wattrans

(2007) 148–165www.elsevier.com/locate/gr

Gondwana Research 11

Water transportation from the subducting slab into the mantle transition zone

Shigenori Maruyama a,⁎, Kazuaki Okamoto b

a Department of Earth and Planetary Sciences, Tokyo Institute of Technology, O-okayama 2-12-1, Meguroku, Tokyo 152, Japanb Institute of Earth Sciences, Academia Sinica, P.O. Box 120, Academic Road Sec. 2, Nankang, Taipei, Taiwan, ROC

Received 1 June 2006; accepted 5 June 2006Available online 14 August 2006

Abstract

Using a recently developed petrogenetic grid for MORB+H2O, we propose a new model for the transportation of water from the subductingslab into the mantle transition zone. Depending on the geothermal gradient, two contrasting water-transportation mechanisms operate at depth in asubduction zone. If the geothermal gradient is low, lawsonite carries H2O into mantle depths of 300 km; with further subduction down to themantle transition depth (approximately 400 km) lawsonite is no longer stable and thereafter H2O is once migrated upward to the mantle wedgethen again carried down to the transition zone due to the induced convection. At this depth, hydrous β-phase olivine is stable and plays a role as ahuge water reservoir. In contrast, if the geothermal gradient is high, the subducted slab may melt at 700–900 °C at depths shallower than 80 km toform felsic melt, into which water is dissolved. In this case, H2O cannot be transported into the mantle below 80 km. Between these two end-member mechanisms, two intermediate types are present. In the high-pressure intermediate type, the hydrous phase A plays an important role tocarry water into the mantle transition zone. Water liberated by the lawsonite-consuming continuous reaction moves upward to form hydrous phaseA in the hanging wall, which transports water into deeper mantle. This is due to a unique character of the reaction, because Phase A can becomestable through the hydration reaction of olivine. In the case of low-pressure intermediate type, the presence of a dry mantle wedge below 100 kmacts as a barrier to prevent H2O from entering into deeper mantle.© 2006 Published by Elsevier B.V. on behalf of International Association for Gondwana Research.

Keywords: MORB+H2O phase diagram; Subduction zone geotherms; Peridotite+H2O phase diagram; Water transportation to deeper mantle

1. Introduction

The role of water in the genesis of magma in subduction zoneenvironments has been widely recognized (e.g. Wyllie, 1988).The dehydration has been attributed either to direct melting of theslab (e.g. Nicolls and Ringwood, 1973; Wyllie and Sekine, 1982;Brophy andMarsh, 1986) or to the release of fluids to decrease inthe melting temperature of the mantle wedge (e.g. Ringwood,1975; Delany and Helgeson, 1978; Tatsumi et al., 1986).Dehydration due to isobaric amphibole breakdown, traditionallytaken as the amphibolite–eclogite transformation reaction, isbelieved to provide the H2O necessary to trigger partial melting ofthe overlying mantle wedge. However, several high-pressurehydrous minerals have been documented to be stable at coesite

⁎ Corresponding author. Fax: +81 03 5734 3538.E-mail address: [email protected] (S. Maruyama).

1342-937X/$ - see front matter © 2006 Published by Elsevier B.V. on behalf of Indoi:10.1016/j.gr.2006.06.001

and diamond isograds in many ultrahigh-pressure metamorphicterranes (see reviews by e.g., Liou et al., 1994; Chopin andSobolev, 1995; Schreyer, 1995). Recent experimental studies(Schreyer, 1988; Poli, 1993; Poli and Schmidt, 1995; Okamotoand Maruyama, 1999) show that hydrous phases such aslawsonite, chloritoid, staurolite, phengite and zoisite–clinozoisiteare passive transporters of considerable amounts of water insubducting oceanic crust to depths greater than that of amphibolestability.

It is important to determine the precise dehydration depth in thesubducting oceanic plate and the mechanism of water trans-portation into the deepmantle, because several recent experimentssuggest that H2O can be stored in several hydrous phases (A, E,hydrous-β, and -γ) in the mantle transition zone (e.g. Inoue et al.,1995; Kawamoto et al., 1996; Smyth and Kawamoto, 1997). Theaim of this study is to propose H2O transportation mechanismsinto the mantle transition zone, combining the experimentalresults for phase diagrams of MORB+H2O and peridotite+H2O,

ternational Association for Gondwana Research.

mailto:[email protected]

http://dx.doi.org/10.1016/j.gr.2006.06.001

149S. Maruyama, K. Okamoto / Gondwana Research 11 (2007) 148–165

together with an additional data set on the T-distribution on theslab surface.

2. Phase diagram of MORB+H2O system

Okamoto and Maruyama (1999) defined the high-pressurestability limit of lawsonite in the MORB+H2O system (Fig. 1).The lawsonite–eclogite facies is bounded by the high-pressurestability limit of a lawsonite–involving a continuous reactiondefined by the compositional enlargement of garnet towards agrossular-rich end member with increasing pressure and/ortemperature. Lawsonite in metabasite is stable up to 10 GPa attemperatures lower than 900 °C.

Fig. 1. A phase diagram of the MORB+H2O system (Modified after Okamoto and M(1) quartz/coesite (Bohlen and Boettcher, 1982), (2) coesite/stishovite (Zhang et al.,albite= jadeite+quartz (modified from Holland, 1983, see Maruyama et al., 1996). Ththe MORB+H2O system (Poli and Schmidt, 1995; Okamoto and Maruyama, 1999). Dsystem. Facies boundaries below 4 GPa are after Maruyama et al. (1996). Abbreviaamphibolite facies, AM=amphibolite facies, GR=granulite facies, HGR=high-pressuEc=zoisite eclogite facies.

The so-called eclogite facies is subdivided into five facies.The dry eclogite facies has the largest P–T space and is boundedby the jadeite eclogite and garnetite facies at P higher than16 GPa (Irifume et al., 1986; Okamoto and Maruyama, 2004)and the dry solidus on the high-T side. The low-pressure limit oflawsonite-bearing eclogite facies is bounded by the zoisiteeclogite subfacies involving a reaction with a gentle positiveslope at 2–3 GPa (Poli and Schmidt, 1995). The low-pressureand low-temperature boundary is separated by a gentle reactioncurve with a negative Clapeyron slope from the blueschist facies(Maruyama et al., 1996; Maruyama, 1997). The amphibole–eclogite facies is stable below the zoisite–eclogite facies, downto 1.0 GPa. Both the zoisite– and amphibole–eclogite facies are

aruyama, 1999, 2004). Thin solid lines delineate polymorphic transformation of1996), and (3) diamond/graphite (Bundy, 1980), and (4) an univariant reaction,e pale-colored shadow area delineates the stability field of lawsonite eclogite inark-colored shadow area delineates the slab melting region in the MORB+H2O

tions; PA=pumpellyite–actinolite facies, GS=greenschist facies, EA=epidote–re granulite facies, BS=blueschist facies, Am Ec=amphibole eclogite facies, Zo

150 S. Maruyama, K. Okamoto / Gondwana Research 11 (2007) 148–165

terminated by wet solidus at T higher than 650 °C. The limits ofmetamorphic facies below 2.0 GPa are after Maruyama et al.(1996) and Maruyama (1997).

3. Fate of water from mid-oceanic ridge to subduction zone

3.1. The average H2O content of hydrated oceanic crust

New oceanic crust is born at mid-oceanic ridges, and suffershydrothermal fluid-alteration (“ocean-floor metamorphism” ofMiyashiro, 1973). The crust generated at a mid-oceanic ridge isabout 7 km thick, and the top 800 m is hydrated by reacting withthe circulating hydrothermal water (Fig. 2). Beneath the 800 mdepth, the crust is essentially dry except along fracture zones.The H2O added to the oceanic crust by hydrothermal meta-morphism is difficult to estimate precisely, simply because itdepends on the degree of development of hydrothermal fracturesystems at the mid-oceanic ridge, which we cannot directlyobserve. The H2O is fixed as loosely bounded water in sheetsilicates or pore space and structural water in the crystallinephases. The measured structural water from the DSDP hole504B (Alt et al., 1986) is approximately 1 to 2 wt.% from thetop of the pillow lava down to 600 m. However, it increases to

Fig. 2. Water circulation sys

4–6 wt.% from 600 down to 800 m depth (the transition fromthe pillow to dike zone). The structural water of the dredgedbasaltic samples by multiple regression analysis also rangesfrom 1 to 8 wt.% (Hart, 1976). Staudigel et al. (1995) estimatedthe structural water at about 2.7 wt.% and the interstitial wateris at least 2 wt.%, based on the void volume of the crust (ap-proximately 5 vol.% by Johnson, 1979). The present study uses5–6 wt.% H2O

+ which is stored in hydrous minerals due toocean-floor hydrothermal metamorphism.

Dredged gabbros and ultramafic rocks from fracture zonesare highly altered and suffered high-grade greenschist to am-phibolite facies metamorphism. These rocks contain severalkinds of hydrous minerals. However, such hydrated rocks arerestricted to fracture zones (Fig. 2).

