In this study, we investigate the age and geochemical variability of volcanic arc rocks found in the Chinese, Kyrgyz, and Tajik North Pamir in Central Asia. New geochemical and geochronological data together with compiled data from the literature give a holistic view of an early to mid-Carboniferous intraoceanic arc preserved in the northeastern Pamir. This North Pamir volcanic arc complex involves continental slivers in its western reaches and transforms into a Cordilleran-style collision zone with arc-magmatic rocks. These are hosted in part by Devonian to Carboniferous oceanic crust and the metamorphic Kurguvad basement block of Ediacaran age (maximum deposition age) in Tajikistan. We discuss whether a sliver of Carboniferous subduction-related basalts and intruded tonalites close to the Chinese town of Mazar was part of the same arc. LA-ICP-MS U-Pb dating of zircons, together with whole rock geochemistry derived from tonalitic to granodioritic intrusions, reveals a major Visean to Bashkirian intrusive phase between 340 and 320 Ma ago. This clearly postdates Paleozoic arc-magmatic activity in the West Kunlun by ~100 Ma. This observation, along with geochemical evidence for a more pronounced mantle component in the Carboniferous arc-magmatic rocks of the North Pamir, disagrees with the common model of a continuous Kunlun belt from the West Kunlun into the North Pamir. Moreover, Paleozoic oceanic units younger than and west of the Tarim cratonic crust challenge the idea of a continuous cratonic Tarim-Tajik continent beneath the Pamir.

A common model for plate tectonic reconstructions of northern Tibet and the Pamir has been the assumption of a continuous magmatic belt extending from the West Kunlun into the northern Pamir [1, 2]. However, comparisons between the Paleozoic-early Mesozoic evolution of the poorly studied North Pamir and the adjacent, well-documented West Kunlun belt in northern Tibet reveal significant differences; these likely explain the different Cenozoic deformation styles of the adjacent regions. The Pamir orogen is the westward extension of the Tibetan Plateau. A common hypothesis is that the Pamir indented into the Tajik-Tarim basin in a late phase of the India-Asia collision, pushing the Pamir several hundred kilometers toward the north with respect to Tibet (e.g., Burtman and Molnar [1] and Schwab et al. [3]). The Pamir and Tibet formed during the Phanerozoic as a result of successive accretions of Gondwana-derived crustal blocks. Today, the Pamir and Tibet are part of the India-Asia collision zone—the largest active continental collision. At the longitude of the Pamir, the highest modern strain rates are found far to the north, along the Main Pamir Thrust (MPT) [4, 5]. In contrast, to the east, N-S shortening rates within northern Tibet are much smaller [6, 7]; the bulk of the convergence is accommodated, along the southern margin of Tibet, within the Himalaya. To understand the Cenozoic MPT and hence the difference between Pamir and Tibetan deformation styles, it is crucial to understand the pre-Cenozoic geologic evolution.

The Pamir has traditionally been subdivided into the North, Central, and South Pamir terranes (Figure 1(a)). The North Pamir terrane was subdivided into the North Pamir Kunlun, which includes middle Paleozoic basalt, gabbro, and felsic plutons, and the North Pamir Karakul-Mazar terrane, which represents a late Paleozoic-early Mesozoic accretionary wedge intruded by felsic plutons [3, 8]. Despite strong deformation and fault dissection, the Chinese Gez valley is today the best-studied locality in the North Pamir. The Karakul-Mazar terrane was correlated with the Songpan-Ganzi-Hoh Xil complex of northern Tibet [9], while the North Pamir Kunlun was correlated with the South Kunlun terrane of the West Kunlun in Tibet (e.g., [10, 11]). The West Kunlun is subdivided into the North and South Kunlun terranes, divided by the early Paleozoic Kudi suture. The North Kunlun corresponds to the margin of the Proterozoic Tarim block [12, 13]. This was described from the Kudi section, the best-studied section crossing the North and South Kunlun south of the town of Kargilik [11, 14, 15]. However, there is little similarity between the North Pamir Kunlun and the South Kunlun terrane of the West Kunlun.

There is an ongoing debate about how and which units of the Tibetan Plateau/West Kunlun are the along-strike equivalent of units within the Pamir plateau. Two of those seemingly laterally contiguous structures—the Kudi-Oytag ophiolites and ophiolites along the Tanymas-Jinsha structure—were interpreted as the remnants of former suture zones related to the closure of the Proto- and Paleo-Tethys [1, 2, 3, 10, 11, 16]. Based on this interpreted lateral continuity, estimates of the amount of largely Cenozoic northward indentation of the Pamir orogen, with respect to the Tibetan Plateau, were made (e.g., more than 300 km by Burtman and Molnar [1], 300–400 km by Burtman [17]). Well-documented early Paleozoic magmatism, of either syn- or postcollisional nature, is thought to be related to the closure of the Proto-Tethys along the Kudi suture zone [1820], which finally closed between 440 Ma (monazite U-Pb age of biotite schist from the Saitula group [13]) and 405 Ma (zircon U-Pb age of the A-type North Kudi Pluton [14]). This was followed by a Late Permian to Triassic intense magmatic phase related to the subduction of the Paleo-Tethys and the collision of the Central Pamir-Qiangtang block with Asia [14, 19, 2123], culminating in the formation of the Tanymas-Jinsha suture zone (Figure 1) between 243 Ma (zircon U-Pb age of anatectic Yuqikapa pluton [21]) and 190 Ma (metamorphic zircon U-Pb age population in amphibolite facies metasediments from Karakul-Mazar accretionary complex [8]). A phase of volcanic quiescence has been proposed between the Silurian and Triassic in the NE Pamir and the West Kunlun [19, 24].

In the North Pamir Kunlun, however, oceanic mafic to intermediate volcanic rocks and marine cherts of Upper Devonian to Bashkirian age [2, 3, 2527] can be found. This narrow, ca. 400 km long zone (Figure 1(a)) spans from the Chinese Gez and Oytag (also called Wuyitake or Aoyitake) valleys [2831] along the strike of the frontal Pamir mountain chain to the Tajik town of Kalai Khumb and continues into the northern reaches of the Hindu Kush/Badakhshan region in Afghanistan [3235]. Oceanic rocks were assigned to the Kalai Khumb-Oytag basin (KOB), a marginal basin of the Paleo-Asian ocean [36]; this has also been named the Kunlun arc [3] or the Pamir arc [33, 37]. Leucocratic granitoids, including tonalites and trondhjemites, are found as large intrusions within mafic volcanic rocks and are dated as Visean to Bashkirian (338–314 Ma [2831]). These units are inferred to represent the remnants of an intraoceanic arc that marks a phase of intraoceanic subduction and the initial closure of an ocean basin [29]. Granitoids in NW Afghanistan also intrude lower Carboniferous marine strata and yield K-Ar ages between 335 and 360 Ma ([38] cited in [32]).Whether the small occurrence of mafic volcaniclastic rocks and associated leucogranites of similar age, found in the Mazar tectonic mélange zone [39] east of the town of Mazar (Figure 1(a), “East Mazar”), is part of the KOB must be discussed; this would increase the eastward extension of the basin.

In this paper, our new geochemical and geochronological data combined with our literature compilation demonstrates that the term North Pamir Kunlun is misleading. There is no evidence for a lateral continuation of the North Pamir Devonian to Carboniferous arc-magmatic sequence into the South Kunlun. Therefore, herein, we use the term North Pamir arc as previously used by Bazhenov and Burtman [33], rather than the North Pamir Kunlun.

Rock units of the North Pamir arc experienced variable degrees of greenschist to lower amphibolite facies metamorphic overprint; higher metamorphic units potentially associated with Pennsylvanian to Permian subduction processes might have been identified by Li et al. [40] in the Tajik North Pamir arc (see Discussion). Moreover, a nonmetamorphic sedimentary succession of upper Permian to Eocene ages overlies Carboniferous rocks in the Chinese Qimgan valley (Figure 2) and shows no sign of a major post-Carboniferous collisional event affecting the NE Pamir. Therefore, the Permo-Triassic Qimgan basin [41] formed on a fragment of Carboniferous oceanic crust that is now situated in the External Pamir and was affected by thin-skinned deformation in Cenozoic time.

Following the initial observation that the North Pamir arc of the NE Pamir has many similarities with units further west [42] and little in common with the West Kunlun to the east, the aim of this study is to better constrain the age and tectonic affinity of the Carboniferous arc-related rocks. Within this contribution, we present new and compiled geochemical and geochronological data from locations in the Tajik, Kyrgyz, and Chinese North Pamir and document that they have a common temporal evolution. We compare this data with the well-known tectonothermal events recorded in the West Kunlun. Thereby, we can test existing models of the northernmost suture zone in the Pamir and its relation to units in the West Kunlun to the east and the Hindu Kush/Badakhshan to the west.

