Spatial and temporal pattern of erosion in the Three Rivers Region, southeastern Tibet

Earth and Planetary Science Letters 433 (2016) 10–20

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

Spatial and temporal pattern of erosion in the Three Rivers Region, southeastern Tibet Rong Yang a,∗ , Maria Giuditta Fellin a , Frédéric Herman b , Sean D. Willett a , Wei Wang c , Colin Maden a a b c

Department of Earth Sciences, ETH Zurich, Sonneggstrasse 5, 8092 Zurich, Switzerland Institute of Earth Surface Dynamics, Université de Lausanne, Geopolis – Bureau 3871, 1015 Lausanne, Switzerland School of Ocean and Earth Science, Tongji University, 200092 Shanghai, China

a r t i c l e

i n f o

Article history: Received 25 March 2015 Received in revised form 16 October 2015 Accepted 19 October 2015 Available online xxxx Editor: A. Yin Keywords: southeastern Tibet thermochronometry river profile inversion erosion rate indentation

a b s t r a c t Convergence and collision between India and Eurasia have produced the Tibetan Plateau, which stands 5 km high over a region of 3 million km2 . Its southeastern margin lies in the restraining bend between the Sichuan basin and the Eastern Himalayan Syntaxis. In this region three parallel rivers, the Salween, the Mekong and the Yangtze, carve gorges up to 3 km deep. Along the longitudinal profiles, large-scale knickzones, defined by very high steepness, correspond to the gorges of the Salween and Mekong. The Yangtze, instead, has a nearly linear profile upstream from its first big bend. New low-temperature thermochronometric data reveal a complex pattern of erosion in the Three Rivers Region. From the Salween in the west to the Yangtze in the east the magnitude and rate of erosion decrease. From southto-north erosion rates exhibit variable gradients in space and time. Along the Salween and the Mekong a northward increase of erosion rate is followed by a decrease with additional distance to the north. Variations of erosion rate in time are characterized by a deceleration along the Salween and a general deceleration with local acceleration along the Mekong. This pattern, together with river profile analysis, is best explained by active coupling between tectonics and river incision related to the indentation and northward migration of the corner of the Indian continent. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The convergence between India and Eurasia has produced the Tibetan Plateau, which stands 5 km high over a region of 3 million km2 , but also extensive mountainous regions over a broader region, including the Three Rivers Region (TRR) of southeast Asia (Fig. 1). Although the onset of continental collision at about ∼40–50 Ma is well established (e.g. DeCelles et al., 2014 and references therein), both the mechanisms and the timing of plateau uplift are the subject of considerable debate (Clark et al., 2005; Clark and Royden, 2000; England and Houseman, 1986; Royden et al., 1997; Tapponnier et al., 1982). Several studies suggest regional incision of the plateau margins since 15 to 10 Ma following a phase of surface uplift (e.g. Clark et al., 2005; Duvall et al., 2012; Ouimet et al., 2010; Wang et al., 2012). Contradicting this model are paleo-altimetry studies indicating that both central and southeast Tibet were already close to their mod-

*

Corresponding author. E-mail address: [email protected] (R. Yang).

http://dx.doi.org/10.1016/j.epsl.2015.10.032 0012-821X/© 2015 Elsevier B.V. All rights reserved.

ern elevation at 35 Ma (DeCelles et al., 2007; Hoke et al., 2014; Rowley and Currie, 2006) and recent studies demonstrating that the geomorphology of the southeast margin of the plateau is also consistent with widespread shortening and river reorganization (Yang et al., 2015). In addition, the time of incision roughly coincides with continental scale climate change at about 8 Ma (Molnar, 2005), so it is difficult to assign a tectonic cause to changes in morphology or incision rate. As a result, there are numerous, disparate models for plateau formation and outward growth of its margins (Royden et al., 1997; Tapponnier et al., 1982; Houseman and Molnar, 2001). A key area for many of these models is the southeastern margin of the Tibetan plateau, in particular, the TRR (Fig. 1a), where three large rivers, the Salween, Mekong, and Yangtze (Chinese names: Nujiang, Lancangjiang, and Changjiang, respectively), originate from the high Tibetan Plateau, run closely parallel to the topographic gradient on the plateau margin and create bedrock gorges up to 3 km deep (Fig. 1b–d). The region is close to the collision boundary in the high-strain zone between the Eastern Himalayan Syntaxis and the Sichuan Basin. As the transition between the plateau and the surrounding lowlands, the timing and pattern of uplift for this region can help differentiate between models.

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

11

Fig. 1. Regional topography and cross sections. (a) Overview map of the regional topography in the Three Rivers Region (TRR) and locations of the Salween, Mekong, and Yangtze rivers. Locations of topographic cross transects A–A , B–B , and C–C are indicated by solid black lines. Inset shows location of the study area within the Himalaya–Tibet system. (b–d) Topographic cross sections of transects A–A , B–B , and C–C , showing the plateau morphology deeply-incised by major rivers.

For example, a gradual outward expansion of the plateau by lower crustal flow would exhibit an outward propagation of topography and thus the focus of erosion. In contrast, uniform uplift would lead to progressive headward incision from the orogen margin to its interior, as transient knickpoints propagate upstream. A number of thermochronometric and geomorphologic studies have concentrated on this issue (Clark et al., 2006, 2005; Duvall et al., 2012; Ouimet et al., 2010; Wilson and Fowler, 2011) and have demonstrated that there has been substantial erosion in this region, but data remain sparse. This study adds new thermochronometric data and their interpretation. 2. Geologic setting The TRR is a complex region of small lithospheric fragments separated by Late Paleozoic–Mesozoic suture zones (Fig. 2). Their reactivation by strike-slip shear has accommodated some of the Cenozoic deformation. Because of the high-degree of strain localization, Cenozoic shortening, metamorphism and exhumation of rocks from middle crustal depths seem mostly limited to

shear zones (Tapponnier et al., 1982; Leloup et al., 2001; Wang and Burchfiel, 1997). Localization occurred in the west along the Gaoligong (Zhang et al., 2012; Lin et al., 2009; Wang et al., 2006) and Chong Shan shear zones (Akciz et al., 2008), and in the east along the northern segments of the Ailao Shan–Red River shear system, specifically the Xuelong Shan and Diancang Shan zones (Gilley et al., 2003; Leloup et al., 2001 and references therein) (Fig. 2). Until ∼26 to 22 Ma transpressional shear along these zones was associated with high-temperature metamorphism and granite emplacement. After ∼22 Ma strike-slip deformation continued during cooling to greenschist facies conditions. Diachronous cooling and exhumation to temperatures below 350 ◦ C occurred earlier in the east, before 15 Ma (Leloup et al., 1993, 2001), and later in the west, continuing until about 10 Ma (Akciz et al., 2008; Zhang et al., 2012). In the west, there was more regional Miocene (∼16 Ma) exhumation of metamorphic rocks (Zhang et al., 2012). After 10 Ma the development of meso- to regional-scale highangle transtensive faults overprinted the shear zones with brittle deformation. At 4.7 Ma, the movement along the Diancang Shan reversed to right-lateral motion (Leloup et al., 1993). At present,

