Geometry and Kinematic Evolution of the Hotan-Tiklik Segment of the Western Kunlun Thrust Belt: Constrained by Structural Analyses and Apatite Fission Track Thermochronology Open Access

The collision and continuous convergence between Indian and Eurasian plates has led to the construction of the Tibetan Plateau, the largest elevated area on Earth (Yin and Harrison 2000). The India-Eurasia collision has created a wide zone of intracontinental deformation that extends thousands of kilometers north of the Himalayan convergence front (e.g., Molnar and Tapponnier 1975; Yin 2010). Understanding the processes by which collision-induced deformation has propagated northward from the convergence front is key to establishing the mechanism of plateau growth (e.g., Tapponnier et al. 2001; Yin et al. 2008). The western Kunlun thrust belt between the relatively stable Tarim Basin in the north and the intensely deformed western Tibetan Plateau in the south is an ideal place to address this problem.

Previous studies in the western Kunlun Mountains and southwestern Tarim Basin have provided a wealth of data and increased understanding of geophysical structures (Matte et al. 1996; Gao et al. 2000; Kao et al. 2001; Li et al. 2001, 2012; Jiang et al. 2004, 2013; Wittlinger et al. 2004), deformation geometry (Cowgill 2001; Liu et al. 2004; Wu et al. 2004; Wang et al. 2011, 2014; Cheng et al. 2012; Liang et al. 2012, 2014; Jiang et al. 2013; Tang et al. 2014; Suppe et al. 2015; Li et al. 2016), mountain building and basin subsidence processes (Lyon-Caen and Molnar 1984; Sobel and Dumitru 1997; Wang et al. 2001, 2003; Wan and Wang 2002; Li et al. 2007; Li 2008; Wei et al. 2013; Jiang and Li 2014), basin sedimentation and stratigraphic architecture (Zheng et al. 2000, 2003, 2006, 2010, 2015a, 2015b; Yin et al. 2002; Jin et al. 2003; Sun and Liu 2006), and the present geodetic velocity fields (Wang et al. 2001; Niu et al. 2005; Gan et al. 2007). Despite progress, some major issues regarding geometry, kinematics, and evolution of deformation related to the development of the Cenozoic western Kunlun thrust belt remain controversial. Taking the Hotan-Tiklik segment, for example, Jiang et al. (2013), using the seismic profiles obtained from SinoPec, suggested that ca. 24.6 km of shortening in the segment was mainly absorbed by the basement-involved Tiklik fault and fault-propagation fold (Hotan anticline) related to the Hotan fault. In contrast, other schools of geologists, with the seismic profiles obtained from PetroChina, interpreted the segment to consist of the basement-involved Tiklik fault and fault-bent fold (Hotan anticline) related to the Hotan fault and suggested that slip of the Hotan fault could reach the Mazatagh fault, ca. 230 km north of the Hotan anticline, along the lowermost Paleogene detachment layer, and absorb a significant portion of shortening in the segment (Liang et al. 2012; Wang et al. 2014; Suppe et al. 2015). Moreover, deformation timing of the Tiklik fault remains poorly understood.

In this study, we combined structural analysis of surface geology and subsurface seismic reflection profiles with growth strata and apatite fission track (AFT) thermochronometry along the Hotan-Tiklik traverse across the western Kunlun Mountains. The results of this work provide new insights into the style, timing, and kinematics of deformation across this segment of the western Kunlun thrust belt.

The western Kunlun defines the south boundary of the Tarim Basin to separate it from the western Tibetan Plateau to the south (fig. 1a). Tectonically, it consists of the western Kunlun (including North and South Kunlun terranes separated by the Tam Karaul fault/Kudi suture zone) and the Tianshuihai and Qiangtang terranes, which assembled onto the Tarim block gradually during the Paleozoic (fig. 1b; Cowgill 2001; Xiao et al. 2002; Cowgill et al. 2003). The western Kunlun terrane is separated from the Tarim block to the north by the north-vergent Tiklik fault and the Tianshuihai terrane to the south by the sinistral transpressional Karakax fault (fig. 1b). The east end of the western Kunlun is offset by the Altyn Tagh fault (ATF), while the west continuation is offset by the Kashgar-Yecheng transfer system (Cowgill 2010; fig. 1b). To the south, the Tianshuihai terrane is separated by the south-vergent Qiertianshan-Hongshanhu fault from the Qiangtang terrane (fig. 1b). To the east, both the Tianshuihai and Qiangtang terranes are bounded and offset by the Longmu Co–Gozha Co fault and the southward extent of the sinistral strike-slip ATF (fig. 1b).