2.1. Overview

Within this study, we compare outcrops, literature data, and new field observations from the following sites (maps in Figure 1(a) and profiles in Figure 2): (1) the Badakhshan region in Afghanistan, (2) the Kalai Khumb and Obikhingou area in Tajikistan (Figure 1(b)), (3) the Kyrgyz Altyn Darya valley (Figure 1(c)), (4) the Chinese NE Pamir (Taergelake valley to Oytag valley, Figure 1(d)), and (5) the East Mazar area. Our study area is focused on locations (2)–(4). However, from excellent, detailed field descriptions, geological observations, and maps made during the last century [32, 33, 43, 44], significant similarities between the Carboniferous units exposed in Badakhshan and the Tajik Pamir can be inferred. This was proposed and discussed by previous studies [37, 4446].

2.2. Badakhshan

From northwestern Afghanistan, the Badakhshan and the southeastern Takhar regions, Lower Mississippian calc-alkaline lavas, volcaniclastics [32, 43], and coral bearing limestones [44] are known (Figure 2). In the Surkhab (aka Kunduz river) valley, Lower Mississippian units are composed of low-grade metamorphosed amygdaloid-basalts, andesites, and tuffs [33]. Above an angular unconformity are nonmetamorphic Visean limestones [32]. These are separated by a second angular unconformity from upper Pennsylvanian to Permian units dominated by littoral to sublittoral clastics and regional marls and platform limestones (Figure 2 [32, 44]). The age of the dioritic intrusive rocks in the Badakhshan region is inferred from relative geologic relationship with disconformably overlying Permian sedimentary rocks [47] and from K-Ar ages of 335–360 Ma from diorite and granodiorite found in the Surkhab river valley ([38] cited in [32]). Diorites, granodiorites, and granites intruded the sequence in the Fakhar area in the Khanabad river valley during the Mississippian [48].

2.3. Kalai Khumb and Obikhingou

A Silurian to Permian succession was described along the Panj and Khingob/Obikhingou river valleys and tributaries located in the northwest Tajik Pamir (Figure 2 [49]). Silurian to Devonian sediments are the oldest exposed low-grade metamorphic units in the region [34, 35, 49]. A low-grade metamorphosed ophiolitic sequence described as Carboniferous [50] is part of a nappe sheet overlying the metamorphosed Kurguvad basement block. The Kurguvad basement block was mapped as Proterozoic in age [51]. Peak metamorphic conditions of the Kurguvad metamorphic suite were 540–650°C and 5.5–7.6 kbar; concordant monazite 206Pb/238U and 208Pb/232Th ages are between 210 and 195 Ma [52]. All of the pre-Carboniferous units are covered by andesitic and basaltic pillow lavas, volcaniclastics, and marly interbeds dated as upper Serpukhovian [53]. Gabbros and leucogranites intruded this sequence. The Carboniferous gabbros and plagiogranites were grouped into the Obikhumbou complex [54]. This lower succession was previously interpreted as an oceanic island arc [36]. Above an erosional unconformity, an amagmatic Bashkirian carbonate sequence covers the top of the volcanic sequence [53]. Our observations (below) show that the section has been metamorphosed to greenschist or lower amphibolite facies. A granitic intrusion covering more than 50 km2 is also known from the Kurguvad basement block.

2.4. Altyn Darya

In the Altyn Darya valley of the Trans-Alai range, basic to acidic volcanic rocks were described (Figure 2 [3]). 40Ar/39Ar hornblende dating of two andesite samples yielded ages of ~356 Ma; zircons from two rhyolite samples yielded a Serpukhovian U-Pb ID-TIMS age of 329 Ma [3]. This volcanic section is overlain by an upper Pennsylvanian to Permian sedimentary sequence consisting of carbonates with intercalated tuff and shales, as previously described from the Sauksai and Beleuli river sections. No unconformity between the lower and upper Carboniferous is reported [55]. Our petrographic observations show that the section reached greenschist facies metamorphic conditions.

2.5. Chinese NE Pamir: Taergelake, Qimgan/Akqi, Gez, and Oytag

Large volumes of Devonian to Mississippian volcaniclastic rocks are mapped in the Chinese Taergelake, Qimgan/Akqi, Gez, and Oytag valleys (Figure 2 [27]). Large leucogranite intrusions are mapped in these locations. In Gez and Oytag, these provided ages between 338 and 314 Ma by zircon U-Pb dating [2831] and are interpreted as island arc derived [29]. The Qimgan/Akqi section exposes a rather complete, unmetamorphosed sequence. We found upper Permian greenish-grey fine clastics and rare conglomerates and cross-bedded sandstones disconformably overlying the Carboniferous volcaniclastics (see Figure 2 and ages in Figure 1). The occurrence of mafic to intermediate volcaniclastic rocks and lava flows in the Permian succession decreases upsection. We interpret the lower, fine-grained part of the sequence as floodplain deposits. Dark, silicified carbonates with monotonous ostracod fauna and remnants of charophytes are interpreted to represent a lacustrine environment. Plant detritus is abundant in siltstones. They are overlain by red, often cross-bedded coarse-grained clastic rocks, interpreted as alluvial deposits of a terrestrial environment. Apparently, parts of the Pennsylvanian and the lower Permian sequence are missing in this section or are extremely condensed compared to the Altyn Darya section.

New field observations from the Bostantielieke valley, between Qimgan/Akqi and Gez, show that the Permian sequence there was overthrusted by a sedimentary sequence containing thick greywackes, shales, phyllites, and prominent layer of dark marls, containing unidentified goniatites, crinoid fragments, brachiopods, and mollusks. We interpret these deposits as fully marine sediments. They are in turn overthrusted by greenschists, amphibolites, and marbles of the Karakul-Mazar derived Shala Tala nappe [56, 57]. Foliation in the greenschist pervasively continues into the underlying dark marls. Therefore, we interpret both nappes to have been emplaced in the same episode. Lineation and microstructures indicate nappe transport to the NNE to NW under ductile conditions (Appendix 6).

2.6. East Mazar

A tectonic horse, to the north east of the town of Mazar, was described as a Mississippian mafic to intermediate volcanic sequence (Figure 2 [39]). The section comprises altered basaltic to andesitic lava flows, intermediate to acidic tuffs, and volcanic breccias which were intruded by granitoids. The outcrop is crosscut by multiple faults. It is disconformably overlain by a thick Triassic conglomerate layer.

3.1. Fieldwork and Sample Preparation

Data from 27 rock samples collected from the Tajik, Kyrgyz, and Chinese North Pamir (Table 1) are presented in this study. We analyzed 12 granitoid samples and 8 mafic and intermediate volcanic rocks for petrology, geochemistry, and geochronology. The samples are from Carboniferous units in the Chinese and Tajik North Pamir as well as two Carboniferous volcanic samples from the Kyrgyz Altyn Darya valley. One sample is from a large granodiorite intruded into the Tajik Kurguvad block. To constrain the paleogeographic affinity of this poorly described basement block, two garnet bearing paragneiss samples were analyzed. Petrographic thin sections were made from all samples. Zircons were separated from 8 granitoid samples, two paragneiss samples, and one aplitic dike sample using a jaw crusher, disc grinder, water table, magnetic separation, and heavy liquids (SPT, DI). The zircons were poured onto a glass plate and arranged in lines on double-sided sticky tape under a binocular microscope. In-line mounting in epoxy helps for better single-grain recognition. Mounted grains were polished to expose an internal surface and imaged with cathodoluminescence (CL) using the electron microprobe facility at the University of Potsdam (UP), Germany.

Two carbonate samples, collected from the Chinese Qimgan basin, were cleaned, cut, and polished. Both zircon and calcite were dated with U-Pb geochronology using a laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS) at the School of Earth and Environmental Sciences, the University of Queensland.

A portion of the granitoid samples and all volcanic samples from the Chinese and Kyrgyz Pamir were also processed for whole-rock geochemistry (Table 2). They were cleaned, crushed, and milled in an agate mill to a particle size <62 μm. Melt tablets for X-ray fluorescence spectroscopy (XRF) analysis to measure major and trace elements were prepared at UP using fluxing agent FX-X65-2 (lithium tetraborate : lithium metaborate, 66 : 34). Samples for rare earth elements (REE) and yttrium and scandium measurements were sintered with sodium peroxide at 480°C and dissolved in hydrochloric acid. REE plus scandium and yttrium were then separated in ion exchange columns. Powder tablets were prepared from seven granite samples to determine the mineral composition by X-ray powder diffraction (see Table 1).