During transport of oceanic crust from the mid-ocean ridgeto the trench, no significant metamorphic and dynamic eventsoccur between mid-oceanic ridge and subduction zone if OIBvolcanism does not affect the MORB crust (Maruyama, 1991).Fluid trapped in the pore space of the subducting oceanic crustand overlying sediments is expelled within the 10–15 km depthrange (e.g. Moore, 1992). Loosely bounded water (H2O

−) isalso released at shallow depths between 15 to 20 km. Subse-quently, the subducted oceanic crust suffers subduction-zone

tem and plate tectonics.

Fig. 3. P–T diagram showing the estimated H2O wt.% of the subducted oceanic crust (modified after Okamoto and Maruyama, 2004). Two representative geotherms(young vs. old slabs) along the slab surface are shown.


metamorphism and continues to release H2O, by dehydrationreactions, to the overlying mantle wedge (Fig. 2). The change inH2O content by dehydration reactions due to subduction zonemetamorphism will be reviewed below.

3.2. The maximum H2O contents in the MORB+H2O system

The maximum H2O content in the hydrated subductingoceanic crust can be calculated, based on the modal variation ofhydrous minerals in metabasalt for given metamorphic facies(e.g. Peacock, 1993). The mineralogical changes in metabasaltas a function of P and T are best illustrated using the concept ofmetamorphic facies (Eskola, 1920), because most hydrousminerals in metabasalt break down by a complex series ofcontinuous reactions. Peacock (1993) calculated the maximumH2O content in each metamorphic facies, except for the eclogitefacies, in the NCMASH system. He calculated a metamorphicmineral norm (that is similar to the CIPW norms) in order toestimate the maximum H2O content in each of the metamorphic

facies. He treated the eclogite facies as completely dry, and fateof hydrous minerals.

However, to estimate the amount of H2O transported fromthe surface to the mantle by a subducting oceanic plate, both theblueschist–eclogite and the epidote amphibolite–eclogitetransformations are of critical importance. The change in theamount of H2O due to the blueschist – lawsonite–eclogitetransformation was estimated to be 3 wt.% by Poli and Schmidt(1995) and that of the amphibolite–eclogite facies transforma-tion to be 1.5 wt.% by Poli (1993). They calculated the mineralproportions employing a mass balance program which includesa Monte Carlo error propagation. In this study modal wt.% ofeach phase was calculated with the least square method formajor elements. The modal wt.% of lawsonite at 8 GPa, 700 °Cwas estimated to 4.0 which corresponds to 0.5 wt.% of water inthe hydrated MORB crust (Okamoto and Maruyama, 1999).The maximum H2O content of hydrated MORB crust has beenestimated by Peacock (1993), Poli (1993) and Poli and Schmidt(1995) to be below 6 GPa, and those are compiled in Fig. 3.


Based on the previous experimental studies, most workersconsidered that most H2O in the subducting oceanic crust isreleased by the amphibolite – eclogite transformation at 60 kmdepth, and less than 1 wt.% H2O is retained beneath the volcanicfront (at ca. 100–120 km depth), if the temperature is below650 °C (within the stability field of lawsonite–eclogite facies).However, if the temperature is above 650 °C at 30–90 kmdepths, subducting oceanic crust reaches the solidus tempera-ture condition, causing the partial melting of amphibole-bearingmetamorphosed oceanic crust. The released H2O dissolves inthe melt; then the remaining subducting oceanic crust trans-forms to dry eclogite. If the eclogite becomes anhydrous, it willnot melt due to the high-T solidus over than 1500 °C at about200 km depth. The H2O content in the subducting oceanic crustdepends on the variation of P–T path along the slab surface.

4. T-distribution on the slab surface

The fate of water in subducting oceanic lithosphere, thus,strongly depends on the T distribution in the subducting slab. Foran accurate estimate, it is critical to apply a multi-disciplinaryapproach, because numerical simulations cannot give a uniquesolution, because there are too many independent parameters tocontrol the thermal structure along the slab surface. In thefollowing, we first discuss the approach by numerical simulation,then summarize the field constraints, and finally define thepredominant parameters which control the thermal structure.

4.1. Numerical simulation

Models presented over the past 25 years demonstrate that thethermal structure of a subduction zone depends on numerousparameters, including: (1) the thermal structure (age) of theincoming lithosphere, (2) the convergence rate, (3) the geometryof subduction, specifically subduction dip angle, (4) thedistribution of radiogenic heat-producing elements, (5) the rateof shear heating along the subduction shear zone, and (6) thegeometry and vigor of induced convection in the overlyingmantle wedge (Fig. 4) (e.g. Oxburgh and Turcotte, 1970; Hasebeet al., 1970; Toksöz et al., 1971; Turcotte and Schubert, 1973,Anderson et al., 1978; Hsui and Toksöz, 1980; Honda andUyeda, 1983; Wang and Shi, 1984; Cloos, 1985; Honda, 1985;

Fig. 4. Schematic cross section of subduction zone and the parameters to contro

van den Beukel and Wortel, 1988; Toksöz and Hsui, 1989;Peacock, 1990, 1991; Davies and Stevenson, 1992; Staudigeland King, 1992; Furukawa, 1993a,b; Peacock et al., 1994).Moreover, although poorly known, (7) the role of fluiddehydrated from the downgoing slab (Anderson et al., 1976,1978; Delany and Helgeson, 1978), and (8) the exothermic orendothermic reaction within it would affect the thermal structureof the top of the downgoing slab.

Proposed P–T paths for the oceanic crust at the top of thesubducted slab differ sharply in the various models shown in theabove papers. The results of selected numerical models aresummarized by Peacock (1996). Although temperature dis-tributions within the mantle wedge and in the subducted slab arerelatively similar in all of these models, dramatic differencesoccur at the interface between the mantle wedge and thesubducted slab. In the absence of shear heating, estimates of thetemperature in the subduction shear zone at 100 km depthranges from 300 to 750 °C. The lower calculated temperaturesresult from experiments in which the mantle wedge is assumedto be rigid. Qualitatively, induced convection in the mantlewedge brings warm mantle material into close proximity to thesubducting slab; induced convection warms the subducting slabat the expense of cooling of the adjacent mantle wedge. Mostmodels suggest that induced mantle–wedge convection heatsthe top of the slab by several hundred degrees. In the Peacock etal. (1994) estimation, induced convection increases thetemperatures of the slab surface at 100 km depth from 150 °Cto 450 °C (Fig. 5a).

In the absence of shear heating, faster convergence ratesresult in cooler subduction shear zone temperatures (Fig. 5b).Below 70 km depth and above 100 km depth, temperatures ofthe faster subducting slabs are lower than that of the slowersubducting slabs. At 100 km depth, differing convergence ratesdo not cause a large temperature difference on the slab surface;T=450 °C if the V=100 mm/yr and T=550 °C if theV=10 mm/yr.

The thermal structure (age) of the oceanic lithosphere prior tosubduction strongly influences the thermal structure of thesubduction zone (Fig. 5c). If the age is 50 Myr, the temperatureof the slab surface is approximately 450 °C at 100 km depth. Ifthe age is less than 2 Myr, although the convergence rate isas fast as 100 mm/yr, subducting slab causes partial melting.

l the thermal structure of subduction zone (modified from Peacock, 1996).

Fig. 5. Steady-state P–T paths along the subduction shear zone calculated by Peacock et al. (1994). (a) Calculated steady-state P–T conditions based on analyticalexpressions (Molnar and England, 1990) and numerical calculations (Peacock et al., 1994). V=100 mm/yr. Age of subducting lithosphere=50 Ma. (b) Subductionshear zone T conditions for the two different convergence rates (10 and 100 mm/yr). Age of incoming lithosphere=50 Ma. (c) Subduction shear zone P–T conditionsas a function of age of the subducting lithosphere. τ=0. V=100 mm/yr. (d) Subduction zone P–T conditions for different shear stresses. Age of incominglithosphere=50 Ma.


Temperature at 100 km depth is 750–1050 °C (Fig. 5c). If the ageof the oceanic plate is older than 50 Myr and shear stress isnegligible, the top of the subducting oceanic plate suffersblueschist facies metamorphism at depth range from 15 km to60–70 km, and transforms to lawsonite–eclogite thereafter. Theaverage age of oceanic plate is approximately 100 Ma on themodern Earth, and the oldest plate dates back to 200 Ma.

Shear heating, caused by shear stresses of the order of100 MPa, dramatically increases calculated temperatures in thesubduction shear zone to values in excess of 1000 °C at 100 km

depth (e.g.Toksöz et al., 1971; Turcotte and Schubert, 1973)(Fig. 5d). Subduction-zone shear stress is poorly known. Recentestimates of shear stresses in subduction zones range from about100 MPa (Honda, 1985; Scholz, 1990; Molnar and England,1990) to several tens of MPa (Bird, 1978; van den Beukel andWortel, 1988; Peacock, 1992; Titchelaar and Ruff, 1993) toapproximately zero (Hyndman and Wang, 1993), and isnegligible at depths greater than 50–60 km, where thesubducting crust and the overlying mantle are mechanicallycoupled (Furukawa, 1993b). Furthermore, shear heating may be


absorbed by dehydration reactions taking place in the subduct-ing slab (Anderson et al., 1976, 1977; Delany and Helgeson,1978). Presence of H2O also lowers rock strengths and frictionaldissipation even lowers shear stress at depth.

