Zircons and monazites from 6 samples of the North Ayilari dextral shear zone (NAsz), part of the Karakorum fault zone (KFZ), have been dated with the U-Th-Pb method, using both ID-TIMS and SIMS techniques. The ages reveal (1) inheritance from several events spanning a long period between the late Archean and the Jurassic; (2) an Eocene-Oligocene magmatic event (∼35–32 Ma); (3) an Oligo-Miocene magmatic event (∼25–22 Ma), at least partly synkinematic to the right-lateral deformation; and (4) a period of metamorphism metasomatism (∼22–14 Ma) interpreted as thermal and fluid advection in the shear zone. The Labhar Kangri granite located ∼375 km farther Southeast along the KFZ is dated at 21.1 ± 0.3 Ma. Such occurrence of several Oligo-Miocene granites along the KFZ, some of which show evidence for synkinematic emplacement, suggests that the fault zone played an important role in the genesis and /or collection of crustal melts. We discuss several scenarios for the onset and propagation of the KFZ, and offset estimates based on the main sutures zones. Our preferred scenario is an Oligo-Miocene initiation of the fault close to the NA range, and propagation along most of its length prior to ∼19 Ma. In its southern half, the averaged long-term fault-rate of the KFZ is greater than 8 to 10 mm/a, in good agreement with some shorter-term estimates based on the Indus river course, or Quaternary moraines and geodesy. Our results show the KFZ cannot be considered as a small transient fault but played a major role in the collision history.
 Whether strike-slip shear zones in continental collision domains are long-term lithospheric features, steadily accommodating large amounts of strain during long time spans, or short-lived and more transient features participating in distributed regional deformation of the crust remains a topic of debate. In the case of the India-Eurasia collision, interest focus on the main strike-slip faults bounding, or cutting across, the Tibetan plateau (inset Figure 1). Some consider that such faults play a major role in absorbing the convergence [e.g., Tapponnier et al., 1986, 2001], while other think that they are negligible [e.g., England and Houseman, 1986].
 Such age constraints are mostly derived from the study of ductile deformation along the KFZ in the Pangong area, as well as in the North Ayilari range (NA range) ∼200 km further to the SE (Figure 1). Here, we present new U-Th-Pb ages obtained by Isotopic Dilution and Thermo Ionisation Mass Spectrometry (ID-TIMS) and Secondary Ion Mass Spectrometry (SIMS), on zircon and monazite grains from the NA range. Combined with structural evidence our ages bring constraints on the time relationships between magmatism, metamorphism and right-lateral deformation in the NA range and allow discussion on the long-term fault rates of the Karakorum fault zone.
2. Geological Setting of the North Ayilari (NA) Range
 Near Gar, the North Ayilari active fault (NAaf) segment of the KFZ trends more northerly and has a normal component of slip, leading to the subsidence of the Gar pull-apart, and the uplift of the ∼6000 m high North Ayilari (NA) range (Figures 1 and 2) [e.g., Armijo et al., 1986, 1989; Chevalier et al., 2005]. Such recent uplift induced the exhumation of granitoids and gneisses corresponding to a ∼5 km wide shear zone parallel to the NAaf: the North Ayilari shear zone (NAsz). This shear zone corresponds to a deep portion of the KFZ [Matte et al., 1996; Lacassin et al., 2004a], continuously deformed, cooled and exhumed since ∼21 Ma [Valli et al., 2007] (Figure 2). Most samples of this study come from the NA range (Figure 2).
 Further to the SE, the KFZ prolongates along the Indus-Tsangpo suture zone eastward of the Kailas range (Figure 1). No other gneisses are found along the fault trace but dextrally sheared schist outcrop south of the Mt Kailas [Lacassin et al., 2004a]. Farther East the active strand of the KFZ bounds the Labhar Kangri granite to the north (Figure 1). In order to discus the age of that granitoid and its relationship with shearing along the KFZ sample K2P30 was taken from its undeformed southern margin.
2.2. Ductile Deformation Linked With the Karakorum Fault Zone: The North Ayilari Shear Zone (NAsz)
 Ductile deformation in the NAsz is described by Lacassin et al. [2004a], and Valli et al. [2007, section 3, Figures 4, 5, and 6]. Along the NA range northeastern margin most rocks are mylonitized and show a steep high temperature (HT) foliation, striking almost parallel to the NAaf (Figure 2a). That foliation becomes flatter in the core of the range where deformation is milder (Figure 2). Close to the NAaf, the HT foliation is overprinted by a green schist foliation and by brittle-ductile deformation related to the active fault zone (Figure 2). The HT foliation bears a nearly horizontal stretching lineation, which is everywhere parallel to the NAaf even when foliation is flat (Figure 2). Deformation is unambiguously rotational, and shear senses are right-lateral where the foliation is steep (Figures 3c and 3d), and top to the southeast where the foliation is nearly horizontal (Figure 3b). Such geometry with flat foliation away from the core of the shear zone is observed in major strike-slip zones as the Ailao Shan Red River in SE Asia [e.g., Leloup et al., 1995], and is consistent with an overall dextral ductile shearing in a ∼5 km wide shear zone parallel to the NAaf: the North Ayilari shear zone (NAsz) [Valli et al., 2007].
 From the study of quartz and feldspar microstructures, Valli et al.  concluded that dextral shearing in the NAsz was continuous from 700–400°C to temperatures lower than 250°C. Unfortunately, in the absence of index minerals, particularly garnet, and due to the overprinting of early structures by lower temperature ones, the peak temperature reached during shearing cannot be precisely documented but is clearly above 400°C.
2.3. Relationships Between Magmatic and Mylonitic Rocks. Structural Setting of Dated Samples
 Locally, undeformed granites and leucocratic dykes crosscut the HT foliation. For example, in the southwest part of section 1, a large granite body intrudes migmatitic gneisses (Figure 1) that show top to the SE directed shearing parallel to the NAsz on flat-lying foliation. The migmatitic gneisses can be found as panels within the granite. Sample L89 corresponds to the leucosome of migmatitic gneisses outside of the granite (Figures 3a and 3b). Undeformed leucocratic veins cut across both the granite and the migmatitic gneisses (Figure 3h).
 Closer to the NAaf, for example in the NE part of sections 2 and 3, such leucocratic veins are strongly sheared (Figure 3e) and interlayered with right-lateraly sheared, biotite-rich and two-mica mylonites (Figure 3g). P20 (K1P20 of Lacassin et al. [2004a, 2004b]) corresponds to such biotite rich (∼40%) orthogneiss (Figure 2). P18, located ∼300 m farther NE is a dextrally sheared two micas orthogneiss (Figures 2 and 3c). In a similar structural location, P34 (Figure 2) is a leucocratic mylonitic orthogneiss (Figure 3d) located in the NE part of section 2.
 Along section 3, the density of leucocratic veins increases toward the SW. Outcrop 3-A exhibits a large amount of leucocratic dykes, either deformed and transposed parallel to the foliation within the gneisses (T) or mildly to undeformed dykes (C) that crosscuts the foliation and the deformed dykes at high angle (Figure 3f). These field relationships indicate that the dykes are synkinematic to the right-lateral deformation. Sample C32 (K1C32 of Lacassin et al. [2004a, 2004b]) is a mildly deformed crosscutting dyke (Table 1, Figures 2 and 3f).
