Plate movements, ductile deformation and geochronology of the Sanbagawa belt, SW Japan: tectonic significance of Ma Lu Hf eclogite ages

Transcription

1 J. metamorphic Geol., 2009, 27, doi: /j x Plate movements, ductile deformation and geochronology of the Sanbagawa belt, SW Japan: tectonic significance of Ma Lu Hf eclogite ages S. R. WALLIS, 1 R. ANCZKIEWICZ, 2,3 S. ENDO, 1 M. AOYA, 4 J. P. PLATT, 5 M. THIRLWALL 3 AND T. HIRATA 6 1 Department of Earth & Planetary Science, Graduate School of Environmental Studies, Nagoya University, Nagoya , Japan (swallis@eps.nagoya-u.ac.jp) 2 Institute of Geological Sciences, Polish Academy of Sciences, Kraków Research Centre ul. Senacka 1, Kraków, Poland 3 Department of Geology, Royal Holloway University of London, Egham, Surrey TW20 OEX, UK 4 Institute of Geology and Geoinformation, National Institute of Advanced Industrial Science and Technology (AIST), Central 7, Higashi 1-1-1, Tsukuba , Japan 5 Department of Earth Sciences, University of Southern California, Los Angeles, CA , USA 6 Department of Earth & Planetary Sciences, Tokyo Institute of Technology, Tokyo , Japan ABSTRACT Prograde P T paths and thermal modelling suggest metamorphism in the Sanbagawa belt represents unusually warm conditions for subduction-type metamorphic belts, and these likely reflect conditions of a convergent margin a few million years before the arrival of an active spreading ridge. Radiometric age data and kinematic indicators of ductile deformation suggest the Sanbagawa belt formed in a Cretaceous convergent margin associated with a plate movement vector that had a large sinistral oblique component with respect to the belt, the East Asian margin. Plate reconstructions for the Cretaceous to Tertiary for this region show that the only plausible plate compatible with such motion at this time is the Izanagi plate. These reconstructions also show that progressively younger sections of the Izanagi plate were subducted beneath eastern Asia, i.e. a spreading ridge approached, until Ma when the Izanagi Plate ceased to exist as an independent plate. The major reorganization of plates and associated movements around this time is likely to be the age of major interaction between the ridge and convergent margin. The ridge-approach model for the Sanbagawa metamorphism, therefore, predicts that peak metamorphism is a few million years older than this age range. New Lu Hf dating of eclogite in the Sanbagawa belt gives ages of Ma, in excellent agreement with the prediction. Combining this estimate for the peak age of metamorphism with published P T-t results implies vertical exhumation rates of greater than 2.5 cm yr )1. This high rate of exhumation can explain the lack of a significant thermal overprint in the Sanbagawa belt during subduction of the ridge. Key words: Lu Hf dating; eclogite; Sanbagawa belt; ridge subduction. INTRODUCTION High P T belts form along convergent margins and their formation conditions are an important source of information about the thermal structure of subduction zones down to depths in excess of 100 km. Thermal modelling has shown that the establishment of low thermal gradient conditions is a result of the inflow of cold lithosphere. However, the peak conditions recorded by the rocks in many such metamorphic belts record much higher temperatures for a given depth than predicted by steady-state thermal models for average subduction zones (Peacock, 1987; Iwamori, 2000; Aoya et al., 2003; Uehara & Aoya, 2005). In continental collision zones, some of the differences may be due to post-collisional heating related to the increased heat input from radioactive decay caused by crustal thickening. However, the same discrepancy is also seen in regions of oceanic subduction where no collision has taken place and there are not significant concentrations of heat producing radioactive elements. The development of hot subduction zones may be due to shear heating along the subduction boundary (Molnar & England, 1990). However, this explanation requires much higher shear stresses than those usually estimated (Peacock, 1996). Another proposed explanation for the relatively high temperatures recorded in subduction type metamorphic belts is that they record transient hot conditions shortly after the onset of subduction, before the subduction zone cooled and reached conditions close to thermal equilibrium (Peacock, 1987; Wakabayashi, 1990). Such cases are characterized by exhumation along cooler P T paths, i.e. with a higher P T gradient, than the subduction part. 93

2 94 S. R. WALLIS ET AL. These anticlockwise P T paths reflect cooling of the subduction zone after peak pressure was attained. The third type of tectonic model that can account for unusually hot subduction zones is the inflow of hot relatively young oceanic lithosphere. A variation of this idea incorporates progressive younging of the plate, and approach of a spreading ridge (Uehara & Aoya, 2005). The Sanbagawa belt of south west Japan (Fig. 1) is a well-characterized example of an oceanic subduction type metamorphic belt (Isozaki & Itaya, 1990; Takasu et al., 1994; Wallis, 1998) that generally shows much higher temperature metamorphism for a given depth than predicted by modelling of average subduction zones (e.g. Iwamori, 2000; Aoya et al., 2003). Aoya et al. (2003) used the results of thermal modelling to suggest that the high temperatures are best explained as the conditions immediately before arrival of a spreading ridge. Two features of the P T paths support this proposal. First, most of the belt is characterized by P T paths that are either similar during subduction and exhumation (e.g. Wallis & Aoya, 2000) or show decompression paths with progressively higher thermal gradients, i.e. lower P T conditions (e.g. reviews in Aoya, 2001; Okamoto & Toriumi, 2005; Wallis, 1998). These characteristics are not compatible with formation in a cooling environment shortly after inception of subduction. We note, however, that more complex P T paths have been proposed locally for the highest grade part of the belt (Takasu, 1989; Ota et al., 2004; Mizukami & Wallis, 2005), suggesting different stages in the subduction history are also locally preserved. Second, prograde subduction P T paths of the Sanbagawa belt show a strongly curved overall P T path with decreasing thermal gradients with increasing depth of subduction (Enami et al., 1994; Enami, 1998; Inui & Toriumi, 2002; Aoya et al., 2003). Uehara & Aoya (2005) used thermal modelling to show this type of curved P T path is best accounted for by a model where there is progressively younger oceanic plate introduced into the subduction zone, and approach of a spreading ridge. Such a model is also compatible with the increase in thermal gradient experienced during exhumation. The tectonic model that the Sanbagawa metamorphism is associated with the approach of a spreading ridge was proposed primarily to account for the observed P T paths. But tectonic models also have implications for deformation histories and, when movements of individual plates are taken into consideration, for the timing of events. In this contribution, these types of data are used to test the proposed tectonic model. First, we re-examine plate reconstructions for the western Pacific realm incorporating on-land deformational characteristics to show ridge approach is likely at the time when the Sanbagawa metamorphism occurred and to constrain the timing of this event. Second, garnet growth ages of the Sanbagawa metamorphism are determined using high precision Lu Hf dating of eclogite. The results are in good agreement with model predictions and offer strong support for the ridge approach model. In addition, combining our new dates with published geochronological data shows that the time-scale for the main part of the Sanbagawa metamorphism was only a few million years. This short time-scale shows that the Sanbagawa metamorphism reflects only a short relatively hot period in the evolution of a subduction zone and it is not representative of more usual colder conditions of subduction zones close to thermal steady