New apatite (U-Th)/He from the northeastern margin of the Tibetan Plateau (north Qilian Shan) indicate rapid cooling began at ∼10 Ma, which is attributed to the onset of faulting and topographic growth. Preservation of the paleo-PRZ in the hanging wall and growth strata in the footwall allow us to calculate vertical and horizontal fault slip rates averaged over the last 10 Myr of ∼0.5 mm/yr and ∼1 mm/yr respectively, which are within a factor of two consistent with Holocene slip rates and geodetic data. Low fault slip rates since the initiation of the northern Qilian Shan fault suggest that total horizontal offset did not exceed 10 km. Further, emergence of the northern Qilian Shan occurs during a period of increased aridity in northern Tibet but is associated with only a minor expansion of the northern plateau perimeter, which is well established near collision time. Outgrowth of the northern Qilian Shan at ∼10 Ma could be simple propagation of the larger Qilian Shan system, occurring in response to decreased slip rates on the Altyn Tagh fault or as a result of the change in GPE of the central plateau.


The large size and youthfulness of the Tibetan orogen make it a prime location to study topographic growth, erosion and the effect of topography on climate. Mountain range exhumation histories and lithologic changes in basin stratigraphy in northeastern Tibet have been used as a proxy for the development of high mean elevations. However, pre-Cenozoic cooling ages from exhumed fault blocks and lack of precise age data in basin deposits make linkages between orogenesis, exhumation/sedimentation, and climate tentative (e.g., Metivier et al., 1998; Zheng et al., 2000; Jolivet et al., 2001; Fang et al., 2003).

The Qilian Shan lies along the northeastern margin of the Tibetan Plateau—a location where the initial timing of plateau growth is poorly known (Fig. 1). Thrust faulting in the Qilian Shan has been linked to crustal thickening and topographic growth, as well as to the accommodation of motion of the Altyn Tagh fault by the transfer of left-lateral strike-slip motion to oblique thrusting (Burchfiel et al., 1989; Peltzer et al., 1989; Meyer et al., 1998; Tapponnier et al., 1990). Previous suggestions for the onset time of Qilian Shan deformation, including that of the North Qaidam terrane, range from Paleocene to Pliocene time (Dupont-Nivet et al. 2004; Horton et al., 2004; Jolivet et al., 2001; Yin et al., 2002, 2008; George et al., 2001; Wang et al., 2004; Fang et al., 2004; Metivier et al., 1998). Analysis of Cenozoic stratigraphy in the western Hexi Corridor and adjacent north Qilian Shan range identifies facies changes in Miocene time related to range growth (Bovet et al., 2009). Growth strata within the dated section of the Niugetao Formation (∼9 Ma) of the Jiuxi Basin (central Hexi Corridor) suggests fault activity of the North Qilian Shan thrust during late Miocene time (Fang et al., 2004), but does not constrain fault initiation or range growth. Cooling histories from fault bounded range blocks are well-suited for determining the initiation of faulting; however, previous apatite fission-track and 40Ar/39Ar dating in hanging wall rocks have been unsuccessful in directly dating the onset of rapid cooling related to erosional exhumation of the range because with exception of a single sample, cooling ages are Mesozoic, which suggests that exhumation in response to Cenozoic thrust and reverse faulting has been insufficient to exhume completely reset ages (< 4–6 km) (Jolivet et al., 2001; George et al., 2001).

Apatite helium ages have a lower closure temperature than thermochronometric techniques previously applied in the Qilian Shan, and so potentially record a thermal history related to fault-induced erosional exhumation. Further, we discuss how the preservation of the paleo-partial retention zone (PRZ) for helium diffusion in hanging wall rocks can be utilized as a passive marker to determine subsequent fault motion. When syn-tectonic sediments are preserved in the footwall, the paleo-PRZ can be used to measure long-term geologic offsets where the appropriate stratigraphic markers in the hanging wall are absent, which is a common structural limitation in basement-cored structures (Clark and Bilham, 2008). Thus in circumstances where erosion is limited to a few kilometers, apatite helium ages provide constraints on both the initiation of faulting and fault offset, and Myr scale faulting rates can be determined.


