Existing models of intracontinental deformation have focused on plate-like rigid body motion v. viscous-flow-like distributed deformation. To elucidate how plate convergence is accommodated by intracontinental strike-slip faulting and block rotation within a fold–thrust belt, we examine the Cenozoic structural framework of the central Qilian Shan of northeastern Tibet, where the NW-striking, right-slip Elashan and Riyueshan faults terminate at the WNW-striking, left-slip Haiyuan and Kunlun faults. Field- and satellite-based observations of discrete right-slip fault segments, releasing bends, horsetail termination splays and off-fault normal faulting suggest that the right-slip faults accommodate block rotation and distributed west–east crustal stretching between the Haiyuan and Kunlun faults. Luminescence dating of offset terrace risers along the Riyueshan fault yields a Quaternary slip rate of c. 1.1 mm a−1, which is similar to previous estimates. By integrating our results with regional deformation constraints, we propose that the pattern of Cenozoic deformation in northeastern Tibet is compatible with west–east crustal stretching/lateral displacement, non-rigid off-fault deformation and broad clockwise rotation and bookshelf faulting, which together accommodate NE–SW India–Asia convergence. In this model, the faults represent strain localization that approximates continuum deformation during regional clockwise lithospheric flow against the rigid Eurasian continent.

Supplementary material: Luminescence dating procedures and protocols is available at https://doi.org/10.17605/OSF.IO/CR9MN

Thematic collection: This article is part of the Fold-and-thrust belts and associated basins collection available at: https://www.lyellcollection.org/cc/fold-and-thrust-belts

Plate tectonics is complicated by intracontinental deformation due to the continental lithosphere's inherent weakness compared with the oceanic lithosphere (Chen and Molnar 1983) and vertical heterogeneity that can lead to lithospheric decoupling and the propagation of detachment horizons (Burchfiel et al. 1989; C.S. Wang et al. 2011; Mouthereau et al. 2013). As a result, continental plate boundaries are often not expressed as single fault systems, but rather wide (hundreds to thousands of kilometres) zones of diffuse deformation that accommodate relative plate motion, such as the Himalaya–Tibet orogen and the San Andreas–Basin and Range system of western North America (Atwater 1970; Davis and Burchfiel 1973; Molnar and Tapponnier 1975; Flesch et al. 2000; Yin 2010; Thatcher et al. 2016). Whether continental tectonics can be quantified by plate-like rigid body motion (Luyendyk et al. 1980; Tapponnier et al. 1982; Weldon and Humphreys 1986; Avouac et al. 1993; Meade and Hager 2005; Meade 2007) or viscous-flow-like distributed deformation (England and Houseman 1986; Yin and Taylor 2011; Haproff et al. 2018) remains a fundamental question.

With regard to the Himalaya–Tibet orogen and the western North American Cordillera, this question focuses on whether major faults such as the left-slip Altyn Tagh fault and the right-slip San Andreas fault act as discrete boundaries of internally rigid crustal blocks or simply localized zones of high strain within a larger deforming continuum (Tapponnier et al. 1982; England and Houseman 1986; Avouac et al. 1993; Platt and Becker 2010; Johnson 2013; Platt and Becker 2013; Evans et al. 2016). A better understanding of the kinematic history of the strike-slip fault systems across the Himalaya–Tibet orogen allows us to explore their role in accommodating India–Asia plate convergence and answer the question of discrete v. distributed intracontinental deformation.

The Qilian Shan fold–thrust belt marks the northeastern margin of the Tibetan Plateau (Fig. 1a) and is bounded by the internally rigid Tarim and Qaidam basins to the WSW and the North China craton to the NE (Braitenberg et al. 2003; Guo et al. 2005; Kusky and Mooney 2015; Cheng et al. 2017; Xu et al. 2020). The fold–thrust belt has accommodated Cenozoic India–Asia convergence through a combination of NE-directed shortening and NW- and WNW-striking strike-slip faulting (Dewey and Burke 1973; Tapponnier et al. 1990; Gaudemer et al. 1995; Murphy et al. 1997; Métivier et al. 1998; Meyer et al. 1998; Yin and Harrison 2000; Yin et al. 2008a; Taylor and Yin 2009; Yin 2010; Zuza et al. 2016; Allen et al. 2017; Cheng et al. 2019b). Major WNW-striking left-slip faults in the Qilian Shan (i.e. the Haiyuan fault) and the Kunlun fault to the south are considered to behave as the respective northern and southern boundaries of clockwise-rotating crustal blocks (Fig. 1b) that resemble a bookshelf fault system (England and Molnar 1990; Zuza et al. 2016). Crustal blocks are bounded to the east and west by NNW-striking right-slip faults such as the Riyueshan and Elashan faults (Fig. 1b).

