The >1000-km-long Oligocene—Miocene left-lateral Red River shear zone (RRSZ) and metamorphic belt and the Pliocene—active right-lateral Red River fault (RRF), stretching from SE Tibet to the South China Sea, has been cited as one of the primary examples of a lithospheric scale strike-slip fault that has resulted in syn-kinematic metamorphism and partial melting and accommodated several hundred to a thousand kilometers of horizontal motion as a result of the indentation of India into Asia. Alternatively we interpret the metamorphic complexes along the RRSZ as exhumed metamorphic core complexes of older rocks, subsequently affected by Oligocene–Early Miocene left-lateral shear and localized partial melting (leucogranite dykes), Miocene low-angle normal faulting along margins (Range Front faults), and Pliocene active dextral strike-slip faulting (RRF). Along the Ailao Shan (ALS) and Diancang Shan (DCS) ranges in Yunnan, SW China, early amphibolite facies metamorphic rocks were intruded by K-feldspar orthogneisses of Triassic age (Indosinian). LA-ICP-MS U-Pb zircon dating reveals a complex history with zircon cores showing evidence of Indosinian (∼239–243 Ma) to Neoproterozoic magmatism. Zircon rims show an Oligocene (∼26 Ma) magmatic or metamorphic overprint. Biotite granodiorites and syenites of mantle origin intruded the gneisses during the Oligocene (∼35 Ma). Later biotite leucogranites intruded the orthogneisses and migmatite host rocks before a significant phase of tight to isoclinal folding. Ductile, left-lateral strike-slip shear fabrics were superimposed on all lithologies at high temperature (∼500–550 °C) for the ALS and lower temperatures (∼250–150 °C) after peak metamorphism and after granite intrusion. A few very small biotite (±Grt ± Tur) leucogranite veins and dykes crosscut the ductile strike-slip shear fabrics at Yuanjiang, in the Ailao Shan. Low-angle normal faulting along the margins of the metamorphic massif accommodated final exhumation of the Red River gneisses. Using published U-Th-Pb and 40Ar/39Ar ages of granites along the shear zone, the age of left-lateral ductile shearing along the RRSZ can be constrained as between the earlier deformed leucogranites (31.9–24.2 Ma) and the later crosscutting dykes (21.7 Ma) with exhumation-related cooling continuing until ∼17 Ma.

The Ailao Shan–Red River (ASRR) metamorphic belt forms a series of NW-SE aligned metamorphic complexes (Xuelong Shan, Diancang Shan, Ailao Shan in Yunnan, and the DayNuiConVoi [DNCV] complex in North Vietnam) showing Oligocene–Early Miocene metamorphism and left-lateral shearing aligned along the Late Pliocene–active, right-lateral Red River Fault (RRF). Both the earlier left-lateral shear zone and the active right-lateral fault have been interpreted as a 1000-km-long, lithospheric-scale strike-slip fault that resulted in syn-kinematic metamorphism and leucogranite melting, and accommodated clockwise rotation and continental extrusion of Indochina–SE Asia as a result of the collision and indentation of India into Asia (Molnar and Tapponnier, 1975; Tapponnier et al., 1986, 1990, 2001; Peltzer and Tapponnier, 1988; Leloup et al., 1993, 1995, 2001). The active right-lateral RRF has been mapped along the northeastern margin of the Ailao Shan (in places also called the “Mid-Valley fault”) or the Yuanjiang Fault (Leloup et al., 1995) and shows a reversal of the earlier left-lateral fabrics within the Ailao Shan massif (Allen et al., 1984; Leloup et al., 1995; Replumaz et al., 2001; Schoenbohm et al., 2005, 2009). However, both the active Mid-Valley and Yuangjiang Faults occur in the heavily vegetated Red River valley, and Schoenbohm et al. (2009) could find no evidence for the existence of the Yuangjiang Fault.

The Red River fault extends from the upper reaches of the Mekong River north of the Eastern Himalayan syntaxis, southeastward through the Three Gorges regions into Yunnan (Fig. 1). Some authors (Tapponnier et al. 1982, 1986, 1990; Leloup et al. 1993, 1995, 2001) proposed that the Red River fault includes all the different strands along its trace, whereas others (Wang and Burchfiel, 1997; E. Wang et al., 1998; Burchfiel et al. 2008) proposed that there is no link between the southern segment, south of the “Midu Gap” (Ailao Shan and SE into North Vietnam), and the northern segment (Diancang Shan and to the NW). Burchfiel et al. (2008) mapped three separate belts of mylonitic rocks in southern Yunnan and Vietnam: the Ailao Shan, a middle belt and the DNCV mylonites separated by zones of weakly to un-metamorphosed Palaeozoic-Triassic sedimentary rocks. E. Wang et al. (1998) and Burchfiel et al. (2008) proposed that the name Red River fault should be restricted to the young and active faults that can be traced from the Ailao Shan south into Vietnam. Following E. Wang et al. (1998) in this paper, we distinguish between the older exhumed high-grade gneisses and mylonites along the Red River shear zone (RRSZ) from the Pliocene–active Red River fault (RRF). We also recognize a low-angle normal fault (Range Front fault of Leloup et al., 1995) that bounds the margins of the metamorphic massifs.

Recent discussions concerning the evolution of the RRSZ and RRF have concerned four major aspects: (a) the timing of motion along these faults, (b) the depth that such faults extend (upper crust, whole crust, or into the mantle), (c) the amounts of geological offset, and (d) the degree of metamorphism and melting produced during strike-slip motion (e.g., Searle, 2006, 2007; Leloup et al., 2007). All the metamorphic and igneous rocks were previously interpreted as having formed as a result of shear heating during Tertiary left-lateral strike-slip shearing along the Ailao Shan–Red River shear zone (Tapponnier et al., 1982, 1986, 1990; Leloup et al., 1993; 1995, 2001; Leloup and Kienast, 1993). An alternative hypothesis correlates the metamorphism and mantle-derived granites to earlier metamorphic events, unrelated to the mylonites of RRSZ. In this scenario, mylonite fabrics formed by left-lateral strike-slip shearing occurred after metamorphism and granite intrusion (Chung et al., 1997, 2008; Jolivet et al., 2001; Searle, 2006, 2007; Anchiewicz et al., 2007; Yeh et al., 2008). Jolivet et al. (2001) first suggested that the RRSZ was purely an upper crustal fault, above a major mid-crust detachment.

