The Miocene Acceleration of Strike-Slip Deformation in the Northern Tian Shan, Central Asia Open Access

The evolution of the Tibetan Plateau and its surrounding regions is key for understanding the continental collision dynamics and the associated climate change in Central Asia. During the late Cenozoic, crustal stress from the converging Indian and Eurasian plates, or due to the growth of Tibetan topography, propagated northward into Central Asia [1, 2]. This process led to the reactivation of weak structures in the Tian Shan that were inherited from the Paleozoic orogeny [3, 4]. The unroofing of the Tian Shan serves as a preeminent example of how the continental interior responded to the distant India–Eurasia collision, which occurred ~1500 km to the south (Figure 1).

The Cenozoic uplift of the northeastern Tibetan Plateau was initially uplifted during the late Eocene–late Oligocene, as recorded by low-temperature thermochronology [5-8], while in the far-field Tian Shan, thermochronological studies suggest an extensive initial late Oligocene–early Miocene onset of the uplift and unroofing [9-12] as a tectonic reactivation of the complex Paleozoic and Mesozoic lithospheric structures [13, 14], with a possible subsequent acceleration over the last ~15–10 Ma [10-12, 14-19]. Therefore, the initial Cenozoic uplift of Tian Shan was either simultaneous or following that of northern Tibet, supporting it as a distant effect of the India–Eurasia collision, possibly due to the stress transferred from the elevated Tibetan Plateau through the rigid Tarim Basin [1, 20]. However, after their initial uplift, the link in evolution between the Tian Shan and northeastern Tibet has not been investigated in detail. For example, the deformation style of northeastern Tibet changed at ~15–10 Ma, from contraction along the plate convergence direction to the lateral, west-east motion, marked by the onset of transcurrent faulting in the region [21, 22]. It remains unclear that whether or how this change has influenced the deformation of the Tian Shan.

While most of the crustal shortening in the Tian Shan has been accommodated by the deformation of fold-and-thrust belts along the mountain boundaries [23, 24], deformation on regional-scale strike-slip faults within the mountain has also been clearly recognized [25, 26]. These structures include several northwest-southeast trending dextral faults, such as the Bolokenu-Aqikekuduk Fault (BAF, also named as Dzhungarian Fault), running parallel to the northern Tian Shan, and the Talas-Fergana Fault (TFF) on the boundary between southwestern Tian Shan and the Pamir, as well as the southwest–northeast trending sinistral Nalati Fault (NF) in the central Tian Shan (Figure 1). These major strike-slip faults are commonly oriented at 60°–75° in map view relative to the maximum compressive stress direction (σ₁), forming V-shaped conjugate fault sets. The transpressional deformation along these faults not only accommodates significant crustal shortening in the nearly north–south direction but also extension in the east–west direction [25-30]. Some studies suggest that the structural style changes are often observed during the later stages of convergent orogen development [22, 31]. Therefore, constraining the Cenozoic kinematic history of strike-slip faults within the Tian Shan, especially the timing and duration of their activity, is crucial for understanding the evolution of Tian Shan and its response to the collision dynamics and plateau uplift farther south.

To investigate the exhumation history of the Tian Shan and its relationship with large-scale strike-slip faults, we constrained thermal history and exhumation models of samples from two sites near the western segment of the BAF and the eastern end of the NF (Figure 1), using new and published apatite fission-track (AFT) and apatite (U-Th)/He (AHe) data, respectively. By comparing the estimated million-year-scale exhumation rates with previously estimated kilo-year-scale late Quaternary fault slip and uplift rates, our results provide constraints on the transpressional deformation history of the Tian Shan. Because of the complex and asynchronous distribution of deformation across the Tian Shan, it is challenging to make generalizations about the entire range based on observations from only a small part of the mountains. To address this, we compiled over 1000 new and existing AFT and AHe ages from the Tian Shan, northeastern Pamir, and the northeastern Tibetan Plateau (Figure 1) to estimate the correlation between the deformation of the Tian Shan and other regions around the Tarim Basin. The compiled ages are used to constrain a large-scale exhumation model, reveal the main features of exhumation rate evolution, and identify the potential impact that the large strike-slip faults might have on the exhumation.