Over the last few decades, geological and geophysical studies have improved our understanding of the geometry and kinematics of Cenozoic deformation in the Qilian Shan fold–thrust belt (Zhang et al. 2004; Ghosh et al. 2006; Bovet et al. 2009; Zheng et al. 2010; Yin and Taylor 2011; Zuza et al. 2016; Pan et al. 2020). Despite disagreement regarding the Cenozoic growth history of the Qilian Shan (Bovet et al. 2009; Allen et al. 2013; He et al. 2018; Zuza et al. 2018a; Cheng et al. 2019b; Li et al. 2020b), most studies agree that strike-slip faulting has been active since the Mid-Miocene (Lin et al. 2011; W.-T. Wang et al. 2011; X. Wang et al. 2011; Duvall et al. 2013; Li et al. 2019). However, the mechanism that generated the sets of NW-striking right-slip faults and WNW-striking left-slip faults and their kinematic evolution since the Mid-Miocene remain inadequately understood (Wang and Burchfiel 1997; Duvall and Clark 2010; Yuan et al. 2013).

Continental strike-slip faults are commonly accompanied by a variety of structures, including horsetail splay faults at their tips, local contractional faults at restraining bends, local extensional faults at releasing bends and bookshelf faults (Fig. 2) (Cunningham and Mann 2007; Duvall et al. 2013; Zuza and Yin 2016). In addition, the geometric arrangement of Riedel shears and associated structures along strike-slip faults can provide information on the kinematics within the wrench zone (Fig. 2) (Bartlett et al. 1981). Here, we integrate the results of satellite imagery analysis and geological field mapping along the right-slip Elashan and Riyueshan faults and existing temporal constraints of deformation associated with these faults to evaluate the mechanism(s) that drove the strike-slip faulting and block rotation in the Qilian Shan. By combining measurements of offset terrace risers along the right-slip Riyushan fault with the luminescence ages of terraces, we determine a late Quaternary slip rate of c. 1.1 mm a−1. Using these results, we discuss the late Cenozoic evolution of the Qilian Shan, which informs how intra-thrust belt block rotation and strike-slip faulting accommodate plate convergence.

Pre-Cenozoic tectonic evolution of the Qilian Shan

The pre-Cenozoic tectonic evolution of the Qilian Shan orogen consists of Neoproterozoic continental break-up, Early Paleozoic subduction and subsequent continental collision, followed by Mesozoic extension (Yin and Harrison 2000; Yang et al. 2001; Yin et al. 2007; Song et al. 2009; Xiao et al. 2009; Song et al. 2013; Huang et al. 2015; T. Wang et al. 2016; Cheng et al. 2017; C. Wang et al. 2017; Wu et al. 2017; Zuza et al. 2018b; Cheng et al. 2019c). Precambrian basement rocks are sporadically exposed in the Qilian Shan and record the tectonic–magmatic events related to the early Neoproterozoic subduction between the South Tarim–Qaidam and North Tarim–North China continents and subsequent late Neoproterozoic continental rifting (Wan et al. 2001; K. Tung et al. 2007; Song et al. 2012; K.-A. Tung et al. 2013; Zuza et al. 2018b; Cheng et al. 2019a; Cheng et al. 2019b).

The Qilian Shan orogen also contains early Paleozoic flysch sequences, plutons, ophiolitic mélange and metamorphic rocks. These rocks record the early Paleozoic closure of the Qilian Ocean as the Kunlun–Qaidam terrane collided with the North China craton (Song et al. 2009; Xiao et al. 2009; Wu et al. 2017). Mesozoic strata overlie Paleozoic strata and Precambrian basement (Zuza et al. 2018b). Triassic plutons that intrude the Paleozoic strata and Precambrian basement probably record the Mesozoic closure of the Palaeotethys and Mesotethys oceans (Pullen et al. 2008; L. Yu et al. 2017). Slab rollback in the south and shearing along the Altyn Tagh fault zone (Cheng et al. 2019c) might have led to regional extension in the Qilian Shan during the Jurassic–Cretaceous.

As summarized by Zuza et al. (2018b) and Wu et al. (2016), the pre-Cenozoic history of this orogen is still ambiguous, with current disputes focused on the polarity of oceanic subduction, the number and location of suture zones, and the onset timing and duration of orogenesis. Several disparate belts of suture zone material are exposed in the Qilian Shan, which form zones of pre-existing weakness that were reactivated in the Cenozoic (Zuza et al. 2018b; Li et al. 2020a).

Cenozoic tectonic evolution of the Qilian Shan

Cenozoic structures of the Qilian Shan consist of series of NW-striking fold–thrust belts and the WNW-striking, left-slip Haiyuan and Kunlun faults (c. 100–110° strike) (Fig. 1b). Several NNW-striking right-slip faults, including the Elashan and Riyueshan faults, are located at high angles between the Haiyuan fault to the north and the Kunlun fault to the south (Fig. 1b) (Meyer et al. 1998; Yin et al. 2008b; Lease et al. 2011; Lease 2014; Cheng et al. 2015b; Zuza et al. 2016; Allen et al. 2017).