Many of the critical interpretations center around whether the metamorphism is pre-kinematic or syn-kinematic with respect to left-lateral shearing and whether the leucogranite sills-dykes are pre- syn- or post-kinematic with respect to left-lateral shearing. During our fieldwork in 2007, we visited the Diancang Shan and Ailao Shan in order to assess the regional structural setting of leucogranites and gneisses along the RRSZ. In this paper, we first discuss the major differences in interpretation of the leucogranite ages with respect to the strike-slip fabrics. Further to our previous studies of the DNCV complex along the RRSZ in Vietnam (Searle, 2006, 2007; Chung et al., 2008; Yeh et al., 2008), this paper contributes more to the debate by presenting some important field constraints from the Diancang Shan and Ailao Shan metamorphic rocks in Yunnan. We also report some new zircon LA-ICP-MS and SHRIMP ages. Finally, we use these data in conjunction with earlier published U-Th-Pb age data (Schärer et al., 1990, 1994; Zhang and Schärer, 1999; Gilley et al., 2003; Sassier et al., 2009) to discuss the differing models for the evolution of the RRSZ.

The Diancang Shan (DCS) range is a NW-SE aligned mountain range ∼80 km long and 10–15 km wide (Fig. 2) and exposes a series of high-grade metamorphic rocks bounded by Mesozoic or Tertiary continental clastic red-beds on either side. We distinguish here between (a) the Red River ductile shear zone mylonites in the margins of the massif, (b) the low-angle ductile to brittle normal fault that bounds the margin of the massif west of Dali, and (c) the active Red River fault, of which we found no trace, at least in the southern DCS region. E. Wang et al. (1998) considered that the NNW trending Tongdian fault, thought to be the continuation of the RRF (Leloup et al., 1995), was not aligned with the DCS and actually swings around east-west to cut off the Diancang Shan metamorphic rocks west of Dali (Fig. 2), also implying that the DCS massif no longer has active faults. Burchfiel et al. (2007) show active outward-dipping normal faults bounding the DCS gneisses with no strike-slip motion. Between the Diancang Shan in the NW and the Ailao Shan to the SE, the 80 km long “Midu Gap” shows no metamorphic rocks. Leloup et al. (1993) interpreted this gap as a very large scale oblique extensional C′ shear plane; there is, however, no evidence of this linking fault at the surface.

The mylonite fabric cuts through several different earlier metamorphic rocks, including K-feldspar augen gneisses, amphibolites, biotite gneisses and migmatites, with rare bands of calc-silicate and quartzite. Leloup et al. (1993) mapped all the Diancang Shan gneisses as a 10–15-km-wide Tertiary shear zone and showed that left-lateral shear fabrics occurred irrespective of the dip direction of the schistosity. These authors and Harrison et al. (1996) used Ar-Ar dating to constrain the age of left-lateral shearing as occurring between ∼35 Ma and 17 Ma with a ductile to brittle phase of normal faulting that began at 4.7 Ma.

Metamorphic rocks are, however, not restricted to the RRSZ (Bureau of Geology and Mineral Resources of Yunnan, 1983). “Older” metamorphic rocks have also been mapped away from the RRSZ, and these include the multi-colored Dali marbles that show extreme ductile deformation. The age of metamorphism is as yet unknown, although we suspect that they could be related to the Triassic Indosinian collision event along the Jingsha (“–Ailao Shan”) suture zone. In the northeast part of the Diancang Shan, Devonian limestones have been mapped unconformably overlying the metamorphic rocks (Bureau of Geology and Mineral Resources of Yunnan, 1983).

New quarry exposures in the southern part of the DCS (GPS: N 25.57198°; E 100.18116°; elevation 1940 m) show excellent structural relationships between the host gneisses and intruding sets of granodioritic and leucogranitic melts. These exposures clearly show that the host gneisses and migmatites have been intruded by an early series of K-feldspar orthogneisses (Fig. 3a) and a later set of biotite leucogranites, sometimes also containing garnet and tourmaline. Schistosity and folds within the host gneisses are truncated by the dyke margins (Fig. 3b). A prominent phase of tight to isoclinal folding affects all these rocks. Stretching lineations associated with left-lateral shearing cut across all lithologies, are sub-horizontal, and were clearly superimposed after metamorphism, folding, and granite intrusion (Fig. 3c). A late set of brittle fractures cuts the host gneisses and the granite dykes (Fig. 3d). Normal faults bound both flanks of the DCS massif and dip at angles between 45 and 25° along the northeast and up to ∼75° along the southwest margin, away from the massif.

The DCS is bounded along both NE and SW margins by outward-dipping normal faults suggesting that the width of the DCS gneisses increases with depth. The foliation along the SW margin of the DCS dips SW at moderate angles (∼45°), whereas along the NE margin dips are steeper toward the NE (60–80°) with steep fold axial planes in the center of the massif (Fig. 4). Earlier pre-strike-slip D1 upright and D2 recumbent folds are cut and transposed into parallelism with the later, steep left-lateral shear fabrics (D3) that run along the NE margin of the DCS. Stretching lineations are mainly horizontal, and strain markers are consistently left-lateral. The Range Front normal faults along both SW and NE margins of the DCS cut all earlier fabrics (D4). In the southern DCS foliations are aligned east-west and are truncated by the low-angle normal faults that bound the southern margin of the DCS gneisses.

The Ailao Shan metamorphic massif (Fig. 5) is up to 20 km wide and more than 300 km long (Bureau of Geology and Mineral Resources of Yunnan, 1983), extending SE into Vietnam into the Sapa metamorphic belt NE of the FanSiPan alkali syenite massif and SW of the DayNuiConVoi (DNCV) metamorphic complex (Leloup et al., 1995, 2001). Whereas the Sapa regional metamorphic rocks show little evidence of strike-slip shearing, the DNCV massif does expose mylonites along two steep faults that bound the margins of the massif (Leloup et al., 2001; Searle, 2006; Anczkiewicz et al., 2007; Yeh et al., 2008). The active RRF supposedly follows the northeastern margin of the Ailao Shan (Leloup et al., 1995) but cuts across the earlier RRSZ to follow the southwestern margin of the DNCV (Song Hong fault). Another active strike-slip fault bounds the northeastern margin of the DNCV (Song Chai fault). Both margins of the DNCV show steep mylonite zones with left-lateral kinematic indicators and brittle normal faulting that cut earlier flat-lying and gently folded schistosities within the DNCV (Jolivet et al., 2001; Searle, 2006, Anczkiewicz et al., 2007; Viola and Anczkiewicz, 2008; Yeh et al., 2008).