In the study, we focus on the Hotan-Tiklik segment of the western Kunlun thrust belt. In map view, the Hotan thrust zone expresses as a simple Hotan anticline (fig. 1c, 1d), which bounds the eastern limb of the Pianamen anticline to the west and is bounded by the NNW-SSE-striking Aqike anticline to the east. The Hotan fault is a blind thrust that is interpreted to structurally control the deformation of the Hotan anticline (fig. 1c; Liang et al. 2012; Jiang et al. 2013). The Hotan anticline is cored by the Pliocene strata that dip to the south in the south limb, with the dip angle increasing to the south, and that dip to the north in the north limb (fig. 1c, 1d). The Tiklik thrust fault defines the south boundary of the Hotan anticline. It dips ca. 40°–75° SSW based on field mapping (fig. 1d; Cowgill 2001; Liang et al. 2012), consistent with the interpretation of Jiang et al. (2013), showing a high-angle thrust fault. The hanging wall of the thrust is composed of the Proterozoic basement and Paleozoic strata but also carries the Mesozoic-Cenozoic sequences in the Buya region to the east (fig. 1c).

The stratigraphic units in the Hotan-Tiklik segment include the Proterozoic, Paleozoic, and Cenozoic strata (e.g., Jiang et al. 2013; fig. 2). The Mesozoic strata (mainly Jurassic and Cretaceous) may distribute in Buya to the southeast and Puska to the west of the study region (fig. 1c) but are not discovered in both outcrops and borehole HeCan 1 (HC 1; e.g., Sobel 1999; Jiang et al. 2013) and therefore are excluded from the stratigraphic discussions in this study (fig. 2). The Proterozoic strata consist of garnet-bearing quartz schist (Cowgill 2001; Wang et al. 2009; fig. 2). The Cambrian and Ordovician strata have been discovered in the Tarim Basin (Shaanxi Geological Bureau 2006), i.e., borehole ShengHe 2, where they consist of limestone, gypsum, and dolomite serving as a detachment layer and overlying detrital rocks (e.g., Yu 2001; He et al. 2002; Wang et al. 2011; fig. 2), although they are determined by neither outcrops nor boreholes along the western Kunlun. The continuous and parallel reflections above the Proterozoic reflectors in the seismic profiles were interpreted as a Cambrian sequence (Wang et al. 2011; Liang et al. 2012), consistent with the interpretation of early Paleozoic strata (Pz₁) by Jiang et al. (2013). The overlying late Paleozoic strata have been confirmed by both outcrops and borehole data. Within these strata, the Devonian units are dominated by sandstone with conglomerate at the bottom (fig. 2; Cowgill 2001; Wang et al. 2001; Shaanxi Geological Bureau 2006). Carboniferous to Permian strata consist of limestone, volcanic rocks, and continental facies detrital rocks (fig. 2; Cowgill 2001; Wang et al. 2001). The Cenozoic sequences are composed of marine sediments (Kashi Group) and overlying terrestrial deposits (Wuqia Group and Artux and Xiyu Formations). The lowermost ca. 300 m of the Kashi Group consists of gypsum beds (fig. 2; Yin et al. 2002; Wei et al. 2013; Bosboom et al. 2014), serving as a regional upper detachment layer (Wang et al. 2011; Liang et al. 2014). The chronology of the Cenozoic sediments has been initially constrained by biostratigraphy, which indicated a Paleogene age for the Kashi Group, a Miocene age for the Wuqia Group, a Pliocene age for the Atushi Formation, and a Quaternary age for the Xiyu Formation (e.g., Hao and Zeng 1980; Hao et al. 1982). More recently, magnetostratigraphic results are consistent with this chronologic division (e.g., Zheng et al. 2000, 2003, 2006, 2010; Sun and Liu 2006; Sun et al. 2008; Chen et al. 2015), except for the results from Zheng et al. (2015a). Under debate was whether there were volcanic tuff beds in the section described by Zheng et al. (2015a), from which the radioisotopic ages were obtained (Sun et al. 2015; Zheng et al. 2015b). These results indicated an age of <1.6–2.0 to 2.6–3.6 Ma for the Xiyu Formation, from 2.6–3.6 to 5.2–5.8 Ma for the Atushi Formation, and from 5.2–5.8 to >22.1 Ma for the Wuqia Group (Zheng et al. 2000, 2003, 2006, 2010; Sun and Liu 2006; Sun et al. 2008; Chen et al. 2015). This division has been employed in interpreting the seismic profiles in this region (e.g., Wang et al. 2011; Jiang et al. 2013; Wei et al. 2013; Liang et al. 2014). In this study, we divided the seismostratigraphy into seven units that correspond to the stratigraphic division in the HC 1 borehole, including the Proterozoic (Pt), the Lower Paleozoic–Devonian (Pz₁-D), the Carboniferous-Permian (C-P), the Paleogene Kashi Group (E), the Miocene Wuqia Group (N₁), the Pliocene Artux Formation (N₂), and the Quaternary Xiyu Formation (Q₁; fig. 2).