Additional variables that lead to significant uncertainties incalculating the thermal structure of subduction zones includehydrothermal circulation in the oceanic lithosphere, the variationof thermal conductivity with T and P, the amount of heat con-sumed and released by metamorphic reactions, and the effects offluid flow at shallow depths.Metamorphic dehydration reactionsin the subducting slab may consume significant amounts of heat(Anderson et al., 1976, 1978; Delany and Helgeson, 1978).

Fig. 6. (a) Yor Yb content in the adakites and island arc volcanics as a function of agethe top surface of young slab calculated by Peacock et al., 1994).

Similarly, hydration reactions in the overlying mantle wedgemay release significant amounts of heat (Peacock, 1987).Thermal calculations by Peacock (1990) that assumed 2 wt.%bound H2O in the subducting oceanic crust indicate that meta-morphic reactions have only a minor effect on the calculatedthermal structure (<5 °C).

Summarizing the numerical simulation, the critical factors tocontrol the thermal structure of the subducting lithosphere are theage of the lithosphere, the convergence rate and the inducedmantle convection. Among these factors, the age the lithospherehas systematic spatial variation in the present- and paleo-subduction zones in circum-Pacific region. Shear stress is difficult

of subducting lithosphere (Drummond and Defant, 1990). (b) P–T conditions on


to estimate, but several studies show that it is less than several tensofMPa, giving a negligible effect at depth larger than 50 km. Thuswe further argue the P–T conditions in various subduction zonesreconstructed from the age of the lithosphere.

4.2. Field constraints

4.2.1. (1) Adakites from slab melting of MORBThe trace element and REE patterns of adakites and high-Al

Archean TTG suites suggest that they are derived from partialmelting of a garnet-bearing MORB source (Kay, 1978; Martin,1986). Recently, several researchers have described adakites inmodern-arc environments where <25 Ma oceanic crust is cur-rently being subducted (Fig. 6a; Drummond and Defant, 1990).These recent volcanics crop out between the main calc–alkalinearc and the trench, do not have typical arc geochemicalsignatures, and may represent direct partial melts of subductingoceanic crust. If the source of these volcanics is direct partialmelting of the downgoing slab, then these magmas provide apowerful constraint on the thermal structure of subductionzones. The slab surface of the subducting lithosphere youngerthan 25 Myr may reach temperatures over 650 °C at 50 kmdepth or further; that is, when it reaches the wet solidus in thehigh-P amphibolite and granulite facies (Fig. 6b). Applying theconvergence rate and the age of subducted lithosphere beneathsouthernmost Chile and Mt. St Helens suites, P–T paths for thetop of the subducted lithosphere were estimated (Peacock et al.,1994). The estimated P–T paths (condition satisfying bothinduced mantle wedge convection- and shear heating-free) alsoreach the wet solidus.

4.2.2. (2) T-estimates based on the depth range of theseismogenic zone of subduction thrust faults

Subduction zone faults generate earthquakes over a limiteddepth range. Hyndman et al. (1997) demonstrated that the updip (upper limit) and the down dip (lower limit) of theseismogenic zone are thermally controlled by the slab surface.The subduction zone faults are aseismic in their seaward updipportions and landward downdip of a critical point (Fig. 7; seeabove). The seaward shallow aseismic zone, commonlybeneath accreted sediments, may be a consequence ofunconsolidated sediments, especially stable-sliding smectiteclays. If the clays are dehydrated and transformed to illite, thenthe fault may become seismogenic where the temperaturereaches 100–150 °C (e.g. Hower et al., 1976), at 5–15 kmdepth (Fig. 7; see above). The downdip seismogenic limit forsubduction of young, hot oceanic lithosphere is also temper-ature-controlled. The maximum temperature for seismicbehavior in crustal rocks is 350 °C (brittle–ductile transition).For subduction beneath thin island arc crust and beneathcontinental crust in some areas, the forearc mantle is attached tothe subducting oceanic crust at temperatures higher than the350 °C. In such a case, the effect of the serpentinization isavoided. Using the determined seismogenic zone fromCascadia and Southwest Japan, where very young and hotplates are subducting, and from Alaska and most of Chile,where the forearc mantle is reached in the crust, updip and

downdip temperatures were determined (see Hyndman et al.,1997; Oleskevich et al., 1999) (Fig. 7, below).

For Cascadia and SW Japan, melting P–T conditions of thesubducting slab causing production of the adakite suites are alsoplotted. The P–T path for SW Japan is higher P and lower T thanthat for Cascadia. This may be due to the greater age of thesubducting lithosphere of the westward descending Pacific plate.The P–T paths for Chile (North Chile, Taltal, Coquimbo, andValparaiso) and for Southern Alaska are steeper than those ofCascadia and SW Japan, showing that they are due to the olderthe subducting slab, and the steeper the geotherm along theBenioff thrust. However, the slab descending beneath SouthChile, with the fastest subducting speed (9 cm/yr) of the presentexamples, has the steeper gradient in spite of being younger age(5 Ma) than the Cascadia (6.5–8 Ma) and the SW Japan (0–15 Ma) analogue. The Valparaiso slab (9 cm/yr) also has thesteeper gradient in spite of being younger (37Ma) than the NorthChile, Taltal, and Coquimbo (45–50 Ma) and the South Alaska(50 Ma) plates. This tendency may be partly due to a higherconvergence rate (9 cm/yr vs. 4–6 cm/yr). Generally speaking,all of these data support the idea that the age of the subductingslab dominantly controls the thermal structure on the Benioffplane although high convergent rate is recognized in someyoung slabs. Normal subduction-zone geotherm passes throughthe lawsonite–eclogite facies.

4.2.3. (3) P–T conditions of HP and UHP metamorphic beltsFossil geotherms in paleo-subduction zones are well-

preserved in HP and UHP metamorphic belts all over theworld (Maruyama et al., 1996). Using these fossil geotherms,radiometric and fossil ages, reconstructed paleo-plate geome-tries and convergence rates, we can estimate the dominant fac-tors controlling the thermal structure on the slab surface in a fewwell-constrained metamorphic belts.

The fossil geotherm and age of the subducted oceanic crustare well reconstructed for the Cretaceous high-P–T Sanbagawabelt, SW Japan (Maruyama and Seno, 1986; Isozaki andMaruyama, 1991; Maruyama, 1997). The reconstructed P–Tpath is shown in Fig. 8. The metamorphic field gradients havebeen determined precisely by Banno and his co-workers for thelast three decades (e.g. Banno, 1964; Higashino, 1975; Banno etal., 1978; Banno and Sakai, 1989; Higashino, 1990; Otsuki andBanno, 1990, etc). The facies series ranges from the pumpel-lyite–actinolite to transitional blueschist–greenschist, greens-chist, epidote–amphibolite, and amphibolite facies, and isclassified of the HP intermediate type (Miyashiro, 1961).Eclogite–facies assemblages occur locally in Fe2+-rich high-est-grade metabasites (Banno et al., 1978). Thus, the metamor-phic grade ranges from 0.4–0.5 GPa and 250–300 °C for thelowest grade (Maruyama and Liou, 1983) to 1 GPa and 550–600 °C for the highest grade (Banno and Sakai, 1989).

The youngest fossils from the blueschist–facies rocks in theSanbagawa belt are latest Jurassic to earliest Cretaceousradiolarians (Iwasaki et al., 1984). Radiometric ages rangefrom 120 Ma to 60 Ma, depending on the distance from the rootzone and metamorphic grade (Itaya and Takasugi, 1988; compi-lation by Isozaki and Maruyama, 1991). 40Ar/39Ar dates yield

Fig. 7. Subduction zone P–T conditions deduced from the depth and temperature of the updip and the downdip seismogenic zone (see text for details). The updip anddowndip temperatures are estimated by the temperatures of smectite–illite transformation and of the brittle–ductile transition of the crustal material, respectively (seeHyndman et al., 1997; Oleskevich et al., 1999).


Fig. 8. The fossil geotherm of the high P–T Sanbagawa metamorphic belt (e.g. Banno and Sakai, 1989; Maruyama et al., 1996), and Alpine UHP–HP metamorphicbelt (e.g. Chopin, 1984; Kienast et al., 1991, etc.).


the same results (Takasu and Dallmeyer, 1990; Takasu et al.,1994; Dallmeyer et al., 1995). Recent U–Pb SHRIMP age of theeclogite ranges 120–110 Ma as peak metamorphic age(Okamoto et al., 2004). Thus the ages of formation of theaccretionary complex of the Sanbagawa protoliths and ofmetamorphism are very close to each other, at ca. 145–130 Maand 120–60 Ma respectively, with a clustered metamorphic agepeak at 80 Ma. The plate geometry from the Jurassic to thepresent in the NW Pacific Ocean margin was reconstructed byEngebretson et al. (1985). According to them, the Kula/Pacificmid-oceanic ridge subducted under NE Asia at about 80 Ma.Information recorded in the accretionary fold belt of theShimanto belt indicates that the Kula/Pacific mid-oceanic ridgesubducted underneath Japan at ca. 75 Ma (Taira, 1985),correlating the exhumation of the Sanbagawa high P–T beltfrom the 30 km depth, with approach of the mid-oceanic ridgeto the subduction zone (Maruyama, 1997). Convergence rate ofthe plate was estimated to be approximately 202 mm/yr(Engebretson et al., 1985).

