mission immiscible: distinct subduction components ...rjstern/pdfs/tamura... · mission immiscible:...

Mission Immiscible: Distinct SubductionComponents GenerateTwo Primary Magmas atPaganVolcano, Mariana Arc

YOSHIHIKO TAMURA1*, OSAMU ISHIZUKA2, ROBERT J. STERN3,ALEXANDER R. L. NICHOLS1, HIROSHI KAWABATA1,4,YUKA HIRAHARA1, QING CHANG1, TAKASHI MIYAZAKI1,JUN-ICHI KIMURA1, ROBERT W. EMBLEY5 ANDYOSHIYUKI TATSUMI1,6

1INSTITUTE FOR RESEARCH ON EARTH EVOLUTION (IFREE), JAPAN AGENCY FOR MARINE^EARTH SCIENCE AND

TECHNOLOGY (JAMSTEC), YOKOSUKA 237-0061, JAPAN2INSTITUTE OF GEOSCIENCE, GEOLOGICAL SURVEY OF JAPAN/AIST, TSUKUBA 305-8567, JAPAN3DEPARTMENT OF GEOSCIENCES, UNIVERSITY OF TEXAS AT DALLAS, 800 W. CAMPBELL ROAD, RICHARDSON, TX

75080-3021, USA4RESEARCH AND EDUCATION FACULTY, KOCHI UNIVERSITY, KOCHI 780-8520, JAPAN5PACIFIC MARINE ENVIRONMENTAL LABORATORY, NOAA/PMEL, 2115 SE O.S.U. DR., NEWPORT, OR 97365-5258, USA6EARTH AND PLANETARY SCIENCES, KOBE UNIVERSITY, KOBE 657-8501, JAPAN

RECEIVED FEBRUARY 27, 2013; ACCEPTED SEPTEMBER 24, 2013ADVANCE ACCESS PUBLICATION NOVEMBER 8, 2013

Pagan is one of the largest volcanoes along the Mariana arc volcanic

front. It has a maximum elevation of 570 m (Mt. Pagan), but its

submarine flanks descend to 2000^3000 m below sea level, and are

unexplored. Bathymetric mapping and ROV Hyper-Dolphindives (HPD1147 and HPD1148) on the submarine NE and SW

flanks of Pagan were carried out during cruise NT10-12 of R.V.

Natsushima in July 2010. There are no systematic compositional

differences between subaerial lavas reported in the literature and dif-

ferentiated submarine lavas collected in HPD1148, with57 wt %

MgO, suggesting they are derived from the same magmatic system.

However, these differentiated lavas show complexities including

magma mixing; thus we concentrate on magnesian submarine

lavas (47 wt % MgO). Twenty least-fractionated basalts

(48·5^50 wt % SiO2) collected during HPD1147 extend to higher

MgO (10^11wt %) and Mg# (66^70) than the subaerial

lavas. Olivine (up to Fo94) and spinel (Cr# up to 0·8) compos-

itions suggest that these Pagan primitive magmas formed from high

degrees of mantle melting. Two basalt types can be distinguished

based on their geochemistry at similar (10^11wt %) MgO; these

erupted recently, 500 m apart. Both contain clinopyroxene and olivine

phenocrysts and are referred to as COB1 and COB2. LowerTiO2,

FeO, Na2O, K2O, incompatible trace element abundances, and Nb/

Yb suggest that COB1 formed from higher degrees of mantle melting.

In addition, light rare earth element (LREE) enrichment and

higherTh/Nb in COB2 contrast with LREE depletion and lower

Th/Nb in COB1. Higher Ba/Th and Ba/Nb and lowerTh/Nb in-

dicate that the main subduction addition in COB1 was dominated

by hydrous fluid, whereas that in COB2 was dominated by sediment

melt. Sr^Nd^Pb^Hf isotopes are also consistent with this interpret-

ation.These observations suggest that the subduction component re-

sponsible for the greater degree of melting of the COB1 source was

mostly hydrous fluid. The origin of such different metasomatic

agents resulted in different primary magmas forming in the same

volcano. Both hydrous fluid and sediment melt components may

have unmixed from an originally homogeneous supercritical fluid in

or above the subducting slab below the volcanic front. These may

*Corresponding author. Telephone: þ81-46-867-9761. Fax:þ81-46-867-9625. E-mail: [email protected]

� The Author 2013. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 55 NUMBER1 PAGES 63^101 2014 doi:10.1093/petrology/egt061

have been added separately to the mantle wedge peridotite (mantle

diapir) and resulted in two neighboring but completely different pri-

mary magmas from the same diapir. Moreover, these primitive

lavas suggest that even for intra-oceanic arcs assimilation^fractional

crystallization is inevitable when these magmas evolve in the crust

and, in addition, that phlogopite is present in their mantle residue

and thus played an important role in their genesis.

KEY WORDS: basalt; subduction; phlogopite; mantle; igneous

petrology; arc basalt; primary magma; Mariana arc; Pagan

I NTRODUCTIONFinding and studying unfractionated arc basalts is funda-mental to understanding the nature of their mantle sourceand the processes that yield primary magmas above sub-duction zones. Unfortunately, arc lavas are characteristic-ally evolved, multiply-saturated, and rich in phenocrysts,and primitive basalts, representing magmas still nearly inequilibrium with mantle peridotite, are rare. New strate-gies, in particular focusing on the lower flanks of intra-oceanic arc volcanoes using submersibles such as ROVs,are allowing us to break through the crustal filter.Previous work in the Izu^Bonin^Mariana (IBM) arc inthe western Pacific shows that small parasitic cones on thesubmarine flanks of larger volcanoes often erupt moreprimitive lavas than the main edifice, which may be sub-aerial or submarine (e.g. Ishizuka et al., 2008; Tamuraet al., 2011). That is certainly true of the Oligocene toRecent volcanic islands of the IBM arc system, whereeven the least fractionated lavas are still significantly frac-tionated (Tamura & Tatsumi, 2002; Tamura et al., 2010).Tamura et al. (2011) identified two primary basalt magmatypes from the flanks of erupting Northwest Rota-1 volcano(NWR1), �40 km behind the Mariana arc volcanic front(Fig. 1). Based on differences in phenocryst assemblages,Tamura et al. (2011) called these cpx^olivine basalt (COB)and plagioclase^olivine basalt (POB). NWR1 COB has agreater subduction component, both hydrous fluid andsediment melt, than POB (Tamura et al., 2011) Subsequentdives on the submarine flanks of other volcanoes alongthe Mariana arc magmatic front (Pagan, Daon, Alama-gan, Tracey) during R.V. Natsushima cruises NT10-12, July9^19, 2010, and NT12-04, February 13^26, 2012, have re-covered mostly undifferentiated olivine-bearing basalts.Such primitive magmas are likely to have followed periph-eral conduits that allowed them to escape storage and dif-ferentiation in magma chambers existing below thevolcanic summit. We will also show in this study thatsome less magnesian lavas show evidence for magmamixing; however, our main target is the primitive basaltswith410wt % MgO, which have survived crustal filtering.Below we build on the recognition of NWR1 primitive

magma diversity by reporting on two types of primitive

basalts erupted on the NE submarine flank of the largePagan volcano. We call these COB1 and COB2 becausethe two varieties have similar phenocryst assemblages(clinopyroxene and olivine) but can be distinguished onthe basis of their ‘subduction component’; that is, what hasbeen added from the subducted Pacific plate to theirmantle source. Using this discovery, we address why andhow these different primary magmas formed beneath thesame volcano.We find that the observed contrasts betweenthese are best described by calling on immiscible subduc-tion components below the volcanic front, and we playfullyoffer the description ‘mission immiscible’ for what theselavas reveal.

GEOLOGICAL BACKGROUNDThe IBM arc system stretches over 2800 km from nearTokyo, Japan, to beyond Guam, USA, and is an excellentexample of an intraoceanic convergent margin (Stern,2010). The volcanic front of the Mariana segment (Fig. 1a)is subdivided into the Northern Seamount Province(NSP), Central Island Province (CIP) and Southern Sea-mount Province (SSP; Fig. 1a) (Dixon & Stern, 1983;Stern et al., 1988; Bloomer et al., 1989), with the recent iden-tification of the Alphabet Seamount Volcanic Province im-mediately south of the SSP (Stern et al., 2013). Among thefour provinces, subaerial CIP volcanoes and their volcanicproducts are best known (e.g. Woodhead, 1989; Elliottet al., 1997; Stern et al., 2003; Pearce et al., 2005;Wade et al.,2005; Kelley et al., 2010; Marske et al., 2011).Subaerial eruptions of lava or tephra have frequently

occurred in CIP in historical times (e.g. eruptions onAnatahan, North Pagan, Asuncion and Uracas). Eruptionrecords are, however, limited to recent subaerial eventsand little is known about the submarine parts of the CIPvolcanoes [mostly from 1980s dredge sampling of submar-ine volcanoes summarized by Bloomer et al. (1989)].Magma production and supply systems beneath the volca-noes are not well constrained. Very high P and S seismicwave attenuation is found at 50^70 km beneath the arc(Pozgay et al., 2009). Interestingly, by using the chemicalcomposition of primitive NWR1 basalts, Tamura et al.(2011) estimated that the primary basalt magmas segre-gated from their mantle source region at pressures of1·5^2·0GPa (equivalent to depths of 50^65 km). Thesedepths are similar to the equilibration depths of hydrousmelts estimated at 34^87 km beneath the Mariana volcanicarc based on thermobarometry (Kelley et al., 2010), indicat-ing that the low-velocity and high-attenuation region rep-resents the mantle source region for Mariana arc magmas.Pagan is a volcanic complex located about 188100N,

1458450E in the Mariana CIP (Fig. 1b) and is one of thelargest (2160 km3; Bloomer et al., 1989) volcanoes along theMariana arc volcanic front. The island is 16 km long,3^6 km wide, and is elongated NE^SW (Fig. 2a),

JOURNAL OF PETROLOGY VOLUME 55 NUMBER 1 JANUARY 2014

64

comprising an extinct volcanic plateau in the south(Marske et al., 2011) and a simple volcanic coneçtheactive volcano Mt. Pagançthat fills a large caldera in thenorth (Banks et al., 1984). Pagan has a maximum elevationof 570m (Mt. Pagan), but its submarine flanks descend to2000^3000m below sea level (mbsl), so most of the volcanois submarine and unexplored (Fig. 2a). Until recently,little was known about the submarine parts of the volcano,although Bloomer et al. (1989) identified a SSW extensiontowards the submarine volcano Daon. ROV Hyper-Dolphin

dives and bathymetric surveys in the southern Marianaarc, including Pagan, were carried out during cruiseNT10-12 of R.V. Natsushima between July 9 and July 19,2010. We studied small knolls on Pagan’s NE and SW sub-marine flanks during dives HPD1147 (2000^1500 mbsl)and HPD1148 (2350^2010 mbsl), respectively (Fig. 2a).Primitive lavas recovered from the NE flank of the volcanoduring HPD1147 (Fig. 2b) are the focus of this study. Thesamples we describe here were all collected from fresh

pillow lava flows with little sediment cover, suggestingthey erupted recently (Holocene). We also report analysesof differentiated lavas from the SW submarine flanks ofPagan, where lavas similar to those on Pagan Island wererecovered during HPD1148.