In the Ailao Shan, high-grade gneisses form a long linear belt between 0–10 km wide (Fig. 5). These rocks are dominantly biotite orthogneisses and K-feldspar augen gneisses with minor amounts of calc-silicate marble, quartzite, and rare pelites. P-T conditions of metamorphism have been estimated at between 3 and 7 kbars and 550–780°C (Leloup et al., 1993; Leloup and Kienast, 1993). Granodiorites, biotite granites and late biotite ± garnet ± tourmaline leucogranites intrude the gneisses. All these rocks have been affected by strong deformation, and late mylonite fabrics related to left-lateral strike-slip shearing along the fault have been superimposed on all lithologies.

The NE margin of the Ailao Shan metamorphic complex is a NE-dipping normal fault (Range Front fault of Leloup et al., 1995) juxtaposing Triassic to Tertiary red-beds directly against high-grade gneisses. The SW margin of the high-grade gneisses is a steep NE-dipping or vertical fault (Ailao Shan fault of Leloup et al., 1995) juxtaposing the gneisses against low-grade schists to the SW. These low-grade schists are not as highly deformed as the gneisses and rarely contain a stretching lineation. Although Leloup et al. (1995) describe both faults as active, we could find no trace of active tectonics along the NE-dipping low-angle Range Front fault, and no outcrop indication of an active Mid-Valley fault along the heavily forested Hong-He (Red River) valley in the central Ailao Shan. The SW margin of the low-grade schists shows a discontinuous belt of ultramafic rocks, mainly serpentinized harzburgites that have presumably been related to the Song Ma suture zone (Leloup et al., 1995). Vertical shear bands cut through the serpentinised harzburgites and low-grade schists indicating that metamorphism preceded strike-slip shearing along the RRSZ.

Yuanjiang Area

Although outcrop is not particularly good in the forested Ailao Shan mountains, we found two small areas along scoured river courses to the west of Yuanjiang where full outcrop permitted detailed structural mapping of the RRSZ mylonites. The first locality (Fig. 6: GPS: N 23.55663°; E 101.91552°; elevation: 712 m) is a small tributary of the Yuanjiang River. Host rock gneisses include garnet amphibolite, biotite orthogneiss and K-feldspar augen gneiss striking between 122 and 132° with very steep dips (between 90 and 75°NE). Biotite bearing leucogranites, sometimes also containing garnet and tourmaline, have intruded the gneisses parallel to the foliation, which has then been subsequently folded (Fig. 7a). There is good evidence for pre-kinematic intrusion of granites into the gneisses (Fig. 7a) and also a later set of post-kinematic dykes that cut folds in the gneisses (Fig. 7b). In some places, small dykes or apophyses of granite break out and intrude across the fabric of the host gneiss, including cutting earlier boudinaged set 1 dykes and layer-parallel set 2 sills (Fig. 7c). NW-SE aligned stretching lineations parallel to the strike-slip shear direction are common in both gneisses and granite (Fig. 7d), although fabrics are less prominent in some of the late cross-cutting apophyses. Numerous kinematic indicators (C-S, C′-S mylonite fabrics, rolled K-feldspar porphyroblasts, asymmetric boudins, etc.) consistently show left-lateral shear. Some outcrops show a mixture of fabrics with both early leucogranites infolded with the host gneisses, and later leucogranites cross-cutting the fabrics (Fig. 7e). Some of these later leucogranites have broken up the host gneisses into detached xenoliths (Fig. 7f) clearly indicating that the gneisses formed first and the leucogranite formed later.

U-Pb zircon and titanite ages from monzonites, monzo-syenites and pegmatites from elsewhere along the Ailao Shan range from 31.9 ± 0.3 Ma to 26.3 – 23.0 ± 0.3 Ma (Schärer et al., 1990; Zhang and Schärer, 1999), ages that date the timing of crystallization of the granite. These authors and Leloup et al. (1995, p. 55) stated that “since the melts are deformed, the ages imply that left-lateral shear was in progress in both massifs [ALS and DCS] at least before 26 Ma until after 22 Ma.” In contrast, Searle (2006, 2007) and Chung et al. (2008) interpreted most of the granites along the DNCV metamorphic belt along the RRSZ in North Vietnam, as well as the alkaline FanSiPan syenite, SW of the RRSZ, (U-Pb titanite age of 35.2 ± 0.4 Ma; Zhang and Schärer, [1999]) as pre-kinematic. Left-lateral strike-slip fabrics were superimposed onto the granites at temperatures (ca 500–550°C) high enough to plastically deform feldspars, but still below melting temperatures as recorded by U-Th-Pb zircon–monazite ages.

Detailed mapping of outcrops in Figure 6, however, shows ambiguity, with evidence that could be interpreted as pre-kinematic, syn-kinematic, or post-kinematic granite intrusion. Pre-kinematic granites are strongly affected by deformation and boudinage and are infolded with the gneisses, as also seen in the Diancang Shan. A few instances also show late veins and apophyses of post-kinematic granite intrusion where very small veins cut across the mylonite fabric. These late veins originate from the cores of parallel granite sheets and are the final melt phase in the RRSZ mylonite belt.