The Cenozoic tectonics of the Qilian Shan and adjacent Hexi Corridor are characterized by folding, thrust faulting, and strike-slip faulting that partially accommodate modern India-Eurasia plate convergence (Tapponnier et al., 1990; Yuan et al., 2004). Geodetic shortening rates are 5.5 (±1.5) mm/yr (Zhang et al., 2004) between the Qaidam and Alashan blocks, and active shortening deformation to the south is distributed throughout the ∼270 km wide Qilian Shan plateau (Institute of Geology, 1993; Metivier et al., 1998; Yuan et al., 2004). The northernmost structure of the Qilian Shan, the North Qilian Shan Thrust, juxtaposes low-grade metamorphic lower Paleozoic rocks (slates, phyllites, limestones, volcanic, and granitic rocks) over Cenozoic sedimentary rocks in the Hexi Corridor basin (Gansu Geological Bureau, 1989; Fang et al., 2004).

The North Qilian Shan Thrust has formed a 2–3 km topographic escarpment above the Hexi Corridor basin. A coarsening-upward succession of lacustrine-fluvial deposits 2000–3000 m thick is preserved within the basin (Gansu Geological Bureau, 1989; Fang et al., 2004) and is dated to be mid-Oligocene to Quaternary depositional age based on paleontology (Bally et al., 1986; Wang and Coward, 1993) and magnetostratigraphy (Fang et al., 2004).

We dated 12 samples for (U-Th)/He thermochronometry that were collected from Paleozoic granites in the hanging wall of the North Qilian Shan Thrust (Fig. 1). Nine samples were collected on a ∼1100 m vertical transect within a single pluton. These samples are used to identify changes in erosion rate that may be related to fault motion. Three additional samples were collected 100–200 km along strike to the east (Fig. 1). Although the density of sampling to the east is not great enough to determine if propagation of erosion occurred along strike of the fault, these samples provide some indication of how erosion magnitude may vary spatially and thus how regional a particular erosion history may be. Samples were analyzed for single-grain apatite (U-Th)/He ages using standard procedures at Caltech (Farley and Stockli, 2002), and sample mean ages are reported as the average of 3 or 4 individual grain analyses (Supplemental Tables 11 and 22).


Apatite helium ages indicate the time at which the sample cooled through its closure isotherm or ∼60 °C for this sample suite based on a radiation damage model for He diffusion kinetics (Farley, 2002; Shuster et al., 2006). Helium ages on the vertical transect increase with elevation with a distinct change in the apparent exhumation rate at ∼2700 m. Below 2700 m elevation, analyses define a steep age/elevation gradient with an increase from 7.2 to 9.5 Ma. Above 2700 m analyses define a shallow age/elevation gradient, with ages increasing from 9.5 to 106 Ma and a pronounced change in age/elevation gradient at ∼9.5 Ma (Fig. 2). Three samples collected ∼200 km southeast of the vertical transect are shown on the same plot (gray) and yield Mesozoic ages at high elevation but younger ages at low elevations compared to the vertical transect (Fig. 2; Supplemental Table 1 [see footnote 1]). While these off-transect samples are too few to constrain a robust change in age/elevation gradient at a second location, these ages are generally consistent with a similar exhumation history along strike of the fault.