Another example is from the Alps (Fig. 8). The fossilgeotherm of the subduction of the Piemont Oceanic crustbeneath the Apulian plate is well-preserved in the HP–UHPmetamorphic belt of the Western Alps (e.g. Chopin, 1984;Kienast et al., 1991, etc.). The metamorphic facies rangefrom the pumpellyite–actinolite to blueschist, amphibole–eclogite facies, and zoisite–eclogite facies (see Maruyama etal., 1996). The Piemont oceanic plate was born at ca.180 Ma (Jurassic to Triassic age; Channel et al., 1979),resulting from the break-up of Pangea and the separation ofLaurasia+Gondwana to initiate the opening of the centralAtlantic Ocean in the Early Jurassic. The metamorphic ageof the UHP rocks was determined as ca. 35 Ma byGebauer et al. (1993, 1997) by ion-probe dating (SHRIMP)of UHP zircons. The age of the subducting Piemont oceanicplate has been assumed to be 145 Ma. The convergencerate was estimated to be very low, approximately 0.2 mm/yr, according to the plate reconstruction of Dewey et al.(1989).


In the above contrasting examples (Fig. 8), an old slab with avery slow convergence rate vs. a young slab with an extremelyfast convergence rate, clearly supports the idea that the age ofthe subducting oceanic plate is a critical parameter in deter-mining the P–T condition of the slab surface of the subductingoceanic plate. Applying this approach to estimate the P–T pathof the slab surface, we can roughly estimate the path in eachrespective subduction zone. The P–T path beneath NE Japan ismuch colder than that of the Alps, because the age of thesubducting oceanic plate is about 100 Myr with much fastersubduction speed than that of Alps. The convergence rate isapproximately 100 mm/yr, which is 50 times higher than that ofAlps, indicating a path within lawsonite–eclogite facies.

5. Phase diagram of peridotite+water

Hydrated oceanic crust releases H2O to the overlying mantlewedge under various P–T conditions. The released H2O shouldbe mainly retained in hydrous phases; some is stored inanhydrous minerals (several tens ppm in olivine, garnet andseveral hundreds ppm in pyroxenes; see Bell and Rossman,1992), in aqueous fluid phase, and in the melt (Thompson,1992). The fate of released water is controlled by the stability

Fig. 9. (a) A phase diagram of the MORB+H2O system and subduction zone geothermstable. Broken lines: phengite stability limit in pelitic rocks (D&H by Domanik andsystem (Kawamoto et al., 1996, 1997; Bose and Ganguly, 1995; Ulmer and Trommminerals are stable. A, B, C and D; representative P–T paths of the slab surfaces.

relations of hydrous phases in the overlying mantle wedge. Thusto discuss the transportation mechanism of water into the deepermantle, we need a phase diagram for peridotite+H2O (Fig. 9).

Stability relations of hydrous minerals for ultramafic compo-sitions in the mantle wedge has been determined recently byultrahigh-pressure experiments (e.g. Yamamoto and Akimoto,1977; Bose and Ganguly, 1995; Ulmer and Trommsdorff, 1995;Kawamoto et al., 1996; Kawamoto and Holloway, 1997). Phaserelations of serpentine, chlorite, amphibole and others are shownschematically in Fig. 9. Amphibole, chlorite and serpentine arestable at depths shallower than 200 km; amphibole is stable below3 GPa, 1000 °C, chlorite below 5 GPa, 800 °C and serpentine(antigorite) at temperatures lower than 500–600 °C below6.5GPa. Thus, with increasing pressure up to 6.5 GPa, maximumtemperatures of hydrousmineral stability decreases from 1000 °Cto 500–600 °C. With a further increase in pressure, hydroussilicates expand their stability fields towards higher temperatures.The hydrous minerals (e.g. phase A, E and hydrous-β, -γ) arecandidates for H2O containers in the mantle wedge (peridotite+H2O system) above 6.5 GPa. The high-Tmaximum stability limitof those phases shifts from 600 °C at 6.5 GPa to 1500 °C at13GPa. The hydrousβ-olivine coexistswithmelt at temperatureshigher than T=1100 °C and P=12 GPa (Fig. 9).

s. Open area: hydrous minerals are unstable, shaded areas: hydrous minerals areHolloway, 1996; S by Schmidt, 1996). (b) A phase diagram of peridotite+H2Osdorf, 1995). Open area: hydrous minerals are unstable, shaded areas: hydrous


The water released from the subducting lithosphere istrapped by the overlying mantle wedge in which hydroussilicates are stable, if the geotherm is cold enough to stabilizehydrous phases. However if the geotherm is hot, the releasedH2O is not stable in the hydrous phases in the overlying mantle.Instead, the H2O would move upwards as porous flow, andcannot be transported into the deeper mantle through draggingdown by the lithospheric counter-flow.

As already discussed above, the P–T path of the slab surfacevaries, and is strongly dependent on the age of the subductingslab, and the convergence rate. Temperature conditions of theoverlying mantle wedge depend also on the T of the slabsurface.

6. Mechanism of water transportation into the mantletransition zone

Combining the two phase diagrams in Fig. 9 in addition toFig. 3, the water transportation mechanism can be estimated asfollows. First, the following four different T path cases along

Fig. 10. Schematic profile showing the transportation mechanism of H2O down to thethe text.

the slab plane (Fig. 9) are discussed: (A) steepest geothermgenerated by old slab; (D) most gentle geotherm produced byvery young slab; and intermediate types (B) and (C).

In the case of subduction of an old oceanic plate along a lowgeothermal gradient, the subducting oceanic crust suffersprogressive high P–T metamorphism from blueschist to law-sonite–eclogite facies (Fig. 9). Lawsonite–eclogite facies as-semblages are stable over a wide P–T field extending from2.5 GPa, T<500 °C, through 8 GPa, T<800 °C to 9.5 GPa,<700 °C. That is, lawsonite is stable at depths ranging from75 km to 300 km. As described above, lawsonite is continuouslydehydrated with increasing temperature in the coesite field, andwith increasing pressure and temperature in the stishovite field.Subduction of old oceanic plate (Path A) releases 3 wt.% H2Ocontinuously from 75 km to 100 km depth by the blueschist–lawsonite–eclogite transformation and retains 1 wt.%H2O in thesubducting slabs. Therefore, the small wedge corner triangle(trenchward from aseismic front) must be hydrated extensivelywith successive subduction of the oceanic plate with time (Fig.10a). Intermediate-depth intraslab earthquake has been

mantle. Case A, B, C and D correspond to the P–T paths in Fig. 9. See detail in


considered to be caused by dehydration embrittlement insubducting oceanic crust (e.g. Kirby et al., 1996). The aseismicfront (AF) in NE Japan may correspond to the pressure-sensitivefacies boundary between the blueschist and lawsonite–eclogitefacies because approximately 5 wt.% of H2O would be releaseddue to continuous dehydration reaction from the blueschistfacies to the boundary. The small triangle mantle wedge may notjoin the dragged counter-flow of the mantle because of thebuoyancy of strongly serpentinized peridotite (Fig. 10a).

At depth in the lawsonite–eclogite facies, continuous dehy-dration of lawsonite releases small quantities of H2O; thesubducting oceanic crust retains about 0.5 wt.% H2O even atdepths greater than 200 km. In the stishovite field, thedescending slab continuously releases all the remaining H2Ountil the lawsonite–eclogite stability limit, at 300 km depth. TheH2O released from the subducting oceanic crust, would migrateupward to hydrate the mantle wedge above the subduction zone.This H2O is stored mainly in chlorite and serpentine in thehanging wall peridotite just above the MORB crust down to200 km depth. The peridotite containing those hydrous phases isdragged further down and is heated up gradually to releasewater-rich fluids. The released H2O moves upward to decreasethe melting temperature of the mantle wedge (Fig. 10a). Island-arc tholeiite magma may be formed by this process (e.g. Tatsumiet al., 1986). In the mantle wedge adjacent to the slab, however,serpentine is stable above 200 km depth; below this depth, phaseA is stable from a depth of 200 km to at least a depth of 300 kmor further down to below 300 km depth. These two phasesretain H2O by slab dehydration through the lawsonite-consuming reaction and the induced mantle wedge convectionmay move all the way down to the mantle transition depth.Below 250 km, phase E is stable at T>700 °C; the hydrous layerin the mantle wedge thickens with continuous subduction be-neath the mantle wedge. Hydrous phase β-olivine is stable atdepths below 350 km, T>1100 °C (Fig. 9b). Thus, the surfacewater trapped in the oceanic lithosphere can be partly transportedinto the mantle transition zone at 410–660 km, if the geotherm iscold enough to stabilize lawsonite down to a depth of 300 km.The water containers in the mantle transition zone are phase E,hydrous β- and γ-olivine. The mantle transition zone at 410–660 km depth is an enormous water reservoir, as demonstratedby the 3 wt.% of H2O content of the most dominant phase of β-olivine (>50%).

