Deposition Timing and Provenance Analyses. Since sedimentary rocks cannot be deposited before their constituent particles are formed, the youngest detrital zircon grains in a given deposit constrain the maximum depositional age40. This “the law of detrital zircons41” is widely proved to be effective in sedimentary basins among numerous tectonic settings42,43. The MDA is consistent with the true depositional age (TDA) of sediments if their depositional sites are proximal to magmatic arcs43. Our study shows that sandstones from different outcrops in the MFAB exclusively yield Paleozoic-late Triassic ages with only a few Precambrian records. Thereinto, among the prominent early Permian (~ 290–296 Ma) age peak, a significant early Triassic population (~ 5%, Figs. 4a-b, 6c) is identified (231.7 ± 1.1 Ma, MSWD = 6.3, n = 4). This implies that the timing of deposition of studied sedimentary strata extended, at least, to the early Triassic.
To conduct the provenance analysis, 2591 zircon Lu-Hf isotopic data, together with ages, from igneous and sedimentary rocks and 7350 detrital zircon U-Pb ages from sedimentary rocks are compiled for comparison. The detrital zircon age spectrums of regional sedimentary rocks sampled on the assumed upper plate (KYCTC), the accretionary wedge (AC, including the STB) and the passive lower continental plate (NTC) are presented in Fig. 6. Sediments from both of above tectonic units (Fig. 5a) display significant amounts (~ 36–77%) of Precambrian detritus population, in contrast with those (< 15%) from Yili Basin and the MFAB in this study. For early Mesozoic-Paleozoic detrital records: 1) sediments from the STB and NTC show consistently binary age peaks, respectively, at 270–310 and 400–500 Ma, interpreted as the reflection of Tarim Permian mantle plume44–46 and temporarily active continental arc developed in the North Tarim Margin during the Silurian-Ordovician44,46−49; 2) broadly continuous early Mesozoic-Paleozoic detrital records characterize the age spectrum of sediments from the AC, KYCTC and Yili Basin, serving as the response of continuous arc magmatism during the north-ward subduction of Paleo STO14,50. Overall, none of any similarity of detrital zircon age spectrum can be identified, from above tectonic units (Fig. 5a), with respect to that of the MFAB (Fig. 6). Besides, Triassic inherited zircon ages are very rare in region, with only one report11 but with eclogite-facies metamorphic origin. This further implies that a local deposition for the detritus provenance of studied sandstones in the MFAB is probably not of possibility, unless a missed Triassic intra-oceanic arc was in fact existed, but have not been found yet. However, the relatively high ICV values (Fig. 3b) of studied sandstone, which imply an immature source typical of active continental margin setting36, and the Th-La-Hf concentrations fairly deviated from detritus provenance from oceanic island arc51 (Fig. 3c-d) essentially rules out this possibility. To be stressed, the identification of minor high-pressure mineral relics-bearing detritus (Fig. 2c-f) and corresponding ~ 315 Ma detrital zircons with Bs/Ec-facies metamorphic origin (Fig. 4c-g) in studied sandstones indicate that at least a few of detrital materials (but not the majority) are deposed from an additional proximal provenance.
A distal source is then of necessity to be considered. The western part of ETS, which extends to the most eastern side of geographic Chinese South Tianshan until the Xingxingxia Fault (Fig. 1a), is also clamped between the KYCTC and the NTC26. Considerable quantity of Triassic igneous rocks had been reported in the ETS region, especially those with adakitic geochemical affinity52,53. According to our data compilation for intermediate rocks among the Central Tianshan-Yamansu arc, two major age peaks, respectively at ~ 230–240 and ~ 280–290 Ma (Fig. 6e) characterize the ETS arc magmatism, and it is completely distinct from those in the KYCTC and the NTC. Such an age pattern, as well as relatively positive εHf(t) value, of the ETS arc intermediate rocks (Fig. 6e), in fact, broadly resembles that of detrital zircon from sandstones in the MFAB (Figs. 4a-b, 6c), implying a potential distal provenance from the ETS region. Additional evidence comes from the sub-angular to moderate-abrased shape (although not all, Fig. 2a, 2c) of most of detritus and these Triassic detrital zircon grains (Fig. S1) in sandstones, which suggests a relatively moderate to long distance of transport, for example via contour current54, prior to their deposition. This speculation is also supported by the presence of climbing ripples on sandstone outcrop (Fig. 1e) which advocate a deposing environment with high suspension load current55. Moreover, the long-lasting (late Carboniferous-early Triassic) regional nearly W-E trend strike-slip movements (as we compiled in Fig. 7f), mainly along the SCTF (Fig. 1a), could also facilitate the trench-paralleled transportation of detritus drive by contour current.
