Tectonic setting of Eocene boninite magmatism in theIzu^Bonin^Mariana forearc

Colin G. Macpherson a;b;*, Robert Hall b

a Department of Geological Sciences, University of Durham, South Road, Durham DH1 3LE, UKb SE Asia Research Group, Department of Geology, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK

Received 5 June 2000; accepted 18 January 2001

Abstract

Middle Eocene boninites were simultaneously generated over a large region during the early history of the Izu^Bonin^Mariana (IBM) arc. However, widespread boninite magmatism is not recognised in younger subduction zones ofsimilar dimensions. This suggests that an additional tectonic or thermal factor influenced the generation of the IBMboninite suite. Regional uplift, ocean island basalt-style magmatism and high heat-flow also characterised the northernPhilippine Sea plate (PSP) at the start of the Middle Eocene. These features are similar to those observed in largevolume basaltic volcanic provinces such as the early Tertiary North Atlantic and suggest that the IBM boninite suitemay have been produced because there was already a thermal anomaly in the mantle beneath the PSP. Thereconstructed Middle Eocene location of the IBM arc and West Philippine Basin lies close to the present day ManusBasin where petrological and geochemical evidence indicate the presence of a mantle plume. A calculated plume tracklinking these locations through time also passes close to the Eauripik Rise, an aseismic ridge on the Caroline Plate,during the Oligocene and Miocene. Therefore, we propose that a thermal anomaly or mantle plume influenced themagmatic and tectonic development of the western Pacific from the Middle Eocene until the present day. ß 2001Elsevier Science B.V. All rights reserved.

Keywords: boninite; Izu^Bonin^Mariana; Philippine Sea Plate; subduction; mantle plumes; Manus Basin

1. Introduction

Boninites are a magnesium- and silica-rich formof volcanic rock believed to form through meltingof residual mantle in supra-subduction zone set-tings [1]. There is a consensus amongst petrolo-gists that several conditions must be met to gen-

erate boninitic magma. First, the magnesium- andsilica-rich nature of the melts and very low abso-lute concentrations of some trace elements (e.g.Nb, Ta, Ti) and heavy rare earth elements re-quires low-pressure melting of residual (harzbur-gitic) peridotite, such as may be found in oceanicmantle lithosphere overriding a subducted slab.Second, because this residual source had previ-ously lost a basaltic melt fraction an abnormallyhigh heat £ux is required to promote melting [2].Third, boninite series lavas display enrichment ofthe large ion lithophile elements (LILEs) and light

0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 2 4 8 - 5

* Corresponding author. Tel. : +44-191-374-4783;Fax: +44-191-374-2510;E-mail: colin.macpherson@durham.ac.uk

EPSL 5764 13-3-01

Earth and Planetary Science Letters 186 (2001) 215^230

www.elsevier.com/locate/epsl

rare earth elements relative to high ¢eld strengthelements (HFSEs). High LILE/HFSE ratios arewidely regarded as indicating a (hydrous) £uid£ux from subducted oceanic lithosphere into thedepleted peridotite [1,3^5]. The presence of such a£uid in the source will also contribute to loweringthe solidus temperature of the residual peridotite.Finally, a mechanism is required to explain theelevation of Zr/Sm, Hf/Sm and Zr/Ti ratios inboninites compared to other types of lavas. Themost common explanations of this feature haveinvolved sources related to ocean island basalt(OIB) magmatism [3,4,6], or partial melting ofthe subducted oceanic crust in the amphibolite¢eld [5,7]. New Hf isotope evidence appears torule out the slab-melting hypothesis in boninitesfrom the western Paci¢c and favours melting ofma¢c components stored in the shallow mantle orlower crust of the overriding plate [8].

Boninitic rocks are found in two main settings.Several convergent margins contain boniniticlavas and these have been used to provide theprimary evidence for the tectonic setting in whichthis style of magmatism occurs. The best studiedlocations of Cenozoic boninite lavas are the Izu^Bonin^Mariana forearc, Cape Vogel in PapuaNew Guinea, the Northern Tonga arc and theSetouchi Volcanic belt of Japan ([1] and referen-ces therein, [9]). Boninitic rocks are also found ina number of ophiolite sequences, most notably theCretaceous Troodos ophiolite in Cyprus [10] andthe Cenozoic Zambales ophiolite of the Philip-pines [11] and in some Palaeozoic and older for-mations around the world [12]. A notable featureof the majority of occurrences is the limited vol-ume of boninites and their absence in many arcs[1]. Boninites are volumetrically subordinate toother styles of magmatism, principally island arctholeiites, in Papua New Guinea [13] and in boththe Troodos [10] and Zambales ophiolites [11,14].Where volume estimates have not been made, bo-ninites still apparently form rare or spatially re-stricted amounts of the total igneous assemblage,e.g. Setouchi belt [15], New Caledonia [16] andthe northern Tonga Trench [9]. In stark contrastto this are the Middle Eocene boninitic lavas thatcomprise a signi¢cant fraction of basementthroughout the Izu^Bonin and Mariana forearcs

[1,5,7,17]. This magmatic suite, which will be col-lectively referred to as the Izu^Bonin^Mariana(IBM) boninite suite, requires that conditionsconducive to boninite genesis occurred over alarge area (2000 km long by 300 km wide) in arelatively short interval [18] in the Eocene. Indeed,recent 40Ar^39Ar dating suggests that the boninitemagmatism was initiated almost simultaneouslythroughout this area in the early part of the Mid-dle Eocene and was, for the most part, a veryshort-lived event [19].

Recent hypotheses for generation of the IBMboninite suite have favoured models in which ahigh geothermal gradient was achieved throughthe combination of two e¡ects. These are that(1) very young, and hence hot, oceanic litho-sphere, or even a spreading centre, was subductedbeneath (2) young, hot lithosphere of the Philip-pine Sea Plate (PSP) [1,5,7]. The combinationwould generate an abnormally high heat £ux inthe subduction zone thus permitting melting ofboth the residual harzburgitic lithosphere in theoverriding plate and the subducted slab. However,there are geodynamic problems with such a tec-tonic arrangement. First, the IBM forearc base-ment is composed largely of the IBM boninitesuite and associated tholeiites with only localisedevidence for pre-Eocene magmatic activity (e.g.DSDP Site 458 [3]). This implies that a new sub-duction zone was initiated close to, and almostparallel to, a spreading centre over a distance ofapproximately 2000 km and that very young,buoyant crust was the ¢rst crust to enter thetrench over its entire length. Comparison withother sites of known ridge^subduction zone colli-sion suggests this is extremely unlikely. Attemptedridge subduction north of Baja California andwest of the Antarctic Peninsula led to a cessationof subduction [20,21] and widespread boniniticmagmatism is not recognised either at these sitesor in locations where subduction of very youngoceanic lithosphere has been successful, such assouthern Chile and northern Cascadia [22,23].Second, new Hf isotope evidence indicates thatslab-melting played a negligible role in genesisof the IBM boninites [8]. This ¢nding removesone of the principal geochemical requirementsfor subduction of a young slab. Third, although

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230216

a young, hot overriding plate can be reconciledwith models for the early Cenozoic tectonics ofthe region this cannot explain the evolution ofgeochemistry in lavas on the overriding plate.Pre-Eocene magmatism produced trace element-depleted Indian Ocean (I)-MORB but this wassucceeded by trace element-enriched and isotopi-cally distinct magmatism during the Middle Eo-cene, followed by a reversion to I-MORB geo-chemistry (see below).