Olivine^cpx basalt (COB1) and cpx^olivinebasalt (COB2)Table 1 shows the phenocryst assemblages and modal pro-portions (vol. %) of primitive basalts from HPD1147.Each sampling point is shown in Fig. 2b. Most HPD1147magnesian basalts have phenocrysts and micro-pheno-crysts of olivine (OL) and clinopyroxene (CPX) and lackplagioclase. Samples R01^R06 contain more cpx thanolivine, so these are termed olivine^cpx basalts. For simpli-city, we refer to Pagan magnesian lavas that contain OLand CPX phenocrysts in any proportion as COB, so thisincludes both CPX^OL basalt and OL^CPX basalt.Pagan COB are further divided into three groups on

Fig. 1. (a) Regional map of the Mariana subduction system and neighboring regions of the Pacific and Philippine Sea Plates: the MarianaTrench, Mariana Arc, Mariana Trough, West Mariana Ridge and Parece Vela Basin. The Pacific Plate is subducting beneath the PhilippineSea Plate at 30^40mm a^1 (Seno et al., 1993). Open circles denote ODP drill Sites 800, 801 and 802. The rectangle shows the area enlarged in(b). (b) Map showing the southern Mariana arc, which includes from west to east the MarianaTrough (back-arc basin), the active Marianaarc, the old Mariana forearc (including the islands of Saipan, Tinian, Aguijan, Rota and Guam), and the MarianaTrench. The magmatic arcin this map is part of the Central Island Province (CIP) mostly defined by volcanic islands, extending from Farallon de Pajaros (208320N) toAnatahan (168200N) and the Southern Seamount Province (SSP), from East Diamante (158550N) toTracey (138380N).White dashed line definesthe Mariana Trough spreading ridge. Red dashed line shows the boundary between the active Mariana arc to the west and the upliftedMariana frontal arc, part of the Mariana forearc, which was volcanically active from the Eocene to the Miocene. Pagan lies on the Marianaarc volcanic front. Rectangle shows the location of Fig. 2a.

TAMURA et al. MISSION IMMISCIBLE

65

the basis of phenocryst modes and geochemistry: COB1,COB2, and ‘others’. COB1 are CPX-rich and contain3·8^10·7 vol. % micro-phenocrysts of CPX. CPX micro-phenocrysts are rare in COB2, which have more OLphenocrysts. ‘Others’ are petrographically similar toCOB2, but they are more differentiated than most COB1and COB2 samples and contain reversely zoned olivineand clinopyroxene phenocrysts, suggesting magma mixing.

ANALYT ICAL METHODSAfter sawing and jaw crushing, all samples were pulverizedin an agate ball mill. Major and selected trace elements(Ba, Ni, Cu, Zn, Pb,Th, Rb, Sr,Y, Zr and Nb) were deter-mined by X-ray fluorescence (XRF) at the Institute forResearch on Earth Evolution (IFREE), Japan Agency forMarine^Earth Science and Technology (JAMSTEC).Trace elements were analyzed on pressed powder discs,and major elements were determined on fused glass discs.A mixture of �0·4 g powdered sample and 4 g of anhyd-rous lithium tetraborate (Li2B4O7) was used; no matrixcorrection was applied because of the high dilution. Allsubsequent discussions refer to analyses that have been nor-malized to 100% on a volatile-free basis with total ironcalculated as FeO.Concentrations of other trace elements, including the

rare earth elements (REE), V, Cr, Rb, Sr, Y, Zr, Nb, Cs,Ba, Hf, Ta, Pb, Th and U were determined by inductivelycoupled plasma mass spectrometry (ICP-MS) using aVGPlatform instrument at the Geological Survey of Japan/

AIST. About 100mg of sample powder was dissolved in anHF^HNO3 mixture (5:1). After evaporation to dryness,the residues were redissolved with 2% HNO3 prior toanalysis. Reproducibility is better than �4% [2 standarddeviations (SD)] for the REE, Rb and Nb, and betterthan �6% (2 SD) for other elements (see JB2 analyses inTable 1). Compiled analyses for JB2 are from Govindaraju(1994) andTaylor & Nesbitt (1998).Isotopic compositions of Sr, Nd, and Pb were deter-

mined on 200mg of hand-picked 0·5^1mm rock chips.The chips were leached in 6M HCl at 1408C for 1h priorto dissolution in HF^HNO3. Sr and Nd isotope ratioswere measured on a seven-collector VG Sector 54 massspectrometer at the Geological Survey of Japan/AIST. Srwas isolated using Sr resin (Eichrom Industries, Illinois).For Nd isotopic analysis, the REE were initially separatedby cation exchange before isolating Nd on Ln resin(Eichrom Industries) columns. Sr and Nd isotopic com-positions were determined as the average of 150 ratios bymeasuring ion beam intensities in multidynamic collectionmode. Isotope ratios were normalized to 86Sr/88Sr¼ 0·1194and 146Nd/144Nd¼ 0·7219. Measured values for NBSSRM-987 and JNdi-1 [143Nd/144Nd¼ 0·512115 (Tanakaet al., 2000)] were 87Sr/86Sr¼ 0·710276�6 (2 SD, n¼ 4)and 143Nd/144Nd¼ 0·512104�12 (2 SD, n¼ 4) during themeasurement period.Pb was isolated using AG1-X8 200^400 mesh anion ex-

change resin. Procedural Pb blanks were530 pg, considerednegligible relative to the amount of sample analyzed. Pb iso-topic measurements were made in multidynamic collection

Fig. 2. (a) Pagan and Daon volcanoes from the Mariana Bathymetric Compilation (S. Merle, PMEL/NOAA, compiler) showing the locationsof ROV Hyper-Dolphin dives during NT10-12: the northeastern flank of Pagan (HPD1147), southern flank of Pagan (HPD1148) and southeasternflank of Daon (HPD1149). (b) Bathymetry of Pagan’s northeastern slopes (bathymetry by S. Merle, NOAA) showing HPD1147 dive tracks.During this dive, two distinct types of primitive clinopyroxene^olivine basalt lavas (COB1 and COB2) were collected only 500m apart.


66

mode using the double spike technique [Southampton-Brest-Lead 207^204 spike SBL74 (Ishizuka et al., 2003)] atGSJ/AIST. Natural (unspiked) measurements were madeon 60^70% of collected Pb, giving 208Pb beam intensitiesof (2·5^3·0)� 10^11 A. Fractionation-corrected Pb isotopiccompositions and internal errors were obtained by a closed-form linear double-spike deconvolution (Johnson & Beard,1999). The reproducibility of Pb isotopic measurement (ex-ternal error of 2 SD) by double spike is5200ppm for all20xPb/204Pb ratios. Measured values for NBS SRM-981during the measurement period were 206Pb/204Pb¼16·9401�0·0011, 207Pb/204Pb¼15·5003�0·0025, and 208Pb/204Pb¼ 36·7236� 0·0041.Hf isotopes were determined at IFREE, JAMSTEC.

Prior to sample digestion, the powder sample splits wereleached with 6N HCl at room temperature for 1h, rinsedwith Milli-Q water, and then dried. The samples were di-gested with HF and HClO4, and then dissolved in HCl.Hf was separated by a single-column method using Lnresin, following the method of Mu« nker et al. (2001). The

total procedural blanks for Hf were 525 pg. Hf isotoperatios were measured on a sector-type multi-collector ICP-MS system (Neptune; Thermo Scientific�) at IFREE,JAMSTEC. Mass fractionation was determined from179Hf/177Hf and isotope ratios were normalized to 179Hf/177Hf¼ 0·7325 using an exponential law. 173Yb and 175Lupeaks were monitored to correct for the interference from176Yb and 176Lu on the 176Hf peak. Repeated measurementof the JMC475 and BCR-2 standards yielded 176Hf/177Hf¼ 0·282141�0·000010 (2SD, n¼ 9) and 0·282868�0·000003 (2SD, n¼ 3), respectively, during these analyses.Reported 176Hf/177Hf was further adjusted to JMC475176Hf/177Hf¼ 0·28216.Major and trace element data, together with Sr^Nd^

Pb^Hf isotope data, are reported in Table 2 for samplesfrom dives HPD1147 and HPD1148 that range from basaltto andesite in composition.Microprobe analyses were carried out on a JEOL JXA-

8900 Superprobe equipped with five wavelength-dispersivespectrometers (WDS) at IFREE, JAMSTEC. Counting

Table 1: Phenocryst assemblages and modal proportions (vol. %) of primitive Pagan

basalts (Mg#457 and MgO47·0wt %) from HPD1147

Magma type: COB1

Sample no: R01 R02 R03 R04 R05 R06 R07 R09 R10 R11 R12

Vol. % excluding vesicles

Olivine 0·6 0·6 1·0 0·6 0·7 1·2 0·6 1·9 1·4 0·3 0·3

Micro Ol 0·9 1·7 1·9 1·1 1·4 1·2 0·0 0·0 0·0 0·4 1·2

Clinopyroxene 1·9 2·1 1·7 1·9 1·6 2·1 0·2 1·0 0·9 0·5 0·4

Micro Cpx 6·1 9·9 8·8 9·3 10·7 8·8 0·0 0·0 0·0 6·2 3·8

Plagioclase 0·0 0·0 0·0 0·0 0·0 þ 0·5 0·0 0·0 0·0 0·0

Micro Pl 0·0 0·0 0·0 0·0 0·0 þ 0·0 0·0 0·0 0·2 0·2

Groundmass 90·6 85·7 86·7 87·0 85·6 86·5 98·7 97·1 97·6 92·4 94·2

Magma type: COB1 COB2 Others

Sample no: R13 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23

Vol. % excluding vesicles

Olivine 2·2 0·8 3·6 6·0 5·8 4·4 2·9 4·6 2·5 6·6 2·0

Micro Ol 1·2 1·4 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0

Clinopyroxene 1·0 0·5 2·7 1·9 2·6 2·0 1·5 2·5 1·5 3·0 0·4

Micro Cpx 5·7 8·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0

Plagioclase 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·4 0·1

Micro Pl þ 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0 0·0

Groundmass 89·8 89·3 93·8 92·1 91·6 93·6 95·6 92·9 96·0 90·1 97·5

Modal proportions (vol. %) based on 2000–3000 points counts. Each sample number isprefixed HPD#1147-R.


67

Table 2: Representative major and trace element and Sr^Nd^Pb^Hf isotope data for Pagan lavas

Locality: Pagan northeastern flank

Sample no.: HPD1147R01 HPD1147R02 HPD1147R03 HPD1147R04 HPD1147R05 HPD1147R06 HPD1147R07

Latitude (N): 18813·9250 18813·9160 18813·8920 18813·8520 18813·820 18813·7940 18813·7350

Longitude (E): 145852·20 145852·1710 145852·1330 145852·0630 145852·0250 145851·990 145851·9190