Table 1. Location (see Also Figure 1 and 2) and Description of Dated Samplesa
 Five kilometers to the SE, section 4 is exclusively made of sheared leucocratic orthogneisses with steep foliation in the NE that shallows to the SW (Figure 2). C43 leucocratic orthogneiss, was sampled where foliations are flatter and deformation mild (Figure 2) but shows top to the southeast shear sense.
 The field relationships described above can be summarized as follows. (1) The migmatitic gneisses (L89) are sheared within the NAsz but are intruded by the granite outcropping in section 1. (2) The leucocratic dykes (C32) intrude all units, including the granite and the dextrally sheared rocks (P18, P34), but (3) close to the NAaf they are all strongly sheared and transposed parallel to the right-lateral HT foliation and appear synkinematic to the right-lateral ductile deformation [Valli et al., 2007].
3. Geochronology Analytical Methods
 Zircon and monazite grains were separated using a Wilfley table, heavy liquids and a Frantz magnetic barrier separator. The final selection of grains according to color and morphology was done using a binocular microscope. The fractions selected for ID-TIMS dating (C43 monazite, C32 zircon and monazite, and P30 zircon) were washed in hot 4 M HNO3 and H2O respectively. The selected grains were dissolved using 29 M HF for zircons or 8 M HCl for monazites in PFA Teflon Ludwig-type Savillex microcapsules [Parrish, 1987] at 220°C, during 24 h (monazite) to 60 h (zircon). Chemical separation and mass spectrometry were performed according to Paquette and Pin . The U and Pb isotopes were measured on a VG Sector 54W mass spectrometer in multicollector static mode. The isotopic ratios are corrected for mass discrimination (0.1 ± 0.015% per amu for Pb and U), isotopic tracer contribution and analytical blanks: 7 ± 2.5 pg for Pb and less than 1 pg for U. Initial common Pb is corrected for each fraction using the Stacey and Kramers  two-step model. Data errors (2σ) of the zircon fractions and ages were calculated using the PBDAT 1.24 and Isoplot/Ex 3.23 programs [Ludwig, 1993, 2003].
 For in situ ion microprobe analyses, the selected grains were mounted together with standard in epoxy resin. The mounts were then abraded and polished to expose at the surface the middle part of the crystals. Each grain was imaged using cathodoluminescence (CL) and backscattered electron (BSE) scanning microscope to characterize the zoning patterns and inner structures. In situ dating of monazite was also performed from rock thin sections, which allows preserving the potential links between the measured ages and the textural location of the grains. Using a diamond saw specific zones from the thin sections containing the radiogenic minerals were extracted. Pieces were then mounted together with zircon and/or monazite standard in epoxy resin and then polished. MOACIR monazite standard [Seydoux-Guillaume et al., 2002] and 91500 zircon standard [Wiedenbeck et al., 1995] were used for samples P20, and TEMORA 1 for all the others [Black et al., 2003]. Zircons in samples P18, P34, C43, L89 and monazites in sample L89 were analyzed for U, Th and Pb isotopes using the sensitive high resolution ion microprobes (SHRIMP II) at the Institute of Geology of Beijing, China, while zircons in sample P20 and monazites in sample P18 were measured using the Cameca IMS 1270 at CRPG in Nancy, France. Calibration parameters, data acquisition and age correction are described by Compston et al.  for the SHRIMP II, and by Deloule et al.  for the Cameca IMS 1270. The error on the calibration curve is taken into account for the age uncertainty calculation. The spot size was between 30 and 60 μm, and their contours were precisely drawn after each analytical session using secondary and backscattered electron (BSE) images.
 Ion probe U-Th-Pb dating of young minerals is an analytical challenge because of the very small amounts of radiogenic daughter isotopes (206Pb, 207Pb, 208Pb). In case of recent minerals, it is now usual for most geochronologists to consider the 238U/206Pb ages as the most reliable for zircons [e.g., Stern and Amelin, 2003], and the 232Th/208Pb ages for monazites [e.g., Catlos et al., 2004]. The isotopic systems of zircons and, to a lesser extent, monazites keep the memory of several distinct magmatic, metamorphic and hydrothermal events. This provides the opportunity to reconstruct complex geological histories but requires cautious interpretation of the analytical results to individualize the different populations.
 Within a given population of ion probe data, it is important to distinguish meaningful ages from outliers, which can always occur in spite of careful selection of rocks and minerals, and of rigorous analytical conditions. Age disparity around a mean value may results either from (1) an overlap of the probe beam on zones of distinct ages, (2) large SIMS analytical errors related to low radiogenic Pb content in young zircon overgrowths [Stern, 1997], (3) the occurrence of common Pb, (4) 230Th radioactive disequilibrium in monazites [Schärer, 1984], (5) a partial lead loss due to (a) subsequent high temperature event(s), (6) a combination of these points. For example, zircon z227 of sample C43 gives significantly distinct ages at its two tips that should a priori give similar ages because they belong to the same growing band (z227-2 & -3, Table 4 and Figure 4b). One of these two ages is among the youngest of its population while the other is the oldest. It is very difficult to determine which result is the most significant. Consequently, we consider that the best age estimate of a given population of ion probe data is its mathematical mean with a two standard deviation uncertainty, which will lower the influence of outlier(s).
4. U-Th-Pb Data and Interpretations
 The Tertiary SIMS data are plotted on the Tera-Wasserburg diagram [Tera and Wasserburg, 1972] (1σ error crosses for readability) while others data are plotted in concordia diagrams (2σ ellipse errors or larger symbols when ellipses are too small). Errors mentioned in the text are at the 2σ level, the weighted averages, and the associate 95% confidence errors, were calculated with Isoplot 3.23 of Ludwig .
4.1. Top to the South Sheared Rocks in the Core of the NA Range
4.1.1. Migmatitic Leucosome L89
 Sixteen zircons were imaged and in situ dated using SHRIMP II (Table 2). The zircon grains are euhedral to subhedral and exhibit distinct rim-core domains and oscillatory zoning (e.g., Figure 4).
Table 2. Migmatitic Gneiss L89 Zircon SHRIMP II Dataa
Results are presented by increasing 206Pb/238U ages. CL domains (within brackets when unclear): C, core; R, Rim; O oscillatory; n.d. not defined. Type, type of event on which the analyzed domain will provide an age constrain, Ma, magmatic; Me, Metamorphic. Th, and U concentration calculated following Stern . n.s., not significant.
 Zircon z28 is euhedral and exhibits an oscillatory zoning pattern (Figure 4a). The crystal core is characterized by a magmatic Th/U ratio and yields a 34.7 ± 2.4 Ma 206Pb/238U age (Table 2). A small euhedral zircon grain (z34) also displaying a magmatic Th/U ratio is dated at 23.4 ± 4.0 Ma. Twelve grains are concordant and have inherited 206Pb/238U ages between 200 and 500 Ma, possibly representing several Triassic to Cambrian magmatic events or a rough discordia trajectory between 23–35 Ma intercepts and ∼500–600 Ma (Figure 5). Two other analyzed grains (z29, z33) are significantly discordant indicating Proterozoic to Archean 207Pb/206Pb ages of 1.7 and 2.6 Ga. Owing to the very small size of the zircon grains, the probe beam often crosscut cores and overgrowth domains, implying that some ages may correspond to mixed values.