A second outcrop was also mapped in detail near a small hydro-electric power station west of Yuanjiang in the Ailao Shan (Fig. 8: GPS: N 23. 55447°; E 101. 91688°; elevation: 701 m). This outcrop is the same locality as shown in Zhang and Schärer (1999, Fig. 2c), Leloup et al. (1995, figure 13, p. 34), and Sassier et al. (2009, figures 4, 5). It shows very strong mylonite fabrics striking 120–125° and clearly shows at least three generations of leucogranites (Fig. 9a, b). Early biotite granites (set 1) intrude K-feldspar augen gneiss, biotite orthogneiss, and uncommon calc-silicate rocks. These granite sheets are all aligned parallel to the fabric in the host gneiss, and also contain the strike-slip shear fabric (Fig.9a, b; set 1 sills), and are therefore interpreted here as having intruded pre-strike-slip shearing. Set 2 dykes clearly cross-cut metamorphic fabrics and set 1 dykes (Fig.9a), but elsewhere in the same outcrop appear to be parallel to the metamorphic fabric (Fig.9b). We interpret these granites as having intruded syn- to post-strike-slip shearing. A few thin (<10–20 cm) late-stage leucogranites (containing Bt, Grt, and Tur) cross-cut the shearing fabrics and are mainly undeformed (Fig. 9a, b, set 3 dykes). These small dykes are interpreted here as intruding during, and mainly after ductile strike-slip shearing. The set 3 dykes are apparently the same set as the one dated by Th-Pb monazite dating by Gilley et al. (2003) at 21.7 ± 0.2 Ma and by Sassier et al. (2009) at 22.55 ± 0.25 Ma. Based on these ages, we now interpret the timing of ductile shearing along this segment of the RRSZ to have occurred between ∼29 – 22 Ma.

The microstructural evolution of the Yuanjiang locality shows the dominant left-lateral S-C fabrics related to shearing along the RRSZ (Fig. 10). These fabrics clearly cut earlier-formed garnets, amphiboles and biotite, suggesting that metamorphic mineral growth preceded ductile strike-slip shearing.

Four gneiss samples from the Diancang Shan and Ailao Shan areas were subjected to zircon separation and U-Pb dating (Fig. 11; Table 1). Zircons were separated from ∼3 kg samples using conventional heavy-liquid and magnetic separation techniques. Cathode-luminescence (CL) images of individual zircon grains were taken at the Institute of Earth Sciences, Academia Sinica, Taipei, for examining the internal structures and selecting suitable positions for U-Pb isotope determinations. Zircon U-Pb isotopic analyses were performed using a New Wave UP213 laser ablation system combined with an Agilent 7500s quadrupole ICP-MS (inductively coupled plasma mass spectrometer) housed at the National Taiwan University, Taipei. The LA-ICPMS operating conditions and analytical procedures were same as those reported in Chiu et al. (2009). Given that in LA-ICP-MS zircon U-Pb isotopic analysis, precise age measurements using 207Pb/235U and 207Pb/206Pb ratios are feasible usually only for zircons older than ∼800 Ma, essentially due to the fact that 235U comprises less than 1% of natural U and thus relatively little 207Pb can be produced in the case. The weighted mean of pooled 206Pb/238U ages are used to represent the ages of the young zircons dated. The analytical results are summarized in Table 1.

Diancang Shan DS07–16 (Fig. 11 A-B)

Sample DS07–16 is a muscovite-biotite-garnet schist from the southern margin of the DCS (GPS: N 25.57624°; E 100.20002°; elevation 1970 m) This sample contains short prismatic zircons between 150–300 μm in length. CL images show growth zoning and resorption features, typically observed in magmatic and metamorphic zircons, respectively (Hoskin and Schaltegger, 2003). The magmatic cores have highly various U concentrations (150–3281 ppm), whereas the rim areas are more uniform in U (256–590 ppm). We performed 35 spots of analyses on both core and rim, if available, of the zircon separates. Among these, 22 analyses (Table 1) form a concordant cluster yielding a mean 206Pb/238U age of 243.0 ± 1.7 Ma (2σ; MSWD = 0.6), which is interpreted to represent the protolith age of the gneiss. Some older spots that range from ∼300–1000 Ma are discordant and considered to represent inherited zircons that may have experienced Pb lost when the protolith was metamorphosed during the Triassic Indosinian time.

Diancang Shan DS07–19A (Fig.11 C-D)

Sample DS07–19A is a biotite orthogneiss that forms the host rock to several granitic intrusions and was collected from the quarry site in the southern DCS (GPS: N 25.57198°; E 100.18116°; elevation 1940 m). Zircons from this sample are stubby to elongate in shape, showing rounded to subrounded terminations. Most zircons that show highly luminescent cores in CL images have low to medium U concentration ranging from 62 to 554 ppm. In two zircon grains, rims with dark CL images and high U (2089–4492 ppm) are recognized. Twenty-five spots on 24 grains were analyzed (Table 1). 24 analyses yielded a concordant cluster with a mean 206Pb/238U age of 234.9 ± 1.8 Ma (MSWD = 0.3), which is interpreted to represent the formation age of the magmatic protolith. Few analyses that plot slightly below the Concordia line show younger ages, probably due to radiogenic Pb loss. The two high-U zircon rims exhibit very low Th/U ratios (0.006–0.010), similar to the zircons crystallizing at high metamorphic grade (Williams et al., 1996), and yield 206Pb/238U ages at ca. 24 Ma, which we interpret as indicating the timing of metamorphic overgrowth.

Ailao Shan AL07–24A (Fig. 11 E-F)

Sample AL07–24A is a mylonitic biotite–K-feldspar orthogneiss collected from the Yuanjiang site in the Ailao Shan (Fig. 6: GPS: N 23. 55663°; E 101. 91552°; elevation: 712 m). This sample represents the host orthogneiss into which the two later sets of leucogranitic dykes have been intruded. The sample contains mostly euhedral, transparent, colorless to pale brown zircons, which have average crystal lengths of ∼150–300 μm and length-to-width ratios from 2:1–3:1. CL images show that most zircons have complex internal structures. The cores show magmatic growth zoning with variable U from 54 to 923 ppm and Th/U ratios from 0.22 to 1.12. Twelve zircon grains gave older 206Pb/238U ages from 239 to 818 Ma, suggesting multiple magmatic sources of the protolith. Most zircons contain rims that are characterized by higher and more variable U concentrations ranging from 280 to 4913 ppm, with Th/U from 0.11 to 1.01. Twenty-two analyses on the rims yielded a weighted mean 206Pb/238U age of 26.2 ± 0.3 Ma (MSWD = 2.0), implying a magmatic or metamorphic origin. These data indicate a complex history of sample AL07–24A, which contains inherited zircons of Indosinian to Neoproterozoic magmatism. The rock was overprinted by an Oligocene magmatism or metamorphism (ca. 26 Ma) during which zircon rims overgrew.

