélange and accreted into the accretionary complex (Festa et al. 2019; Raymond 2019; Wakabayashi 2019). Although no ultramafic blocks/serpentinized or otherwise crop out in the Kanguer mélange, some ophiolites were reported by Li et al. (2008) in the mélange to the northwest of our study area. Thus, the lithologies, field relationships, and structural characteristics indicate that the Kanguer mélange is a classic ophiolite-bearing tectonic mélange that formed at a convergent margin and was created by subduction-accretion processes (Kusky et al. 2020; Wakabayashi 2015; Xiao et al. 2015) (Fig. 8).

All the blocks of basalt from the GPS, HLH, YMW, and HSS areas in the Kanguer mélange are tholeiitic and have consistent geochemical compositions, suggesting they formed by the same petrogenetic processes. Their REE patterns (Fig. 9a, c, e, g) are similar to those of N-MORBs, as also are the similar trace element patterns with their positive Rb, Ba, and Sr anomalies (Fig. 9b, d, f, h). Their enrichments in fluid-soluble elements and ratios such as Sr–Nd isotopes indicate seawater alteration (Kelley et al. 2003) (Fig. 12b). In Th/Yb–Nb/Yb and V-Ti/1000 diagrams (Fig. 10c, d), most of the basalts plot in the field of N-MORB, but some in the island arc field, probably because of fluid enrichments due to seawater alteration (Furnes and Safonova 2019; Reagan et al. 2010). All the geochemical and Sr–Nd isotopic features demonstrate that the basalts in the Kanguer mélange belt were probably generated in a Mid-Ocean-Ridge, where erupted lavas were influenced by seawater penetration. We envisage that most of the basaltic blocks in the Kanguer mélange belt were scraped off a downgoing slab of oceanic crust, which had been generated in a Mid-Ocean-Ridge.

MDA of the Kanguer mélange

There are several methods for calculating the MDA of sedimentary rocks from their detrital zircon U–Pb ages (Coutts et al. 2019). Here we use the youngest three grains that overlap at a 2σ uncertainty. The sandstone samples (08K01, 08K02, and 08K03) yield mean ages of 234 ± 14 Ma, 242.5 ± 1.3 Ma, and 236.9 ± 2.0 Ma, respectively (Fig. 11b, d, f). All the youngest three grains overlap in age at a 2σ uncertainty, which indicates that the mean age for each sample is robust. The lag time between zircon crystallization and real deposition can be very short in some geologic settings, e.g., when a sediment is sourced from an active magmatic arc (Cawood et al. 2012). We interpret the MDA of each sample to indicate that the overall deposition was at 234 Ma in the late Middle Triassic (Ladinian).

The youngest MDA of the sandstone matrix of the Kanguer mélange belt was in the mid-Triassic, which indicates that the Kanguer Ocean still existed at 234 Ma, and that the final accretion of the Kanguer mélange was after the Ladinian.

Provenance of the Kanguer mélange

The Kanguer mélange is located between the intra-oceanic Dannanhu arc in the north and the Yamansu-CTS arc in the south (Fig. 1b). Therefore, these intra-oceanic island arcs were the main potential provenance for the sandstone matrix of the Kanguer mélange.

In the cross-sections of Fig. 6a and b the three sandstones (08K01, 02 and 03) have similar and consistent detrital zircon U–Pb age populations (Fig. 11a, c, e) with dominant age peaks in the period 288–292 Ma, a second peak at 415–436 Ma, and some scattered Precambrian ages (less than 8% of the total concordant ages for each sample). Lu–Hf isotopic analyses of these detrital zircons yielded εHf(t) values from − 16.5 to + 23.7 (Fig. 12a). The Dannanhu intra-oceanic arc contains major Permian and Ordovician magmatic rocks (but no Precambrian rocks), which are largely tholeiitic basalts and calc-alkaline andesites and granitoids that have subduction-related geochemical features with positive whole-rock εNd(t) and zircon εHf(t) values (Du et al. 2018a; Mao et al. 2019; Mao et al. 2014b; Qin et al. 2011; Wang et al. 2018; Yuan et al. 2010). The Dananhu arc is interpreted as an intra-oceanic arc that formed in the early Ordovician (Du et al. 2019; Xiao et al. 2004). Both the detrital zircon age spectra and εHf(t) values are comparable with the magmatic record in the Dananhu intra-oceanic arc. The sedimentary samples (08K01, 02, and 03) were probably derived from the Dananhu intra-oceanic arc in the north, which means that Kanguer mélange probably formed in the forearc of the Dananhu intra-oceanic arc.