While their association with the onset of wide-spread subduction in the IBM arc is well estab-lished [17], boninites have not been identi¢ed inmore recently initiated subduction zones such asthe Bicol arc in the eastern Philippines [24], the

New Britain arc of Papua New Guinea [25] andthe Banda arc in eastern Indonesia [26]. There areprobably several ways of achieving the combina-tion of heat, harzburgite melting and hydrous£uxing required for boninite magmatism as re-£ected in the presence of boninitic rocks in anumber of subduction zones and supra-subduc-tion zone ophiolites. However, the exceptionalvolume of boninitic lavas erupted in the MiddleEocene IBM, compared to other Cenozoic arcsand Cretaceous^Cenozoic supra-subduction zoneophiolites, suggests that unusual thermal, chemi-cal and/or tectonic conditions may have prevailedat that time. In order to identify the conditionsthat generated the IBM suite it is necessary to

Fig. 1. Map of the western Paci¢c and SE Asia. The PSP is outlined by the dashed line. Relevant geographic and bathymetricfeatures are labelled. Abbreviations: WPB, West Philippine Basin; NWSB, Northwest sub-basin; SSB, Southern sub-basin, RT,Ryukyu Trench; Phil T., Philippine Trench. Symbols represent DSDP/ODP drill sites and IBM boninite series sampling sites ofMiddle Eocene magmatism. Di¡erent symbols represent petrogenetic types identi¢ed using trace element characteristics[3,27,37,43]; see Fig. 2 for key. Thin lines represent magnetic anomalies from Hilde and Lee [34].

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 217

understand the Middle Eocene plate tectonics ofthe western Paci¢c.

2. Middle Eocene tectonics of the equatorialwestern Paci¢c

Boninites of the IBM suite were erupted at thepresent eastern margin of the PSP during theMiddle Eocene [17^19,27]. The geographic loca-tions of relevant tectonic elements are highlighted

in Fig. 1, and Fig. 2 shows a reconstruction of thearea at 50 Ma after Hall [28].

2.1. Pre-Middle Eocene PSP

Fragments of oceanic crust dredged from theinner trench wall of the Izu^Bonin Trench beara distinctive I-MORB isotopic signature [27]. Theage of these crustal fragments is unknown but isprobably similar to or greater than Late Creta-ceous ages determined radiometrically for rocks

Fig. 2. Plate reconstruction of Southeast Asia and the western Paci¢c at 50 Ma, at the start of the Middle Eocene [28]. The loca-tions of the Izu^Bonin and Mariana arcs are indicated, along with estimated extents of pre-Middle Eocene crust of the Amami^Oki Daito Province and the Southern sub-basin of the West Philippine Basin (WPB). Sampling sites of Middle Eocene magma-tism on the present day PSP and the IBM forearc are shown with symbols representing the geochemical nature of the magmaserupted during the Middle Eocene. The I-MORB sample from the Izu^Bonin Trench wall is probably pre-Middle Eocene [27].Radiometric ages of most of the WPB lavas plotted are within error of 50 Ma [36,37,40,72]. Ages for the IBM boninite suitevary between 55 and 44 Ma with a mode at 45 Ma [19]. Classi¢cation of these lavas as MORB, E-MORB or OIB is based ontrace element characteristics [3,27,37,43]. Small diamonds represent sites where trace element data are not available but ocean is-land magmatism is inferred from the petrology and major element chemistry. The inferred location of the Manus Basin plumecentre is plotted and has been migrated such that it remains ¢xed with respect to the Hawaiian hotspot. A small circle of 750km radius is drawn around this point to represent the size of the plume head inferred for the Afar plume [51].

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230218

dredged from the Amami^Oki Daito Province,which forms the northern part of the PSP ([29];Fig. 1). This suggests that prior to the MiddleEocene the PSP was comprised of the Amami^Oki Daito Province, which is probably of arc ori-gin [18,29,30], and some oceanic crust generatedin the I-MORB domain.

2.2. Middle Eocene spreading in theWest Philippine Basin

The West Philippine Basin forms the largesttract of the PSP and is bounded by the PhilippineTrench to the west, the Ryukyu Trough to thenorthwest and the Palau^Kyushu Ridge to theeast (Fig. 1). It comprises three main sections; anorthwestern sub-basin, the main or northernsub-basin and a southern sub-basin [31]. 40Ar^39Ar dating and geophysical constraints suggestcrust in the northwestern sub-basin is oceanicand of Early Cretaceous and Eocene age in thewest and east, respectively [32]. The northern sub-basin can be further divided into the Amami^OkiDaito province in the north and the deeper centralportion containing the NW^SE striking CentralBasin Fault. The Central Basin Fault representsan extinct ocean ridge system that facilitated sep-aration of the Amami^Oki Daito province andthe undated southern sub-basin by sea£oorspreading [31,33,34].

Mrozowski [31] identi¢ed paired magneticanomalies 20^17 trending NW^SE in the centralpart of the basin giving a minimum age of 46 Mafor the initiation of spreading at the Central BasinFault. Hilde and Lee [34] interpreted magneticanomalies extending through the Amami^OkiDaito province as indicating spreading fromabout 60 Ma. However, Okino [35] have sug-gested that the sea£oor immediately south of theAmami^Oki Daito Province may preserve a dif-ferent, N^S spreading fabric. Radiometric ages of49 and 19 Ma have been obtained by 40Ar^39Ardating of magmatic rocks from DSDP Site 294,which sampled basement immediately south of theOki Daito Ridge (Fig. 1; [36,37]). Excess argonmay have confounded the latter determinations[37] since the older age is similar to that of nearbymagnetic anomalies and the age inferred for the

sedimentary cover drilled at Site 294. Further-more, the older age is in closer accord with Mid-dle Eocene ages (ca. 45 Ma) determined for base-ment at DSDP Site 291 on the opposite margin ofthe basin ([37]; Fig. 1). The radiometric ages andtheir overlying sediments indicate that the onsetof spreading in the central West Philippine Basinbegan in the latest part of the Early Eocene or theearliest Middle Eocene.

The oceanic crust of the southern sub-basin isdistinct from the central part of the West Philip-pine Basin. Acoustic basement in the southernsub-basin is smoother and is at a shallower depthbelow sea-level than the crust closer to the CentralBasin Fault, suggesting greater crustal thickness[31,35]. Thicker, smoother crust devoid of fracturezones is a characteristic found in many parts ofthe North Atlantic close to Iceland, which White[38] attributed to higher melt £uxes and slowercooling rates as a result of higher mantle potentialtemperature. The di¡erences in crustal structuresuggest the initial, Middle Eocene, phase of open-ing in the West Philippine Basin involved morevoluminous melt generation than during subse-quent spreading, possibly as a result of hottermantle and higher heat-£ow. The initial oceaniccrust may also have formed with a higher spread-ing rate [31,35].

2.3. Non-boninitic Middle Eocene magmatism

The IBM boninite suite was not the only mag-matic event to a¡ect the PSP during the MiddleEocene. Magmatism also occurred in the newlyforming West Philippine Basin and in the Ama-mi^Oki Daito province on the northern margin ofthe rift (Figs. 1 and 2). Rifting was responsible forthe opening of the West Philippine Basin but therange of trace element and isotopic compositionscharacterising magmatism in these two areas can-not be reconciled solely with decompression melt-ing of depleted upper mantle.

Pre-Middle Eocene lavas from the West Philip-pine Basin that are depleted in the most incom-patible elements have Sr, Nd and Pb isotopic ra-tios that resemble I-MORB [27] while MiddleEocene lavas tend to be incompatible element-en-riched and show isotopic evidence for the pres-

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 219

ence of an additional component similar to EM-2[37]. This latter signal is strong in the igneousbasement of DSDP Sites 291, 292 and 294 onthe southernmost and northernmost edges of theWest Philippine Basin (Fig. 1). These sites repre-sent the ¢rst crust that formed during Middle Eo-cene spreading. Incompatible element-enrichedsills with EM-2 isotopic ratios were also emplacedinto early Middle Eocene sediments in the DaitoBasin (at DSDP Site 446; [37,39]) and haveyielded early Middle Eocene radiometric ages[37,40] suggesting this source was availablethroughout the newly forming rift. Further north,Middle Eocene conglomerates from DSDP Site445 (Fig. 1) commonly contain fragments of alkalibasalt with titanaugite phenocrysts [30,41] andpotassic trachyandesite dated as 48.5 þ 2 Ma hasbeen dredged from the Amami Plateau [42]. Hick-ey-Vargas [37] refers to lavas displaying trace el-ement enrichment and EM-2 isotopic a¤nity col-lectively as PSP-OIB. Drilling and dredgingthroughout the northern part of the West Philip-pine Basin suggest that alkaline magmatic activitywas relatively widespread during the Middle Eo-cene. Furthermore, all analyses conducted thusfar for alkaline and tholeiitic lavas with reliableradiometric or stratigraphic age controls indicatethat this part of the basin was characterised bymagmatism with EM-2 OIB geochemistry [37].