As indicated by existing low-temperature thermochronological data and sedimentological records, the southern part of Qilian Shan experienced significant crustal shortening during the Paleocene to Eocene shortly after India–Asia collision, followed by Oligocene exhumation of the central and northern parts of the Qilian Shan (Jolivet et al. 2001; Bovet et al. 2009; Zhuang et al. 2011; He et al. 2017; He et al. 2018; Cheng et al. 2019b). Given the general northwards younging initiation ages of Cenozoic structures within the Qilian Shan, some studies advocate for a progressive northwards propagation of deformation from the Miocene to Pliocene (George et al. 2001; Bovet et al. 2009; D. Zheng et al. 2010, 2017; W. Wang et al. 2016; Cheng et al. 2019b). However, several pulses of out-of-sequence thrusting have been reported throughout the Qilian Shan (Zuza et al. 2018a; Li et al. 2020b). Aside from the NE–SW-directed shortening across the Qilian Shan fold–thrust belt, the >1000 km long, WNW-striking, left-slip Haiyuan fault (Fig. 1) probably played an important part in accommodating crustal deformation in northeastern Tibet from the Mid-Miocene (Burchfiel et al. 1991; Duvall et al. 2013; Yuan et al. 2013; Zuza et al. 2016; Li et al. 2019).

Rotating bookshelf faulting

The Elashan and Riyueshan faults are sub-parallel, NW-striking (c. 325–340° strike), right-slip faults situated between the left-slip Haiyuan fault to the north and the left-slip Kunlun fault to the south, forming a rotating bookshelf fault system (Fig. 1b). Slip initiated along the Elashan and Riyueshan faults at 9 ± 3 and 10 ± 3 Ma, respectively (Duvall et al. 2013; Yuan et al. 2013). The reported late Quaternary slip rates along the middle segments of the Elashan and Riyueshan faults are 1.1 ± 0.3 and 1.2 ± 0.4 mm a−1, respectively (Yuan et al. 2011).

Geological field mapping, satellite image analysis and seismic profile interpretation

To investigate the geometry and kinematics of the structures within the Qilian Shan, we conducted geological field mapping along the Riyueshan and Elashan faults within the central part of the Qilian Shan fold–thrust belt (sites 1, 2, 3 and 4 in Fig. 1b). Detailed field observations across much of the Qilian Shan have been reported previously (Wu et al. 2017; Zuza et al. 2018a; Zuza et al. 2018b; Cheng et al. 2019a). Google Earth imagery and field observations were used to document the surface trace and kinematics of these two faults. In addition, we reinterpreted a previously published seismic reflection profile across Lake Qinghai adjacent to the Riyueshan fault (Fig. 1b) (An et al. 2006) to further understand the kinematics of the bookshelf faults.

Luminescence geochronology

Two samples (RF001 and RF002) were collected for optically simulated luminescence dating from one site (site 4) along the northern segment of the Riyueshan fault near its termination at the Haiyuan fault (Fig. 1b). Sample RF001 was collected from a north-facing wall along the south side of a stream drainage at a depth of 85 cm below the channel-incised wall or riser (site 4a). Sample RF002 was collected south of sample RF001 from the north-facing wall along the south side of a stream drainage at a depth of 65 cm below the channel-incised wall (site 4c). The sampled sediment is mainly composed of massive pebbly silty clay with rootlets extending 90 cm below the ground surface and the uppermost 20–25 cm consists of modern day soil. Angular pebbles appear to be locally derived and therefore the sediments are probably colluvial with an aeolian component.

Samples RF001 and RF002 were collected by hammering steel tubes into the cleaned surfaces of the terrace walls. The tubes were subsequently removed from the terrace wall and wrapped in light-proof plastic for transport to the Desert Research Institute E.L. Cord Luminescence Laboratory (Reno, NV, USA) for further sample preparation. Detailed sample preparation and analytical procedures are described in the Supplementary Material.

Initial luminescence measurements and dose recovery tests showed that potassium-rich feldspar (Brookfield) grains were better suited to dating than the quartz grains in these samples (see Supplementary Material for details). Infrared-stimulated luminescence (IRSL) ages were therefore measured from small (1 mm diameter) aliquots of fine sand-sized grains. The age distributions from c. 45 measurements from both samples were tight with an overdispersion of <10%. This indicates that the measured fine sand-sized feldspar grains were well bleached prior to burial. We therefore interpret an aeolian origin for much of the sediment and/or shallow slope aggradation, which supports an aeolian and/or shallow colluvial origin for the deposit.