12

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

Fig. 2. Sample location and schematic geologic map of the TRR. The three rivers are highlighted in white. White filled circles are samples from this study. Grey filled circles are samples from Ouimet et al. (2010) and Wilson and Fowler (2011). Inset shows major active structures of the Himalayan–Tibetan Plateau. The black box on the inset shows the location of the TRR. TP: Tibetan Plateau. EHS: Eastern Himalayan Syntaxis. SCB: Sichuan Basin. MFF: Main Frontal Fault. ATF: Altyn Tagh Fault. KLF: Kunlun Fault. XSH-XJF: Xianshuihe-Xiaojiang Fault. LMSF: Longmen Shan Fault. ALS-RRF: Ailaoshan-Red River Fault. SGF: Sagaing Fault. The suture zones are indicated by white lines in the inset. Both the schematic geologic map and the inset were modified after: Akciz et al. (2008), Kirby (2003), Leloup et al. (1995), MGMR (1986), Wang and Burchfiel (1997), Yin and Harrison (2000), Zhang et al. (2010).

active faults in our study area include the Xuelong Shan (Leloup et al., 2001) and the strike-slip Zhongdian fault (near the town Deqin, Fig. 2), which has triggered large earthquakes (i.e. magnitude 5.9 in 2013, Chang, 2015). Existing low-temperature thermochronometric data are limited to the north of the ductile shear zones described above (Fig. 2) and reflect exhumation from shallow depths (2–3 km) starting in the Early–Middle Miocene and continuing till the present (Ouimet et al., 2010; Wilson and Fowler, 2011). The age pattern is related to elevation with pre-Miocene ages at high elevations and the youngest ages (3–4 Ma) at low elevations along the Mekong between 29◦ N and 30◦ N. Although brittle structures are dominated by strike-slip faults, crustal thickening must be occurring at depth to give the high elevations observed. Clark and Royden (2000) have suggested that

this occurs at lower crustal depths by viscous flow from under the plateau. Significant active crustal thickening in southeast Tibet appears to be restricted to the east of the TRR (Wang and Burchfiel, 2000), while in our study area Eocene-to-Miocene shortening related to folding and local thrusting has so far been described only in the region south of Xuelong Shan shear zone (Wang and Burchfiel, 1997). 3. River profile analysis The evolution of topography is coupled to changes in the river channel network, which is part of a dynamic system adjusting towards a balance between tectonic uplift and erosion. As a result, morphometric analysis of the river channel is a useful tool to assess the transient state of a landscape. In particular, the chan-

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

13

Fig. 3. River profile for the (a) Salween, (b) Mekong, and (c) Yangtze with maximum topography (grey line), annual rainfall (blue line), and channel steepness (small circles). River profiles (see river location in Fig. 2 inset) and maximum topography profiles were extracted from the SRTM3 DEM with filling artifacts removed and are shown by solid black lines and solid grey lines, respectively. Annual rainfall profiles were obtained from the TRMM data and are shown by solid blue curves. Channel steepness indices were calculated according to Willett et al. (2014) and are shown by small circles. The red segment on each river profile indicates river section that is the focus of thermochronometry study and is shown in Fig. 4. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

nel steepness, which is the slope of the channel normalized by drainage area, is predictive of channel response to uplift rate (e.g. Kirby, 2003; Ouimet et al., 2009). In this study, we analyze the channel steepness to identify where temporal or spatial changes in uplift rate may occur or have occurred, and, in the next section, combine these results with the thermochronometric data analysis. We extract the longitudinal profiles of the Salween, Mekong, and Yangtze (Fig. 3a–c). All three rivers have their headwaters at about 5000 m on the high Tibetan plateau. The Salween and Mekong have similar longitudinal profiles. They both have a convex section in their mid-reaches, with a low gradient below 1000 m, a steep gradient between 1000–3000 m and a low gradient above 3000 m. This is also reflected in their steepness index, which shows high values of >100–300 at the steep reaches. Furthermore, the transition in steepness index coincides with a change in maximum topography and precipitation, in particular, in the Salween. The maximum elevation increases by about 3 km over 1000 km in the upstream direction and is accompanied by a 3 to 4 fold decrease in precipitation rates. As the rivers reach the plateau interior, the peak elevation is almost uniform at 5000–5500 m and rainfall is below 500 mm/yr. In contrast, the Yangtze exhibits a different morphology, with a nearly linear longitudinal profile below 4000 m and local steep segments at ∼500 m, ∼1500 m, and ∼2500 m. Above 4000 m, the river flows on the plateau with generally low gradient. The channel steepness varies within the range of 200–100 below 4000 m but decreases dramatically at the plateau edge. Unlike the Salween and Mekong where the maximum elevation decreases significantly in the downstream direction, the Yangtze shows a gentle decrease in peak elevation. In addition, the Yangtze receives rainfall below 500 mm/yr in the upper reaches and overall less rainfall with respect to the Mekong and Salween. 4. Low-temperature thermochronometry 4.1. New thermochronometric ages Low-temperature thermochronometry provides constraints on the time-temperature history of rock cooling below ∼300 ◦ C. Given a temperature field near the Earth’s surface, a suite of thermochronometric ages can be converted into an erosion rate history over the timescale provided by the measured ages. Here we combine three dating systems, apatite (U–Th–Sm)/He (AHe), zircon (U– Th)/He (ZHe) and apatite fission-track (AFT), providing constraints on the erosion history of the uppermost ∼6 km of the crust. It is also worth noting that incision of less than ∼6 km (dependent on the geothermal gradient) will only bring rocks to the surface that were residing above the ZHe partial retention zone (PRZ). In this case, the ZHe age mostly reflects the pre-incision history and

provides only a minimum erosion rate estimate. Likewise, for the AHe system, incision of less than 2 km only exposes pre-incision ages. The TRR is appropriate for our purpose as thermochronometric studies have shown there has been enough erosion since the Miocene to exhume rocks from below the PRZ (Ouimet et al., 2010). Bedrock samples were collected along the trunk valleys of the three rivers but also from tributaries in some places (Fig. 2, Supplementary data Table 1). For the Salween and Mekong we collected samples along the steepest section of the channels (Fig. 3a, b). Along the Yangtze, samples came from north of the first big bend where the river is entrenched in a deep gorge (Fig. 3c). We present here our new data, organized by river basin. In the Salween, we obtained six mean AHe ages, seven AFT ages, and twelve mean ZHe ages (Fig. 4a, Supplementary data Tables 2–5). All samples but one were collected at elevations close to the valley bottom, within 500 m from the river. All three thermochronometers yield ages younger than 10 Ma and also younger than the Ar/Ar ages along the Gaoligong shear zone that range between 32 and 10 Ma (Zhang et al., 2012; Lin et al., 2009; Wang et al., 2006). AHe ages are between 2.5 and 5.8 Ma. AFT ages are between 3.4 and 9.8 Ma. ZHe ages are between 3.4 and 7.9 Ma. All three thermochronometers exhibit a clear younging trend in the upstream direction with the youngest age of 2.5 Ma (AHe) at 27.5◦ N, followed by a reversal in trend with older ages between 3.4 Ma (AHe) and 9.8 Ma (AFT) to the north of 27.5◦ N. Four Salween samples (NJ1, NJ2, NJ3, NJ5, Supplementary data Table 5) yield indistinguishable AFT and ZHe ages and one sample (NJ21) yields an AFT age older than its mean ZHe age. The age overlap indicates rapid cooling through a ∼180–120 ◦ C temperature window and could also suggest complications affecting either zircons or apatites and their ages. These factors could include U–Th enriched rims or high radiation damage in zircon or very low spontaneous track-density in apatite. Rims enriched in U and Th can produce anomalously young ZHe ages (Hourigan et al., 2005) but we cannot control for zonation of parent nuclides because we used the total dissolution technique. Increasing radiation damage can cause a decrease in helium diffusivity by trapping helium atoms (Shuster et al., 2006). However, radiation damage over a certain threshold can also cause an increase in helium diffusivity by interconnecting damage zones (Guenthner et al., 2013; Nasdala et al., 2004). A first order estimate of this effect is given by the effective uranium concentration (eU) of zircons (Flowers et al., 2007) that in most samples varies over a low-to-medium range between 331 and 4353 ppm (Supplementary data Table 3). In our dataset eU values >5000 ppm characterize two samples only, NJ8 and NJ12, and if this corresponded to high radiation damage, it could result in minimum effective closure temperatures as low as 107 ◦ C (Supplementary material Table 3). Low spontaneous track-