Provenance variations in the Kanguer mélange

As discussed above, the Kanguer mélange is an accretionary complex, which indicates that the matrix of the mélange was scraped off from sediments of the downgoing slab. So, the provenance variations of the matrix can shed light on the spatiotemporal framework of the arc and accretionary complex.

The Kanguer Ocean subducted to the north and generated the Kanguer ophiolite (494 Ma) in a supra-subduction zone in the Late Cambrian (Li et al. 2008) and the Dananhu intra-oceanic arc in the Ordovician (Xiao et al. 2004). Therefore, the Kanguer mélange probably formed as early as the Ordovician. Chen et al. (2019) reported detrital zircon U–Pb age spectra in late Devonian sedimentary rocks in the Dananhu arc which have a unique age peak at ~ 400 Ma with no Precambrian ages (Chen et al. 2020), which suggests that the Kanguer Ocean between the Dananhu intra-oceanic arc and the Yamansu-CTS composite continental arc was quite wide in the Devonian. Our Triassic samples (08K01, 02, and 03) from the Kanguer mélange received sediments mainly from the Dananhu intra-oceanic arc in the north and very few from the Yamansu-CTS composite arc in the south, suggesting that the Kanguer Ocean was sufficiently narrow for their proximity. This provenance variation supports the idea that the Kanguer mélange formed in the forearc of the Dananhu intra-oceanic arc.

Tectonic evolution and implications for the Altaids

The Kanguer mélange belt has been interpreted as: (1) an inter-arc basin between the Dannanhu and Yamansu arcs (Han and Zhao 2018; Xiao et al. 2004); (2) a mélange in the terminal suture zone between the approaching Siberian craton and Tarim craton (Ma et al. 1997); (3) a short-lived limited ocean rifted on the accretionary wedge (Wang et al. 2019). The third interpretation is inappropriate, because, if the rifting was caused by an upwelling mantle plume, the basalts in the Kanguer mélange should display an oceanic island basalt (OIB) affinity, and the contact between basalts and their wall-rocks should be intrusive. However, all the basalts in this study have an N-MORB signature and have been thrust within mélange matrix. The Kanguer ophiolite was reported to have formed in a supra-subduction zone at 494 Ma (SHRIMP zircon U–Pb age) (Li et al. 2008), which suggests that the Kanguer Ocean was open and being subducted in the late Cambrian. Therefore, the Kanguer Ocean was a long-lived branch of the Paleo-Asian Ocean in the southern Altaids.

The final closure time of the Paleo-Asian Ocean is controversial: 1. late Carboniferous (Han and Zhao 2018; Wang et al. 2016; Wu et al. 2016; Zhang et al. 2005); 2. late Permian to early Triassic (Chen et al. 2019; Huang et al. 2018; Mao et al. 2020; Mao et al. 2014a; Xiao et al. 2018; Zhang et al. 2007; Zheng et al. 2020). The late Carboniferous time of final closure of the Paleo-Asian Ocean in the Eastern Tianshan is mainly based on the following evidence: (1) No ophiolites or ultrahigh-pressure metamorphic rocks younger than 300 Ma have been found in the eastern North Tianshan suture (Han and Zhao 2018). (2) an 40Ar/39Ar age of 278–254 Ma on biotite and hornblende from granitic plutons reflects collision-induced regional metamorphism and deformation (Wang et al. 2008). However, this evidence is circumstantial and not robust. First, the formation age of an ophiolite only indicates the time when oceanic crust was generated and does not provide information on the time of ophiolite emplacement and the ocean closure (Coleman 1981; Dilek and Furnes 2011). Second, the 40Ar/39Ar age of biotite and hornblende can also be interpreted as the time of a magmatic reset of the Ar–Ar system, particularly in view of the fact that ca. 275–267 Ma mafic-intermediate intrusions occur in the same area (zircon U–Pb ages, (Wang et al. 2014)). Moreover, juvenile magmatic arc accretion and continental growth are characteristic of the Altaids, which was commonly produced by oceanic slab subduction and intrusion in accretionary prisms (Jahn 2004; Jahn et al. 2000; Windley et al. 2007; Xiao et al. 2015). Therefore, only the age of the youngest subduction-related intrusions can be used to define the final amalgamation age in a long-lived accretionary orogenic belt such as the Altaids.