The Tian Shan is a ∼2500 km long and 300–400 km wide orogenic belt, comprising a series of basement-cored ranges divided by strike-slip faults and intermontane basins. It is located between the Tarim Basin to the south and the Kazakh platform and Junggar Basin to the north (Figure 1). The ancestral Tian Shan orogen was formed during the Paleozoic plate convergence, through the accretion and amalgamation of microcontinents, accretionary wedges, and volcanic arcs of the paleo-Asian Ocean, which separated Eurasia and Gondwana [4, 32, 33]. The Mesozoic experienced localized deformation and small-magnitude rock exhumation within the Tian Shan [14, 16]. The entire orogen was reactivated in the late Cenozoic in response to the collision of India into Eurasia [1].

Ongoing indentation is the main driving force behind the present-day tectonic activity in the Tibetan Plateau and Tian Shan. GPS data reveal large velocity gradients existing across major strike‐slip faults [34]. Strike-slip earthquakes mainly occur in the southwestern Tian Shan, near the boundary with the Pamir block, and within the interiors of the northern and central Tian Shan, for example along the BAF and NF, respectively. These faults were (re)activated episodically in response to distal plate collisions . The BAF, stretching over 1000 km from the Kazakh Platform to the Turpan Basin and transecting the northern Tian Shan obliquely, is an inherited structure from the Paleozoic deformation. The late Quaternary mean dextral slip rate on the BAF is up to 5 mm/yr near the western Junggar Basin and decreases to 1 mm/yr toward its eastern end [35]. The NF, generally running along the strike of the southwestern Tian Shan, was also part of the Paleozoic suture zone between the Kazakh and Tarim blocks (Figure 1). This fault zone experienced significant transpressive deformation during the Paleozoic collision and accumulated a total of 60 km sinistral displacement [36]. Further to the southwest, the TFF, another seismically active inherited Paleozoic structure, has accumulated a total of over 200 km right-lateral offset since its formation [25].

Low-temperature thermochronology is based on concentrations of parent isotopes and daughter products (e.g. fission tracks and alpha particles) of radioactive decay measured in minerals and is widely used in tectonic studies for estimating rock cooling and denudation histories. AFT and AHe thermochronometers are sensitive to the thermal evolution of rocks at temperatures below ~120 and ~ 70°C, respectively [37, 38], so are the most effective in resolving the rock exhumation history within the upper 4 km of the crust. Increases in exhumation rates are a common consequence of high topographic relief generated by dip-slip faulting but could also be related to deformation on strike-slip faults with a measurable vertical component [22, 39].

We collected a total of eleven samples along the northern Tian Shan that were analyzed for AFT and AHe dating (Figure 2, Table 1). Among these, three samples (TS1918, TS1919, and TS1922) were from granitic intrusion rocks around Jinghe county, and one was from the middle segment of BAF (TS1933). The remaining seven samples were collected from the Jurassic and Carboniferous sandstone along the Toutun river and Hutubi river on the southern margin of the Junggar Basin. Samples were crushed, sieved, and processed through traditional magnetic and density method to separate apatite crystals.

The AFT analysis was performed at the State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, using the laser ablation-inductively coupled plasma mass spectrometry (LA-ICP-MS) method [40, 41]. Apatite grains were mounted on glass slides with araldite epoxy. After grinding and polishing to expose internal surfaces, spontaneous tracks were etched with 5.5 M HNO₃ at 21°C for 20 seconds [38]. ²³⁸U concentration was measured using LA-ICP-MS, and the National Bureau of Standards trace element glass NIST612 was used as the external standard. Fission-track density and length were measured under a Zeiss Axioplan 2 microscope, with a dry objective and total magnification of 1000. Horizontal confined tracks for each sample were measured where possible. AFT ages were calculated using the formula described by Gleadow et al. [42].