Depth (mbsl): 1992 1978 1944 1893 1851 1812 1749

Magma type: COB1 COB1 COB1 COB1 COB1 COB1 COB1

wt %

SiO2 48·89 49·24 49·04 49·44 49·00 49·37 48·82

TiO2 0·40 0·41 0·40 0·41 0·40 0·40 0·56

Al2O3 13·03 13·15 13·06 13·16 13·04 13·12 16·43

Fe2O3 9·62 9·77 9·61 9·75 9·63 9·76 10·32

MnO 0·17 0·17 0·17 0·17 0·17 0·17 0·17

MgO 10·81 10·86 10·90 11·10 10·93 11·15 7·97

CaO 14·92 14·97 14·98 15·02 14·96 15·02 12·94

Na2O 1·27 1·29 1·27 1·32 1·27 1·28 1·85

K2O 0·24 0·25 0·24 0·25 0·23 0·24 0·38

P2O5 0·06 0·07 0·07 0·07 0·06 0·07 0·09

Total 99·41 100·17 99·73 100·68 99·69 100·59 99·54

Trace element (ppm) by XRF

Ba 83·7 94·1 97·9 81·8 87·2 96·6 111·7

Ni 63·8 62·3 64·5 65·5 64·5 62·4 41·7

Cu 71·9 76·4 73·1 71·5 70·5 68·9 109·3

Zn 60·3 60·4 60·0 60·0 59·5 60·5 71·2

Pb 2·3 2·7 2·1

Th

Rb 3·9 4·1 3·9 4·1 3·9 4·0 6·4

Sr 258·2 259·8 257·9 260·2 260·3 256·6 333·0

Y 9·3 8·7 9·4 9·2 9·3 9·3 11·8

Zr 17·0 17·2 17·3 17·0 17·3 17·1 26·0

Nb 0·5

Trace element (ppm) by ICP-MS

V 228·8 242·0 248·5 246·8 250·5 229·3 276·2

Cr 436·61 436·59 436·17 442·65 467·51 399·78 137·23

Ni 70·2 70·0 66·8 68·6 75·0 66·4 51·4

Rb 3·11 3·24 3·08 3·19 2·92 3·15 5·25

Sr 251 269 257 254 256 262 333

Y 10·2 10·0 9·9 9·9 11·2 10·2 13·2

Zr 17·4 16·9 17·1 17·4 17·1 16·8 26·3

Nb 0·23 0·22 0·22 0·23 0·23 0·23 0·39

Cs 0·14 0·15 0·15 0·16 0·16 0·16 0·23

Ba 70·5 71·9 71·8 73·5 70·2 72·0 109·6

La 1·40 1·38 1·44 1·40 1·42 1·40 2·51

Ce 3·66 3·70 3·85 3·71 3·71 3·68 6·28

Pr 0·55 0·54 0·56 0·56 0·53 0·55 0·87

Nd 2·93 3·01 3·09 3·03 2·94 3·05 4·62

Sm 1·06 1·11 1·10 1·10 1·04 1·15 1·55

(continued)


68

Table 2: Continued



Latitude (N): 18813·9250 18813·9160 18813·8920 18813·8520 18813·820 18813·7940 18813·7350

Longitude (E): 145852·20 145852·1710 145852·1330 145852·0630 145852·0250 145851·990 145851·9190

Depth (mbsl): 1992 1978 1944 1893 1851 1812 1749


Eu 0·44 0·44 0·48 0·45 0·43 0·46 0·61

Gd 1·49 1·40 1·49 1·51 1·42 1·46 1·84

Tb 0·25 0·26 0·26 0·25 0·25 0·25 0·34

Dy 1·62 1·68 1·65 1·59 1·56 1·68 2·09

Ho 0·36 0·35 0·36 0·36 0·35 0·35 0·47

Er 1·05 1·02 1·03 1·06 1·05 1·01 1·31

Tm 0·15 0·15 0·15 0·16 0·15 0·15 0·19

Yb 1·06 1·03 1·01 1·00 1·03 1·01 1·29

Lu 0·15 0·16 0·16 0·15 0·16 0·16 0·20

Hf 0·54 0·53 0·53 0·54 0·53 0·54 0·79

Ta 0·020 0·021 0·019 0·033 0·021 0·019 0·032

Pb 1·09 1·08 1·12 1·10 1·11 1·08 1·50

Th 0·108 0·105 0·105 0·102 0·123 0·099 0·193

U 0·067 0·076 0·070 0·062 0·086 0·060 0·109

87Sr/86Sr 0·703453 0·703455

143Nd/144Nd 0·513059 0·513055

206Pb/204Pb 18·8754 18·8712

207Pb/204Pb 15·5661 15·5635

208Pb/204Pb 38·423 38·4140

176Hf/177Hf 0·283225 0·283212



Latitude (N): 18813·6780 18813·6780 18813·4870 18813·4750 18813·4610 18813·4470 18813·2910

Longitude (E): 145851·8580 145851·8580 145851·5990 145851·5830 145851·5370 145851·4930 145851·1830

Depth (mbsl): 1697 1697 1765 1752 1711 1658 1702


wt %

SiO2 49·31 49·12 48·92 48·73 48·85 48·66 49·16

TiO2 0·57 0·57 0·42 0·41 0·41 0·42 0·56

Al2O3 16·45 16·43 14·40 14·34 14·27 14·36 14·85

Fe2O3 10·59 10·54 9·88 9·82 9·83 9·75 10·33

MnO 0·18 0·18 0·17 0·17 0·17 0·17 0·18

MgO 8·08 7·87 10·45 10·34 10·63 10·24 10·08

CaO 12·95 12·87 14·58 14·55 14·59 14·52 12·84

Na2O 1·87 1·86 1·36 1·36 1·35 1·37 1·71

K2O 0·40 0·40 0·23 0·24 0·25 0·22 0·44

P2O5 0·09 0·09 0·07 0·07 0·07 0·07 0·10

Total 100·48 99·93 100·47 100·03 100·42 99·76 100·23

(continued)


69

Table 2: Continued



Latitude (N): 18813·6780 18813·6780 18813·4870 18813·4750 18813·4610 18813·4470 18813·2910

Longitude (E): 145851·8580 145851·8580 145851·5990 145851·5830 145851·5370 145851·4930 145851·1830

Depth (mbsl): 1697 1697 1765 1752 1711 1658 1702



Ba 118·9 120 77·6 100·9 94·8 97·9 119·9

Ni 45·1 44·3 72·2 70·6 75·0 73·2 109·5

Cu 110·2 108·1 73·8 65·8 72·1 77·4 97·8

Zn 73·6 72·8 62·3 61·4 61 61·9 72·7

Pb 1·7 2·1 2·6 1·9 2·7

Th 0·7 0·8

Rb 6·6 7·2 4·4 4·2 4·7 4·1 6·9

Sr 331·7 332·6 280·9 278·2 277·0 282·3 326·3

Y 12·7 12·0 9·7 9·6 9·6 10·0 12·5

Zr 27·2 26·9 17·9 17·5 17·6 17·9 28·5

Nb 0·5 0·5


V 300·5 245·9 239·6 235·4 234·6 268·4

Cr 155·47 386·61 405·71 409·28 372·31 341·92

Ni 51·3 80·5 78·0 84·2 71·1 112·2

Rb 5·45 2·99 2·97 2·94 2·90 6·13

Sr 332 275 264 266 268 308

Y 14·4 11·4 10·6 10·6 10·9 15·2

Zr 26·2 16·7 16·9 17·3 17·3 29·6

Nb 0·38 0·21 0·23 0·20 0·24 0·43

Cs 0·24 0·15 0·16 0·15 0·14 0·27

Ba 113·6 70·4 70·6 70·8 71·3 112·5

La 2·58 1·45 1·53 1·49 1·52 3·37

Ce 6·40 3·73 3·82 3·89 3·95 7·55

Pr 0·88 0·56 0·56 0·56 0·57 1·03

Nd 4·82 3·05 3·20 3·04 3·17 5·39

Sm 1·56 1·08 1·16 1·15 1·10 1·70

Eu 0·63 0·46 0·48 0·47 0·44 0·69

Gd 1·88 1·43 1·57 1·53 1·47 2·13

Tb 0·35 0·25 0·25 0·26 0·25 0·35

Dy 2·18 1·57 1·69 1·67 1·62 2·25

Ho 0·46 0·35 0·38 0·36 0·36 0·47

Er 1·38 1·05 1·12 1·05 1·04 1·42

Tm 0·20 0·15 0·16 0·16 0·15 0·20

Yb 1·38 0·98 1·02 1·02 1·03 1·36

Lu 0·21 0·16 0·16 0·16 0·15 0·21

Hf 0·79 0·52 0·56 0·52 0·53 0·86

Ta 0·035 0·014 0·023 0·011 0·020 0·035

Pb 1·56 1·05 1·08 1·07 1·05 1·70

Th 0·208 0·116 0·107 0·114 0·109 0·303

(continued)


70

Table 2: Continued



Latitude (N): 18813·6780 18813·6780 18813·4870 18813·4750 18813·4610 18813·4470 18813·2910

Longitude (E): 145851·8580 145851·8580 145851·5990 145851·5830 145851·5370 145851·4930 145851·1830

Depth (mbsl): 1697 1697 1765 1752 1711 1658 1702


U 0·12 0·065 0·065 0·075 0·074 0·148

87Sr/86Sr 0·703473 0·703476 0·703480 0·703475

143Nd/144Nd 0·513042 0·513044 0·513034 0·513033

206Pb/204Pb 18·8596 18·8615 18·8642 18·8960

207Pb/204Pb 15·5718 15·5679 15·5703 15·5737

208Pb/204Pb 38·435 38·417 38·424 38·461

176Hf/177Hf 0·283219 0·283209 0·283198 0·283208



Latitude (N): 18813·2780 18813·2620 18813·2520 18813·2450 18813·2340 18813·2340 18813·0160

Longitude (E): 145851·1430 145851·0920 145851·0750 145851·0530 145851·0220 145851·0220 145850·4460

Depth (mbsl): 1654 1597 1575 1538 1508 1508 1536

Magma type: COB2 COB2 COB2 COB2 COB2 COB2

wt %

SiO2 48·80 49·25 48·77 49·03 48·60 49·16 48·59

TiO2 0·55 0·56 0·55 0·55 0·51 0·55 0·83

Al2O3 14·62 14·76 14·61 14·63 14·31 14·68 14·93

Fe2O3 10·26 10·40 10·25 10·35 10·12 10·37 11·40

MnO 0·18 0·18 0·18 0·18 0·18 0·18 0·18

MgO 10·25 10·44 10·31 10·45 10·87 10·45 9·36

CaO 12·72 12·79 12·69 12·74 12·87 12·77 11·42

Na2O 1·68 1·73 1·69 1·73 1·60 1·72 2·08

K2O 0·43 0·44 0·43 0·42 0·40 0·43 0·77

P2O5 0·10 0·10 0·10 0·10 0·09 0·10 0·17

Total 99·58 100·64 99·57 100·18 99·54 100·43 99·75


Ba 120·7 127·1 125·3 119·8 106·6 133·7 175·9

Ni 119·8 115·7 124·0 117·2 130·1 122·4 133·3

Cu 107·0 91·7 99·2 97·2 92·4 100·6 142·0

Zn 71·5 72·8 72·0 72·7 69·8 73·1 90·1

Pb 1·9 2·3 3·0 2·4 2·1 2·5

Th

Rb 7·5 7·3 7·0 7·5 6·8 7·2 15·4

Sr 322·7 322·6 322·9 319·9 320·6 321·4 314·7

Y 12·6 13·1 13·0 12·5 11·4 12·7 16·8

Zr 28·5 29·0 28·5 29·5 25·8 28·6 50·4

Nb 0·7 0·5 0·6 0·6 0·6 0·6 1·2

(continued)


71

Table 2: Continued



Latitude (N): 18813·2780 18813·2620 18813·2520 18813·2450 18813·2340 18813·2340 18813·0160

Longitude (E): 145851·1430 145851·0920 145851·0750 145851·0530 145851·0220 145851·0220 145850·4460