Ailao Shan AL07–26A (Fig. 11 G-H)

Sample AL07–26A was collected from the Yuanjiang dam site (Fig. 8: GPS: N 23. 55447°; E 101. 91688°; elevation: 701 m). This rock is an orthogneiss and is the host rock into which the three sets of dykes have been intruded (Fig. 9). Zircons from this sample are dominated by clear to brownish, rounded to short prismatic, partly irregular grains, ranging in size between ∼100 and 300 μm. Most zircons reveal sector zoning in CL images. Rim structures that are occasionally observed have an outer zone of high luminescence, indicative of low U concentration. Thirty-eight spots were analyzed on 31 zircons from this sample. A cluster of 26 analyses gave a weighted mean 206Pb/238U age of 239.4 ± 1.9 Ma (MSWD = 1.3), interpreted to reflect the emplacement age of the magmatic protolith. Several older grains, ranging from early Palaeozoic and Neoproterozoic, indicate the presence of earlier magmatic events in the region. Four young ages from 29.0 to 34.0 Ma were obtained from the rims (Table 1), which have medium to high U concentrations (429–3785 ppm) and low Th/U ratios (0.02–0.05). These data suggest that the sample was derived from an Indosinian magmatic event that later underwent a metamorphic overprint during the Oligocene.

In the Ailao Shan, U-Pb dating of zircons, monazites, and xenotimes from granite veins parallel to the foliation and granite veins affected by left-lateral shearing gave ages ranging from 26.3 ± 0.2–23 ± 0.2 Ma (Schärer et al., 1990, 1994). Two “layered granitic intrusions (YS-53 and YS-13) within the gneiss massif” gave monazite ages of 31.9 ± 0.3 Ma and 25.8 ± 0.2 Ma (Zhang and Schärer, 1999, p. 70). In the Diancang Shan, ages from a “cross-cutting aplitic layer (MD-1)” are between 24.4 ± 0.2 Ma and 22.3 ± 0.3 Ma, suggesting that the fabric must have been pre-24.4 Ma (Zhang and Schärer, 1999, p. 69). Schärer et al. (1990, 1994) and Zhang and Schärer (1999, p. 67) correctly state that such ages date the crystallization of the granite melt. However, they also concluded that “magmatism and left-lateral slip movements occurred coevally during latest Eocene – Oligocene times from 33 to 22 Ma” even though some of their samples (e.g., MD-1 Diancang Shan) cross-cut fabrics. Leloup et al. (1995) further stated that “since the melts are deformed, the ages imply that left-lateral shear was in progress in both massifs from at least before 26 Ma until after 22 Ma”.

The possible exhumation path through time of a typical rock from the RRSZ is illustrated in Figure 12. U-Pb ages are interpreted as dating the timing of crystallization of the granite at temperatures probably >750°C (Parrish, 2001). Following crystallization, transpressional exhumation up to mid-crust levels resulted in the rock cooling through 40Ar/39Ar closure isotherms for hornblende (∼ 550°C), K-feldspar (500 °C), and biotite (∼ 300 °C) (Parrish, 2001). High-temperature ductile mylonite fabrics (Hanmer and Passchier, 1991) were superimposed at temperatures high enough for feldspars to deform plastically (∼ 450–500 °C) and quartz to deform plastically (∼ 300 °C). These fabrics, although still high temperature (∼ 450–550 °C), were all formed after granite crystallization, so the U-Pb granite ages will only give a maximum age of strike-slip initiation (Searle, 2006, 2007). Anczkiewicz et al. (2007) showed that microstructures from the RRSZ mylonites in the DNCV in North Vietnam were compatible with maximum deformation temperatures around 500–550 °C, temperatures that are less than those required for granite melting. Complete dynamic recrystallization of calcite occurs at ∼300 °C (Passchier and Trouw, 1996). Above this, at temperatures less than ca. 300 °C, brittle faulting produces cataclasites in the upper seismogenic layer.

Searle (2006, 2007) suggested that the majority of the leucogranites along the RRSZ were pre-kinematic and deformed at high temperatures within the solid-state ductile regime after crystallization of the granite. Photographs of the deformed leucogranites in Leloup et al. (1995) also suggest that they have been deformed subsequent to crystallization and emplacement in the host gneisses, with extensive boudinaged and stretched pods. Subsequently, new Th-Pb ages from a few minor cross-cutting leucogranite veins have been published by Sassier et al. (2009); these new data now require a revision of timing constraints of ductile shear along the RRSZ.

Only a few dykes cross-cut the prominent regional left-lateral shear fabrics, notably the later set of cross-cutting dykes at Yuanjiang in the Ailao Shan (Zhang and Schärer, 1999, fig. 2c; Leloup et al., 1995, fig. 13a; Sassier et al., 2009, figs. 4, 5). The latter authors published 232Th-208Pb high-resolution ion microprobe ages that accurately constrain the emplacement age of the three sets of leucogranites. Ages of the three groups were ca. 22.5 Ma for the later cross-cutting dykes, 26 Ma and 30 Ma for the earlier more deformed dykes parallel to the matrix fabric. The late cross-cutting dykes at Yuanjiang are apparently the same as the sample (YU-4a-00) dated by Gilley et al. (2003, p. 14–10) at 21.7 ± 0.2 Ma. This suggests that the ductile shear fabric that is dominant along the Ailao Shan must have been older than 21.7 ± 0.2 Ma. We subsequently visited this site and others around Yuanjiang in 2007 and confirm the structural succession as published by Sassier et al. (2009). During our fieldwork we also found numerous examples of folded and boudinaged fabrics in leucogranites within gneisses. The critical observation that both amphibolite layers and leucogranite layers show similar boudinage textures suggest that high-temperature ductile strain was imposed in the solid state after both metamorphism and leucogranite intrusion. The only exception seems to be the youngest set of cross-cutting leucogranite dykes in the Yuanjiang section (Figs. 8, 9).