Elashan fault

The northern tip of the Elashan fault is expressed as a horsetail splay structure consisting of a series of NW- to NNW-striking faults that merge to the south (site 1, Fig. 3a). Regional geological maps show that these faults offset Triassic shallow marine and terrestrial strata along a right-slip sense (Fig. 3b, c) (Qinghai-Geology-Bureau 1976). By contrast, the northwesternmost tip of the Elashan fault features minor faults with both right-slip and left-slip separation (Fig. 3d, e). Previous geological mapping (Qinghai-Geology-Bureau 1976) and satellite-based observations suggest that these structures are high-angle faults as a result of their relatively linear traces across the topography and the displacement of gently SW-dipping strata (Fig. 3d, e). We suggest that the right-slip and left-slip map view separation along these faults reflects dominantly dip-slip normal kinematics.

To the south of site 1, just north of the city of Wulan (site 2, Fig. 1b), the middle segment of the Elashan fault has two right-stepping releasing bends that form a local pull-apart basin featuring several north- and NNW-striking faults (Fig. 4a, b). The north-striking fault traces are sub-parallel to the trend of the local river drainage and several triangular facets are observed along the faults (Fig. 4a, b). In addition, the north-striking linear faults do not laterally displace the local stream drainage (Fig. 4a) or topography (Fig. 4b). We therefore interpret the north-striking faults to have dominantly dip-slip kinematics. By contrast, the NW-striking faults feature sub-horizontal striations in vertical fault planes (Fig. 4d), suggesting dominantly strike-slip kinematics (Fig. 4e).

Riyueshan fault

The northern segment of the Riyueshan fault shows a complex variability between strike-slip and normal-slip kinematics. The Riyueshan fault along the western flank of the Lakeshan Range (site 3 on Fig. 1b), which strikes c. 338°, features triangular facets and displays no evidence of laterally displaced stream drainage (Fig. 5), which suggests dominantly dip-slip kinematics. However, the northernmost segment of the Riyueshan fault (site 4 on Fig. 1b), which strikes c. 328°, features right-laterally displaced stream drainage (Fig. 6a). We did not observe any dip-slip displacement at this latter site. This structural geometry is similar to that of the Elashan fault, wherein the more northerly striking fault segments have normal-slip normal kinematics segments, whereas the fault segments with a more NW-striking orientation have strike-slip kinematics.

At the northernmost site along the Riyueshan fault, we surveyed right-laterally displaced fluvial terrace risers at three locations (sites 4a–4c, Fig. 6a). To mitigate potential errors in measuring the displaced fluvial terrace risers (Cowgill 2007), we systematically measured the right-lateral displacements of the thalweg and bounding risers on both sides of the drainage to determine a range of plausible displacement magnitudes. Based on riser morphology, we report an average displacement magnitude and standard deviation (1σ), which typically excludes measurements of rounded, eroded and/or ambiguous riser crests.

At site 4a, we determined that the northern riser, thalweg and southern riser are displaced by 7.41, 8.04 and 5.60 m, respectively (Fig. 6b). At site 4b, we determined that the northern riser, thalweg and southern riser are displaced by 9.45, 7.66 and 5.63 m, respectively (Fig. 6c). At site 4c, the northern riser, thalweg and southern riser are displaced by 8.30, 7.90 and 8.85 m, respectively (Fig. 6d). The average (1σ) right-slip displacement magnitudes along the Riyueshan fault at sites 4a, 4b and 4c are 7.73 ± 0.45 m (Fig. 6b), 8.56 ± 1.27 m (Fig. 6c) and 8.10 ± 0.28 m (Fig. 6d), respectively. Erosion and degradation of the southern risers at each site added to the uncertainty of the displacement measurements. For this reason, our reported average displacement estimates exclude the displacement measurements of the southern risers.

A previous geophysical survey by An et al. (2006) across Lake Qinghai (Fig. 1b) shows a series of WNW-striking, high-angle faults (Fig. 7). By comparing the isobaths of subsurface reflectors on either side of these faults, we observed that the faults have both apparent right-slip and left-slip separation (Fig. 7) and divide the Lake Qinghai Basin into several internally rigid blocks (Fig. 7a).

The orientation of these WNW-striking, high-angle faults compared with the larger WNW-striking Haiyuan fault to the north and the Kunlun fault to the south are roughly consistent with the expected orientation of subsidiary normal faults parallel to the minimum compressive stress within a strike-slip system (Fig. 7b). In a north–south-trending seismic reflection profile, we observe sets of growth strata that indicate a >5 Ma initiation timing of normal faulting (Fig. 7c) (Fu et al. 2013).

Samples RF001 and RF002 yield respective IRSL ages of 7.15 ± 0.42 and 7.16 ± 0.42 ka (Table 1). The nearly identical IRSL ages of samples RF001 and RF002, despite being collected c. 550 m apart from two displaced stream drainages (Fig. 6), suggest spatially uniform erosion and aggradation processes over time and relatively minor aggradation during most of the Holocene. Both samples were taken within a metre of the ground surface and below the soil horizon and no paleosol was observed at depth in the terrace walls. It is therefore possible that aggradation rates were higher in the late Pleistocene–early Holocene before slowing at c. 7 ka. Assuming that the sampled sediments aggraded prior to channel incision and fault slip, the average IRSL ages reflect the maximum ages for the most recent displacement along the Riyueshan fault.