Depth (mbsl): 1654 1597 1575 1538 1508 1508 1536

Magma type: COB2 COB2 COB2 COB2 COB2 COB2


V 255·2 280·8 242·0 263·0 268·6 313·3

Cr 311·73 460·38 323·89 342·04 361·07 339·02

Ni 132·4 149·6 138·2 121·6 129·4 139·3

Rb 6·13 6·18 6·22 6·09 6·41 14·69

Sr 310 305 294 318 316 318

Y 15·1 13·9 14·3 13·8 14·8 19·8

Zr 28·7 29·6 28·8 29·1 30·0 50·6

Nb 0·43 0·46 0·43 0·46 0·44 1·03

Cs 0·27 0·26 0·26 0·26 0·27 0·54

Ba 114·2 109·0 108·6 113·8 116·1 186·9

La 3·17 3·29 3·13 3·29 3·20 5·63

Ce 7·29 7·66 7·31 7·66 7·57 13·21

Pr 1·01 1·06 0·98 1·05 1·02 1·85

Nd 5·39 5·41 5·15 5·30 5·19 9·12

Sm 1·61 1·66 1·57 1·60 1·55 2·61

Eu 0·63 0·63 0·61 0·59 0·67 0·98

Gd 1·93 2·10 2·00 1·97 1·96 3·28

Tb 0·34 0·34 0·35 0·34 0·34 0·52

Dy 2·16 2·20 2·13 2·15 2·23 3·29

Ho 0·46 0·47 0·46 0·47 0·47 0·67

Er 1·34 1·40 1·36 1·40 1·35 2·05

Tm 0·20 0·20 0·20 0·20 0·19 0·29

Yb 1·40 1·33 1·38 1·34 1·35 1·96

Lu 0·20 0·21 0·20 0·20 0·21 0·30

Hf 0·80 0·83 0·80 0·81 0·80 1·50

Ta 0·035 0·038 0·033 0·038 0·036 0·083

Pb 1·69 1·78 1·68 1·80 1·61 2·45

Th 0·295 0·293 0·280 0·305 0·286 0·512

U 0·150 0·160 0·149 0·167 0·145 0·271

87Sr/86Sr 0·703476 0·703479 0·703477

143Nd/144Nd 0·513027 0·513036 0·513030

206Pb/204Pb 18·8982 18·8968 18·8927

207Pb/204Pb 15·5735 15·5714 15·5683

208Pb/204Pb 38·466 38·464 38·453

176Hf/177Hf 0·283206 0·283198 0·283202

(continued)


72

Table 2: Continued

Locality: Pagan southwestern flank


Latitude (N): 18813·0040 17858·5930 17858·6240 17858·6520 17858·6780 17858·990 17859·0110

Longitude (E): 145850·3670 145838·3250 145838·3270 145838·3140 145838·3110 145838·3040 145838·2980

Depth (mbsl): 1505 2331 2307 2287 2262 2175 2151

Magma type:

wt %

SiO2 49·62 47·89 48·10 48·04 47·89 52·05 51·69

TiO2 0·67 0·77 0·79 0·79 0·77 0·82 0·81

Al2O3 16·57 16·95 17·23 17·12 16·92 16·39 16·28

Fe2O3 8·84 10·45 10·50 10·17 10·49 10·41 10·28

MnO 0·14 0·17 0·17 0·17 0·17 0·18 0·18

MgO 8·15 7·43 7·26 7·16 7·48 5·48 5·40

CaO 13·29 12·83 12·85 12·84 12·76 10·41 10·37

Na2O 2·04 1·81 1·83 1·85 1·80 2·52 2·52

K2O 0·71 0·86 0·86 0·91 0·87 1·28 1·26

P2O5 0·15 0·21 0·21 0·21 0·21 0·20 0·20

Total 100·18 99·38 99·81 99·26 99·36 99·73 98·99


Ba 159·4 128·7 137·0 148·0 130·6 200·5 186·2

Ni 78·7 46·7 44·4 42·7 48·6 20·6 19·7

Cu 151·4 113·7 128·4 133·2 120·2 122·7 135·6

Zn 71·5 71·7 75·5 74·0 72·4 84·8 85·5

Pb 3·2 3·3 3·4 2·2 2·4 3·8 4·0

Th 0·9 0·8 1·0 1·5

Rb 12·0 15·3 14·2 14·8 16·3 24·4 24·5

Sr 372·7 422·1 428·2 432·2 420·7 348·8 348·2

Y 14·3 16·1 16·1 15·8 15·4 21·5 20·8

Zr 42·3 43·3 44·1 45·5 43·8 79·8 79·2

Nb 1·0 1·0 1·1 1·2 0·9 1·8 1·6


V 275·4 345·0 344·2 318·7 330·1 255·5 291·1

Cr 294·81 86·13 129·88 86·26 102·84 33·97 36·51

Ni 96·0 50·9 52·3 49·6 58·7 32·6 24·3

Rb 13·92 15·49 16·37 15·04 15·14 23·50 27·53

Sr 336 406 393 415 408 370 329

Y 15·0 16·8 16·9 19·3 17·5 25·2 22·6

Zr 42·3 43·9 42·7 47·8 44·5 80·8 81·8

Nb 1·09 1·22 1·20 1·13 1·09 1·67 1·78

Cs 0·54 0·47 0·54 0·50 0·44 0·75 0·75

Ba 158·3 131·2 129·2 141·9 131·5 201·6 201·2

La 5·86 7·46 6·76 7·38 6·95 8·29 8·66

Ce 12·63 15·30 14·35 15·46 15·08 17·96 18·65

Pr 1·65 2·08 1·97 2·23 2·08 2·49 2·63

Nd 8·24 9·83 9·53 10·50 10·15 12·09 12·41

Sm 2·28 2·64 2·77 2·97 2·82 3·34 3·44

(continued)


73

Table 2: Continued



Latitude (N): 18813·0040 17858·5930 17858·6240 17858·6520 17858·6780 17858·990 17859·0110

Longitude (E): 145850·3670 145838·3250 145838·3270 145838·3140 145838·3110 145838·3040 145838·2980

Depth (mbsl): 1505 2331 2307 2287 2262 2175 2151

Magma type:

Eu 0·82 1·00 0·94 1·06 1·02 1·11 1·10

Gd 2·52 3·10 2·99 3·27 3·19 3·81 3·80

Tb 0·44 0·48 0·51 0·53 0·51 0·62 0·64

Dy 2·64 2·92 2·99 3·19 2·99 4·02 3·87

Ho 0·55 0·60 0·61 0·64 0·62 0·85 0·81

Er 1·64 1·69 1·69 1·82 1·74 2·45 2·48

Tm 0·24 0·24 0·25 0·26 0·27 0·38 0·38

Yb 1·56 1·51 1·51 1·68 1·63 2·42 2·34

Lu 0·24 0·22 0·24 0·26 0·25 0·38 0·36

Hf 1·13 1·07 1·16 1·25 1·23 2·27 2·20

Ta 0·064 0·066 0·072 0·076 0·074 0·124 0·121

Pb 2·19 1·99 1·96 2·26 2·13 3·18 3·22

Th 0·603 0·566 0·583 0·616 0·625 1·120 1·057

U 0·250 0·265 0·326 0·314 0·308 0·574 0·529

87Sr/86Sr 0·703435 0·703349 0·703482

143Nd/144Nd 0·512983 0·512984 0·513005

206Pb/204Pb 18·8283 18·8875 18·8330

207Pb/204Pb 15·5683 15·5718 15·5714

208Pb/204Pb 38·429 38·467 38·433

176Hf/177Hf 0·283190 0·283158



Latitude (N): 17859·030 17859·0650 17859·070 17859·0910 17859·1060 17859·30 17859·3290

Longitude (E): 145838·2930 145838·2890 145838·2940 145838·2790 145838·2820 1458380 145838·0060

Depth (mbsl): 2121 2090 2077 2047 2037 2200 2178

Magma type:

wt %

SiO2 52·00 51·81 52·04 52·02 60·03 59·01 59·14

TiO2 0·82 0·81 0·82 0·82 0·76 0·80 0·76

Al2O3 16·31 16·32 16·35 16·36 15·49 16·41 15·35

Fe2O3 10·38 10·26 10·33 10·30 8·05 8·73 8·14

MnO 0·18 0·18 0·18 0·18 0·17 0·18 0·18

MgO 5·48 5·31 5·41 5·35 2·16 2·57 2·33

CaO 10·45 10·22 10·35 10·29 5·36 6·40 5·60

Na2O 2·51 2·55 2·56 2·54 3·81 3·70 3·77

K2O 1·27 1·25 1·30 1·24 2·11 1·84 2·02

P2O5 0·20 0·20 0·20 0·20 0·21 0·21 0·21

Total 99·59 98·92 99·52 99·29 98·14 99·85 97·48

(continued)


74

Table 2: Continued



Latitude (N): 17859·030 17859·0650 17859·070 17859·0910 17859·1060 17859·30 17859·3290

Longitude (E): 145838·2930 145838·2890 145838·2940 145838·2790 145838·2820 1458380 145838·0060

Depth (mbsl): 2121 2090 2077 2047 2037 2200 2178

Magma type:


Ba 199·7 193·2 188·5 199·4 379·4 341·5 365·1

Ni 19·2 17·8 19·6 18·7

Cu 130·1 133·6 138·8 136·1 54·6 66·6 58·0

Zn 83·3 84·6 84·1 84·9 78·1 80·2 79·9

Pb 2·9 3·0 4·2 4·4 6·3 5·5 5·5

Th 1·6 0·9 1·1 3·0 1·3 2·8

Rb 23·4 25·1 24·1 24·8 41·9 36·0 37·5

Sr 350·1 350·1 349·3 350·0 279·4 316·0 309·4

Y 20·9 21·1 21·0 21·1 35·6 31·6 33·0

Zr 80·1 81·0 80·6 81·1 170·9 145·8 153·3

Nb 1·6 1·7 1·8 1·4 3·3 2·8 2·6


V 286·0 292·7 274·0 282·1 137·0 158·7 145·7

Cr 35·01 30·12 32·28 36·96 1·43 1·33 4·14

Ni 23·9 29·8 22·6 23·3 3·9 4·3 4·2

Rb 25·17 25·53 24·75 26·04 38·63 38·97 43·19

Sr 336 336 339 349 270 303 295

Y 22·9 23·9 23·5 23·4 37·9 36·2 37·7

Zr 85·9 86·1 84·2 83·7 153·4 155·1 176·4

Nb 1·71 1·58 1·58 1·61 3·05 2·85 3·16

Cs 0·69 0·69 0·68 0·68 1·12 1·06 1·14

Ba 204·9 200·9 202·9 208·7 361·5 356·3 390·0

La 8·54 8·77 8·62 8·52 12·60 11·99 13·09

Ce 18·51 18·41 19·06 18·36 28·13 25·92 28·91

Pr 2·55 2·51 2·55 2·54 4·04 3·65 3·97

Nd 11·83 12·10 12·47 12·12 18·24 17·08 18·07

Sm 3·35 3·26 3·38 3·47 5·08 4·62 4·96

Eu 1·07 1·09 1·11 1·06 1·43 1·31 1·33

Gd 3·95 3·96 3·74 3·71 5·78 5·38 5·75

Tb 0·63 0·64 0·64 0·63 1·04 0·89 0·98

Dy 3·85 3·83 3·99 3·76 6·18 5·49 5·96

Ho 0·84 0·86 0·86 0·83 1·36 1·22 1·34

Er 2·37 2·36 2·38 2·41 4·11 3·56 3·96

Tm 0·36 0·36 0·37 0·35 0·59 0·56 0·60

Yb 2·28 2·34 2·38 2·30 3·97 3·59 3·99

Lu 0·35 0·35 0·37 0·34 0·63 0·55 0·62

Hf 2·16 2·18 2·25 2·16 4·55 3·96 4·41

Ta 0·120 0·116 0·126 0·119 0·232 0·202 0·225

Pb 3·17 3·11 3·31 3·11 5·69 5·11 5·80

Th 1·055 0·988 1·068 1·028 1·960 1·818 2·098

(continued)


75

Table 2: Continued



Latitude (N): 17859·030 17859·0650 17859·070 17859·0910 17859·1060 17859·30 17859·3290

Longitude (E): 145838·2930 145838·2890 145838·2940 145838·2790 145838·2820 1458380 145838·0060

Depth (mbsl): 2121 2090 2077 2047 2037 2200 2178

Magma type:

