Magma Genesis Beneath Northeast Japan Arc: a New Perspective on Subduction Zone Magmatism

Gondwana Research 16 (2009) 446–457

Contents lists available at ScienceDirect

Gondwana Research

journal homepage: www.elsevier.com/locate/gr

Magma genesis beneath Northeast Japan arc: A new perspective on subduction zone magmatism

Tetsu Kogiso a,⁎, Soichi Omori b, Shigenori Maruyama b a Graduate School of Human and Environmental Studies, Kyoto University, Yoshida-nihonmatsu, Sakyo, Kyoto 606-8501, Japan b Department of Earth and Planetary Sciences, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro, Tokyo 152-8551, Japan article info abstract

Article history: It is being accepted that earthquakes in subducting slab are caused by dehydration reactions of hydrous Received 28 February 2009 minerals. In the context of this “dehydration embrittlement” hypothesis, we propose a new model to explain Received in revised form 6 May 2009 key features of subduction zone magmatism on the basis of hydrous phase relations in peridotite and basaltic Accepted 8 May 2009 systems determined by thermodynamic calculations and seismic structures of Northeast Japan arc revealed Available online 18 May 2009 by latest seismic studies. The model predicts that partial melting of basaltic crust in the subducting slab is an inevitable consequence of subduction of hydrated oceanic lithosphere. Aqueous ﬂuids released from the Keywords: Subduction subducting slab also cause partial melting widely in mantle wedge from just above the subducting slab to just Arc magma below overlying crust at volcanic front. Hydrous minerals in the mantle wedge are stable only in shallow Dehydration (b120 km) areas, and are absent in the layer that is dragged into deep mantle by the subducting slab. The Slab melting position of volcanic front is not restricted by dehydration reactions in the subducting slab but is controlled by Phase equilibrium dynamics of mantle wedge ﬂow, which governs the thermal structure and partial melting regime in the mantle wedge. © 2009 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

1. Introduction seismic velocity structures in the mantle wedge beneath NE Japan (Hasegawa et al., 2009-this issue). In addition, new experimental and Subduction zone is a location where significant chemical differ- theoretical data on hydrous phase relations in peridotite and basalt entiation processes occur, such as the generation of continental crust systems (e.g., Omori et al., 2002, 2004; Komabayashi et al., 2004; and transfer of chemical heterogeneity into the deep mantle, both of Grove et al., 2006) allowed us to know precise pressure–temperature which play fundamental roles in the chemical evolution of the Earth. conditions of fluid and magma generation in subduction zones. Chemical differentiation in subduction zones is promoted mainly by Moreover, it is widely accepted that embrittlement caused by liberation of aqueous fluids from subducting oceanic lithosphere liberation of fluids from hydrous minerals in subducting slab is a (slab) and resultant magma generation. Therefore, quantitative potential cause of intraslab earthquakes (Raleigh, 1967; Kirby et al., modeling of fluid liberation in the subducting slab and its relevance 1996; Seno and Yamanaka, 1996, Jung et al., 2004). In the context of to magma genesis in subduction zones is critical for better under- this “dehydration embrittlement” hypothesis, the distribution of standing of the Earth's chemical evolution. Generation of aqueous intraslab seismicity can be linked with stability limits of hydrous fluids and silicate magmas in subduction zones has also been one of minerals in subducting slab, which in turn allows estimating detailed the most important issues in the area of igneous petrology, and many thermal structure of subducting slab (e.g., Omori et al., 2002; Hacker researchers have proposed a variety of petrologic models for the fluid et al., 2003; Omori et al., 2004, 2009-this issue). The thermal structure and magma genesis in subduction zones (e.g., Nicholls and Ringwood, of subducting slab is an essential parameter for quantitative modeling 1972; Wyllie and Sekine, 1982; Kushiro, 1983, Tatsumi, 1989, Schmidt of magma genesis in subduction zones. In this paper, we combine the and Poli, 1998; Kimura and Yoshida, 2006). However, there is little latest results of seismic studies of NE Japan with hydrous phase consensus on what are keys to explain the important features of relations in peridotite–basalt systems, and propose a new petrologic subduction zone magmatism, such as the existence of volcanic front model for subduction zone magmatism on the basis of the dehydra- and across-arc variations of magma chemistry and volume. tion embrittlement hypothesis. Recent progress in seismological studies of Northeast Japan arc (e.g., Nakajima and Hasegawa 2007; Kita et al., 2006) provided us with highly precise distribution of intraslab earthquakes and detailed 2. Subduction zone magmatism: key features and previous models

There are several important features that characterize subduction ⁎ Corresponding author. Fax: +81 75 753 2919. zone magmatism (Fig. 1). Here we review key features of subduction E-mail address: [email protected] (T. Kogiso). zone magmatism, and illuminate some problems that have not been

1342-937X/$ – see front matter © 2009 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.gr.2009.05.006 T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 447

surface around 100–130 km (Gill, 1981; Tatsumi, 1986)(Fig. 1). This feature has led some researchers to propose that the locus of magma genesis is controlled by depth-dependent dehydration reactions within and/or above subducting slab (e.g., Tatsumi, 1986, 1989). Tatsumi (1989) attributed the constancy of the slab surface depth beneath volcanic front to dehydration reaction of amphibole in peridotite, which he thought occurs at ∼110 km depth just above the subducting slab and forms a hydrous column to cause ﬂux melting in the mantle wedge. However, recent compilations of high-pressure hydrous phase relations (e.g., Schmidt and Poli, 1998; Iwamori, 1998; Omori et al., 2002) revealed that amphibole in the peridotitic system breaks down at pressures less than 3.0 GPa, which corresponds to b∼90 km depth. Moreover, it has been shown that serpentine and