In a compressional tectonic setting, an abrupt increase in apparent exhumation rate on an age/elevation plot typically signals accelerated erosion most likely related to the upward motion of the hanging wall over the footwall (e.g., Wagner and Reimer, 1972; Wagner et al., 1977; Fitzgerald et al. 1995; Stockli et al., 2000; Reiners and Brandon, 2006). Ages that predate this transition define the base of the fossil helium partial retention zone prior to rapid exhumation, i.e., the “lower break in slope” on an age/elevation diagram and can therefore be used to reconstruct the relative elevation of the land surface prior to faulting (Clark and Bilham, 2008). Such a reconstruction provides a marker horizon in the hanging wall that can be used to determine relative motion across the fault and is particularly useful in geologic settings where stratigraphic markers are absent in the hanging wall. Because non-vertical pathways of rocks can complicate quantitative interpretation of erosion rate from age/elevation information alone (e.g., Huntington et al., 2007), we focus only on timing of abrupt change in apparent erosion rate. As described below, we derive fault slip rates from offset markers across the fault and initiation age of faulting, not from the apparent erosion rate recorded by age/elevation data.

Timing of Fault Motion

An abrupt increase in erosion rate in the hanging wall of a thrust or reverse fault likely signals fault activity and enhanced erosion at that time. The attribution of increased erosion rate to fault activity may be invalid if climate conditions caused enhanced erosion of pre-existing topography or delayed erosion following fault motion to a climate period of greater erosivity. However, ∼9 Ma growth strata found at the base of the Niugetao Fm. in the Hexi Corridor represent the initiation of deposition related to fault motion (Yang et al., 2007; Fang et al., 2004) (Fig. 1). The correlation of growth strata with an increase in exhumation rate is strong evidence of synchronous fault motion and accelerated erosion.

Regional climate conditions were also unlikely to have caused the erosion signal we observe at ∼10 Ma. Isotopic and lithostratigraphic evidence from basins across northern Tibet suggests arid conditions began in Oligocene time and is attributed to the retreat of the Paratethys epicontinental sea, global climate changes, or the early rise of mountain ranges in northern Tibet, which block moisture from the south and east (Wang et al., 2003; Graham et al., 2005; Dupont-Nivet et al., 2006). Climate proxies from Linxia Basin in northeastern Tibet suggest an increase in aridity at 12–13 Ma with the period of greatest aridity occurring between 9.6 and 8.5 Ma (Dettman et al., 2003; Fan et al., 2007). A Neogene increase in δ18O values from sediments in the Tarim and Qaidam Basins may also represent a regional shift to more arid conditions (Kent-Corson et al., 2009), correlative with an increase in dust deposition in the North Pacific and an increase in loess deposition within northern China at 7–8 Ma (Rea et al., 1998; Sun et al., 1998).

Decreased precipitation would likely cause a decrease in climatically driven erosion rates, so climatic forcing at ca. 10 Ma is not likely to be the cause of the exhumation rate increase indicated by our data. Given the presence of synchronous growth strata and evidence for a lack of climate forcing, we attribute the increase in exhumation rate at ca. 10 Ma to fault initiation and the likely generation of steep topography of the northern-most escarpment of the modern Tibetan Plateau. Helium ages are 2–3 Myr younger at low elevations east of the vertical transect, which suggests that either faulting initiated a few million years later or higher erosion rates occur along strike assuming no warping of isotherms between the two sample collection sites.

Fault Slip Rate

Vertical separation of marker horizons across the fault, combined with the dip of the fault and age constraints, can be used to calculate long-term (Myr) vertical and horizontal fault slip rates. The base of the PRZ can be used to reconstruct the relative elevation of the ca. 10 Ma land surface (i.e., the land surface just prior to fault initiation with respect to the modern elevation of the PRZ) in the hanging wall (Fig. 3). This reconstruction provides a marker horizon in the hanging wall, which can be correlated across the fault in the footwall to the foreland basin stratigraphic horizon that represents the initial deposition at the onset of faulting identified from growth strata (Fig. 3) (i.e., Clark and Bilham, 2008). We discuss the relative offset of markers across the fault in terms of modern elevation for ease of discussion as we have no constraints on paleoelevation.