 Eleven subhedral to anhedral monazites coming from sample L89 were dated in thin section with the SHRIMP II. One grain (M1) shows Mesozoic inheritance (Table 3). The ten remaining monazite grains yield 208Pb/232Th ages ranging from 25.4 ± 3.6 Ma to 13.1 ± 5.1 Ma (Figure 6). Note that the oldest monazites, found in inclusions within biotites and chlorites (M3-1, 2 and M25), have an average age broadly synchronous with magmatic zircon z34 suggesting crystallization of the rock at ∼24.2 ± 2.4 Ma (Figure 6). Younger ages can be related to several crystallization events between ∼22–13 Ma, or to partial opening of the Th-Pb system at ≤13 Ma.
Table 3. Migmatitic Gneiss L89 Monazite SHRIMP II Data, Results are Presented by Ascending 206Pb/238U ages
Structural location refers to the shape and position of the monazite crystal with respect to the surrounding minerals (Figure 10): It, interstitial; Ic, included, and the following abbreviation indicates the mineral in which the monazite is included. Bt., biotite; fs., Feldspar; Chl, Chlorite. Th and U concentrations are calculated following Stern and Sanborn .
Represent the radiogenic component.
4.1.2. C43 Leucocratic Orthogneiss
 Eleven zircon grains were selected for ID-TIMS and SHRIMP II dating. Zircon rims are dark in CL images indicating a higher U-content relative to their rounded or prismatic cores (Figure 4b). Ten zircon rims yield concordant ages ranging from 25.2 ± 1.7 Ma to 13.0 ± 0.8 Ma with a median value for the 206Pb/238U ages of 17.8 + 4.4/−3.4 Ma (Table 4 and Figure 7a). The Th/U ratios are systematically low (0.01–0.05), pointing out for strong Th depletion relatively to U, which is often related to the (re)-crystallization of the rims during metamorphism or metasomatism [Rubatto et al., 2001]. These rims are strongly U-enriched (6800–20,000 ppm), compared to the cores. This implies that the crystallization of the rims happened in equilibrium with a fluid enriched in trace elements and especially in U [e.g., Rubatto and Gebauer, 1998]. This very high U concentration could have favored Pb-loss, producing the discordance of the analytical points along a chord close to the Concordia. However, this possibility is ruled out by the lack of any correlation between discordance level and U content, the rim growth is therefore most probably related to several hydrothermal pulses between ∼25 and ∼13 Ma. In addition, two monazite fractions analyzed by ID-TIMS are concordant at 14.4 ± 0.7 Ma (Table 5), implying the (re)crystallization of monazites at that time.
Similar caption as Table 2.  Number of grains of the analyzed fraction. ab. = air abraded, an. = anhedral, cl. = colorless, lp. = long prismatic, need. = needle shaped (acicular), pi. = pink, sp. = short prismatic, un. = unabraded, ye. = Yellow. The ± numbers represent the errors made on the last digit.
C32 Leucocratic Dyke
<100  need.cl.un.
0.00353 ± 1
0.0227 ± 1
0.0502 ± 3
<100  need.cl.un.
0.00602 ± 2
0.0441 ± 3
0.0531 ± 3
>100  sp.pi.un.
0.00726 ± 2
0.0583 ± 3
0.0582 ± 2
>100  sp.pi.ab.
0.00822 ± 2
0.0692 ± 5
0.0610 ± 4
>100  sp.pi.ab.
0.00942 ± 2
0.0840 ± 3
0.0647 ± 2
<100  ye.an.
0.00245 ± 3
0.0156 ± 2
<100  ye.an.
0.00246 ± 1
0.0158 ± 1
C43 Leucocratic Orthogneiss
<100  ye.an.
0.00222 ± 1
0.0142 ± 1
<100  ye.an.
0.00220 ± 1
0.0141 ± 1
 Among eleven spots on zircon cores, a single one (z244-1) yields a subconcordant age of ∼22 Ma (Figure 7a) coinciding with zircon rims and displays intermediate Th/U ratio (0.09) and U-content (1379 ppm). All the other cores yield older ages ranging from 206Pb/238U ∼170 Ma to 207Pb/206Pb ∼3.0 Ga (Figure 7b). Most of these cores show high Th/U ratios (0.18–2.2) suggesting a magmatic origin (Table 5), but the analytical points are mostly discordant and it is impossible to distinguish a simple age pattern or to draw a significant discordia array (Figure 7b).
4.2. Right-Laterally Sheared Gneisses in the Core of the North Ayilari Shear Zone
4.2.1. P34 Leucocratic Mylonitic Orthogneiss
 Nine euhedral to subhedral grains (e.g., Figure 4c) were selected for in situ SHRIMP II dating. Most of the cores and rims exhibit oscillatory zoning, which are often disrupted in the complex and recrystallized cores.
 Five zircon rims characterized by low metamorphic Th/U ratio (0.01–0.08) yield concordant and nearly concordant tertiary 206Pb/238U ages ranging from 26.9 ± 1.6 to 18.2 ± 1.2 Ma with a mean of 22.1 ± 4.7 Ma (Table 6, and Figure 8a). U contents are higher in the rims (1400–2500 ppm) than in the cores (170–1300 ppm). This suggests rim formation by (re)crystallization in equilibrium with a metamorphic fluid enriched in trace elements [Hoskin and Schaltegger, 2003; Rubatto and Gebauer, 1998]. One core (z120-1, not plotted in Figure 8 since no reliable 207Pb/235U age was obtained), characterized by low metamorphic Th/U ratio (0.01), and low U-content (230 ppm), and by the absence of oscillatory zoning, yielded a 20.9 ± 2.2 Ma 206Pb/238U age. Three cores (z147-1, z156-1, z153-1) and one rim (z145-1, 3) yielded concordant ages around 300 Ma (Figure 8b and Table 6). The five other analyzed cores are significantly discordant and indicate Proterozoic 207Pb/206Pb ages between 0.9 Ga and 1.8 Ga.
Table 6. Leucocratic Mylonite P34 Zircon SHRIMP II Dataa
 Twenty-four ages where obtained from ten zircon crystals using the SHRIMP II (Table 7). All zircons are euhedral, colorless, nonfractured, and contain rounded to prismatic cores. On CL images, eight of these crystals show rims much darker than cores (e.g., Figure 4d) indicative of higher U-content at their borders [Hanchar and Rudnick, 1995; Rubatto and Gebauer, 1998]. Some cores or rims exhibit clear oscillatory zoning. Only two zircons crystals (z8 & z41) present uniform oscillatory zoning patterns over the whole grain without any rim-core transition (Figure 4d). The 206Pb/238U ages measured in the central parts of these two grains are similar and yield a mean value of 34.4 ± 1.3 Ma (Table 7 and Figure 9a). Their Th/U ratios (0.21 to 1.40) rather favor a magmatic origin for the zircons [Rubatto and Gebauer, 1998]) as well as their continuous oscillatory-zoning patterns [Gebauer, 1996; Rubatto and Gebauer, 1998; Schärer et al., 1995; Vavra et al., 1996]. This suggests that magmatic zircons crystallized around 34 Ma are present in sample P18.