AHe analysis was also conducted at the Institute of Geology, China Earthquake Administration following the standard procedures described in Farley and Stockli [43]. Euhedral crystals were handpicked from separated apatite grains. For each sample, five apatite crystals were selected and photographed, and grain dimensions were measured under a high-power binocular microscope with a magnification of 100. Apatite grains were degassed by heating with an Nd‐YAG laser (8 A current) for 5 minutes, and He was measured with an Australian Scientific Instruments Alphachron noble gas mass spectrometer. After degassing, grains were dissolved in the 7 mol/L nitric acid with spikes of ²³⁵U and ²³⁰Th. The solutions of U and Th were analyzed by an Agilent 7900 ICP-MS. The raw AHe ages were corrected for α-ejection according to Farley et al. [44].

Near the BAF in the northern Tian Shan, we simulated the rock cooling histories using new AFT and AHe data from the plutonic rocks. On the southern margin of the Junggar Basin, five of the seven Mesozoic sedimentary rock samples were used for thermal history modeling. Despite the relatively older AFT ages, AHe data of these samples yield consistent Cenozoic ages, indicating that they had been fully reset after deposition. Two samples (TS1902 and TS1907) were not used for modeling, due to their scattered and mostly pre-Cenozoic single-grain AHe ages. In addition to the simulation of new data, we also simulated the cooling history on the southern margin of the Nalati range near the eastern NF (Figure 1), using AFT and AHe data from two samples (Bay D and Bay E) reported by Jolivet et al. [16]. This site was chosen due to its proximity to the NF and the availability of both AFT and AHe. The modeling procedure optimizes the thermal history using a Bayesian transdimensional Markov Chain Monte Carlo (MCMC) method [45]. Each forward model of the thermal history is defined as a limited number of time-temperature points, and the model likelihood is quantified by the fit between the observed and predicted thermochronological data. To avoid overcomplex models, the method adds a penalty to the likelihood of increasing the number of time-temperature points to calculate the posterior probability. AFT and AHe data were predicted using the annealing model of Ketcham et al. [46] and the He retention model of [47], respectively. During the inversion, a temperature offset between the uppermost and lowermost samples on a near-vertical profile is also sampled, allowing an estimate of the paleogeothermal gradient. The random walk is performed in the three-dimensional parameter space consisting of time (0–200 Ma), temperature (0–150°C), and temperature gradient (20–40°C), assuming a uniform prior distribution for all model parameters. The present-day temperature of the uppermost sample on a profile is prescribed at 5–25°C. For each inversion, we ran 200,000 MCMC iterations and discarded the initial 100,000 as the “burn-in” with the rest kept in the “post-burn-in” ensemble.

We compiled >1000 published and new AFT and AHe ages that are younger than 100 Ma from the Tian Shan and adjacent regions (Figure 1, online supplementary Table S1). The compiled ages were modeled using the method described by Fox et al. [48] and modified by Herman and Brandon [49] to infer the exhumation rate. This modeling approach considers the closure depth of a thermochronological age as a summation of exhumation rates of a limited number of time steps, over the period from the time indicated by the apparent cooling age to the present day. Dodson [50] approximation is used to estimate effective closure temperature of an age, which is dependent on the cooling rate when the rock passed through the closure temperature. The thermal model of the crust is solved in one dimension at each sample location. The a priori rate of 0.05 km/Ma was set according to the observation that in regions with slow exhumation in Tian Shan, the total exhumation (<4 km) in the Cenozoic has not reset the AFT ages (e.g. Dumitru et al. [51]). Our thermal model has a thickness of 30 km with surface and base temperatures at 15 and 615°C, respectively. This initial model setup mimics the current mean geothermal gradient (~20°C/km) in the Tarim Basin [52], where we assume negligible exhumation has taken place. We used a spatial correlation distance (L_c) of 30 km and a time interval length (Δt) of 5 Ma for our reference model, but also tested different values for L_c (2–50 km) and Δt (3, 4, and 5 Ma). Each inversion contains twenty iterations. Detailed descriptions of the inversion method as well as discussions of its limitations were provided by Fox et al. [53], Herman and Brandon [49], Stalder et al. [54], and Willett et al. [55].