state. WESTERN PACIFIC CRETACEOUS PLATE RECONSTRUCTIONS Kinematics of deformation of the Sanbagawa belt and plate motion Cretaceous oceanic plate motion vectors can be estimated with some precision from oceanic magnetic anomalies and bathymetry (e.g. Engebretson et al., 1985). However, the position of former plate boundaries that have been subducted away is subject to large uncertainties. On-land deformation kinematics from former convergent plate boundary zones can help constrain the motion direction of the subducting plate and hence identify the subducting plate. The correlation between plate motion direction and kinematics of deformation in convergent margins is complex (e.g. McCaffrey, 1992). Nevertheless, two useful general statements can be made: (1) highly oblique convergence will be associated with strong strike-parallel motion along the length of the orogen, and (2) the sense of obliquity will be reflected in the overall sense of shear recorded in the rocks (Platt, 2000). In both cases, it is important to examine deformation on a large scale, because there are likely to be localized complications because of deformation partitioning. Polyphase ductile deformation has been documented in the Sanbagawa belt by numerous workers (Faure, 1985; Wallis, 1990, 1998; Hara et al., 1992). Recent investigations have documented a series of wellpreserved deformation stages in the high-grade units, such as those found in the eclogite-bearing Besshi area of central Shikoku (Aoya, 2002; Mizukami & Wallis, 2005). These are the oldest tectonic fabrics in the belt. In these high-grade regions, the direction of the dominant stretching direction is variable. Outside of the eclogite-bearing regions, the kinematics of deformation is more consistent with a stretching direction oriented oblique to the orogen (Faure, 1985; Toriumi & Noda, 1986; Hara et al., 1992; Wallis et al., 1992; Abe et al., 2001; Takeshita & Yagi, 2004). This deformation occurred after the growth of albite porphyroblasts and the resulting microstructural relationships assist in correlating deformation phases on a regional scale. We refer to this dominant deformation of the Sanbagawa belt as Ds following earlier work (e.g. Wallis, 1998; Aoya, 2002). This is broadly

3 89 88 MA LU - HF AGES OF SANBAGAWA ECLOGITE 95 equivalent to the D1 defined by other workers (e.g. Faure, 1985; Takeshita & Yagi, 2004). Microstructural studies show that Ds starts close to the peak of temperature and reflects the main exhumation stage for the non-eclogitic part of the metamorphic belt (Wallis et al., 1992). Ds deformation developed during exhumation and is, therefore, not directly related to plate motion. Here, it is assumed that even during exhumation, deformation will be influenced by the direction of plate motion. In particular, we use the sense of obliquity with respect to the orogen and the sense of shear to place first-order constraints on the direction of plate motion. The Ds stretching lineation data are shown in Fig. 2. The stretching direction is generally rotated clockwise from the length of the orogen by about 30 (Fig. 2) and is generally oriented from W-E to NW-SE (Faure, 1985; Toriumi & Noda, 1986; Wallis, 1990, 1995; Hara et al., 1992). In the Kanto mountains region, in the eastern part of the belt, the Ds stretching direction is oriented roughly WNW roughly parallel to the trend of the Sanbagawa belt (Fig. 2). This can be explained as due to localized effects of the Miocene collision between the Japanese mainland and the Izu-Ogasawara Arc that caused localized flexure of the structure of SW Japan (Takahashi & Saito, 1997). In addition to stretching direction, many workers have also documented the sense of shear associated with Ds (Wallis, 1990; Wallis et al., 1992; Abe et al., 2001; Yagi & Takeshita, 2002; Takeshita & Yagi, 2004). The results are consistently top to the western quadrant with a sinistral component of deformation with respect to the margin. Localized zones of top to the east deformation are also recognized in the Sanbagawa belt; available kinematic data suggest that this is the dominant sense of shear in the low-grade region including the Mikabu domain to the south of the Sanbagawa belt (Wallis, 1995). This may represent either localized domains of deformation partitioning (Wallis, 1995) or, alternatively, a phase of deformation that post dates Ds deformation. The obliquity of the stretching direction with respect to the length of the orogen and the overall sinistral sense of shear are recorded in the rock record throughout the 800 km length of the Sanabagawa belt. The scale over which these observations have been made implies they reflect large-scale tectonic processes. The above results can, therefore, help place constraints on the motion of the plate that was subducting when the Sanbagawa subduction zone was active. However, before comparing these data with the plate reconstructions, we first need to take into consideration the Miocene rigid-body rotation of the Japanese islands associated with the opening of the Japan Sea. Palaeomagneitc studies have shown that the rotation of Japan can be described by a rigid body rotation by an angle of (Otofuji & Matsuda, 1987). Geochronological work constrains this rotation to have taken place since 20 Ma (Otofuji & Matsuda, 1987). Restoring the pre-japan Sea configuration results in a roughly NNE-SSW orientation of the Sanbagawa belt. The obliquity of the stretching direction with respect to the orogen suggests oblique plate convergence with respect to this margin. In addition, the sinistral sense of shear with respect to the margin implies movement of the plate from the south to the north (Fig. 1a) in the restored reference frame. Movement of the Izanagi-Pacific ridge The dominant ductile deformation of the Sanabagawa belt, Ds, occurred during exhumation and cooling (e.g. Wallis et al., 1992). Cooling ages reported in the Sanbagawa belt are all Cretaceous (e.g. Itaya & Fukui, 1994; Takasu et al., 1996; Wallis et al., 2004a) and discussion of plate reconstructions can be restricted to this period. The plates to consider are then: the Farallon, Pacific, Izanagi and Kula plates (Engebretson et al., 1985). Enough of the Cretaceous parts of these plates remain intact for the appropriate motion vectors to be reconstructed. Using the stage poles of Engebretson et al. (1985), the motion vectors of these plates were calculated with respect to the Sanbagawa Belt of southwest Japan rotated to its original position (Fig. 1b). This clearly shows that there is only one Cretaceous plate with a motion vector compatible with the observed sinistral shear associated with Ds: the Izanagi Plate. After subduction of the Izanagi plate, the Kula plate is predicted to subduct beneath the Japanese margin this time with a dextral component of shear. A second important feature of these plate reconstructions is they predict the presence of a spreading ridge between the Izanagi Plate and the Pacific Plate to the south. This ridge is predicted to interact with the Japanese archipelago sometime in the late Cretaceous as has been pointed out by several previous workers (Uyeda & Miyashiro, 1974; Maruyama, 1997; Osozawa, 1997). The interaction of plate boundaries with subduction zones is likely to result in a major reorganization of plates including the formation of new plates and changes in the motion vector of preexisting plates. An examination of changes in the plate movements with time shows a major reorganization of plate motion in the Cretaceous in the period Ma (Fig. 1b). It is at this time that the new plate, the Kula plate, is formed. The remains of the Izanagi plate became fused to the Pacific Plate. Errors on the estimated timing of changes in plate motion are a few million years. In particular, the major reorganization of plate motion associated with the cessation of independent movement of the Izanagi plate and formation of the Kula plate took place somewhere between 85 and 80 Ma. For simplicity, in Fig. 1b, the reconstruction for the Farallon plate is given only for the region north of the Mendocino fracture zone.

