Th or collective redistirbution or collective of this by of any article portion SPECIAL ISSUE FEATURE in published been has is article

Manufacturing Continental in , Volume 4, a quarterly journal of Th 19, Number the or other means reposting, is only photocopy machine, permitted

Factory Copyright 2006 by Th Society. e Oceanography BY YOSHIYUKI TATSUMI AND ROBERT J. STERN with the approval of Th of approval the with WERE BORN dehydration reactions and/or partial length of ’s 55,000 km of conver- IN THE ? melting during subduction dissolve par- gent margins (Stern, 2002) (Figure 2),

Subduction zones have been consuming ticular elements and carry them into the provide vital information on how the Permission reserved. Society. All rights e Oceanography is gran oceanic since the dawn of overlying wedge, leading to gener- subduction factory works. They are all correspondence to: Society. Send [email protected] e Oceanography or Th on this planet (Figure 1). ation of chemically distinct arc situated on thinner and dominantly Raw materials such as pelagic and ter- that differentiate, solidify, and produce more mafi c crust than systems at con- rigenous sediments, altered and fresh juvenile arc crust; the site of these trans- tinental margins and hence signifi cant basaltic , and lithosphere formations is known as the “subduction contamination of magmas by easily fus- are conveyed into this “factory” as the factory.” Collisional coalescence and ible (light-colored rocks high in subducted plate sinks deeper into the magmatic thickening of juvenile arcs and silica, for example, gran- mantle. Assuming steady-state subduc- ultimately yield , an im- ite) crust may not occur. Along such tion of the entire 7-km-thick oceanic portant subduction-factory product. The margins, we can also better measure crust for 3 billion years, the accumulated factory inevitably emits waste materials, crustal growth rates, now estimated to ted in teaching to and research. copy this for use Repu article crustal materials occupy ~ 10 percent of such as chemically modifi ed, residual be 30–95 km3 km-1 Myr-1 (Dimalanta et the . Aqueous fl uids and/or oceanic crust and sediments and delami- al., 2002). Intra-oceanic arcs, however, silicate melts that are progressively ex- nated arc mafi c (high in iron and mag- are signifi cantly less well studied than MD 20849-1931, USA. 1931, Rockville, Box PO Society, e Oceanography tracted from these raw materials through nesium) lower crust, which sink deeper continental arcs. The main reason for the into the mantle. These waste materials, paucity of research is that except for the Yoshiyuki Tatsumi (tatsumi@jamstec. because of their great mass and residual tops of the largest volcanoes, most of the is Program Director, Geochemical crustal composition, contribute greatly arc is submerged below level. Ocean Evolution, Institute for Frontier Research on to the thermal and chemical evolution drilling at intra-oceanic arcs, therefore, Earth Evolution, Japan Marine Science and of the mantle and ultimately may return should provide key observations for Technology Center, Yokosuka, Japan. Robert to Earth’s surface as mantle plumes and comprehending the solid-Earth cycle and blication, systemmatic reproduction, J. Stern ([email protected]) is Professor, related magmas (Figure 1). evolution. Because of the importance of Department of Geosciences, University of Intra-oceanic subduction systems, this problem, both the MARGINS Sci- at Dallas, Dallas, TX, USA. which occupy ~ 40 percent of the global ence Plans (2003) and Integrated Ocean

104 Oceanography Vol. 19, No. 4, Dec. 2006 Figure 1. Role of the subduction factory in the evolution of the solid Earth. Raw materials, such as ocean- ic sediments, oceanic crust, mantle lithosphere, and wedge materials, are fed into the factory and are manufactured into arc magmas and continental crust. Th e waste ma- terials or residues processed in this factory, such as chemically modifi ed components (oceanic crust and sediments) and delaminated mafi c lower arc crust, are transported and stored in the deep mantle, and re- cycled as raw materials for mantle- plume-related hot-spot .