In contrast to PSP-OIB, PSP crust created priorto the opening of the West Philippine Basin re-sembles I-MORB [27]. Similarly, the youngestdrilled and dredged sections of the West Philip-pine Basin at DSDP Site 447 and the CentralBasin Fault possess I-MORB trace element andisotopic signatures [8,43,44]. Furthermore, thespreading rate in the main part of the basin mayhave decreased in the latter part of the MiddleEocene [31,34,35]. Therefore, an isotopically dis-tinct source that yielded lavas with large enrich-ments of incompatible trace elements and an EM-2 isotopic signature appears to have been intro-duced into the mantle beneath the West Philip-pine Basin at, or immediately before, the timethat sea£oor spreading began. Towards the endof the Middle Eocene magmatism forming thecentral portion of the West Philippine Basinclosely resembled I-MORB again [8,43]. Suitable

samples are not available to test for the presenceof similar magmatism in the southern sub-basin ofthe West Philippine Basin but its morphologicalsimilarities suggest that it may be a complemen-tary margin to the Amami^Oki Daito province(see above).

2.4. Middle Eocene uplift in the West PhilippineBasin

Several authors have suggested that the EarlyEocene or earliest Middle Eocene saw a period ofuplift a¡ecting the Amami^Oki Daito Province. Alack of in situ shallow water fossils in Upper Cre-taceous to Lower Eocene sediments of the Ama-mi^Oki Daito province correlates with a trans-gression of Paci¢c reefs [30]. However,Nummulites boninensis is a common constituentof the Middle Eocene sections of DSDP/ODPholes and dredge hauls throughout the Amami^Oki Daito province suggesting proximity to ashallow sea environment at this time[29,30,41,47]. An increased abundance of chlorite,serpentine and quartz in Middle Eocene sedi-ments, relative to smectite-dominated Lower Eo-cene beds, led Chamley [46] to propose either adecrease in volcanic activity or regional uplift ofthis area at the beginning of the Middle Eocene.The presence of Middle Eocene conglomeratesand coarse grains and clasts in other contempora-neous deposits of DSDP Site 445, along with theexposure of Cretaceous rocks throughout theAmami^Oki Daito province, are more consistentwith signi¢cant uplift [46,48]. Furthermore, schistclasts in the Daito Ridge conglomerates preserveevidence of two metamorphic events. Radiometricdating identi¢ed an intermediate- to high-pressureregional event during the Late Cretaceous whilelow-pressure contact metamorphism occurred at49 þ 4 Ma [41]. The inclusion of clasts of theyounger event in Middle Eocene conglomeratesindicates signi¢cant erosion, probably as a resultof rapid uplift, at this time. Finally, Misawa et al.[48] suggested that an unconformity observed be-tween the (Cretaceous) basement and the overly-ing Nummulites-bearing strata on seismic linesthroughout the Amami^Oki Daito province re-sulted from uplift of the region at some time dur-

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230220

ing the Early Eocene or the earliest part of theMiddle Eocene.

Subaerial exposure that a¡ected parts of theAmami^Oki Daito province was relatively short-lived and was succeeded by the regional develop-ment of shallow marine, Nummulites-bearing beds[30,41,48]. Conglomerates are coarsest at the baseof the Middle Eocene section but are replaced up-section by sandstones and mudstones [41] and theprovenance of Upper Eocene sediments in thisarea is wholly marine [46]. These observationssuggest relatively rapid elevation to above sea-lev-el occurred in the Amami^Oki Daito provincebetween the end of the Cretaceous and the MiddleEocene which was quickly succeeded by subsi-dence to shallow marine depths [30,41,46,48].

3. Discussion

The evidence outlined in the previous sectionsuggests that several notable tectonic events ac-companied the generation of the IBM boninitesuite. The reconstruction in Fig. 2 places theseevents in a palaeogeographic context. During theMiddle Eocene a curvi-linear array over 2000 kmlong on the northern margin of the Amami^OkiDaito province was subject to IBM boninite mag-matism, which locally succeeded and was region-ally interspersed with arc tholeiites. To the south,sea£oor spreading was initiating between theAmami^Oki Daito province and the southernsub-basin of the West Philippine Basin. The initialcrust formed by spreading, and contemporaneousmagmatism in the Amami^Oki Daito province,had trace element and isotopic characteristics ofOIB [37] and was slightly thicker than crustformed subsequently at the same spreading centre[31,35]. In addition, the Amami^Oki Daito prov-ince was temporarily uplifted and experiencedhigh-temperature, low-pressure metamorphism.Any geodynamic model for genesis of the volumi-nous Middle Eocene boninite magmatism in theIBM forearc must explain these events. In addi-tion a source of excess heat, relative to most con-vergent margins, must have been available in themantle to produce the distinctive features of theboninites themselves.

3.1. A thermal anomaly in the western Paci¢cmantle

There are a number of similarities between theMiddle Eocene western Paci¢c and the large vol-ume basaltic provinces generated during continen-tal rifting (e.g. [49,50]).

1. There was a short-lived, voluminous ma¢cigneous event inferred to result from unusuallyhigh mantle temperatures. In volcanic riftedmargins a combination of hot mantle and de-compression causes genesis of large volumes ofbasaltic magma [38,49,51]. At convergent mar-gins, the combination of a hot mantle sourceand dehydration of downgoing oceanic litho-sphere would also generate large volumes ofbasic magma but with distinctive, subductionzone, characteristics. These are the character-istics displayed by the IBM boninite suite. Rel-ative to average arc growth rates the excessvolume of crust generated in the IBM duringthe Eocene is comparable to excess crustal gen-eration in hotspot settings. Bloomer et al. ([17]following [52]) calculated that crustal produc-tion along 2000 km of the IBM forearc duringa 10 Ma period in the Eocene exceeded meanrates of arc growth by 50^150 km3/km Ma.Such rates are comparable with estimates forthe amount of excess crust generated in theNorth Atlantic for 1000 km to the north andsouth of Iceland (125 km3/km Ma [53]) relativeto normal oceanic crust. The inferred total ex-cess forearc crust produced in the IBM (1^3U105 km3/Ma) is also comparable with esti-mated crustal production rates for Iceland(2.5U105 km3/Ma [53]), Hawaii (1.6U105

km3/Ma [54]) and for the Columbia River ba-salts (1^9U105 km3/Ma [55]) but is a factor of2^10 lower than estimates for large igneousprovinces in oceanic settings, such as OntongJava and Kerguelen [55]. However, subsequent40Ar^39Ar age data indicate that much of theIBM forearc magmatism occurred in a veryshort period [19], therefore, the Stern andBloomer [52] calculations may underestimatethe actual excess growth rate by as much as afactor of two.

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 221

2. The Amami^Oki Daito province experienced atransitory phase of uplift of similar magnitudeto that a¡ecting Greenland during the onset ofNorth Atlantic Tertiary magmatism (91 km[56]). Due to the complexity of the inferredtectonic setting of the Amami^Oki Daito prov-ince it is impossible to unequivocally identifythe mechanism(s) responsible for this uplift.Uplift can occur on the margins of rifts, andrifted arcs in particular can experience complexhistories of uplift and subsidence in the ab-sence of thermal anomalies in the mantle[17,57]. However, subsidence of the Amami^Oki Daito province from the Middle Eoceneto Quaternary is in excess of that expectedthrough isostatic subsidence alone [45] suggest-ing some component of dynamic uplift was as-sociated with the original uplift.