The eleven new AFT ages reported vary between 37.6 and 213 Ma (Table 2). Ten of them yield Mesozoic ages while only one has Cenozoic age. Figure 3 shows the radial plots of grain ages. Track length distributions are also shown for all samples except TS1909 and TS1933, in which only a very small number (≤5) of confined track lengths could be measured. The mean D_par values from these samples span between 1.58 and 3.19 µm, with the measured fission-track lengths ranging from 11.3 to 13.65 µm (Figure 3). All ages passed the chi-squared (χ²) test with low dispersion. The range and average values of length measurements remain relatively consistent despite differences in the AFT ages of the samples. AHe ages derived from same set of samples with the AFT samples. A total of fifty-four single-grain ages show a significant variability, ranging from 8.11 to 237.73 Ma (Table 3). We excluded one sample (TS1902) from interpretation due to its large grain age dispersion. Additionally, we excluded single grain AHe ages prior to the Cenozoic from our thermal history modeling.

At the western BAF site (Figure 4(a)), thermal history modeling results suggest that the rock had been exhumed to shallow levels of the crust, in the temperature range of ~55–65°C, by ~50 Ma (Figure 4(c)). Then, the samples experienced a long-term thermal quiescence until their final exhumation to the surface during the late Cenozoic. As the modeled temperature prior to the final cooling was near the upper boundary of the partial annealing/retention zones of the AFT and AHe methods, the predicted onset age of the final rapid cooling should be treated with caution (see Discussion 5.1). Throughout the Cenozoic, the modeling results suggest a paleogeothermal gradient at ~20°C/km (Figure 4(b)). This final cooling event (<70°C; Figure 4(c)) is of moderate magnitude and therefore has not fully “reset” the AFT and AHe ages.

At the east NF site (Figure 4(d)), the thermal history models suggest very similar rock cooling paths as that from the western BAF: the rocks were cooled to <60°C before ~50–40 Ma and had most likely remained at the same temperature until the final cooling during the late Cenozoic (Figure 4(f)). At this site, the models are based on data from only two samples on a profile with <100 m offset in elevation, and thus, the post-burn-in ensemble contains a larger uncertainty (Figure 4(f)). As a result, the time for final exhumation of the rocks during the late Cenozoic tectonics cannot be tightly constrained. Based on a paleogeothermal gradient of 20–30°C/km, the predicted total magnitude of the final exhumation is less than 2 km.

Among the five modeled samples on the southern margin of the Junggar Basin, four thermal history models (TS1909, TS1912, TS1913, and TS1914) show an apparent late Cenozoic increase in the cooling rate (Figure 5). The models of TS1909 and TS1913 indicate the onset of cooling rate acceleration around ~20 Ma. This is consistent with the rock cooling histories inferred from AFT and AHe data on the margin along the Manas and Guertu Rivers [9, 56]. Compared to the regions further west, the total magnitude of exhumation and cooling appear to lower near the eastern end of northern Tian Shan (Figure 2), where all AFT and some AHe data have not been reset. As a consequence, some of the modeled thermal histories either show no late Cenozoic acceleration in cooling rates (TS1911; Figure 5(b)) or the timing of this acceleration is not well constrained (TS1912 and TS1914; Figures 5(c) and 5(e) ).

The performances of inverse exhumation models, which were set up with various L_c and Δt values, are shown in Figures 6 and 7, respectively. All optimized models with different L_c and Δt predict normal distributions of age misfit centered at 0 Ma (see Figures 6(b) and 7(b)). For models with the same Δt, those set up with a higher L_c show faster decreases in the residual; models with L_c >20 km achieved minimized residuals after ten iterations (Figure 6(a)). Similarly, among models with the same L_c, those with longer Δt show quicker reductions in residuals (Figure 7(a)). The pre-Cenozoic part of the exhumation history was poorly solved in our inversion. The model with 30 km correlation distance and 5 Ma time interval yields good convergence performance and reasonable fit to the data and thus is presented here; the resolved exhumation rates and their corresponding resolution for this model are shown in Figures 8 and 9.