Given the 7.73 ± 0.45 m displacement and 7.15 ± 0.42 ka IRSL age at site 4a, we determine a late Quaternary slip rate of 1.1 ± 0.1 mm a−1 (1σ) along the Riyueshan fault. Given the 8.35 ± 0.47 m displacement and 7.16 ± 0.42 ka IRSL age at site 4c, we determine a late Quaternary slip rate of 1.2 ± 0.1 mm a−1 along the Riyueshan fault. We thus estimate the mean late Quaternary slip rate along the northern part of the Riyueshan fault to be 1.1 ± 0.1 mm a−1 (1σ).

Strike-slip faulting and block rotation related to the Riyueshan and Elashan faults

Our field- and satellite-based observations allow us to better characterize the geometry and kinematics of the right-slip Elashan and Riyueshan faults. The northwestern termination of the Elashan fault constitutes a horsetail splay fault system (Fig. 3) that probably distributes lateral shear. Directly north of this structure is the western segment of the WNW-striking Haiyuan fault, which is not displaced right-laterally by the Elashan fault (Fig. 1b). This requires right-slip faulting to terminate at the mapped tip of the Elashan fault and not continue northwards across the Haiyuan fault (Fig. 1b). The middle segment of the Elashan fault is expressed as a pull-apart basin (Fig. 4). The northern and middle segments of the Riyueshan fault are right-lateral normal oblique-slip faults. The development of these subsidiary structures suggests that both the Riyueshan and Elashan faults probably accommodated regional extension since their inception. Based on their geometries and transtensional setting, these faults accommodate west–east stretching of Qilian Shan.

Previous studies have inferred that right-lateral slip along the Elashan and Riyueshan faults initiated at 9 ± 3 and 10 ± 3 Ma, respectively (Yuan et al. 2011; Duvall et al. 2013). The west- to WNW-striking Qinghai–Nanshan and Gonghe–Nanshan thrusts (Fig. 1b), which form kinematically linked thrust ramps between the Elashan and Riyueshan faults, initiated at 11–9 Ma (Zhang et al. 2012) and 10–7 Ma, respectively (Craddock et al. 2011). This coeval Late Miocene initiation of the timing of deformation (Fig. 8a) implies a regional rather than local driving force, possibly induced by a NE-trending maximum principal compressive stress related to NE–SW India–Asia convergence. As a consequence, right-lateral slip along the Elashan and Riyueshan faults resulted in coeval Miocene extension by the Lake Qinghai fault system in the north and shortening by the Qinghai–Nanshan and Gonghe–Nanshan thrusts (Fig. 1b). A Miocene onset of faulting in the Lake Qinghai fault system is roughly supported by the >5 Ma age revealed by the growth strata (Fig. 7) (Fu et al. 2013).

Given that the Miocene onset of left-slip motion along the Kunlun and Haiyuan faults (Jolivet et al. 2003; Lin et al. 2011; W.-T. Wang et al. 2011; X. Wang et al. 2011; Duvall et al. 2013; Cheng et al. 2014; Li et al. 2019) is roughly consistent with the initiation ages of the right-slip structures between these left-slip faults, we propose that the Elashan and Riyueshan right-slip faults initiated in the Miocene as part of a bookshelf fault system embedded within the left-lateral shear zone between the Haiyuan and Kunlun faults (e.g. Duvall and Clark 2010). For a bookshelf fault system that is controlled by a broad left-slip shear zone, the internal fault-bounded blocks are predicted to rotate counterclockwise (Fig. 8b). This kinematic prediction is complicated because northern Tibet is part of a broad NE-trending right-lateral shear zone that experiences clockwise rotation against the fixed Tarim block and left-slip Altyn Tagh fault (Cobbold and Davy 1988; England and Molnar 1990; Zuza and Yin 2016), including the Haiyuan and Kunlun left-slip faults, their fault-bounded wallrock and the right-slip fault system.

Previous palaeomagnetic studies adjacent to this right-slip fault domain between the Haiyuan and Kunlun faults do not yield significant rotation since the Miocene (Dupont-Nivet et al. 2003; Fang et al. 2003) (Fig. 8a). However, if most of northern Tibet experienced clockwise rotation, local counterclockwise rotation within the bookshelf fault system might be obscured as the integrated rotation magnitudes cancel each other out within uncertainties (Fig. 8c). We therefore propose a hybrid model with regional right-lateral shear and clockwise rotation in northern Tibet that drives left-slip motion along the Haiyuan and Kunlun faults, which ultimately causes smaller scale counterclockwise rotation and right-slip bookshelf faulting between the left-slip faults (Fig. 8d).