Fig. 1. Schematic illustration of the key features of subduction zone magmatism. chlorite have the largest contribution to the H2O budget in peridotite and that amphibole breakdown plays only minor role (Schmidt and Poli, 1998; Omori et al., 2002). Because dehydration reactions of well explained by previous petrologic models for subduction zone serpentine and chlorite are highly temperature-dependent, there are no magmatism. major depth-dependent dehydration reactions that can produce con- siderable amounts of aqueous fluids in mantle wedge at ∼110 km depth. Another explanation for the constancy of the slab surface depth is 2.1. Occurrence of magmatism that flow and thermal regime in mantle wedge controls the location of volcanic front (e.g., Kushiro, 1983; Spiegelman and McKenzie, 1987; The most important feature of subduction zone magmatism is the Iwamori, 1998). Kushiro (1983) suggested that the location of volcanic occurrence of magmatism itself, that is, the fact that hot magmas are front should be controlled by the position of wet solidus of peridotite generated at settings in which the mantle is being cooled by in mantle wedge because magmas are generated by continuous subduction of cold oceanic lithosphere. There is a consensus that addition of aqueous fluids from subducting slab. Spiegelman and aqueous fluids released from hydrous minerals brought by the McKenzie (1987) numerically demonstrated that streamlines of melt subducting slab play critical roles in generating magmas at subduction in mantle wedge can converge to one point beneath overlying crust zones, because H O considerably lowers solidus temperatures of 2 owing to corner flow of solid mantle drugged by subducting slab. magma source materials (e.g., Kushiro, 1969). However, detailed Iwamori (1998) numerically simulated migration and chemical mechanisms of magma genesis in subduction zones have been reactions of aqueous fluids in solid peridotite flow in mantle wedge, disputed. and suggested that extensive melting in mantle wedge occur at fixed One possible mechanism is hydrous partial melting of oceanic depths because similar flux of fluids and flow pattern of solid can be basaltic crust in the subducting slab (e.g., Nicholls and Ringwood, attained regardless of age of subducting slab. 1972; Kay, 1980; Wyllie and Sekine, 1982). This mechanism had been proposed as the main cause for basaltic–andesitic magmas that 2.3. Across-arc variations of magma chemistry and volume dominate subduction zones. However, later melting experiments on hydrous basaltic systems (e.g., Beard and Lofgren, 1991; Rushmer, Across-arc variations of chemistry and volume of erupted magmas 1991; Wolf and Wyllie, 1994; Rapp and Watson, 1995) demonstrated are also important. Kuno (1959, 1960) reported alkali contents of that hydrous melting of basaltic crust dominantly produces silicic magmas increase toward the backarc side in the NE Japan arc. After magmas with particular chemical characteristics like tonalite or Kuno's pioneer works, many researchers reported such increasing adakite, which are limited in volcanic arcs beneath which relatively alkali contents toward backarc side in many volcanic arcs in the world young (=hot) oceanic lithosphere is subducting. In addition, (e.g., Miyashiro, 1974; Dickinson, 1975; Tatsumi and Eggins, 1995; analytical and numerical models on the thermal structure of Kimura and Yoshida, 2006). The across-arc chemical variations in subduction zones suggested that subducting oceanic crust older magmas are observed not only in major elements but also in than 20 Myr are unlikely to partially melt beneath volcanic arcs (e.g., incompatible trace element concentrations (e.g., Gill, 1981; Sakuyama Drummond and Defant, 1990; Molnar and England, 1995; Bourdon and Nesbitt, 1986) and radiogenic and stable isotopic ratios (e.g., et al., 2002). On the other hand, high geotherms of some arcs Nohda and Wasserburg, 1981; Hawkesworth et al., 1993; Ishikawa and petrologically estimated from metamorphic rocks suggest that Nakamura, 1994). The volume variation was first pointed out by subducting oceanic crust as old as 80 Myr can partially melt beneath Sugimura et al. (1963) who showed that the volume of erupted volcanic arcs (Kelemen et al., 2003). Partial melting of subducting magmas in NE Japan is largest in the volcanic front and sharply oceanic sediments has also been proposed from basalt geochemistry decreases toward the Japan sea. The magma volume decrease toward for old subduction zones such as Izu-Mariana (e.g., Ishizuka et al., backarc side has been reported in many volcanic arcs since then 2003). (Marsh, 1979; Tatsumi and Eggins, 1995). Some researchers suggested The other possible mechanism is hydrous partial melting in mantle that two volcanic chains occur in many single arcs because the magma wedge caused by addition of aqueous fluids from dehydration reaction volume does not seem to decrease monotonously to the backarc side in subducting slab (e.g., McBirney, 1969; Kushiro, 1983; Tatsumi et al., (Miyashiro, 1974; Marsh, 1979; Tatsumi and Eggins, 1995), although it 1983). This has been thought to be more plausible because subducting is not accepted as a general feature of subduction zone magmatism slab in any subduction zones has been hydrated by interaction with (Ishikawa and Nakamura, 1994; Kondo et al., 1998; K. Kaneko, pers. seawater before subduction, which necessarily cause dehydration comm. 2009). reactions during subduction. These across-arc variations in magma chemistry and volume have been attributed to physical conditions of magma genesis that tend to 2.2. Formation of volcanic front be affected by subduction of oceanic lithosphere. Before the concept of plate tectonics was established, Kuno (1959) already proposed that Another key feature is the formation of volcanic front (Sugimura, arc magmas are generated along the deep seismic zone (Wadati– 1960), in particular its correspondence to the depth of subducting slab Benioff zone) and attributed the chemical variations of arc magmas to 448 T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 the different depth of magma generation. Later high-pressure melting experiments on basalt systems (e.g., Green and Ringwood, 1967; Kushiro, 1968) revealed that the depth of the Wadati–Benioff zone (i.e., subducting slab) is too deep to explain chemical characteristics of magmas. Nevertheless, it is widely accepted that the depth of magma generation and/or segregation plays a key role in producing across-arc chemical variations, because further high-pressure experiments (Kushiro, 1983; Tatsumi et al., 1983) has demonstrated that alkali basalts that dominate in backarc side volcanism can segregate from peridotite at deeper part in mantle wedge than tholeiitic basalts in the volcanic front. Chemistry of aqueous fluids released from subducting slab is also an important parameter that affects the across-arc chemical variations, especially in trace element concentrations (Gill, 1981; Tatsumi and Kogiso,1997; Hochstaedter et al., 2001; Kimura and Yoshida, 2006; Nakamura et al., 2008). Thecauseofthevolumedecreasetowardbackarcsideis controversial. One explanation is that flux of slab-derived fluids, Fig. 2. Summary of notable seismic structures recently discovered in NE Japan arc. which promote partial melting in mantle wedge, is largest beneath volcanic front and decreases toward backarc side (Gill, 1981; Tatsumi, 1989; Cagnioncle et al., 2007). However, as described above, there are slab (Umino and Hasegawa, 1975; Hasegawa et al., 1978). The two no dehydration reactions that can supply large amounts of fluids just planes of the double seismic zone converge to one plane at around below volcanic front. Another explanation is the existence of partial 180 km depth (Hasegawa et al., 1978), beneath which the earthquake melting zone inclined toward backarc side, in which the largest extent activity decreases in the slab. If dehydration embrittlement induces of melting sufficient for melt segregation can be achieved at the upper intraslab earthquakes, the location of the double seismic zone should end below volcanic front (Marsh, 1979; Kushiro 1983; Schmidt and be loci of dehydration reactions in the subducting slab (Omori et al., Poli, 1998). Such inclined partial melting zone in mantle wedge can be 2002; Hacker et al., 2003), which imposes strong constraints on the formed because strong focusing of flow in the wedge corner caused by thermal structure of the subducting slab. Recently, Kita et al. (2006) Newtonian temperature-dependent or non-Newtonian rheology precisely relocated intermediate-depth (50–150 km) earthquakes produces high temperature zone slanted toward backarc side beneath NE Japan by applying the double-difference hypocenter (Furukawa, 1993; van Keken et al., 2002; Kelemen et al., 2003). This location method (Waldhauser and Ellsworth, 2000), and revealed thermal structure in the wedge corner can also explain the across-arc dense concentration of seismicity at depths of 70–90 km in the upper difference in magma segregation depth. plane of the double seismic zone (Fig. 2). This “upper-plane seismic belt” is located near the upper boundary of the subducting Pacific slab, 3. New view of subduction zone magmatism probably in the basaltic crust in the slab (Kita et al., 2006). The existence of the upper-plane seismic belt implies that major The most remarkable progress in recent studies of subduction zone dehydration reactions occur in the basaltic slab at these depths is the increasing acceptance of the dehydration embrittlement (Hacker et al., 2003; Omori et al., 2004; Omori et al., 2009-this issue; hypothesis as the cause of intraslab earthquakes (e.g., Kirby et al., see discussion below). Kita et al. (2006) also show that thicknesses of 1996; Jung et al., 2004; Hasegawa et al., 2009-this issue). As Hasegawa both the upper and lower planes of the double seismic zone are et al. (2009-this issue) summarized, an assumption that intraslab ∼15 km or thinner, suggesting that dehydration reactions in the slab earthquakes occur where hydrous minerals dehydrate in subducting occur just within restricted zones. slab can well explain many important phenomena in subduction zone: Tsuji et al. (2008) determined detailed seismic velocity structure of the existence of double seismic zone in the slab, uneven distribution of the subducting Pacific slab by double-difference tomography method earthquakes within it, seismic velocity structure of the slab and so on (Zhang and Thurber, 2003). The obtained tomography image shows (Omori et al., 2002, 2004; Kita et al., 2006; Tsuji et al., 2008; Omori the existence of a thin (b10 km) low-velocity layer at the top of the et al., 2009-this issue; Nakajima et al., 2009). The dehydration slab (Fig. 2). The low-velocity layer gradually disappears at depths of embrittlement hypothesis has great significance for modeling of 70–90 km, which corresponds to the depth of the upper-plane seismic subduction zone magmatism because it allows us to link intraslab belt found by Kita et al. (2006). Tsuji et al. (2008) also show that a earthquakes to the thermal structure of subducting slab and mantle low-velocity layer exists in the mantle wedge just above the Pacific wedge by comparing loci of earthquakes with pressure–temperature slab at depths greater than 70–90 km (Fig. 2). This low-velocity layer limits of hydrous mineral stabilities (Hacker et al., 2003; Omori et al., corresponds to that already found by previous studies (Matsuzawa 2002; Yamasaki and Seno, 2003; Omori et al., 2004; Nakajima et al., et al., 1986; Kawakatsu and Watada, 2007), which continues to deeper 2009). The thermal structure of subducting zone, which is critical to than 200 km (Tonegawa et al., 2008). These observations suggest that understand subduction zone magma genesis, has been estimated only dehydration reactions take place in subducting hydrated basaltic slab by numerical simulations (e.g., Davies and Stevenson, 1992; Furukawa, at depths around 70–90 km where expelled aqueous fluids move to 1993; Iwamori, 1998; Kelemen et al., 2003), but now it can be the mantle wedge and form a hydrous layer just above the slab (Tsuji determined by seismic observations and hydrous phase relations. et al., 2008; Nakajima et al., 2009) and that the hydrous layer in the Here we summarize recent progress in studies on seismology of NE mantle wedge is dragged into depths greater than 200 km by the Japan with regard to the relevance to modeling magma genesis (Fig. 2) subducting slab (Kawakatsu and Watada, 2007; Tonegawa et al., as well as recent studies on phase relations of hydrous minerals and 2008). It is also shown that another thin low-velocity layer exists melts in basalt and peridotite systems. around the lower plane of the double seismic zone within the subducting slab (Tsuji et al., 2008; Nakajima et al., 2009), suggesting 3.1. Seismic observations significant amounts of aqueous fluids released also along the lower seismic plane. A notable feature of the seismic activities in NE Japan is the Intensive seismic tomography studies beneath NE Japan (e.g., presence of a double-planed seismic zone in the subducting pacific Hasegawa et al.,1991; Zhao et al.,1992; Tsumura et al., 2000; Nakajima T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 449 et al., 2001; Hasegawa and Nakajima, 2004) have revealed the velocity zone, and that their positions well correspond to the locations existence of a prominent low-velocity zone in the mantle wedge that of volcanic clusters and loci of low-frequency microearthquakes is inclined toward backarc side subparallel to the subducting slab observed (Okada and Hasegawa, 2000). On the basis of these (Fig. 2). This low-velocity zone is observed in the whole region of NE observations, Hasegawa and Nakajima (2004) proposed that partial Japan, that is, the low-velocity zone has sheet-like shape. The low- melts segregated from the very low-velocity areas in the low-velocity velocity zone also has high Vp/Vs ratios and high seismic attenuation zone rise vertically to the overlying crust and cause volcanisms and (Zhao et al., 1992; Nakajima et al., 2001; Tsumura et al., 2000), and low-frequency microearthquakes. extends from depths around 150 km to the Moho just beneath the volcanic front (Zhao and Hasegawa, 1993; Zhao et al., 1994; Nakajima 3.2. Subsolidus phase equilibria in hydrous systems et al., 2001; Hasegawa and Nakajima, 2004). The existence of the sheet-like inclined low-velocity high-attenuation zone is thought to Subsolidus phase relations of hydrous minerals in peridotite and correspond to the locus of upwelling flow in the mantle wedge, in basalt systems have been determined in detail by high-pressure which peridotite is partially molten (e.g., Iwamori and Zhao, 2000; experiments and thermodynamic calculations in the last decades (e.g., Hasegawa and Nakajima, 2004). The presence of small amounts of Schmidt and Poli, 1998; Iwamori, 1998; Omori et al., 2002; Iwamori, partial melts in the inclined low-velocity zone is supported by detailed 2004; Omori et al., 2004; Okamoto and Maruyama, 2004; and references investigations of its seismic attenuation structure (Nakajima and therein). Omori et al. (2009-this issue) newly constructed P–T phase