Drilling Program (IODP) Initial Sci- component because it is the storehouse crust. The very distinctive composition ence Plan (ISP) (Coffi n, McKenzie et al., of most natural resources and serves as of the continental crust indicates that 2001) have identifi ed understanding the substrate for agriculture and civiliza- understanding how this crust forms is operation of Earth’s subduction factory tion. Chemically, the continental crust is essential for understanding solid Earth as a research priority. a reservoir for “incompatible” elements evolution. The IODP ISP includes conti- that do not fi t easily in mantle minerals, nental-crust formation and growth as a PERSPECTIVES ON such as K, Rb, U, Th, and the light Rare priority drilling target (Coffi n, McKenzie CONTINENTAL CRUST Earth Elements. Continental crust thus et al., 2001). FORMATION possesses “differentiated” compositions, It is generally accepted that conti- Continental crust occupies only 0.4 per- quite distinct from a chondritic (compo- nental crust has an andesitic bulk com- cent of the total mass of the solid Earth; sition similar to that of asteroids) bulk position (Rudnick and Fountain, 1995; however, it is by far the most important Earth, primitive mantle, or even oceanic Taylor and McLennan, 1995). Because

Oceanography Vol. 19, No. 4, Dec. 2006 105 andesitic magmatism typifi es subduction Reconstructing crustal-growth processes to and more felsic crustal com- zones, many researchers consider that for continental arcs is, however, challeng- positions typical of continental crust. continental crust has been created domi- ing. Low-density continental crust acts as Seismic profi ling of the Izu-Bonin-Mari- nantly in arc-trench settings. However, a density fi lter, trapping mantle-derived ana (IBM) (Figure 3) arc crust (Suyehiro modern-day, mantle-derived magmatism mafi c melts that then are likely to frac- et al., 1996; Takahashi et al., 2006) has in such settings is dominated by . tionate, assimilate, and melt the crust, further demonstrated the presence of This dilemma faces anyone interested in strongly affecting the compositions of middle crust that has a P-wave velocity continental-crust formation. and plutons. In contrast, intra-oce- of ~ 6 km s-1 (Figure 4a), interpreted to To explain the characteristic andesitic anic arcs are built on high-density, mafi c correspond broadly to intermediate-to- composition of the average continen- oceanic crust. Oceanic crust is much felsic “tonalitic” composition (Figure 4b). tal crust, several mechanisms have been less effective as a density fi lter and melts (A tonalite is an igneous intrusive proposed (e.g., Rudnick, 1995; Tatsumi, at a higher temperature, thus effects of of felsic composition.) Furthermore, 2005). These hypotheses have been as- pre-existing crust are minimized for tonalitic rocks are found as xenoliths in- sessed petrologically, geochemically, and intra-oceanic arcs. Nevertheless, intra- cluded in lavas and as exposures along geophysically at various continental arcs. oceanic arc crust appears to differentiate intra-oceanic arcs such as the IBM and

0˚ 60E˚ 120E˚ 180˚ 120W˚ 60W˚

60˚ Eurasia

W. Aleutians Kuriles N. America Lesser 30˚ Izu-Bonin Antilles Mariana Africa 0˚ New Britain Vanuatu S. America New 30˚ Hebrides Kermadec Scotial/ S. Sandwich Arc



Figure 2. Locations of modern intra-oceanic arc-trench systems. Trenches associated with these systems are shown by barbed lines. Intra-oceanic arcs are sites where the eff ects of pre-existing continental crust can be neglected and represent relatively simple natural laboratories for the study of modern crustal growth.