The final closure of the South Tianshan Ocean. It is of challenges to constrain the timing of final closure of a long-lived accretionary subduction system. Miscellaneous methods such as the timing of a regional unconformity, bimodal magmatism, stitching plutons, and extensional deformation were used to constrain the time of final closure of the South Tianshan Ocean, which consequently led to different conclusions. Much of viewpoints, for the issue of the final closure of the STO among the STOB in the southern Altaids, had been proposed (as reviewed by ref14). Some authors argued the final assembly of southern Altaids occurred along the North Tianshan Suture Zone in the late Carboniferous subsequent to the collision of the KYCTC and NTC during late Devonian to early Carboniferous times7,10,56,57. This model is mainly based on the following assumptions that: 1) the regional angular unconformity during late Devonian-early Carboniferous times was related to the collision of the STB and the KYCTC, 2) eclogite-facies peak metamorphism as reflected by the UHP terranes took place at ~ 350 − 345 Ma, and 3) top-to-north thrusting of the UHP rocks over the KYCTC resulted from south-ward oceanic subduction. Alternatively, some authors14 suggested the final suturing of southern Altaids was achieved by northward subduction of the South Tianshan Ocean and subsequent collision of the KYCTC and the Karakum-Tarim cratons at ~ 320 Ma, based on below arguments: 1) ~ 320–330 Ma ophiolite/ophiolitic mélange had been identified as the youngest MOR-type ophiolite along the SCTF (in Guluogou and Wuwamen areas14,58), broadly consistent with the timing of LT-UHP eclogite-facies metamorphism in Akeyazi and Atbashi areas21,22,31; 2) the latest Carboniferous molasse-type conglomerate overlying the Atbashi (U)HP rocks28, and the ~ 285 Ma post-orogenic S-type leucogranite dike crosscutting the Akeyazi UHP complex12; 3) the resumption of widespread magmatism in the STB and the NTC at ~ 270–310 Ma59–62.
It is worth noticed, yet, a significantly younger Permian (~ 265 Ma) MOR-type ophiolitic mélange63 from the Bindaban area in the eastern Central Tianshan along the SCTF, the late Carboniferous radiolarian64 in chert from Kyrgyzstan Atbashi range, and the late Permian radiolarian65 in clastic rock from Chinese western South Tianshan had been recently identified, further jointly advocating the possible existence of a much younger ocean basin in the southern Altaids. This speculation is echoed by the find of ~ 300 Ma glaucophane-bearing BS/GS facies meta-volcanoclastic units in Akeyazi metamorphic complex which suggests the subduction of the Paleo STO was probably still active during late Carboniferous22. In addition, the concept that “the metamorphic age of deeply recovered high grade protolith formation constrains the timing of oceanic closure and/or continental collision” only works in place where late exhumation associated with oceanic closure and continental subduction occurred (Yet, only found in the western Alps, New Caledonia and Central Cuba)66. This further implies that most of metamorphic age of deeply recovered high grade rock essentially has no link with the timing of oceanic closure. Recent work22 also highlighted that the final juxtaposition, at crustal level, of the diverse sub-units with various metamorphic grades making the metamorphic dome in the AMC could have occurred at around ~ 280–290 Ma, broadly consistent with the regional ~ 285 Ma crosscut post-orogenic S-type leucogranite dike12, prior to the STO closure. Moreover, the preservation of climbing ripples (Fig. 1e), petrographic characters of Triassic zircon population (and most of detritus, Figs. 2a, 2c, S2) and the detrital zircon age pattern of studied sandstones in the MFAB, jointly call for the existence of a relatively board ocean basin (Fig. 8), until the early Triassic, to facilitate relatively moderate to long distance of detritus transport, likely via contour current67 with high suspension load current55, from a potential distal provenance in the ETS region.
In short, our new data, integrated with the compilation of regional multi-disciplinary evidence, suggest that oceanic subduction could be still operating until the early Triassic in the STOB. Such Triassic closure of accessory ocean basins, which paleo-geographically belongs to the PAO, were reported among the Altaids, e.g., the Kanguer accretionary mélange in East Tianshan68, the Alxa block42, central Inner Mongolia69 and Solongker suture zone70,71, implying that the PAO closed almost synchronously along the western, central and eastern parts of the Altaids.