3. Magmatism in and around the nascent WestPhilippine Basin possesses trace element andisotopic characteristics atypical of subductionzone environments and more akin to magma-tism associated with aseismic ridges and vol-canic ocean island chains that are interpretedto represent focussed upwelling of mantleplumes. In some cases aseismic ridges arelinked to £ood basalt provinces associatedwith continental break-up where a plumehead may have been present beneath a verysubstantial area of lithosphere [51]. There issome evidence (outlined above) that the initialoceanic crust at the northern and southernmargins of the West Philippine Basin is thickerand morphologically distinct from the crustgenerated later. Although the thickness doesnot match that generated at present day hot-spots the decrease in thickness and change inmorphology are consistent with a drop in man-tle potential temperature following the initialopening of the basin. Since the exact locationof the basin relative to any thermal anomalycannot be pinpointed it is the changes in thecrust that are the tectonically revealing proper-ties, rather than the actual crustal thickness.

A mantle plume is a potential mechanism forgenerating all the unusual geological and geo-chemical features associated with the PSP in the

Middle Eocene; namely excess heat, high crustalproduction rates, uplift and an OIB-like reservoir.

Mantle plumes are believed to be robust mantlefeatures on timescales of tens of millions of years.Therefore, it might be expected that the presenceof a plume a¡ecting the PSP during the MiddleEocene would have left evidence in the subsequentgeological record, although the rapid rotation ofthe plate during the Eocene and Oligocene [58]means that such evidence would appear as tempo-rary features. Hickey-Vargas [37] suggested that amantle hotspot currently beneath the Caroline is-lands may have been responsible for an OIB sig-nature in West Philippine Basin lavas but thepresent position of Kusaie, the youngest Carolineisland, was signi¢cantly to the northeast of thePSP during the Middle Eocene. The only otherproposed location for a plume in the equatorialwestern Paci¢c is in the Manus Basin. Based onthe petrology, geochemistry and 3He/4He of activemagmatism in the eastern Bismarck Sea a plumeis postulated to exist in the vicinity of the ManusSpreading Centre and the St. Andrew Strait Is-lands [59,60]. Mantle tomography also revealsthat a low velocity anomaly exists in the mantlebeneath the Bismarck Sea [61]. The position ofthe Manus plume, adjusted such that it remains¢xed with respect to the Hawaiian hotspot, isplotted on Fig. 2 and lies within the area thatexperienced the tectonic phenomena outlinedabove. A small circle plotted around this pro-jected location is of similar size to the hypothe-sised plume head associated with impact of theAfar plume [51]. There is a striking similarity be-tween the scale of plume in£uence at Afar and theregion of the western Paci¢c a¡ected by uplift andmagmatism in the Middle Eocene. Upwelling andmagmatism in Greenland and northwest Europeduring the early Tertiary also occurred in a zoneof similar dimensions [38,49].

Fig. 3 explores this coincidence further by com-paring Nd and Pb isotopic compositions of west-ern Paci¢c Middle Eocene lavas with those oflavas recently erupted in the Manus Basin.Although the data are limited, the Manus Basinlavas are transitional between those erupted in theOligocene to Miocene backarc basins of the east-ern PSP and the Middle Eocene PSP-OIB lavas of

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230222

the West Philippine Basin (Fig. 3). The ManusBasin specimen possessing the highest 3He/4Heratio lies very close to the ¢eld of trace element-enriched Middle Eocene PSP-OIB. Also plotted inFig. 3 are isotopic data for Eocene IBM boninites(grey symbols). The majority of these describe anear vertical array between the ¢eld of Paci¢cMORB, with relatively high ONd, and a low ONd

component. The low ONd contaminant could besediment like that thought to contaminate thesource of the active Mariana arc ([62], Fig. 3).The low ONd of Chichijima lavas is consistentwith an input from a very low ONd componentthat could be subducted sediment (see also [8]).However, for a given 206Pb/204Pb the remainingIBM boninites have lower 207Pb/204Pb and 208Pb/204Pb than active Mariana arc lavas with similarONd. This suggests either a di¡erent sediment con-taminant or a di¡erent component altogether. Analternative low ONd contaminant for the majorityof IBM boninite lavas is mantle similar to thesource of PSP-OIB. A role for OIB-type mantlehas previously been proposed to explain both thelow ONd and the high Zr/Sm and Hf/Sm ratios ofIBM boninites [3,6] but recent Hf isotope datasuggest the excess Hf (and Zr) in IBM boninitesmay be acquired from basaltic lithologies at very

shallow depths in the mantle lithosphere or lowercrust [8]. Therefore, use of Hf and Zr may lead toerroneous conclusions regarding the presence [3,6]or absence [5,7] of OIB-type mantle during genesisof IBM boninite. Previous models of OIB involve-ment proposed disparate veins or zones containedin the mantle lithosphere that were composed offrozen metasomatic melts derived from OIB`plums' in the asthenosphere [3,6]. If an OIBsource was responsible for low ONd in IBM bonin-ites the geologic evidence summarised in this worksuggests that it originated in a discrete thermalanomaly in the convecting mantle that waspresent throughout much of the nascent IBMarc/West Philippine Basin system during the earlyMiddle Eocene.

To assess the subsequent role of a thermalanomaly at the present location of the ManusBasin we have traced the movement of plates us-ing a recent plate reconstruction that models thetectonic development of SE Asia and the westernPaci¢c in one million year intervals [28]. The cal-culated `plume track' is illustrated in Fig. 4. Atthe earliest point in the reconstruction the trackruns through the Amami^Oki Daito province to-wards the IBM forearc at ca. 50 Ma. At this stagethe track can be modelled as part of either the

Fig. 3. Initial 206Pb/204Pb and 143Nd/144Nd isotope ratios of lavas from the West Philippine Basin, the IBM forearc, the Palau^Kyushu Arc, Oligocene to Miocene backarc basins of the eastern PSP and the Manus Basin [3^8,25,27,37,43]. Ranges of modernPaci¢c and Indian MORB are shown for comparison. Classi¢cation of PSP lavas as MORB, E-MORB or OIB is based on traceelement characteristics [3,27,37,43].

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 223

northern (Amami^Oki Daito province) or south-ern (southern sub-basin) plate fragments thatwere spreading in the West Philippine Basin(Fig. 2). The change in crustal thickness in thenorthern and southern sections of the West Phil-ippine Basin, compared with the central portion(see above), suggests that both plates may havebeen in£uenced by the anomaly at around 50Ma, therefore, the track can been modelled asbelonging to both fragments. The separation ofthe two tracks during this period (50^45 Ma) isdue to subsequent (N^S, present orientation) ex-tension of the West Philippine Basin (and Palau^Kyushu Ridge) about the Central Basin Fault

(Fig. 4). Both tracks for this period trend towardsthe east under the Palau^Kyushu Ridge suggest-ing the IBM forearc rotated over the thermalanomaly during the late Eocene. Data pertainingto the cessation of boninitic magmatism in theIBM forearc is equivocal. However, it is possiblethat boninites were generated in Guam and Palau,close to the parts of the track crossing the arc, forsome time after the main phase of forearc mag-matism had ended [19]. Later opening of the Par-ece Vela Basin and the Mariana Trough has fur-ther disrupted the northern track (Fig. 4). Aftertraversing the arc the anomaly would have lainbeneath part of the Paci¢c Plate that was subse-

Fig. 4. Map of western Paci¢c and SE Asia showing the 55 Ma^present track of a thermal anomaly currently located beneaththe Manus Basin calculated from the tectonic reconstruction of Hall [28]. Labelled squares show positions at 5 Ma intervals andcircles are positions at intervening 1 Ma intervals from 55 to 30 Ma. A line links subsequent 5 Ma intervals. There are two seg-ments from 50 to 45 Ma since the track can be calculated for either the northern or southern segments of the West PhilippineBasin, which was actively spreading at that time. The northern track is further divided into two parts due to subsequent spread-ing in the Parece Vela Basin and Mariana Trough. From 45 to 33 Ma the track is not marked since it would have been locatedon the Paci¢c Plate lithosphere that has now been subducted. Calculation of the 10 Ma to present day section (dotted line) iscomplicated by tectonics (see text for discussion).