The analytical methods applied in the study include the interpretation and balanced cross section of seismic profiles to determine the timing and shortening magnitude of the Hotan thrust zone and AFT thermochronology to estimate those of the Tiklik thrust zone.

The S-N-trending dip line profile of the two-way-travel (TWT) time data and borehole data was obtained from Tarim Oil Company, PetroChina. The conversion from TWT time profile into depth section relies on the velocities of each stratigraphic unit obtained from the HC 1 borehole (fig. 3), which is verified by outcrop data on the surface (fig. 1c, 1d). The stratigraphic division and logging data of the borehole HC 1 and outcrop data were used to correlate the seismic reflections to the stratigraphic units in the seismic profile. The structural and growth strata style interpretation of the seismic reflections was based on fault-related fold models (e.g., Suppe et al. 1992; Shaw et al. 2005).

The deformation restoration was constructed using balanced cross section with the inverse kinematic model using 2D Move software. The timing of the fault slip and related folding is determined from the overlying growth strata. The inverse models were conducted with a fault-bend fold geometry. The construction of balanced cross sections relies on three assumptions: (1) the folding and rock uplifting are both controlled by the fault slip, (2) stratigraphic layers were originally horizontal before the deformation, and (3) layer lengths and bed areas are both constant during the deformation.

AFT samples were collected in the Tiklik Mountains in the hanging wall of the Tiklik thrust fault, with sample locations and coordinates shown in figure 1c and table 1. All the samples are from garnet-bearing quartz schists of Proterozoic age (Wang et al. 2009). Apatite grains were obtained by standard density and magnetic separation of the crushed samples, with final hand picking from the mineral concentrate under a binocular microscope. AFT analyses of the samples were carried out by W. M. Yuan at the Institute of High Energy Physics, Chinese Academy of Sciences (now working at the China University of Geosciences, Beijing).

AFT analysis followed the approach outline by Yuan et al. (2006). Apatite grains were mounted in epoxy resin on glass slides and ground and polished to an optical finish to expose internal grain surfaces. Spontaneous tracks were revealed by etching in 5.5% HNO₃ for 20 s at 21°C. The samples were then irradiated with thermal neutrons in the 492 light-water reactor in Beijing, with muscovite used as the external detector. After irradiation, the muscovite detectors were detached and etched in 40% hydrofluoric acid for 20 min at 25°C. Densities of both natural and induced fission tracks were measured with a dry objective at ×1500 magnification. Neutron flux was determined using dosimeter glass CN5 with a known uranium content of 11 ppm (Hurford and Green 1982) included at the ends of the irradiation package. AFT ages were calculated using the International Union of Geological Sciences–recommended zeta calibration approach (Hurford and Green 1982), with a weighted mean zeta value of 386.8 ± 18.1 (1σ). Horizontal confined fission tracks were measured to obtain track lengths (Green et al. 1986). The χ² test was used to detect the probability that all age grains analyzed belong to a single population of ages (Galbraith 1981). A probability of <5% is indicative of an asymmetric spread of single-grain ages indicating variability in population age (i.e., multiple discrete age populations within a given sample).

The TWT seismic reflection profile images three packages of reflections (fig. 3a). Outcrop and borehole data have been used to correlate the reflections with the seven seismostratigraphic units noted above.

The upper-right (upper-north) profile is characterized by a package of well-defined continuous, subparallel, and horizontal reflections (fig. 3a), which has been correlated with the Cenozoic strata (E, N₁, N₂, and Q₁; fig. 3b; Liang et al. 2012, 2014; Jiang et al. 2013). The horizontal feature of the reflections suggests that the strata have not been significantly deformed. A horizontal and continuous reflector that represents the lowermost of the Paleocene strata (E) and is parallel to both the overlying and underlying reflections defines the lower boundary of the package (fig. 3a, 3b). To the left (south), the package is bounded by a north-direct-dipping (down to the right side of the profile) well-defined reflector (arrows in fig. 3a) that separates the package from the one to the south (the upper-left package of reflections), with the boundary reflector reaching a depth corresponding to ca. 4.2 s (fig. 3a).

The upper-left (upper-south) reflection package is featured by well-defined continuous, parallel, and bent reflections (depths corresponding to ca. 1.2–3.2 s; fig. 3a) overlying chaotic to subparallel reflections (depths corresponding to ca. 3.2–4.7 s; fig. 3a). The bent feature of the reflections demonstrates that the strata have been deformed. According to stratigraphic data obtained in outcrops (fig. 1d) and borehole HC 1 (fig. 3a), the continuous and parallel reflectors correspond to N₂ to C-P strata (fig. 3a, 3b; Wang et al. 2001; Jiang et al. 2013), while the lower chaotic to subparallel reflectors correlate with Pz₁-D strata (fig. 3a, 3b; Cowgill 2001; Wang et al. 2001; Jiang et al. 2013). The package is bounded by a north-direct-dipping reflector to the north, as noted above (arrows in fig. 3a), and by a subhorizontal, but bent at the southernmost, reflector at the bottom (triangles at a depth corresponding to ca. 4.7 s in fig. 3a).