Hasegawa, 2003; Nakajima et al., 2005). Detailed tomographic images diagrams in the model system MgO–Al2O3–SiO2–H2O(MASH)for of the inclined low-velocity zone obtained by Hasegawa and Nakajima peridotite compositions and Na2O–CaO–FeO–MgO–Al2O3–SiO2–H2O (2004) show that there are several very low-velocity areas in the low- (NCFMASH) for basalt composition by thermodynamic calculations

Fig. 3. (a) and (b). Calculated P–T diagrams of stable minerals and H2O contents for peridotite (a: lherzolite and b: harzburgite) in the MASH system (Omori et al., 2009-this issue). The thick lines are boundary of the main dehydration ﬁeld (see text). Stable phases and boundaries of mineral assembly are shown. Abbreviations are: br = brucite, chl = chlorite, ol = olivine (forsterite), opx = orthopyroxene, phA = phase A, srp = serpentine (antigorite), tlc = talc. (c) Calculated P–T diagram of stable minerals and H2O contents for average MORB in the NCFMASH system (Omori et al., 2009-this issue). Thick broken lines show approximate position of the boundaries between major metamorphic facies. Thick gray arrow indicates the estimated P–T path of the slab surface (Omori et al., 2009-this issue). Thin line is the H2O-saturated solidus of MORB (see panel d). Note that the maximum value of the color scale for H2O contents is 15 wt.% for lherzolite and harzburgite, and 5 wt.% for MORB. Abbreviations are: Am = amphibolite, AmEc = amphibole eclogite, BS = blueschist, EC = eclogite,

EpAm = epidote amphibolite, EpEc = epidote eclogite, GS = greenschist, LwBS = lawsonite blueschist, LwEc = lowsonite eclogite. (d). H2O-saturated solidus of lherzolite (Grove et al., 2006) and that of MORB (Liu et al., 1996; Schmidt and Poli, 1998). Also shown is critical curve in peridotite–H2O system above which aqueous fluid and silicate melt are indistinguishable (Mibe et al., 2007). 450 T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 and semi-quantitative analysis of high-pressure experiments. Details of 4. Modeling for magma genesis calculation, chemical compositions and thermodynamic data of phases encountered in the model are described in Omori et al. (2009-this issue). On the basis of the recent seismic and petrologic studies The phase diagrams determined for model lherzolite, harzburgite, and summarized above, we propose a new petrologic model for magma oceanic crust (mid-oceanic ridge basalt: MORB) compositions are shown in Fig. 3. Principal hydrous phases that can release large amounts of fluids in peridotite systems are serpentine, brucite, talc and chlorite (Fig. 3a–b). Although amphibole is also a major hydrous mineral in natural peridotite systems including Na2O(e.g.,Schmidt and Poli, 1998; Grove et al., 2006), its contribution to H2O budget in subducting slab peridotite is much smaller than above three minerals (Iwamori, 1998, 2007). So the phase diagram in the MASH system is a good approximation of hydrous phase equilibria in peridotite of subducting slab and mantle wedge. Thus the major dehydration reactions in subducting slab occur under P–T conditions where serpentine, brucite, talc and chlorite decompose (between thick lines in Fig. 3a–b). The high- pressure end of this “main dehydration field” is around 6–7GPa (∼200 km depth) (Omori et al., 2002, 2009-this issue), which is close to the depth at which the two planes of the double seismic zone converge (Fig. 2). It should be noted that decomposition reactions of these minerals are highly temperature-dependent, that is, they occur within narrow temperature interval (∼200 °C: Fig. 3a–b), suggesting these reactions are useful to estimate temperature distribution in subducting slab in the context of the dehydration embrittlement hypothesis (e.g., Omori et al., 2002). In the hydrous basalt system (Fig. 3c), major dehydration reactions are decomposition of Ca-amphibole (b2 GPa), epidote (1–2.5 GPa), lawsonite and Na-amphibole (N2 GPa). The Ca-amphibole decomposition reaction is pressure-dependent, that is, its decomposition occurs at nearly constant pressure over a range of temperature (Fig. 3c). In contrast, decomposition reactions of lawsonite and Na- amphibole are rather temperature-dependent than pressure-dependent. Thus, loci of dehydration in basaltic slab is sensitive to temperature at depths greater than ∼70 km. This means that the uneven distribution of earthquakes within the basaltic slab (Kita et al., 2006) is also useful to estimate temperature distribution in subducting slab (Omori et al., 2009-this issue).