3.2. LA-ICP-MS Zircon U-Pb Geochronology

262 zircon grains from 8 granitoid samples were dated (Appendix 1(a) and Appendix 7). Additionally, we dated 109 zircons from two paragneiss samples RT15-11-and RT13-148 and 56 zircons from an aplitic dike (17NP439). Paragneiss sample RT15-11 was mislabeled during processing as a granite; therefore, only 51 zircons were mounted. Zircon 91500, with a 206Pb/238U age of 1062.4±0.4Ma and 206Pb/207Pb age of 1065.4±0.3Ma [58] was used as a primary reference material. Temora 2 zircons, with a 206Pb/238U age of 416.78±0.33Ma [59], was used as secondary reference material. Laser ablation was carried out with an ASI RESOlution 193 nm ArF excimer laser system in the Radiogenic Isotope Laboratory (RIF) at the University of Queensland. After air evacuation, He carrier gas was introduced into the laser chamber at a flow rate of 0.35 l/min. A flow of 0.005 l/min N2 gas was also introduced into the laser chamber to enhance the sensitivity of the measurements. The gas mixture was then transferred into the plasma torch of a Thermo iCAP RQ quadruple ICP-MS with 0.85 l/min Ar nebulizer gas. No reaction gas was employed. The laser spot size was 30 μm in diameter. Laser frequency was set to 10 Hz, with a measured instrument laser fluence of 2.9 J/cm2. For each ablation spot, 3 s of blank was collected, followed by 20 s of ablation and 5 s of wash out. Before starting data acquisition, the ICP-MS signals were optimized by signal tuning. The zircons were measured during two analytical sessions, in October 2018 and in October 2019. Typically, we achieved over 300 cps for 238U counts, ~1 for 238U/232Th, and 0.22–0.25 for 206Pb/238U when measuring NIST612 glass using a line scan of 3 μm/s, 10 Hz, 50 μm round laser pit, and 3 J/sm2. Laser spot locations on the sample zircons were carefully chosen using CL images. Fractures and zones with strongly differing Th/U values were avoided.

The following isotopes were counted (dwell time in brackets): 88Sr (0.005 s), 91Zr (0.001 s), 200Hg (0.01 s), 204Pb (0.01 s), 206Pb (0.045 s), 207Pb (0.055 s), 208Pb (0.01 s), 232Th (0.01 s), and 238U (0.01 s). As a single cycle takes 0.155 s, during a 20 s ablation, approximately 120 measurements are taken for each mass. Reduction of the raw data was done in the program Iolite v2.5, which runs within the Igor Pro environment [60, 61] using the VizualAge [62] data reduction scheme. The primary standard zircon 91500 was used to bracket each five unknown analyses to correct for machine drift. 91500 data were not corrected for common lead. Temora 2 zircons were treated as unknowns. The age of the primary standard was reproduced as a 206Pb/238U age of 1062.7±0.2Ma, and Temora 2 gave a 206Pb/238U age of 420.70±0.21Ma and 417.80±0.69Ma in the 2018 and 2019 measurement sessions, respectively. All unknowns were filtered for concordance and strontium content. LA-ICP-MS U-Pb dating and data processing were similar to these described in Zhou et al. [63].

3.3. LA-ICP-MS Calcite U-Pb Geochronology

Two limestone samples were chosen for calcite U-Pb age determination (Appendix 7). Suitable ablation spots were identified by thin section examination of the samples. In situ calcite U-Pb analysis was also performed using the RIF LA-ICP-MS system. An ASI Resolution 193 nm excimer UV ArF laser ablation system equipped with a dual-volume Laurin Technic ablation cell was employed with an on-sample fluence of ~3 J/cm2 and a spot size of 100 μm. All samples were measured using a Thermo iCAP RQ quadruple ICP MS. Analyses consisted of 250 pulses at a repetition rate of 10 Hz. Each analysis consists of 20 s background acquisition followed by 30 s of sample ablation and 7 s washout. We used NIST 614 to correct for 207Pb/206Pb fractionation and for instrument drift in the 206Pb/238U ratio [64]. Data reduction and production of carbonate U-Pb isotopic ratios were undertaken with the software Iolite v2.5 [60, 61]. An in-house carbonate reference material (AHX-1a) of known age (209.80±0.48Ma[65, 66]) is used for normalization of 206Pb/238U ratios. Corrected data were regressed on Tera-Wasserburg plots using IsoplotR [67] software to calculate the lower intercept ages. Our protocol is similar to recent carbonate in situ U-Pb LA-MC-ICP-MS dating reported elsewhere [6870], with the exception of using a different calcite reference material for the 206Pb/238U mass-bias correction.

3.4. XRF and ICP-AES Whole Rock Geochemistry

XRF analysis (X-ray fluorescence spectroscopy) was done at the GeoForschungsZentrum Potsdam (GFZ) on a PANalytical AXIOS Advances XRF system. Calibrations were validated by analysis of international reference materials. “Monitor” samples and 130 certified reference materials were used for the correction procedures. The detection limit of the XRF system is generally 10 ppm. The rare earth elements plus yttrium and scandium were measured by inductively coupled plasma atomic emission spectroscopy (ICP-AES) at UP on an Agilent ICP5100 machine following the procedure of [71]. Long-term precision for ICP-AES at UP is generally <5%. High field strength elements were measured at the GFZ using standard ICP-MS procedures [72, 73]. The detection limit is generally ±5% [74]. Concentrations of H2O and CO2 were determined from 20 mg of sample powder, weighed in tin foil. A Euro EA 3000 Elemental Analyser at UP was used for analysis.

3.5. XRD Mineral Phase Analysis from Powder Samples and EDX Mineral Phase Analysis from Thin Sections

Mineral analysis of 7 granitoid samples were obtained at UP using a PANalytical Empyrean powder X-ray diffractometer (XRD) with a Bragg-Brentano geometry. The XRD is equipped with a PIXcel1D detector using Cu K_α radiation (λ=1.5419Å) operating at 40 kV and 40 mA. θ/θ scans were run in a 2θ range of 4–70° with step size of 0.0131° and a sample rotation time of 1 s. It was equipped with a programmable divergence and antiscatter slit and a large Ni-beta filter. The detector was set to continuous mode with an active length of 3.0061°. The total detection time was 21 min.

Double-polished thin sections of garnet bearing paragneiss from the Kurguvad basement block were analyzed with the scanning electron microscope (SEM: JEOL JSM-6510 SEM) at UP. Samples were carbon coated prior to analysis. The SEM is equipped with an Inca-Xact Energy Dispersive X-ray detector (EDX) from Oxford Instruments. INCA analysis software [75] was used to distinguish the element distribution in the minerals. Analyses were performed with a measurement time of 60 s using a tungsten cathode at 20 kV.

4.1. New and Previous Findings from Field Geology and Petrology

Field investigations show that the North Pamir arc experienced three phases of volcanism and related igneous intrusions. The first phase featured the emplacement of basaltic to andesitic lava flows, containing pillow lavas, mafic to intermediate tuff layers, and cherts [3, 28]. These voluminous volcanic deposits can be found along the strike from Oytag in China into North Afghanistan [33]. As previously described, the first period of volcanism was followed by the intrusion of voluminous plagiogranites [30]. In a third phase, the plagiogranites were crosscut by mafic to intermediate dikes [29]. Those dikes can be found in the Oytag plagiogranites as well as in the granitic intrusion in the Taergelake valley. Our field data and U-Pb data from the Chinese Qimgan basin, northwest of Oytag, show that the basaltic to andesitic Carboniferous lava flows are overlain by an upper Permian to Triassic continental sequence (Figure 2 [41]). The lower part of the sequence is characterized by fine-grained greenish clastics; subordinate cross-bedded, red alluvial sandstones; and up to two-meter-thick lacustrine carbonate layers. The entire lower part of the Permo-Triassic continental basin hosts a large number of mafic to intermediate lava flows. We have obtained a new maximum age constraint for the continental basin by dating upper Permian sedimentary carbonate and an aplitic dike in the lower part of the section.