4 96 S. R. WALLIS ET AL. (a) (b) Fig. 1. (a) Southwestern part of the Japanese archipelago and Asian continental margin before and after the opening of the Japan Sea. Areas for which stretching lineation data are reported in Fig. 2 are shown as follows: WS = Western Shikoku, CS = Central Shikoku, ES = Eastern Shikoku, and KM = Kanto Mountains. (b) Motion history of plates in western Pacific realm with respect to a point with present location 33.8 N E within the Sanbagawa belt in central Shikoku island constructed using the poles of Engebretson et al. (1985). The original location of the reference point prior to the opening of the Japan Sea was calculated using a pole of rotation 34 N 129 E and an anticlockwise rotation angle of with respect to eastern Asia following Otofuji & Matsuda (1987). Comparison with results of Irving (1977) suggests that rotation angles up to 60 of southwest Japan with respect to central Asia are possible. This would increase the angle h recorded on Fig. 1b, but have no significant affect on the conclusions presented here. Errors on the estimated timing of changes in plate motion are a few million years. In particular, the major reorganization of plate motion associated with the cessation of independent movement of the Izanagi plate and formation of the Kula plate took place somewhere between 85 and 80 Ma. For simplicity the reconstruction for the Farallon plate is given only for the region north of the Mendocino fracture zone. Ridge subduction changes the boundary conditions for plate movement and is one of the major causes of major reorganizations of plate motion vectors such as that seen in the Cretaceous in the Pacific realm. These results suggest that the Izanagi-Pacific ridge arrived at the Japanese convergent margin in the period between 85 and 80 Ma. This interaction is likely to have been diachronous along the Japanese arc, but the major changes in plate movement direction are complete within a few million years suggesting that any trench-parallel variations in the timing of interaction with the ridge will be of the same order. If, as we suggest, the Sanbagawa metamorphism formed shortly before the arrival of this active spreading ridge, the peak of the Sanbagawa metamorphism should be a few million years older than the Ma age range described above. This is a restrictive prediction of the model that can be tested with accurate geochronology.

5 89 88 MA LU - HF AGES OF SANBAGAWA ECLOGITE 97 Fig. 2. Stretching direction and orientation of the Sanbagawa belt. Data are a compilation of our own observations and data from Toriumi & Noda (1986), Shimizu (1988), and Muta(1999). Equal area lower hemisphere plots. The general areas where the data were collected are shown in Fig. 1. The data for the Kanto Mountains Area are in agreement with the results of Abe et al. (2001), however, these results are not incorporated here, because this publication does not distinguish between stretching and crenulation lineations and these have very different kinematic significance. AGEOFECLOGITE Previous studies Geochronological studies in the Sanbagawa belt have defined the cooling history (Itaya & Takasugi, 1988; Takasu & Dallmeyer, 1990; Wallis et al., 2004a), however, the age of peak metamorphism is unclear. U-Pb zircon SHRIMP ages of Ma are proposed as the age of eclogite metamorphism (Okamoto et al., 2004), but no eclogite facies inclusions were identified and the link between the obtained age and metamorphism is unclear. Zircon may grow over a range of pressures and temperatures and linking U-Pb ages to metamorphism is ambiguous unless chemical and textural constraints can be used to directly link growth to mineral reactions in the rock. Dating minerals that characteristically form under eclogite facies conditions provides a clear link between the resulting age and metamorphism. One widely employed method is Sm-Nd dating of garnet and omphacite. However, attempts to apply this method in the Sanbagawa belt have been unsuccessful. This is largely due to very high abundance of Nd-rich titanite and epidote inclusions, which dominate the Sm-Nd budget and considerably reduce the spread in isotopic ratios (Anczkiewicz et al., 2004) with the result that a meaningful isochron cannot be constructed. In this study, the first successful dating of eclogite from the Sanbagawa belt is reported using Lu Hf dating of garnet and omphacite. The advantage of using the Lu Hf method is the strong preference of garnet for Lu over Hf, which leads to fractionation and very high Lu Hf ratios. This allows high precision dating. Eclogite sample description For the present Lu Hf geochronology study, two eclogite samples were selected that are well characterized both in terms of their petrological history and structural geology and for which the metamorphic and structural relationships to the rest of the Sanbagawa belt are well established. The selected samples are from the Seba (ESB-45) and Kotsu (KKT-3) eclogite bodies (Fig. 3). Both samples have a well-developed schistosity and mineral lineation. The main constituent minerals are garnet, amphibole (barroisite in ESB-45 and glaucophane in KKT-3), omphacite, phengite, quartz, epidote, calcite, titanite and rutile. Small amounts of plagioclase are present in both samples but only as retrograde products (Aoya, 2001; Matsumoto et al., 2003). Details of mineral chemistry are given in Matsumoto et al. (2003) and Aoya (2001). Electron microprobe analyses and laser Raman spectroscopy were used to identify inclusions in garnet larger than 1 lm. In both ESB-45 and KKT-3, the garnet inclusions are similar: omphacite, epidote, titanite, calcite and quartz. Garnet of KKT-3 shows a change in Ti phase from titanite in the core to rutile in the rim. Chemical mapping clearly shows that garnet from both samples preserves growth zonation (Fig. 4). Petrological studies using Gt-Cpx thermometry (Aoya, 2001; Matsumoto et al., 2003) have established peak temperatures of metamorphism for the Seba and Kotsu areas of C and around 600 C, respectively. The peak pressures are less well defined, but in both areas the jadeite component in omphacite coexisting with quartz gives minimum constraints of around 13 kbar. In addition, both areas contain paragonite both as an inclusion and a matrix phase.