U 0·519 0·486 0·522 0·485 0·987 0·861 0·973

87Sr/86Sr 0·703440

143Nd/144Nd 0·513030

206Pb/204Pb 18·8249

207Pb/204Pb 15·5716

208Pb/204Pb 38·429

176Hf/177Hf


Sample no.: HPD1148R16 HPD1148R17 HPD1148R18 HPD1148R19 HPD1148R21 HPD1148R22

Latitude (N): 17859·340 17859·3580 17859·3720 17859·4130 17859·4830 17859·5350

Longitude (E): 145838·0020 145838·0060 145838·0090 145837·9880 145837·9760 145837·9740

Depth (mbsl): 2172 2148 2132 2073 2033 2016

Magma type:

wt %

SiO2 59·34 51·82 59·58 59·53 59·79 57·74

TiO2 0·76 0·81 0·80 0·77 0·78 0·82

Al2O3 15·70 16·26 16·22 16·14 15·92 16·46

Fe2O3 8·17 10·29 8·65 8·24 8·23 9·13

MnO 0·18 0·18 0·18 0·18 0·18 0·18

MgO 2·30 5·34 2·50 2·31 2·22 2·89

CaO 5·69 10·25 6·14 6·00 5·67 7·06

Na2O 3·82 2·51 3·80 3·80 3·83 3·57

K2O 2·00 1·26 1·92 1·94 2·02 1·70

P2O5 0·21 0·20 0·22 0·22 0·22 0·21

Total 98·15 98·93 100·01 99·11 98·85 99·75


Ba 344·3 189·1 340·0 372·9 365·3 324·0

Ni 18·2

Cu 59·3 130·4 63·5 59·6 48·4 78·4

Zn 78·8 83·3 80·3 80·9 81·4 78·8

Pb 6·2 3·7 5·2 6·7 5·6 5·6

Th 1·5 1·1 2·0 3·0 1·4 1·5

Rb 38·5 23·9 37·5 38·0 39·8 32·9

Sr 305·0 347·2 309·4 307·5 300·4 322·9

Y 33·7 21·4 33·6 33·8 34·7 31·2

Zr 157·4 81·2 153·6 156·2 161·5 134·6

Nb 2·7 1·6 2·7 3·0 2·9 2·5

(continued)


76

Table 2: Continued


Sample no.: HPD1148R16 HPD1148R17 HPD1148R18 HPD1148R19 HPD1148R21 HPD1148R22

Latitude (N): 17859·340 17859·3580 17859·3720 17859·4130 17859·4830 17859·5350

Longitude (E): 145838·0020 145838·0060 145838·0090 145837·9880 145837·9760 145837·9740

Depth (mbsl): 2172 2148 2132 2073 2033 2016

Magma type:


V 145·6 271·6 149·0 137·0 123·0 181·9

Cr 0·98 34·46 2·30 1·19 1·80 3·56

Ni 3·9 23·1 4·5 3·9 10·4 6·1

Rb 40·40 24·86 39·50 41·26 37·03 34·43

Sr 307 368 309 307 287 328

Y 37·2 25·3 38·3 37·8 38·2 34·0

Zr 164·2 83·3 159·5 166·4 156·1 138·4

Nb 3·11 1·69 3·02 3·21 3·09 2·79

Cs 1·13 0·73 1·11 1·13 1·14 0·97

Ba 369·8 212·8 359·7 371·9 357·3 317·7

La 13·01 8·34 12·19 12·52 12·56 11·59

Ce 27·60 18·89 26·65 27·55 27·88 25·41

Pr 3·82 2·52 3·61 3·76 3·76 3·35

Nd 18·04 12·23 17·04 17·30 17·91 16·02

Sm 4·64 3·44 4·73 4·70 4·85 4·31

Eu 1·34 1·10 1·27 1·29 1·37 1·25

Gd 5·33 3·80 5·25 5·51 5·66 4·75

Tb 0·94 0·67 0·90 0·92 0·94 0·82

Dy 5·77 3·99 5·74 5·68 5·94 5·26

Ho 1·27 0·85 1·22 1·27 1·30 1·18

Er 3·75 2·52 3·65 3·73 4·03 3·30

Tm 0·59 0·36 0·55 0·57 0·58 0·51

Yb 3·78 2·40 3·67 3·71 3·80 3·34

Lu 0·59 0·39 0·56 0·58 0·63 0·50

Hf 4·19 2·35 4·05 4·07 4·30 3·52

Ta 0·222 0·131 0·212 0·213 0·233 0·176

Pb 5·34 3·30 5·31 5·36 5·69 4·70

Th 1·961 1·135 1·931 1·888 1·938 1·677

U 0·915 0·539 0·914 0·901 0·961 0·776

87Sr/86Sr

143Nd/144Nd

206Pb/204Pb

207Pb/204Pb

208Pb/204Pb

176Hf/177Hf


77

times on the olivines were 100 s, with 20 kV acceleratingvoltage, 25 nA beam current, and 5 mm spot diameter,which ensured reliable Ni analyses. Pyroxene, spinel andplagioclase analyses were conducted with 20 s countingtime, 15 kV accelerating voltage, and 15 nA beam current.Representative mineral compositions for HPD1147 andHPD1148 samples are given in Electronic Appendix Table1, which may be downloaded from http://www.petrology.oxfordfournals.org/.

MAGMATIC VAR IAT ION OFPAGAN VOLCANOFigure 3 shows silica variation diagrams for lavas fromPagan, including those previously analysed from theisland (Woodhead, 1989; Elliott et al., 1997; Marske et al.,2011) and those we have collected from its submarineflanks (HPD1147 and HPD1148, Fig. 2a). The data rangefrom 48 to 64wt % SiO2, defining a low- to medium-Ksuite as defined by Gill (1981) (Fig. 3). Subaerial and sub-marine lavas with 452wt % SiO2 exhibit no systematicdifference, suggesting that they came from the samePagan magmatic system. However, more primitive basaltlavas (47wt % MgO and 58^70Mg#) are observed andsampled only from the NE submarine flanks (Fig. 3). Thusthe submarine samples provide the only opportunity to

study the primitive lavas of Pagan volcano. Access to suchprimitive lavas is critical to understand magma genesis inthe mantle wedge above the Mariana subduction zone.Wetherefore focus here on Pagan primitive and near-primitivesubmarine basalts with MgO47wt %.

MINERAL CHEMISTRYOlivineMagnesian basalts (Mg# 458, MgO 47wt %) fromPagan volcano contain olivine phenocrysts (0.3^6.6 vol. %),with rare Cr-spinel inclusions. Cores and rims of 20^30 oliv-ine phenocrysts from each of eight samples of COB1, threeof COB2 and two other basalts recovered by HPD1147(Fig. 2) were measured by electron microprobe. NiO con-tents correlate with Fo content [Fo¼ olivine 100Mg/(MgþFe)], suggesting olivine fractionation. Fo contents ofPagan olivines range from 70 to 94, the most magnesian ofwhich (Fo92^94) contain �0·4wt % NiO, indistinguishablefrom mantle olivines; moreover, some olivines are moremagnesian than mantle olivines (Fig. 4). COB2 olivines con-tain more NiO than those from COB1, tending to be moreiron-rich at the same NiO contents (Fig. 4a and 4b). Figure4c shows olivines in the ‘other lavas’ (HPD1147R22 andR23). These lavas are more differentiated (510wt % MgO)than COB1 and COB2, but the forsterite contents of their

Fig. 3. SiO2 variation diagrams for subaerial Pagan island lavas (Woodhead, 1989; Elliott et al., 1997; Marske et al., 2011) and those collectedfrom Pagan’s submarine flanks by ROV Hyper-Dolphin (HPD1147, 1148). High-, medium- and low-K boundaries after Gill (1981).


78

http://www.petrology.oxfordfournals.org/

http://www.petrology.oxfordfournals.org/

Fig. 4. Variation of Fo (mol % [100Mg/(MgþFe)]) and NiO (wt %) for olivine phenocrysts in Pagan and NW Rota-1 primitive basalts, andolivines in mantle wedge peridotites. The field for mantle wedge olivines in each panel shows the compositional variation of olivines from peri-dotite xenoliths from Avacha volcano, Kamchatka (Ionov, 2010) and serpentinized peridotites recovered by ODP Leg 125 from TorishimaForearc Seamount and Conical Seamount of the Izu^Bonin^Mariana arc (Ishii et al., 1992). Olivines from (a) COB1, (b) COB2, and (c) otherlavas recovered in dive HPD1147. (d) Calculated olivine fractionation trends determined for R13 (representative of COB1) and R15 (representa-tive of COB2) compared with fields for COB1 and COB2. (See text for explanation.) (e) Comparison between COB1 and COB2 from Pagan,COB from NW Rota-1, and mantle olivines. (f) Average Fo�1SD and average wt % NiO�1SD of olivines from Avacha (91·2�0·4 and0·38�0·02, respectively), Torishima Forearc Seamount (91·8�0·6 and 0·38�0·03, respectively) and Conical Seamount (92·0� 0·3 and0·38�0·06, respectively).


79

olivines are in the range of Fo70^93, and overlap COB1 andCOB2 on Fo^NiO diagrams (Fig. 4c). Some olivines in the‘other lavas’ are reversely zoned in terms of Fo content,which might have resulted from magma mixing. Figure 4dshows trends of olivine fractionation calculated from thebulk-rock compositions of HPD1147-R13 (COB1) andHPD1147-R15 (COB2), as described byTamura et al. (2000),assuming that the ratio of oxidized iron to total iron [Fe3þ/(Fe2þþFe3þ)] in Pagan basalt melts was 0·3 (Kelley &Cottrell, 2009). The temperature-dependent olivine^liquidNi partitioning of Li & Ripley (2010) was used to reproducethe olivine fractionation trends. R13 (COB1) and R15(COB2) lavas contain 75 and 110 ppm Ni, respectively, andare calculated to be in equilibrium with Fo90·86 and Fo90·03olivines having 0·109 and 0·135wt % NiO, respectively(Fig. 4d). Addition of 15 and 16% of equilibrium olivine re-produces COB1and COB2 fractionation trends, respectively(Fig. 4d). A slightly lower temperature is required to pro-duce the steeper COB1 (12008C) trend compared with thatof COB2 (12508C) (Fig. 4d).The calculated primary olivinesof R13 and R15 are Fo93·4 and Fo93·0, respectively, whichare similar to the most magnesian olivines in these lavas.Figure 4e compares olivines in NWR1 COB (Tamura

et al., 2011), Pagan COB1 and COB2. At the same forsteritecontent (4Fo85), the NiO content in the olivines increasesfrom Pagan COB1 through COB2 to NWR1 COB, withsome overlap. The most magnesian olivines in NWR1COB (Fo93), however, are less magnesian than those ofPagan (up to Fo94). Figure 4f shows the range and averagesof olivines in mantle wedge peridotite xenoliths fromAvacha, a Kamchatka arc volcano (Ionov, 2010) and IBMmantle wedge peridotites (Torishima Forearc Seamountand Conical Seamount) recovered by Ocean DrillingProgram (ODP) Leg125 (Ishii et al., 1992). The NWR1pri-mary magmas have previously been discussed in terms ofthese mantle wedge peridotite olivines (Tamura et al., 2011).

SpinelFigure 5a shows the relationship between the Fo contentsof olivine and Cr-number [¼ Cr/(AlþCr) atomic ratio]of spinel in Pagan lavas with47wt % MgO. There areno systematic differences in Cr-number between spinels inCOB1 and COB2 (Fig. 5a). Primitive lavas should haveolivine and spinel compositions that are similar to the oliv-ine^spinel mantle array (OSMA), which represents thespinel peridotite residual trend (Arai, 1994). The Cpx/(OpxþCpx) ratio also reflects the fertility of spinel peri-dotite, decreasing from lherzolite (40·1) to harzburgite(50·1) (Arai, 1994). This change accompanies increases inboth olivine Fo and Cr-number of chromian spinel towardsthe refractory (high-Fo, high-Cr-number) end of OSMA.A Cr-number of 0·5^0·6 approximates the boundary be-tween fertile lherzolite and depleted harzburgite (Arai,1994).