14

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

Fig. 4. River profile plotted with sample locations and thermochronologic ages for the (a) Salween, (b) Mekong, and (c) Yangtze. River profile was extracted from the SRTM3 DEM with filling artifacts removed and is shown by solid black lines in the upper panel. Squares in the upper panel indicate sample locations of this study (referred to the white circles in Fig. 2) with samples from Ouimet et al. (2010) and Wilson and Fowler (2011) (grey circles in Fig. 2) along the river course. The lower panel shows thermochronometric ages for samples in the upper panel. Each thermochronometric system is represented by a different symbol. Ages in the lower panel and sample locations in the upper panel are aligned by latitudes. Mean ages are reported with 1 − σ error. Fill colors in all symbols indicate sampling distance above the thalweg. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

density results in very high uncertainty on the age, as is the case for samples NJ1, NJ5 and NJ21. Sample NJ1 has zero tracks in some counted areas and a total number of countable spontaneous tracks of 35 resulting in a 20% uncertainty on the age (Supplementary data Table 4). Sample NJ5 has a total of 59 spontaneous tracks giving an uncertainty of 14% (Supplementary data Table 4). Sample NJ21 has only 20 countable spontaneous tracks resulting in a high χ 2 probability (94%) and a minimum age uncertainty of 22% (Supplementary data Table 4). Therefore the reversal between the AFT (5.7 Ma) and ZHe (3.4 Ma) ages of this sample NJ21 is likely associated with very low spontaneous track density producing a spurious AFT age. Our new data from the Mekong include eight mean AHe ages, four AFT ages, and ten mean ZHe ages. Our ages scatter over a wide range depending on elevation and/or latitude (Fig. 4b, Supplementary data Tables 2–5). South of 28◦ N, four samples come from elevations within 500 m from the river-bed but one of these is from a tributary. One sample comes from a tributary valley at high elevation (>1000 m above the main trunk thalweg). North of 28◦ N, out of seven samples only two samples (MK04 and MK09) come from elevations close to the river (within 500 m from the river bed) and two (MK12 and MK13) are from tributaries. AHe ages range from 1.3 to 6.3 Ma but one AHe age is 81.8 Ma (MK13): this sample is located at ∼4000 m along a tributary. AFT ages range from 5.2 to 13.5 Ma and they are consistent with ages reported by Wilson and Fowler (2011) at similar locations (Fig. 2, supplementary data Table 6). ZHe ages range from 5.5 to 156 Ma. Most zircons from the Mekong contain modest eU (<1000 pm) with 3 exceptions (Supplement data Table 3) indicating only a minor influence of radiation damage on ZHe ages. Two samples (LCJ6 and LCJ9) yield indistinguishable AFT and ZHe ages. The AFT ages of these two samples have large uncertainties due to low spontaneous track-density as zero tracks were counted in some areas. Sample LCJ6 was collected near the Xuelong Shan shear zone and all the three thermochronometric ages of this sample are younger than the Ar/Ar ages from this zone (Leloup et al., 2001). The Mekong AHe ages exhibit two spatial patterns. First, the ages near the valley bottom become younger in the upstream direction: from 5.6 Ma to 1.3 Ma. Second, the ages along the valley walls vary from those near the valley bottom. For instance, at ∼28.5◦ N they range from 1.3 Ma at 2000 m to 5.8 Ma above 3000 m. Our AFT data combined with those reported by Wilson and Fowler (2011) exhibit similar patterns, although variations of age with elevation are much larger from 3 to 59.4 Ma. In contrast to AHe and AFT data,

ZHe ages close to the valley bottom of the Mekong reveal older ages towards the upper reaches: from 5.5 Ma to 15.5 Ma. Five ZHe ages at high elevations (>1 km above the river bed) yield ages older than 80 Ma indicating the preservation of the PRZ of the ZHe system along the Mekong. Three samples were analyzed from the Yangtze. All three samples are from tributaries and they yield old AHe ages, ranging from 29 to 38.4 Ma. One sample has an AFT age of 53.5 Ma and one sample has a ZHe age of 130.6 Ma (Fig. 4c, Supplementary data Tables 2–5). This last sample is characterized by a modest eU of about 2000 ppm (Supplementary data Table 3). Previous studies along the Yangtze upstream from our sample locations (Ouimet et al., 2010; Wilson and Fowler, 2011) (Fig. 2, Supplementary data Table 6) presented AHe ages that were younger than 9 Ma, ZHe ages older than 20 Ma and AFT ages from 5.9 to 23 Ma (Fig. 4c). 4.2. Quantitative Interpretation of thermochronologic data 4.2.1. Converting thermochronometric age to mean erosion rate To convert thermochronometric ages into erosion rates, we first used an analytical method (Willett and Brandon, 2013). This method makes use of a solution based on the assumption of a constant rate of cooling, such that observed ages represent the time since samples passed the closure temperature Dodson (1973) and therefore the time they took to travel from the closure depth to the modern elevation. This travel time defines the time-averaged erosion rate for any given thermochronometric sample. The kinetic parameters for helium diffusion in apatite and zircon are from Farley (2000) and Reiners et al. (2004), respectively, and for fission track annealing in apatite from Ketcham et al. (1999). Given that, with erosion, rocks advect heat upward and thus disturb the geotherm, this approach includes the transient effect of erosion on the closure isotherms such that the geotherm is a function of erosion rate and time. The closure depth is determined by the intersection of the geotherm with its corresponding closure temperature. This approach assumes steady-state topography and the thermal perturbation due to topography is wave-length-dependent and approximated through the calculation of an appropriate mean elevation (Willett and Brandon, 2013). For each thermochronometer the effect of surface topography on an isotherm is largest for wavelengths longer than 2π zc (Braun, 2002), where zc is the depth of the closure isotherm. Thus an averaging circle with a radius of half this length provides a reasonable estimate for averaging the topography and is used to calculate the mean elevation. The temperature history for samples that are located below or above the