3.3. Melting phase relations of hydrous systems

Grove et al. (2006) precisely determined melting phase equilibria of hydrous lherzolite at near-solidus conditions by careful observations of high-pressure experiment products, and found that the solidus temperature of peridotite is as low as 800 °C at 3.2 GPa (Fig. 3d). A surprising result of their experiments is that chlorite appears as a stable phase above the solidus at pressures higher than 2.8 GPa. This means that hydrous peridotite will melt even below the stability limit of chlorite. Melting phase equilibria of hydrous basaltic system determined for a typical MORB composition (Liu et al., 1996; Schmidt and Poli, 1998) showed that the solidus of hydrous MORB basalt locates around 700 °C at 1–2 GPa, which is also well below the stability limit of chlorite in hydrous peridotite (Fig. 3d). These data Fig. 4. Seismic data in the cross section beneath Kurikoma–Chokai line of NE Japan arc. imply that basalt and peridotite necessarily melt at depths greater The position of the cross section is shown as a line in the inset. Red triangles in the inset are Quaternary volcanoes, among which Sengan area (see text) is circled. (a). S-wave than ∼80–100 km if their temperatures are higher than the chlorite velocity perturbation determined by Nakajima et al. (2001). Gray and red circles are stability limit. microearthquakes and low-frequency microearthquakes, respectively (Okada and The other important feature of hydrous melting of peridotite is Hasegawa, 2000). (b). Detailed S-wave velocity perturbation determined with complete miscibility between aqueous fluid and silicate melt at high double-difference tomography method by Nakajima et al. (2009). The image (a) was ∼ pressures (Mibe et al., 2007). In situ observation of high-pressure and obtained so that the long wave-length ( 20 km) structure around the central part of the mantle wedge is clearly determined (Nakajima et al., 2001). On the other hand, the high-temperature experimental runs by synchrotron radiation X-ray image (b) was obtained so that the shorter wave-length (∼5 km) structure near the slab radiography (Mibe et al., 2007) revealed that aqueous fluid and surface is clearly determined (Nakajima et al., 2009). Thus, the difference between (a) hydrous silicate melt in a peridotite system become indistinguishable and (b) is due to the differences not only in tomographic inversion method but also in above 3.8 GPa and ∼1000 °C (Fig. 3d). Therefore, at depths greater region that is well resolved by each method. (c). Locations of intraslab earthquakes determined with double-difference hypocenter location method by Kita et al. (2006, than ∼110 km, there can exist one supercritical liquid at higher 2008). The inclined line is the position of the slab surface (Kita et al., 2008). Colors temperatures, which changes its composition with increasing tem- indicate distance from the slab surface: red b10 km, green=10–23 km, blue N23 km, perature gradually from H2O-rich to silicate-rich at around 1000 °C. gray=above slab surface. T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 451 genesis in NE Japan arc. The seismic data we employed for modeling modeling (e.g., Iwamori, 1998; Kelemen et al., 2003; Nakajima and are the vertical cross sections of seismic tomography in the mantle Hasegawa, 2003; Nakajima et al., 2005). wedge (Nakajima et al., 2001), high-resolution tomography near the The temperature distribution in the mantle wedge is difficult to top of the subducting slab (Nakajima et al., 2009), and distribution of estimate because few petrologic constraints are available. Ueki and earthquakes in the subducting slab (Kita et al., 2006, 2008) along the Iwamori (2007) investigated thermal and chemical conditions of line connecting Kurikomayama and Chokaisan volcanoes (Fig. 4). magma genesis beneath the volcanic front of NE Japan by dense sampling and petrologic analyses of volcanic rocks in Sengan area 4.1. Thermal structure (Fig. 4), and revealed by thermodynamic calculations with pMELTS (Ghiorso et al., 2002) that parental magmas were produced at 1250–

For estimating temperature distribution in subduction zone, we 1300 °C and 1.0–1.5 GPa with 0.3–0.5 wt.% H2O. Thus we assume that employ several major assumptions in the context of the dehydration the temperature at the Moho just below the volcanic front is around embrittlement hypothesis as follows. 1300 °C. To estimate temperatures around the inclined low-velocity zone in (1) The location of the double seismic zone in the subducting Pacific the mantle wedge, we focused our attention to the change of velocity slab corresponds to the main dehydration field in the P–T anomaly values near the center of the low-velocity zone. The S-wave diagram of the hydrous peridotite (Omori et al., 2002)(Fig. 3a). velocity perturbation value abruptly changes around +1% to −3% on The inner edge of the double seismic zone is where brucite in both side of the low-velocity zone (Fig. 4a). We interpret that this harzburgite or talc in lherzolite decomposition starts, and the abrupt change corresponds to a sudden increase of melt fraction, outer edge is where chlorite decomposition ends. because velocity reduction rates are correlated with melt fraction in (2) Thickness of basaltic crust in the subducting slab is 7 km. addition to aspect ratio of melt pore and temperature (Nakajima et al., The earthquakes within 7 km from the upper boundary of the 2005). Melting experiments on hydrous depleted peridotite (Kubo, subducting slab are caused by dehydration reactions in the 2003) demonstrated that melt fraction of depleted peridotite under basaltic system. hydrous condition remains very low (b5%) between the hydrous (3) The bulk composition of peridotite in the mantle wedge and in solidus and the dry solidus and steeply increases around dry solidus the subducting slab is depleted lherzolite, except that the temperature. Therefore, we assume that the position of the −1to uppermost portion of the peridotite slab consists of harzburgite −3% contours of the S-wave velocity perturbation roughly matches (i.e., the upper plane of the double seismic zone is within the the solidus of dry depleted peridotite (Robinson et al., 1998; harzburgite portion). The bulk of subducting oceanic crust is Wasylenki et al., 2003; Laporte et al., 2004). The temperature average MORB (Omori et al., 2002, 2009-this issue). distribution in the subducting slab and mantle wedge determined When employing the assumption (1), we didn't take into account with these assumptions is shown in Fig. 5. the earthquakes between the upper and lower plane of the double seismic zone at ∼60–90 km depths. The wide distribution of these 4.2. Distribution of hydrous minerals and aqueous fluids “interplane” earthquakes at these depths can be explained by continuous decomposition of brucite if we include Fe in the bulk From the temperature distribution determined above and hydrous system (Omori et al., 2002; Omori and Komabayashi, 2007). Other phase relations (Figs. 3 and 5), we estimated the stability fields of assumptions we employed to determine the thermal structure are hydrous minerals, aqueous fluids and silicate melts in the subducting described and discussed below. We used here constraints obtained slab and mantle wedge (Fig. 6). We assumed that aqueous fluids from petrologic studies only in order to clarify the difference from the liberated from the double seismic zone migrate vertically much faster thermal structure determined by seismic studies and numerical than slab subduction and mantle wedge flow just for simplicity. As the

Fig. 5. Estimated temperature distribution within the subducting slab and the mantle wedge beneath Chokai–Kurikoma line of NE Japan. Blue digits with color dots are temperatures (°C) estimated with the assumption that intraslab earthquakes are caused by dehydration in the subducting slab (Omori et al., 2009-this issue). Dot colors indicate bulk compositions assumed for estimating temperatures. Light blue ovals are approximate positions of dry solidus of depleted peridotite. Dark blue oval indicates the temperature and depth of the generation of primitive arc magmas (Ueki and Iwamori, 2007). Red contours are smoothed isotherms. Gray crosses are the location of intraslab earthquakes (Kita et al., 2006, 2008). 452 T. Kogiso et al. / Gondwana Research 16 (2009) 446–457

Fig. 6. (a) Stability fields of hydrous minerals and partial melt estimated from the temperature distribution (Fig. 5). Black and dark red arrows schematically show movement of aqueous fluids and silicate melts, respectively. A thick broken line shows the boundary of critical curve below which aqueous fluids and silicate melts exists as supercritical liquid. Abbreviations are as in Fig. 3. (b) Estimated possible distribution of free aqueous fluid, silicate melt, and supercritical liquid.

slab continues subducting, aqueous fluids are continuously supplied, tion of the outer boundary of the double seismic zone (=stability and the liberated fluids hydrate the overlying portion of the slab and limits of hydrous phases in peridotite) into the mantle wedge, which the mantle wedge along migration paths. Thus the slab above the crosses the slab surface at around 120 km depth. This means that no double seismic zone and the mantle wedge would be locally saturated hydrous minerals exist at the bottom layer of the mantle wedge with H2O along fluid migration paths. Therefore, we drew the stability deeper than 120 km, that is, the bottom layer dragged by the fields of hydrous minerals and silicate melts, and possible distribution subducting slab into deep mantle is free from hydrous minerals even of free aqueous fluids for H2O-saturated conditions (Fig. 6), but we do though aqueous fluids are being supplied from the subducting slab. In not mean that these phases are uniformly distributed within the the basaltic slab, hydrous minerals are stable up to ∼120 km depth, respective stability fields. but major dehydration reactions occur around 60–80 km depths Our estimation shows that hydrous minerals (mainly serpentine) (Fig. 3c). This corresponds to the depth of the lower end of the low- can be present in the wedge corner and in the subducting slab velocity layer at the top of the slab (Fig. 4b; Nakajima et al., 2009), between the double seismic zone if aqueous fluids are supplied there suggesting that the low-velocity layer in the slab is consistent with the (Fig. 6a). This is consistent with high Poisson's ratios observed within presence of hydrated basaltic slab. the wedge corner and subducting slab beneath Kanto area of Japan Major dehydration reactions occur between talc/brucite and (Kamiya and Kobayashi, 2000; Omori et al., 2002). It should be noted chlorite stability limits in the peridotite slab (the main dehydration that the stability fields of hydrous minerals in the mantle wedge are field of Fig. 3a–b) as well as at shallower (b120 km) part of the limited to the wedge corner shallower than 120 km (Fig. 6a). This basaltic slab (Fig. 3c). Therefore, free aqueous fluids can be present results from the temperature distribution determined by extrapola- within the double seismic zone and basaltic slab because aqueous T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 453