Fig. 5. (a) Relationship between spinel Cr-number (Cr#¼ [Cr/(CrþAl) atomic ratio]) and Fo content of its host olivine in primitivebasalts (COB1, COB2 and other lavas) recovered in dive HPD1147from Pagan Volcano. Field labeled OSMA (olivine^spinel^mantlearray) defines residual spinel peridotites (Arai, 1994). The most mag-nesian olivine^spinel pairs lie within the OSMA, suggesting that theseare from the primary magmas in equilibrium with mantle peridotite.(b) Frequency distribution diagrams for spinel Cr# from primitivebasalts of Pagan and NW Rota-1 volcanoes. The average Cr# foreach volcano is 0·66 (1 SD¼ 0·08) and 0·57 (1 SD¼ 0·08), respectively.


80

Thehighlydepletednature of thePagan residualmantle isshown by the higher Cr-numbers of HPD1147 COB spinels(0·66� 0·08) compared with those of NWR1 (Fig. 5b).NWR1 spinels have Cr-numbers typically from 0·5 to 0·6,averaging 0·57�0·08 (Tamura et al.,2011). IBM forearcperi-dotites (Conical andTorishima Seamounts) contain spinelswith Cr-numbers like those of NWR1basalts (0·5^0·6); Cr-numbersofConicalSeamount(0·61�0·14)aremorevariablethan those of Torishima Seamount (0·57�0·05) and reach amaximum of�0·83 (Ishii et al.,1992). Peridotite xenoliths ofAvachavolcanohave chromian spinels with Cr-number ran-ging from 0·5 to 0·8, giving an average of 0·61�0·04(Ishimaru et al.,2007; Ionov,2010).

Olivine and clinopyroxeneFigure 6 shows frequency diagrams of Fo content of olivineandMg-number (Mg#¼ 100Mg/(MgþFe) atomic ratio)of augite phenocrysts in COB1, COB2 and‘other lavas’.MostCOB1 and COB2 lavas are primitive, with 10^11wt %MgO, except R07 and R10. Olivine cores of both COB1andCOB2 with 10^11wt % MgO have Fo contents with modesof Fo88^92. On the other hand, the modes of the Mg# ofCOB1augite cores range from 88 to 90, higher than those ofthe augite cores inCOB2 (Mg#¼ 86^88) (Fig.6).Some olivine and clinopyroxene phenocrysts in ‘other

lavas’ contain reversely zoned olivine and clinopyroxene.Mixing between primitive and differentiated magmasmight play a role in producing these magmas; this is con-sistent with the wide range of olivine Fo contents (Fo70^93;Fig. 4c) and the presence of chromian spinel (Cr# �0·8)in the more iron-rich olivine (Fo84) (Fig. 5a).

TWO BASALT MAGMA TYPESOlivine and clinopyroxene phyric basalts COB1and COB2and ‘other lavas’ (Table 1) contain47 wt % MgO; we willshow below that these magnesian basalts show significantdifferences in major and trace element abundances, incom-patible element ratios and Sr^Nd^Pb^Hf isotope compos-ition. Moreover, Pagan COB1 and COB2 are similar toprimitive ‘island arc ankaramites’ from Western Epi,Vanuatu (Barsdell & Berry, 1990). Thus, this detailedstudy of primitive Pagan basalts has implications forunderstanding arc magmas in other subduction zones.

MgO diagramsFigure 7 shows MgO (wt %) variation diagrams forPagan COB1, COB2 and ‘other lavas’ and ankaramitesfrom western Epi (Barsdell & Berry, 1990). Regressionlines fitted by eye through the COB1 and COB2 dataare shown in each diagram. At the same MgO content(10^11wt % MgO), COB1 and COB2 define distincttrends, except for SiO2. Compared with COB2, COB1 islow inTiO2, FeO* (total Fe as FeO), Na2O and K2O, butis high in CaO, CaO/Al2O3 and Mg# (Fig. 7a). Cpx

fractionation is important for COB1, because CaO andCaO/Al2O3 co-vary with MgO. Figure 7b shows variationdiagrams of MgO (wt %) versus selected trace elements(ppm). All incompatible elements are lower in COB1 thanin COB2 at the same MgO content. However, as shownbelow based on incompatible element ratios, differencesbetween COB1 and COB2, or COB2/COB1 ratios, at thesame MgO (10^11wt %) are not uniform and differ fromelement to element. For example, in COB2 Rb andTh arethree times higher, Nb and Zr two times higher, and Sronly 1·2 times higher compared with COB1. Compatibleelements also behave differently; COB1 and COB2 havesimilar Cr contents, but in COB1 Ni contents (60^80 ppm) are much lower than in COB2 (120^150 ppm).

Trace element abundance: incompatibleelement and rare earth element (REE)patternsFigure 8 shows (a) mid-ocean ridge basalt (MORB)-nor-malized incompatible trace element patterns for COB1and COB2 from Pagan, (b) MORB-normalized patternsfor COB and POB from NWR1, (c) chondrite-normalizedREE plots for COB1, COB2 and ‘other lavas’ from Paganand (d) chondrite-normalized plots for COB and POBfrom NWR1. Interestingly, the differences between PaganCOB1and COB2 incompatible elements that are not mobi-lized by hydrous fluids (Pearce et al., 2005), such as Nb,Ta,Zr, Hf and heavy REE (HREE), are similar to the differ-ences between COB and POB at NWR1.REE patterns reveal differences during COB1 fraction-

ation and between COB1 and COB2 (Fig. 8c). As COB1MgO contents decrease from 11 to 8wt %, REE patternsshift upward in almost parallel trends to each other. Atthe same MgO content (10^11wt %), COB2 have higherREE contents than COB1; light and middle REE (LREEand MREE; from La to Dy) of COB2 are especially en-riched relative to COB1, and thus COB2 show steeperREE patterns (Fig. 8c). On the other hand, NWR1 COBREE patterns are steeper than those of POB (Fig. 8d),which results in the REE patterns for POB and COB over-lapping between La and Dy.

Incompatible element ratiosFigure 9a shows MgO variation diagrams for trace elementratios. These incompatible element ratios show little vari-ation with MgO content, and thus their differences areused as proxies for different subduction and mantle pro-cesses (Pearce et al., 2005). Ba, but not Th, is significantlypartitioned into aqueous fluids derived from the subductionzone, whereas both Ba and Th are significantly partitionedinto siliceous melts (Pearce et al., 2005, and referencestherein). In contrast, Nb is probably mobilized from arutile-bearing slab only in melts at the highest temperatures(Pearce et al., 2005, and references therein). Thus, Ba/Nb,Ba/Th and Th/Nb highlight subduction input; Ba/Nb is a


81

proxy for total subduction addition, Ba/Th is sensitive toshallow (hydrous fluid) subduction addition and Th/Nb isa proxy for deep subduction addition (sediment melting)(Pearce et al., 2005). Assuming that the source mantle peri-dotite is similar and that melting occurs in the spinel

peridotite stability field, the lower Nb/Yb of COB1 relativeto COB2 suggests greater depletion and/or higher degreesof melting of its source, which is also consistent with thelower abundances of COB1 incompatible elements relativeto COB2 at the same MgO content (Figs 7 and 8).

Fig. 6. Frequency distributions for olivine and clinopyroxene phenocryst compositions in COB1, COB2 and other lavas. Black and white barsindicate core and rim compositions, respectively. Dashed vertical lines show Fo90 and Mg#¼ 90 of olivine and clinopyroxene, respectively, tofacilitate comparison between the diagrams.


82

Figure9eandf showsBa/ThvsTh/NbandNb/YbvsBa/Nb.Thehigher Ba/Thand lowerTh/NbofCOB1indicate that themain subduction component added to COB1 is slab-derivedfluid, but in COB2 it is sediment melt. Moreover, Ba/Nb and

Nb/Yb negatively correlate, suggesting that higher totalsubductionaddition resulted inhigher degrees ofmelting.Figure 10 plots primitive mantle normalized (Sun &

McDonough, 1989) La/Sm vs Ba/Th for Pagan lavas

Fig. 7. MgO (wt %) variation diagrams for primitive basalts (MgO47wt %) from dive HPD1147 on Pagan compared with those from westernEpi (Barsdell & Berry, 1990). The arrow on the SiO2^MgO diagram indicates a model trend line for olivine fractionation. Trend lines fitted byeye through HPD1147 COB1 and COB2 data are shown in other diagrams. (a) Variation diagrams of MgO vs major element oxides (all wt %),CaO/Al2O3 and Mg-number (Mg#¼ [100Mg/(Mgþ�Fe)]). (b) Variation diagrams of MgO (wt %) vs selected trace elements (ppm).


83

(continued)

(COB1, COB2 and ‘other lavas’) and for COB and POBfrom NWR1. The fields for the Izu arc, Mariana arc, otherarcs and MORB are from Elliott (2003). Ba/Th vs La/Smdata for arc basalts anti-correlate and define a crescent-shaped field (Elliott, 2003; Turner et al., 2003). Significantly,

primitive Pagan basalts cover the whole range of Marianaarc lavas, and there is a gap between COB1 and COB2.Figure 10 also suggests that COB1 (with higher Ba/Th) isenriched in fluid from altered oceanic crust and that COB2(with higher La/Sm) is enriched in partial melt from

Fig. 7. (continued)


84

subducted sediment. At NWR1, however, Ba/Th of COBand POB are similar, but COB has higher (La/Sm)N com-pared with POB (Tamura et al., 2011).

Pb^Sr^Nd^Hf isotopesPb isotopes

Our new Pb isotopic data for Pagan lavas are integratedwith previously published data for the Southern Mariana

Trough (SMT) (Gribble et al., 1996) and Mariana CentralIsland Province (CIP) (Fig. 11a; Woodhead, 1989; Elliottet al., 1997; Wade et al., 2005; Stern et al., 2006; Marskeet al., 2011). Mariana CIP lavas show positive trends on208Pb/204Pb^206Pb/204Pb and 207Pb/204Pb^206Pb/204Pb dia-grams, extending from Southern Mariana Trough basaltsto Site 801 sediment (Fig. 11a). Southern Mariana Troughbasalts could represent the composition of the underlying

Fig. 8. Normal (N)-MORB-normalized incompatible element patterns for Pagan (COB1and COB2) (a) and NW Rota-1 (COB and POB) (b).N-MORB composition is from Sun & McDonough (1989). (c) C1 chondrite-normalized REE patterns of Pagan primitive basalts (COB1,COB2) and other lavas and (d) NW Rota-1 (COB and POB). C1 chondrite composition is from McDonough & Sun (1995).


85

Fig. 9. (a^d) Variation of selected trace element ratios with MgO (wt %). (a) Ba/Nb is a proxy for total subduction addition, (b) Ba/Th forshallow subduction addition, (c) Th/Nb for deep subduction addition, and (d) Nb/Yb for degree of melting (Pearce et al., 2005). (e) Ba/Th vsTh/Nb, and (f) Nb/Yb vs Ba/Nb.