The lower reflection package is composed of well-defined continuous and parallel reflections overlying less continuous to chaotic ones (depths corresponding to ca. 4.7 s to the bottom of the profile; fig. 3a). These reflections correlate with C-P, Pz₁-D, and Pt strata, respectively (fig. 3a; Wang et al. 2009; Jiang et al. 2013). The generally horizontal feature of these reflections implies that they have not been deformed.

In the TWT profile, the upper-left reflection package exhibits bent reflections (fig. 3a) that have been interpreted as a symmetrical anticline with deformed strata from the Paleozoic to the Pliocene (Pz₁-D, C-P, E, N₁, and N₂; fig. 3b; Liang et al. 2012, 2014; Suppe et al. 2015). The bottom of the anticline, the boundary reflector between the upper-left package and the lower one, has been interpreted as a thrust fault (fig. 3b; Liang et al. 2012, 2014; Suppe et al. 2015), supported by the repetition of Pz₁-D and C-P strata in these two packages (fig. 3a). The right (northern) boundary of the anticline is defined by a north-direct-dipping reflector (fig. 3a), which truncates the reflections of both the upper-right and upper-left packages, showing an onlapping contact relation. This type of contact relation is consistent with the time-transgressive angular unconformity between foreland onlapping growth strata and the forelimb and crest of an anticline (fig. 3b–3d; Shaw et al. 2005).

Using the velocity data obtained from the borehole HC 1, the TWT profile has been converted into a depth section (fig. 3c), in which the relation between the anticline, the thrust fault, and the time-transgressive angular unconformity is clearly presented. The symmetrical Hotan anticline has been interpreted to be a fault-bend fold controlled by the underlying Hotan thrust fault detaching along the lowest Paleogene detachment layer (fig. 3c; Shaw et al. 2005). This interpretation is supported by a clear bend of the fault at the left (south) end of the section that divides the fault plane into upper fault-flat and fault-ramp (fig. 3c). To the north of the anticline, the onlapping contact between the horizontal strata and the strata involved within the anticline results in a time-transgressive angular unconformity. The termination of the boundary line and its foreland horizontal extent (dashed line in mid- to upper N₁ in fig. 3c) define the boundary between growth and pregrowth strata (fig. 3b–3d; Liang et al. 2012, 2014), suggesting that the growth strata formed since the mid- to upper Miocene. To create this time-transgressive angular unconformity, the growth strata accompanied with anticline formation should have witnessed varying ratios of sedimentation rate relative to anticline uplift rate (fig. 3d).

To obtain the slip magnitude of the Hotan thrust fault and the shortening magnitude of the Hotan anticline, balanced cross section has been conducted against a regional section (fig. 4). This regional section has been established with the interpretations of seismic profile (fig. 3) and field observations (fig. 1). The geometry of the south-direct-dipping Tiklik fault is constrained by field measurement (fig. 1d) and in accord with the interpretation of Jiang et al. (2013). The Hotan thrust fault detached along both the lowermost Paleogene and Paleozoic detachments, forming upper and lower fault-flats (fig. 4). This is consistent with the observation that no pre-Paleozoic rocks (Pt; Wang et al. 2009) have been observed in the hanging wall of the Hotan fault (fig. 3c) and the fact that the lowermost Paleozoic and lowermost Paleogene strata could serve as detachment layers (Wang et al. 2011; Liang et al. 2014). The Tiklik and Hotan faults have been inferred to share the same detachment layer along the lowest Pz₁ strata (fig. 4; Suppe et al. 2015). Within the structural balanced cross section, activity of the Tiklik fault has not been taken into consideration, and, therefore, the slip and shortening magnitudes relate only to the Hotan thrust fault and the Hotan fold, respectively (fig. 4).

Results of balanced cross section indicate that the slip magnitude of the Hotan fault along its lower fault-flat (equal to the motion of the hinterland panel; terminology following the definition in Lin et al. 2010) reaches ca. 27.3 km in total, resulting in the section length being shortened from ca. 84.2 km in the original undeformed section to ca. 56.9 km in the present deformed section (fig. 4). The slip magnitude of the fault along its upper fault-flat (equal to the offset between the footwall cutoff at the top of the fault-ramp and the corresponding hanging wall cutoff at the top of the C-P bed, also equal to the motion of the foreland panel; terminology following the definition in Lin et al. 2010) is ca. 23.2 km (fig. 4). The difference between the slip magnitude along the lower fault-flat and that along the upper fault-flat has been absorbed by shortening of the Hotan anticline, suggesting a shortening magnitude of ca. 4.1 km in the anticline (fig. 4).