In the case of Path B (shown in Fig. 9), lawsonite is stabledown to 270 km depth in the subducting oceanic crust. In theoverlying mantle wedge, chlorite and serpentine are not stablebelow 150 km and no hydrous phase is stable at depths between150 km (T>600 °C) and 240 km (T>700 °C). Phase A should bestable below 240 km depth if the H2O is present in the mantlewedge. However, no hydrous silicates are stable between150 km and 240 km in the convected mantle wedge. Never-theless lawsonite is stable down to 270 km, so less than 0.5 wt.%H2O can be supplied from the slab down to 270 km depth bycontinuous dehydration of the remaining lawsonite. Concur-rently, the released H2O from the downgoing slab movesupwards to hydrate the mantle wedge to form hydrous phase Abelow 240 km to 270 km (Fig. 10b). Once retained in phase A,

H2O can be carried to greater depths due to the inducedconvection of the mantle wedge. Below 270 km, phase Ebecomes the main H2O container, and is stable in the mantletransition zone. At greater depth, phase E is not stable, and H2Ois stored in hydrous β-phase of olivine containing up to 3 wt.%H2O (Fig. 10b).

From the 150 to 240 km depth, the H2O released from thesubducting oceanic crust by lawsonite-consuming reaction,migrates upward through the overlying mantle wedge if thefluid is dominated by H2O. The ascending aqueous fluid causespartial melting of the mantle wedge, and produces island-arcmagma. However, the fluid dissolves considerable amounts ofSi, Mg and other elements under ultrahigh-pressure conditions(Fujii et al., 1997). The addition of aqueous fluid may producehydrous phases such as K-richterite and Ti-clinohumite in themantle wedge.

A moderately old subducting oceanic plate may follow pathC of Fig. 9. The P–T trajectory passes from the blueschist tozoisite–eclogite facies at 70 km depth. The transformation fromzoisite–eclogite to dry eclogite occurs at 110 km, within thecoesite stability field (Fig. 9a). Most of the H2O is leached out at110 km depth when the slab is heated up to 650–700 °C. H2Ocontent of the subducting slab decreases from 0.5 wt.% at 75 kmdepth to nearly 0 at 110 km depth. Released H2O could bestored in serpentine and chlorite of the overlying mantle wedgealong the Benioff thrust. However, no hydrous mineral is stableat depths below 130 km in the mantle wedge right above theslab surface. Stored H2O in these hydrous minerals at shallowlevels is migrated upward in the convected mantle wedge at130 km. The ascending aqueous fluids cause the partial meltingof the highest temperature region of the mantle wedge, toproduce island-arc magma (Fig. 10c).

In the case of subduction of young oceanic crust (path D inFig. 9), slab melting occurs at depths 30–70 km and the releasedH2O is stored in the melt ascending to the mantle wedge.Extensive melting of hydrated MORB crust forms calc–alkalinemagma, in which all H2O is incorporated. Hence, H2O cannotbe transported into the mantle wedge below 70 km depth. OnceH2O is leached out of the subducting oceanic crust, it is difficultto produce melt again, because the melting temperature definedby the dry solidus of the MORB is higher than 1220 °C at 70 kmdepth. The hydrated SiO2-rich felsic melt would react with α-olivine during the magma ascent into the overlying mantle, toform orthopyroxene as an interface curtain wall in the meltpathway (Fig. 10d; Maruyama, 1997).

7. Role of serpentinized peridotites along transform faultson the oceanic floor

Oxygen isotope analyses of ophiolites (e.g. Gregory andTaylor, 1981) and direct drilling of oceanic crust at DSDP site504B (Alt et al., 1986) indicate that hydrothermal circulation atmid-oceanic ridges is shallower than 800 m based on boilingstudy of the seawater detected at 504B, or at least less than5 km, estimated by water/rock ratio by oxygen isotope onophiolite. Maede and Jeanloz (1991) estimated the maximumthickness of the hydrated zone to be 12 km, which is far thinner


compared to the thickness of slab 600–100 km if it is older than50 Myr. If this is true, major parts of oceanic peridotite under theMORB crust are predominantly dry, and are unrelated to thetransportation of surface water into the deeper mantle.

However, it is well-known that serpentinized peridotites areexposed along transform faults on the ocean-floor (Fig. 2).Therefore, some parts of oceanic peridotite are undoubtedlyhydrated and may transport surface water into the mantle tran-sition zone, although it is difficult to estimate its rolequantitatively. If the lower plane of the double seismic zone isrepresenting the dehydration of the hydrated peridotite, thehydrous peridotite in the slab is not only the part hydrated at thetransform faults. Further speculative hypothesis has beenproposed by Seno and Yamanaka (1996), therefore we do notargue further here. Thus, the presence of those hydrated peri-dotites promotes the water transportation into deeper mantle.

8. Role of phengite in subducting pelitic rocks

Capping the top of the oceanic lithosphere, the hemipelagicsediments and trench turbidites may subduct together withoceanic crust in deeper parts of a subduction zone, althoughmost would be offloaded to form an accretionary complex atshallow depths on the hanging wall of the overlying plate. It isnot certain whether they could subduct below Moho depth intothe deep upper mantle, lie above the transition zone. Thestability relations of hydrous phases in a pelitic system havebeen investigated on a wide P–T range (Schmidt, 1996;Domanik and Holloway, 1996). Phengite–muscovite as amost common hydrous phase in pelitic rocks has a maximumP-stability up to 8–10 GPa and up to 1060 °C. Other accessoryhydrous phases are topaz–OH, Mg–pumpellyite, lawsonite, andK–cymrite (Domanik and Holloway, 1996). Phengitic mica hasa stability field 100–200 °C higher than the maximum stabilitylimit of lawsonite in metabasite at P=3–10 GPa. Therefore,some amounts of water may be transported by pelitic rocksdown to 300 km depth only if the geothermal gradients are lowenough including A, B and C in Fig. 9. Thus, pelitic rockscould, possibly help the transportation of H2O to great mantledepth.

9. Discussion

Before the advent of plate tectonics in Earth Science, mostearth scientists never discussed the fate of surface water in termsof global material circulation on a whole mantle scale, becauseof the lack of critical petrologic observations, experimental datasets, and geophysical data on mantle circulation. However, thepresence of phlogopite and K–amphibole in mantle xenolithsclearly indicate that water is present as structural OH− in mantleminerals. When plate tectonics appeared, to interpret subductionzone magmatism, most volcanologists inferred that calc–alkaline magmas are derived from slab-melting of hydratedMORB or partial melting of the mantle wedge, to which thereleased seawater is added to lower the melting T. Somepetrologists considered that almost all dehydration occurs togenerates arc magma at depths shallower than 150–200 km

(Ernst, 1999), hence no H2O penetrates into the mantle transitionzone with time. Therefore, the Earth's mantle transition zone(410–660 km), if hydrated when the magma ocean hadconsolidated during Early Earth at 4.55 Ga, must have graduallydehydrated by plume activity through geologic time (Kawamotoet al., 1996). In contrast, there is an opposing idea thatsubduction zones deliver six times more water into the mantle(8.7×1011 kg yr−1) than is delivered to the surface by arcvolcanism (1.4×1011 kg yr−1) (Peacock, 1990), implying theon-going water transportation into the mantle transition zone. Ifthis is true, most subducted water is being swallowed into themantle transition zone or even into the lower mantle at present,implying that the total amount of seawater in the oceans is nowbeing reduced. From the standpoint of mantle mineralogy andpetrology, dense hydrous magnesium silicates (DHMS) whichare stable under mantle conditions, count more than 28 species(e.g. review by Thompson, 1992). For example, for upper mantleconditions, after the pioneering work by Ringwood and Major(1967), Yamamoto and Akimoto (1977) investigated the systemMgO–SiO2–H2O at P=2.9–7.7 GPa and T=470–1225 °C, andfound DHMS such as brucite, phase A, phase D, clinohumite,serpentine, talc, 10 Å phase, and chondrodite. Since then,superhydrous B (Pacalo and Parise, 1992; Burnley andNavrotsky, 1996), phase A (Bose and Ganguly, 1995; Pawleyand Wood, 1996), and phase E (Luth, 1995) have beensynthesized and analyzed. Akaogi and Akimoto (1980), Liu(1986, 1987), Kanzaki (1991), and Gasparik (1993) haveinvestigated the phase relations near and in the mantle transitionzone. Under lower mantle conditions, phase D (Li and Jeanloz,1991; Yang et al., 1997; Shieh et al., 1998), phase F (Ohtani etal., 1995), and phase G (Ohtani et al., 1997; Kudoh et al., 1997)have been described as stable hydrous silicates if we presume thepresence of water in the lower mantle.

Thus, a variety of hydrous silicates has been demonstrated aspossible stable phases within the whole depth range of uppermantle and topmost parts of the lower mantle.

In spite of wide and high-T stability fields of those DHMS inthe deeper mantle, surface water cannot be simply trans-ported into the deeper mantle, because (1) the subduction zonegeotherm that they assumed is too warm, resulting indehydration of all DHMS at relatively shallow depth ranges(see Fig. 9b), (2) subducting oceanic lithosphere is capped by7 km thick MORB, and MORB cannot carry water into mantledeeper than 300 km. Although the core is the coldest part of thesubducting slab, hydrous parts of the slab are less than thetopmost 12 km (Maede and Jeanloz, 1991), or even less(probably less than 800 m as discussed before), and (3) watertransportation into the mantle transition zone depends stronglyon the subduction-zone geotherm as explained below. If it iswarm, no water can enter into the deeper mantle. If the presenceof limited lateral arrangement of arc volcanoes usually 120–200 km right above the slab surface, means that all DHMSdehydrate below 200 km depth, then a suggestion arises that nowater enters into mantle depths greater than 200 km (Kawamotoet al., 1996).