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

15

Fig. 5. Time-averaged erosion rates for the (a) Salween, (b) Mekong, and (c) Yangtze. Erosion rates are estimated with a range of thermal parameters (see 2.5.1). The envelopes show the range of erosion rates for each thermochronometric age, where the range reflects a range for the modern geothermal gradient, time of initiation of erosion and radius of the averaging window for locally mean elevation. Grey triangles in (b) are AFT ages from Wilson and Fowler (2011), which are not considered in constructing the envelope. Squares are millennial erosion rates derived from 10 Be (Henck et al., 2011).

mean elevation is corrected by a vertical extension of the cooling path. To extract erosion rate from a thermochronometric age, estimates of the modern geothermal gradient, time of initiation of erosion, and surface temperature are required. In this study, we chose a range of geothermal gradients, times for onset of erosion, and radius of the averaging window in order to test the sensitivity of the results. In our study area there are few heat flux data to constrain the modern geothermal gradient. Along the eastern margin of Tibet, a geothermal gradient between 30–35 ◦ C/km fits best the observed thermochronometric data (Clark et al., 2005). Given the very young ages (<2 Ma) observed in the TRR, we chose bounds to the geothermal gradient of 30 and 40 ◦ C/km. We assumed a range of initiation of erosion from 20 Ma to 30 Ma and we only calculated erosion rates for ages that are younger than 20 Ma. Using such a range is intended to record the changes in erosion rates associated with the evolution of the eastern Tibetan plateau margin since the Miocene, when uplift of this region is likely to have occurred (Clark et al., 2005; Ouimet et al., 2010) and when the major shear zones of the TRR cooled below 350 ◦ C. The surface temperature is set at 15 ◦ C. The radius of the elevation-averaging circle is 5 km for the AHe, 10 km for the AFT, and 15 km for the ZHe. We also used a radius of 5 km for all the three thermochronometers in order to check sensitivity to this parameter. For AHe and ZHe ages, we also tested the sensitivity of the models to grain size. No significant differences were observed and therefore in the following sections we only discuss results that were modeled with the kinematic parameters mentioned above. As shown in Fig. 5, we obtain consistent relative values of erosion rate, but the model parameters affect the magnitude of these rates. Among them, the assumed modern geothermal gradient plays the biggest role in the estimated magnitude of erosion rates. In Fig. 5, the upper and lower limits of each colored envelope correspond to geothermal gradients of 30 and 40 ◦ C/km, respectively. These colored ranges do not represent independent

uncertainty ranges for each thermochronometer: thus, they do not imply that proximate samples have experienced widely different geothermal gradients. Variations in time of initiation of erosion and radius of averaging window have small effects on the estimated erosion rate. Uncertainties are correlated between ages, so that the spatial and temporal patterns remain robust unless the geothermal gradient changes dramatically along the course of the river. Erosion rates derived from the ZHe data on the Salween (Fig. 5a) for a geothermal gradient of 40 ◦ C/km range from 0.5 to 1.4 mm/yr (lower limit of color envelope) whereas at a geothermal gradient of 30 ◦ C/km they range from 0.6 to 3.2 mm/yr (upper limit of color envelope). However, the general erosion pattern always shows a 2.5-fold increase of the erosion rate from 26◦ N to 27.7◦ N followed by a decrease of the same magnitude towards 28◦ N. Samples NJ8 (at ∼27◦ N) and NJ12 (at ∼27.4◦ N) with high radiation damage were included, although erosion rates at these two locations are likely to be overestimated. The other two thermochronometers show a similar spatial pattern with northward increases in erosion rate followed by a decrease, but the magnitude of the erosion rate is everywhere lower than that inferred from the ZHe ages: AFT-derived erosion rates range between 0.15 and 1 mm/yr and AHe-derived erosion rates range from <0.1 to 0.4 mm/yr. The differences in erosion rates estimated from the three thermochronometric ages of the same sample give information on the temporal changes in erosion, reflecting the different time spans over which erosion rate is averaged. Thus, all the samples from the Salween exhibit a decrease of erosion rate with time. Depending on the geothermal gradient, erosion rates estimated from both the ZHe and AFT data of the Mekong (Fig. 5b) show a spatial pattern characterized by a decreasing trend towards the north. The magnitude of this decrease is from ∼1.2–0.7 mm/yr to ∼0.4–0.2 mm/yr for the ZHe ages and from 0.6–0.4 mm/yr to ∼0.4–0.2 mm/yr for the AFT ages. Two samples from the main stem show a slightly increase in the ZHe erosion rate south of 28◦ N. The spatial pattern of erosion rates for the AHe system is constrained by 2 samples to the south of 28◦ N and 4 samples to

16

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

Fig. 6. Erosion pattern of the TRR determined by age inversion using method of Herman and Brandon (2015). The white circles are sample locations as in Fig. 2.

the north of 28◦ N: to the south they show rates between <0.2 and 0.6 mm/yr, whereas to the north they range between 0.2 and >1 mm/yr. Comparison of the three erosion rates from the Mekong reveals that samples south of 28◦ N experienced a deceleration of erosion from >0.6 mm/yr to 0.2–0.6 mm/yr. Samples north of 28◦ N show locally erosion rates derived from the AHe system higher than those derived from ZHe and AFT. Unlike the Salween and the Mekong, we could not find a solution for the Yangtze when the geothermal gradient exceeds 37 ◦ C/km, which is in fact consistent with its lower erosion rate relative to the Salween and Mekong (Fig. 5c). Due to the limited number of samples, the erosion rates are only resolved at ∼29.1◦ N and ∼29.5◦ N with rates <0.5 mm/yr. 4.2.2. Inference of erosion rates in space and time In order to further resolve erosion rates in both space and time, we applied a modified version of the inversion method of Fox et al. (2014). This approach is based on the fact that the depth to the closure isotherm is the integral of erosion rates from the cooling age to the present day. The heat transfer equation with both heat conduction and advection is solved with a one-dimensional thermal model, accounting for the effects of topography on the shape of the isotherms. The same kinetic parameters as the previous inversion method are used to calculate the closure temperature for each thermochronometer. The time dimension is discretized over a finite number of time intervals and erosion rate is determined for each of these intervals. The erosion rate is calculated by initializing from an a priori erosion rate that is iteratively updated to a posterior erosion rate that maximizes the fit to the data using the nonlinear least squares method (e.g. Tarantola, 2005). This approach imposes a condition that the erosion rate is treated as a spatially correlated random variable, with a given mean and covariance. As a result, although each datum is treated independently, solutions are linked by requiring that the erosion-rate functions are spatially correlated and thereby vary smoothly in space. Under these conditions, we expect that the influence of the two samples with high radiation damage (NJ8 and NJ12) on the regional erosion pattern will be small. (We used a modified version of this algorithm that imposes a positivity constraint through a logarithmic transformation and causes the problem to become non-linear Herman and Brandon, 2015.) The inferred erosion rates in space and time are shown in Fig. 6. We used an a priori erosion rate of 0.4 mm/yr with an a priori variance of 3.5 and a spatial correlation length of 20 km. We chose an initial unperturbed geothermal gradient of about 30 ◦ C/km. The

Fig. 7. Residuals for the inversion. The nonlinear inverse problem is solved using the steepest-gradient method and convergence is typically attained after 20 iterations. a) Plot of relative frequency against residuals of closure depth for an iteration number of 20. b) Plot of measured age against predicted age.