Crustal stretching and lateral displacement along strike-slip faults in northern Tibet

It remains debatable whether there has been significant crustal stretching and lateral displacement along strike-slip faults in northern Tibet since the Miocene (Gaudemer et al. 1995; Ding et al. 2004; Harkins et al. 2010; Cheng et al. 2015b). This type of potential deformation is likely to be limited to the upper crust (Hubbard and Shaw 2009; Tian et al. 2016; Shen et al. 2019). As shown by recent earthquake focal mechanisms in the Qilian Shan region (Pan et al. 2020), most earthquakes are associated with NW-striking reverse faults and strike-slip faults (Fig. 9a). Earthquake focal mechanisms indicate right-slip motion along the Elashan and Riyueshan faults and reverse fault motion along the Qinghai–Nanshan and Gonghe–Nanshan thrusts (Fig. 9a). The recent earthquake activity and geometrical relationships between these faults suggest a kinematic linkage between the reverse faults and the Riyueshan and Elashan faults.

Our field observations and IRSL ages allow us to constrain a late Quaternary average slip rate of 1.1 ± 0.1 mm a−1 along the northernmost segment of the Riyueshan fault. This estimate is consistent with the previously estimated late Quaternary slip rates of 1.2 ± 0.4 mm a−1 along the middle segment of the Riyueshan fault (Yuan et al. 2011) and 1.1 ± 0.3 mm a−1 along the middle segment of the Elashan fault. We therefore infer that both the Riyueshan and Elashan faults have shared a c. 1 mm a−1 uniform slip rate since the Late Quaternary. Assuming this constant long-term slip rate, eastwards crustal displacement in the Qilian Shan would correspond to >10 km of displacement along each of the right-slip faults.

As shown in the compilation of the Quaternary slip rates in the Qilian Shan region in Figure 9b, the estimated slip rate of the Haiyuan fault ranges from 1–3 to 19 ± 5 mm a−1. However, the study sites of Lasserre et al. (1999, 2002), from which faster slip rates of 19 ± 5 and 12 ± 4 mm a−1 were derived, have come into question due to a bias towardsthe systematic use of lower terrace reconstructions in interpreting the age of offset (Zheng et al. 2013b; Yao et al. 2019). With the exception of these two possibly overestimated slip rates, the Haiyuan fault slip rates vary from 1–3 to 6.4 ± 0.7 mm a−1 along the western fault segment and 3.2 ± 0.2 to 8–10 mm a−1 along the eastern fault segment (Li et al. 2009; Yuan et al. 2011; Zheng et al. 2013b; Matrau et al. 2019; Yao et al. 2019). Assuming a constant long-term slip rate (c. 4 and c. 6 mm a−1 along the western and eastern segments, respectively), a left-slip initiation at 15 Ma along the Haiyuan fault (Duvall et al. 2013; Li et al. 2019; Yu et al. 2019) yields c. 60–90 km of total displacement. This slip estimate overlaps published offset measurements (Gaudemer et al. 1995; Ding et al. 2004). For the West Qinling fault, given the estimated slip rate of 2.5–3 mm a−1 (Fig. 9b) (Chen and Lin 2019), c. 38–45 km of displacement would have occurred since 15 Ma.

Published slip rate estimates for the Kunlun fault are c. 16 mm a−1 along its western segment and gradually decrease to 8.9 ± 0.7 to 10.9 ± 0.5 mm a−1 along its central segment and to 2.0 ± 0.4 mm a−1 along its eastern segment (Fig. 9b) (Van Der Woerd et al. 1998, 2002; Li et al. 2005; Kirby et al. 2007; Harkins and Kirby 2008; Lin and Guo 2008; Harkins et al. 2010). Assuming a constant long-term slip rate (c. 16, 8.9 ± 0.7 to 10.9 ± 0.5 and 2.0 ± 0.4 mm a−1 along the western, central and eastern segments, respectively) and left-slip initiations of 12–8, 20–15 and 8–5 Ma along the western, central and eastern segments of the Kunlun fault, respectively, displacement magnitudes of 130–200, 120–230 and 8–20 km would have occurred for these fault segments.

Although the use of a constant slip rate may overestimate or underestimate the total displacement along these faults, these estimates provide a perspective on tens to hundreds of kilometres of eastwards displacement of crust in northern Tibet since the Miocene. Whether these amounts of eastwards displacement of crust in the Qilian Shan along strike-slip faults can be balanced with the magnitude of crustal lateral displacement transferred from >300 km of left-slip displacement along the Altyn Tagh fault (Meyer et al. 1998; Searle et al. 2011; Cheng et al. 2015b, 2016b) remains uncertain.