fluids are continuously liberated there as the slab sinks (Fig. 6b). In the chemistry of slab-derived fluids, differences in amounts of slab- wedge corner, aqueous fluids are continuously supplied from the derived components and so on (see Kimura and Yoshida (2006) for subducting slab, but the mantle flow is thought to be quite slow (e.g., details). Some of the previous studies suggested that partial melting of Kelemen et al., 2003). Therefore, the supplied fluids can hydrate subducting sediments contribute to produce the across-arc variations peridotite in the corner until it become oversaturated with H2O along even in old subduction zones like Izu-Mariana (e.g., Johnson and fluid migration paths. Thus free fluids can also be present in the wedge Plank, 1999; Ishizuka et al., 2003). Because the solidus temperatures of corner (Fig. 6b). On the other hand, in the slab between the upper and sediment and basalt are similar under hydrous conditions (Schmidt lower planes of the double seismic zone, it is unclear whether free and Poli, 1998; Johnson and Plank, 1999), partial melting of basaltic fluids can be present. Because the supply of liberated fluids from the slab can occur if sediment melting occurs. In our model, subducting lower seismic plane will stop when all hydrous minerals completely slab releases aqueous fluids into overlying mantle wedge at shallower decompose, the liberated fluids will be completely consumed by depths (b∼120 km), whereas it supplies partial melts (or supercritical hydration of surrounding peridotite if their flux is small. The fluid flux liquid) from the basaltic slab at deeper depths (N∼120 km). Thus from the lower seismic plane depends on the amounts of hydrous liquid phases released from the subducting slab change their chemical minerals there, which is to be quantitatively investigated in future compositions with increasing depth, and such changes in chemistry of studies. slab-derived fluid/melt can be another important factor to make across- arc chemical variations of basalt chemistry. Detailed experimental 4.3. Locus of partial melting studies for chemistry of fluids, melts and supercritical liquids released from hydrous basalt and peridotite compositions will advance our Since aqueous fluids are supplied to the basaltic slab from the understanding of the role of slab melting to arc basalt chemistry. double seismic zone below, the basaltic slab can be saturated with H2O along fluid migration paths even above the stability limits of hydrous 5.2. Low-velocity layer at the bottom of the mantle wedge minerals in the basalt system (Fig. 3c). This causes partial melting within the basaltic slab at depths around 120 km and deeper (Fig. 6) Another remarkable aspect of our model is the absence of hydrous where the temperature of the basaltic slab exceeds the hydrous minerals in the mantle wedge just above the subducting slab at depths solidus of basalt (Fig. 3d). In other words, slab melting necessarily deeper than 110 km. Many previous models have proposed that occurs beneath NE Japan if we accept the dehydration embrittlement hydrous minerals are stable in the bottom layer of the mantle wedge if hypothesis. subducting slab is as cold as the Pacific slab beneath NE Japan and that Partial melting also occur in a wide area of the mantle wedge (Fig. 6), the layer is dragged into deep mantle by subducting slab (Tatsumi, because the temperatures in most parts of the mantle wedge exceed the 1989; Schmidt and Poli, 1998; Iwamori, 2004). These previous models hydrous solidus of peridotite (Figs. 3 and 5). It should be noted that the suggested that hydrous minerals in the dragged layer are stable partial melting area extends from just above the subducting slab. Melt beneath volcanic front and deeper. In our model, however, hydrous fraction increases toward the center of the inclined low-velocity zone minerals are stable only at shallow (b120 km) depths in the mantle and become highest just below the volcanic front, if we assume that the wedge corner. Therefore, the thin low-velocity layer just above the S-wave velocity perturbation correlates with melt fraction (Nakajima slab around 70–90 km depths found by the double-difference et al., 2005). Thus the position of the volcanic front corresponds to the tomography (Tsuji et al., 2008; Nakajima et al., 2009)correspondsto area of maximum melt fraction in the mantle wedge, and is not related to the hydrous mineral stability area of our model (Figs. 2 and 6). However, hydrous phase stabilities (see discussion below). the low-velocity layer found at the deeper portion (Kawakatsu and Watada, 2007; Tonegawa et al., 2008) does not indicate the presence of 5. Discussion hydrous minerals there. The deep low-velocity layer, which was found by the receiver 5.1. Slab melting function technique (Kawakatsu and Watada, 2007; Tonegawa et al., 2008), is not necessarily a thin “layer”, because the thickness of the An important feature of the petrologic model we proposed here is low-velocity “layer” cannot be constrained solely by the receiver that partial melting occurs in the subducting slab. In previous studies functions. Instead, the receiver function technique is sensitive to an on subduction zone magmatism, it has widely been thought that slab sharp velocity change within a short length scale. Such a sharp melting occurs only where relatively young oceanic lithosphere is velocity change can be attained in our model. Our model predicts that subducting (e.g., Drummond and Defant, 1990; Peacock, 2003). partial melting in the mantle wedge occurs just above the subducting Therefore, in the NE Japan, where a very old (∼130 Ma) oceanic slab and in the slab at deeper than ∼120 km (Fig. 6). Below this partial plate is subducting, slab melting has never been considered to occur melting zone, hydrous minerals and/or free aqueous fluids are (e.g., Tatsumi, 1989; Iwamori, 2007). However, in the context of the present, and the boundary of these zones are nearly parallel to the dehydration embrittlement hypothesis, the subducting Pacific oceanic slab surface (Fig. 6b). Although amounts of aqueous fluids and crust inevitably undergoes partial melting because of the melting hydrous minerals in the slab are difficult to constrain, it is clear that phase relations of hydrous basalt (Fig. 3). This means that slab melting the fraction of “liquid” (silicate melt or aqueous fluid) in a certain commonly occur in subduction zones worldwide. This is consistent volume of peridotite/basalt abruptly increases around the solidus with with the numerical model incorporating temperature-dependent increasing temperature in a hydrous peridotite and basalt systems viscosity (Kelemen et al., 2003), which demonstrated that slab surface (Fig. 7). Therefore, the fraction of partial melts in the partial melting temperature can exceed wet solidus of basalt even if subducting slab is zone is much higher than the fraction of aqueous fluids in the slab, at older than 80 Myr. least along fluid migration paths. Thus, the bottom of the partial Slab melting can be one of the main factors to produce across-arc melting zone can be a sharp velocity-jump boundary (Fig. 7), which variations in arc basalt chemistry. The across-arc geochemical can be detected by the receiver function technique. variations of arc basalts were well documented by thorough petrological and geochemical studies (e.g., Sakuyama and Nesbitt, 5.3. Position of volcanic front 1986; Shibata and Nakamura, 1997; Hochstaedter et al., 2001; Ishizuka et al., 2003; Kimura and Yoshida 2006). These studies attributed the Since the major dehydration reactions in the slab are essentially across-arc variations to many different processes, such as different temperature-dependent (Fig. 3), there is no rationale for the idea that degrees of partial melting in the mantle wedge, difference in the position of volcanic front is controlled chiefly by pressure- 454 T. Kogiso et al. / Gondwana Research 16 (2009) 446–457