Trench-paralleled slab geometry “anisotropy” and geodynamic implication. Relatively fast and young oceanic plate that subducts beneath less steep plate, including horizontal slab segments, is known as the Peru-Chile type subduction72, with symbolic features like marginal subduction erosion, inboard migration of upper-plate deformation and weak to absent arc magmatism73. Its rareness74 and remarkable influences on overlying continental plate73 and economically important ore deposits75 make it of great significance and diagnostic especially for ancient orogens76,77.
To reconcile numerous controversies, we suggest the possibility of flat subduction tectonics, providing an alternative hypothesis (Fig. 8) for the geodynamic model of the STOB. Common models, whatever advocating mainly north- or south-ward subduction beneath the KYCTC12,14 or the NTC10,78, all based on a priori assumption that the down-going STO slab is ultimately coherent object with almost no topographical and geometrical “anisotropy” (i.e., differences). In fact, significant impacts on the CTS arc magmatism during the Carboniferous, by trench-perpendicular topographical “anisotropy”(i.e., seamount chain79), had been highlighted in region, calling for the possible existence of trench-paralleled slab geometry “anisotropy” during the north-ward subduction of the paleo STO.
As we presented above, studied early Triassic, HP mineral relics-bearing sandstones likely received detritus mainly from distal provenance-the ETS region whose arc magmatism characterize by similar age (broadly, also the εHf(t), Fig. 7a) pattern of Triassic population with a major early Permian peak (Figs. 4a-b, 6c). In this case, at least a part of geographical ETS region (i.e., the west side of Xingxingxia fault, Fig. 1a), which previously has not been considered yet, is non-negligible with respect to the geodynamic model of the STOB12,50,78. According to our compilation (Fig. 7), the age gap of ~ 220–280 Ma, between the continuous ETS (~ 220–330 Ma) and the intermittent CTS (~ 280–330 Ma, Fig. 5) arc magmatism, is speculated as the period of the absence of CTS arc magmatism, during which crustal thickening (Fig. 7b) and mantle “signal” soaring (Fig. 7a, 7c, 7d) in the ETS are accompanied. Such phenomenon is well observed in present-day Chilean subduction zone where area suffered from flat subduction is characterized by to-some-extent broadening (in trench-paralleled and -perpendicular directions) of magmatic arc and subsequent cessation of magmatism74,80,81. In addition, Nd-Hf isotopic decoupling (Fig. 7g) is evident for Permian arc magmatic rocks in the CTS region, suggesting a “zircon effect” because of the addition of subducted terrigenous sediments into magma sources82 and could imply the Permian arc magmatic in the CTS region was derived from a mixed mantle source due to the interaction between arc inherited magmatism and upwelling magma of oceanic lithosphere induced by flat subduction and potential subsequent slab rollback, although such a decoupling was interpreted as the inducing of the activity of Tarim Plume17. Flat subduction could be also facilitated by strong discontinuities in the oceanic (i.e., oceanic plateau and seamount chain as the buoy83) and overriding plates structure (craton and micro-continent with fairly deep Moho depth67). Similar processes had, in fact, been identified in region, coincidentally supporting the speculation of temporary flat subduction beneath the CTS region during the north-ward subduction of paleo STO: 1) Long-lasting Carboniferous-Devonian seamount chain subduction79; the relative thickened crust, which could maintain extra compositional buoyancy, of seamount, oceanic plateau and aseismic ridge effectively prohibit the slab from sinking into the mantle84,85; 2) Localized micro-continents with cratonic lithosphere were identified among the CTS region86,87, and their spatial distribution (as revealed by Hf isotope contour map; Fig. 5c) is broadly in coincidence with the area where the early Triassic to early Permian absence of CTS arc magmatism is confirmed with speculated flat subduction; numerical modeling shows that a thick cratonic root can increase the magnitude of suction acting on the subducting plate due to the mantle wedge flow and this suction effect will vary along strike if craton has a finite width88,89. Other features associated with flat subduction, for example the crustal contractional deformation (e.g., in Mexican90) and migration of arc volcanism (e.g., in Chile, Andes72), had not been observed yet in the STOB, and are probably overprinted by the pervasive nearly W-E trend strike-slip deformation22,91,92 along the SCTF (Figs. 1a, 6f).