6 98 S. R. WALLIS ET AL. This suggests peak pressures did not exceed 25 kbar; the upper stability limit of paragonite at the estimated temperatures. Application of the garnet-omphacitephengite barometer to the Kotsu eclogite yields pressures around 20 kbar compatible with the constraint from the presence of paragonite (Matsumoto et al., 2003). In the Seba eclogite, estimates by Zaw et al. (2005) give a minimum estimate of 18 kbar. Both the Kotsu and Seba Eclogite bodies lie at high structural levels in the orogen and the Seba area preserves deformational fabrics that are older than those in the lower grade units (Aoya, 2002). A similar deformational history is likely for the Kotsu area (Matsumoto et al., 2003). Microstructural and petrological studies have shown that the non-eclogite units have a prograde history that corresponds to the (a) (b) (c) Fig. 3. (a) Location of the Seba and Kotsu eclogites. (b) Geological map of the Kotsu area. The location of the boundary to the Kotsu eclogite unit is after Matsumoto (2002). Geological map of the Seba area. The age data are from Dallmeyer & Takasu (1991).

7 89 88 MA LU - HF AGES OF SANBAGAWA ECLOGITE 99 exhumation stage of the eclogite units (Aoya, 2001; Mizukami & Wallis, 2005). Prograde garnet growth in the eclogite units predates all the significant metamorphic history of the main part of the lower grade Sanbagawa metamorphic belt. Our Lu Hf dates, therefore, give an upper limit on the formation age of not only the eclogite unit dated here, but also the underlying Sanbagawa belt, which comprises nearly all of this belt. Lu Hf isotopic analysis Garnet and omphacite were separated from the samples using conventional techniques of crushing, sieving, heavy liquids, magnetic separation, and handpicking. Zircon contamination is a potential problem with Lu Hf dating because of its high Hf content, but no grains of this mineral were observed in the present study. Two garnet separates and one clinopyroxene fraction from each sample were chosen for age determination. Sample dissolution and column chemistry were then performed following the protocol of Anczkiewicz & Thirlwall (2003). The analyses were carried out using MC ICPMS IsoProbe TM at Royal Holloway University of London. Mass spectrometric procedures are outlined in (Thirlwall & Anczkiewicz, 2004). A summary of isotopic results is presented in Table 1. Leaching was carried out on garnet, but this had no appreciable effect. This reflects the high purity of the samples used in this study. Lu Hf laser ablation analyses In addition to the above analyses, Lu and Hf contents were also measured in garnet using laser ablation ICPMS. The LA-ICPMS analysis was carried out at the Tokyo Institute of Technology using a VG PlasmaQuad2 Omega laser. Details of the analytical procedure given in Iizuka & Hirata (2004). Laser settings were kept at a 10 Hz repetition rate and a 30 lm spot size, 4 mj laser power and 10 s counting time. NIST 612 glass (Pearce et al., 1997) was used as an external standard. Peak height was used for data reduction. Estimated analytical errors were better than 15% (2RSD). Euhedral garnet grains cut close to the crystal centres were selected from both ESB-45 and KKT-3 for LA-ICPMS analysis (Fig. 4). The garnet grains are relatively clear and have few inclusions larger than 1 lm along the analysed profiles. Both garnets show similar trends: Lu is strongly concentrated in the cores, but shows a sharp decrease to around 1 ppm in the rim. The Hf content is close to the detection limit in many analyses, but also shows a higher concentration in the core. These steep concentration gradients suggest the distribution of Lu and Hf in the garnet has not been affected by diffusion in keeping with the generally inferred high closure temperature for the Lu Hf system in garnet (e.g. Anczkiewicz et al., 2007). RESULTS OF Lu Hf DATING Garnet from both ESB-45 and KKT-3 has high 176 Lu 177 Hf ratios, which group rather tightly between 6.2 and 6.9. In contrast, omphacite plots very closely to the initial 176 Hf 177 Hf value, which allowed two highly precise ages to be achieved. KKT-3 has a well-defined three point isochron of 88.2 ± 0.5 Ma (MSWD = 0.3) with an initial Hf ratio of (Fig. 5a). ESB-45 was treated in the same way. Unfortunately, one of the garnet fractions was lost during preparation and only two points were available for the isochron construction, which resulted in an age of 88.8 ± 0.6 Ma. Despite the fact that the latter age relies on only one garnet fraction, the two ages are concordant within their analytical error to very high precision and strongly support each other. Hence, we treat them both as robust time constraints. Additionally, both samples are very similar and most likely derived from the same source as suggested by their very similar initial 176 Hf 177 Hf isotopic ratios that are identical within error (Fig. 5). Loss of Lu or Hf by diffusion after the growth of garnet would affect the estimated age. The Lu content of omphacite is very low and the main function in the isochron for the analyses of this mineral is to establish the Hf isotopic ratio. The following observations show that diffusion of Lu and Hf within garnet is minimal and has had a negligible effect on the original elemental concentrations. First, mapping of major elements, such as Mn, reveals a well-defined growth zonation parallel to the crystal faces, which shows that diffusion has not significantly affected the distribution of these elements. Diffusion of both Lu and Hf in garnet is expected to be lower than these major elements (van Ormon et al., 2002) and, therefore, also negligible. Second, laser ablation ICP-MS analyses show strong concentration of Lu in the core of garnet in both samples in keeping with the expected distribution for growth zoning (Fig. 4e, f). A less clear concentration of Hf is also seen in sample KKT3 (Fig. 4f). Third, support for slow diffusion even at elevated temperatures is given the study of Anczkiewicz et al. (2007) who reported a natural example where diffusion of Lu and Hf is negligible even at metamorphic temperatures around 900 C. For these reasons, we interpret the obtained ages as reflecting prograde garnet growth close to the peak of metamorphism. DISCUSSION Time-scales of orogeny and preservation of high-p metamorphic belts Thermal modelling and comparison with P T paths in the eclogite and lower grade units suggest that the Sanbagawa metamorphism is related to the thermal conditions a few million years prior to the arrival of the ridge (Aoya et al., 2003; Uehara & Aoya, 2005). An

8 100 S. R. WALLIS ET AL. (a) (b) 0.5 mm 0.5 mm (c) (d) (e) 18 (f) 10 Concentration (ppm) Nd Sm Lu Hf Concentration (ppm) Nd Sm Lu Hf Distance from rim (mm) Distance from rim (mm) Fig. 4. (a), (b) Mn chemical zoning of garnet from the Seba ESB-45 (a) and Kotsu KKT-3 (b) eclogite samples. The chemical zonation is parallel to the present growth surfaces of the garnet and preserves a hexagonal shape throughout the grain implying that diffusion has not significantly altered the original growth zonation. (c) Backscattered electron image of garnet shown in (a) with the pits caused by laser ablation clearly visible. (d) Photomicrograph of garnet shown in (b) with pits caused by laser ablation. (e),(f ) Concentrations of Lu, Hf, Sm, and Nd in the two analysed garnet grains. The Sm content of garnet is very low. The peak shown in KKT-3 is from a spot where an inclusion of epidote was ablated.