Fig. 7. Schematic phase diagrams at below (a) and above (b) the second critical end point in a simplified silicate–H2O system. When the bulk H2O content is fixed, the fraction of fluid is f/g at below-solidus temperature Y. At above-solidus temperature Z, the fraction of melt is f/h, which is much higher than f/g. Thus, the fraction of “liquid” (fluid or melt) abruptly increases across the solidus temperature. Even under conditions above the second critical end point, similar abrupt increase in liquid fraction occurs as shown in (b). Therefore, the bottom boundary of the partial melting zone (thick broken line in panel c) can be a sharp velocity-jump boundary. dependent dehydration reactions in the mantle wedge (Tatsumi, 1986, similar position in various subduction zones. However, it is not 1989). Therefore, such “single phase dehydration model” is no longer theoretically clarified why the flow pattern in the mantle wedge can valid for subduction zone magmatism, as already pointed out by be similar with varying age and angle of subducting slab. Since there Schmidt and Poli (1998). Our model predicts that chlorite and are no petrologic constraints to rationalize the formation of volcanic serpentine decompose beneath the volcanic front in the mantle front so far, more theoretical investigations on the dynamics of mantle wedge (Fig. 6), which seems to indicate that these decomposition wedge flow is needed. reactions in the mantle wedge control the position of volcanic front. However, the position of these decomposition reactions is chiefly 5.4. Temperature distribution in the mantle wedge governed by the temperature distribution in the mantle wedge. In addition, partial melting widely occurs in the mantle wedge, so the In our model, we assumed that the temperature in the mantle position of chlorite and serpentine decomposition does not have wedge exceed the dry solidus of depleted peridotite near the central direct relevance to the position of volcanic front. Thus, the thermal portion of the inclined low-velocity zone (∼1300 °C just below the structure in the mantle wedge is the key to the formation of volcanic volcanic front: Fig. 5) based on recent petrologic studies on arc basalts front and the constancy of the depth to the slab surface beneath (Ueki and Iwamori, 2007). This estimation is too high as compared volcanic front. with the mantle wedge temperature estimated by numerical model- Many researchers proposed that the inclined low-velocity zone ing (e.g., Drummond and Defant, 1990; Iwamori, 1998; Peacock, corresponds to the locus of partial melting (Kushiro, 1987; Iwamori 2003). This discrepancy mainly derives from uncertainties in the and Zhao, 2000; Hasegawa and Nakajima, 2004; Nakajima et al., rheological properties of mantle wedge incorporated in numerical 2005). Hasegawa and Nakajima (2004) suggested that the inclined calculations, as demonstrated by van Keken et al. (2002) and Kelemen low-velocity zone indicates the position of a sheet-like upwelling flow et al. (2003). Our understanding of the rheological and other physical in the mantle wedge where significant partial melting occurs. Iwamori properties of mantle materials is not yet adequate to rigidly constrain (1998, 2007) numerically demonstrated that flow pattern in the the temperature in the mantle wedge. mantle wedge and fluid flux from the slab are not changed We did not draw high temperature anomaly below the backarc- significantly even if angle and age of subducting slab vary, suggesting side volcanism (i.e., Chokai volcanic zone) in our model (Fig. 6), that a similar thermal structure can be achieved in various subduction because the seismic velocity structure (Fig. 4a) doesn't show zones. These studies imply that the position of volcanic front is significant anomalies beneath the backarc-side volcanoes. This can controlled by inclined upwelling flow, which can be formed at a be explained by localization of melt migration. The occurrence of low- T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 455 frequency microearthquakes just below the Chokai volcano (Fig. 4)is References thought to be caused by migration of liquid phases (Hasegawa et al., 1991). This suggests that partial melts are migrating upward there Beard, J.S., Lofgren, G.E., 1991. Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1, 3, and 6.9 GPa. Journal of (Hasegawa and Nakajima, 2004). Therefore we interpret that magmas Petrology 32, 365–401. of the backarc-side volcanoes are segregated from the inclined partial Bourdon, E., Eissen, J.P., Monzier, M., Robin, C., Martin, H., Cotten, J., Hall, M.L., 2002. melting zone and migrate upward in small length scales that cannot Adakite-like lavas from Antisana volcano (Ecuador): evidence for slab melt metasomatism beneath Andean Northern Volcanic Zone. Journal of Petrology 43, be detected by seismic tomography. 199–217. Cagnioncle, A.M., Parmentier, E.M., Elkins-Tanton, L.T., 2007. Effect of solid flow above a subducting slab on water distribution and melting at convergent plate boundaries. 5.5. Implication of slab melting Journal of Geophysical Research 113, B09402. doi:10.1029/2007JB004934. Davies, J.H., Stevenson, D.J., 1992. Physical model of source region of subduction zone volcanics. Journal of Geophysical Research 97, 2037–2070. If slab melting is a general feature of the subduction zone Dickinson, W.R., 1975. Potassium–depth (K–h) relations in continental margin and magmatism, it has an important implication for deep mantle recycling intraoceanic magmatic arcs. Geology 3, 53–56. of crustal material. As partial melting of subducting oceanic crust Drummond, M.S., Defant, M.J., 1990. A model for trondhjemite–tonalite–dacite genesis and crustal growth via slab melting: Archean to modern comparisons. Journal of proceeds, the composition of its residue approaches bimineralic Geophysical Research 95, 21503–21521. eclogite (comprised solely of garnet+clinopyroxene) owing to the Furukawa, Y., 1993. Magmatic processes under arcs and formation of the volcanic front. loss of silica-rich components as partial melts (Kogiso et al., 2004). Journal of Geophysical Research 98, 8309–8319. – Ghiorso, M.S., Hirschmann, M.M., Reiners, P.W., Kress, V.C., 2002. The pMELTS: a Since bimineralic eclogite is on the thermal divide in the peridotite revision of MELTS for improved calculation of phase relations and major element basalt melting regime (O'Hara, 1968; Kogiso et al., 2004), the residue partitioning related to partial melting of the mantle to 3 GPa. Geochemistry, composition remains bimineralic until partial melting ceases. There- Geophysics, Geosystems 3. doi:10.1029/2001GC000217. fore, bimineralic eclogite can be the representative lithology of the Gill, J.B., 1981. Orogenic Andesites and Plate Tectonics. Springer-Verlag, New York. Green, D.H., Ringwood, A.E., 1967. The genesis of basaltic magmas. Contributions to crustal material subducting into deep mantle. If such bimineralic Mineralogy and Petrology 15, 103–190. fl eclogite is recycled in the deep mantle and involved in upwelling Grove, T.L., Chatterjee, N., Parman, S.W., Médard, E., 2006. The in uence of H2Oon – mantle plumes, it can generate nepheline-normative melts similar to mantle wedge melting. Earth and Planetary Science Letters 249, 74 89. Hacker, B.R., Abers, G.A., Peacock, S.M., 2003. Subduction factory: 1. Theoretical mineralogy, alkalic ocean island basalts (OIB) that are dominant product in many densities, seismic wave speed, and H2O contents. Journal of Geophysical Research 108, hotspot volcanoes (Kogiso et al., 2003; Kogiso and Hirschmann, 2006). 2029. doi:10.1029/2001JB001127. Thus, the idea that slab melting is a general phenomenon in Hasegawa, A., Umino, N., Takagi, A., 1978. Double-planed structure of the deep seismic zone in the northeastern Japan arc. Tectonophysics 47. subduction zone is consistent with OIB genesis that requires common Hasegawa, A., Zhao, D., Hori, S., Yamamoto, A., Horiuchi, S., 1991. Deep structure of the occurrence of bimineralic eclogite in their source regions as recycled northeastern Japan arc and its relationship to seismic and volcanic activity. Nature crustal component. 352, 683–689. Hasegawa, A., Nakajima, J., 2004. Geophysical constraints on slab subduction and arc magmatism. In: Sparks, R.S.J., Hawkesworth, G.J. (Eds.), The State of the Planet: Frontiers and Challenges in Geophysics. In: Geophysical Monograph, vol. 150. 6. Summary American Geophysical Union, Washington, D.C., pp. 81–94. Hasegawa, A., Nakajima, J., Uchida, N., Okada, T., Zhao, D., Matsuzawa, T., Umino, N., Recent remarkable progress in seismic studies of NE Japan and 2009. Plate subduction and generation of earthquakes and magmas in Japan as inferred from seismic observations: an overview. Gondwana Research 16, 370–400 petrologic studies of hydrous phase relations in the mantle enabled us (this issue). to newly build a petrologic model for subduction zone magmatism Hawkesworth, G.J., Gallagher, K., Hergt, J.M., McDermott, F., 1993. Mantle and slab from the viewpoint of the dehydration embrittlement hypothesis. Our contributions in arc magmas. Annual Reviews for Earth and Planetary Sciences 21, 175–204. model demonstrates that subducting slab inevitably undergoes partial Hochstaedter, A., Gill, J., Peters, R., Broughton, P., Holden, P., 2001. Across-arc melting in the basaltic portion. Hydrous minerals in mantle wedge are geochemical trends in the Izu-Bonin arc: contributions from the subducting slab. stable only in shallow (b120 km) part of the wedge corner, and no Geochemistry, Geophysics, Geosystems 2 2000GC000105. Ishikawa, T., Nakamura, E., 1994. Origin of the slab component in arc lavas from across- dehydration reactions occur at deeper areas in the mantle wedge. arc variation of B and Pb isotopes. Nature 370, 205–208. Aqueous fluids are supplied into the mantle wedge chieflyby Ishizuka, O., Taylor, R.N., Milton, J.A., Nesbitt, R.W., 2003. Fluid–mantle interaction in an temperature-dependent decomposition reactions of serpentine, talc, intra-oceanic arc: constraints from high-precision Pb isotopes. Earth and Planetary Science Letters 211, 221–236. brucite and chlorite in subducting slab. Partial melting occurs in a Iwamori, H., 1998. Transportation of H2O and melting in subduction zones. Earth and wide area of mantle wedge, from just above subducting slab to just Planetary Science Letters 160, 65–80. below volcanic front. The formation of volcanic front is controlled by Iwamori, H., 2004. Phase relations of peridotites under H2O-saturated conditions and distribution of partial melts in mantle wedge, which is governed by ability of subducting plates for transportation of H2O. Earth and Planetary Science Letters 227, 57–71. fl dynamics of mantle ow and melt migration. Petrologic constraints do Iwamori, H., 2007. Transportation of H2O beneath the Japan arcs and its implications for not provide a rationale for the constancy of slab depth beneath global water circulation. Chemical Geology 239, 182–198. volcanic front. Iwamori, H., Zhao, D., 2000. Melting and seismic structure beneath the northeast Japan arc. Geophysical Research Letters 27, 425–428. Johnson, M.C., Plank, T., 1999. Dehydration and melting experiments constrain the fate of subducted sediments. Geochemistry, Geophysics, Geosystems 13 Acknowledgments 1999GC000014. Jung, H., Green, H.W., Dobrzhinetskaya II, L.F., 2004. Intermediate-depth earthquake faulting by dehydration embrittlement with negative volume change. Nature 428, We are extremely grateful to Junichi Nakajima and Saeko Kita for 545–549. providing us with their up-to-date seismic data and related figures. Kawakatsu, H., Watada, S., 2007. Seismic evidence for deep-water transportation in the We deeply thank Prof. Hasegawa for his outstanding seismic studies, mantle. Science 316, 1468–1471. which stimulated us to review subduction zone magmatism. The Kamiya, S., Kobayashi, Y., 2000. Seismological evidence for the existence of serpentinized wedge mantle. Geophysical Research Letters 27, 819–822. model proposed here has been developed through extensive discus- Kay, R.W., 1980. Implication of a melting-mixing model for element recycling in the sions in the subduction zone meetings held by Prof. Hasegawa's and crust–upper mantle system. Journal of Geology 88, 497–522. Maruyama's groups. We thank Profs. Hasegawa, Matsuzawa, Umino, Kelemen, P.B., Rilling, J.L., Parmentier, E.M., Mehl, L., Hacker, B.R., 2003. Thermal structure due to solid-state flow in the mantle wedge beneath arcs. In: Eiler, J.M. Zhao, and other members who attended the meetings and for (Ed.), Inside the Subduction Factory. In: Geophysical Monograph, vol.138. American constructive and helpful discussion. Discussions with Katsuya Kaneko Geophysical Union, Washington, D. C., pp. 293–311. fl and Mamoru Kato were helpful to improve our model. The manuscript Kimura, J.I., Yoshida, T., 2006. Contributions of slab uid, mantle wedge and crust to the origin of Quaternary lavas in the NE Japan. Journal of Petrology 47, 2185–2232. has been improved with constructive comments by two anonymous Kirby, S., Engdahl, E.R., Denlinger, R., 1996. Intermediate-depth intraslab earthquakes and reviewers. arc volcanism as physical expressions of crustal and upper mantle metamorphism in 456 T. Kogiso et al. / Gondwana Research 16 (2009) 446–457