Results of balanced cross section also reveal how the fault slip and anticline shortening were distributed in each episode of deformation. The Hotan fault and the related Hotan anticline formed when the growth strata began to deposit at the mid- to upper Wuqia Group (fig. 4). During the deposition of the mid- to upper Wuqia Group, the Hotan fault slipped ca. 1.9 km along its lower fault-flat and ca. 1.3 km along its upper fault-flat, with ca. 0.6 km absorbed by the Hotan anticline (fig. 4). During the deposition of the Artux Formation (N₂), the section underwent an intense deformation stage, with the Hotan fault slipping ca. 16.8 km along the lower fault-flat and ca. 13.8 km along the upper fault-flat, suggesting a ca. 3.0 km shortening absorbed by the Hotan anticline (fig. 4). During the deposition of the Xiyu Formation (Q₁), the fault slip of the Hotan fault decreased to ca. 8.6 km along its lower fault-flat and ca. 8.1 km along its upper fault-flat, indicating a shortening of ca. 0.5 km absorbed by folding in the Hotan anticline (fig. 4).

Six samples were collected for AFT analysis in the Tiklik thrust belt (table 1). The apparent AFT ages span from 14.3 +2.1/−1.9 to 22.5 +3.1/−2.7 Ma, with mean track lengths ranging from 12.09 ± 2.23 to 12.96 ± 2.05 μm (table 1).

The AFT analyses yielded Miocene ages (table 1). These ages are younger than the metamorphic age (U-Pb ages of ca. 736–810 Ma; Wang et al. 2009), indicating an imprint of a postmetamorphic cooling event. Most of the samples show disperse grain ages, including samples KLKS-1, KLKS-3, KLKS-6, and KLKS-7, with the P (χ²) values far below 5% (table 1). Likely attributions of the obtained disperse grain ages could include the following: (1) apatite minerals in schist samples coming from different sources would result in various kinetics of annealing behavior and consequently cause age dispersal and/or (2) abundant fission tracks shortened by annealing can introduce variability in age calculations because of the difficulty in accurately recognizing heavily shortened tracks (e.g., Gleadow et al. 1986; Green 1988; Lin et al. 2011). The former attribution is open for discussion because of annealing kinetic parameters not being measured in the analysis, while the latter seems reasonable, as abundant shortened fission tracks were observed (table 1). Grain ages of the remaining two samples, KLKS-2 and KLKS-5, pass the χ² test and suggest Miocene AFT ages (table 1), indicating that imprint of Miocene cooling was significant.

The thermal histories of the samples can be assessed, at least to some degree, by thermal history modeling with grain age and track length data. The modeling could help to establish the tectonothermal histories of the samples through a partial annealing zone (PAZ) of AFT, ca. 60°–120°C (e.g., Gleadow and Duddy 1981). By using the annealing model proposed by Ketcham et al. (1999), we investigated the potentially complex thermal histories of the AFT samples through HeFTy software (Ketcham et al. 2000; Ketcham 2005).

Ideally, some temperature-time (T-t) constraints imposed into the modeling would be useful to link the modeling results with the geological process. Without such reliable T-t constraints, only minimal limits were imposed on the thermal history modeling: (1) all the samples were cooled to the temperature of ca. >120°C at a time older than the apparent AFT ages (ca. 50% older than the apparent ages, Ketcham et al. 2000) and (2) all the samples are at Earth’s surface today, with the temperature ca. 10° ± 10°C (fig. 5). The samples were all obtained from a mountain area, leading to the hypothesis that the samples have experienced monotonic cooling without subordinate burial and heating process until their ultimate exhumation onto Earth’s surface.

The ending condition was set as when the modeling returned 100 good paths. Two of the six analyzed samples (KLKS-1 and KLKS-2) returned 100 good paths, with goodness-of-fit values of measured ages and track lengths against modeled ones both above 0.50 in the two samples (fig. 5). Modeling of the other samples did not return enough good paths, and therefore the modeling was ended artificially (e.g., see sample KLKS-3 in fig. 5). The modeling results indicate that the samples have undergone two episodes of cooling, one during the late Oligocene–early Miocene and the other possibly since the mid- to late Miocene (fig. 5). The former cooling event exhumed the samples from the upper limit (>120°C) to approximately the lower limit (ca. 60°C) of the PAZ, while the latter one further exhumed the samples from the temperature ca. 60°C to Earth’s surface (fig. 5). In particular, the exact beginning timing of later cooling stage, with most paths out of PAZ (fig. 5), should be treated as uncertain.

The seismic data indicate that the Hotan fold is a fault-bend fold controlled by the Hotan fault. Therefore, the growth strata developed over the Hotan fold provide constraints to evaluate the deformation timing of the Hotan fault. For the Tiklik fault, the AFT data and related thermal history modeling results from the samples in the hanging wall can be used to determine the deformation timing of the fault.