a priori erosion rate and geothermal gradient were chosen based on estimates for south–eastern Tibet as discussed above. Both a priori variance and spatial correlation length act as trade-off parameters between solution resolution in space and the need to average out noise. Erosion rates were resolved over time intervals of 2 Ma. Erosion rates are represented by the a priori value when there is no age constraint. Fig. 7a shows the residual distribution following 20 iterations. It appears that the residual is symmetrically distributed between −0.8 and 0.8. In Fig. 7b, we show the misfit between measured ages and predicted ages for the best-fit model. It appears that our model can fit most of the AHe ages but it slightly overestimates the ZHe and AFT ages. Before 10 Ma, erosion rates are only resolved along the Yangtze and upper Mekong. Both rivers are characterized by erosion rates less than 0.25 mm/yr, with the lowest erosion rates in the range of 0.01 mm/yr along the Yangtze south of 28◦ N. Rates are higher (∼0.25 mm/yr) in the north. It seems that this erosion pattern on the Yangtze has been sustained for the last 10 Ma but it is also possible that changes in erosion rate are too small to be resolved (see further discussion in supplement of Herman et al., 2013). From 10 to 6 Ma, a significant change occurs in the downstream reach of the Salween. In this region between 10 and 8 Ma, an erosion rate of ∼0.35 mm/yr is resolved at 26◦ N, and between 8 and 6 Ma an erosion rate of ∼0.5 mm/yr is resolved around 27◦ N. From 6 to 4 Ma, along the Salween there is a significant increase in erosion with rates more than doubled. In particular, the erosion rates at 27◦ N peak at >2.5 mm/yr. During the same time interval, along the Mekong, erosion rates of up to >0.5 mm/yr are resolved south of 28◦ N and slow erosion rates persist to the north.

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

Fig. 8. Inversion sensitivity tests at location of sample NJ2. a) The black curve indicates results using parameters as in Fig. 6, whereas the blue line indicates a different prior erosion rate as illustrated on the figure. b) The black curve indicates the same parameters used as Fig. 6 while the blue line indicates result using only two thermochronometric systems used as indicated on the figure. c) The black curve indicates result using the parameters as in Fig. 6 whereas the blue line indicates a different initial geothermal gradient as indicated on the figure. d) The black curve shows result using the parameters as in Fig. 6 whereas the blue line indicates a different spatial correlation length as indicated on the figure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Between 4 to 2 Ma, along the Salween the locus of rapid erosion migrates from 27◦ N–26◦ N to 27.5◦ N–28◦ N and increases to ∼0.6 mm/yr while south of 27◦ N erosion rates drop from >0.9 mm/yr to ∼0.25 mm/yr. At the same time, erosion rates south of 28◦ N along the Mekong slow to ∼0.25 mm/yr. Finally in the last 2 Ma interval, the locus of rapid erosion along the Mekong migrates further upstream to 29◦ N. During this interval, erosion rate is higher than 0.75 mm/yr near 28◦ N on both the Salween and Mekong. The erosion history we obtained from this inversion method may depend on the choice of the a priori erosion rate, the initial geothermal gradient, and the spatial correlation length. Fig. 8 shows an example of the erosion history at the location of sample NJ2 using different inversion parameters. It appears that the magnitude of erosion rate is affected by the choice of the a priori erosion rate (Fig. 8a) and the initial geothermal gradient (Fig. 8c) but the signal of increase/decrease in erosion rate is robust as long as the change in erosion rate is large enough to be detected (see further discussion in supplement of Herman et al., 2013). Erosion rates are weakly dependent on the spatial correlation length in this case (Fig. 8d). Given that some of our ZHe ages and AFT ages are very close, especially on the Salween, and some of the AFT ages might have large uncertainties due to zero track counts, we also inverted for erosion rates using only the ZHe and AHe ages. This results in a systematically different magnitude of erosion rate, but a similar erosion pattern (Fig. 8b). 5. Discussion Two inverse modeling methods were used to derive erosion rates from ages. The first solution provides time-averaged erosion rates while the second one resolves erosion rate over specified time intervals, thus enabling the comparison of erosion in space during the same time period as well as the evaluation of erosion rate in time at the same location. The two methods described above (time-averaged and transient) yield consistent results indicating: (i) a deceleration of erosion with time along the Salween, (ii) a deceleration of erosion south of 28◦ N but an acceleration

17

north of 28◦ N along the Mekong, (iii) low erosion rates south of 29◦ N and (iv) no resolvable temporal variation along the Yangtze. Both methods also indicate a decrease in erosion rates over a 100 km distance from west to east: i.e. from ∼3 mm/yr along the Salween to <0.5 mm/yr along the Yangtze. The west–east erosion rate gradient corresponds also to a west– east gradient in the magnitude of erosion. In fact, along the Salween not only all our thermochronometric ages but also the Ar/Ar ages on muscovite and biotite along the Gaoligong and Chong Shan shear zones are reset by Neogene exhumation (Zhang et al., 2012; Lin et al., 2009; Wang et al., 2006). In contrast, along the Mekong, ZHe ages are older than 80 Ma indicating the preservation of a PRZ of the ZHe system. Finally, along the Yangtze south of 29◦ N all ages are ≥29 Ma. Thus in the last 22 Ma, along the Salween, rocks were exhumed from more than 10 km deep, along the Mekong they were exhumed from about 6 km deep and along the Yangtze from less than 2 km. Our long-term (Ma-scale) west-to-east gradient in erosion rate is similar to the short-term (millennial) spatial gradient observed by Henck et al. (2011) using detrital cosmogenic nuclide dating (Fig. 5). From west to east the mean annual rainfall also decreases at 28◦ N from 2000 mm/yr to less than 1000 mm/yr (Fig. 3a–c), the valley depth of the three rivers decreases from almost 3 km to about 2 km (Fig. 1b–d). In contrast, the rock erodibility likely increases towards the west reflecting the exposure of high-grade gneisses and granites along the Salween and low-grade metasediments and volcanics along the Mekong and Yangtze. Henck et al. (2011) concluded that tectonics is the primary control and the west–east gradient mirrors a gradient in rock uplift rates, and this is supported by our data. Our results also show south-to-north variations of erosion rates. The south-to-north patterns are not easily comparable between rivers because of the different latitudes and elevations of the sample locations. Nevertheless, the Salween and the Mekong both show a northward increase in erosion rate followed by a decrease in erosion rate between 26◦ N and 30◦ N. This region is where the two rivers are most closely spaced (<30 km in distance), where they reach the highest steepness and where the upstream decay in steepness index coincides with the plateau edge marked by an increase in maximum elevation and an abrupt decrease in precipitation rate. This is also the region where it could be argued that the northeast India plate corner indents the furthest, and low-relief and high-elevation landscapes are absent (Clark et al., 2006). At these latitudes the Yangtze falls off this pattern, as the ages are too old to resolve any potential change. It could also be that the river is located too far to the east (>70 km east of the Mekong) to be impacted by deformation. Along the Salween and the Mekong, modern rainfall, long-term and short-term erosion rates and river steepness have different patterns: immediately south of the edgeplateau they all increase but further to the south rainfall increases steadily while erosion rates and steepness decrease (Figs. 3 and 4). Thus, if during the Neogene the climatic gradient was similar to the modern one, the south-to-north variations in erosion rates are unlikely related to coupling between rock uplift and precipitation. Taken together, the south-to-north and west-to-east gradients are better explained as a result of a non-uniform deformation and uplift field associated with the indenting of India into Eurasia, with deformation decaying towards the south and the east with the distance from the collisional front. Such a deformation field seems to be consistent with models of the indentation of a strong India plate into a weaker Asian plate (England and Houseman, 1986; England and McKenzie, 1982) such that strain is concentrated in the vicinity of the indenter and decays towards the distant regions. A second observation of this study is that both the Salween and Mekong are characterized by a general deceleration of erosion rate with local acceleration along the Mekong accompanying the