4.1.1. Granitoids

Several large leucocratic granitoid bodies are mapped along the length of the North Pamir arc (Figure 1). They are all dominated by quartz and plagioclase; some show minor amounts of K-feldspar. Twelve granitoid samples from different intrusions in the North Pamir arc were analyzed in this study (thin section photographs in Figure 3 and Appendix 3 and modal mineral content in Appendix 10). They are medium to coarse grained and of light grey to pale green color (Figure 3). They are generally composed of 25–45 vol% of quartz and 35–55 vol% of plagioclase. Sample 17NP426 from the Oytag plagiogranite with a gneissic fabric yields higher amounts of quartz (60 vol%), while the plagioclase cumulate sample 16NP342 (Appendix 3(g) and (h)) from an outcrop close to the Tajik town of Kalai Khumb shows almost no quartz. Sample RT13-146 of a granite gneiss, found in the Kurguvad block in Tajikistan, is composed of 50 vol% quartz and just 10 vol% plagioclase. Plagioclases show common polysynthetic twinning and saussuritization up to complete replacement of the primary crystals (Figure 3(d)). In several samples, plagioclase starts to be replaced by epidote from its center (Figure 3(e)). Gez plagiogranites have strongly zoned plagioclase (Figure 3(f)). Orthoclase is largely absent and was only found in larger amounts (≤10 vol%) in samples 15NP247, RT13-146, and RT13-149. Tajik plagiogranites have a significant amount of primary biotite. The biotite grains show green to brown colors under plane light. Coarse, greenish grains are interpreted as magmatic biotite that is in part replaced by brown, metamorphic biotite. In the Chinese samples, biotite and other mafic minerals are rare. Samples from the Tajik Pamir (RT13-108, RT13-115) host euhedral, unoriented aggregates of metamorphic stilpnomelane (Appendix 3(f)). In three Tajik plagiogranite samples (RT13-108, RT13-115, and 16NP341), we found minor amounts (≤10 vol%) of altered amphiboles (Appendix 3(a) and (b)). Amphiboles have light brown pleochroic colors and are often replaced by epidote. Samples RT13-146 and RT13-149 have a two-mica mineralogy with brown biotite and light grey to colorless muscovite (Figure 3(c)). All samples contain minor amounts of primary and secondary opaque minerals. Sample RT13-109 contains secondary calcite. Samples RT13-146 and 17NP426 plot in the quartz-rich granitoid field of the Streckeisen diagram [76]. The Kurguvad granitoid samples RT13-146 and RT13-149 and the Taergelake sample 15NP247 are classified as granodiorites; all other samples are classified as tonalites when plotted in the Streckeisen diagram. All samples show evidence for ductile straining, as quartz grains show undulous extinction accompanied by various degrees of subgrain rotation recrystallization. Feldspars show idiomorphic habit and only gentle fracturing in the strained samples. Exceptions are the samples RT13-146 and RT13-149 from the Kurguvad intrusion. Their feldspars are fine-grained and show recrystallisation patterns (Appendix 3(c)). In the mylonitic sample 17NP426, feldspar formed an augen gneiss fabric and started to recrystallize to epidote and chlorite in the pressure shadows. From deformation patterns, we estimate a maximum temperature of clearly above 300°C and below 500°C for most samples and maximum temperatures above 500°C for RT13-146 and RT13-149 from the Kurguvad granite. Due to the abundant epidote and chlorite, we presume a greenschist facies metamorphic overprint of all granitoids of the North Pamir arc. Peak metamorphic conditions of the Kurguvad samples are estimated as lower amphibolite facies.

4.1.2. Volcanic Rocks

The North Pamir arc hosts a wide variety of volcanic and volcaniclastic rocks (thin section photographs in Appendix 2). We examined 8 samples from basaltic to andesitic lava flows from Altyn Darya and the Qimgan valley. The lava flows are fine-grained and dark brown to green. Abundant vesicles are filled with secondary minerals, such as calcite. The greenish color and the abundance of calcite in all samples suggest metasomatism and spilitization. The samples from Altyn Darya show a clear greenschist facies metamorphic overprint and ductile straining. The lava flow samples show similar microscopic features: plagioclase, as phenocrysts or microliths, is abundant, with hornblende and rare olivine as phenocrysts. The samples show hyalopilitic textures, sometimes trachytic textures.

4.1.3. Paragneiss

We made thin sections from two Kurguvad paragneiss samples, RT13-148 and RT13-185 from the Panj valley. Quartz and biotite are the dominant mineral phases (thin section photographs in Appendix 4), accompanied by garnet and minor amounts of feldspar and muscovite. Quartz and feldspar show dynamic recrystallisation structures. The ~150 μm garnets in sample RT13-148 have an inclusion-rich core and a clear rim. Garnets in sample RT13-185 are remnants of larger garnets (up to 500 μm) that were replaced partially by chlorite and biotite (Appendix 4(a) and (b)). The remaining garnet fragments have a dark core, packed with small inclusions and a clear rim. The complex structure of the garnets in RT13-185 hints at a multistage metamorphic history of the Kurguvad paragneiss.

4.1.4. Qimgan Aplitic Dike and Carbonates

Carboniferous pillow basalts are disconformably overlain by a predominantly clastic sequence in the Chinese Qimgan valley. A light grey, aplitic dike was sampled in the stratigraphically lower part of the Qimgan basin (Figure 2). It crosscuts upper Permian fine-grained clastic strata. The primary mineral phases have been substantially replaced by secondary minerals. A primary porphyritic texture can be inferred from the matrix replaced by fine-grained, dirty secondary minerals and phenocrysts replaced by brownish-colored secondary minerals.

In the lower part of the Qimgan basin, we collected two carbonate samples for in situ U-Pb dating (Figure 4). Sample 17NP438 is a sparry, light grey carbonate pebble from a red, conglomeratic, clay-rich debris flow deposit. The conglomerate contained clasts of serpentinites and sparry carbonates. The sampled carbonate clast contains planktonic foraminifera (Figure 4(a)) and unidentified shell fragments. Therefore, we interpret the carbonate pebbles to be derived from marine carbonate deposits. The primary fabric of carbonate sample 17NP472 is overprinted by sparry calcite. However, ostracod shells and remnants of lacustrine algae can be discerned (charophytes, Figure 4(c)). The carbonate strata, interpreted as lacustrine based on their fossil content, are interbedded by fine-grained greenish siltstones and mudstones and occasional cross-bedded sandstones. Most carbonate layers show silicification and chert nodules.

4.2. New and Published Radiometric Age Data

An overview of the LA-ICP-MS U-Pb age data from all thirteen samples measured for this study as well as a compilation of literature data is provided in Table 2 (Figure 5). Detailed single-grain information from our own data is available in 1(a) and Appendix 7.

We report three ages for the igneous samples: (1) the age of the youngest grain, (2) the median age of the largest age cluster calculated using the TuffZirc function of [77], and (3) the peakfit function of [67]. (2) and (3) give largely similar results for our data sets. For (1) and (2), we used the more conservative discordance criteria d=1206/238/207/206; this is recommended as both isotope ratios are measured directly in LA-ICP-MS dating and the lead isotope ratio is more sensitive to lead loss. Method (3) uses an age-dependent discordance criterion, i.e., for grains<1000Ma, d=1206/238/207/235, and for grains>1000, d=1206/238/207/206. For the detrital zircon samples, we report ages (1) and (3). The two carbonate samples gave lower intercept ages in the Tera-Wasserburg diagram. Generally, peakfit ages were preferred for interpretations. Therefore, we take into account that single young discordant grains may have experienced lead loss but still are mathematically concordant due to their large age error [78] or due to the very narrow space between concordia and the discordia line for young metamorphic events. Zircons from the granites and the aplitic dike show clear magmatic zoning in CL imaging. The zircons of paragneiss samples RT13-148 and RT15-11 are in part metamict. They do not show metamorphic overgrowths.

4.2.1. Granitoids

We compiled zircon U-Pb ages from the Chinese [2831, 39] and Kyrgyz North Pamir [3] and reinterpreted these ages based on the criteria described above (see Table 1). The Chinese studies focus on the tonalite outcrops in the Gez and Oytag (Aoyitake) valleys 80 km SW of Kashgar. Of the published age data from the Oytag and Gez plagiogranites in the Chinese Pamir, we extracted those ages which fall into a range of ±10% of discordance. Those grains reveal two age populations: a younger one around 330 Ma and an inherited component of Ordovician to Silurian age [29]. The younger age population is often left skewed, suggesting lead loss [78]. Therefore, we do not use the youngest concordant zircon age for our interpretation. Previously published data of samples from the Oytag tonalite gave ages between 314 Ma and 331 Ma, the published data from samples in the Gez valley yielded an age of 338 Ma.

Our zircon U-Pb analyses of samples from the Tadjik plagiogranites gave ages between 322 Ma and 340 Ma. Samples from the westernmost, most extensive intrusion (~370 km2), spanning from Darvaz in the south to the Khingob valley in the north, gave ages between 334 Ma to 339 Ma. The samples taken from intrusions further to the east—two from an intrusion cropping out in the Khingob valley and one from the Kurguvad granite—gave slightly younger ages between 322 Ma and 334 Ma. Contrasting to those large intrusions, sample 15NP245 from a seemingly smaller intrusion in the Chinese Taergelake valley, 120 km southwest of Kashgar, gave an age population of about 360 Ma. Zircons younger than 360 Ma from that sample do not form a coherent age group and are interpreted to be affected by lead loss.