106 Oceanography Vol. 19, No. 4, Dec. 2006 Kyushu-Palau systems. This compelling now understand that the subduction fac- to the upper level of the subduction fac- evidence suggests that the ingredients tory has two “fl oors.” The lower level of tory—the arc crust—where it is further for continental crust are indeed created the factory, the subducted slab and man- processed by fractionation, partial melt- today in intra-oceanic arcs. IODP drill- tle wedge, extracts fl uids and/or melts ing, and delamination to generate tonal- ing in intra-oceanic arcs will “be a tre- from subducted materials. These fl uids itic/andesitic continental crust. mendous step towards understanding the and melts are enriched in the “continen- We cannot visit (by drilling) the lower origin of the andesitic continental crust” tal component.” This mixture is trans- fl oor of the subduction factory, so our (Coffi n, McKenzie et al., 2001). ferred upward, away from the slab, and information about its operations must mixes with normal asthenospheric man- be indirect. One method to evaluate the THE FLOOR PLAN OF THE tle, enriching it and lowering its melting processes in the lower fl oor of the facto- SUBDUCTION FACTORY temperature. This hybrid mantle melts ry is to analyze primitive, least-fraction- Like all factories, the subduction factory to form primitive (Mg-rich) arc basaltic ated arc lavas, which are relatively rare, follows specifi ed procedures. To under- that is enriched in continental- especially on large volcanic . It is stand them, one must know where the type incompatible elements. This primi- easier to fi nd primitive lavas by sampling various operations are carried out. We tive yet distinctive magma is then sent small submarine arc volcanoes, which

125˚E 130˚ 135˚ 140˚ 145˚ 150˚ Figure 3. Location map of the Philippine Sea

Eurasia Plate region. Th e Izu-Bonin-Mariana (IBM) arc-trench system forms the convergent margin between 40˚N Pacifi c and Philippine Sea plates. Backarc ba- sins such as Shikoku Basin, Parece Vela Basin, and Mariana Trough were created by seafl oor spreading between the formerly contiguous remnant arc (Kyushu-Palau and West Mariana ridges) and the active IBM arc. At its northern 35˚ Honshu Isl. Arc-Arc Collision Zone tip, the IBM arc has collided with Honshu Is- land since 15 Ma. Th e X-X’ red line locates the structural model cross section (Figure 4). Num- Nankai Trough Amami X X' bers show proposed drill sites. Sankaku Izu-Bon

Basin Izu-Bonin Arc 30˚ Kyusyu-Palau Ridge

in Trench Shikoku Amami Basin Plateau


W Mariana Ridge

Mariana Trench Daito West Philippine Ridge Basin Mariana Trough Mariana Arc

20˚ Parece Vela Basin

Philippine Sea Plate


Oceanography Vol. 19, No. 4, Dec. 2006 107 Volcanic X Rear-arc Front Fore-arc Trench X' 0

4 5 5 Upper crust

Middle crust 6.0-6.5 4 10 6.8-7.4 6.0-6.3 6.8-6.9 15 Lower crust Uppermost Lower crust late 20 7.1-7.3 (a) Moho After Suyehiro et al. 1996 -7.8 km/s 25 Subducting Pacific P

Serp. Mud Pre-existing Rear Arc Middle Arc Initial Arc Crust & Mantle Magmatism Crust Crust Deformed Lavas, Sediments, & Shallow Intrusions Wedge Intermediate Plutons 10 Gabbro, Pyroxenite, and Amphibolite 20 Trapped Oceanic Lithosphere PartiallySerpentinized 30 (b) km After DeBari et al. 1999 0 50 100 150 200 250 300 350 400 450 Distance (km)

Figure 4. (a) Seismic structure of the crust and based on seismic refraction and wide-angle refl ection data along a X-X’ cross-arc section within the IBM system (Figure 3). Th e arc crust seismic structure is explained by the distribution of upper, middle, and lower crustal layers as schemati-

cally shown in (b). IBM is the only juvenile intra-oceanic arc known to have a well-developed, low-velocity middle crust with Vp=6.0-6.3 km/s, possibly of similar composition to the average continental crust and provides key constraints on the process of continental crust formation. Numbers in (b) show schematic representation of proposed drill sites.

are too young to have developed effective els for thermal structure of the mantle nevertheless immense. Intra-oceanic arc magma chambers where magmatic frac- wedge, and models for how melts rise systems have the dimensions of gigan- tionation occurs, or by drilling the old- from the mantle wedge into the crust. tic fl at noodles, much longer than they est, most primitive parts of the . To understand how continental crust are wide and much wider than they are The composition of primitive lavas is is generated from primitive arc magma, thick. For example, the IBM arc system used to constrain models for arc magma we must visit the upper fl oor of the sub- is ~ 3000-km long, ~ 200-km wide, and genesis, which must be integrated with duction factory, the arc crust. Physically, ~ 25-km thick (Figures 3 and 4). The our understanding of the composition of the upper fl oor is much smaller in size physical dimensions of intra-oceanic the downgoing slab and sediments, mod- than the lower fl oor, but intra-oceanic arcs are thus subcontinental in scale; our els of how fl uids and melts migrate from arc systems—the simplest example of ability to understand these processes is the slab up into the mantle wedge, mod- the subduction factory upper fl oor—are challenged not only by the tremendous