Accordingly, we update our tectonic model of the Eastern Tianshan in the southern Altaids (Fig. 13) using our new and published data in the light of the above discusssion. Our new detrital zircon U–Pb ages of matrix sandstones indicate that one major branch of the Paleo-Asian Ocean, the Kanguer ocean, was still open at ca. 234 Ma. The youngest 234 Ma age of the matrix in the Kanguer accretionary mélange that contains blocks of N-MORB basalts provides a snapshot of the mid-Triassic northward subduction of the Paleo-Asian Ocean in the Eastern Tianshan (Fig. 13a). This tectonic scenario is consistent with the presence of 243–234 Ma adakites, which were generated by slab-melted, lower crust of the Dananhu arc (Mao et al. 2020). In the Middle to Late Triassic (234–217 Ma), the Kanguer ocean closed, and the Dananhu intra-oceanic arc was amalgamated and welded with the Yamansu-CTS continental arc (Fig. 13b). Large-scale syn- to post-collisional granites intruded in the Eastern Tianshan in the Late Triassic (Li et al. 2012; Wang et al. 2016).

Fig. 13
figure 13

Tectonic model showing two stages of evolution in the Eastern Tianshan. a In the Permian to Middle Triassic, the Kanguer ocean, a branch of the Paleo-Asian Ocean, remained open and was subducting on both sides. The blocks of N-MORB-affinity, diabase, limestone, and chert were accreted to the fore-arc of the Dananhu arc of the growing Kanguer accretionary complex, in which detrital zircons in the youngest matrix sediments have MDAs of 234 Ma. b In the Middle to Late Triassic (234–217 Ma), the Kanguer ocean closed by double subduction, and the Dananhu intra-oceanic island arc was amalgamated with the Yamansu-CTS continental arc. Large syn- to post-collisional granodiorite plutons intruded the two arcs and their accretionary complexes

Full size image

Combined with the Mid-Late Triassic peak metamorphic ages of eclogites in the Western Tianshan (Sang et al. 2017, 2020; Zhang et al. 2007), we consider that the final time of amalgamation of the southern Altaids, recorded in the whole Tianshan orogenic belt, was most likely in the Late Triassic.

Conclusions

(1) Field relations and map data indicate that the Kanguer mélange has a characteristic block-in-matrix structure and it underwent intense top-to-the-south thrusting and shearing. The blocks of basalt, diabase, gabbro, limestone, and chert are probably fragments derived from a downgoing oceanic plate, which were accreted to the evolving accretionary complex.

(2) Geochemical and isotopic data of basalts from the Kanguer mélange belt have the same diagnostic signatures as those of N-MORBs.

(3) The MDAs of sandstones from the matrix of the Kanguer mélange belt are 234 ± 14 Ma, 242.5 ± 1.3 Ma, 236.9 ± 2.0 Ma and 274 ± 3 Ma, which indicates that the Kanguer Ocean was still open at 234 Ma.

(4) The basalts and sandstone occur in imbricate thrust slices, suggesting the accretion of the Kanguer mélange belt should be later than 234 Ma. The final time of amalgamation of the southern Altaids in the whole Tianshan orogenic belt was probably in the mid-late Triassic.