4.2.2. Volcanic Rocks

There is almost no geochronologic data available in the literature concerning the volcanic sequence of the North Pamir arc. Two andesites from the Kyrgyz Altyn Darya valley yielded lower Mississippian ages of 355 Ma and 357 Ma (hornblende 40Ar/39Ar). Rhyolites dated in the same study with ID-TIMS on zircon gave a U-Pb age of 329 Ma [3].

4.2.3. Paragneiss

Two samples were taken from the metamorphic Kurguvad basement complex in Tajikistan. RT15-11 is from a unit mapped as granite. It was identified as a metasediment from the hand specimen; zircon ages show a similar distribution as detrital grains from paragneiss sample RT13-148 (Figure 6(a)). Both samples, RT15-11 and RT13-148 have age peaks at around 580 Ma, 722 Ma, and 943 Ma and two minor age peaks at 2.0 Ga and 2.6 Ga (Figure 6, Appendix 1(b) and (c)). Age peaks were calculated for both samples together, using the discrete mixture modelling algorithm [79] implemented in the peakfit function of IsoplotR [67]. We show that the Kurguvad complex was intruded by a tonalite (RT13-146) at around 322 Ma. Therefore, a concordant grain in RT15-11 which yielded an age of 200.8±1.8Ma was classified as metamorphic and not included in the age peak calculation. The limited number of measured detrital zircon grains (RT15-11: 51 mounted, 25 measured; RT13-148: 120 mounted, 82 measured) allows for identification of major age peaks.

4.2.4. Qimgan Aplitic Dike and Carbonates

To constrain the minimum age limit of the activity of the North Pamir arc, we dated two carbonates and a crosscutting aplitic dike from the lower part of the Qimgan basin, 50 km to the northwest of the Oytag valley. The aplitic dike (sample 17NP439) yields two zircon age populations: 250 Ma and 417 Ma (Appendix 1(d) and (e)). It crosscuts the lower part of the Qimgan basin. The inherited, older age peak is younger than inherited grains reported from the Gez plagiogranite (448 Ma and 468 Ma [29]). Carbonate samples were collected from a red conglomerate containing serpentinite and limestone pebbles (17NP438) and a dark lacustrine limestone (17NP472). Sample 17NP438 gave an age of roughly 347 Ma from 60 single ablation spots; sample 17NP472 yielded an age of 260 Ma from 34 single ablation spots (Figures 4(b) and 4(d)). These are lower intercept ages from the Tera-Wasserburg diagram. Outliers were carefully excluded in order to optimize the goodness of fit (weighted MSWD value).

4.3. New and Published Whole Rock Geochemistry

We compiled published whole rock geochemistry data for granitoids from the Oytag [28, 29, 31] and Gez sections [29] in the Chinese North Pamir and a tectonic sliver near Mazar [39] along the Karakax fault system in the Chinese West Kunlun. There is an extensive database of major element whole rock geochemical data from intrusive rocks from the North Pamir arc in Tajikistan by Mamadjanov et al. [54] with which we compare our data and the data from the Chinese granitoids (Figures 7(a) and 7(b)). They interpret the occurrence of gabbro and tonalites/plagiogranites as a continuous series of five major phases: emplacement of (1) gabbro, (2) quartz diorites, (3) tonalites, (4) plagiogranites, and (5) leucoplagiogranites. Mamadjanov et al. [54] report trace elements as bulk analysis; those data were not used in our comparison. The granitoids intruded in most places into a thick pile of basic to intermediate volcanic and volcaniclastic rocks.

We also compiled whole rock geochemical data from volcanic rocks from the Oytag section [28] in the Chinese North Pamir, East Mazar [39] in the Chinese West Kunlun, and the Altyn Darya valley [3] in Kyrgyzstan (Figure 8). There is a large similarity amongst the volcanic rocks from the Chinese North Pamir and the Altyn Darya valley. Therefore, we present and plot our whole rock geochemical data from the volcanic rocks together with literature data from the Chinese volcanic rocks in Oytag and Gez and Altyn Darya, as they complement each other. The new geochemical data from this study are presented in Appendix 8.

4.3.1. Granitoids

The granitoids of the North Pamir arc show SiO2 content between 63 and 78 wt%, MgO values between 0.3 wt% and 3.1 wt%, and Al2O3 content between 12 and 16 wt%. They show relatively high Na2O values between 4 and 6 wt% and low K2O values between 0.1 and 4.2 wt%. They can be classified as peraluminous to metaluminous using the A/NK versus A/CNK plot of Shand (Figure 9 [80]). CIPW normed samples from the Oytag/Gez plot in the trondhjemite field of the albite-anorthite-orthoclase diagram (Figure 7(b) [81]). Samples from the Tajik plagiogranites show a larger variability and higher normative anorthite (Ca) content (RT13-149, RT13-111) or higher normative orthoclase (K) content (RT13-108, RT13-115, and 15NP247). Samples RT13-146 and RT13-149, which show orthoclase and plagioclase in the thin section, plot along the tonalite-granodiorite border. The plagioclase cumulate sample 16NP342 from the Tajik Kalai Khumb intrusion is an exception, with a low SiO2 content of 43 wt% and high Al2O3 (27 wt%), MgO (5 wt%), and CaO (15 wt%) content. All samples can be classified as calc-alkaline. Rare earth element (REE) data normalized to chondrite [82] from the Oytag tonalites plot parallel to the N-MORB composition (Figure 7(c)) and show no to slightly negative Eu anomalies. The Tajik granitoids show enriched light REE compared to chondrite, more pronounced depleted Eu, and lower amounts of heavy REE compared to the Oytag tonalites. Plagioclase cumulate sample 16NP342 shows very low REE amounts and a strong positive Eu anomaly. C1 chondrite [83] normalized La/Lu ratios fall between 0.6 and 1.6 in the Chinese tonalites and between 4 and 10.5 in the Tajik granitoids. The Taergelake granite has a normalized La/Lu ratio of 10.8. The granite and two monzonite samples from East Mazar have the highest normalized La/Lu ratios of 11.8, 14.2, and 17.6 [39]. Trace element data reveals Rb content between 0.6 and 14 ppm in the Gez and Oytag tonalites [28, 29, 31] and between 26 (16NP341) and 131 ppm (RT13-115) in the Tajik granitoids. Ni content generally ranges between 0.34 and 3.20 ppm with enrichment in RT13-111 (15.73 ppm) and 16NP342 (48.51 ppm). Th content is between 1.03 ppm (17NP426) and 23.37 ppm (RT13-108). U concentrations are between 0.29 ppm (17NP426) and 4.67 ppm (RT13-108). Plagioclase cumulate sample 16NP342 shows a very primitive signature with Th content of 0.04 ppm and U content of 0.01 ppm. Plotting the granitoids in the Rb versus Y and Nb classification scheme [84], most samples fall in the field of volcanic arc granites (Figure 10). The granitoids show a typical arc signature; however, this signature is not homogenous along the strike of the North Pamir arc. The tonalites in Oytag are depleted and show a strong mantle influence, while the Tajik and East Mazar granites are more enriched.

4.3.2. Volcanic Rocks

The SiO2 content in volcanic rocks from the North Pamir arc ranges from 38 wt% in the Qimgan valley to 54 wt% in Altyn Darya. These samples show Al2O3 content between 13 and 17 wt% and MgO content between 3.7 and 9.3 wt%. K2O content is relatively low (0.05 to 1.59 wt%) while Na2O values are high (1.8 to 5.4 wt%). They can be classified as metaluminous using the A/NK versus A/CNK plot of Shand (Figure 9 [80]). As is evident from hand specimens and thin section observations, all volcanic samples were altered. Ocean floor metasomatism caused spillitization of the basaltic to andesitic lava flows. The volcanic rocks in the Altyn Darya valley have been strongly affected by greenschist facies metamorphism. General effects of low-T/low-P overprints on geochemistry are described elsewhere [8587]. In all samples, we presume that mobile elemental abundances have been disturbed. Therefore, classification schemes based on mobile components should be handled with care. All samples fall within a tholeiitic series based on the FeOt/MgO versus silica diagram (Figure 8(a) [88]). However, samples from the Altyn Darya valley show higher FeOt/MgO ratios than the samples from the Gez and Oytag valleys but are similar to those from Qimgan. The K2O versus silica plot does not show such a trend, as potassium, similar to sodium, rubidium, and strontium, is highly mobile even during low temperature alteration [89, 90]. To get a robust geotectonic classification, we use the multiple major element classification scheme of Verma et al. [91]. The scheme is only applied to samples with SiO2<52wt%. Most samples plot in the island arc basalt field (Figure 11(a)). Likewise, in the Zr/Y versus Y plot [92, 93], most samples plot in the island arc basalt field (Figure 11(b)). Rare earth element (REE) data normalized to chondrite [82] show variable patterns, varying from flat to slightly depleted light REE to moderately enriched light REE pattern (Figure 8(c)). C1 chondrite [83] normalized La/Lu ratios fall between 1 and 2 for most samples (Figure 8(b)). A basalt from Oytag shows the lowest ratio of 0.5, similar to N-MORB [94]. Most andesites from Altyn Darya (e.g., AD1a, AD7c, and AD6e [3]) have values above 2; samples from East Mazar (D534/6, D1029/1-4, and D1029/1-3 [39]) have the highest values between 6.9 and 8.5. Enrichment of light REE (Figure 8(c)) may be connected to contamination with continental material. Cr values range between 178 ppm (15NP233-1) and 431 ppm (15NP234); two samples yield Cr<10ppm (2011T49 and 01-09-13-01).