Table 7. Sheared Two-Mica Granitoid P18 Zircons SHRIMP II Dataa
 Twelve analyses of the rims were performed, all ages being concordant or subconcordant within 2σ error limits, with 206Pb/238U ages spanning between 27.4 ± 1.9 and 19.8 ± 1.4 Ma (Table 7 and Figure 9a) with a 206Pb/238U median age of 23.3 + 4.0/−2.2 Ma.
 Six zircon cores yield concordant 206Pb/238U ages spreading between 181 ± 13 and 438 ± 27 Ma (Table 7). Such ages could correspond either to several Lower Jurassic to Cambro-Ordovician magmatic events, or to a discordia line between ∼34–23 Ma and ∼500 Ma intercepts (Figure 9b). Two cores give older 207Pb/206Pb apparent ages at ∼1.0 and ∼1.8 Ga, suggesting inheritance from Proterozoic magmas.
 Sixteen subhedral to anhedral monazites were dated in situ within thin sections (Table 8) using the SHRIMP II. On BSE images the grains display concentric to patchy zoning patterns (Figure 10). 208Pb/232Th apparent ages span between 27.9 ± 3.6 Ma and 13.5 ± 1.7 Ma with a relationship between the ages and the textural position of the grains (Figure 11 and Table 8). The four oldest monazites are included in biotite or feldspar crystals, with a mean 208Pb/232Th age of 25.2 ± 1.6 Ma, whereas the twelve remaining interstitial monazite grains display ages ranging from 23.7 ± 3.0 Ma to 13.5 ± 1.7 Ma with a mean of 18.8 ± 2.8 Ma.
Table 8. Sheared Two-Mica Granitoid P18 Monazite SHRIMP II Dataa
 Included monazites are broadly synchronous with zircon magmatic rims which favor the hypothesis of the crystallization of the rock at ∼25–23 Ma. Th-Pb ages of interstitial monazites are younger, down to 13.5 ± 1.7 Ma (Figure 11). This could be related to the occurrence of one or several thermal pulses between ∼23 and 13 Ma, as interstitial monazites are potentially more sensitive to hydrothermal fluids than the included grains. The Th/Pb dating providing only apparent ages, this 25–13 Ma spread in age could also be related to the discordancy of the analyzed spots toward a lower intercept at ≤13 Ma.
4.2.3. P20 Biotite-Rich Gneiss
 Four zircon grains were dated in thin sections with the Cameca IMS 1270. 206Pb/238U ages range between 23.9 ± 6.9 Ma and 19.9 ± 1.6 Ma with an average value of 21.7 ± 3.6 Ma (Table 9 and Figure 12). The high Th/U ratios suggest crystallization or recrystallization of magmatic zircon grains at ∼22 Ma.
 Five zircon fractions were selected for ID-TIMS dating. The most acicular zircon grains (z1) yield a concordant age of 22.7 ± 0.1 Ma, while the four other fractions are discordant, and define a chord yielding a lower intercept at 32.5 ± 2.6 Ma and an upper intercept at 1296 ± 120 Ma (Figure 13 and Table 5).
 The concordant fraction at ∼23 Ma most likely represents the crystallization of zircons during a high temperature event. The lower intercept at ∼33 Ma may represent either a high temperature event, or may be related to a fortuitous alignment of inherited zircon grains affected by multiepisodic Pb loss. The good alignment and the position of the analytical points close to the lower intercept rather favor the first hypothesis. The syn-kinematic leucogranite C32 crystallized at ∼23 Ma and probably represents the result of partial melting of a ∼33 Ma old magmatic intrusion. The upper intercept at ∼1300 Ma is poorly defined by discordant fractions and may reflect either a single crystallization event or the average of a complicate inheritance pattern.
 Two monazite fractions were dated by ID-TIMS (m1 and m2, Table 5). The concordant, slightly overlapping analytical ellipse errors yield a mean age of 15.8 ± 0.2 Ma (2σ) (Figure 13). This implies the primary crystallization of newly formed hydrothermal or metamorphic monazite crystals or the resetting of older magmatic grains within the leucogranite C32 at that time.
4.4. Labhar Kangri Granite K2P30
 In sample P30, the zircons are colorless and translucent needle-shaped crystals, as well as elongated light pink and yellow short-prismatic euhedral grains. Six fractions were selected for ID-TIMS dating (Figure 14). One fraction of acicular grains yield a concordant age of 21.1 ± 0.3 Ma (Table 10), comparable to those measured on westernmost granitoids from the Ayilari range. The five remaining fractions are discordant and indicate two different inherited component with upper intercept ages at 494 ± 45 Ma and 1447 ± 38 Ma.
Table 10. K2P30 ID-TIMS U–Pb Isotope Data for Zircona
 All U-Th/Pb ages obtained in the Ayilari range are summarized in Table 11 and Figure 15. All samples unless P20 reveal at least three magmatic or metamorphic / metasomatic episodes. All samples record Cenozoic ages comprised between 35 and 14 Ma. The age inheritance patterns derive from several events spanning a long period between the late Archean (∼3 Ga) and the Jurassic (∼ 170 Ma). Three samples (C32, P18, and L89) indicate an Eocene-Oligocene (∼35–32 Ma) magmatic episode. These three samples also show a younger Oligo-Miocene magmatic event for which populations of monazite and zircon grains yield average ages between ∼25 and 23 Ma. This magmatic event is also indicated by P20 zircons (21.7 ± 3.6 Ma). A ∼25–22 Ma magmatic event is thus recorded in most samples of the NAsz.
Table 11. Summary of ID-TIMS, SHRIMP II and Cameca IMS 1270 Agesa
Migmatitic Gneiss L89
Migmatitic Gneiss L89
Leucocratic Orthogneiss P34
Two-Mica Orthogneiss P18
Two-Mica Orthogneiss P18
Biotite-Rich Gneiss P20
Leucocratic Dyke C32
Leucocratic Dyke C32
Leucocratic Orthogneiss C43
Leucocratic Orthogneiss C43
Average ages in Ma of zircon/monazite grains populations are reported in bold with the corresponding 95% confidence level error (xx.x ± x.x). Coresponding individual ages range of the population (xx.x − xx.x) with extremum 2σ errors (+x.x, −x.x) and number of individual ages [n] are also given. Magmatic and, metamorphic / Hydrothermal zircons are distinguished from their crystal shape, color, Th/U ratio, U content and BSE – CL patterns, see text for details.