The structural analysis of the seismic profile suggests that the growth strata started to develop along the north limb of the Hotan fold since the mid- to late Miocene (fig. 3b, 3c; Liang et al. 2012, 2014). The mid- to late Miocene deformation timing is consistent with the uncertain later cooling episode obtained from the thermal history modeling results of the samples from the hanging wall of the Tiklik fault (fig. 5), suggesting that this stage of cooling should be very likely. In addition, the AFT results of the samples from the hanging wall of the Tiklik fault suggest that the fault underwent significant activity at the late Oligocene to early Miocene, resulting in the exhumation of the hanging wall (fig. 5).

Taken together, data in this study suggest two episodes of deformation in the Hotan-Tiklik segment of the western Kunlun. The earlier stage during the late Oligocene to early Miocene was expressed by activity of the Tiklik fault, indicated by the earlier cooling stage recorded by the AFT data (fig. 5). The later stage since the mid- to late Miocene involved activity of both the Tiklik and Hotan faults, documented by cooling paths (fig. 5) and related folding and development of growth strata.

The shortening magnitude of the Hotan-Tiklik segment is well constrained along the Hotan thrust zone with the interpretation of its seismic profile, while in the Tiklik thrust zone, without data to determine the subsurface structure, the shortening magnitude can be only roughly constrained by thermochronological data combined with fault geometry. As noted above, in light of the Hotan thrust zone, the interpretation of a seismic profile and its structural balanced cross section results suggest that the total fault slip of the Hotan fault reaches ca. 27 ± 1 km (given 1 km for measurement error) south of the Hotan anticline and decreases to ca. 23 ± 1 km (given 1 km for measurement error) north of the anticline, with the slip difference ca. 4 ± 2 km absorbed by the Hotan anticline (fig. 4). This suggests that the total shortening magnitude of the Hotan anticline is only ca. 4 ± 2 km, which is much less than that (total shortening of ca. 24.6 km) suggested by Jiang et al. (2013), due to different geometry and kinematic interpretation models between their studies and this one. In our model, the hanging wall block of the Hotan fault slips along the detachment layer in the lowermost Paleogene strata, and most of the slip magnitude (ca. 23 ± 1 km) might be absorbed by the Mazatagh fault and its related fold, consistent with the model proposed by Liang et al. (2012, 2014) and Suppe et al. (2015).

Regarding the Tiklik thrust zone, the AFT analyses and related thermal history modeling results suggest that the samples have been exhumed into the upper limit of PAZ of AFT (ca. 120° ± 10°C) during the late Oligocene to early Miocene (fig. 5). Supposing a constant geothermal gradient of ca. 20° ± 5°C/km and a surface temperature of ca. 10° ± 10°C, it means that the samples have been exhumed by ca. 4–9 km since that time. Assuming that rock uplift = surface uplift + exhumation (England and Molnar 1990), it indicates a rock uplift of ca. 5–10 km. Within the calculation, the exhumation is ca. 4–9 km, and the surface uplift is ca. 1 km. The surface uplift of ca. 1 km was calculated from the difference between the sample elevations (mean elevation of ca. 2 km above sea level) and the elevation of the Tarim Basin (mean elevation of ca. 1 km above sea level; elevation data from the digital elevation model [DEM], http://srtm.csi.cgiar.org/). This calculation is based on several assumptions, including (1) the elevation of Tarim has not changed, (2) there was no paleo-relief in the Tiklik range before the period of exhumation, and (3) isostatic adjustment was not taken into consideration. Given a dipping angle of 40°–75° for the Tiklik fault (Cowgill 2001; Liang et al. 2012), this leads to an estimation of horizontal shortening of ca. 7 ± 5 km (2–12 km). Considering that significant crustal thickening (e.g., Wittlinger et al. 2004) and more complex structures in the upper crust (Shaanxi Geological Bureau 2006) have been excluded from the calculation, the exact shortening of the Tiklik thrust zone should be more than ca. 7 ± 5 km (2–12 km).

The above discussions suggest that the total horizontal shortening magnitude of the Hotan-Tiklik traverse reaches more than ca. 34 ± 6 km. This is compatible with previous estimations (e.g., Lyon-Caen and Molnar 1984; Cowgill et al. 2003; Jiang et al. 2013; Wang et al. 2014; Suppe et al. 2015). In addition, our study determined the partition of the total horizontal shortening magnitude in specific structures. Within the total horizontal shortening, the Tiklik fault absorbed more than ca. 7 ± 5 km, the Hotan anticline absorbed ca. 4 ± 2 km, and the Hotan detachment fault extending to the Mazatagh fault (Liang et al. 2012, 2014; Wang et al. 2014; Suppe et al. 2015) absorbed ca. 23 ± 1 km of horizontal shortening magnitude.