To solve the debate, phase relations of MORB+H2O andperidotite+H2O under the whole upper mantle conditions, are


crucial. The summarized phase relations in Fig. 9b provide aconcept regarding the fate of water. Two critical points at X1

(6.5 GPa, 550 °C) and X2 (5.5 GPa, 600 °C), are marked on Fig.9b. X1 corresponds to the lowest T at which dry mineralassemblages are stable in mantle peridotite. From this pointtoward higher-P and -T, the stability field of anhydrous mineralassemblages is spread over a wide range. Therefore, if theBenioff geotherm passes the lower-T side than X1, draggeddownflow of the mantle wedge can carry the water withinhydrous phases into the deeper mantle. If the Benioff geothermpasses on the higher-T side of X2, water cannot be transportedinto deeper mantle not only by downflow of the wedge mantlebut also by hydrated MORB crust. If the Benioff geothermpasses between X1 and X2, lawsonite can transport thecrystalline water down to 240 km depth (point X3) to berelayed to the overlying downflow of mantle in which phase Ais stable. Note a unique character of the reaction to form phaseA connecting X1 with X3, because it is a hydration reactionforsterite+water=phase A (Pawley and Wood, 1996). If nowater is supplied by the downgoing oceanic crust, the mantlewedge cannot be hydrated. In the case of the Benioff geothermpassing on the higher-T side of X2, water cannot be transportedinto the deeper mantle at all. All hydrous phases in downgoingMORB crust and in serpentinized peridotites along transformfaults would be dehydrated by 200 km depth, thereafter nowater moves into the transition zone (Fig. 9b). Thus a smalltriangular area defined by X1, X2, X3 is critical to control thefate of water in a subduction zone. The significance of the roleof triangular zone in the P–T space is portrayed in Fig. 10b andc where, the presence of a dry hanging wall peridotite preventsdehydrated water moving down as hydrous phases along theslab surface. Instead, free-water moves upwards to trigger thepartial melting of peridotite.

Thus, the geothermal gradients along the slab surfacedetermine whether or not the surface water can be transportedinto the deeper mantle. The variable P–T gradients along the slabsurface are a key to understanding the fate of water in subductionzones. If subduction zones do not follow P–T paths like A or B,then we must support the interpretation by Kawamoto et al.(1996). But, as discussed before, if subduction zone geothermsare very variable and depend highly on the age of the slab, andfollow paths such as A or B, wemust support the idea by Peacock(1990). If the slab is older than 50 Myr, it would pass into thelawsonite stability fields or even much colder P–T ranges, aswell-demonstrated by (1) numerical simulation (Peacock, 1990,1996), (2) lawsonite–eclogites (reviewed by Okamoto andMaruyama, 1999), (3) a few young regional metamorphic beltswith independent information on age of subducted slab andrelative plate motion, and (4) P–T estimates by frequency ofactive seismicity in several active subduction zones discussed inthis paper. Considering the average age of subducting slabs (ca.100 Ma) on the modern Earth, we must conclude that surfacewater moves down into the deeper mantle in most presentsubduction zones. This leads to an idea that surface water is nowdecreasing.

Additional supporting observations are derived from seis-mological aspects of the mantle. Earthquakes deeper than

100 km, down to 650 km in the Earth may easily be explained ifdehydration is on-going at those depths by subduction (Maedeand Jeanloz, 1991; Stein, 1995). At those depths rocks areexpected to deform by ductile flow rather than brittle fracturingor frictional sliding on fault surfaces (Maede and Jeanloz,1991).

Acknowledgements

We express sincere thanks to Dr. Chris Parkinson, Profs. E.Takahashi, J.G. Liou and M. W. Schmidt who gave us usefulcomments and suggestions. We appreciate detailed reviews byProf. W.G. Ernst.

References

Akaogi, M., Akimoto, S., 1980. High-pressure stability of a dense hydrousmagnesian silicate Mg23Si8O42H6 and some geophysical implications.Journal of Geophysical Research 85, 6944–6948.

Alt, J.C., Honnorez, J., Laverne, C., Emmermann, R., 1986. Hydrothermalalteration of a 1 km section through the upper oceanic crust, deep sea drillingproject Hole 504 B: mineralogy, chemistry, and evolution of seawater–basaltinteractions. Journal of Geophysical Research 91, 10,309–10,335.

Anderson, R.N., Uyeda, S., Miyashiro, A., 1976. Geophysical and geochemicalconstraints at converging plate boundaries: Part I. Dehydration in thedowngoing slab. Geophysical Journal of the Royal Astrological Society 44,333–357.

Anderson, R.N., DeLong, S.E., Schwarz, W.M., 1978. Thermal model forsubduction with dehydration in the downgoing slab. Journal of Geology 86,731–739.

Banno, S., 1964. Petrologic studies of the Sanbagawa crystalline schists in theBessi–Ino district, central Shikoku, Japan. Journal of Faculty of Sciences,University of Tokyo, Sec II 15, 203–319.

Banno, S., Sakai, C., 1989. Geology and metamorphic evolution of theSanbagawa metamorphic belt, Japan. In: Daly, J.S., Cliff, R.A., Yardley, B.W.D. (Eds.), Evolution of Metamorphic Belts. . Special Publication, vol. 43.Geological Society of London, pp. 519–532.

Banno, S., Higashino, T., Otsuki, M., Itaya, T., Nakajima, T., 1978. Thermalstructure of the Sanbagawa belt in central Shikoku. Journal of Physics of theEarth 345–356 (Supplement).

Bell, D.R., Rossman, G.R., 1992. Water in Earth's mantle: the role of nominallyanhydrous minerals. Science 255, 1391–1397.

Bird, P., 1978. Stress and temperatures in subduction shear zones: Tonga andMariana. Journal of Royal Astrological Society 55, 411–434.

Bohlen, S.R., Boettcher, A.L., 1982. The quartz–coesite transformation: aprecise determination and the effects of other components. Journal ofGeophysical Research 87, 7073–7078.

Bose, K., Ganguly, J., 1995. Experimental and theoretical studies of the stabilitiesof talc, antigorite and phase A at high pressures with applications tosubduction processes. Earth and Planetary Sciences Letters 136, 109–121.

Brophy, J.G., Marsh, B.D., 1986. On the origin of high-alumina arc basalt andthe mechanics of melt extraction. Journal of Petrology 27, 763–789.

Bundy, F.R., 1980. The P,T phase reaction diagrams for elemental carbon.Journal of Geophysical Research 85, 6930–6936.

Burnley, P.C., Navrotsky,A., 1996. Synthesis of high-pressure hydrousmagnesiumsilicates: observations and analysis. American Mineralogist 81, 317–326.

Channel, J.F.T., D'Argenio, B., Horvath, F., 1979. Adria, the Africanpromontary. Mesozoic Mediterranean Paleogeography. Earth ScienceReview, vol. 15, pp. 213–292.

Chopin, C., 1984. Coesite and pure pyrope in high-grade blueschists of thewestern Alps: a first record and some consequences. Contributions toMineralogy and Petrology 86, 107–118.

Chopin, C., Sobolev, N.V., 1995. Principal mineralogic indicators of UHP incrustal rocks. In: Coleman, R.G., Wang, X. (Eds.), Ultrahigh PressureMetamorphism. Cambridge University Press, pp. 96–131.


Cloos, M., 1985. Thermal evolution of convergent plate margins: thermalmodeling and re-evaluation of isotopic Ar-ages for blueschists in theFranciscan complex of California. Tectonics 4, 421–433.

Dallmeyer, R.D., Takasu, A., Yamaguchi, K., 1995. Mesozoic tectonothermaldevelopment of the Sanbagawa, Mikabu and Chichibu belts, southwestJapan: evidence from 40Ar/39Ar whole-rock phyllite ages. Journal ofMetamorphic Geology 13, 271–286.

Davies, J.H., Stevenson, D.J., 1992. Physical model of source region ofsubduction zone volcanics. Journal of Geophysical Research 97, 2037–2070.

Delany, J.M., Helgeson, H.C., 1978. Calculation of the thermodynamicconsequences of dehydration in subducting oceanic crust to 100 kb and>800 °C. American Journal of Sciences 278, 638–686.

Dewey, J.F., Helman,M.L., Turco, E., Hutton, D.H., Knott, S.D., 1989. Kinematicsof the western Mediterranean. In: Coward, M.P., Dietricu, D.D., Park, R.G.(Eds.), Alpine Tectonics. . Special Publication, vol. 45. Geological Society ofLondon, pp. 265–283.

Domanik, K.J., Holloway, J.R., 1996. The stability and composition of phengiticmuscovite and associated phases from 5.5 to 11 GPa: implications for deeplysubducted sediments. Geochimica Cosmochimica Acta 60, 4133–4150.

Drummond, M.J., Defant, M.S., 1990. A model for trondjemite–tonalite–dacitegenesis and crustal growth via slab melting: Archean to moderncomparisons. Journal of Geophysical Research 95, 21503–21521.

Engebretson, D.C., Cox, A., Gordon, R.G., 1985. Relative plate motion betweenoceanic and continental plates in the Pacific basin. Special Paper, vol. 206.Geological Society of America, pp. 1–59.

Ernst, W.G., 1999. Hornblende, the continent marker – evolution of H2O duringcircum-Pacific subduction versus continental collision. Geology 27, 675–678.