34.7 ± 2.4 
34.4 ± 1.3 
32.5 ± 2.6 
23.4 ± 4.0 
24.2 ± 2.4 (25.4 − 23.1 +3.6, −6.9) 
23.3 +4 −2.2 (27.4 − 19.8 +1.9, − 1.7) 
25.2 ± 1.6 (27.9 − 24.1 +3.6, −3.1) 
21.7 ± 3.6 (23.9 − 19.9 +6.9, −1.6) 
22.7 ± 0.1 
21.1 ± 0.3 
Metamorphic or hydrothermal
18.2 ± 1.8 (21.8 − 13.1 +2.6, −5.1) 
22.1 ± 4.7 (26.9 − 18.2 +1.6, −1.2) 
18.8 ± 2.8 (23.7 − 13.5 +3.0, −1.7) 
15.8 ± 0.2 
21.7 ± 1.4 
14.4 ± 0.7 
Metamorphic or hydrothermal
17.8 +4.4 −3.4 (25.2 − 13.1 +1.7, −0.8) 
 In samples P34 and C43, metamorphic zircons, characterized by low or intermediate U/Th ratio and U-contents, crystallized around 21 Ma. Hydrothermal zircons characterized by high to very high U-content, and very low U/Th ratio crystallized between ∼22 and 17 Ma while monazite populations in samples L89, P34, C32, P18, and C43 yield ages between ∼19 and ∼14 Ma. We interpret these ages as the result of metamorphism and metasomatism either during several pulses during the ∼22–14 Ma time period, or as a single ∼14 Ma event producing partial Pb loss and discordance of the analytical points.
 The cooling history of the NA range is constrained by Ar/Ar, U-Th/He and fission track data (Figure 16) [Valli et al., 2007; Lacassin et al., 2004a]. These data reveal rapid cooling below ∼350°C starting at ∼15 Ma along section 1 and at ∼13 Ma in all other sections. Such cooling history fits well with the fact that no zircon, nor monazite populations, gives U-Th/Pb age younger than ∼14 Ma.
 The picture that emerges is that major Oligo-Miocene (23.4 ± 1.4 Ma) magmatic, event(s) affected rocks containing Eocene-Oligocene and pre-Mesozoic inherited zircons. These rocks were then affected by metamorphism and metasomatism, and after ∼15 Ma rapidly cooled below 350°C (Figure 16).
5.2. Age of Deformation in the Ayilari Range
 Because C32 leucocratic dyke is synkinematic to the right-lateral deformation (see section 2.3, Figures 3e, 3f, and 3g), Lacassin et al. [2004a, 2004b] inferred that deformation started prior to the emplacement of that dyke. At a larger scale, the fact that numerous undeformed leucocratic dykes cut across the deformed migmatites (Figure 3a) corroborate that such dykes postdate the onset of deformation in the NAsz. An age of 22.7 ± 0.1 Ma is given to C32 by the youngest concordant zircon fraction it contains (Figure 13 and Table 5). This age is a lower bound for the onset of right-lateral shear.
 The three other samples that record Oligo-Miocene magmatism (L89, P18, and P20) have been affected by ductile deformation related to the NAsz. Their U/Pb ages spanning from 25 to 21 Ma could thus be interpreted as an upper bound for the onset of shearing (i.e., the age of the protolith prior to deformation). However, the fact that these ages are identical within errors to that of C32 (Figure 15), and that S-C fabrics, typical of shear under high to medium-grade temperature conditions and often of syn-kinematic granitoid [e.g., Gapais, 1989a, 1989b; Gapais and Barbarin, 1986], are ubiquitous in the NAsz (Figure 3), suggest that migmatisation (L89) and Oligo-Miocene magmatism (P18 and P20) are at least partly synkinematic. This implies that the magmatic ages of these rocks do not provide a strict upper bound, but that deformation started some time during or prior to the Oligo-Miocene magmatic episode at 22–25 Ma. On the other hand, there is no clear argument to link the older Eocene-Oligocene phase of magmatism seen in samples L89, C32, and P18 (35–32 Ma) to right-lateral deformation deformation along the KFZ.
 These conclusions are in agreement with the cooling histories based on Ar/Ar data that suggest that the shear zone temperature dropped below 400°C before 15 to 21 Ma depending on the samples considered (Figure 16) [Valli et al., 2007], thus implying that ductile deformation was chiefly acquired prior to 21 Ma at least for some part of the shear zone. This interpretation is however strongly contested by authors that consider that there is no evidence for synkinematic partial melting, and thus that right-lateral shear should have started after, not before, ∼21 Ma [Searle and Phillips, 2004, 2007]. The crucial point in that controversy is that the crosscutting dykes are clearly syn to late kinematics (C32 see Figure 3) and dated at 22.7 ± 0.1 Ma (Figure 13), implying that deformation started before that time. Postponing the onset of ductile deformation in NAsz would imply (1) to contest the age of C32, (2) to infer that temperatures stayed above 400°C more recently than suggested by the Ar data and (3) to provide an alternative explanation for the ages of metamorphic and metasomatic zircons and monazites that fill the gap between the last magmatic event and the onset of rapid cooling linked with right-lateral / normal deformation.
 The most straightforward interpretation of our data is thus to confirm the analysis of Lacassin et al. [2004a, 2004b] that links the Oligo-Miocene magmatism, and the following metamorphism and fluid circulation in the NA range to right-lateral shear in the Karakorum shear zone. Metamorphism and fluid advection is typical of major strike-slip shear zones as documented in several natural examples [e.g., Leloup et al., 1999, and references therein; Moore et al., 2001]. This in turns implies that right-lateral deformation initiated at high temperature in the NAsz prior to ∼22 Ma, probably during or just prior to the 25–22 Ma magmatic episode. The deformation pursued under decreasing temperature conditions until ∼300°C were reached ∼15 to 10 Ma ago [Valli et al., 2007]. Since then right-lateral / normal brittle deformation occurs along the NAaf.
5.3. Insights Into Pre-Tertiary History of the Ayilari Rocks
 The Ayilari range is located in the southeastern prolongation of the Ladakh batholith (Figure 1) that emplaced between ∼103 and 50 Ma [e.g., Schärer et al., 1983; Weinberg and Dunlap, 2000; Schwab et al., 2004]. Surprisingly, despite the large number of analyzed zircons in our study (∼50) we did not find any inherited core recording such crystallization ages (Table 11). As already discussed, the age inheritance patterns of most samples are multigenetic with zircon grains being derived from several events spanning a long period between the Jurassic and the late Archean (between ∼170 and 3000 Ma) (Table 11, Figures 5, 7b, 8b, 9b, 13, and 14). This large age span together with the high mica content of the samples (between 13 and 40%) suggests that most of the Ayilari Range Tertiary granitoids initially derive from melting of metasedimentary rocks [Chappell and White, 1974], possibly through several melting episodes. This indirectly confirms that the eastern part of the Ladakh batholith emplaced on a continental basement as proposed by Rolland et al.  and Rolland . The youngest concordant zircon grain ages, older than the Ladakh batholith emplacement, are around 400 Ma. This age yield an upper bound for the deposition of the sediments constituting the protolith. Considering the Tertiary events as lower intercepts, discordant inherited zircons younger than ∼400 Ma broadly define a discordia line with this Early Devonian upper intercept (Figures 8b and 9b). This s.uggests a post Early Devonian deposition of the original sediments. In the absence of any Himalayan age, deposition probably took place prior to the emplacement of the Ladakh batholith. Such sediments possibly correspond to the Paleozoic-Mesozoic Tethyan series, which are actually outcropping on the Qiangtang and Lhasa blocks [Jiao et al., 1988].