The geometries of the Riyueshan and Elashan faults imply that c. 1 mm a−1 of slip along each fault corresponds to c. 0.5 mm a−1 of west–east-oriented stretching per fault, or c. 1 mm a−1 of west–east stretching across the Lake Qinghai fault system. This interpretation based on Quaternary slip rates is comparable with the west–east-oriented stretching rates of 1–2 mm a−1 in the same region derived from global positioning system (GPS) velocities (Duvall and Clark 2010). The NE–SW-oriented convergence rate across the central Qilian Shan is 6–7 mm a−1 based on a tabulation of shortening estimates over the past 10 Ma (Zuza et al. (2018a) and c. 5–7 mm a−1 based on modern convergence rates from GPS velocities (Zhang et al. 2004; Gan et al. 2007). We therefore infer that, since the Mid- to Late Miocene, the Qilian Shan has accommodated c. 6 mm a−1 of NE–SW-oriented crustal shortening and slower west–east-oriented stretching of 1–2 mm a−1.

Mechanism of Late Cenozoic deformation in the Qilian Shan

Most studies have reached a consensus that strike-slip faulting in the Qilian Shan fold–thrust belt has been active since the Mid-Miocene (Lin et al. 2011; W.-T. Wang et al. 2011; X. Wang et al. 2011; Duvall et al. 2013; Li et al. 2019) and accommodated a considerable amount of Cenozoic crustal deformation in northern Tibet (Dewey and Burke 1973; Murphy et al. 1997; Yin and Harrison 2000; Dupont-Nivet et al. 2002; Yin et al. 2002, 2008a; E. Wang et al. 2006; Taylor and Yin 2009; Yin 2010; Cheng et al. 2016a; Zuza et al. 2016; L. Wang et al. 2020).

Two broad end-member models have been proposed to explain this Cenozoic deformation associated with this strike-slip faulting: the eastwards lateral displacement model (Meyer et al. 1998; Cheng et al. 2015b) and the non-rigid passive bookshelf faulting model (Zuza and Yin 2016). The eastwards lateral displacement model emphasizes that the strike-slip faults (e.g. the Kunlun, Haiyuan, Riyueshan, Elashan and West Qinling faults) are the dominant structures in accommodating convergence in northern Tibet. By contrast, the non-rigid passive bookshelf faulting model involves discrete left-slip faulting in northern Tibet, which accommodates distributed off-fault deformation during the regional clockwise rotation of crustal blocks and faults. Despite the merits of the non-rigid passive bookshelf faulting model, it is challenged by palaeomagnetic studies that indicate limited rotation in northern Tibet (Dupont-Nivet et al. 2002, 2003; Fang et al. 2003; Yu et al. 2014) v. others that demonstrate appreciable Neogene clockwise rotation (Cogné et al. 1999; Chen et al. 2002; Halim et al. 2003; Liu et al. 2003; Yin et al. 2008a). The eastwards displacement model is questioned by whether the NW-striking strike-slip faults (e.g. the Riyueshan and Elashan faults) within the Qilian Shan can absorb such a large amount of eastwards crustal motion across northern Tibet (Meyer et al. 1998; Cheng et al. 2015b).

As a result of this study, we suggest that the non-rigid passive bookshelf faulting model can be reconciled with a new integrative model that better explains the kinematics framework of northern Tibet by taking the significant fault-parallel west–east stretching into consideration. We thus propose a hybrid model that combines these two end-member models to describe the Cenozoic deformation pattern of northern Tibet.

In the early Cenozoic, as a result of India–Asia convergence, shortening strain in northern Tibet caused the translation of the Qaidam Basin northwards against the Altyn Tagh fault (Ritts and Biffi 2000; Cowgill et al. 2003; Cheng et al. 2015a, 2015b, 2016b) (Fig. 10a). Greater shortening in the west than in the east (Yin et al. 2008a; Zuza et al. 2016) led to broad right-lateral shear strain and net clockwise rotation against the rigid Tarim Basin and may have resulted in greater crustal thickening in the west than in the east. Progressive displacement along the Altyn Tagh fault was ultimately transferred into the fold–thrust Qilian Shan (Meyer et al. 1998; Searle et al. 2011; Cheng et al. 2015b, 2016b) (Fig. 10b).

As a result of progressive crustal thickening in northern Tibet, it became more mechanically efficient for strike-slip faulting to initiate at c. 15 Ma (Duvall et al. 2013; Li et al. 2019) along the pre-Cenozoic suture zones (Wu et al. 2017, 2019; Zuza et al. 2018b) and continued clockwise rotation drove left-slip bookshelf faulting of the Haiyuan and Kunlun faults (England and Molnar 1990; Zuza and Yin 2016) (Fig. 10b). The initiation of a broad left-slip transpressional system established a four-quadrant strain pattern, with fault-parallel shortening in the NW and SE quadrants and fault-parallel extension in the NE and SW quadrants. This included the activation of the right-slip Elashan and Riyeshan faults at the SW tip of the Haiyuan fault, which simultaneously accommodated fault-parallel west–east stretching and counterclockwise rotation embedded between the left-slip Kunlun and Haiyuan faults (e.g. Duvall and Clark 2010; this study). The overprinting clockwise and counterclockwise rotation led to variable net rotation that might appear negligible (Fig. 8).