4.4. Results of XRD Analysis of Selected Granite Samples

Powder XRD analysis was performed on seven granitoid samples to determine the generally very fine-grained secondary and alteration mineral phases. When the abundance of mica in the sample is low, white and dark mica cannot be distinguished by their X-ray diffraction spectra. Therefore, data must be cross-checked with results from petrographic thin section examination. Results are summarized in Table 1. The XRD measurements confirm the presence of typical greenschist facies mineral assemblages. Chlorite and epidote are common. The spectra of sample 16NP342 show evidence of clinozoisite and zoisite. The spectra of sample RT13-111 suggest the presence of actinolite. All plagioclases are identified as albite and in some samples as oligoclase. Potassium feldspars are of microclinic composition. XRD results agree well with whole rock CIPW normed feldspar classification (Figure 7(b)).

4.5. Results of EDX Analysis of Garnets from Kurguvad Basement Block

We used EDX to analyze polished thin sections of two biotite-garnet gneiss samples from the Kurguvad basement block. They yielded two types of garnet. Semiquantitative EDX analysis on a scanning electron microscope reveals gentle zoning for the large, multistage garnets of sample RT13-185 and stronger zoning for the small, clearer garnets from sample RT13-146. Garnets show high almandine (XFe2+) values around 80 mol% in sample RT13-185 and around 70 mol% in sample RT13-146. The pyrope (XMg) values are generally highest in the core, and spessartine (XMn) values increase toward the rim (Appendix 5).

5.1. Ages and Tectonic Setting

Our new geochemical results and compiled data confirm the existence of a volcanic North Pamir arc, which was active from the latest Devonian to Bashkirian—for at least 50 Ma. Basaltic to andesitic rocks are characterized by typical arc tholeiitic compositions and low normalized La/Lu ratios. Granitoids with tonalitic to granodioritic composition intruded the volcanic sequence.

A minimum age for arc volcanism initiation in the North Pamir is given by the 360 Ma Taergelake granodiorite (sample 15NP245, Figure 5) and 40Ar/39Ar hbl ages of ~357 Ma from andesites in the Altyn Darya valley [3]. The 360 Ma Taergelake intrusion has the oldest U-Pb age and the highest REE enrichment of all Chinese granites in this study. This is remarkable, as it represents a more evolved granitoid compared with the later-intruded Oytag tonalite, for example. So far, there is no good explanation for this. The younger age limit is presently defined by the youngest plutons in the Oytag valley, Chinese North Pamir [28], by the onset of upper Permian continental deposition in the Chinese Qimgan basin (Figure 2), and by the overlying, amagmatic, and Bashkirian carbonate sedimentation in the Tajik North Pamir [53]. The major preserved intrusive phase happened between 340 and 320 Ma, coevally in the Chinese and the Tajik North Pamir. Despite differences in geochemical composition, all granitoids from that time interval show similar petrographic properties and a relatively primitive composition. The granitoids of the Tajik North Pamir exhibit a generally more enriched REE pattern and a more pronounced negative Eu anomaly compared with those from the Oytag and Gez tonalites (Figure 7 and Appendix 8 [28, 29, 31]). The more enriched nature of the Tajik granitoids might reflect a higher grade of continental contamination within the mantle wedge due to the presence of the Kurguvad continental basement block. The same can be stated for the East Mazar granites, which probably intruded in the vicinity of the Tianshuihai complex as a result of a Carboniferous reactivation of the Proto-Tethys subduction zone [24]. However, interpretation of the East Mazar granites is hindered by their incorporation into the Karakax fault zone.

The vergence of the subduction zone cannot be constrained by our data. The southern margin of the North Pamir arc is largely overprinted by the late Triassic-early Jurassic Cimmerian orogeny. The northern margin is defined by the Cenozoic Main Pamir thrust, which has a larger offset in the western portion of the North Pamir arc. To the east of Taergelake, the position of the Main Pamir thrust is not well-constrained. Our field data from Qimgan valley (Figure 2 [41]) shows that the northern margin of the North Pamir Arc is covered by Permo-Triassic to Eocene sedimentary rocks.

5.2. The West Kunlun and the Kudi Suture

The Kudi section, situated along the Xinjiang-Tibet highway south of Kargilik, is the type locality for the West Kunlun. The North and South Kunlun terranes are separated by the mid-Paleozoic Kudi suture [10, 11, 18, 95, 96]. The South Kunlun block is an accretionary complex formed between the colliding North Kunlun (the southern margin of Tarim) and Tianshuihai blocks [13, 24]. The South Kunlun terrane is characterized by an early Paleozoic accretionary sequence, pre-Cambrian sedimentary sequences, Cambrian ophiolites, and Silurian to Ordovician arc volcanic rocks [13, 19, 97]; none of these are found in the northeastern Pamir. Closure of the Kudi suture is dated as lower Silurian by a 440 Ma monazite age of the Saitula group of the South Kunlun [13], reflecting initiation of large scale obduction of metamorphic units due to collision of South and North Kunlun.

The West Kunlun experienced two major intrusive phases: between 530 Ma and 400 Ma and between 240 Ma and 200 Ma (Figure 12). The older intrusive phase in the West Kunlun domain extends as far north as the shoshonitic Datong pluton (~450 Ma [98, 99]). The Yirba granodiorite (Figure 12(a)), cropping out in the Kudi section, intruded between 460 and 471 Ma during the mature phase of the Andean-type magmatic arc in the West Kunlun [19]. The Datong pluton (Figure 12(a)), emplaced between 448 and 473 Ma [100], likewise belongs to the mature phase. It marks the present, northwesternmost extent of the early Paleozoic plutons preserved in the West Kunlun. The Andean-type arc developed on top of a north (e.g., [19]) or south (e.g., [13, 24]) dipping subduction zone with Proto-Tethyan oceanic crust being subducted between the North Kunlun, regarded as Tarim crust, and the Tianshuihai block. An accretionary complex—the South Kunlun terrane—formed along the subduction zone. By that time, the South Kunlun accretionary complex experienced amphibolite facies metamorphism due to crustal thickening, subduction of the Proto-Tethys, and later collision of the continental block related to the Tianshuihai complex [13, 19]. Proto-Tethys subduction terminated in a late Silurian to early Devonian postorogenic stage with emplacement of highly evolved A-type granites [14, 101]. Since at least the emplacement of the A-type North Kudi pluton at around 405 Ma [14], the postorogenic extensional phase marks the termination of compressional, subduction-related tectonics in the West Kunlun. Plutons associated with the younger, Permo-Triassic phase can be found in the West Kunlun as well as in the Tianshuihai and in the Karakul-Mazar accretionary complex, which forms the southern part of the North Pamir.

5.3. Comparison between the West Kunlun and the North Pamir

The North Pamir arc is clearly distinguishable from the two major volcanic arcs recorded in the West Kunlun, where volcanism occurred during the closure of the Proto-Tethys and collision of the Tianshuihai, South Kunlun, and North Kunlun/Tarim blocks in the Silurian and Ordovician times [14, 19, 20, 98, 102], as well as during the closure of the Paleo-Tethys and accretion of the Cimmerian blocks [13, 21, 23, 103]. Volcanic rocks overlapping in age with the two major tectonothermal events of the West Kunlun have not been discovered in the North Pamir arc. Similarly, there are no arc volcanic rocks of Carboniferous age described from the West Kunlun. Instead, West Kunlun Carboniferous deposits represent amagmatic, shallow marine siliciclastic and platform carbonate successions [104]. Although pillow basalts, lava flows, and related clastic rocks from the Yixiekekhgou area, north of the town of Kudi, were previously interpreted as Carboniferous (Wang, 1996), a dacite from that section yielded a zircon U-Pb age of 492±9Ma [19].