9 89 88 MA LU - HF AGES OF SANBAGAWA ECLOGITE 101 Table 1. Summary of Lu Hf dating results. Sample Samp wt Lu ppm Hf ppm 176 Lu 177 Hf 176 Hf 177 Hf Age Hf(t) KKT-3 CPX ± ± GRT A ± 25 GRT B ± 30 ESB-45 CPX ± ± GRT B ± 27 Errors are 2SE and relate to the last significant digits. 176 Lu 177 Hf errors are 0.5%. Standard JMC475 yielded ± 28 (2 SD, n = 11) over the period of analyses. Daily variations in 176 Hf 177 Hf ratios were normalized to 176 Hf 177 Hf = Age errors are 95% confidence level and were propagated for standard reproducibility. Age calculations conducted using Isoplot 3.0 (Ludwig, 2001). Mass bias correction to 179 Hf 177 Hf = Decay constant k176lu = )11 yr )1 (Dalmasso et al., 1992; Scherer et al., 2001; Anczkiewicz et al., 2004). Values used for Hf(t) calculations: 176 Hf 177 Hf CHUR(0) = and 176 Lu 177 Hf CHUR(0) = (Blichert-Toft & Albarede, 1997). examination of the plate reconstructions for the area suggest that ridge subduction beneath Japan occurred in the interval Ma. The results of the Lu Hf dating give estimates of eclogite formation at Ma in very good accord with the model prediction. A major difficulty with the ideas that the formation and exhumation of high-pressure metamorphic belts are related to the approach of a spreading ridge is that they do not commonly show overprinting by high temperature metamorphism as would be expected if a ridge passed directly beneath. This apparent contradiction may be explained if the duration of orogenesis including exhumation to shallow levels is short (a few million years); in this circumstance the metamorphic belt will have exhumed and cooled sufficiently to be largely unaffected by the subsequent subduction of the ridge. Combining our new results for the peak of metamorphism with already published geochronological studies can reveal the time-scale of orogeny in the Sanbagawa belt. There is considerable information available on the cooling history of the Sanbagawa belt in adjacent regions to the Seba area: phengite Ar Ar and K-Ar (90 80 Ma) (Itaya & Takasugi, 1988; Takasu & Dallmeyer, 1990), and zircon fission track dating (69 57 Ma) (Wallis et al., 2004b). Erosion of the Sanbagawa belt had probably begun by 50 Ma (Narita et al., 1999). These age constraints are shown in Fig. 4. In addition, exhumation P T paths of rocks from these areas have also been well-constrained (Enami et al., 1994; Aoya, 2001; Okamoto & Toriumi, 2005; Zaw et al., 2005). The ranges of these age estimates superimposed on the P T histories are shown on Fig. 4. The cooling age data for the Seba area can be combined with appropriate closure temperatures to date one part of the P T path. Two Ar Ar ages from the eclogitic schist and a shear zone in the Seba area (Dallmeyer & Takasu, 1991) the same unit as the one from which the Lu Hf sample was collected give phengite plateau ages of Ma, including 2r interlaboratory errors (Fig. 6). The Ar isotope analyses of amphibole from the same eclogitic schists (Dallmeyer & Takasu, 1991) have irregular age spectra and there is a large discrepancy between the best plateau ages of Ma and the isotope correlation ages of around 83.5 Ma. These features suggest that the results are not reliable metamorphic ages and they are not used here. The Lu Hf dates represent the age of garnet growth, which is closely related to peak pressures in excess of 18 kbar. The cooling path given by Okamoto & Toriumi (2005) passes through 400 C at around 3.5 kbar. Taking 400 C as the closure temperature for Ar Ar dating of phengite implies there was a decrease in pressure of 14.5 kbar in the space of <1.8 Myr. This time period is calculated as the maximum difference between the Lu Hf eclogite formation age (a) Seba Eclogite ESB-45 Gt B (b) Kotsu Eclogite KKT-3 Gt B Gt A 177 Hf/ 176 Hf Age = 88.8 ± 0.6 Ma 177 Hf/ 176 Hf Age = 88.2 ± 0.5 Ma Omph Initial 176 Hf/ 177 Hf = ± Omph Initial 176 Hf/ 177 Hf = ± Lu/ 177 Hf Lu/ 177 Hf Fig. 5. (a) Lu Hf isochron for the Seba eclogite. (b) Lu Hf isochron for Kotsu eclogite. In both cases, the age errors are given as 95% confidence limits.