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230224

quently subducted under the east PSP margin andtherefore the track disappears from ca. 46 to 33Ma.

The track reappears at the southern margin ofthe Caroline Plate in the early Oligocene (ca. 33Ma) and crosses the central part of the plate de-scribing a northward path to 25 Ma, after whichit reverses to trend south until 10 Ma (Fig. 4).This path lies close to the Eauripik Rise, a bathy-metric high trending N^S across the CarolinePlate (Fig. 1). Unfortunately the nature of theEauripik Rise is virtually unknown from directsampling. However, bathymetric and seismic pro-¢les suggest it formed through excess volcanism[63,64]. The coincidence of the calculated plumetrack with this feature suggests that the EauripikRise may record the Late Oligocene to Miocenelocation of the plume postulated to presently existin the Manus Basin. Calculation of the track dur-ing the Late Miocene to Quaternary is compli-cated by the large number of plate fragments re-quired in the reconstruction of north New Guineaand the Bismarck Sea. The calculated position at10 Ma lies slightly southwest of Manus Island,but this may result from subsequent subductionof the southernmost Caroline Plate at the ManusTrench. This would result in the 10 Ma plumeposition appearing on the plate that has overrid-den the actual plume-stained lithosphere. Traceelement characteristics of lavas from Manus Is-land and the Admiralty Islands [59,65] were the¢rst evidence used to propose a mantle plume inthe eastern Bismarck Sea. A possible hotspottrace trending northwest beyond Manus Islandand towards the southern Eauripik Rise includeslarge volumes of £at-lying lava £ows that mayhave formed in the Manus Trench forearc [66]but these remain unsampled.

In summary, the geological evidence suggeststhat the mantle currently lying beneath the ManusBasin has generated excess magmatism for sub-stantial portions of the Cenozoic. At the start ofthe Middle Eocene crust formed in the nascentWest Philippine Basin had an EM-2 isotopic sig-nature [37] and was slightly thicker than the crustsubsequently generated in the same basin. In ad-dition, boninitic and basaltic magma were derivedfrom a hot mantle source close to the IBM fore-

arc [18] in volumes that exceed normal arc pro-duction by amounts similar to those characteris-ing active hotspots. From the Oligocene to thepresent day the central Caroline Plate, where theEauripik Rise is located, tracked south then northacross the present position of the Manus Basin.

3.2. Origin of the IBM boninite suite

Taken together these pieces of evidence suggestthat a thermal anomaly has existed in the westernPaci¢c mantle since at least the Middle Eoceneand has in£uenced tectonics and magmatism onseveral plates. Middle Eocene magmatism wasdispersed over an area comparable to that postu-lated for the initial phases of magmatism associ-ated with Iceland and Afar [49,51]. We speculate,therefore, that the Middle Eocene magmatism ofthe western Paci¢c developed its particular char-acter because subduction was initiated in a regionin which there was already a `start-up plume', orin which plume material was very e¤ciently trans-ported laterally in a short period. Such a scenariowould provide the excess heat required to upliftthe Amami^Oki Daito province and to generateIBM boninitic magmatism, and could also pro-vide the distinctive reservoir sampled by lavas inthe overriding plate during the Middle Eocene. Acombination of the excess heat provided by alarge thermal anomaly and £uid released fromthe slab would promote melting of mantle litho-sphere in the overriding plate that is refractory inmost subduction zones [1]. Under these circum-stances ma¢c veins or domains in the shallowlithosphere or lower crust would also be expectedto melt providing the high Zr/Ti, Zr/Sm and Hf/Sm of the boninites [8]. Where temperatures werenot su¤cient to cause melting of the overridinglithosphere £uid £uxing of anomalously hot man-tle wedge would produce large volumes of rela-tively depleted island arc tholeiite [17].

Boninitic magmatism lasted only as long as hotmaterial associated with the thermal anomaly wasavailable to provide heat to the mantle lithosphereof the overriding plate. By the Late Eocene andOligocene magmatism in the West Philippine Ba-sin had reverted to an I-MORB a¤nity suggestingthat the basin had been removed from the zone of

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 225

in£uence of the thermal/chemical anomaly byclockwise rotation [58] and/or a diminution ofthe area in£uenced by the anomaly. In addition,mantle circulation through the mantle wedgewould eventually replace anomalously hot materi-al with upwelling asthenosphere from deeper lev-els. As noted above, localised boninite genesis inGuam and Palau may have post-dated the mainphase of forearc magmatism [19]. This would beconsistent with conversion from a `start-up plume'during the Middle Eocene to a `plume tail' thatsubsequently generated a restricted Oligocene^Miocene aseismic rise on the Caroline Plate.

3.3. Initiation of the IBM arc

There is evidence for limited subduction zonemagmatism in the Izu^Bonin and Mariana arcsprior to the Middle Eocene. Island arc tholeiiteslie beneath boninite lavas at DSDP Site 458 [3]and arc magmatism was active at DSDP Site 296at approximately 48 Ma [36], but at other sitesboninites and arc tholeiites are interbedded[17,18]. The inception of subduction in the IBMarc may have been a localised phenomenon thatwas enhanced by the presence of large contrasts inmantle structure associated with a thermal anom-aly. Warm, buoyant mantle associated with theanomaly would make the overlying lithosphereresistant to subduction while cooler asthenosphereunderlay the lithosphere to the north and east(Fig. 2). Incipient or existing subduction close tothe boundary between these domains could ex-ploit this contrast to rapidly propagate alongthe boundary. Magmatism could commencenearly simultaneously along the margin of thethermal anomaly and would possess a boniniticcharacter due to the unusual thermal structureof the mantle. From this it can be inferred thatthe widespread boninitic magmatism of the IBMforearc is a characteristic of interaction between asubduction zone and a thermal anomaly, ratherthan a characteristic of infant subduction zones.

Since the excess heat required to promote melt-ing of the oceanic mantle lithosphere is providedby the mantle, this model does not require, butdoes not preclude, subduction of an unusually hot(young) slab. Subduction of a young slab at the

initiation of the north-facing IBM arc was ¢rstproposed to account for convergent motion be-tween the arc and the Paci¢c Plate in tectonicreconstructions [67]. The Paci¢c Plate was movingnorthwards prior to the plate motion change at 43Ma suggested by the bend in the Hawaiian-Em-peror seamount chain. A young slab entering theIBM subduction zone was postulated as part of a(now fully subducted) North New Guinea Plate,which was spreading southwards from a WNW^ESE striking spreading centre in the western Pa-ci¢c. As discussed above there are dynamic prob-lems with subduction of young crust over such alarge distance and this explanation also fails tosatisfactorily answer the question as to how thearc was initiated. The absence of substantialamounts of IBM arc crust older than the MiddleEocene requires that subduction of the NorthNew Guinea Plate must have been initiated paral-lel to the spreading ridge in relatively young litho-sphere along a distance of some 2000 km. Remov-al of the North New Guinea Plate from tectonicreconstructions reintroduces the problem of howto accommodate Middle Eocene convergence be-tween the Philippine Sea and Paci¢c plates. Onepossibility is that there was no 43 Ma change inPaci¢c Plate motion [68]. Alternatively, clockwiserotation of the PSP ([28] and references therein),West Philippine Basin spreading [34] and exten-sion in the IBM forearc [52] are all mechanismsthat could have operated at this time to providesuitable plate vectors. There may also have beenother small plates in the western Paci¢c at thistime.