18

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

Fig. 9. Erosion history extracted from Fig. 6. Sample locations are indicated by small circles and formatted as in Fig. 2. The red curve indicates the erosion history. The blue curve indicates the variance reduction indicating how well the posterior erosion rate is constrained with respect to the a priori erosion rate and its variance. Smaller variance indicates that the data is able to constrain better the erosion rate. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

northward migrating high erosion rate as revealed by our transient model (Fig. 6). Along the Salween the locus of rapid incision was focused at latitudes between 27 and 26◦ N from 6 to 4 Ma constrained by sample NJ2, NJ4 and NJ10. After 4 Ma it shifted northwards to the region containing samples NJ14 and NJ21 (Fig. 9). However, the low erosion rate before 4 Ma derived from sample NJ21 is constrained locally by only one AFT age (Fig. 4a) indicating that our transient model is strongly dependent on that sample. Along the Mekong, data to the north of 28◦ N indicate a northward shift of the locus of high erosion rate during the last 2 Ma (Fig. 9). The northward migration of high erosion rate towards the plateau interior is inconsistent with the plateau expansion scenario, which predicts an outward migration of an erosional wave. Alternatively, the northward migration could represent a transient incision event associated with recent, regional rock uplift. Regional uplift would trigger a pulse of incision that would propagate upstream as a channel knickpoint. In this model, the knickpoint migrates as a kinematic wave with a predictable speed that is a function of rock erodibility and water discharge (Whipple and Tucker, 1999). We calculated the speed of incision migration along the Salween based on the estimate of rock erodibility according to the method by Goren et al. (2014). In this approach, erosion rate is calculated according to the stream power law and rock erodibility is then calibrated with external information of total exhumation such as exhumation constrained by low-temperature thermochronomet-

Fig. 10. Knickpoint migration velocity along the Salween. Knickpoint migration velocity (C ) was calculated according to the equation: C = K Q m S n−1 in which K is rock erodibility, Q is water discharge, S is channel slope, m and n are constants. We chose an m of 0.45 and n of 1 for the TRR. We used the calibration method of Goren et al. (2014) for the rock erodibility. Along the Salween, the rock erodibility was calibrated from sample NJ2 where a total amount of ∼2.5 km erosion since ∼8 Ma was estimated from its AFT age by assuming a geothermal gradient of 30 ◦ C and a surface temperature of 15 ◦ C which yields a rock erodibility of 2.19 × 10−6 m0.45 /yr. We assumed spatially constant rock erodibility for the knickpoint migration velocity calculation.

ric data. We obtain a rock erodibility of 2.19 × 10−6 m0.45 /yr, which is comparable to estimates for resistant rocks in other studies (Stock and Montgomery, 1999), and a migration rate of the knickpoint between 26◦ N and 30◦ N of ∼180–205 km/Ma (Fig. 10). For comparison, we calculated the migration rate of our high erosion rate locus. From 6 Ma to 2 Ma, the high erosion locus moved

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

from NJ2 to NJ21 over a distance of ∼160 km (Fig. 9), resulting in a migration velocity of ∼40 km/Ma. This velocity is at least 4 times smaller than that calculated from the kinematic wave model. Although variations associated with lithology and drainage area changes contribute to the uncertainties in estimation of the kinematic wave speed, we argue that the differences between these two rates are too large for the thermochronometers to be reflecting a river transient. Taken together, the high rates of erosion near the corner of the indenting Indian plate support the idea that this process dominates the uplift, and hence, the erosion pattern. The northward propagation of high erosion rates could reflect the northward advance of the India indenter corner, similarly to what described for the Eastern Himalayan Syntaxis (Seward and Burg, 2008), where the deformation associated with the syntaxis expanded northwards. In southeast Tibet, the northward migration of the indenter triggered successive deformation along strike-slip faults (Wang and Burchfiel, 1997; Zhang et al., 2015) and partitioning between shortening and strike-slip shear (Zhang et al., 2004). For instance, the Zhongdian fault (Fig. 2) absorbed displacement from one segment of the Red River fault and became active in the Pliocene (Replumaz and Tapponnier, 2003). Also, along the Salween the northward migration of high erosion rates between 10 and 4 Ma could be related to the transtensional activity that after 10 Ma reactivated the Gaoligong shear zone diachronically along strike. Thus, several factors associated with the fault system may have led to isolated and transient uplift associated with transpression or the development of individual splays. Unfortunately, the spatial distribution of our data is insufficient to reconstruct the activity of individual faults. Regional shortening or transpression would also lead to variability in erosion rates driven by fluvial processes of river reorganization (Yang et al., 2015). Shortening by plate indentation or river channel defeat by strike-slip faulting can lead to large-scale river capture, a process which has been proposed for this region (Clark et al., 2004; Yang et al., 2015). Large or progressive capture of large rivers could also lead to variations in erosion rate in space and time. 6. Conclusion Our results reveal both spatial and temporal variations of erosion rates in the TRR. Variation of erosion rate in space is characterized by an eastward decrease from the Salween to the Yangtze and a northward increase followed by a decrease further north along the Salween and Mekong. Variation of erosion rate in time is characterized by a deceleration along the Salween and a deceleration with local acceleration along the Mekong. Together with river longitudinal profile analysis, we infer that such erosion patterns in the TRR reflects an active coupling between tectonics and river incision related to the indentation and northward migration of the Indian plate corner where erosion rates are focused in the maximum strain areas. Acknowledgements We thank two anonymous reviewers and the editor An Yin for their constructive comments and suggestions on the present work. We thank Li Yuansen and Liu Weinan for field assistance. R.Y. is grateful to Prof. Zhou Zuyi who has initially made this cooperation possible. This work was partially supported by Chinese National Science Foundation (Program 4117218). F.H. was funded through SNF grant PP00P2_138956. Appendix A. Supplementary material Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2015.10.032.