Our tests of models with varying time interval lengths suggest that the choices of L_c and Δt have a more significant influence on the estimated exhumation rates than the timing of rate changes (Figures 10 and 11). We selected six representative locations (labeled a–f) along major strike-slip faults and northeastern Tibet frontal thrusts (Figure 8) to show the predicted exhumation rates from models with different correlation distances (Figure 10) and time intervals (Figure 11). Our results suggest that while the Cenozoic unroofing of the Tian Shan may have occurred since 25–20 Ma, substantial increases in exhumation rates at many locations also occurred between approximately 15 and 10 Ma.

Our thermochronological data offer constraints on the Cenozoic deformation of the BAF in northern Tian Shan. Throughout the early Cenozoic, the modeled thermal history suggests no significant cooling that could be related to exhumation along the fault planes. During this period, based on the modeled temperature offset between the uppermost and lowermost samples on the elevation profile, the simulated geothermal gradient is 20°C/km. This value aligns with current value measured from the Tarim Basin [52], which may reflect the thermal structure of the crust in this region with no strong tectonic activity. On the BAF, although the predominant motion during the Cenozoic is strike-slip, a significant dip-slip component exists [57]. Therefore, we infer that this Paleozoic structure was not likely active over the early Cenozoic.

During the late Cenozoic, based on the modeled geothermal gradient at ~20°C/km, the final exhumation since the Miocene had a magnitude of <2.3 km. The best-fit models from the post-burn-in ensemble suggest that rapid cooling started at ~10 Ma and thus indicates an exhumation rate of ~0.2 km/Ma. It is worth noting that prior to the final cooling, the samples resided at a relatively low temperature (<70°C) for a long period (>30 Ma), and thus, the thermochronological data may not be sensitive to provide a tight constraint on the onset age of the final exhumation and therefore the rate of this exhumation. However, we argue that the predicted temperature ranges for the samples during the early Cenozoic quiescence are reliable. Therefore, a trade-off should exist between the exhumation rate and the onset time of the final exhumation, such that a shorter duration of the final exhumation requires a higher exhumation rate. For example, if the final exhumation took twice as long as our best-fit model prediction, that is it started at ~20 Ma, the exhumation rate would be half of the model prediction, that is at ~0.1 km/Ma. During the Cenozoic, thermal history models from previous work [9, 51, 56] and this study (Figure 5) suggest that the northern Tian Shan started to unroof during the Oligocene to early Miocene. Sedimentary records along the southern margin of the Junggar Basin indicate possible acceleration in the middle Miocene [58]. We propose that the BAF was most likely reactivated during these periods, and near the Jinghe region, displacement on the fault resulted in an average exhumation rate at ~ 0.1–0.2 km/Ma.

At the site near the east NF, due to the limited elevation difference between samples on the vertical profile (Figure 4(d)), the uncertainties in estimating the magnitude and rate of the final cooling and exhumation (Figure 4(f)) are even larger compared to those from the BAF. This uncertainty is clearly presented by thermal history models in the post-burn-in ensemble, as well as in the probability distributions of the onset age of the final exhumation and the sample depths during the early Cenozoic (Figure 12(b)). Similarly, our exhumation rate models, simulated with different time interval lengths, also reveal a discernible negative linear correlation between the rate and duration for the last phase of rapid exhumation at some sites (Figure 11). This implies that low-temperature thermochronological data provide a better estimate of the total magnitude than either the age or rate of the late Cenozoic exhumation independently.