108 Oceanography Vol. 19, No. 4, Dec. 2006 size of intra-oceanic arc systems but also tory, IODP scientists must integrate data passage of arc magmas; oceanic crustal because even the roof of the subduction from many sources, such as multichan- remnants could also make up an impor- factory is submerged, with only a few nel seismic refl ection and wide-angle tant part of the lower-arc crust. Typically, chimneys (volcano summits) reaching seismic profi les imaging the crust and the presence of such relict oceanic crust . These dimensions and the sub- upper mantle, heat-fl ow measurements, is assumed, because recovery of samples marine nature of the subduction factory and petrological/chronological examina- from sub-arc depths of 15–20 km is im- second fl oor are very suitable for marine tion of seafl oor and on-land samples. possible. At the IBM, pre-existing oce- geophysical surveys (e.g., seismic refl ec- anic crust is present west of the arc, un- tion and refraction; electromagnetic and ARC EVOLUTION AND der 1–1.5 km of sediments in the Amami gravity surveys), which routinely extend CONTINENTAL CRUST Sankaku Basin adjacent to the Kyushu- for several hundred kilometers and can FORMATION Palau Ridge remnant arc (Figure 3) (Tay- image tens of kilometers into the crust. Understanding how continental crust lor and Goodliffe, 2004) and may crop Interpretation of the geophysical sections forms at intra-oceanic arcs requires out on the lower fore-arc slope of the that are generated, however, requires that we fi rst know how intra-oceanic Bonin Trench (DeBari et al., 1999). calibrating the seismic profi les with drill arcs form and mature. Key questions The backarc basins of the eastern core, which at present is impossible. for comprehending arc crust forma- Philippine Sea Plate are underlain by The deepest drillhole to date on Earth tion are: (1) What is the nature of the asthenosphere of Indian Ocean charac- ( in Russia) crust and mantle in the region prior to ter, geochemically distinct from mantle only penetrates ~ 12 km into the con- the beginning of subduction? (2) How sources beneath the Pacifi c Plate to the tinental crust and the deepest scientifi c does subduction initiate and initial arc east (Hickey-Vargas, 1998). IBM subduc- drillhole at sea only penetrates ~ 2 km crust form? (3) How do the middle and tion-zone inception may have therefore below the seafl oor. The new IODP drill- arc crusts evolve? (4) What are the spa- occurred along a major mantle domain ship Chikyu can drill with a riser (to re- tial changes of arc magma and crust boundary of the type presently exposed move rock bits from the drilled hole and compositions of the entire arc? Possible at the Australian-Antarctic Discordance. stabilize the hole) to depths of ~ 7 km strategies for answering these questions But, it is also clear that the initial con- below the seafl oor, which is still less than include drilling by IODP at the IBM struction of the IBM arc system, rather half way to the bottom of arc crust. But, arc-system, which is one of the best-sur- than developing solely upon oceanic depth of seafl oor and total depth below veyed intra-oceanic arcs. Figure 4a crust, transected a series of Cretaceous- the seafl oor are not the only limitations shows the across-arc seismic structure of Paleocene ridges (e.g., Amami Plateau on scientifi c drilling of intra-oceanic the IBM (line X-X’ in Figure 3), which and Daito and Oki-Daito ridges) and arc systems; we are also not able to drill is representative for the entire IBM arc intervening basins that formed, at least into very hot (> 200°C) rocks. Drilling (Suyehiro et al., 1996). IODP has pro- in part, an arc-backarc system (Figure 3) deeply into arc crust thus requires that posals to drill at the IBM, including three (Taylor and Goodliffe, 2004; Hickey-Var- we avoid the itself and fo- non-riser holes (1, 2, and 4 in Figures 3 gas et al., 2005). Alternatively, these pro- cus on forearc and perhaps backarc set- and 4b) and one riser, ultra-deep hole to-continental ridges and plateaus may tings, where lower geothermal gradients (3 in Figures 3 and 4). have rifted away from Asia. Decoding can be found. the nature of the magma source in the In the following sections we discuss Nature of the Original Crust upper mantle that existed immediately how to take advantage of the Chikyu’s and Mantle before IBM subduction inception is key new drilling capabilities to better under- Pre-existing, non-arc oceanic crustal to understanding the cause of initiation stand the subduction factory. To get the components should contribute to arc of subduction zones and intra-oceanic most information and to more tightly magma chemistry through assimilation arc formation. This decoding can be ac- constrain models of the subduction fac- and triggered during complished only by examining pre-arc