subducting slabs. In: Bebout, G.E., Platt, J., Scholl, D.W., Kirby, S. (Eds.), Subduction: Top Okada, T., Hasegawa, A., 2000. Activity of deep low-frequency microearthquakes and to Bottom. In: Geophysical Monograph, vol. 96. American Geophysical Union, their moment tensor in northeastern Japan (in Japanese with English abstract). Washington, D.C. Bulletin of Volcanological Society of Japan 45, 47–63. Kita, S., Okada, T., Nakajima, J., Matsuzawa, T., Hasegawa, A., 2006. Existence of a seismic Okamoto, K., Maruyama, S., 2004. The eglogite–garnetite transformation in the MORB +

belt in the upper plane of the double seismic zone extending in the along-arc H2O system. Physics of the Earth and Planetary Interiors 146, 283–296. direction at depths of 70–100 km beneath NE Japan. Geophysical Research Letters Omori, S., Komabayashi, T., 2007. Subuduction zone: the water channel to the mantle. 33. doi:10.1029/2006GL028239. In: Yuen, D.A., Maruyama, S., Karato, S., Windley, B.F. (Eds.), Superplumes: Beyond Kita, S., Okada, T., Nakajima, J., Matsuzawa, T., Hasegawa, A., Katsumata, K., 2008. Plate Tectonics. InSpringer, Dordrecht, pp. 113–138. Anomalous deepening of upper-plane seismic belt in the Pacific slab beneath the Omori, S., Kamiya, S., Maruyama, S., Zhao, D., 2002. Morphology of the intraslab seismic Hokkaido corner. 7th General Assembly of Asian Seismological Commission and zone and devolatilization phase equilibria of the subducting slab peridotite. Bulletin Seismological Society of Japan, Tsukuba, Japan. of Earthquake Research Institute, University of Tokyo 76, 455–478. Kogiso, T., Hirschmann, M.M., Frost, D.J., 2003. High-pressure partial melting of garnet Omori, S., Komabayashi, T., Maruyama, S., 2004. Dehydration and earthquakes in the pyroxenite: possible mafic lithologies in the source of ocean island basalts. Earth subducting slab: empirical link in intermediate and deep seismic zones. Physics of and Planetary Science Letters 216, 603–617. the Earth and Planetary Interiors 146, 297–311. Kogiso, T., Hirschmann, M.M., Pertermann, M., 2004. High pressure partial melting of Omori, S., Kita, S., Maruyama, S., Santosh, M., 2009. Pressure-temperature condition of mafic lithologies in the mantle. Journal of Petrology 45, 2407–2422. ongoing regional metamorphism beneath the Japanese Islands. Gondwana Kogiso, T., Hirschmann, M.M., 2006. Partial melting experiments of bimineralic eclogite Research 16, 458–469 (this issue). and the role of recycled mafic oceanic crust in the genesis of ocean island basalts. Peacock, S.M., 2003. Thermal structure and metamorphic evolution of subducting slabs. Earth and Planetary Science Letters 249, 189–199. In: Eiler, J.M. (Ed.), Inside the Subduction Factory. In: Geophysical Monograph, vol. Komabayashi, T., Omori, S., Maruyama, S., 2004. Petrogenetic grid in the system MgO– 138. American Geophysical Union, Washington, D. C., pp. 7–22.