In order to calculate the closure isotherm depth, we first calculate the helium closure temperature based on an average eU (53 ppm) (Supplemental Table 2 [see footnote 2]) and an average cooling rate (4.5–9 °C/Myr) determined from the apparent exhumation rate of the interval of fast exhumation (0.3 mm/yr) (Fig. 2 inset) and a range of typical continental geothermal gradients (15–30 °C/km). Based on the radiation damage trapping model of Shuster et al., (2006), we determine a closure temperature of 58–62 °C. Using a surface temperature of 10 °C and a range of typical continental geothermal gradients (15–30 °C/km), we calculate an average closure isotherm depth of 2.6 ± 1 km.

The base of the fossil PRZ is located at 2700 m elevation (Fig. 2), and the ca. 10 Ma surface is reconstructed to 5.3 ± 1.0 km elevation based on the calculation of closure isotherm depth (Fig. 3). Growth strata formed at the base of the Niugetao Fm. (900 m depth beneath the surface or 600 m elevation) in the Hexi Corridor represent the initiation of deposition related to fault motion (Yang et al., 2007) and have been dated with magnetostratigraphy at ∼9 Ma (Fang et al., 2004). Fault throw of 4.7 km is determined from the separation between the base of the Niugetao Fm. and the reconstructed ca. 10 Ma land surface in the hanging wall, and we assume faulting has been continuous since the onset time and thus the calculated rates reflect Myr scale average of fault motion (Fig. 3). Using estimates of fault throw (4.7 ± 1.0 km), fault initiation (9–10 Ma), and a fault dip (30°; Yang et al. 2007) that is assumed to remain unchanged in time, we calculate Myr time scale vertical and horizontal fault slip rates of ∼0.5 mm/yr and ∼1 mm/yr respectively, and a horizontal offset of 8.2 ± 1.8 km. Uncertainty on the horizontal offset is likely to be underestimated because it includes only the uncertainty on the vertical offset and not the fault geometry. Changes in fault geometry at depth or through time would increase the uncertainty in the horizontal offset.

Fault slip rates averaged over the past ∼10 Myr are consistent with late Pleistocene–Holocene rates determined in the Yumu Shan (vertical rates of 0.4–1.9 mm/yr) (Tapponnier et al., 1990) and for the Zhangye Thrust (0.6–0.9 mm/yr and 0.4–1.1 mm/yr vertical and horizontal rates, respectively), a correlative thrust ∼100 km to the east (Hetzel et al. 2004). Ten million year rates are also broadly consistent with the average geodetic velocity (horizontal) (5.5 ± 1.5 mm/yr; Zhang et al., 2004) measured between the Qaidam and Alashan blocks assuming that deformation is distributed throughout the ∼270 km wide Qilian Shan plateau (Institute of Geology, 1993; Metivier et al., 1998; Yuan et al., 2004). Low slip rates that were maintained over the past 10 million years imply that total horizontal fault offsets in the North Qilian range are likely to be small (< 10 km).


We interpret slow cooling of the northern Qilian Shan during 106–10 Ma as an indication that faulting did not reach this portion of the range until relatively late in the orogen's history. Fault initiation at ca. 10 Ma in the northern Qilian Shan range appears to be synchronous along strike for at least ∼100 km (Bovet et al., 2009). Such a young age of faulting broadly conforms to the step-wise growth model of plateau formation (Tapponnier et al., 2001) only in the sense that northern Tibet represents the youngest or most recent stage of orogenisis. Unlike the step-wise model, we suggest that the late Miocene initiation of the northern Qilian Shan fault represents only a modest advancement of ∼100 km of the plateau perimeter based on deformation ages in the central and southern parts of the range. Motion on the northern range front occurs significantly later than initial faulting in the central Qilian Shan (>33 Ma) (Yin et al., 2002), southern Qilian Shan/N. Qaidam terrane (broadly speaking since Paleocene time) (Yin et al., 2008) though locally faults initiate at later times (e.g., Sun et al., 2005; Wang et al., 2004), broad deformation in northeastern Tibet between 55 and 52 Ma (Dupont-Nivet et al., 2004; Horton et al., 2004) and locally along the West Qinling fault at 45–50 Ma (Clark et al., 2010). Late Miocene initiation of the north Qilian fault also significantly postdates the initial Oligocene motion along the Altyn Tagh fault (Ritts et al., 2004).