Oceanography Vol. 19, No. 4, Dec. 2006 109 oceanic crust, such as that of the Amami of the additional proposed drilling intra-oceanic arcs, coupled with a loss of Sankaku Basin in the Northeast Philip- there by the IODP will be spent passing -rich lower crustal residues (Jull pine Sea Plate (1 in Figure 3), or by ex- through a signifi cant thickness of sheet- and Kelemen, 2001); this process may amining crust that might be encountered ed dikes. Whether or not a signifi cant represent an early stage in the produc- in deep drilling of the fore-arc. section (about 1000-m thick) of sheeted tion of continental crust. Establishing the dikes exists at the IBM for-arc is an im- composition of IBM arc middle crust, Initial Arc Crust Formation and portant test of the hypothesis that the correlating this composition to profi les Subduction Initiation IBM fore-arc formed by seafl oor spread- of crustal seismic velocities, and inferring The nature of arc magmatism, and hence ing. Existing geophysical data provide no genetic relationships between in situ in- the nature of the arc crust, at the very constraints on this question. Attempt- tra-oceanic arcs and upper crustal volca- initial stage of IBM subduction is dis- ing to penetrate a sheeted dike complex nic rocks and mid-crustal sequences ex- tinct from that of later stages. To better also comprises an important test of the posed on land, can only be accomplished constrain the initial arc crust formation hypothesis that most form by scientifi c ocean drilling. We propose at the IBM, deepening of the previous by seafl oor spreading during the initial returning to the region of ODP Site 792 Ocean Drilling Program (ODP) hole stages of subduction. We should be able (3 in Figures 3 and 4) to drill through (Hole 786B; shown as 2 in Figures 3 to determine the strike of dikes and thus the Eocene-Oligocene upper crust down and 4b) is proposed. the local direction of spreading, which to middle crust. The mid-crustal layer ODP Leg 125 drilled a series of holes may provide critical constraints on tec- in this area is shallow enough to be in the IBM fore-arc (Arculus, Pearce et tonic regime at the very initial stage of reached by drilling, and heat fl ow is low al., 1992) to recover samples of initial arc evolution. enough for drilling to proceed at mid- arc crust. The last of these (Hole 786B) crustal temperatures. cored 666 m of Eocene volcanic base- Evolution of Middle Crust ment. The recovered rocks were a com- Extensive seismic refraction studies reveal Rear-Arc Magma and Arc Crust plex series of pyroclastics (fragmented that IBM crustal structure is not only Compositions bits of rock) and lavas, followed by much thicker (25 km vs. 7 km) than typi- Previous drilling efforts in IBM during dikes, exhibiting compositions distinct cal oceanic crust, it also has a thick layer the Drilling Project (DSDP)- -1 from the later stage magmas. The recov- of low-velocity (Vp=6.0–6.3 km s ) mid- ODP focused mainly on the forearc (the ered extrusive rocks likely were pillow dle crust (Figure 4a). Yet, intra-oceanic area between the trench and the volcanic lavas that formed the fl anks of a vol- island arcs are thought to be constructed arc) and documented the temporal vari- cano, which was intruded by dikes and of basalt, similar to oceanic crust formed ation of magmatism (e.g., increasing K20 sills—similar stratigraphy to normal at mid-ocean ridges (~ 50 percent SiO2). with time) along the volcanic front (the oceanic crust. These observations, in- A basaltic bulk composition for intra- trenchward boundary of a volcanic arc) cluding the structural and compositional oceanic arcs is also expected because the preserved in tephra and turbidites (Gill similarity between the IBM fore-arc mantle partially melts to produce basalt. et al., 1994; Arculus et al., 1995; Bryant crust and ophiolites (oceanic crust that As noted previously, subduction is also et al., 2003; Straub, 2003). However, the has been thrust onto land), led Pearce thought to ultimately be responsible for magmatic history of the rear-side of the et al. (1992), Stern and Bloomer (1992), generating continental crust, which is an- IBM arc has not been similarly well stud- and Ishiwatari et al. (2006) to con- desitic in bulk composition (~ 60 percent ied, partly because rear-arc volcanoes are clude that the IBM fore-arc formed by SiO2). The paradox posed by the basaltic submarine so this tephra is not dispersed seafl oor spreading. composition of mantle melts and the an- as far as that from subaerial volcanoes of The abundant dikes drilled at the bot- desitic composition of continental crust the magmatic front. tom of ODP Hole 786B (Arculus, Pearce may partially be resolved if a low-veloc- Analysis of dredged rocks suggests et al., 1992) suggest that much (~ 1 km) ity, mid-crustal felsic layer is generated in that there is a consistent difference in