Eskola, P., 1920. The mineral facies of rocks. Norsk Geologisk Tidsskrift 6,143–194.

Fujii, T., Mibe, K., Yasuda, A., 1997. Aqueous fluid at high pressure: anothersilicate magma. Kazan 42, S321–S326 (in Japanese).

Furukawa, Y., 1993a. Magmatic processes under arcs and formation of thevolcanic front. Journal of Geophysical Research 98, 8309–8319.

Furukawa, Y., 1993b. Depth of the decoupling plate interface and thermalstructure under arcs. Journal of Geophysical Research 98, 20,005–20,013.

Gasparik, T., 1993. The role of volatiles in the transition zone. Journal ofGeophysical Research 98, 4287–4299.

Gebauer, D., Tilton, G.R., Shertl, H.-P., Schreyer, W., 1993. Eocene/Oligoceneultrahigh-pressure (UHP)-metamorphism in the Dora-Maira massif (westernAlps) and its geodynamic implications. Fourth International EclogiteConference (Calabria, Italy). Terra Abstract, vol. 4, pp. 10–11.

Gebauer, D., Schertl, H.-P., Brix, M., Schreyer, W., 1997. 35 Ma old ultrahigh-pressure metamorphism and evidence for very rapid exhumation in the DoraMaira massif, Western Alps. Lithos 41, 5–24.

Gregory, R.T., Taylor Jr., H.P., 1981. An oxygen isotope profile in a section ofCretaceous oceanic crust, Samail ophiolite, Oman: evidence for δ18Obuffering of the oceans by deep (>5 km) seawater-hydrothermal circulationat mid-oceanic ridges. Journal of Geophysical Research 86, 2737–2755.

Hart, R.A., 1976. Progressive alteration of the oceanic crust. In: Yeats, R.S., et al.(Eds.) Initial Reports of the DSDP, vol. 34, pp. 433–437.

Hasebe, K., Fujii, N., Uyeda, S., 1970. Thermal processes under island arcs.Tectonophysics 10, 335–355.

Higashino, T., 1975. Biotite zone of the Sanbagawa metamorphic terrain in theShiragayama area, central Shikoku. Journal of Metamorphic Geology 8,413–423.

Higashino, T., 1990. The higher grade metamorphic zonation of the Sanbagawametamorphic belt in central Shikoku, Japan. Journal of MetamorphicGeology 8, 413–423.

Holland, T.J.B., 1983. The experimental determination of activities indisordered and short range ordered jadeitic pyroxenes. Contributions toPetrology and Mineralogy 82, 214–220.

Honda, S., 1985. Thermal structure beneath Tohoku, northeast Japan – a casestudy for understanding the detailed thermal structure of the subductionzone. Tectonophysics 112, 69–102.

Honda, S., Uyeda, S., 1983. Thermal process in subduction zones – a review andpreliminary approach on the origin of arc volcanism. In: Shimozuru, D.,Yokoyama, I. (Eds.), Arc Volcanism: Physics and Tectonics. TERRAPUB,Tokyo, pp. 117–140.

Hower, J., Eslinger, W.V., Hower, M., Perry, E.A., 1976. Mechanism of burialmetamorphism of argillaceous sediments: I. Mineralogical and chemicalevidence. Geological Society of America Bulletin 87, 725–737.

Hsui, A.T., Toksöz, M.N., 1980. The evolution of thermal structures beneath asubduction zones, in oceanic ridges and arcs. In: Toksöz, M.N., Uyeda, S.,Francheteau, J. (Eds.), Geodynamic Processes. Development of Geotec-tonics, vol. 14. Elsevier, New York, pp. 325–342.

Hyndman, R.D., Wang, K., 1993. Thermal constraints on the zone of majorthrust earthquake failure: the Cascadia subduction zone. Journal ofGeophysical Research 98, 2039–2060.

Hyndman, R.D., Yamano, M., Oleskevich, D.A., 1997. The seismogenic zone ofsubduction thrust faults. The Island Arc 6, 244–260.

Inoue, T., Yurimoto, H., Kudoh, Y., 1995. Hydrous modified spinel, Mg1.75-SiH0.5O4: a new water reservoir in the mantle transition region. GeophysicalResearch Letters 22, 117–120.

Irifume, T., Sekine, T., Ringwood, A.E., 1986. The eclogite–garnet transfor-mation at high pressure and some geophysical implications. Earth andPlanetary Sciences Letters 77, 245–256.

Isozaki, Y., Maruyama, S., 1991. Studies on orogeny based on plate tectonics inJapan and new geotectonic division of the Japanese Islands. Journal ofGeography (Tokyo) 100, 697–761.

Itaya, T., Takasugi, H., 1988. Muscovite K–Ar ages of the Sanbagawa schists,Japan and argon depletion during cooling and deformation. Contributions toMineralogy and Petrology 100, 281–290.

Iwasaki, M., Ichikawa, K., Yao, A., Faure, M., 1984. Fossil age of greenstoneconglomerate in eastern Shikoku. Report of Kansai Branch, vol. 97.Geological Society of Japan, p. 21.

Johnson, D.M., 1979. Crack distribution in the upper oceanic crust and itseffects upon seismic velocity, seismic structure, formation permeability, andfluid circulation. In: Donnelly, T., et al.(Eds.) Initial Reports of the Deep SeaDrilling Project, Washington, DC, pp. 1473–1490.

Kanzaki, M., 1991. Stability of hydrous magnesium silicates in the mantletransition zone. Physics of the Earth and Planetary Interiors 66, 307–312.

Kawamoto, T., Holloway, J.R., 1997. Melting temperature and partial meltchemistry of H2O-saturated mantle peridotite to 11 Gigapascals. Science276, 240–243.

Kawamoto, T., Hervig, R.L., Holloway, J.R., 1996. Experimental evidence for ahydrous transition zone in the early Earth's mantle. Earth and PlanetarySciences Letters 142, 587–592.

Kay, R.W., 1978. Aleutian magnesian andesites: melts from subducted PacificOcean crust. Journal of Volcanology and Geothermal Research 4, 117–132.

Kienast, J.R., Lombardo, B., Biino, G., Pinardon, J.L., 1991. Petrology of very-high-pressure eclogitic rocks from the Brossasco Isasca Complex, DoraMaira Massif, Italian Western Alps. Journal of Metamorphic Geology 9,19–34.

Kirby, S., Engdahl, E.R., Denlinger, R., 1996. Intermediate-depth intraslabearthquakes and arc volcanism as physical expressions of crustal anduppermost mantle metamorphism in subducting slabs. In: Bebout, G.E., et al.(Eds.) Subduction: Top to Bottom. . Geophysical Monograph, vol. 96. AGU,pp. 195–214.

Kudoh, Y., Nagase, T., Mizohata, H., Ohtani, E., Sasaki, S., Tanaka, M., 1997.Structure and crystal chemistry of phase G, a new hydrous magnesiumsilicate synthesized at 22 GPa and 1050 °C. Geophysical Research Letters24, 1051–1054.

Li, X., Jeanloz, R., 1991. Phase and electrical conductivity of a hydrous silicateassemblage at lower-mantle condition. Nature 350, 332–334.

Liou, J.G., Zhang, R., Ernst, W.G., 1994. An introduction to ultrahigh-pressuremetamorphism. The Island Arc 3, 1–24.

Liu, L., 1986. Phase transformations in serpentine at high pressures andtemperatures and implications for subducting lithosphere. Physics of theEarth and Planetary Interiors 42, 255–262.

Liu, L., 1987. Effects of H2O on the phase behavior of the forsterite–enstatitesystem at high pressures and temperatures and implications for the Earth.Physics of the Earth and Planetary Interiors 49, 142–167.

Luth, R.W., 1995. Is phase a relevant to the Earth's mantle? GeochimicaCosmochimica Acta 59, 679–682.

Maede, C., Jeanloz, R., 1991. Deep-focus earthquakes and recycling of waterinto the Earth's mantle. Science 252, 68–72.


Martin, H., 1986. Effect of steeper Archean geothermal gradient ongeochemistry of subduction-zone magmas. Geology 14, 753–756.

Maruyama, S., 1991. Ocean-floor metamorphism: a review on the DSDP, IPODand ODP. EOS vol. 72, 552.

Maruyama, S., 1997. Pacific-type orogeny revisited: Miyashiro-type orogenyproposed. The Island Arc 6, 91–120.

Maruyama, S., Liou, J.G., 1983. The stability of Ca–Na pyroxene in low-grademetabasites of high-pressure intermediate facies series. American Mineral-ogist 70, 16–29.

Maruyama, S., Seno, T., 1986. Orogeny and relative plate motions – an exampleof the Japanese islands. Tectonophysics 127, 1–25.

Maruyama, S., Liou, J.G., Terabayashi, M., 1996. Blueschists and eclogites ofthe world and their exhumation. International Geology Review 38, 485–594.

Miyashiro, A., 1961. Evolution of metamorphic belts. Journal of Petrology 2,277–311.

Miyashiro, A., 1973. Metamorphism and Metamorphic Belts. John Wiley andSons, New York. 492 pp.

Molnar, P., England, P.C., 1990. Temperatures, heat flux, and frictional stressnear major thrust faults. Journal of Geophysical Research 95, 4833–4856.

Moore, C.J., 1992. Fluids in accretionary prisms. Review of Geophysics 30,113–135.