Field structural mapping can be used to construct a relative chronology of metamorphism, granite intrusion and deformation along the RRSZ in the DCS and ALS. The age of regional metamorphism has not been constrained yet by U-Th-Pb dating in this area, but our U-Pb zircon data provide compelling evidence of a Triassic Indosinian metamorphic event in the protolith (D1). In Vietnam, there is geochronological evidence of high-temperature granulite facies events during the Ordovician and the Permo-Triassic (Roger et al., 2007). A widespread regional Permo-Triassic Indosinian metamorphism was caused by the collision of the Sibumasu-Qiangtang terrane with the South China terrane (Carter et al., 2001; Maluski et al., 2001). High-temperature and ultra-high temperature granulites were formed at 250–240 Ma (U-Pb ages, Roger et al., 2007; Sm-Nd ages, Nakano et al., 2007) in the Kontum massif, central Vietnam, along with both garnet granites and orthopyroxene–bearing granites (Owada et al., 2007).

In the RRSZ, K-feldspar augen gneisses were intruded into the host orthogneiss and both underwent regional metamorphism during which a regional metamorphic schistosity was formed (D2). Th-Pb ages from matrix monazites from the Ailao Shan range from 74 to 20 Ma, and from the DNCV in North Vietnam from ca. 220–44 Ma (Gilley et al., 2003). The wide range of Th-Pb monazite ages makes interpretation very difficult. Some old ages could be recording a mixture of incompletely reset detrital monazites; some of the oldest could actually be recording an Indosinian metamorphic event. Clearly these ages cannot be related to either shear heating along the RRSZ (Leloup and Kienast, 1993) or to strike-slip shearing along the fault (Leloup et al., 1995, 2001, 2007). Gilley et al. (2003) showed by time-dependent conductive thermal modeling that the heat supply through strike-slip shearing would not have been sufficient to raise temperatures up to those required for high-grade metamorphism or melting.

The metamorphic fabric, both in the DCS and the ALS was clearly present prior to the intrusion of biotite granodiorites and later set 1 biotite ± garnet ± tourmaline leucogranite dykes. Host gneisses and the granite intrusive rocks were then folded together by tight to isoclinal folding (D1-D2). All these rocks were then subjected to high-strain during left-lateral strike slip motion along the RRSZ (D3). Sub-horizontal stretching lineations parallel to the fault (120–140°) were superimposed on all rocks within the shear zone. The final magmatic phase was a set of very small, thin leucogranite dykes and veins that crosscut the mylonitic fabric (set 3 dykes in Figure 9a, b). Whereas all previous granites are all pre-kinematic, these dykes and veins are syn-to post-kinematic with respect to the ductile strike-slip shear fabrics. The final deformation involved brittle strike-slip faults at high structural levels and the range-bounding normal faults (D4) that accommodated the final few km of exhumation of the DCS and AS metamorphic rocks.

Searle (2006, Fig. 5) reviewed all the geochronological data from rocks along the RRSZ in Yunnan and North Vietnam. He concluded that strike-slip shearing occurred after 21 Ma, the U-Pb age of the youngest leucogranite showing post-magmatic, high-temperature (∼550 °C) strike-slip shear fabrics. Subsequently Leloup et al. (2007) and Sassier et al. (2009) reported new U-Th-Pb ages from the Yuangjiang site (Fig. 8). The earlier, more deformed dykes (set 1) are 29.9 ± 0.5 Ma, the intermediate dykes (set 2) are 26.8 ± 0.7 - 24.2 ± 0.4 Ma, and the late set 3 cross-cutting leucogranite dykes are 22.5 ± 0.2 Ma (Sassier et al., 2009). Using all the U-Th-Pb age data (Schärer et al., 1990, 1994; Zhang and Schärer, 1999; Gilley et al., 2003), as well as the new Sassier et al. (2009) ages, we interpret the timing of ductile shearing along the RRSZ in the Ailao Shan to be after 31.9 (age of early set 1 deformed granites) and before 21.7 ± 0.2 Ma, the age of the set 3 cross-cutting dykes at this locality. The revised time chart for the RRSZ (Fig. 13) now shows that the ductile left-lateral shearing along the RRSZ in the Ailao Shan occurred within a very narrow span during the latest Oligocene – earliest Miocene, and was followed by brittle faulting. The timing of ending of brittle faulting is difficult to constrain in the DCS and AS, but in the Gulf of Tonkin, offshore North Vietnam, seismic profiles show that Red River related strike-slip faults are truncated by a prominent flat-lying unconformity dated at 5.5 Ma (Rangin et al., 1995).

40Ar/39Ar dating of micas and fission track dating from apatite and zircon provide information on cooling histories of exhumed metamorphic and magmatic rocks. 40Ar/39Ar ages only record a point on, or a small part of the cooling curve; they do not give any information on timing of peak metamorphism, melting or strike-slip shearing. 40Ar/39Ar ages from the DayNuiConVoi gneisses along the RRSZ in Vietnam record a rapid cooling event at 27–21 Ma (P-L. Wang et al., 1998, 2000) immediately following leucogranite melting ages (Schärer et al., 1990, 1994; Zhang and Schärer, 1999). The RRSZ mylonites in Vietnam have fission track ages from 30 ± 4–20 ± 3 Ma (Viola and Anczkiewicz, 2008). 40Ar/39Ar ages from the Ailao Shan range from ∼22–19 Ma, and ages from the Diancang Shan gneisses range from ∼23–7 Ma (Leloup et al., 1993, 1995; Harrison et al., 1996). These authors reported a transition of 40Ar/39Ar cooling ages along the RRSZ from 17 Ma in the NW part of the Ailao Shan to 25 Ma in Vietnam and proposed that the RRSZ propagated from SE to NW at a rate of 4.5 cm/a (“zipper tectonics”). However, if the RRSZ was a result of the indentation of the Indian plate into Asia as required by the block extrusion model (Tapponnier et al., 1982. 1986, 1990; Leloup et al., 1993, 1995, 2001) the first affects, and hence the oldest ages, should be recorded nearer to the Indian indenter, not farthest away from it. The “zipper” model is in direct contradiction to the 40Ar/39Ar and fission track data.