110 Oceanography Vol. 19, No. 4, Dec. 2006 chemistry between the IBM volcanic SUMMARY ern Pacifi c. B. Taylor and Natland J., eds. Ameri- can Geophysical Union, Washington, D.C. front and rear-arc magmas throughout We are confi dent that ocean drilling Bryant C.J., R.J. Arculus, and S.M. Eggins. 2003. the Neogene (Tatsumi et al., 1992; Hoch- at intra-oceanic arcs, such as the IBM, The geochemical evolution of the Izu-Bonin arc staedter et al., 2000, 2001; Ishizuka et is the way to answer the fundamental system: A perspective from tephras recovered by deep-sea drilling. Geochemistry, Geophysics, al., 2003; Machida and Ishii, 2003); this question of how continental crust is cre- Geosystems 4:1094, doi:10.29/2002GC000427. difference has long been recognized for ated; to test the hypothesis of whether Coffi n, M.F., J.A. McKenzie et al. 2001. Earth, arc systems globally (Dickinson, 1975; continents themselves are fi rst born in , and Life. Scientifi c Investigations of the Earth System using Multiple Drilling Platforms Tatsumi and Eggins, 1995). The record the ocean; and to understand the role and New Technologies. Integrated Ocean Drilling for the IBM, however, is based on a small of subduction-factory processes in solid Program Initial Science Plan, 2003–2013. Inter- number of rocks dredged and dated Earth evolution. The key objective of national Working Group Support Offi ce, Inte- grated Ocean Drilling Program, Washington, from rear-arc cross chains (including re- this drilling effort is to sample rocks that D.C., 110 pp. [Online] Available at http://www. sults from Marianas as well as Izu: Stern form the deeper part of the arc crust, [last accessed October 8, 2006]. et al., 1993; Stern et al., 2006). The only which is feasible only by riser drilling Dimalanta, C., A. Taira, G.P. Yumul, Jr., H. Tokuyama, and K. Mochizuki. 2002. New rates thing known about rear-arc with Chikyu. To implement these IODP of western Pacifi c magmatism from between the Neogene cross chains is that expeditions, pre-expedition, non-drilling seismic and gravity data. Earth and Planetary the 7 Ma basalt and andesite dredged efforts will be required, including acqui- Science Letters 202:105–115. DeBari, S.M., B. Taylor, K. Spencer, and K. Fujioka. from one such site are similar to volcanic sition of multichannel seismic refl ection 1999. A trapped Philippine Sea plate origin for front rocks (Hochstaedter et al., 2001). and wide-angle seismic data for better MORB from the inner slope of the Izu–Bo- nin trench. Earth and Planetary Science Letters Therefore, it may be that the apparent imaging of the crust and upper mantle, 174:183–197. difference between volcanic front and heat fl ow measurements for confi rm- Dickinson, W.R. 1975. Potash-depth (K-h) rela- rear-arc magmatism is restricted to the ing the deep drilling, and petrological/ tions in and intra-oceanic magmatic arcs. Geology 3:53–56. cross chains. The temporal variation in chronological examination of seafl oor Gill, J.B., R.N. Hiscott, and P. Vidalc. 