Over the late Quaternary, a substantial portion (23%) of crustal shortening in the northern Tian Shan has been accommodated by strike-slip faulting [59]. On the Jinghe segment of the BAF based on the displaced geomorphic markers and age constraints, Hu et al. [59] calculated a late Quaternary dextral slip rate at 3.2 + 1.4/−1.1 mm/yr and a secondary but significant uplift of 8 m over 74 ka; the strike-slip rate is consistent with the post-Pliocene average of 3 ± 1 mm/yr [60]. Lu et al. [61] compared the published late Pleistocene–Holocene deformation rates and the modern GPS measurements in the northern Tian Shan and suggested that the deformation pattern and rates of both the mountain range and its foreland have remained constant throughout the late Quaternary. Assuming that this steady deformation pattern also applies to the segment of BAF in the northern Tian Shan, the observed ratio between the strike-slip and uplift rates on the fault is expected to have remained constant over the Quaternary, leading to a total uplift of >200 m. Therefore, both the post-Pleistocene and the Quaternary average uplift rate estimates are on the same order of magnitude as the prediction of our thermal history model (Figure 12(c)), which suggests a total exhumation of ~2 km over the past 10 ± 5 Ma for the lowermost sample on the profile at the Jinghe site (Figure 12(a)). Even if the late Cenozoic activity on the BAF started in the early Miocene 25–20 Ma and the mean exhumation is slower (see Section 5.1), this uncertainty does not affect the consistency between the inferred exhumation rate and the Quaternary uplift rates. Given the arid climate, the average denudation rate of Tian Shan may be outpaced by the tectonic uplift [62]. However, near the boundary between mountain and foreland, especially close to the fault scarps, the localized exhumation rate could be much higher and comparable to the uplift rate [63]. This consistency in uplift or exhumation rates over different time scales implies that the uplift and deformation rates on the BAF have been relatively constant since its Miocene reactivation (Figure 12(c)).

Near the Bayanbulak Basin on the eastern NF, Wu et al. [30] measured a vertical offset of >7 m on an alluvial fan that was abandoned at ∼59 ka, which is similar to the observed displacement on the BAF during the same period (Figure 12(c)). This uplift rate estimate is also consistent with the mean exhumation rate (∼0.1 km/Ma) on the basin margin predicted by our thermal history model, despite its large uncertainties (Figure 12(c)).

The spatial variation in precipitation rate has been invoked as a key factor for the different erosion rates in Central Asia [64], and climatic change is unlikely to be the driver for the late Cenozoic increase in exhumation rate in the Tian Shan. Since the early Eocene, the uplift of the Tibetan Plateau gradually blocked the northward transport of monsoonal moisture from South Asia, and as a result, the climate in interior Asia has been dominantly controlled by the midlatitude westerlies [65]. During the Neogene, the uplift of Tian Shan and other mountain ranges in the region decreased the moisture transport by the westerlies, leading to a progressive aridity of the interior of Central Asia [66]. Therefore, we attribute the Miocene increase in exhumation rate around the Tarim Basin to the evolution of the tectonic deformation rather than climate change.

In both northern and southwestern Tian Shan, the Miocene increases in exhumation rates occurred not only along the mountain-basin boundaries but also in the interior of the ranges (Figures 8, 10(b), and 10(c) and 11(b)). This cannot be explained solely by a sequential thrust model, where deformation propagates from the high mountain toward basin margin [67, 68]. In addition, the general spatial correlation between rapid exhumation (i.e. sites b, c, and d in Figures 8, 10, and 11) and large strike-slip faults (i.e. the BAF and NF) suggests a tectonic control on the renewed or intensified episode of crustal deformation [69-71], implying a possible transition in deformation style from predominant thrust faulting to more active strike-slip faulting. It is worth noting that after 15 Ma around the Tarim Basin, the NE margin of the Tibetan Plateau (Figures 10(f) and 11(f)) and the NE Pamir (Figures 10(a) and 11(a)) also experienced important changes in the deformation style, suggesting a potential correlation in the deformation mechanism of these regions surrounding the Tarim Basin.