Our data and interpretations reveal the geometry, deformation timing, and shortening magnitude of the Hotan-Tiklik traverse. However, the kinematic evolution and its relationship with related surface geology, in particular in the Aqike and Buya regions, remain unclear and need more discussions.

The AFT data and related thermal history modeling results suggest that the hanging wall of the Tiklik fault underwent two stages of exhumation, an earlier stage during the late Oligocene to early Miocene followed by a later stage since the mid- to late Miocene (fig. 5), caused by activity of the fault during these periods. The later stage corresponded to synchronous activity of the Hotan fault, documented by growth strata related to the Hotan anticline (fig. 3b, 3c). The AFT results in this study indicate that the samples from Pt strata have been uplifted by ca. 5–10 km since the late Oligocene–early Miocene, suggesting a depth level already above the detachment layer along the lowermost Pz₁ strata and allowing the Tiklik fault to already have the detachment fault-flat segment before this time (Suppe et al. 2015; fig. 6). However, slipping along the detachment would not allow the Pt strata to be carried in the hanging wall of the Tiklik fault, requiring another fault-ramp farther south truncating the Pt strata (Suppe et al. 2015; fig. 6). Suppe et al. (2015) proposed another deeper detachment bed along the midcrust to generate the Tiklik fault, consistent with seismic ambient-noise tomography suggested by Li et al. (2012). We tentatively summarized three major stages for kinematic evolution of the Hotan-Tiklik traverse. The first stage prior to the late Oligocene–early Miocene was featured by the four-segmented (lower fault-flat and fault-ramp and upper fault-flat and fault-ramp) Tiklik fault, during which the AFT samples in the study were located at the depth level below the upper fault-flat (fig. 6). The second stage during the late Oligocene–early Miocene was characterized by hanging wall slipping along the Tiklik fault to uplift the AFT samples in this study to temperatures at PAZ of AFT corresponding to a depth level above the upper fault-flat (figs. 5, 6). The third stage since the mid- to late Miocene was featured by that Hotan fault generated along the upper fault-flat (detachment layer along the lowermost Pz₁ strata) of the Tiklik fault and formed a fault-ramp until it slipped along the upper detachment layer along the lowermost Paleogene strata to form the fault-bend-fold Hotan anticline (fig. 6). During the third stage, the hanging wall of the Tiklik fault slipped along the fault to uplift AFT samples in the study through the upper part of the PAZ onto Earth’s surface (figs. 5, 6).

The geometry of structures would be more complex in the Aqike region, where a WSW-dipping thrust fault carries D, C, P, E, N₂, and Q₁ strata in its hanging wall to form the Aqike anticline (fig. 1c; Shaanxi Geological Bureau 2006). Yang et al. (2009), based on detailed field mapping, interpreted the anticline as a fault-propagation fold. Given the fact that the thrust fault is bounded to the south by the Tiklik fault and Paleozoic strata (D, C, and P) crop out at the core of the anticline, we tentatively interpret the surface geology to be caused by a lateral fault-ramp that generated from the upper detachment layer at the footwall of the Tiklik fault. The contact relationships between the Cenozoic strata are compatible with the proposed kinematic evolution model in this study. We speculate that the lateral fault-ramp associated with the Aqike anticline in its hanging wall formed during the third stage (mid- to late Miocene), synchronous with the Hotan fault, causing the N₁ sequence to be eroded and N₂ to overlie the E strata in an unconformity (fig. 1c).

With regard to the Buya region, the Tiklik fault carries Cenozoic (E, N₁, N₂, and Q), J, and C-P strata in its hanging wall (fig. 1c). The contact relationships between N₂, N₁, and E are unconformities, whereas N₁ sits above C north of Buya and N₂ overlies Pt farther north (fig. 1c; Shaanxi Geological Bureau 2006). Three likely factors or their integration would cause these surface geology features and stratigraphic relationships. (1) Shortening magnitude decreased to the east in the Tiklik thrust zone, with the Tiklik fault plunging to the east (e.g., Jiang et al. 2013). (2) Fault-bend folding (consistent with a fault concave) controlled the formation of the Buya anticline (Suppe et al. 2015). (3) Lateral fault ramps, respectively, at the east end of the Tiklik thrust zone and the west end of the Buya region associated with a footwall fault ramp just north of Buya controlled the deformation in the Buya region. These factors or their integration would result in shallower stratigraphic units cropping out toward the east, Cenozoic strata being deposited at the east end of the Tiklik thrust zone, and Buya syncline forming at the hanging wall of the Tiklik fault. In addition, the kinematic evolution noted above is compatible with the contact relationships between the Cenozoic units. Activity of the Tiklik fault during the late Oligocene to early Miocene would result in the unconformity between N₁ and E to the west of Buya, erosion of E and overlying of N₁ above older strata (C) just north and northeast of Buya (fig. 1c). Later activity of the Tiklik fault since the mid- to late Miocene is consistent with the unconformity between N₂ and N₁ and the erosion of N₁ associated with N₂ sitting above Pt strata (fig. 1c).