86

‘ambient mantle’ (Woodhead et al., 2012); thus, these trendssuggest mixing between the local mantle wedge and sub-ducting sediments represented by Site 801 (Plank &Langmuir, 1998).Wade et al. (2005) concluded that Mariana arc lavas form

a steep array in 206Pb/204Pb^207Pb/204Pb space, which isconsistent with mixing between ‘mantle’ Pb and bulk sedi-ment Pb, and moreover, 94% of the Pb in the ‘mantle’end-member appears to be derived from subducted PacificMORB. On the other hand, Hauff et al. (2003) suggestedthat a mixture of �80^84% unaltered ocean crust and�20^16% average basaltic crust (similar in compositionto the average Site 801 basaltic crust shown in Fig. 11a)could serve as the unradiogenic end-member for Marianaarc lavas. However, the shapes of the distribution ofMariana CIP lavas (Fig. 11a) do not suggest mixing, butrather smear between Site 801 sediments and basaltic crust.Primitive Pagan lavas lie within the Pb isotopic range of

Mariana CIP lavas, similar to those of subaerial Paganlavas, shown byWoodhead (1989), Elliott et al. (1997) andMarske et al. (2011). In detail, however, the Pb isotopicratios of subaerial Pagan lavas (Mt. Pagan, South Paganand the Central Volcanic Region) (206Pb/204Pb¼18·72^

18·87 and 208Pb/204Pb¼ 38·34^38·46) extend to slightlylower 206Pb/204Pb and 208Pb/204Pb compared with sub-marine primitive Pagan lavas (18·82^18·91 and 38·41^38·47, respectively) (Fig. 11a).

208Pb/204Pb^206Pb/204Pb and 207Pb/204Pb^206Pb/204Pbvariations show slight but systematic differences betweenPagan COB1 and COB2 lavas (Fig. 11a). COB2 lavas havehigher 208Pb/204Pb and 206Pb/204Pb than those of COB1.207Pb/204Pb ratios overlap, but on average, COB2 arehigher than COB1.

Sr^Nd isotopes

The 87Sr/86Sr (0·703349^0·703482) and 143Nd/144Nd(0·512983^0·513059) ratios of Pagan primitive lavas plotwithin the Sr^Nd isotopic range of Mariana CIP lavas(Fig. 11b). The range of 143Nd/144Nd is similar to that ofsubaerial Pagan Island (0·51297^0·51306) shown byWoodhead (1989), Elliott et al. (1997) and Marske et al.(2011); the 87Sr/86Sr of primitive lavas mostly overlap withthe range of subaerial Pagan Island (0·70340^0·70354),except for one lava with a lower 87Sr/86Sr of 0·703349.COB1 and COB2 have similar 87Sr/86Sr ratios, but most

COB1 have higher 143Nd/144Nd (40·513042) than COB2

Fig. 10. (a) Variation of La/Sm normalized to primitive mantle (Sun & McDonough,1989) vs Ba/Th for Pagan lavas (COB1, COB2 and otherlavas). The fields for the Izu arc, Mariana arc, other arcs and MORB are from Elliott (2003). Significantly, primitive Pagan basalts encompassthe whole range exhibited by the Mariana arc. (b) Enlargement of the area in which Pagan and NW Rota-1 primitive lavas plot.


87

Fig. 11. (a) Pb isotope variations in Pagan, Southern Mariana Trough (SMT) and Mariana Central Island Province (CIP) lavas. NHRL,Northern Hemisphere Reference Line (Hart, 1984). Average ODP Site 801 sediment and sediment field is after Plank & Langmuir (1998).Average Site 801 basaltic crust, average unaltered ocean crust and basaltic crust field are after Hauff et al. (2003). SMTdata are from Gribbleet al. (1996) and Mariana CIP data are fromWoodhead (1989), Elliott et al. (1997),Wade et al. (2005), Stern et al. (2006) and Marske et al. (2011).(b) 143Nd/144Nd vs 87Sr/86Sr isotope variations in Pagan, NW Rota-1, SMTand Mariana CIP lavas. Data for the SMTand the Mariana CIPare the same as in (a). (c) 176Hf/177Hf vs 143Nd/144Nd in Pagan, NW Rota-1, Mariana CIP lavas and MarianaTrough (MT) lavas. The continu-ous line shows the isotopic variability in the MT mantle array of Woodhead et al. (2012). Primitive Pagan and NW Rota-1 basalts plot betweenthe MT mantle array and CIP lavas. Data for the Mariana CIP are fromWoodhead et al. (2001, 2012) andWade et al. (2005). The dashed lineshows a possible mixing line between ambient mantle and subducting sediment (Woodhead et al., 2012), which can explain the primitive Paganisotopic Hf-Nd isotopic compositions. (See text for further discussion.)


88

(143Nd/144Nd50·513036), with ‘other lavas’ having the lowest143Nd/144Nd (Fig. 11b). This vertical trend for Pagan, whereCOB2 has lower 143Nd/144Nd than COB1, is different fromthe NWR1 array, which shows a negative trend in Fig. 11b,with NWR1 COB having lower 143Nd/144Nd and higher87Sr/86Sr than NWR1POB (Fig. 11b).

Nd^Hf isotopes

Figure 11c shows 176Hf/177Hf vs 143Nd/144Nd for Pagan,NWR1, Mariana CIP lavas and the Mariana Trough(MT). The straight line shows the isotopic variability inthe MT mantle array of Woodhead et al. (2012), and corres-ponds to the composition of the mantle wedge (ambientmantle) prior to slab additions. The dashed line definesthe mixing curve between the appropriate MT basalt (am-bient mantle) and bulk subducted sediment comprising amixture of 40% pelagic clay and 60% volcanogenic sedi-ment (Woodhead et al., 2012). Pagan and NWR1 primitivebasalt data (COB1, COB2, POB and COB) lie near themixing curve or between the mixing curve and the MTmantle array (ambient mantle). However, subaerial Paganlavas and lavas from Mariana CIP islands plot on thelow-143Nd/144Nd side of the primitive Pagan lavas.

PR IMARY MAGMASThe compositions of the Pagan primary basalt magmasare estimated based on the method described byTamura et al. (2000, 2011), assuming that Fe3þ/(Fe2þþFe3þ)¼ 0·3 (Kelley & Cottrell, 2009) and usingthe temperature-dependent olivine^liquid Ni partition-ing of Li & Ripley (2010) to reproduce olivine frac-tionation trends (Fig. 4). Pagan primary basalt majorelement compositions were calculated by incrementallyadding equilibrium olivine to the measured compos-itions and recalculating the basalt compositions untilolivines in equilibrium with the recalculated compositionhad Fo and NiO contents similar to those of harzburgiteor dunite (Fo93^94 and 0·35^0·42wt % NiO).Incompatible trace element concentrations in the primarybasalts were calculated with a Rayleigh crystal fraction-ation model using the fraction of liquid (F) obtained bythe olivine addition modeling [see Tamura et al. (2011) forfurther explanation]. Table 3 lists the major and traceelement compositions of the estimated primary COB1 andCOB2 magmas.All primary COB1 magmas have lower TiO2, Al2O3,

Na2O and K2O and higher MgO contents than those ofprimary COB2. Based on peridotite melting experiments[see Kushiro (2001) for review], these differences suggesthigher degrees of melting of the COB1mantle source com-pared with the COB2 mantle source. Low SiO2 (�48·2wt%) and high olivine components on the olivine^plagio-clase^SiO2 projection of Walker et al. (1979) (not shown)suggest segregation pressures of �2·5GPa for both COB1

and COB2. There is, however, one enigma; COB1 containmore CaO than COB2, which is inconsistent with higherdegrees of melting of the COB1 source and expected harz-burgite and dunite residues. This enigma is explored inthe discussion below.Figure 12 shows primitive mantle (PM)-normalized in-

compatible trace element patterns for average primaryPagan COB1 and COB2 and average NWR1 COB andPOB.We assume that the mantle wedge below Pagan andNWR1 before melting was compositionally similar to themost enriched Horoman peridotite and/or the estimatedprimitive mantle (McDonough & Frey, 1989; McDonough& Sun, 1995; Takazawa et al., 2000; Tamura et al., 2011).Based on major element compositions, Tamura et al. (2011)estimated that NWR1 COB and POB formed from 24%and 18% melting, respectively, which is consistent withtheir parallel HREE patterns. Figure 12 shows distinct par-allelism from Dy to Lu in Pagan COB1 and COB2 andNWR1 COB and POB. Thus, based on the concentrationsof these elements in the primary magmas, the degrees ofmelting required to produce Pagan COB1 and COB2 areestimated to be 36% and 27%, respectively.

SUMMARY AND DI SCUSSIONThe subducting slab, mantle wedge and overlying crustcontribute to the genesis of arc magmas. The study ofprimitive arc basalt lavas from Pagan Volcano makes itpossible to characterize and quantify each contribution inmore detail.

COB1 versus COB2Based on phenocryst modes and assemblages (Table 2), thecrystallization sequence of both COB1 and COB2magmas is spinelþolivine, spinelþolivineþ cpx, and fi-nally spinelþolivineþ cpxþplagioclase. There is no sig-nificant difference in spinel composition between COB1and COB2; both come from a depleted mantle source. Focontents of Pagan olivines range from 70 to 94, the mostmagnesian of which (Fo92^94) contain �0·4wt % NiO, in-distinguishable from mantle olivines. COB2 olivines at agiven Fo content contain more NiO than those fromCOB1. COB1 clinopyroxene phenocrysts are more magnes-ian than those of COB2 (Fig. 6) at the same MgO content,which suggests that clinopyroxenes in COB1 crystallizedsooner after olivine began to crystallize than clinopyrox-enes in COB2. Clinopyroxene crystallization starts earlywhen magmas have not degassed water [see Tamura et al.(2011) for references]. Thus the difference between COB1and COB2 clinopyroxene Mg# suggests that the COB1magma was more water-rich than the COB2 magma,which is consistent with the higher Ba/Th, a proxy ofhydrous fluid addition, in COB1 (Fig. 9).All incompatible element abundances at the same MgO

content (10^11wt %) and at the MgO content of


89

Table 3: Estimated primary arc basalts of Pagan volcano



Added Ol %: 17 17 17 17 17 17 15

F: 0·843 0·843 0·843 0·843 0·843 0·843 0·860

Fo: 93·76 93·70 93·79 93·79 93·79 93·79 93·30

wt %

SiO2 48·25 48·23 48·24 48·19 48·22 48·16 47·98

TiO2 0·34 0·35 0·34 0·34 0·34 0·34 0·36

Al2O3 11·13 11·16 11·13 11·11 11·12 11·09 12·43

FeO* 8·67 8·74 8·64 8·67 8·66 8·69 8·86

MnO 0·16 0·17 0·17 0·17 0·17 0·17 0·17

MgO 17·26 17·22 17·31 17·38 17·33 17·44 16·11

CaO 12·83 12·78 12·85 12·76 12·83 12·77 12·66

Na2O 1·09 1·10 1·08 1·11 1·08 1·08 1·17

K2O 0·21 0·21 0·20 0·21 0·20 0·21 0·20

P2O5 0·05 0·06 0·06 0·06 0·05 0·05 0·06

Mg-no. 78·0 77·8 78·1 78·1 78·1 78·1 76·4

ppm

Ni 487 476 491 497 491 472 471

Rb 2·62 2·73 2·59 2·69 2·46 2·65 2·57

Ba 59·4 60·6 60·5 61·9 59·2 60·7 60·6

Th 0·915 0·914 0·940 0·923 0·935 0·909 0·903

U 0·056 0·064 0·059 0·053 0·073 0·051 0·056

Nb 0·19 0·18 0·19 0·19 0·19 0·19 0·18

Ta 0·017 0·017 0·016 0·028 0·017 0·016 0·012

La 1·18 1·16 1·21 1·18 1·20 1·18 1·25

Ce 3·09 3·12 3·24 3·13 3·13 3·10 3·21

Pb 0·92 0·91 0·94 0·92 0·93 0·91 0·90

Pr 0·46 0·45 0·47 0·47 0·45 0·47 0·48

Sr 211 227 217 215 216 221 237

Nd 2·47 2·54 2·60 2·56 2·47 2·57 2·62

Zr 14·7 14·3 14·4 14·6 14·4 14·2 14·4

Hf 0·45 0·45 0·45 0·45 0·45 0·46 0·44

Sm 0·89 0·93 0·93 0·93 0·88 0·97 0·93

Eu 0·37 0·37 0·40 0·38 0·36 0·39 0·40

Gd 1·25 1·18 1·26 1·27 1·20 1·23 1·23

Tb 0·21 0·22 0·22 0·21 0·21 0·21 0·21

Dy 1·36 1·41 1·39 1·34 1·31 1·42 1·35

Y 8·6 8·4 8·4 8·3 9·4 8·6 9·8

Ho 0·31 0·30 0·30 0·31 0·29 0·30 0·30

Er 0·88 0·86 0·87 0·90 0·89 0·85 0·91

Tm 0·13 0·13 0·13 0·13 0·13 0·13 0·13

Yb 0·89 0·87 0·85 0·84 0·87 0·86 0·85

Lu 0·13 0·13 0·13 0·13 0·13 0·14 0·14

(continued)