The Oytag (or Wuyitake) suture—namely, a small outcrop of ultramafic rocks—is only found in the Gez valley [105]; no definitively correlative outcrops are known from the North Pamir arc to the west. A connection between Kudi and Oytag ultramafic rocks has long been proposed [3, 11, 106108]. This led to the interpretation that the Kudi-Oytag suture zone, separating the North and South Kunlun terranes, stretches from the West Kunlun all around the North Pamir. This terminology has been applied despite mismatching geochemical and especially geochronological data from the granitoids in the Gez/Oytag section [29, 30]. The arc volcanic rocks of the Gez/Oytag section were related to an intraoceanic subduction zone [29, 106] and have an island arc tholeiitic character. Volcanic rocks from the Kudi ophiolite show a wide range of geochemical characteristics; they are interpreted as N- and E-MORB ridge basalts [19], back-arc basin tholeiites, low-Ti island arc tholeiites, and island arc tholeiites [18, 20]. As geochronological information on the volcanic sequences is sparse, discrimination was difficult. However, amphibole and biotite 40Ar/39Ar ages from the Kudi ophiolite suggest a formation age prior to 460 Ma [18]. Hornblende 40Ar/39Ar ages from andesites of the Kyrgyz Altyn Darya valley yield ages of roughly 350 Ma [3], so 100 Ma younger than Kudi.

The difference between the North Pamir and the West Kunlun is even more obvious when comparing granitic intrusion history and metamorphism. Granites of the North Pamir arc show relatively uniform tonalitic to granodioritic composition and are not older than 360 Ma, with the main intrusive phase between 340 and 320 Ma. No postcollisional granitoids have been identified. In contrast, the granitoids of the West Kunlun are much older and show a higher compositional variability, including postcollisional A-type granitoids. Granitoids from all three volcanic phases (early Paleozoic, late Paleozoic, and Triassic) are related to subduction processes. Thus, they show similar trace element features. However, the Carboniferous granitoids are the least enriched ones (Figure 12), suggesting a higher mantle influence during that phase. Metamorphism in the North Pamir arc reached a maximum of lower amphibolite facies post-320 Ma, while the West Kunlun experienced amphibolite metamorphism between 440 and 430 Ma [13, 109].

If there is a continuation of the Oytag subduction zone toward the east, it should be situated south of the South Kunlun but not within the collision zone between the North Kunlun, the South Kunlun, and the Tianshuihai complex. The East Mazar sliver might give a hint of a continuation of the subduction zone between the Tianshuihai complex and the South Kunlun in the westernmost part of the West Kunlun. Based on sedimentary facies and the lack of evidence for Paleozoic orogenesis in the NE Pamir, it is also argued that a remnant of the Proto-Tethys remained open in the NE Pamir until the Cimmerian orogeny [24]. Therefore, we speculate that collisional tectonics in the West Kunlun terminated westward, toward the North Pamir, during the early Paleozoic orogeny.

Geochemical data compiled from previous studies of the early Paleozoic and early Mesozoic intrusive rocks show certain similarities with the geochemical signature of intrusive rocks discussed in this study (Figure 12). All are related to subduction processes. The magmatic suites show negative Nb, P, and Ti anomalies. The tonalites from the Oytag/Gez section show the least enrichment in large ion lithophile elements (LILE). Their enrichment is higher in the Tajik granitoids. Compared with the Triassic and early Paleozoic intrusive rocks, the Carboniferous magmatic suite is less enriched in LILE. The enrichment in LILE and depletion of high field strength elements (HFSE) such as Nb, P, and Ti is typical for subduction-related environments. Therefore, the Carboniferous granitoids show a stronger mantle component than the Triassic and early Paleozoic granitoids.

5.4. Relationship of the North Pamir Arc with Carboniferous Rocks in the Afghan Badakhshan

For the Carboniferous units of the Tajik North Pamir and their continuation into Afghanistan, a simple division into two zones was proposed by Bazhenov and Burtman [33]. Zone 1 represents oceanic crust, characterized by early Paleozoic volcanics and intercalated open marine sediments. Carboniferous ultramafic units were also described from this zone [34, 50]. Zone 2 represents an active margin, characterized by arc volcanic rocks and accreted microcontinents (Kurguvad block, Fayzabad microcontinent). Middle Paleozoic rocks sharing characteristics with those zones can be found from the North Pamir into the Herat area and further west [33]. We do not follow this bipartite division, as arc tonalites and granodiorites appear to intrude both units. Instead, we interpret the arc to have formed on top of both pre- to early Carboniferous oceanic crust and continental slivers. It is likely that the Kurguvad microcontinental block separated from other Tarim-related crustal blocks of the region (e.g., Garm) in the early Paleozoic, leaving behind a patchwork of oceanic and continental crustal fragments—which later were accreted during subduction processes in the Carboniferous. Subduction zones may have formed along the flanks of microcontinental blocks during the middle-late Carboniferous compressive phase.

Another feature that can be followed into the Badakhshan area is the sedimentary hiatus between the Mississippian and Pennsylvanian (Figure 2). This hiatus has been recognized in many profiles, separating a magmatic phase in the Mississippian from an amagmatic and predominantly marine sedimentary phase in the Pennsylvanian. The duration of this hiatus seems to increase from west to east, when comparing the similar setting within the Chinese North Pamir. Sedimentary Pennsylvanian units are only found in the Chinese North Pamir as allochthonous units [25]. Hiati within sedimentary sections from Afghanistan, Tajikistan, and Kyrgyzstan [3, 32, 33, 53, 55] are temporally variable. They show a pattern of diachronous marine sedimentary environments changing from platform carbonate sedimentation to clastic shelf sedimentation and a more or less erosional, terrestrial phase in the upper Carboniferous to Permian [44].

5.5. The Kurguvad Block: Detrital Signal and Metamorphism

The two Kurguvad paragneiss samples investigated using detrital zircon, RT13-148, and RT15-11 (Figure 6) have age peaks at 580 Ma, 722 Ma, and 943 Ma. These are similar to newly published ages from metamorphic units of the Kurguvad-Badakhshan complex (Figure 6 and Li et al. [40]). Two metasedimentary garnet-staurolite schists sampled by Li et al. (DV-7-27-15-1 and LY-7-18-17-2 [40]) gave a youngest age peak at around 310–360 Ma. This age peak likely reflects magmatic activity of the North Pamir arc. All other samples of that dataset together with our data from samples RT13-148 and RT15-11 lack that peak. We show that a granodiorite intruded into the metamorphic Kurguvad basement in the Bashkirian. The new findings can be integrated to a sequence of processes affecting rock units of the NW Pamir. (1) The lower metamorphic grade of the Early Carboniferous Kurguvad granodiorite, compared to the surrounding metasediments (e.g., garnet-biotite schist and garnet-staurolite schist), strongly supports the existence of at least one preintrusion metamorphic phase in the Kurguvad basement. (2) The presence of Carboniferous zircon in metasedimentary rocks, presented by Li et al. [40] just a few kilometers southeast of the Kurguvad intrusion, indicates late- to post-Carboniferous metamorphism. Metasediments yielding Carboniferous age peaks might reflect nappes of a Carboniferous accretionary wedge.

Peaks similar to the ones found in the Kurguvad block are known from gneisses of the Garm block in the Tajik Tien Shan [110112] and the Kyzylkum segment of the South Tien Shan (Mirkamalov et al. [113] cited in [110] and Konopelko et al. [42]). These ages indicate an Ediacaran maximum depositional age (Figure 6) and suggest a linkage of the Kurguvad basement block to the Tien Shan basement at that time. A very pronounced 590 Ma age peak is also reported from NE Gondwana [114]. The 943 Ma age peak may be related to igneous activity in Tarim, as suggested for a similar age peak between 1150 Ma and 800 Ma in the Garm gneisses [110]. The older age peaks at 2 Ga and 2.6 Ga are less pronounced in the Kurguvad gneiss. Similar peaks are reported from samples taken in the Quruqtagh, Central Tien Shan, that represents the Precambrian Tarim basement [115]. Shu et al. [115] link the late Archean and early Paleo-Proterozoic age peaks to two poorly defined Proterozoic tectonovolcanic events between 1.8 and 2.0 Ga and 2.4 and 2.6 Ga that could be related to the early formation of the Tarim block.