19

References Akciz, S., Burchfiel, B.C., Crowley, J.L., Jiyun, Y., Liangzhong, C., 2008. Geometry, kinematics, and regional significance of the Chong Shan shear zone, eastern Himalayan Syntaxis, Yunnan, China. Geosphere 4, 292–314. http://dx.doi.org/ 10.1130/GES00111.1. Braun, J., 2002. Estimating exhumation rate and relief evolution by spectral analysis of age-elevation datasets. Terra Nova 14, 210–214. http://dx.doi.org/ 10.1046/j.1365-3121.2002.00409.x. Chang, Z.-F., 2015. Seismotectonic background of the 2013 benzilan m5.9 earthquake, Yunnan Province. Seismol. Geol. 37, 192–207 (in Chinese with English abstract). Clark, M.K., House, M.A., Royden, L.H., Whipple, K.X., Burchfiel, B.C., Zhang, X., Tang, W., 2005. Late Cenozoic uplift of southeastern Tibet. Geology 33, 525–528. http://dx.doi.org/10.1130/G21265.1. Clark, M.K., Royden, L.H., 2000. Topographic ooze: building the eastern margin of Tibet by lower crustal flow. Geology 28, 703–706. Clark, M.K., Royden, L.H., Whipple, K.X., Burchfiel, B.C., Zhang, X., Tang, W., 2006. Use of a regional, relict landscape to measure vertical deformation of the eastern Tibetan Plateau. J. Geophys. Res. 111, F03002. http://dx.doi.org/10.1029/ 2005JF000294. Clark, M.K., Schoenbohm, L.M., Royden, L.H., Whipple, K.X., Burchfiel, B.C., Zhang, X., Tang, W., Wang, E., Chen, L., 2004. Surface uplift, tectonics, and erosion of eastern Tibet from large-scale drainage patterns. Tectonics 23. http://dx.doi.org/ 10.1029/2002TC001402. DeCelles, P.G., Kapp, P., Gehrels, G.E., Ding, L., 2014. Paleocene–Eocene foreland basin evolution in the Himalaya of southern Tibet and Nepal: implications for the age of initial India–Asia collision. Tectonics 33. http://dx.doi.org/10.1002/ 2014TC003522. DeCelles, P.G., Quade, J., Kapp, P., Fan, M., Dettman, D.L., Ding, L., 2007. High and dry in central Tibet during the Late Oligocene. Earth Planet. Sci. Lett. 253, 389–401. http://dx.doi.org/10.1016/j.epsl.2006.11.001. Dodson, M.H., 1973. Closure temperature in cooling geochronological and petrological systems. Contrib. Mineral. Petrol. 40, 259–274. Duvall, A.R., Clark, M.K., Avdeev, B., Farley, K.A., Chen, Z., 2012. Widespread late Cenozoic increase in erosion rates across the interior of eastern Tibet constrained by detrital low-temperature thermochronometry. Tectonics 31, TC3014. http://dx.doi.org/10.1029/2011TC002969. England, P., Houseman, G., 1986. Finite strain calculations of continental deformation. 2. Comparison with the India–Asia collision zone. J. Geophys. Res. 91, 3664–3676. England, P., McKenzie, D., 1982. A thin viscous sheet model for continental deformation. Geophys. J. R. Astron. Soc. 70, 295–321. Farley, K., 2000. Helium diffusion from apatite: general behavior as illustrated by Durango fluorapatite. J. Geophys. Res., Solid Earth (1978–2012) 105, 2903–2914. Flowers, R.M., Shuster, D.L., Wernicke, B.P., Farley, K.A., 2007. Radiation damage control on apatite (U–Th)/He dates from the Grand Canyon region, Colorado Plateau. Geology 35, 447. http://dx.doi.org/10.1130/G23471A.1. Fox, M., Herman, F., Willett, S., May, D., 2014. A linear inversion method to infer exhumation rates in space and time from thermochronometric data. Earth Surf. Dyn. 2, 47–65. http://dx.doi.org/10.5194/esurf-2-47-2014. Gilley, L.D., Harrison, T.K., Leloup, P.H., Ryerson, F., Lovera, O.M., Wang, J.-H., 2003. Direct dating of left-lateral deformation along the Red River shear zone, China and Vietnam. J. Geophys. Res. 108 (B2), 2127. http://dx.doi.org/10.1029/ 2001JB001726. Goren, L., Fox, M., Willett, S.D., 2014. Tectonics from fluvial topography using formal linear inversion: theory and applications to the Inyo Mountains, California. J. Geophys. Res., Earth Surf. 119, 1651–1681. http://dx.doi.org/10.1002/ 2014JF003079. Guenthner, W.R., Reiners, P.W., Ketcham, R.A., Nasdala, L., Giester, G., 2013. Helium diffusion in natural zircon: radiation damage, anisotropy, and the interpretation of zircon (U–Th)/He thermochronology. Am. J. Sci. 313, 145–198. http://dx.doi.org/10.2475/03.2013.01. Henck, A.C., Huntington, K.W., Stone, J.O., Montgomery, D.R., Hallet, B., 2011. Spatial controls on erosion in the Three Rivers Region, southeastern Tibet and southwestern China. Earth Planet. Sci. Lett. 303, 71–83. http://dx.doi.org/10.1016/ j.epsl.2010.12.038. Herman, F., Brandon, M., 2015. Mid-latitude glacial erosion hotspot related to equatorial shifts in southern Westerlies. Geology, G37008-1. Herman, F., Seward, D., Valla, P.G., Carter, A., Kohn, B., Willett, S.D., Ehlers, T.A., 2013. Worldwide acceleration of mountain erosion under a cooling climate. Nature 504, 423–426. Hoke, G.D., Liu-Zeng, J., Hren, M.T., Wissink, G.K., Garzione, C.N., 2014. Stable isotopes reveal high southeast Tibetan Plateau margin since the Paleogene. Earth Planet. Sci. Lett. 394, 270–278. http://dx.doi.org/10.1016/j.epsl.2014.03.007. Hourigan, J.K., Reiners, P.W., Brandon, M.T., 2005. U–Th zonation-dependent alphaejection in (U–Th)/He chronometry. Geochim. Cosmochim. Acta 69, 3349–3365. http://dx.doi.org/10.1016/j.gca.2005.01.024. Houseman, G., Molnar, P., 2001. Mechanisms of lithospheric rejuvenation associated with continental orogeny. Geol. Soc. (Lond.) Spec. Publ. 184, 13–38. http:// dx.doi.org/10.1144/GSL.SP.2001.184.01.02.