10 102 S. R. WALLIS ET AL. (89.4 Ma) and the phengite Ar Ar cooling age (87.6 Ma) including the associated errors. Assuming an average density of the overburden of 3.3 g cm )3, this equates to a vertical exhumation rate of at least 2.5 cm yr )1. The presence of any crustal rocks and sepentinized mantle will lower the average density of the overburden and the estimated exhumation rate would have to be increased accordingly. Even allowing for a much higher closure temperature for phengite than usually considered, up to 500 C, would reduce this estimate by only 2 mm per year. Similar rates of exhumation have also been recognized for eclogites in other subduction settings, e.g. in eastern Papua New Guinea (Baldwin et al., 2004). After this early stage of rapid exhumation, the rate of exhumation decreased to around 0.3 mm yr )1 (q = 2.9 g cm )3 ); a reduction in two orders of magnitude (Fig. 6b) that implies a first-order change in exhumation mechanism. The initial phase of rapid exhumation revealed here implies the main body of the Sanbagwa belt cooled and rose to shallow crustal levels within a few million years of reaching peak metamorphic conditions. This rapid ascent can explain why the eclogite and other high pressure rocks of the Sanbagawa belt were able to survive the approach of a hot spreading ridge without being strongly overprinted by associated high temperature affects. The Sanbagwa belt and ridge subduction We propose the following history for the Sanbagawa subduction zone and metamorphic belt (Fig. 7). (1) Early History (>90 Ma, Fig. 7a): Subduction of the Izanagi Plate causes high-pressure low-temperature metamorphism. Plate reconstructions suggest that subduction was active in this period, but the ages reported here constrain most of the exposed Sanbagawa belt to be younger than around 90 Ma. The implication is that, most, if not all, subducted crustal material was recycled into the mantle and does not return along the subduction zone. The old ages reported by Okamoto et al. (2004) may represent an occurrence of locally preserved older subducted material. (2) Exhumation Stage (89 85 Ma, Fig. 7b): Approach of the Izanagi-Pacific spreading ridge caused warming of the subduction zone. Crustal rocks entering the subduction zone at this time undergo a characteristic P T path with decreasing thermal gradients with increasing pressure. Rocks (a) (b) Fig. 6. (a) North-south cross-section through the Sanbagawa belt of Shikoku showing the main tectonic units with the eclogite facies rocks at the highest structural level. The eclogite nappe is interfolded with the underlying Besshi nappe, which in turn overlies the Oboke nappe (Wallis & Aoya, 2000; Aoya, 2002). Chl, Gt and Ab-Bt refer to the chlorite, garnet and albite-biotite metamorphic zones within the Besshi nappe. (b) P T path and radiometric dates for the Seba eclogite area and the albite biotite zone of the Besshi nappe. Early exhumation rates of the Seba eclogite were on the order of cm yr )1 and are clearly much faster than later. The Seba area Ar-Ar data do not include data from the metagabbro body, which is associated with anomalously old ages. Data sources are as follows. Age data: Dallmeyer & Takasu (1991), Narita et al. (1999), Wallis et al. (2004b). P T data: A01 (Aoya, 2001), Z05 (Zaw et al., 2005), E94 (Enami et al., 1994), O&T 05 (Okamoto & Toriumi, 2005). Metamorphic facies boundaries are shown with the following abbreviations: GS greenschist, BS blueschist, EC eclogite, AM amphibolite, GR granulite, EA epidote amphibolite.

11 89 88 MA LU - HF AGES OF SANBAGAWA ECLOGITE 103 (a) (b) (c) Fig. 7. Summary of proposed relationship between approach of a spreading ridge and formation of the Sanbagawa metamorphic belt. (a) Prior to 90 Ma there is normal approximately steady-state cold-subduction of the Izanagi Plate that results in most or all of the subducted crustal material being recycled into the mantle. (b) Around 90 Ma the close approach of a spreading ridge and associated younging of the subducting slab causes the subduction zone to warm and triggering the rapid rise of relatively buoyant units of the Sanbagawa belt (EC = clogite nappe; BS = Besshi nappe; OB = Oboke nappe). The preserved metamorphic domain does not, therefore, reflect the conditions of steady-state subduction, but a snap shot of a particularly warm period in the evolution of the subduction zone. (c) Proposed relationship between the present architecture of the Sanbagawa belt and the former subduction zone. that were undergoing eclogite facies metamorphism began to return along the subduction zone. Details of the process of exhumation are beyond this present discussion, but we suggest that the temperature rise because of the approach of the ridge was the trigger needed to start exhumation. The implication is that the metamorphic conditions of the Sanbagawa belt are not representative of its long-term history, but only a snap shot of a particularly warm period in its evolution. (3) Later Development (<85 Ma, Fig. 7c): Exhumation of the three main units (Eclogite, Besshi and Oboke Nappes) of the Sanbagawa belt to midcrustal levels is followed by much slower exhumation, probably reflecting a change in dominant exhumation mechanism. CONCLUSIONS Lu Hf dating of eclogite from the Sanbagawa belt gives garnet growth ages of around 89 Ma. Plate reconstructions for this period suggest that a ridge interaction with the Sanbagawa subduction zone occurred around Ma. Thermal modelling and derivation of P T paths implies that the Sanbagawa metamorphism is best explained as the result of metamorphism immediately before such a ridge interaction. The Lu Hf eclogite ages of Ma are in excellent agreement with the model prediction suggesting that the metamorphic conditions recorded in the Sanbagawa belt are a snap shot of the history of the Sanbagawa subduction when conditions were relatively warm and not representative of the more normal cooler conditions. This conclusion has important implications for the use of metamorphic conditions of high P T metamorphic belts to estimate secular cooling of the Earth. The reason the Sanbagawa metamorphism was able to escape strong thermal overprinting associated with subduction of the ridge is that the metamorphic rocks exhumed and cooled rapidly, with initial vertical exhumation rates of cm yr )1. ACKNOWLEDGEMENTS Part of the analytical work was carried out under the tenure of a JSPS-Royal Society exchange fellowship