Metamorphism of the Kurguvad basement block has not been thoroughly documented. Peak amphibolite facies metamorphism was suggested by petrographic analysis and garnet-biotite (GARB) and garnet-biotite-muscovite-plagioclase (GBMP) thermobarometry of seven gneiss samples from the Kurguvad block [52]. These samples yielded temperatures and pressures of 540–650°C and 5.5–7.6 kbar without staurolite and 600–650°C and 6.5–8.2 kbar with staurolite, respectively. A monazite age of around 200 Ma reflects Cimmerian metamorphism. We obtained one zircon U-Pb age of 200.8±1.8Ma from sample RT15-11, which reflects the Cimmerian metamorphic imprint on the zircon fraction. The Kurguvad basement block must have also experienced pre-Carboniferous metamorphism, based on relative age relations. The complex garnets (Appendix 5) suggest a multiphase metamorphic event. However, Konopelko et al. [110] report concordant zircon U-Pb ages between 303 Ma and 406 Ma (Figure 6(c)) from the Garm basement block; these were interpreted to be metamorphic. This hints at a different metamorphic history in Garm compared to the Kurguvad basement block.

5.6. Metamorphism of the North Pamir Arc

The metamorphic grade of the North Pamir arc has been estimated by petrological thin section analysis, supported by XRD mineral phase analysis. Excepting the Kurguvad block, no indication of metamorphic imprint higher than greenschist facies on the Carboniferous volcanic rocks and granitoids was found. The Kurguvad block experienced amphibolite facies metamorphism during the Cimmerian orogeny [52]. Our data cannot clarify whether the greenschist facies metamorphism of the North Pamir is related to the late Carboniferous to Permian arc obduction history or to a later, Cimmerian overprint.

New and published geochemical and geochronological data provide a detailed view of the along-strike variations of the North Pamir Carboniferous granitoid intrusions and their host rocks. Geochemical data from the tholeiitic and mafic to intermediate rocks, cropping out in the Chinese Oytag and Qimgan valley and in the Kyrgyz Altyn Darya valley, document the formation of an island arc complex in the Upper Devonian to Bashkirian [3, 28]. The North Pamir arc is chronologically and geochemically distinct from the volcanic arcs known from the West Kunlun.

Compiled literature data from the early Paleozoic and Triassic magmatic arc successions of the West Kunlun are compared with new and published data from the Carboniferous magmatic arc rocks of the North Pamir. The early Paleozoic succession in the West Kunlun records major intrusive activity of a mature arc between 470 Ma and 450 Ma. The major Carboniferous intrusive activity in the North Pamir lasted from 340 Ma to 320 Ma. The Triassic arc-magmatic activity previously described from the West Kunlun and Karakul-Mazar lasted from 240 Ma to 200 Ma. As all three magmatic environments are subduction related, their granitoid rocks show similarities in geochemistry: enrichment in LILE and depletion of HFSE such as Nb, P, and Ti (Figure 12). However, the North Pamir arc Carboniferous granites show less enrichment of LILE and a stronger mantle influence than the West Kunlun early Paleozoic and the Triassic granites. Collision and exhumation of metamorphic rocks of the West Kunlun started in the Silurian. A-type magmatism occurring in the lower Devonian marks a postorogenic extensional phase [14]. No magmatic rocks of that age have been recognized in the North Pamir.

Despite the loose stratigraphic control on the age of the volcanic sequence in the North Pamir, we show that all analyzed samples and complementary literature data share common geochemical characteristics. As the mafic and intermediate volcanic complexes show an arc signature, the name Kalai Khumb-Oytag basin (KOB [36]) might be used for the subbasin of the Paleoasian ocean which hosted the arc. The arc itself should better be named the North Pamir arc, following Bazhenov and Burtman [33]. As we have shown, the term Kunlun arc [3] is inappropriate. The Kurguvad basement block with Ediacaran maximum depositional age and complex metamorphic history was part of the volcanic arc, at least for part of its history. This sliver of a continental basement block is quite similar to the Garm block in the Tian Shan, suggesting a joint Precambrian geologic evolution.

Starting with the emplacement of the Taergelake granite at around 360 Ma, large granitoids were emplaced within the arc until 314 Ma. Granitoids in the Chinese North Pamir have a more primitive composition and are classified as island arc granites within an intraoceanic subduction zone [29]. The coeval tonalites and granodiorites that intruded volcanic sequences and the Kurguvad basement block in the Tajik North Pamir show more enriched REE patterns and are classified as continental volcanic arc granites. Our samples from the Tajik North Pamir arc fit well in stages 3–5 (tonalites to leucoplagiogranites) established by Mamadjanov et al. [54] for Late Paleozoic intrusives of that region. To date, there are no age constraints for the stage 1 and 2 intrusive rocks (i.e., gabbro and quartz diorites). We interpret the geochemical along-strike variance as the transition from intraoceanic island arc subduction in the Chinese Pamir to Cordilleran-type subduction in the Tajik North Pamir.

Granitoids in the East Mazar sliver of the Chinese West Kunlun show similar geochemical patterns as the Tajik granitoids, suggesting another Cordilleran-type subduction zone to the east (Figure 13). However, there is also a connection to continental material in the Chinese Pamir, documented by an inherited zircon population with an age of 448 Ma from the Oytag granite [29] and the 417 Ma age population from the Triassic aplitic dike in Qimgan. Both age populations might be derived from the West Kunlun, where lower Devonian to Ordovician granitoids related to the closure of the Proto-Tethys are common [101, 116]. We infer a Carboniferous sediment flux from the Proto-Tethys suture zone in the West Kunlun toward today’s Chinese North Pamir and recycling of that detritus in the accretionary wedge.

The preserved length of the segment between Fakhar (Afghanistan) and Oytag (China) is around 500 km. However, it is likely that the geometry of the subduction system was modified by post-Carboniferous deformation. A partial obduction and erosion of the North Pamir arc in Pennsylvanian to Permian time can be inferred from disconformities and facies variations found in the Tajik and Chinese North Pamir. Facies distributions indicate a longer phase of crustal uplift and erosion in the Chinese North Pamir, where the Upper Devonian to Bashkirian arc volcanic units are followed by a continental Guadalupian to Triassic sequence. Whether the widespread presence of greenschist metamorphism in the North Pamir arc is related to this arc obduction event is not clear.

The existence of pre-Mesozoic oceanic crust to the north—in present day coordinates—of the advancing Cimmerian continents in the region of the todays North Pamir is in sharp contrast to the situation in the West Kunlun. New and existing data indicate the existence of a major oceanic domain in the North Pamir in late Paleozoic time and argue against a continuous Tarim-Tajik cratonic continent. This also challenges the hypothesis that the West Kunlun and North Pamir formed a continuous linear belt prior to the India-Asia collision; this removes a key paleogeographic constraint on the magnitude of Cenozoic indentation of the Pamir. A Carboniferous compressive phase partly closed the oceanic basin along an intraoceanic subduction zone that laterally continued into a Cordilleran-style subduction zone. In the east, closure and exhumation resulted in continental conditions, as shown by the Permo-Triassic volcanosedimentary sequence found in the Qimgan valley. The lack of high-grade metamorphic units all along the North Pamir arc and ongoing marine sedimentation in the western parts of the basin, however, hint at a soft collision or slowdown of Carboniferous compressive tectonics. That left behind weaker crust to the west of Tarim compared to the well-amalgamated early Paleozoic terranes of the West Kunlun. The West Kunlun incorporated the Tarim basement, in the North Kunlun domain [12]. This E-W difference in rheology facilitated the advance of the Central Pamir further to the north compared to its eastward lateral equivalents, which may have caused a bending of the Cimmerian orogen.

New geochronological and geochemical data used for this study can be found in Appendixes 7 and 8. Sources of compiled data are cited in the text.

The authors declare that they have no conflict of interest.

We thank Langtao Liu for help with fieldwork in China. Likewise, we thank Ilhomjon Oimahmadov, Mustafo Gadoev, Sherzod Abdulov, and Umed Sharifov for help with fieldwork in Tajikistan. For support with sample preparation and laboratory work, we thank Christine Fischer, Antje Musiol, Baiansuluu Terbishalieva, and Christina Günther. We also thank Martin Timmermann, Roland Oberhänsli, Patrick O’Brian, Robert Trumbull, and Romain Bousquet for helpful discussions on geochemical data. This project was funded by Deutsche Forschungs Gesellschaft e.V. (DFG) grant SO 436/12-1 to Edward R. Sobel and DFG grant KL 495/27-1 to Jonas Kley. Sampling in Tajikistan and Kyrgyzstan by Edward R. Sobel was funded by the National Geographic Society grant GEFNE105-14. Sampling of the granites and the Kurguvad basement in the Tajik Punj and Obikhingou valleys was funded by DFG grant TH 1317-5 to Rasmus Thiede.

Exclusive Licensee GeoScienceWorld. Distributed under a Creative Commons Attribution License (CC BY 4.0).

Supplementary data