There is abundant evidence that major topography existed in the Tibetan Plateau region during the late Oligocene–early Miocene, if not even older. Apatite and zircon fission track thermochronology from sands along the Red River show accelerated cooling in the source regions since 25 Ma (Clift et al., 2006). Major Miocene incision in the upper Red River drainage basin was probably 13–9 Ma in Yunnan (Clark et al., 2004). A later, important Pliocene surface uplift is also recorded by (U-Th)-He low-temperature thermochronology (Schoenbohm et al., 2006). Clift et al. (2006) and Clift and Sun (2006) showed that the volume of sediment in the Yinggehai-Song Hong basin offshore Vietnam is much greater than the volume of rock removed from the modern river basin, supporting evidence that the ancient Red River may have included the upper Mekong and upper Yangtze rivers, before their capture ca 24 Ma. The Yinggehai-Song Hong basin in the Gulf of Tonkin was part of the rifted South China Sea margin offshore North Vietnam, but records renewed rifting and subsidence as a result of transtension along the Red River fault. This basin contains 17 km of sediment eroded from the southeastern margin of the Tibetan Plateau and the Red River hinterland. Multichannel seismic reflection data and oil well data show that the basin started to open ca. 45 Ma and more rapidly at ca. 35 M, but began to invert diachronously between 21 and 14 Ma (Clift and Sun, 2006). The timing of the rapid basin deepening corresponds roughly to the timing of initiation of ocean floor spreading in the South China Sea (32–17 Ma; Briais et al., 1993, or 30–16 Ma; Cande and Kent, 1992). Ocean floor spreading accelerated to 7.3 cm/yr at 25 Ma, and ended at 20.5 Ma (Zhu et al., 2009). The timing of the basin inversion (21–14 Ma) corresponds to our proposed timing of cessation of ductile left-lateral shearing and exhumation along the RRSZ but before the more recent brittle right-lateral strike-slip faulting (Burchfiel et al. 2008).

Our model for the RRSZ in Yunnan is based on field structural evidence and timing constraints as described above. The geometry of the RRSZ gneisses in the Diancang Shan (Fig. 14a) and the Ailao Shan (Fig. 14b) do not conform to models of classical strike-slip fault zones. The DCS is bounded along both NE and SW margins by outward-dipping normal faults suggesting that the width of the DCS gneisses increases with depth. Foliation planes within the internal and northern part of the DCS and the DNCV in Vietnam are sub-horizontal and cut by steeper left-lateral shear zones along the margins. Stretching lineations are mainly horizontal and strain markers are consistently left-lateral. Metamorphic rocks are present outside and adjacent to the DCS massif (Dali marbles) and the DNCV (Sapa marbles to the SW and Yen Bai marbles to the NE) are unrelated to strike-slip shearing and are possibly of Indosinian age.

The RRSZ in the Ailao Shan (Fig. 14b) also has a Range Front fault along the NE margin of the Ailao Shan that dips NE, away from the massif. The precise angle of dip is difficult to determine because the fault is poorly exposed along the forested slopes of the Ailao Shan and the Red River, but the true dips are likely to be between 40 and 70°NE. The Ailao Shan fault along the center of the massif (Leloup et al., 1995) separates low-grade mica schists to the SW from the high-grade amphibolites, orthogneisses, augen gneisses, and mylonites of the RRSZ to the NE. Foliations in the low-grade schists have dips less than 45° and strike-slip lineations are much less common, so we do not include these rocks in the RRSZ sensu stricto. Uncommon serpentinized harzburgites occur close to the shear zone in the NW part of the Ailao Shan, but to the SE around Mojiang, they are actually within low-grade schists 5–15 km SW from the Red River mylonite zone. The fault does not follow an old suture zone, but rather cuts obliquely across different geological zones. Since we propose that the mantle-sourced granodiorites and early granites were intruded prior to strike-slip ductile shearing, there is now no evidence that the RRSZ tapped down into the mantle as proposed by Tapponnier et al. (1990) and Leloup et al. (1995). It is also clear that other Oligocene alkali granite–syenite intrusions, both in Yunnan and North Vietnam (e.g., FanSiPan granite), were intruded prior to strike-slip shear (Chung et al., 1997, 2008; Searle, 2006) and not during strike-slip shearing (Leloup et al., 1995; Zhang and Schärer, 1999; Liang et al., 2007). Undeformed syenites–alkali granites on Mount FanSiPan 10 km south of the RRSZ in North Vietnam have been cut through by a few discrete left-lateral shears with minor offsets, but the granite and the metamorphic rocks around Sapa are not formed by shear heating along the RRSZ (Jolivet et al., 2001; Searle, 2006, 2007). The K-rich magmas are not restricted to the RRSZ, and were part of a regional high-K magmatic event that lasted from 40 to 30 Ma, and possibly up to 20 Ma (Chung et al., 1997, 2005, 2008; Guo et al., 2004).

We suggest that the RRSZ gneisses were not formed from shear heating during strike-slip faulting, but are uplifted basement rocks, possibly of Indosinian age, with an Oligocene–Early Miocene high-temperature overprint event that resulted in localized partial melting to form leucogranite dykes. In Yunnan, the Dali marbles outside of the RRSZ show that metamorphism is not restricted to the shear zone. In Vietnam, high-grade marbles around Sapa to the SW of the RRSZ and around Luc Yen to the NE of the RRSZ also show regional metamorphism outside the RRSZ. Miocene leucogranite dykes also occur outside of the RRSZ (Anczkiewicz et al., 2007), suggesting that their origin was not related to the strike-slip fault. Ubiquitous left-lateral strike-slip shear fabrics in RRSZ gneisses are post-metamorphic, post-sillimanite growth (Jolivet et al., 2001; Searle, 2006; Anczkiewicz et al., 2007; Yeh et al., 2008). However, transpressional motion along the Red River fault was related to their uplift and exhumation. Along the NE flank of the Ailao Shan and Diancang Shan, the metamorphic fabrics are truncated by the later steep NE-dipping range-bounding brittle normal fault. Harrison et al. (1992, 1996) proposed that exhumation was related to transtension and propagated from SE to NW with time. However, extension and transtension will not exhume any rocks without a push from below. It is far more likely that early transpression caused the upward motion of the RRSZ rocks and this was synchronous, or followed by, Miocene extensional faulting along the flanks of the metamorphic massifs (Range Front fault). If 40Ar/39Ar mica ages are used as a rough proxy for timing of cooling through the ductile-brittle transition (∼350°C), then this transition and rapid cooling occurred between ca. 25–22 Ma in the Ailao Shan (Leloup et al., 2001).