Nicolls, I.A., Ringwood, A.E., 1973. Effect of water on olivine stability intholeiites and production of silica-saturated magmas in the island arcenvironment. Journal of Geology 81, 285–300.

Ohtani, E., Shibata, T., Kubo, T., Kato, T., 1995. Stability of hydrous phases inthe transition zone and the uppermost part of the lower mantle. GeophysicalResearch Letters 22, 2553–2556.

Ohtani, E., Mizohata, H., Kudoh, Y., Nagase, T., Arashi, H., Yurimoto, H.,Miyagi, I., 1997. A new hydrous silicate, a water reservoir in the upper partof the lower mantle. Geophysical Research Letters 24, 1047–1050.

Okamoto, K., Maruyama, S., 1999. The high pressure stability limits oflawsonite in the MORB+H2O system. American Mineralogist 84,362–373.

Okamoto, K., Maruyama, S., 2004. The eclogite-garnetite transformation in theMORB+H2O system. Physics of the Earth and Planetary Interiors 146,283–296.

Okamoto, K., Shinjoe, H., Katayama, I., Terada, K., Sano, Y., Johnson, S., 2004.SHRIMP U-Pb dating of quartz-bearing eclogite from the Sanbagawa Belt,SW Japan; Implications for Protolith and Metamorphic evolution. TerraNova 16, 81–89 (Oleskevich).

Oleskevich, D.A., Hyndman, R.D., Wang, K., 1999. The updip and downdiplimits to great subduction earthquakes: thermal and structural models ofCascadia, south Alaska, SW Japan, and Chile. Journal of GeophysicalResearch 104, 14,965–14,991.

Otsuki, M., Banno, S., 1990. Prograde and retrograde metamorphism ofhematite-bearing basic schists in the Sanbagawa belt in central Shikoku.Journal of Metamorphic Geology 8, 425–439.

Oxburgh, E.R., Turcotte, D.L., 1970. Thermal structure of island arcs.Geological Society of America Bulletin 81, 1665–1688.

Pacalo, R.E.G., Parise, J.B., 1992. Crystal structure of superhydrous B, ahydrous magnesium silicate synthesized at 1400 °C and 20 GPa. AmericanMineralogist 77, 681–684.

Pawley, A.R., Wood, B.J., 1996. The low-pressure stability of phase A,Mg7Si2O8(OH)6. Contributions to Mineralogy and Petrology 124, 90–97.

Peacock, S.M., 1987. Thermal effects of metamorphic fluids in subductionzones. Geology 15, 1057–1060.

Peacock, S.M., 1990. Fluid processes in subduction zones. Science 248,329–337.

Peacock, S.M., 1991. Numerical simulation of subduction zone pressure–temperature–time paths: constraints on fluid production and arc magma-tism. Philosophical Transaction of Royal Society of London A 335,341–353.

Peacock, S.M., 1992. Blueschist–facies metamorphism, shear heating and P–T–t paths in subduction shear zone. Journal of Geophysical Research 97,17,693–17,707.

Peacock, S.M., 1993. The importance of blueschist – eclogite dehydrationreactions in subducting oceanic crust. Geological Society of AmericaBulletin 105, 684–694.

Peacock, S.M., 1996. Thermal and petrological structure of subduction zones. In:Bebout, G.E., et al.(Eds.) Subduction: Top to Bottom. Monograph, vol. 96.American Geophysical Union, pp. 119–133.

Peacock, S.M., Rushmer, T., Thompson, A.B., 1994. Partial melting ofsubducting oceanic crust. Earth and Planetary Science Letters 121,227–244.

Poli, S., 1993. The amphibolite–eclogite transformation: an experimental studyon basalt. American Journal of Science 293, 1061–1107.

Poli, S., Schmidt, M.W., 1995. H2O transport and release in subduction zones:experimental constraints on basaltic and andesitic systems. Journal ofGeophysical Research 100, 22,299–22,314.

Ringwood, A.E., 1975. Composition and Petrology of the Earth's Mantle.McGraw-Hill, New York. 618 pp.

Ringwood, A.E., Major, A., 1967. A high pressure transformations inpyroxenes. Earth and Planetary Science Letters 1, 351–357.

Schmidt, M.W., 1996. Experimental constraints on recycling of potassium fromsubducting oceanic crust. Science 272, 1927–1930.

Scholz, C.H., 1990. The Mechanics of Earthquakes and Faulting. CambridgeUniv. Press, Cambridge. 439 pp.

Schreyer, W., 1988. Experimental studies on metamorphism of crustal rocksunder mantle pressures. Mineralogical Magazine 52, 11–26.

Schreyer, W., 1995. Ultradeep metamorphic rocks: the retrospective viewpoint.Journal of Geophysical Research 100, 8353–8366.

Seno, T., Yamanaka, Y., 1996. Double seismic zones, compressional deep trench–outer rise events, and superplumes. In: Bebout, G.E., et al.(Eds.) Subduction:Top to Bottom. Geophysical Monograph, vol. 96. AGU, pp. 347–355.

Shieh, S.R., Ming, L., Mao, H., Hemley, R.J., 1998. Decomposition of phase Din the lower mantle and the fate of dense hydrous silicates in subductingslabs. Earth and Planetary Science Letters 159, 13–23.

Smyth, J.R., Kawamoto, T., 1997. Wadsleyite: II. A new high pressure hydrousphase in the peridotite–H2O system. Earth and Planetary Science Letters146, E9–E16.

Staudigel, H., King, S.D., 1992. Ultrafast subduction: the key to slab recyclingefficiency and mantle differentiation? Earth and Planetary Science Letters109, 517–530.

Staudigel, H., Davies, G.R., Hart, S.R., Marchant, K.M., Smith, B.M., 1995.Large scale isotopic Sr, Nd and O isotopic anatomy of altered oceanic crust:DSDP/ODP sites 417/418. Earth and Planetary Science Letters 130,169–185.

Stein, S., 1995. Deep earthquakes: a fault too big? Science 268, 49–50.Taira, A., 1985. Sedimentary evolution of Shikoku subduction zone. The

Shimanto Belt and Nankai Trough. In: Nasu, N., et al.(Eds.) Formation ofActive Ocean Margins. Terra Science Publication, Tokyo, pp. 273–290.

Takasu, A., Dallmeyer, R.D., 1990. 40Ar/39Ar age constraints for thetectonothermal evolution of the Sanbagawa metamorphic belt, centralShikoku, Japan: a Cretaceous accretionary prism. Tectonophysics 185,303–324.

Takasu, A., Wallis, S., Banno, S., Dallmeyer, R.D., 1994. Evolution of theSanbagawa metamorphic belt, Japan. Lithos 33, 119–133.

Tatsumi, Y., Hamilton, D.L., Nesbitt, R.W., 1986. Chemical characteristics offluid phase released from a subducted lithosphere and origin of arc magmas:evidence from high-pressure experiments and natural rocks. Journal ofVolcanology and Geothermal Research 29, 293–309.

Thompson, A.B., 1992. Water in the Earth's upper mantle. Nature 358,295–302.

Titchelaar, B.W., Ruff, L.J., 1993. Depth of seismic coupling along subductionzones. Journal of Geophysical Research 98, 2017–2037.

Toksöz, M.N., Hsui, A.T., 1989. A review of the thermal–mechanical structuresat convergent plate boundaries and their implications for crust/mantlerecycling. In: Hart, S.R., Gulen, L. (Eds.), Crust/Mantle Recycling atConvergent Zones. Kluwer Academic, Norwell, Mass, pp. 75–80.

Toksöz, M.N., Minear, J.W., Julian, B.R., 1971. Temperature field andgeophysical effects of a downgoing slab. Journal of Geophysical Research76, 1113–1138.

Turcotte, D.L., Schubert, G., 1973. Frictional heating of the descendinglithosphere. Journal of Geophysical Research 78, 5876–5886.

Ulmer, P., Trommsdorff, V., 1995. Serpentine stability to mantle depths andsubduction-related magmatism. Science 268, 858–861.


van den Beukel, J., Wortel, R., 1988. Thermo-mechanical modeling of arc–trench regions. Tectonophysics 154, 177–193.

Wang, C.-Y., Shi, Y.-L., 1984. On the thermal structure of subduction complexes:a preliminary study. Journal of Geophysical Research 89, 7709–7718.

Wyllie, P.J., 1988. Magma genesis, plate tectonics, and chemical differentiationof the Earth. Review of Geophysics 26, 370–404.

Wyllie, P.J., Sekine, T., 1982. The formation of mantle phlogopite in subductionhybridization. Contributions to Mineralogy and Petrology 79, 375–380.

Yamamoto, K., Akimoto, S., 1977. The system MgO–SiO2–H2O at highpressures and temperatures – stability field for hydroxyl–chondrodite,

hydroxyl–clinohumite and 10 Å-phase. American Journal of Science 277,288–312.

Yang, H., Prewitt, C.T., Frost, D.J., 1997. Crystal structure of the dense hydrousmagnesium silicate, Phase D. American Mineralogist 82, 651–654.

Zhang, J., Li, B., Utsumi, W., Liebermann, R.C., 1996. In situ X-rayobservations of the coesite–stishovite transition: reversed phase boundaryand kinetics. Physics of Chemistry and Mineralogy 23, 1–10.

Documents

wattrans