1994. Turbi- rear-arc magma chemistry also has not and on-land samples. dite geochemistry and evolution of the Izu-Bo- been well documented. This variation nin arc and continents. Lithos 33:135–168. Hickey-Vargas, R. 1998. Origin of the Indian can be addressed only by examining the ACKNOWLEDGEMENTS Ocean-type isotopic signature in from tephra and turbidite record stored in the We thank Tim Byrne, Donald Fisher, Philippine Sea plate spreading centers: An as- basins located between the rear-arc vol- and Robert Burger for their constructive sessment of local versus large-scale processes. Journal of Geophysical Research 103:20,963– canic chains. In the northern IBM, these comments on the manuscript and Miki 20,979. basins should be largely free of ash from Fukuda for preparing the manuscript Hickey-Vargas, R., I.P. Savov, and M. Bizimis. 2005. the IBM volcanic front because prevail- and fi gures. We are also grateful to our Relationship between the West Philippine Ba- sin and the early Izu-Bonin-Mariana Arc: Arc ing winds blow east. Finally, nothing at colleagues in the Project IBM work- initiation and ancestry. Eos Transactions of the all is known about Oligocene rear-arc ing group. American Geophysical Union, 86(52), Fall Meet- volcanism. Such information affects how ing Supplement, Abstract T44A-07. Hochstaedter, A., J.B. Gill, B. Taylor, O. Ishizuka, M. crust formed and evolved across the REFERENCES Yuasa and S. Morita. 2000. Across-arc geochem- arc, and can be learned only by drilling. Arculus, R.J., J.A. Pearce, B.J. Murton, and S.R. Van ical trends in the Izu-Bonin arc: Constraints on der Laan. 1992. Igneous stratigraphy and major source composition and mantle melting. Jour- We thus propose a drilling expedition element geochemistry of Holes 786A and 786B. nal of Geophysical Research 105:495–512. at the rear-arc side of the Izu arc (4 in Pp. 143–69 in Proceedings of the Ocean Drilling Hochstaedter, A., J. Gill, R. Peters, P. Broughton, Figures 3 and 4b) along an across-arc Program, Scientifi c Results, L.H. Dearmont, E.K. P. Holden, and B. Taylor. 2001. Across-arc Mazzullo, N.J. Stewart, and W.R. Winkler, eds. geochemical trends in the Izu-Bonin arc: Con- section that passes other proposed drill Ocean Drilling Program, College Station, TX. tributions from the subducting slab. Geochem- sites (Figures 3 and 4). Arculus, R.J., J.B. Gill, H. Cambray, W. Chen, and istry, Geophysics, Geosystems 2:doi:10.1029/ R.J. Stern. 1995. Geochemical evolution of arc 2000GC000105. systems in the western Pacifi c; The ash and tur- Ishiwatari A., Y. Yanagida, Y.-B. Li, T. Ishii, S. Hara- bidite record recovered by drilling. Pp. 45–65 in guchi, et al. 2006. Dredge petrology of the Active Margins and Marginal Basins of the West- boninite- and -bearing Hahajima Sea-