3.4. Other examples of plume^subduction zoneinteraction

Interaction between hot mantle and subductionzones on a scale comparable with that suggestedfor the Middle Eocene IBM arc is not recognisedin any active convergent margins. With the excep-tion of the Izu^Bonin and Mariana forearcs, bo-ninites are present in relatively small volumescompared to coeval magmatism in Cenozoic arcsand in Phanerozoic ophiolites [10^16]. This prob-ably re£ects the di¤culty in promoting shallowmelting of residual harzburgite under the in£u-

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230226

ence of typical arc geotherms, and the generationof other Cenozoic boninite suites may re£ect op-eration of tectonic phenomena that can only lo-cally elevate the geotherm to a su¤cient degree.Proximity to a magmatic rift or spreading centrein the overriding plate, or the subduction ofyoung oceanic lithosphere are among the mostpopular theories for providing excess heat to con-vergent margins [3,5,7,16].

One example of present interaction between aplume tail and a subduction zone may be in theTonga arc. Elevated 3He/4He ratios in magmatismfrom the northern Lau Basin are believed to rep-resent Samoan plume material that entered themantle beneath the Tonga Trench at 3^4 Maand subsequently migrated a short distance tothe south [69,70]. High-Ca boninites have beendredged from the northern termination of theTonga arc [9] and could be the result of interac-tion between hot, depleted mantle from the Samo-an plume and the northern Tonga subductionzone [2].

Examples of Late Archean plume^subductionzone interaction have been postulated in thegreenstone belts of North America. Boninitic-type lavas are interbedded with calc-alkaline andkomatiitic lavas in the Abitibi Belt in Canada [71].Temporal associations of the di¡erent types oflavas vary between di¡erent locations leading Wy-man [71] to suggest that a single plume may haveinteracted with di¡erent arcs over a period of ap-proximately 50 Ma. Interaction between a thermalanomaly and subduction zones as outlined abovefor the PSP and Caroline Plate suggests the west-ern Paci¢c may provide a suitable modern ana-logue for the Archean Abitibi Belt.

4. Summary

Boninites occur in restricted volumes in severalCenozoic locations [1] but the large areal extent ofthe IBM boninite suite is problematic and re-quires an unusual geodynamic setting in the west-ern Paci¢c during the Middle Eocene. This paperprovides a model that explains the voluminousIBM boninitic magmatism in the context of other,simultaneous tectonic events in the region. The

model suggests that widespread boninitic magma-tism of the IBM forearc is a characteristic of in-teraction between a subduction zone and a ther-mal anomaly, rather than a characteristic ofinfant subduction zones. This model may not ap-ply to all locations in which boninites are foundbut may be relevant to northern Tonga and toArchean magmas with boninitic a¤nities.

Plate tectonic reconstruction suggests that OIB-style magmatism and uplift of the northern riftmargin accompanied rifting in the West Philip-pine Basin. The location of these events lies closeto a point in the mantle where a thermal anomalyis presently inferred from petrology and geochem-istry. Furthermore, an aseismic ridge, the Eauri-pik Rise, lies on the part of the Caroline plate thatpassed over the same point during the Oligoceneand Miocene. These facts suggest that excess mag-matism may have been generated close to thisparticular location from the Middle Eocene untilthe present day. The region of high heat-£ow dur-ing the Early Eocene may have interacted with anexisting but restricted subduction zone, or zones,leading to development of a 2000 km magmaticarc and generation of the IBM boninite suite. Theabsence of analogous extensive boninitic suites inthe Phanerozoic rock record may result from theunusual juxtaposition of zones of upwelling(plumes) and downwelling (subduction) inthe mantle. The more restricted volumes of bo-ninite suites found elsewhere in the Cenozoicre£ect interaction between hot mantle and sub-duction on much smaller scales or the operationof other mechanisms that facilitate a suitablecombination of sources and geotherms in subduc-tion zones.

Acknowledgements

Funding from the Southeast Asia ResearchGroup supported by Arco, Canadian Petroleum,Exxon, LASMO, Minorco, Mobil, Union Texasand Unocal is gratefully acknowledged. MartinMenzies, Cindy Ebinger and Brian Taylor arethanked for comments on the manuscript. Sher-man Bloomer provided a review that was bothconstructive and instructive.[AC]

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 227

References

[1] A.J. Crawford, T.J. Falloon, D.H. Green, Classi¢cation,petrogenesis and tectonic setting of boninites, in: A.J.Crawford (Ed.), Boninites, Unwin Hyman, London,1989, pp. 1^49.

[2] T.J. Falloon, L.V. Danyushevsky, Melting of refractorymantle at 1.5, 2 and 2.5 GPa under anhydrous and H2O-undersaturated conditions: implications for the petrogen-esis of high-Ca boninites and the in£uence of subductioncomponents on mantle melting, J. Pet. 41 (2000) 257^283.

[3] R. Hickey-Vargas, Boninites and tholeiites from DSDPSite 458, Mariana forearc, in: A.J. Crawford (Ed.), Bo-ninites, Unwin Hyman, London, 1989, pp. 339^356.

[4] R.J. Stern, J. Morris, S.H. Bloomer, J.W. Hawkins, Thesource of the subduction component in convergent marginmagmas: trace element and radiogenic evidence from Eo-cene boninites, Mariana forearc, Geochim. Cosmochim.Acta 55 (1991) 1467^1481.

[5] J.A. Pearce, M.F. Thirlwall, G. Ingram, B.J. Murton, R.J.Arculus, S.R. van der Laan, Isotopic evidence for theorigin of boninites and related rocks drilled in the Izu^Bonin (Ogasawara) forearc, Leg 125, in: P. Fryer, J.A.Pearce, L. Stokking et al. (Eds.), Proceedings of theOcean Drilling Program, Scienti¢c Results 125, 1992,pp. 237^261.

[6] R. Hickey-Vargas, M.K. Reagan, Temporal variation ofisotope and rare earth element abundances in volcanicrocks from Guam: implications for the evolution of theMariana arc, Contrib. Miner. Petrol. 97 (1987) 497^508.

[7] R.N. Taylor, R.W. Nesbitt, P. Vidal, R.S. Harmon, B.Auvray, I.W. Croudace, Mineralogy, chemistry and gen-esis of the boninite series volcanics, Chichijima, BoninIslands, Japan, J. Petrol. 35 (1994) 577^617.

[8] J.A. Pearce, P.D. Kempton, G.M. Nowell, S.R. Noble,Hf^Nd element and isotope perspective on the nature ofmantle and subduction components in western Paci¢c arc-basin systems, J. Petrol. 40 (1999) 1579^1611.

[9] T.J. Falloon, A.J. Crawford, The petrogenesis of high-calcium boninite lavas from the northern Tonga ridge,Earth Planet. Sci. Lett. 102 (1991) 375^394.

[10] W.E. Cameron, Petrology and origin of primitive lavasfrom the Troodos ophiolite, Cyprus, Contrib. Miner. Pet-rol. 89 (1985) 256^262.

[11] C.A. Evans, G. Casteneda, H. Franco, Geochemical com-plexities preserved in the volcanic rocks of the Zambalesophiolite, Philippines, J. Geophys. Res. 96 (1991) 16251^16262.

[12] L.M. Parson, B.J. Murton, P. Browning, Ophiolites andtheir Modern Oceanic Analogues, Geol. Soc. Spec. Publ.60, 1992.

[13] G. Jenner, Geochemistry of high-Mg andesites from CapeVogel, Papua New Guinea, Chem. Geol. 33 (1981) 307^332.

[14] J. Encarnacion, S.B. Mukasa, C.A. Evans, Subductioncomponents and the generation of arc-like melts in the

Zambales ophiolite, Philippines: Pb, Sr and Nd isotopicconstraints, Chem. Geol. 156 (1999) 343^357.

[15] Y. Tatsumi, S. Maruyama, Boninites and high Mg-ande-sites: tectonics and petrogenesis, in: A.J. Crawford (Ed.),Boninites, Unwin Hyman, London, 1989, pp. 50^71.