12 104 S. R. WALLIS ET AL. awarded to SW. JPP, MT, and RA acknowledge support grant NERC A S from the Natural Research Council of Great Britain. Part of this work was also supported by Grant-in-aid for scientific research No awarded to SW. We thank J. Wakabayashi and J. Inglis for careful and useful reviews. REFERENCES Abe, T., Takagi, H., Shimada, K., Kimura, S., Ikeyama, K. & Miyashita, A., Ductile shear deformation of the Sambagawa metamorphic rocks in the Kanto Mountains. Journal of the Geological Society of Japan, 107, Anczkiewicz, R. & Thirlwall, M. F., Improving precision of Sm-Nd garnet dating by H 2 SO 4 leaching a simple solution to phosphate inclusions problem. In: Geochronology: Linking the Isotopic Record with Petrology and Textures (eds Vance, D., Mueller, W. & Villa, I. M.), pp Geological Society, London, Special Publications, 220. Anczkiewicz, R., Platt, J. P., Thirlwall, M. F. & Wakabayashi, J., Franciscan subduction off to a slow start: evidence from high-precision Lu-Hf garnet ages on high-grade blocks. Earth & Planetary Science Letters, 225, Anczkiewicz, R., Szczepan ski, J., Mazur, S. et al., Lu-Hf geochronology and trace element distribution in garnet: implications for uplift and exhumation of ultra-high pressure granulites in the Sudetes, SW Poland. Lithos, 95, Aoya, M., P-T-D path of eclogite from the Sanbagawa belt deduced from combination of petrological and microstructural analyses. Journal of Petrology, 42, Aoya, M., Structural position of the Seba eclogite unit in the Sambagawa belt: supporting evidence for an eclogite nappe. The Island Arc, 11, Aoya, M., Uehara, S., Matsumoto, M., Wallis, S. R. & Enami, M., Subduction-stage pressure-temperature path of eclogite from the Sambagawa belt: Prophetic record for oceanic-ridge subduction. Geology, 31, Baldwin, S. L., Monteleone, B. D., Webb, L. E., Fitzgerald, P. G., Grove, M. & Hill, E. J., Pliocene eclogite exhumation at plate tectonic rates in eastern Paua New Guinea. Nature, 431, Blichert-Toft, J. & Albarede, F., The Lu-Hf isotope geochemistry of chondrites and the evolution of the mantle-crust system. Earth & Planetary Science Letters, 148, Dallmeyer, R. D. & Takasu, A., Tectonometamorphic evolution of the Sebadani eclogitic metagabbro and the Sambagawa schists, central Shikoku, Japan; 40 Ar 39 Ar mineral age constraints. Journal of Metamorphic Geology, 9, Dalmasso, J., Barci-Funel, G. & Ardison, G. J., Reinvestigation of the decay of the long-lived odd-odd 176 Lu nucleus. Applied Rad. Isot, 43, Enami, M., Pressure-temperature path of Sanbagawa prograde metamorphism deduced from grossular zoning of garnet. Journal of Metamorphic Geology, 16, Enami, M., Wallis, S. & Banno, Y., Paragenesis of sodic pyroxene-bearing quartz schists: implications for the P-T history of the Sanbagawa belt. Contributions to Mineralogy and Petrology, 116, Engebretson, D. C., Cox, A. & Gordon, G., Relative plate motions between oceanic and continental plates in the Pacific basin. Geological Society of America Special paper, 206, pp Faure, M., Microtectonic evidence for eastward ductile shear in the Jurassic orogen of SW Japan. Journal of Structural Geology, 7, Hara, I., Shiota, T., Hide, K. et al., Tectonic evolution of the Sambagawa schists and its implications in convergent margin processes. Journal of Science, Hiroshima University, Series C, 9, Iizuka, T. & Hirata, T., Simultaneous determinations of U-Pb age and REE abundances for zircons using ArF excimer laser ablation-icpms. Geochemical Journal, 38, Inui, M. & Toriumi, M., Prograde pressure temperature paths in the pelitic schists of the Sanbagawa metamorphic belt, SW Japan. Journal of Metamorphic Geology, 20, Irving, E., Drift of the major continental blocks since the Devonian. Nature, 270, Isozaki, Y. & Itaya, T., Chronology of Sambagawa metamorphism. Journal of Metamorphic Geology, 8, Itaya, T. & Fukui, S., Phengite K-Ar ages of schists from the Sanbagawa southern marginal belt, central Shikoku, Southwest Japan; influence of detrital mica and deformation on age. Island Arc, 3, Itaya, T. & Takasugi, H., Muscovite K-Ar ages of the Sanbagawa schists, Japan and argon depletion during cooling and deformation. Contributions to Mineralogy and Petrology, 100, Iwamori, H., Thermal effects of ridge subduction and its implications for the origin of granitic batholith and paired metamorphic belts. Earth & Planetary Science Letters, 181, Ludwig, K. R., Users manual for Isoplot In: Berkeley Geochronology Center Special Publications 1a, pp. 58. University of California at Berkeley. Maruyama, S., Pacific-type orogeny revisited: Miyashirotype orogeny proposed. Island Arc, 6, Matsumoto, M., Tectonic evolution of the glaucophane eclogite from the Kotsu area, eastern Shikoku, Japan: structural and petrological constraints. Unpublished Masters Thesis, Nagoya University. Matsumoto, M., Wallis, S. R., Aoya, M., Enami, M., Kawano, J. & Seto, Y., Petrological constraints on the formation conditions and retrogrde P T path of the Kotsu eclogite unit, central Shikoku. Journal of Metamorphic Geology, 21, McCaffrey, R., Oblique plate convergence, slip vectors, and forearc deformation. Journal of Geophysical Research, 97(B6), Mizukami, T. & Wallis, S. R., Structural and petrological constraints on the tectonic evolution of the garnet-lherzolite facies Higashi-Akaishi peridotite body, Sanbagawa belt, SW Japan. Tectonics, 24(TC6012), pp doi: /2004TC Molnar, P. & England, P. C., Temperature, heat flux, and friction al stress near major thrust fault. Journal of Geophysical Research, 95, Muta, K., Reconstruction of past plate motion based on kinematic studies of the Sanbagawa metamorphic belt. Unpublished Masters Thesis, Kyoto University. Narita, K., Yamaji, T., Tagami, T., Kurita, H., Obuse, A. & Matsuoka, K., The sedimentary age of the Tertiary Kuma Group in Shikoku and its significance. Journal of the Geological Society of Japan, 105, Okamoto, A. & Toriumi, M., Progress of actinoliteforming reactions in mafic schists during retrograde metamorphism: an example from the Sanbagawa metamorphic belt in central Shikoku, Japan. Journal of Metamorphic Geology, 23, Okamoto, K., Shinjoe, H., Katayama, I., Terada, K., Sano, Y. & Johnson, S., SHRIMP U-Pb zircon dating of quartzbearing eclogite from the Sanbagawa Belt, south west Japan: implications for metamorphic evolution of subducted protolith. Terra Nova, 16, van Ormon, J. A., Grove, T. L., Shimizu, N. & Layne, G. D., Rare earth element diffusion in a natural pyrope single crystal at 2.8 GPa. Contributions to Mineralogy and Petrology, 142,