Oceanography Vol. 19, No. 4, Dec. 2006 111 mount of the Ogasawara (Bonin) forearc: An terms of major element chemistry. Geochemis- or a serpentinite ? The Island try, Geophysics, Geosystems 4:1018, doi:10.29/ Arc 15:102–118. 2002GC000357. Ishizuka, O., K. Uto, and M. Yuasa. 2003. Volcanic Suyehiro, K., N. Takahashi, Y. Ariie, Y. Yokoi, R. history of the back-arc region of the Izu-Bonin Hino, et al. 1996. Continental crust, crustal un- (Ogasawara) Arc. Pp. 187–205 in Intra-Oceanic derplating, and low-Q upper mantle beneath an Subduction Systems: Tectonic and Magmatic Pro- oceanic island arc. Science 272:390–392. cesses, R.D. Larter and P.T. Leat, eds. Geological Takahashi, N., S. Kodaira, S. Klemperer, Y. Tatsumi, Society of London, London, UK. Y. Kaneda, and K. Suyehiro. 2006. Structure and Jull, M., and P.B. Kelemen. 2001. On the conditions evolution of Izu-Ogasawara (Bonin)-Mariana for lower crustal convective instability. Journal oceanic island arc crust. Geology submitted. of Geophysical Research 106:6,423–6,446. Tatsumi, Y. 2005. The subduction factory: How Machida, S., and T. Ishii. 2003. Backarc volcanism it operates in the evolving Earth. GSA Today along the en echelon : Enpo sea- 15:4–10. mount chain in the northern Izu-Ogasawara Tatsumi, Y., and S. Eggins. 1995. Subduction Zone Arc. Geochemistry, Geophysics, Geosystems Magmatism. Blackwell Science, Boston, MA, 4(8):9,006, doi: 10.1029/2003GC000554. 211 pp. MARGINS Science Plans. 2003. Science Plans 2004. Tatsumi, Y., M. Murasaki, and S. Nohda. 1992. NSF MARGINS Program. MARGINS Offi ce Across-arc variation of chemistry in the at Lamont-Doherty Earth Observatory of Co- Izu-Bonin Arc: Identifi cation of subduction lumbia University, 170 pp. [Online] Available components. Journal of Volcanology and - at: thermal Research 49:179–190. SciencePlans/MARGINS_SciencePlans2004.pdf. Taylor, B., and A.M. Goodliffe. 2004. The west Pearce, J.A., S.R. Van der Laan, R.J. Arculus, B.J. Philippine Basin and the initiation of subduc- Murton, T. Ishii, et al. 1992. Boninite and harz- tion, revisited. Geophysical Research Letters burgite from Leg 125 (Bonin-Mariana forearc): 31:10.1029/2004GL020136. A case study of magma genesis during the ini- Taylor, S.R., and S.M. McLennan. 1995. The geo- tial stages of subduction. Pp. 623–659 in Pro- chemical evolution of the continental crust. ceedings of the Ocean Drilling Program, Scientifi c Reviews of Geophysics 33:241–265. Results, P. Fryer, J.A. Pearce, L.B. Stokking, J.R. Ali, R. Arculus, et al., eds. Ocean Drilling Pro- gram, College Station, TX. Rudnick, R.L. 1995. Making continental crust. Na- ture 378:571–578. Rudnick, R.L., and D.M. Fountain. 1995. Nature and composition of the continental crust: A lower crustal perspective. Reviews of Geophysics 33:267–309. Stern, R.J. 2002. Subduction Zones Reviews of Geo- physics 40, 10.1029/2001RG000108 Stern, R.J., and S.H. Bloomer. 1992. Subduc- tion zone infancy: Examples from the Eocene Izu-Bonin-Mariana and California arcs. Geological Society of America Bulletin 104:1,621–1,636. Stern, R.J., M.C. Jackson, P. Fryer, and E. Ito. 1993. O, Sr, Nd, and Pb isotopic composition of the Kasuga Cross-Chain in the Mariana Arc: A new perspective on the K-h relationship. Earth and Planetary Science Letters 119:459–475. Stern, R.J., E.J. Kohut, S.H. Bloomer, M. Leybourne, M. Fouch, and J. Vervoort. 2006. Subduction factory processes beneath the Guguan Cross- Chain, Mariana Arc: No role for sediments, are serpentinites important? Contributions to Min- eralogy and Petrology 151:202–221. Straub, S.M. 2003. The evolution of the Izu Bo- nin-Mariana volcanic arcs (NW Pacifi c) in

112 Oceanography Vol. 19, No. 4, Dec. 2006