[16] W.E. Cameron, Contrasting boninite^tholeiite associa-tions from New Caledonia, in: A.J. Crawford (Ed.), Bo-ninites, Unwin Hyman, London, 1989, pp. 314^338.

[17] S.H. Bloomer, B. Taylor, C.J. MacLeod, R.J. Stern, P.Fryer, J.W. Hawkins, L. Johnson, Early arc volcanismand the ophiolite problem: a perspective from drilling inthe western Paci¢c, in: B. Taylor, J. Natland (Eds.), Ac-tive Margins and Marginal Basins of the Western Paci¢c,American Geophysical Union Geophysical Monograph88, 1995, pp. 1^30.

[18] B. Taylor, Rifting and the volcanic-tectonic evolution ofthe Izu^Bonin^Mariana arc, in: B. Taylor, K. Fujioka etal. (Eds.), Proceedings of the Ocean Drilling Program,Scienti¢c Results 126, 1992, pp. 627^651.

[19] M.A. Cosca, R.J. Arculus, J.A. Pearce, J.G. Mitchell,40Ar/39Ar and K^Ar geochronological age constraintsfor the inception and early evolution of the Izu^Bonin^Mariana arc system, Island Arc 7 (1998) 579^595.

[20] G. Rogers, A.D. Saunders, Magnesian andesites fromMexico, Chile and the Aleutian Islands: implications formagmatism associated with ridge-trench collision, in: A.J.Crawford (Ed.), Boninites, Unwin Hyman, London, 1989,pp. 416^445.

[21] M.J. Hole, A.D. Saunders, G. Rogers, M.A. Sykes, Therelationship between alkaline magmatism, lithospheric ex-tension and slab window formation along continental de-structive plate margins, in: J.L. Smellie (Ed.), Volcanismassociated with Extension at Consuming Plate Margins,Geol. Soc. Spec. Publ. 81, 1995, pp. 265^285.

[22] R.M. Bruce, E.P. Nelson, S.C. Weaver, D.R. Lux, Tem-poral and spatial variations in the southern Patagonianbatholith; constraints on magmatic arc development, in:R.S. Harmon, C.W. Rapela (Eds.), Andean Magmatismand its Tectonic Setting, Geol. Soc. Am. Special Paper265, 1991, pp. 1^12.

[23] N.L. Green, D.L. Harry, On the relationship betweensubducted slab age and arc basalt petrogenesis, Cascadiasubduction system, North America, Earth Planet. Sci.Lett. 171 (1999) 367^381.

[24] A.F. Divis, The petrology and tectonics of recent volcan-ism in the central Philippine Islands, in: D.E. Hayes (Ed.),The Tectonic and Geologic Evolution of Southeast AsianSeas and Islands: Part 1, American Geophysical Union,Geophysical Monograph 23, 1980, pp. 127^144.

[25] J.D. Woodhead, S.M. Eggins, R.W. Johnson, Magmagenesis in the New Britain island arc: further insightsinto melting and mass transfer processes, J. Petrol. 39(1998) 1641^1668.

[26] G.E. Wheller, R. Varne, J.D. Foden, M.J. Abbott, Geo-chemistry of Quaternary volcanism in the Sunda^Bandaarc, Indonesia, and three-component genesis of island-arc

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230228

basaltic magmas, J. Volcanol. Geotherm. Res. 32 (1987)137^160.

[27] S.M. DeBari, B. Taylor, K. Spencer, K. Fujioka, Atrapped Philippine Sea plate origin for MORB from theinner slope of the Izu^Bonin trench, Earth Planet. Sci.Lett. 174 (1999) 183^197.

[28] R. Hall, Reconstructing Cenozoic SE Asia, in: R. Hall,D.J. Blundell (Eds.), Tectonic Evolution of SoutheastAsia, Geol. Soc. Spec. Publ. 106, 1996, pp. 153^184.

[29] T. Shiki, A. Mizuno, K. Kobayashi, Data listing of thebottom materials dredged and cored from the northernPhilippine Sea, in: T. Shiki (Ed.), Geology of the North-ern Philippine Sea, Tokai University Press, Tokyo, 1985,pp. 23^41.

[30] W. Mills, Analysis of conglomerates and associated sedi-mentary rocks of the Daito Ridge, Deep Sea DrillingProject Site 445, in: G. de Vries Klein, K. Kobayashi etal. (Eds.), Initial Reports of the Deep Sea Drilling Project58, US Government Printing O¤ce, Washington, DC,1980, pp. 643^657.

[31] C.L. Mrozowski, S.D. Lewis, D.E. Hayes, Complexities inthe tectonic evolution of the West Philippine Basin, Tec-tonophysics 82 (1982) 1^24.

[32] A. Deschamps, P. Monie, S. Lallemand, S.-K. Hsu, K.Y.Yeh, Evidence for Early Cretaceous oceanic crust trappedin the Philippine Sea Plate, Earth Planet. Sci. Lett. 179(2000) 503^516.

[33] S.D. Lewis, D.E. Hayes, C.L. Mrozowski, The origin ofthe West Philippine Basin by inter-arc spreading, in: G.R.Balce, A.S. Zanoria (Eds.), Geology and Tectonics of theLuzon^Marianas Region, Philippine SEATAR Commit-tee Spec. Publ. 1, 1982, pp. 31^51.

[34] T.W.C. Hilde, C.-S. Lee, Origin and evolution of the WestPhilippine Basin a new interpretation, Tectonophysics 102(1984) 85^104.

[35] K. Okino, Y. Ohara, S. Kasuga, Y. Kato, The PhilippineSea: new survey results reveal the structure and history ofthe marginal basins, Geophys. Res. Lett. 26 (1999) 2287^2290.

[36] M. Ozima, I. Kaneoka, H. Ujiie, 40Ar^39Ar age of rocksand the development mode of the Philippine Sea, Nature267 (1977) 816^818.

[37] R. Hickey-Vargas, Origin of the Indian Ocean-type iso-topic signature in basalts from Philippine Sea platespreading centres: an assessment of the local versuslarge-scale processes, J. Geophys. Res. 103 (1998)20963^20979.

[38] R.S. White, Rift^plume interaction in the North Atlantic,Phil. Trans. R. Soc. Ser. A 355 (1997) 319^339.

[39] D.A. Wood, J.-L. Joron, N.G. Marsh, J. Tarney, M.Treuil, Major- and trace element variations in basaltsfrom the north Philippine Sea drilled during Deep SeaDrilling Project Leg 58: a comparative study of back-arc basin basalts with lava series from Japan and mid-ocean ridges, in: G. de Vries Klein, K. Kobayashi et al.(Eds.), Initial Reports of the Deep Sea Drilling Project 58,

US Government Printing O¤ce, Washington, DC, 1980,pp. 873^894.

[40] E.H. McKee, P.R. Klock, K^Ar ages of basalt sills fromDeep Sea Drilling Project Sites 444 and 446, ShikokuBasin and Daito Basin, Philippine Sea, in: G. de VriesKlein, K. Kobayashi et al. (Eds.), Initial Reports of theDeep Sea Drilling Project 58, US Government PrintingO¤ce, Washington, DC, 1980, pp. 921^929.

[41] H. Tokuyama, M. Yuasa, A. Mizuno, Conglomerate andsandstone petrography, Deep Sea Drilling Project Site445, Philippine Sea, in: G. de Vries Klein, K. Kobayashiet al. (Eds.), Initial Reports of the Deep Sea DrillingProject 58, US Government Printing O¤ce, Washington,DC, 1980, pp. 629^641.

[42] J. Matsuda, K. Saito, S. Zashu, K^Ar age and Sr isotopeof rocks of manganese nodule nuclei from the AmamiPlateau, the West Philippine Sea, in: K. Nakazawa etal. (Eds.), Geological Problems of the Philippine Sea,Geol. Soc. Japan, Tokyo, 1975, pp. 99^101.