Finally, although there are numerous active faults in the Yunnan region (e.g.: E. Wang et al., 1998; Burchfiel and Wang, 2002; Burchfiel et al., 2007, 2008; Schoenbohm et al., 2009), there is no apparent active faulting linking the three metamorphic segments in the DCS, ALS, and DNCV complex, Vietnam. Although numerous authors cite evidence for Quaternary or Recent motion along the RRF with some displaced stream channels crossing the fault, it is difficult to see how it could be active today, given the seismic quiescence and the fact that the GPS velocity field cuts across the strike of the fault at right angles (Shen et al., 2005).

We suggest that the metamorphism along both the DCS and AS was not formed during the Tertiary by shear heating along the RRSZ, but occurred mainly prior to strike-slip shearing, similar to the DNCV complex in Vietnam (Jolivet et al., 2001; Searle, 2006, 2007; Anchiewicz et al., 2007; Yeh et al., 2008). U-Pb zircon geochronology suggests that the RRSZ gneisses were metamorphosed during the Triassic Indosinian event but have an Oligocene thermal overprint prior to intrusion of the Late Oligocene–Early Miocene leucogranite dykes. Metamorphic rocks have been exhumed along the RRSZ but are not restricted to the shear zone; similar metamorphic rocks occur away from the shear zone in Yunnan (e.g., Dali marbles) and in Vietnam (e.g., Song Chai dome, Sapa, and Luc Yen marbles, etc.). Metamorphism may have reached hornblende melting temperatures (ca 900 °C) resulting in in situ tonalitic partial melts in orthogneiss migmatites. Maximum P-T conditions were reached prior to strike-slip shearing, and the source of heat was not from shear heating along the fault. Transpression during early Miocene left-lateral strike-slip shearing exhumed deep crustal metamorphic rocks and superimposed strike-slip ductile mylonite fabrics onto earlier metamorphic fabrics.

Strong, steeply inclined, left-lateral mylonite fabrics with very strong NW-SE aligned stretching lineations (D5) were superimposed on all earlier metamorphic fabrics (D1-D4) by large-scale, left-lateral strike-slip shearing along the RRSZ. The ALS-DCS metamorphic massifs form NE-dipping (45°-vertical) ductile shear zones, thickening toward deeper structural levels. They do not form classic strike-slip “flower structures,” with outward verging thrusts; they are more like an exhumed deep-crust SW-verging slab placing high-grade gneisses over low-grade schists, subsequently overprinted and cut by NW-SE strike-slip shearing fabrics relating to left-lateral shear along the RRSZ. The overall structure resembles an elongate gneiss dome similar to those in metamorphic core complexes, although the origin of the DCS and ALS is not the same as Cordilleran-type metamorphic core complexes.

Early granodioritic melts were mantle-derived, whereas later leucogranitic melts containing Bt ± Grt ± Tur were derived from crustal melting. All these granites (except for a few very small set three cross-cutting veins) were formed prior to mylonite fabric formation and shearing along the RRSZ. U-Th-Pb ages of K-feldspar augen gneisses, granodiorites, and many leucogranites will only give a maximum age of initiation of strike-slip shearing along the RRSZ. U-Th-Pb ages from the few, very small leucogranite dykes that cross-cut the left-lateral strike-slip shear fabrics will give a minimum age of left-lateral ductile shearing along the RRSZ. Ductile shearing along the RRSZ occurred after Oligocene metamorphic zircon rim growth and post-early folded leucogranite intrusions (31.9 – 24.2 Ma), but prior to the later cross-cutting small dykes and veins (21.7 Ma). Brittle faulting, mainly transtensional normal faulting along the flanks of the AS and DCS metamorphic massifs, occurred after this time.

U-Pb ages of the alkali granites must be pre-strike slip shearing, although the 40Ar/39Ar ages may well relate to transpressional exhumation during the Miocene. Leloup et al. (1995, 2001) interpreted the high temperature shear as coeval with felsic magmatism that lasted from 33 to 22 Ma (Schärer et al. 1990, 1994; Zhang and Schärer, 1999). However, if the shear fabrics were superimposed onto earlier granite as we suggest, then the timing of fault initiation must be post-25–24 Ma.

The precise age of metamorphism is not yet known for certain, and RRSZ gneisses may well show multiple periods of metamorphism. Triassic metamorphism is known from North Vietnam (244 ± 7 Ma; Carter et al., 2001) as well as Early Miocene melting (26.0–23.7 ± 1.7 Ma; Nagy et al., 2000) in the Bu Kang dome, NE of the RRSZ. Gilley et al. (2003) dated monazites from the Xuelong Shan, Diancang Shan and Ailao Shan by in situ Th-Pb ion microprobe dating, with ages ranging from 34 – 21 Ma. Matrix monazite ages from the DNCV complex are far more complicated with ages ranging from 208–21 Ma. Monazite inclusion ages from DNCV gneisses at Bao Yen in North Vietnam were 117 ± 2 Ma and 85.1 ± 1.1 Ma (Gilley et al., 2003), clearly long before initiation of the RRSZ. These authors, however, interpreted the old ages as representing inheritance. Clearly much more high-precision U-Th-Pb dating of metamorphic and magmatic rocks that are structurally well constrained is now required along the RRSZ.

MPS thanks NERC grant NER/K/S/2000/951 for funding fieldwork in Vietnam, and the Royal Society for funding fieldwork in Yunnan. M.-W. Yeh and T.-H.L. and S.-L.C. thank the National Science Council, Taiwan, for financial support for fieldwork. We also thank Y. Iizuka at IES and H.-Y. Chiu at NTU for their help with CL-imaging and LA-ICPMS experimentation. We are grateful to B.C. Burchfiel for discussions, P.H. Leloup for extremely detailed comments on an earlier version of the manuscript, and P. Kapp and an anonymous person for insightful reviews.