13 89 88 MA LU - HF AGES OF SANBAGAWA ECLOGITE 105 Osozawa, S., The cessation of igneous activity and uplift when an actively spreading ridge is subducted beneath an island arc. Island Arc, 6, Ota, T., Terabayashi, M. & Katayama, I., Thermobaric structure and metamorphic evolution of the Iratsu eclogite body in the Sanbagawa belt, central Shikoku. Lithos, 73, Otofuji, Y.-I. & Matsuda, T., Amount of clockwise rotation of southwest Japan - fan shape opening of the southwestern part of the Japan Sea. Earth & Planetary Science Letters, 85, Peacock, S. M., Creation and preservation of subduction related inverted metamorphic gradients. Journal of Geophysical Research, 92, Peacock, S. M., Thermal and petrologic structure of subduction zones. Geophysical Monograph, 96, Pearce, N. J. G., Perkins, W. T., Westgate, J. A. et al., A compilation of new and published minor and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandards Newsletter, 21, Platt, J. P., Calibrating the bulk rheology of active obliquely convergent thrust and accretionary wedges from surface profiles and velocity distributions. Tectonics, 19, Scherer, E. E., Munker, C. & Mezger, K., Calibration of the lutetium hafnium clock. Science, 293, Shimizu, I., Ductile deformation in the low-grade part of the Sambagawa metamorphic belt in the northern Kanto Mountains, central Japan. Journal of the Geological Society of Japan, 94, Takahashi, M. & Saito, K., Miocene intra-arc bending at an arc-arc collision zone, central Japan. Island Arc, 6, Takasu, A., P-T histories of peridotite and amphibolite tectonic blocks in the Sanbagawa metamorphic belt, Japan. In: The Evolution of Metamorphic Belts. Vol. 43 (eds J. S. Daly, R. A. Cliff, B. W. D. Yardley), pp Geological Society of London Special Publications, 43. Takasu, A. & Dallmeyer, R. D., Ar- 39 Ar mineral age constraints for the tectonothermal evolution of the Sambagawa metamorphic belt, central Shikoku, Japan: a Cretaceous accretionary prism. Tectonophysics, 185, Takasu, A., Wallis, S. R., Banno, S. & Dallmeyer, R. D., Evolution of the Sambagawa metamorphic belt, Japan. Lithos, 33, Takasu, A., Dallmeyer, R. D. & Y., H., Ar 39 Ar muscovite ages of the Sambagawa schists in the Iimori district, Kii Peninsula, Japan: Implications for orogen-parallel diachroism. Journal of Geological Society of Japan, 102, Takeshita, T. & Yagi, K., Flow patterns during exhumation of the Sanbagawa metamorphic rocks, SW Japan, caused by brittle-ductile, arc-parallel extension. In: Vertical Coupling and Decoupling in the Lithosphere (eds Grocott, J., McCaffrey, K. J. W., Taylor, G. & Tikoff, B.), pp Geological Society of London Special Publications, 224. Thirlwall, M. F. & Anczkiewicz, R., Multidynamic isotope ratio analysis using MC-ICP-MS and the causes of secular drift in Hf, Nd and Pb isotope ratios. International Journal of Mass Spectrometry, 235, Toriumi, M. & Noda, H., The origin of strain patterns resulting from contemporaneous deformation and metamorphism in the Sambagawa metamorphic belt. Journal of Metamorphic Geology, 4, Uehara, S.-I. & Aoya, M., Thermal model for approach of a spreading ridge to subduction zones and its implications for high-p high-t metamorphism: Importance of subduction versus ridge approach ratio. Tectonics, 24(TC4007), pp doi: /2004tc Uyeda, S. & Miyashiro, A., Plate tectonics and the Japaense Islands: a synthesis. Bulletin of the Geological Society of America, 85, Wakabayashi, J., Counterclockwise P-T-t paths from amphibolites, Franciscan Complex, California: relics from the early stages of subduction zone metamorphism. Journal of Geology, 98, Wallis, S. R., The timing of folding and stretching in the Sanbagawa belt, the Asemigawa region, central Shikoku. Journal of the Geological Society of Japan, 96, Wallis, S. R., Vorticity analysis and recognition of ductile extension in the Sanbagawa belt, SW Japan. Journal of Structural Geology, 17, Wallis, S. R., Exhuming the Sanbagawa metamorphic belt; the importance of tectonic discontinuities. Journal of Metamorphic Geology, 16, Wallis, S. R. & Aoya, M., A re-evaluation of eclogite facies metamorphism in SW Japan: proposal for an eclogite nappe. Journal of Metamorphic Geology, 18, Wallis, S. R., Banno, S. & Radvanec, M., Kinematics, structure and relationship to metamorphism of the east-west flow in the Sanbagawa belt, southwest Japan. Island Arc, 1, Wallis, S. R., Moriyama, Y. & Tagami, T., 2004a. Chronology of the Sanbagawa metamorphism, SW Japan new constraints from zircon fission track analysis. Journal of Metamorphic Geology, 22, Wallis, S. R., Moriyama, Y. & Tagami, T., 2004b. Exhumation rates and age of metamorphism in the Sanbagawa belt: new constraints from zircon fission track analysis. Journal of Metamorphic Geology, 22, Yagi, K. & Takeshita, T., Regional variation in exhumation and strain rate of the high-pressure Sambagawa metamorphic rocks in central Shikoku, southwest Japan. Journal of Metamorphic Geology, 20, Zaw, W. K., Enami, M. & Aoya, M., Chloritoid and barroisite-bearing pelitic schists from the eclogite unit in the Besshi district, Sanbagawa metamorphic belt. Lithos, 81, Received 30 January 2007; revision accepted 23 November 2008.