To the southeast of the Tarim Basin, deformation of the northeastern Tibetan Plateau is mainly accommodated by sinistral slip on the Altyn Tagh Fault (ATF) and shortening on the northwest–southeast trending fold-and-thrust belts [72]. Currently, the strike-slip deformation on the northwestern and southeastern margins of the Tarim Basin complies with a model in which the internally stable Tarim block rotates clockwise relative to the Eurasian plate [34] (Figure 1). Similar to the Tian Shan, the present-day deformation rates in northeastern Tibetan Plateau are also consistent with millennial-scale rates estimated from ages and offsets of geomorphic features [73, 74]. Extrapolation of the late Quaternary slip rates on two faults conjugate to the ATF suggests a total displacement accumulated over 8–12 Ma [75], and thus, the faulting activity probably started when the deformation of northeastern Tibet changed from nearly north-south shortening to eastward expansion at ∼15–8 Ma [21]. This transition is coincident with or younger than our inferred age of the accelerated strike-slip deformation in Tian Shan, suggesting that by then a clockwise rotation of the Tarim block relative to Eurasia has been established. Moreover, if the rotation rate (0.63°/Ma) modeled from geodetic data [34] is representative of the mean rate over the geological time scale, the total late Cenozoic rotation of the Tarim block (7° ± 2.5°) [13] would also suggest a mid-Miocene (~11 Ma) onset of the deformation around the basin margins.

Between the Tian Shan and northeastern Tibetan Plateau, the apparent synchronization in their transitions of deformation styles supports the significant role of the Tarim block in the strain distribution within the interior of the Eurasian Plate during the continental collision. It has been demonstrated that the presence of the strong Tarim block not only transformed the deformation of the Tibetan Plateau from contraction in the plate convergence direction to W-E motions toward weaker zones but also, due to its weak internal deformation [76], effectively transferred strains to the far-field domains from the collision boundary [77, 78]. Our work further indicates that the transferred strains were not only restricted to crustal shortening on the direction of India–Asia convergence and consequent thickening but also the lateral deformation on the south margins of the Tarim block. The relative rotation of the Tarim is a result of differential shortening across as well as strike-slip deformation along the basin margins [76], which may be related to the asymmetrical lateral extrusion occurring between the northwestern (i.e. the NE Pamir) and northeastern parts of the Tibetan Plateau. Due to the complexity of the preexisting Paleozoic structures, this lateral deformation in Tian Shan is accommodated by motions on a series of dextral, northwest–southeast trending faults (e.g. the BAF and TFF) and their conjugate, the sinistral NF.

We presented new AFT and AHe data from the northern Tian Shan near the western BAF and the southern margin of the Junggar Basin, which are used to invert thermal history models to investigate the relationship between rock exhumation and the major strike-slip structures. The modeling results suggest that following the initial unroofing of northern Tian Shan during the late Cenozoic, there was a significant increase in the exhumation rate near the western BAF during the Miocene, with a total exhumation magnitude of ~2 km. This increase is consistent with the uplift magnitude estimated by extrapolating the late Quaternary uplift rate on the BAF to the modeled onset of final exhumation, supporting that the Paleozoic fault was reactivated during the Miocene. This inferred timing of the strike-slip deformation acceleration in the northern Tian Shan coincides with other major changes in the deformation pattern along the margins of the Tarim Basin, such as the eastward expansion of the northeastern Tibetan Plateau and the northward advance of the Pamir. These observations imply that the current lateral deformation around the Tarim Basin was established by the Miocene.

Readers can access the data online at doi:10.5281/zenodo.7693579 via the Creative Commons Attribution 4.0 International license.

The authors declare that there is no conflict of interest regarding the publication of this article.

S.W. was supported by the China Scholarship Council and the Graduate Award from the University of Victoria. R.J. was funded by the Natural Sciences and Engineering Research Council of Canada (RGPIN-2019-04405).

Table S1 includes the thermochronological data used to model exhumation. We only compiled and modeled the thermochronological data younger than 100 Ma.