SiO2–H2O up to 30 GPa,1600 °C: applications to hydrous peridotite subducting into the Raleigh, C.B., 1967. Tectonic implications of serpentine weakening. Geophysical Journal Earth's deep interior. Journal of Geophysical Research 109, B03206. doi:10.1029/ of the Royal Astronomical Society 14, 113–118. 2003JB002651. Rapp, R.P., Watson, E.B.,1995. Dehydration melting of metabasalt at 8–32 kbar; implications Kondo, H., Kaneko, K., Tanaka, K.,1998. Characterization of spatial and temporal distribution for continental growth and crust–mantle recycling. Journal of Petrology 36, 891–931. of volcanoes since 14 Ma in the Northeast Japan arc. Bulletin of Volcanological Society of Robinson, J.A.C., Wood, B.J., Blundy, J.D., 1998. The beginning of melting of fertile and Japan 43, 173–180. depleted peridotite at 1.5 GPa. Earth and Planetary Science Letters 155, 97–111. Kubo, K., 2003. Melting experiments on fertile and depleted peridotites: chromian Rushmer, T., 1991. Partial melting of two amphibolites: contrasting experimental results spinel, index for melting processes. Ph.D. Thesis, Tokyo Institute of Technology. under ﬂuid-absent conditions. Contributions to Mineralogy and Petrology 107, Kuno, H., 1959. Origin of Cenozoic petrographic provinces of Japan and surrounding 41–59. areas. Bulletin of Volcanology 20, 37–76. Sakuyama, M., Nesbitt, R.W., 1986. Geochemistry of the Quaternary volcanic rocks of the Kuno, H., 1960. High-alumina basalt. Journal of Petrology 1, 121–145. northeast Japan arc. Journal of Volcanology and Geothermal Research 29, 413–450. Kushiro, I., 1968. Composition of magmas formed by partial zone melting of the Earth's Schmidt, M.W., Poli, S., 1998. Experimentally based water budgets for dehydrating slabs upper mantle. Journal of Geophysical Research 73, 619–634. and consequences for arc magma generation. Earth and Planetary Science Letters Kushiro, I., 1969. The system forsterite–diopside–silica with and without water at high 163, 361–379. pressures. American Journal of Science 267A, 269–294. Seno, T., Yamanaka, Y., 1996. Double seismic zones, compressional deep trench-outer Kushiro, I., 1983. On the lateral variation in chemical composition and volume of rise events and superplumes. In: Bebout, G.E., Platt, J., Scholl, D.W., Kirby, S. (Eds.), Quaternary volcanic rocks across Japan arc. Journal of Volcanology and Geothermal Subduction: Top to Bottom. In: Geophysical Monograph, vol. 96. American Research 18, 435–447. Geophysical Union, Washington D.C. Kushiro, I., 1987. Viscosity of partial melts in the upper mantle. Journal of Geophysical Shibata, T., Nakamura, E., 1997. Across-arc variations of isotope and trace element Research 91, 9343–9350. compositions from Quaternary basaltic rocks in northeastern Japan: implication for Laporte,D.,Toplis,M.J.,Seyler,M.,Devidal,J.L.,2004.Anewexperimentaltechniquefor interaction between subducted oceanic slab and mantle wedge. Journal of extracting liquids from peridotite at very low degrees of melting: application to Geophysical Research 102, 8051–8064. partial melting of depleted peridotite. Contributions to Mineralogy and Petrology 146, Spiegelman, M., McKenzie, D., 1987. Simple 2-D models for melt extraction at mid-ocean 463–484. ridges and island arcs. Earth and Planetary Science Letters 83, 137–152. Liu, J., Bohlen, S.R., Ernst, W.G., 1996. Stability of hydrous phases in subducting oceanic Sugimura, A.,1960. Zonal arrangement of some geophysical and petrological features in Japan crust. Earth and Planetary Science Letters 143, 161–171. and its envisions. Journal of Faculty of Science, University of Tokyo, Series II 12, 133–153. Marsh, B.D., 1979. Island arc development: some observations, experiments, and specula- Sugimura, A., Matsuda, T., Chinzei, K., Nakamura, K., 1963. Quantitative distribution of tions. Journal of Geology 87, 687–713. late Cenozoic volcanic materials in Japan. Bulletin of Volcanology 26, 125–140. Matsuzawa, T., Umino, N., Hasegawa, A., Takagi, A., 1986. Upper mantle velocity Tatsumi, Y., 1986. Formation of volcanic front in subduction zones. Geophysical Research structure estimated from PS-converted wave beneath the northeastern Japan arc. Letters 13, 717–720. Geophysical Journal of the Royal Astronomical Society 86, 767–787. Tatsumi, Y., 1989. Migration of ﬂuid phases and genesis of basalt magmas in subduction McBirney, A.R., 1969. Andesitic and rhyolitic volcanism of orogenic belts. In: Hart, P.J. zones. Journal of Geophysical Research 94, 4697–4707. (Ed.), The Earth's Crust and Upper Mantle. In: Geophysical Monograph, vol. 13. Tatsumi, Y., Eggins, S.M., 1995. Subduction Zone Magmatism. Blackwell, Cambridge. American Geophysical Union, Washington D.C., pp. 501–506. Tatsumi, Y., Kogiso, T., 1997. Trace element transport during dehydration processes in Mibe, K., Kanzaki, M., Kawamoto, T., Matsukage, K.N., Fei, Y., Ono, S., 2007. Second the subducted oceanic crust: 2. origin of chemical characteristics in arc magmas.

critical endpoint in the peridotite–H2O system. Journal of Geophysical Research 112, Earth and Planetary Science Letters 148, 207–221. B03201. doi:10.1029/2005JB004125. Tatsumi, Y., Sakuyama, M., Fukuyama, H., Kushiro, I., 1983. Generation of arc basalt Miyashiro, A., 1974. Volcanic rock series in island arcs and active continental margins. magmas and thermal structure of mantle wedge in subduction zones. Journal of American Journal of Science 274, 321–335. Geophysical Research 88, 5815–5825. Molnar, P., England, P., 1995. Temperatures in zones of steady-state underthrusting Tonegawa, T., Hirahara, K., Shibutani, T., Iwamori, H., Kanamori, H., Shiomi, K., 2008. of young oceanic lithosphere. Earth and Planetary Science Letters 131, 57–70. Water flow to the mantle transition zone inferred from a receiver function image of Nakajima, J., Hasegawa, A., 2003. Estimation of thermal structure in the mantle wedge the Pacific slab. Earth and Planetary Science Letters 274, 346–354. of northeastern Japan from seismic attenuation data. Geophysical Research Letters Tsuji, Y., Nakajima, J., Hasegawa, A., 2008. Tomographic evidence for hydrated oceanic 14, 1760. doi:10.1029/2003GL017185. crust of the Pacific slab beneath northeastern Japan: implications for water Nakajima, J., Hasegawa, A., 2007. Tomographic evidence for the mantle upwelling transportation in subduction zones. Geophysical Research Letters 35, L14308. beneath southwestern Japan and its implications for arc magmatism. Earth and doi:10.1029/2008GL034461. Planetary Science Letters 154, 90–105. Tsumura, N., Matsumoto, S., Horiuchi, S., Hasegawa, A., 2000. Three-dimensional Nakajima, J., Matsuzawa, T., Hasegawa, A., Zhao, D., 2001. Three-dimensional structure attenuation structure beneath the northeastern Japan arc estimated from spectra of of Vp, Vs, and Vp/Vs beneath northeastern Japan: implications for arc magmatism small earthquakes. Tectonophysics 319, 241–260. and fluids. Journal of Geophysical Research 106, 843–857. Ueki, K., Iwamori, H., 2007. Melting condition and origin of arc magmas beneath the Nakajima, J., Takei, Y., Hasegawa, A., 2005. Quantitative analysis of the inclined low- Sengan region, Northeastern Japan. Geochimica et Cosmochimica Acta 71, A1045. velocity zone in the mantle wedge of northeastern Japan: a synthetic change of Umino, N., Hasegawa, A.,1975. On the two-layered structure of a deep seismic plane in the melt-filled pore shapes with depth and its implications for melt migration. Earth northeastern Japan arc (in Japanese with English abstract). Journal of Seismological and Planetary Science Letters 234, 59–70. Society of Japan 28, 125–139. Nakajima, J., Tsuji, Y., Hasegawa, A., 2009. Seismic evidence for thermally-controlled van Keken, P.E., Kiefer, B., Peacock, S.M., 2002. High-resolution models of subduction dehydration reaction in subducting oceanic crust. Geophysical Research Letters 36, zones: implications for mineral dehydration reactions and the transport of water L03303. doi:10.1029/2008GL036865. into the deep mantle. Geochemistry, Geophysics, Geosystems 3, 1056. doi:10.1029/ Nakamura, H., Iwamori, H., Kimura, J., 2008. Geochemical evidence for enhanced fluid 2001GC000256. flux due to overlapping subducting plates. Nature Geoscience 1, 380–384. Waldhauser, F., Ellsworth, W.L., 2000. A double-difference earthquake location algorithm: Nicholls, I.A., Ringwood, A.E., 1972. Production of silica-saturated magmas in island arcs. method and application to the northern Hayward fault, California. Bulletin of the Earth and Planetary Science Letters 17, 243–246. Seismological Society of America 90, 159–177. Nohda, S., Wasserburg, G.J., 1981. Nd and Sr isotopic study of volcanic rocks from Japan. Wasylenki, L.E., Baker, M.B., Kent, A.J.R., Stolper, E.M., 2003. Near-solidus melting of the Earth and Planetary Science Letters 52, 264–276. shallow upper mantle: partial melting experiments on depleted peridotite. Journal O'Hara, M.J., 1968. The bearing of phase equilibria studies in synthetic and natural of Petrology 44, 1163–1191. systems on the origin and evolution of basic and ultrabasic rocks. Earth-Science Wolf, M.B., Wyllie, P.J., 1994. Dehydration melting of amphibolite at 10 kbar: effect of Reviews 4, 69–133. temperature and time. Contributions to Mineralogy and Petrology 115, 369–383. T. Kogiso et al. / Gondwana Research 16 (2009) 446–457 457

Wyllie, P.J., Sekine, T., 1982. The formation of mantle phlogopite in subduction zone Zhao, D., Hasegawa, A., 1993. P-wave tomographic imaging of the crust and upper hybridization. Contributions to Mineralogy and Petrology 79, 375–380. mantle beneath the Japan Islands. Journal of Geophysical Research 98, 4333–4353. Yamasaki, T., Seno, T., 2003. Double seismic zone and dehydration embrittlement of the Zhao, D., Hasegawa, A., Horiuchi, S., 1992. Tomographic imaging of P and S wave velocity subducting slab. Journal of Geophysical Research 108. doi:10.1029/2002JB001918. structure beneath northeastern Japan. Journal of Geophysical Research 98, 4333–4353. Zhang, H., Thurber, C.H., 2003. Double-difference tomography: the method and its Zhao, D., Hasegawa, A., Kanamori, H., 1994. Deep structure of Japan subduction zones as application to the Hayward Fault, California. Bulletin of the Seismological Society of derived from local, regional, and teleseismic events. Journal of Geophysical Research America 93, 1875–1889. 99, 22313–22329.