[43] R. Hickey-Vargas, Isotope characteristics of submarinelavas from the Philippine Sea: implications for the originof arc and basin magmas of the Philippine tectonic plate,Earth Planet. Sci. Lett. 107 (1991) 290^304.

[44] K. Fujioka, K. Okino, T. Kanamatsu, Y. Ohara, O. Ishi-zuka, S. Haraguchi, T. Ishii, Enigmatic extinct spreadingcentre in the West Philippine backarc basin unveiled,Geology 27 (1999) 1135^1138.

[45] A. Mizuno, Y. Okuda, S. Niagumo, H. Kagami, N. Nasu,Subsidence of the Daito Ridge and associated basins,north Philippine Sea, Am. Assoc. Petrol. Geol. Mem. 29(1979) 239^243.

[46] H. Chamley, Clay sedimentation and paleoenvironment inthe area of Daito Ridge (northwest Philippine Sea) sincethe Early Eocene, in: G. de Vries Klein, K. Kobayashi etal. (Eds.), Initial Reports of the Deep Sea Drilling Project58, US Government Printing O¤ce, Washington, DC,1980, pp. 683^693.

[47] H. Tokuyama, Y.S. Kong, H. Kagami, N. Nasu, Colli-sion of the Amami Plateau with the Ryukyu island arc,in: N. Nasu (Ed.), Formation of Active Ocean Margins,Terra Scienti¢c, Tokyo, 1985, pp. 853^873.

[48] Y. Misawa, H. Aoki, S. Kuroki, Results of seismic re£ec-tion pro¢lings and dredge works around the Daito Ridgegroup region, in: T. Shiki (Ed.), Geology of the NorthernPhilippine Sea, Tokai University Press, Tokyo, 1985, pp.23^41.

[49] R.S. White, D. McKenzie, Mantle plumes and £ood ba-salts, J. Geophys. Res. 100 (1995) 17543^17585.

[50] J.G. Fitton, A.D. Saunders, M.J. Norry, B.S. Hardarson,R.N. Taylor, Thermal and chemical structure of the Ice-land plume, Earth Planet. Sci. Lett. 153 (1997) 197^208.

[51] C.J. Ebinger, N.H. Sleep, Cenozoic magmatism through-out east Africa resulting from impact of a single plume,Nature 395 (1998) 788^791.

[52] R.J. Stern, S.H. Bloomer, Subduction zone infancy: ex-amples from the Eocene Izu^Bonin^Mariana and Jurassic

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230 229

California arcs, Geol. Soc. Am. Bull. 104 (1992) 1621^1636.

[53] G. Ito, J. Lin, C.W. Gable, Dynamics of mantle £ow andmelting at a ridge-centred hotspot: Iceland and the Mid-Atlantic Ridge, Earth Planet. Sci. Lett. 144 (1996) 53^74.

[54] S. Watson, D. McKenzie, Melt generation by plumes astudy of Hawaiian volcanism, J. Petrol. 32 (1991) 501^537.

[55] M.F. Co¤n, O. Eldholm, Scratching the surface: estimat-ing dimensions of large igneous provinces, Geology 21(1993) 515^518.

[56] P.D. Clift, J. Turner, Dynamic support by the Icelandicplume and vertical tectonics of the northeast Atlantic con-tinental margins, J. Geophys. Res. 100 (1995) 24473^24486.

[57] P.D. Clift, ODP Leg 135 Scienti¢c Party, Volcanism andsedimentation in a rifting island-arc terrain: an examplefrom Tonga, SW Paci¢c, in: J.L. Smellie (Ed.), VolcanismAssociated with Extension at Consuming Plate Margins,Geol. Soc. Spec. Publ. 81, 1995, pp. 29^51.

[58] R. Hall, M. Fuller, J.R. Ali, C.D. Anderson, The Philip-pine Sea Plate: magnetism and reconstructions, in: B.Taylor, J. Natland (Eds.), Active Margins and MarginalBasins of the Western Paci¢c, American GeophysicalUnion, Geophysical Monograph 88, 1995, pp. 371^404.

[59] R.W. Johnson, I.E.M. Smith, S.R. Taylor, Hot-spot vol-canism in St Andrew Strait, Papua New Guinea: geo-chemistry of a Quaternary bimodal rock suite, BMR,J. Aust. Geol. Geophys. 3 (1978) 55^69.

[60] C.G. Macpherson, D.R. Hilton, J.M. Sinton, R.J. Poreda,H. Craig, High 3He/4He ratios in the Manus backarcbasin Implications for mantle mixing and the origin ofplumes in the western Paci¢c Ocean, Geology 26 (1998)1007^1010.

[61] W. Spakman, H. Bijwaard, Mantle structure and large-scale dynamics of South-East Asia, in: P. Wilson, G.W.Michel (Eds.), The Geodynamics of S and SE Asia (GEO-DYSSEA) Project, GeoForschingsZentrum, Potsdam,1998, pp. 313^339.

[62] T. Elliott, T. Plank, A. Zindler, W. White, B. Bourdon,Element transport from slab to volcanic front, J. Geo-phys. Res. 102 (1997) 14991^15019.

[63] K.A. Hegarty, J.K. Weissel, Complexities in the develop-

ment of the Caroline Plate region, western equatorial Pa-ci¢c, in: A.E.M. Nairn, F.G. Stehli, S. Uyeda (Eds.), TheOcean Basins and Margins 7B, Plenum Press, New York,1988, pp. 277^301.

[64] S. Altis, Origin and tectonic evolution of the CarolineRidge and Sorol Trough, western tropical Paci¢c, fromadmittance and a tectonic modelling analysis, Tectono-physics 313 (1999) 271^292.

[65] A.L. Jaques, Quaternary volcanism on Manus andM'buke Islands, in: R.W. Johnson (Ed.), Cooke-RavianVolume of Volcanological Papers, Geol. Surv. Papua NewGuinea Mem. 10, 1981, pp. 213^219.

[66] R.W. Johnson, M.R. Per¢t, B.W. Chappell, A.L. Jaques,R.D. Shuster, W.I. Ridley, Volcanism in the New IrelandBasin and Manus Island region: notes on the geochemis-try and petrology of some dredged volcanic rocks from arifted-arc region, in: M.S. Marlow, S.V. Dadisman, N.F.Exon (Eds.), Geology and O¡shore Resources of Paci¢cIsland Arcs: New Ireland and Manus region, Papua NewGuinea, Circum-Paci¢c Council for Energy and MineralResources, Houston, TX, Earth Science Series 9, 1988, pp.113^130.

[67] T. Seno, S. Maruyama, Paleogeographic reconstructionand origin of the Philippine Sea, Tectonophysics 102(1984) 53^84.

[68] I.O. Norton, Tertiary relative plate motions in the NorthPaci¢c: the 43 Ma non-event, Tectonics 14 (1995) 1080^1094.

[69] R. Poreda, Helium-3 and deuterium in back-arc basalts:Lau Basin and the Mariana Trough, Earth Planet. Sci.Lett. 73 (1985) 244^254.

[70] J.I. Wendt, M. Regelous, K.D. Collerson, A. Ewart, Evi-dence for a contribution from two mantle plumes to is-land-arc lavas from northern Tonga, Geology 25 (1997)611^614.

[71] D.A. Wyman, A 2.7 Ga depleted tholeiite suite: evidenceof plume^arc interaction in the Abitibi Greenstone Belt,Canada, Precambrian Res. 97 (1999) 27^42.

[72] M. Ozima, Y. Takigami, I. Kaneoka, 40Ar^39Ar geochro-nological studies on rocks of Deep Sea Drilling Projectsites 443, 445 and 446, in: G. de Vries Klein, K. Kobaya-shi et al. (Eds.), Initial Reports of the Deep Sea DrillingProject 45, US Government Printing O¤ce, Washington,DC, 1980, pp. 917^920.

EPSL 5764 13-3-01

C. G. Macpherson, R. Hall / Earth and Planetary Science Letters 186 (2001) 215^230230