5.4. Timing of Onset of the KFZ: Conflicting Constraints From Various Places Along the Fault?
 Because the KFZ stands out as the main active strike-slip fault bounding the Tibetan plateau to the southwest its timing and slip history have been the subject of several studies. Various constraints on the timing of the fault onset have been proposed in different places. These constraints are summarized in Table 12 and Figure 17.
Table 12. Published Timing Constraints Ages of Shearing Onset and Miocene Magmatism Along the KFZa
Onset Age of Right-Lateral Deformations Along the KFZ
2nd magmatic episode (P18, P20, P34, C32, C43 and L89)
21.1 ± 0.3 Ma
 Along the Pangong and Nubra ranges (Figure 1), the KFZ was first thought to be younger than ∼5 Ma [Searle, 1996]. Since then, the description of ductile deformation in these ranges, in a restraining bend between two strands of the KFZ, and the dating of granitoids led to propose older ages for the onset of deformation [e.g., Searle et al., 1998]. By linking rapid cooling, starting at ∼17 Ma according to cooling histories based on Ar data, with transpression along the fault Dunlap et al.  proposed that motion on the KFZ started at that time ( on Figure 17 and Table 12). Considering that all granitoids, unless few late dykes, are strictly prekinematics, led to bracket the onset of right-lateral shear between 15.7 and 13.7 Ma ( and  on Figure 17 and Table 12), [Phillips et al., 2004; Searle and Phillips, 2004, 2007; Phillips and Searle, 2007].
 Across the Baer basin, Murphy et al. , propose that the South Kailas thrust has been offset 66 ± 5.5 km by the KFZ. The age of this thrust, deduced from the cooling history based on a single Kf Ar data from its footwall [Yin et al., 1999], is ∼13 Ma. This would constrain the KFZ to be younger than 13 Ma at this location ( on Figure 17 and Table 12).
 On the basis of the age constraints at Pangong and Baer, the KFZ might have started in Pangong at ∼14 or 17 Ma and propagated to the SE reaching Baer less than 13 Ma ago ([A] on Figure 17, Murphy et al. ). However, this would imply a shear onset after ∼13 Ma in the NA range, which is in contradiction with our results ( and  in Figure 17 and Table 12), (section 5.2) [see Lacassin et al., 2004a, 2004b; Valli et al., 2007]. It could thus be envisaged that the KFZ nucleated in the North Ayilari 25–22 Ma ago or before, prior to propagate northwestward, reaching Pangong after ∼15.7 Ma, and southwestward reaching Baer after 13 Ma, thus reconciling previous interpretations ([B] on Figure 17). This would imply propagation rates on the order of 20 to 30 mm/a (northwestward), and ≤1 mm/a (southwestward). Such rates would be very low compared for example with the propagation rates of 138 to 200 mm/a inferred for the North Anatolian fault [Armijo et al., 1999]. Furthermore, they would not allow the KFZ to reach its total length before present time. More complicated scenarios, such as variations in fault propagation rate or simultaneous initiation in distant parts of the fault cannot be ruled out, but remain conjectures given the dearth of data.
 To the contrary, there is evidence that the KFZ was active in Tangtse and Baer prior to ∼15 and ∼13 Ma respectively [Lacassin et al., 2004a], implying an older history of the KFZ. According to Murphy et al. [2000, 2002] the KFZ ends southward in the Gurla Mandhata detachement system (GMDS) (Figure 1), implying that the KFZ started to slip less than 13 Ma ago in the Mt Kailas area. Lacassin et al. [2004a] showed that this hypothesis is implausible because (1) there is no demonstrable connection between the KFZ and the GMDS, the ophiolitic rocks in between being not significantly offset (Figure 1); (2) structural mapping in the Kailas range rather indicates that most of the dextral motion is transferred east of the GMDS along the Yarlung Tsangpo suture zone; (3) the South Kailas thrust cannot be used as a marker to define a piercing point because it is a part of the KFZ flower structure mapped in this area. The 13 Ma age derived for this thrust [Yin et al., 1999] is thus only an evidence for KFZ activity at that time. It is thus probable that the KFZ initiated in Baer prior to 13 Ma ago.
 Other lines of evidence also concur to suggest that the Tangtse shear zone in the Pangong area was active prior to ∼16 Ma. (1) Several generations of variously deformed dykes with late ones less deformed and crosscutting earlier transposed ones suggest synkinematic intrusion. (2) The field relationships and microstructures depicted by Searle and Phillips [2007, Figure 3] and Phillips and Searle  have been interpreted to show that ductile deformation started after 15.6 and stopped at ∼13.7 Ma. However, Ar/Ar thermochronology indicates conditions compatible with ductile deformation (≥300°C) until 10–7 Ma [e.g., Dunlap et al., 1998]. This would imply that right-lateral shear took place during 2 Ma (15.6–13.7 Ma), then stopped during 4 to 7 Ma, before to resume with brittle deformation along the Karakorum fault. Such strange behavior would need to be justified. (3) As in the Ayilari range, pervasive C/S fabrics affect the Tangtse granite [Searle et al., 1998; Rolland et al., 2000]. Such structures are the telltale sign that the granite cooled below solidus during shear [e.g., [Gapais, 1989a, 1989b; Gapais and Barbarin, 1986]. The fact that strike-slip deformation occurred at and below temperatures close to the granitic solidus (750°C) was indeed mentioned by Dunlap et al. [1998, p. 904] and Weinberg and Searle [1998, pp. 885 and 890]. (4) The Tangtse granite is intrusive within granulitic (800°C, 5.5 Kb) and amphibolitic (700–750°C, 4–5 Kb) rocks that have been penetratively deformed in a dextral transpressive regime until greenschist conditions were reached [Rolland et al., 2008]. Like the Oligo-Miocene magmatism in the NA range, the intrusion age of the Tangtse granite thus possibly only provides a lower bound to the onset of deformation [Rolland et al., 2008].
 If, as seems to be the case, the KFZ affected the Tangtse and Baer area prior to ∼18 and ∼13 Ma respectively, a completely different history of the fault zone, with an early Miocene onset along most of its length, should be proposed ([C] on Figure 17). Because the remaining age constraints only provide a minimum age for initiation of faulting it is not possible to fully reconstruct such history. However, hints are given by the age of Miocene plutonic rocks that appear to emanate from the KFZ (Figure 17, Table 12, and see section 5.5).
 A few ages in the NA range and in Tangtse have been interpreted to provide even older constrains for the onset of right-lateral shear: prior to ∼35 and ∼32 Ma respectively ( and , Table 12, [D] on Figure 17) [Lacassin et al., 2004a; Rolland et al., 2008]. However, in the Ayilari range, there is no structural argument to link the Eocene-Oligocene magmatic event with right-lateral shear (see section 5.2). In Tangtse, the age only rests on the last three heating steps of an amphibole that do not define a plateau, and correspond to ∼25% of the total gas release. We thus consider this hypothesis ([D] on Figure 17b) as unsubstantiated.