Our data indicate that the Hotan-Tiklik segment of the western Kunlun thrust belt underwent significant activity starting around the late Oligocene to early Miocene, which is consistent with previously reported AFT ages in the region (e.g., Sobel and Dumitru 1997; Arnaud et al. 2003; Wang et al. 2003; Li et al. 2007; Li 2008) and also the basin subsidence acceleration at this time (e.g., Jiang and Li 2014). This stage of tectonic event has been documented by thermochronological ages around the margins of the western Tarim Basin (e.g., Hendrix et al. 1994; Sobel and Dumitru 1997; Yang et al. 2003; Sobel et al. 2006). The tectonic event was interpreted to mark thrusting and exhumation of the surrounding orogens around the Tarim Basin (e.g., Sobel and Dumitru 1997; Sobel et al. 2006), consistent with synchronous underthrusting of the Tarim block documented by geophysical data (e.g., Matte et al. 1996; Gao et al. 2000; Kao et al. 2001; Li et al. 2001, 2012; Jiang et al. 2004; Wittlinger et al. 2004; Schneider et al. 2013; Sippl et al. 2013a, 2013b; Schurr et al. 2014). Combining these thermochronological and geophysical results, synchronous activity of the Kashgar-Yecheng transfer system has been interpreted as a subduction-transform edge propagator fault (Sobel et al. 2013).

The tectonic event since the mid- to late Miocene has been documented by the activity of the Hotan fault and the cooling of samples from the hanging wall of the Tiklik fault in this study. The event is consistent with previous thermochronological results along the orogens around the Tarim Basin (Wang et al. 2001, 2003; Wan and Wang 2002; Li et al. 2007; Li 2008). To our knowledge, it may directly represent basinward propagation of the thrusting along the western Kunlun (Wu et al. 2004; Liang et al. 2012, 2014; Jiang et al. 2013; Wei et al. 2013; Jiang and Li 2014) and also in the western South Tian Shan (e.g., Heermance et al. 2008) north of the Tarim Basin. The mountain-building process during this stage has also been observed, supported by significant activity of the Tam Karaul and Karakax faults documented by the thermochronological ages since the mid- to late Miocene obtained from the South Kunlun and Tianshuihai terranes (e.g., Wang et al. 2001, 2003; Wan and Wang 2002; Li et al. 2007; Li 2008). In addition, shortening partition results in the study suggest that within the more than ca. 34 ± 6 km total shortening, 27 ± 1 km shortening was absorbed by structures (Hotan anticline and Hotan detachment thrust) related to detachment layers (along lowermost Pz₁ and lowermost E strata) during basinward propagation of thrusting in this stage, suggesting that the detachment layers played an important role in propagating thrusting from the western Tibetan Plateau into the Tarim Basin.

The interpretation of seismic profile in the Hotan-Tiklik segment of the western Kunlun thrust belt indicates that the segment consists of Hotan and Tiklik thrust zones. The Hotan zone is characterized by the three-segmented (two fault-flats and a fault-ramp) Hotan fault and related fault-bend anticline (the Hotan anticline) in the hanging wall, while the Tiklik zone is likely featured by the four-segmented (two fault-flats and two fault-ramps) Tiklik fault. The growth strata suggest that the Hotan fault initiated activity since the mid- to late Miocene, while the fission track data and related thermal history modeling results document the Tiklik fault activated initially during the late Oligocene–early Miocene and again since the mid- to late Miocene. The late Oligocene–early Miocene event may mark the underthrusting of the Tarim block beneath the western Tibetan Plateau, while the mid- to late Miocene event likely represents basinward propagation of the thrusting.

The structural balanced cross section and fission track results suggest that the Hotan-Tiklik traverse contributes more than ca. 34 ± 6 km of total horizontal shortening. The total shortening was absorbed by major structures, with the Tiklik fault absorbing more than ca. 7 ± 5 km, the Hotan anticline absorbing ca. 4 ± 2 km, and the Hotan detachment fault extending to the Mazatagh fault absorbing ca. 23 ± 1 km of horizontal shortening magnitude. This shortening partition suggests that a detachment layer in the Tarim Basin played an important role in propagating deformation from the Tibetan Plateau into the cratonal regions.

This work was funded by the National Science Foundation of China (grants 41472181, 41472182, 41402170, 41330207, 41372206, 41102128, and 41072154), the National S&T Major Project (grants 2016ZX05008-001 and 2016ZX05003-001), Fundamental Research Funds for the Central Universities (grant 2016FZA3007), and the Scientific Research Fund of the Zhejiang Provincial Education Department (grant Y201019040). We thank A. Yin for comments on an earlier draft and X. Wang for help in revising the manuscript. We are grateful for comments from E. Sobel and Editor D. Rowley that significantly improved the manuscript.