20

R. Yang et al. / Earth and Planetary Science Letters 433 (2016) 10–20

Ketcham, R.A., Donelick, R.A., Carlson, W.D., 1999. Variability of apatite fission-track annealing kinetics: III. Extrapolation to geological time scales. Am. Mineral. 84, 1235–1255. Kirby, E., 2003. Distribution of active rock uplift along the eastern margin of the Tibetan Plateau: inferences from bedrock channel longitudinal profiles. J. Geophys. Res. 108. http://dx.doi.org/10.1029/2001JB000861. Leloup, P.H., Arnaud, N., Lacassin, R., Kienast, J.R., Harrison, T.M., Phan Trong, T.T., Replumaz, A., Tapponnier, P., 2001. New constraints on the structure, thermochronology, and timing of the Ailao Shan–Red River shear zone, SE Asia. J. Geophys. Res. 106, 6683–6732. Leloup, P.H., Harrison, T.K., Ryerson, F., Chen, W.J., Li, Q., Tapponnier, P., Lacassin, R., 1993. Structural, petrological and thermal evolution of a Tertiary ductile strikeslip shear zone, Diancang Shan, Yunnan. J. Geophys. Res. 98, 6715–6743. Leloup, P.H., Lacassin, R., Tapponnier, P., Schärer, U., Zhong, D., Liu, X., Zhang, L., Ji, S., Trinh, P.T., 1995. The Ailao Shan–Red River shear zone (Yunnan, China), tertiary transform boundary of Indochina. Tectonophysics 251, 3–84. Lin, T.-H., Lo, C.-H., Chung, S.-L., Hsu, F.-J., Yeh, M.-W., Lee, T.-Y., Ji, J.-Q., Wang, Y.-Z., Liu, D., 2009. 40 Ar/39 Ar dating of the Jiali and Gaoligong shear zones: implications for crustal deformation around the Eastern Himalayan Syntaxis. J. Asian Earth Sci. 34, 674–685. MGMR (Ministry of Geology and Mineral Resources), 1986. Geological Map of Nujiang, Lancang and Jinsha Rivers Area. Geological Publishing House, Beijing. Molnar, P., 2005. Mio-Pliocene growth of the Tibetan Plateau and evolution of East Asian climate. Palaeontol. Electronica 8, 1–23. Nasdala, L., Reiners, P.W., Garver, J.I., Kennedy, A.K., Stern, R.A., Balan, E., Wirth, R., 2004. Incomplete retention of radiation damage in zircon from Sri Lanka. Am. Mineral. 89, 219–231. Ouimet, W.B., Whipple, K.X., Granger, D.E., 2009. Beyond threshold hillslopes: channel adjustment to base-level fall in tectonically active mountain ranges. Geology 37, 579–582. http://dx.doi.org/10.1130/G30013A.1. Ouimet, W., Whipple, K., Royden, L., Reiners, P., Hodges, K., Pringle, M., 2010. Regional incision of the eastern margin of the Tibetan Plateau. Lithosphere 2, 50–63. http://dx.doi.org/10.1130/L57.1. Reiners, P.W., Spell, T.L., Nicolescu, S., Zanetti, K.A., 2004. Zircon (U–Th)/He thermochronometry: He diffusion and comparisons with 40 Ar/39 Ar dating. Geochim. Cosmochim. Acta 68, 1857–1887. Replumaz, A., Tapponnier, P., 2003. Reconstruction of the deformed collision zone between India and Asia by backward motion of lithospheric blocks. J. Geophys. Res. 108, 2285. http://dx.doi.org/10.1029/2001JB000661. Rowley, D.B., Currie, B.S., 2006. Palaeo-altimetry of the late Eocene to Miocene Lunpola basin, central Tibet. Nature 439, 677–681. http://dx.doi.org/10.1038/ nature04506. Royden, L.H., Burchfiel, B.C., King, R.W., Wang, E., Chen, Z., Shen, F., Liu, Y., 1997. Surface deformation and lower crustal flow in eastern Tibet. Science 276, 788–790. http://dx.doi.org/10.1126/science.276.5313.788. Seward, D., Burg, J.-P., 2008. Growth of the Namche Barwa Syntaxis and associated evolution of the Tsangpo Gorge: constraints from structural and thermochronological data. Tectonophysics 451, 282–289. http://dx.doi.org/10.1016/ j.tecto.2007.11.057. Shuster, D.L., Flowers, R.M., Farley, K.A., 2006. The influence of natural radiation damage on helium diffusion kinetics in apatite. Earth Planet. Sci. Lett. 249, 148–161. http://dx.doi.org/10.1016/j.epsl.2006.07.028.

Stock, J.D., Montgomery, D.R., 1999. Geologic constraints on bedrock river incision using the stream power law. J. Geophys. Res. 104, 4983–4993. Tapponnier, P., Peltzer, G., Le Dain, A., Armijo, R., Cobbold, P., 1982. Propagating extrusion tectonics in Asia: new insights from simple experiments with plasticine. Geology 10, 611–616. Tarantola, A., 2005. Inverse problem theory and methods for model parameter estimation. Soc. Ind. Math. 214, 68–70. Wang, E., Burchfiel, B.C., 1997. Interpretation of Cenozoic tectonics in the rightlateral accommodation zone between the Ailao Shan Shear zone and the eastern Himalayan syntaxis. Int. Geol. Rev. 39, 191–219. http://dx.doi.org/10.1080/ 00206819709465267. Wang, E., Burchfiel, B.C., 2000. Late Cenozoic to Holocene deformation in southwestern Sichuan and adjacent Yunnan, China, and its role in formation of the southeastern part of the Tibetan Plateau. Geol. Soc. Am. Bull. 112, 413–423. Wang, Y., Fan, W., Zhang, Y., Peng, T., Chen, X., Xu, Y., 2006. Kinematics and 40 Ar/39 Ar geochronology of the Gaoligong and Chongshan shear systems, western Yunnan, China: implications for early Oligocene tectonic extrusion of SE Asia. Tectonophysics 418, 235–254. Wang, E., Kirby, E., Furlong, K.P., van Soest, M., Xu, G., Shi, X., Kamp, P.J.J., Hodges, K.V., 2012. Two-phase growth of high topography in eastern Tibet during the Cenozoic. Nat. Geosci. 5, 640–645. http://dx.doi.org/10.1038/ngeo1538. Whipple, K.X., Tucker, G.E., 1999. Dynamics of the stream-power river incision model: implications for height limits of mountain ranges, landscape response timescales, and research needs. J. Geophys. Res., Solid Earth 104, 17661–17674. Willett, S.D., Brandon, M.T., 2013. Some analytical methods for converting thermochronometric age to erosion rate. Geochem. Geophys. Geosyst. 14, 209–222. http://dx.doi.org/10.1029/2012GC004279. Willett, S.D., McCoy, S.W., Perron, J.T., Goren, L., Chen, C.Y., 2014. Dynamic reorganization of river basins. Science 343, 1248765. http://dx.doi.org/10.1126/ science.1248765. Wilson, C.J.L., Fowler, A.P., 2011. Denudational response to surface uplift in East Tibet: evidence from apatite fission-track thermochronology. Geol. Soc. Am. Bull. 123, 1966–1987. http://dx.doi.org/10.1130/B30331.1. Yang, R., Willett, S.D., Goren, L., 2015. In situ low-relief landscape formation as a result of river network disruption. Nature 520, 526–529. http://dx.doi.org/10.1038/ nature14354. Yin, A., Harrison, T.M., 2000. Geologic evolution of the Himalayan–Tibetan orogen. Annu. Rev. Earth Planet. Sci. 28, 211–280. Zhang, B., Zhang, J., Zhong, D., 2010. Structure, kinematics and ages of transpression during strain-partitioning in the Chongshan shear zone, western Yunnan, China. J. Struct. Geol. 32, 445–463. Zhang, B., Zhang, J., Zhong, D., Yang, L., Yue, Y., Yan, S., 2012. Polystage deformation of the Gaoligong metamorphic zone: structures, 40 Ar/39 Ar mica ages, and tectonic implications. J. Struct. Geol. 37, 1–18. Zhang, P.-Z., Shen, Z., Wang, M., Gan, W., Bürgmann, R., Molnar, P., Wang, Q., Niu, Z., Sun, J., Wu, J., Hanrong, S., Xinzhao, Y., 2004. Continuous deformation of the Tibetan Plateau from global positioning system data. Geology 32, 809. http:// dx.doi.org/10.1130/G20554.1. Zhang, Y.-Z., Replumaz, A., Wang, G.-C., Leloup, P.H., Gautheron, C., Bernet, M., van der Beek, P., Paquette, J.L., Wang, A., Zhang, K.-X., Chevalier, M.-L., Li, H.-B., 2015. Timing and rate of exhumation along the Litang fault system, implication for fault re-organization in South East Tibet. Tectonics 34. http://dx.doi.org/ 10.1002/2014TC003671.