Miocene growth of the northern Qilian Shan could be local and simply represent the latest outgrowth of faulting within the larger Qilian Shan/N. Qaidam region in the direction of plate convergence. As Bovet et al. (2009) note, Miocene growth of the northern Qilian Shan also follows a decrease in fault slip accumulation of the Altyn Tagh fault (post-early Miocene; Yue et al., 2004) and uplift of the Altun Shan and SE Tarim Basin since 15–16 Ma (Ritts et al., 2008). Therefore between early to middle Miocene time, the kinematics of the Altyn Tagh fault system in northern Tibet may have evolved from fast strike-slip motion to distributed uplift and reverse faulting (Ritts et al., 2008). Following this change from strike-slip to shortening, the northern plateau margin expands to the northern Qilian Shan by 10 Ma, which may represent the final stage of distributed, reverse faulting and range growth that follows in the wake of decreased slip along the Altyn Tagh fault.

Alternatively, or in concert with the aforementioned changes to the Altyn Tagh fault system, convective removal of an overthickened, gravitationally unstable mantle lithosphere beneath north-central Tibet with associated elevation gain of 1–2 km of the central plateau (Molnar et al., 1993) would augment the force per unit length that Tibet applies to its surroundings, perhaps sufficiently to displace northward the Qaidam Basin as an effectively rigid block. The thinner and less deformed crust of the Qaidam Basin may act as a rigid block capable of deforming the region to the east or north by rotation or northward propagation (Dupont-Nivet et al., 2002). Increased topographic gradients may cause compressional stresses that translate the Qaidam Basin northward as a secondary indenter that causes deformation of the Qilian Shan to step northward and faulting in northeastern Tibet to extend eastward to the Liupan Shan (Zheng et al., 2006).

Since 10 Ma, the northeastern perimeter of the Tibetan Plateau advanced for nearly the first time since collision began, but it is important to note that the magnitude of this change was small compared to the north-south extent of the plateau prior to 10 Ma. Faulting within the central and southern Qilian Shan and in northeastern Tibet begins near the time of Indo-Asian collision (Dupont-Nivet et al., 2004; Horton et al., 2004; Yin et al., 2008; Clark et al., 2010). In late Miocene time, the areal extent of faulting expands northward by ∼100 km from the central to the northern Qilian Shan and eastward by ∼200 km to the Liupan Shan (at ∼8 Ma; Zheng et al., 2006) (Fig. 1 inset). Such a modest expansion in northern Tibet can be correlated with the more dramatic development of the eastern plateau by lower crustal flow, where more than 2.25 × 107 km3 crustal volume has been added east of the main collision zone (Royden et al., 1997; Clark and Royden, 2000; Clark et al., 2005a) since mid-late Miocene time (Kirby et al., 2002; Clark et al., 2005b; Ouimet et al., 2010). This suggests that the late Miocene to recent period of plateau growth was mainly east, and not north as might be simply predicted from plate boundary stresses applied by the northward motion of India.

We thank Lindsey Hedges, Alison Duvall, Peter Molnar, and support by the National Science Foundation of China (40672134 and 40234040), and DFIGCEA0607122, and the US National Science Foundation, Continental Dynamics Program (EAR-0507431). Reviews by G. Dupont-Nivet, A. Carter, K. Huntington, and three anonymous reviewers improved the clarity of this manuscript.

1Supplemental Table 1. Excel file of sample location and summary age data. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00523.S1 or the full-text article on www.gsapubs.org to view Supplemental Table 1.
2Supplemental Table 2. Excel file of (U-Th)/He replicate analyses. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00523.S2 or the full-text article on www.gsapubs.org to view Supplemental Table 2.