5.5. Miocene Magmatism Along the KFZ
 The Baltoro plutonic unit (Figure 1) crystallized in two pulses dated at 25.5 ± 0.3 Ma [Schärer et al., 1990] ([M2] on Figure 17 and Table 12), and 21 ± 0.5 Ma [Parrish and Tirrul, 1989; Schärer et al., 1990] ([M3] on Figure 17 and Table 12), at relatively low temperature (750–600°C) [Searle et al., 1992]. The batholith has a sigmoidal shape, with its southern and northern edges striking ∼N110°, 100 km west of the KFZ and progressively bending southeastward to become parallel to the KFZ (N142° strike, Figure 17a). South directed thrusting occurred during granite emplacement along the southern edge of the batholith [Searle et al., 1992]. The eastern border of the batholith is bounded by the KFZ active strand and shows dextral S-C structures [Searle et al., 1998]. Like the Tangtse granite, the Baltoro batholith has been interpreted by Searle  and Searle et al.  to strictly predate the onset of dextral motion along the KFZ. However, its large-scale sigmoid shape suggests intrusion at least in part synchronous with right-lateral shear along the KFZ, coevally with thrusting along the batholith southern edge and dextral high to medium temperature deformation along its eastern border. In that case, the 25.5 or 21 Ma ages of the granitoid would, as already suggested by Mahéo et al. , represent a lower bound for the KFZ timing initiation.
 Other evidence for ∼20 Ma magmatism exists along the KFZ trace, while little magmatism of this age is observed far from the fault (Figure 17a) [Mahéo et al., 2002; Schwab et al., 2004, and references therein]. ∼200 km north of the Baltoro pluton, west of Tash Gurgan, an undeformed alkali granite striking parallel to the KFZ (Figure 17a) crystallized at ∼20–18 Ma [Arnaud, 1992; Xie et al., 1992], with very rapid cooling from 18 to 11 Ma, according to Ar/Ar data [Arnaud, 1992; Ronghua et al., 1996; Xie et al., 1992]. ∼375 km southeast of the Ayilari range, the Labhar Kangri granite, just south of the Zangbo suture branch of the KFZ (Figures 1 and 17a), yields a crystallization age of 21.1 ± 0.3 Ma (section 4.4). The alignment of all these Oligocene-Miocene plutonic units, several of them exhibiting evidence for synkinematic emplacement, suggests that the KFZ acted at that time as a heat source through shear heating and/or as a conduit promoting heat advection for magma ascent. This would imply that the KFZ has been a major discontinuity reaching at least into the lower crust. The occurrence of mantle-derived magmatic rocks (lamprophyres) along the KFZ north of the Baltoro granite [Pognante, 1991; Searle et al., 1992], and in the Tash Gurgan alkaline complex [Xie et al., 1992], further suggests that the KFZ roots at or below the crust-mantle transition. Furthermore, in the Pangong Range, the occurrence of deformed granulites was interpreted as evidence for heat advection along a lithospheric-scale shear zone [Rolland, 2002; Rolland and Pêcher, 2001; Rolland et al., 2000].
 The KFZ shares many characteristics with the ASRR shear zone in Yunnan (SE Asia) which has been interpreted as the Oligo-Miocene continental transform boundary between South China and Indochina [e.g., Tapponnier et al., 1990; Leloup et al., 1995, 2001]. The kinematic link between motion along the ASRR and seafloor spreading in the South China Sea [e.g., Briais et al., 1993; Harrison et al., 1996; Leloup et al., 2001] requires that the shear zone affects the whole lithosphere. In both cases, >1000 km long continental faults absorb large offsets (see section 5.6) during a time period of a couple of tens of Ma long. In both cases, the corresponding shear zone, where exhumed, is rather narrow (less than 20 km) and shows nearly pure strike-slip deformation coeval with magmatism partly derived from lower crust and mantle partial melting [e.g., Zhang and Schärer, 1999]. This leads us to interpret the KFZ as a lithospheric-scale fault shear zone comparable with the Red River, Altyn-Tagh, and North Anatolian fault zones.
5.6. Finite Offsets Across the KFZ, Implication for Long-Term Slip-Rate
 The fact that the KFZ is a major boundary that plays an important role in Tibet tectonics is challenged by authors that consider that its long-term slip-rate is low: between 3 and 10 mm/a [e.g., Phillips and Searle, 2007]. After a discussion on the most reliable finite offsets, our new timing data will allow us to estimate the KFZ long-term slip rate.
 The course of the Indus River is offset ∼120 km across the fault north of Shiquanhe (Figure 1) [Gaudemer et al., 1989]. This value only corresponds to a minimum offset on the fault since the incision of the present river course at its present location. Southwest of the active trace of the fault entrenchment probably occurred during or since rapid exhumation in the North Ayilari range ∼16 to 12 Ma ago, thus implying a long-term fault rate ≥8.5 ± 1.5 mm/a [Valli et al., 2007].
Searle et al. , Searle and Phillips , and Phillips and Searle  correlate the Baltoro with the Tangtse granite, which they take to constrain a maximum offset of 120–150 km across the fault near Tangtse. However, because the Tangtse granite is located between two strands of the KFZ, this correlation would only provides the offset on the Southern branch of the fault, hence, a lower bound on the total offset. Furthermore, as the Tangtse granite is possibly synkinematic (see section 5.4), it cannot be used to define a finite offset at all.
 A more detailed discussion of offsets across the KFZ is limited by the lack of accurate mapping and poor knowledge of geological structures and of their age. The most reliable markers that can be used to define piercing points along the KFZ at this stage thus appear to be major suture zones. The Indus-Tsangpo suture can be followed continuously from the eastern syntaxis to the Kailas area (Figures 1 and 17). West of the Kailas, the suture zone is smeared into a ∼50 km wide zone bounded to the south by the KFZ southern branch [Lacassin et al., 2004a; Tapponnier et al., 1986; Peltzer and Tapponnier, 1988]. This branch possibly connects with the Thanglasgo shear zone within the Ladakh batholith (TSZ, Figure 1) [Weinberg and Dunlap, 2000], as suggested by right-lateral shear evidences along the Indus suture zone south of Leh [Stutz and Steck, 1986; de Sigoyer et al., 2004]. Note that according to cooling histories deduced from Ar/Ar data, right-lateral ductile deformation took place in the TSZ prior to 22 Ma [Weinberg and Dunlap, 2000; de Sigoyer et al., 2004], in good accordance with an Oligo-Miocene age for the KFZ. Offset of the Indus suture measured on the south branch of the KFZ reaches ∼200 km [Ratschbacher et al., 1994], or ≥220 km [This study] (Figures 17a, 18, and Table 13). A large scale offset ≤400 km is obtained taking into account the whole Karakorum deformation zone [Lacassin et al., 2004a].