The thicker crust and greater shortening in the west established a local stress state where eastward crustal stretching was favoured, which may be analogous to the conditions in southern Tibet (Yin et al. 1999; Bian et al. 2020). This strain condition was superimposed over the bookshelf fault kinematics outlined here and the crust between the Kunlun and Haiyuan faults extended eastwards. This explains the prevalence of normal fault structures and bulk west–east stretching (Figs 4 and 5). In this sense, progressive crustal thickening and north–south shortening in the Qilian Shan is balanced by net eastwards extrusion, as suggested by Cheng et al. (2015b). Therefore, net north-directed convergence across the Qilian Shan (Zhang et al. 2004; Wang and Shen 2020) is accommodated via a combination of north–south crustal shortening, tens to hundreds of kilometres of displacement along strike-slip faults, eastwards crustal stretching and clockwise rotation (Fig. 10).

As revealed by GPS and seismic anisotropy studies (Gan et al. 2007; Li et al. 2011; Pan et al. 2020), there is a regional clockwise flow pattern across northern Tibet. Our work suggests that major strike-slip faults accommodate the continuum deformation of this region. None of these faults is a rigid block boundary, but rather they are structures of localized high strain approximating viscous flow. We acknowledge that pre-existing weaknesses throughout the Qilian Shan (e.g. suture zones) and crustal heterogeneities are reactivated as discrete faulting within this flowing body. As a result of conservation of mass requirements, this NE–SW-oriented contractional system requires NW-SE-oriented stretching, which causes the strike-slip faults to have subsidiary normal splay faults.

A better understanding of the geometry and kinematics of the strike-slip faults and associated structures within the Qilian Shan allows us to explore how intra-fold–thrust belt block rotation and strike-slip faulting accommodate plate convergence. We examined the Elashan and Riyueshan right-slip faults in the central Qilian Shan and our observations and slip rate constraints lead to the following conclusions.

  1. We document horsetail termination splays, discrete right-slip faults, off-fault normal faulting and releasing bends, which suggest that the right-slip faults actively accommodated block rotation and distributed west–east-oriented crustal stretching between the left-slip Haiyuan and Kunlun faults.

  2. New IRSL dating of displaced terrace risers along the Riyueshan fault provides a c. 1.1 mm a−1 slip rate estimate, similar to other published rates along this fault.

  3. The pattern of Cenozoic deformation in northern Tibet is compatible with west–east-oriented crustal stretching, non-rigid off-fault deformation and broad clockwise rotation and bookshelf faulting, which together accommodate the NE–SW-oriented India–Asia convergence.

  4. The faults in northern Tibet represent strain localization that together accommodates continuum deformation during regional clockwise lithospheric flow against the rigid Eurasian continent. Pre-existing weaknesses and strength heterogeneities cause the partitioning of deformation along dip-slip and strike-slip faults.

FC and AVZ are grateful to the Chief Editor Prof. Graham Shields and Maggie Simmons and Patricia Pantos for the invitation to publish this paper. We are grateful to Editor Dr Yuntao Tian, reviewer Dr Xiubin Lin and one anonymous referee for providing insightful comments. We thank Dr Zhengyang Pan for providing the raw data for focal mechanism solutions of recent earthquakes in North Tibet.

FC: conceptualization (equal), formal analysis (equal), investigation (equal), methodology (lead), visualization (lead), writing – original draft (lead); AVZ: conceptualization (equal), formal analysis (equal), funding acquisition (lead), investigation (equal), methodology (equal), supervision (equal), writing – review and editing (lead); PJH: formal analysis (equal), validation (equal), writing – review and editing (equal); CW: investigation (equal), validation (equal); CN: investigation (equal), methodology (equal), validation (equal), writing – review and editing (equal); HC: validation (supporting), writing – review and editing (supporting); XL: validation (supporting), writing – review and editing (supporting); BL: formal analysis (supporting), writing – review and editing (supporting).

This work was funded by an open grant from the Key Laboratory of Deep-Earth Dynamics of Ministry of Natural Resource, Institute of Geology, Chinese Academy of Geological Sciences (J1901), an open grant from the State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences (SKLLQG1903), and a grant from the National Natural Science Foundation of China (41930213 and 41888101) to FC. The National Science Foundation (1914501), National Science Foundation (1914501) and The Tectonics Program of the National Science Foundation supported this research (EAR 1914501) with grants to AVZ and PJH. BL acknowledges grants from the Natural Science Foundation of China (U1902208) and from the Youth Innovation Promotion Association Chinese Academy of Sciences (Y201759). CN acknowledges funding from the National Science Foundation (No. 1914566) and National Natural Science Foundation of China (41888101).

All data generated or analysed during this study are included in this published article (and its supplementary information files).

Scientific editing by Yuntao Tian