Table 13. Estimates of Suture Zones Offsets Across the KFZ and Corresponding Finite Fault Ratesa
Suture Zone West of the Karakorum Fault Zone
Match East of the Karakorum Fault
Finite Fault Rate
Oligo-Miocene KKF [C]
Miocene KKF [A]
Preferred offsets are in bold. Some offsets of Lacassin et al. [2004a, 2004b] were measured for the whole ∼80 km wide Karakorum deformation zone. In this study, corresponding offsets are measured from piercing points on the trace of the active fault. In any case, uncertainties on measured offsets are difficult to estimate but could reach a few tenths of kilometres. Rates are calculated using initiation ages [A] (italic) and [C] shown in Figure 17b. Preferred rates are bold. See text section 5.6 and Figure 18.
 Much farther north, the Late Palaeozoic-Early Mesozoic South KunLun suture zone is strongly deflected along the Pamir syntaxis east flank but remains essentially continuous from the western KunLun to the Pamir ranges (Figures 17a and 18) [e.g., Schwab et al., 2004]. Between the Indus and the KunLun three other sutures are found on both sides of the KFZ (Figures 17a and 18). However, their precise location and age has been disputed, leading to conflicting offset estimates across the KFZ (Figure 18 and Table 13).
 In the Pamir, the Tanymas suture is probably an equivalent of the Jinsha suture of Tibet [Schwab et al., 2004]. The Jinsha suture and associated rock assemblages in eastern Tibet can be followed westward within central Tibet only to the LongMuCo area [e.g., Matte et al., 1996, Figure 17]. In the absence of detailed mapping, some authors speculate that it extends westward across the Tianshuihai terrane to reach the KFZ North of the K2 [e.g., Schwab et al., 2004; Yin and Harrison, 2000] (Jinsha (a) on Figure 18), which would correspond to a ∼100 to 130 km offset on the KFZ. The bulk of the large-scale geological evidence [e.g., Geology Publishing House, 1998] suggests instead that this suture has been left-laterally offset and dragged along the Gozha fault implying a much larger apparent offset of 435 to 565 km across the KFZ (Jinsha (b) on Figure 18 and Table 13); [Valli, 2005].
 The Late Cretaceous (88–80 Ma) Shyok suture of Pakistan [Petterson and Windley, 1985; Weinberg et al., 2000] appears to be a remnant of a back-arc basin that formed between 108 and 92 Ma [Rolland et al., 2000]. The most obvious match of the Shyok suture across the Karakorum fault is the Shiquanhe suture. This yields a minimum offset of 200 to 240 km (Figures 17, 18, and Table 13). In this, we follow Matte et al.  who interpret the Shiquanhe mafic and ultramafic rocks as a Late Cretaceous-Paleocene suture continuing eastward across South Tibet to Xainza. Alternatively, Kapp et al.  interpreted these rocks as far south travelled klippes from the Bangong suture zone, as did Girardeau et al. [Girardeau et al., 1985] for the Xainza – NamCo ultramafics rocks northwest of Lhasa. In that case the Shyok suture would have no counterpart in southern Tibet. Finally despite their different ages, Searle and Phillips  match the Shyok and Bangong sutures across the KFZ, which brings the offset down to ∼150 km (Table 1 and Figure 18).
Figure 18 summarizes these diverging views on suture offsets across the KFZ. One view favor small offsets (italic numbers in Table 13) [Searle and Phillips, 2007; Murphy et al., 2000]. However, the corresponding interpretations raise the following major problems. (1) The Tanymas suture in the Pamir would have no counterpart in Tibet (Figure 18), while most authors agree that it corresponds both in age and geodynamic significance to the Jinsha suture. (2) Because the Rushan-Pshart and the Jinsha sutures not only have a different age but also a different vergency, it is impossible to match them together. (3) Similarly, matching the Shyok suture with the Bangong suture is just as unlikely. In fact the Shyok back-arc basin (90–110 Ma) was forming at a time when the Bangong realm of the Tethys was in its final stage of closure (100–140 Ma). It seems thus clear that the larger offsets of ∼550, ∼480, and ≥200 from north to south along the KF are the only ones that make sense (Table 13).
 Offsets appear to decrease from NW to SE along the KFZ. This might indicate that the Tertiary KFZ initiated in its NW part and/or has a faster slip rate in that section. Alternatively, part of the offset in the NW could be inherited from older deformation phases. For example the two largest offsets, between the Triassic (Jinsha / Tanymas) and the Early Cretaceous (Bangong / Rushan-Pshart) sutures, could have accrued along a proto KFZ. Since there is little additional evidence to argue for such a proto KFZ, we have chosen to calculate the slip rates corresponding to the above offsets assuming that all are of Tertiary age (Table 13). Such rates thus correspond to maximum estimates. Dismissing the new age results presented here and assuming that the KFZ is of upper Miocene age (A on Figure 17b), and considering the most improbable smaller offsets, would yield rates between ∼6.6 and 11 mm/a (Table 13, italic). With the larger, and more realistic, offsets the rates increase in the 15 to 38 mm/a range (Table 13, bold italic). For an Oligo-Miocene KFZ (C on Figure 17b) the larger offsets yield rates between ≥8 and 27 mm/a (Table 13, bold).
 For the southern half of the KFZ, the long-term rates (≥8–10 mm/a, Table 13) are in good agreement with that deduced from the Indus river offset (≥8.5 ± 1.5 mm/a, Valli et al., 2007], with those derived from Quaternary moraines (10.7 ± 0.7 mm/a, Chevalier et al., 2005], as well as with the geodectic rate of Barnejee and Bürgmann  (11 ± 4 mm/a). Reasons why other geodetic rates, 3.4 ± 5 mm/a [Jade et al., 2004] and 1 ± 3 mm/a [Wright et al., 2004], appear to be lower than even the lowest geologic rate (6.6 mm/a Murphy et al. ) remain to be understood.
 The North Ayilari shear zone was right-lateral prior to ∼22.7 Ma ago. This suggests that the Karakorum fault zone is active since at least the Oligo-Miocene. The occurrence of several Oligo-Miocene granites outcropping along the fault zone, some of which showing evidence for synkinematic emplacement, suggest that the KFZ may have played an important role in the creation and /or collection of crustal melts. Considering the most realistic reconstructions of suture zones on both sides of the fault yield integrated rates ≤27 mm/a in the northern section of the fault zone decreasing to ≥8 to 10 mm/a along the Southern section. Dismissing the geochronological results presented in this paper and considering a Miocene age for the onset of the KFZ would significantly increase the fault rates. The KFZ cannot be considered as a small transient fault as it appears to have stayed stable through the Miocene, and have absorbed more than 200 km of displacement for an integrated fault rate on the order of 1 cm/a.
 We would like to thank W. J. Dunlap, R. Weinberg, and R. Phillips for constructive and detailed reviews of an earlier version of the paper; as well as Y. Rolland, an anonymous reviewer, and L. Ratschbacher for the final version. Field and analytic work were supported by the programs “Intérieur de la Terre” (IT) and “Dynamique et Evolution de la Terre Interne (DYETI)” of CNRS-INSU and by the Institute of Geology, Chinese academy of Geological Sciences, Beijing, China. Access to the Cameca IMS 1270 (CRPG Nancy) was possible in the framework of INSU Service National. This is IPGP contribution 2411.