90

Table 3: Continued



Added Ol %: 15 15 15 16 15 15 14

F: 0·860 0·860 0·860 0·851 0·860 0·860 0·869

Fo: 93·29 93·39 93·30 93·00 92·95 92·96 92·86

wt %

SiO2 48·00 47·94 48·05 48·22 48·26 48·20 48·32

TiO2 0·36 0·36 0·36 0·48 0·48 0·48 0·49

Al2O3 12·43 12·33 12·48 12·73 12·74 12·73 12·86

FeO* 8·84 8·81 8·81 9·28 9·29 9·31 9·28

MnO 0·17 0·17 0·17 0·18 0·18 0·18 0·18

MgO 16·06 16·29 15·99 16·11 15·98 16·05 15·68

CaO 12·69 12·67 12·70 11·07 11·15 11·09 11·23

Na2O 1·18 1·17 1·19 1·47 1·46 1·49 1·49

K2O 0·21 0·22 0·19 0·38 0·37 0·38 0·38

P2O5 0·06 0·06 0·06 0·09 0·09 0·09 0·09

Mg-no. 76·4 76·7 76·4 75·6 75·4 75·4 75·1

ppm

Ni 461 486 480 609 615 593 591

Rb 2·56 2·53 2·49 5·22 5·27 5·31 5·40

Ba 60·7 60·9 61·3 95·8 98·2 93·8 94·3

Th 0·929 0·918 0·905 1·446 1·456 1·529 1·460

U 0·056 0·065 0·063 0·126 0·129 0·137 0·129

Nb 0·20 0·17 0·20 0·37 0·37 0·40 0·37

Ta 0·020 0·010 0·017 0·030 0·030 0·032 0·029

La 1·31 1·28 1·31 2·87 2·73 2·83 2·72

Ce 3·28 3·35 3·40 6·43 6·27 6·59 6·35

Pb 0·93 0·92 0·91 1·45 1·46 1·53 1·46

Pr 0·48 0·48 0·49 0·88 0·87 0·91 0·86

Sr 227 229 231 262 267 263 255

Nd 2·75 2·62 2·72 4·59 4·63 4·65 4·48

Zr 14·6 14·9 14·9 25·2 24·7 25·5 25·0

Hf 0·48 0·44 0·45 0·73 0·68 0·72 0·69

Sm 1·00 0·99 0·95 1·45 1·39 1·43 1·37

Eu 0·41 0·41 0·38 0·58 0·54 0·54 0·53

Gd 1·35 1·32 1·27 1·81 1·66 1·81 1·74

Tb 0·21 0·22 0·22 0·30 0·29 0·29 0·30

Dy 1·45 1·43 1·39 1·92 1·86 1·89 1·85

Y 9·1 9·1 9·4 13·0 12·9 11·9 12·5

Ho 0·33 0·31 0·31 0·40 0·40 0·40 0·40

Er 0·96 0·91 0·89 1·21 1·15 1·20 1·18

Tm 0·14 0·13 0·13 0·17 0·17 0·18 0·17

Yb 0·88 0·88 0·88 1·16 1·21 1·14 1·20

Lu 0·14 0·14 0·13 0·18 0·17 0·18 0·18

(continued)


91

Table 3: Continued

Sample no.: HPD1147R19 HPD1147R21 HPD1147 HPD1147

Magma type: COB2 COB2 COB1average COB2average

Added Ol %: 15 14 16 15

F: 0·860 0·869 0·850 0·862

Fo: 92·98 92·86 93·59 92·93

wt %

SiO2 48·20 48·30 48·13 48·25

TiO2 0·48 0·48 0·35 0·48

Al2O3 12·68 12·82 11·64 12·76

FeO* 9·31 9·31 8·74 9·30

MnO 0·18 0·18 0·17 0·18

MgO 16·10 15·73 16·84 15·94

CaO 11·10 11·21 12·75 11·14

Na2O 1·50 1·50 1·12 1·49

K2O 0·36 0·38 0·20 0·37

P2O5 0·09 0·09 0·06 0·09

Mg-no. 75·5 75·1 77·4 75·3

ppm

Ni 599 583 481 598

Rb 5·24 5·57 2·59 5·34

Ba 97·9 100·9 60·6 96·8

Th 1·550 1·398 0·919 1·473

U 0·144 0·126 0·060 0·132

Nb 0·39 0·39 0·19 0·38

Ta 0·032 0·031 0·017 0·031

La 2·83 2·78 1·23 2·79

Ce 6·59 6·58 3·20 6·47

Pb 1·55 1·40 0·92 1·47

Pr 0·90 0·89 0·47 0·88

Sr 274 274 223 266

Nd 4·56 4·51 2·59 4·57

Zr 25·1 26·0 14·5 25·3

Hf 0·70 0·70 0·45 0·70

Sm 1·38 1·34 0·94 1·39

Eu 0·51 0·58 0·39 0·55

Gd 1·70 1·71 1·26 1·74

Tb 0·29 0·30 0·22 0·30

Dy 1·85 1·94 1·39 1·88

Y 11·8 12·9 8·9 12·5

Ho 0·40 0·41 0·31 0·40

Er 1·20 1·18 0·89 1·19

Tm 0·17 0·17 0·13 0·17

Yb 1·15 1·17 0·87 1·17

Lu 0·17 0·18 0·13 0·17

F, fraction of liquid; Fo is equilibrium olivine Fo.


92

calculated primary magmas (15·7^17·4wt % MgO) arelower in COB1 than in COB2 (Figs 7 and 8, Table 3). Allprimary COB1 magmas have lower TiO2, Al2O3, Na2Oand K2O, and higher MgO contents than those of primaryCOB2 (Table 3). Moreover, Nb/Yb is lower in COB1 (Fig.9), suggesting higher degrees of melting (�36%) of theCOB1 mantle source compared with the COB2 mantlesource (�27%) (Fig. 12).

‘Mission Immiscible’Ba/Nb and Ba/Th variations (Fig. 9) suggest that COB1lavas have total and shallow subduction additions (Pearceet al. 2005) that are larger than those for COB2 lavas. Incontrast, Th/Nb and La/Sm indicate that sediment meltsare more important for COB2 than for COB1 (Figs 9 and10). The negative correlation between Ba/Nb and Nb/Yb(Fig. 9f) suggests that both total subduction addition anddegree of melting of the COB1 source are higher than forthe COB2 source. Higher total subduction addition mighthave resulted in a higher degree of melting of the COB1mantle source. Importantly, the subduction addition thatcaused higher degrees of melting of the COB1 source wasmostly hydrous fluid, not sediment melt (Figs 9 and 10).Pb isotope ratios (Fig. 11a) are consistent with this inter-

pretation. Significantly, even the Pagan primitive basaltlavas, which were collected in the small area shown inFig. 2, do not define a linear trend in Pb isotope space,thus making mixing between two end-members, a singlesubduction component and a single mantle component,unlikely. We suggest that Pb from subducted sediment is

important for COB2 lavas, but Pb in aqueous fluid ismore important for COB1 lavas (Fig. 11a). These lines ofevidence suggest that aqueous fluid and sediment melt co-exist when they are released from the subducting slab(Mibe et al., 2011; Kawamoto et al., 2012).These slab compo-nents might be added separately to the source mantle,which independently generated COB1 and COB2magmas. The mission of the subducted slab is to add sub-duction components to the overlying mantle wedge to pro-duce arc magmas. Thus we refer to this coexistence offluid and melt as ‘Mission Immiscible’.

Pagan versus NW Rota-1 (NWR1);immiscible below volcanic front butmiscible below rear arcPagan and NWR1 are located at the volcanic front and40 km behind the volcanic front of the Mariana arc, re-spectively (Fig. 1). The chemical differences betweenPagan and NWR1 could be attributed to whether the sub-duction components (hydrous fluid and sediment melt)are immiscible or miscible. The subducting Pacific Platebeneath the Mariana arc is old (�160 Ma), thick, coldand dense, and itsWadati^Benioff zone dips steeply (Sternet al., 2003). Thus, the depths to the top of the MarianaWadati^Benioff zone just below Pagan and NWR1 are�100 and �200 km, respectively (Pozgay et al., 2009).Pagan COB2 and NWR1 POB lavas are enriched in

high field strength elements (HFSE) and HREE com-pared with Pagan COB1 and NWR1 COB lavas, respect-ively (Fig. 8). However, in terms of fluid- and melt-mobileincompatible elements, such as Rb, Ba, Th, U, K, LREEand MREE, Pagan and NWR1 lavas differ. In short, thepatterns for the different types of Pagan lavas do notcross, whereas NWR1 COB and POB intersect (Fig. 8).COB2 lavas have higher amounts of sediment melt thanCOB1 lavas, which results in almost parallel incompatibleelement patterns (Fig. 8a) and enrichments in LREE(Fig. 8c). On the other hand, NWR1 COB have higher orsimilar contents of subducted sediment components com-pared with NWR1 POB, which causes the COB and POBpatterns to intersect (Figs. 8b and 8d) and La/Sm to behigher in the COB lavas (Fig. 10). In contrast to the Pagansuites shown in Fig. 11a, NWR1COB and POB lavas showsimpler linear trends on Pb isotope diagrams (Tamuraet al., 2011), suggesting mixing between ambient mantleand sediment melt.In summary, NWR1 COB have a greater subduction

component, of both hydrous fluid and sediment melt,than POB, perhaps reflecting the fact that the subductingslab below NWR1 is 4100 km deeper than it is beneathPagan. At these higher pressures, hydrous fluid and sedi-ment melt could have mixed into a uniform supercriticalfluid (Mibe et al., 2011; Kawamoto et al., 2012), resulting in

Fig. 12. Primitive mantle (PM)-normalized incompatible traceelement patterns for average primary Pagan COB1 and COB2, andNW Rota-1 COB and POB. Primitive mantle values are fromMcDonough & Sun (1995).


93

‘Mission Miscible’, with higher addition of the supercriticalfluid yielding the distinct NWR1COB.

Pagan primitive basalt versus ankaramitefrom western Epi, Vanuatu; does hydrouscarbonatite play a role in ‘MissionImmiscible’?Pagan COB1 and COB2 are similar to primitive island arcankaramites from western Epi, Vanuatu (Fig. 4; Barsdell& Berry, 1990). Basalts from western Epi plot betweenCOB1and COB2 inTiO2, CaO, K2O and CaO/Al2O3 dia-grams at the same MgO (Fig. 4a). Limited trace elementdata for western Epi were published by Barsdell & Berry(1990) and show that Ba and Sr are higher than in PaganCOBs at the same MgO, with Rb, La, Zr, Yb, Cr and Nisimilar to, or between, COB1 and COB2 (Fig. 7b) at thesame MgO content.The terms ankaramite and picrite were used by Barsdell

& Berry (1990) to describe primitive basaltic lavasdominated by clinopyroxene and olivine phenocrysts,respectively, and are independent of the degree of silica-saturation and alkali content. On the other hand, theGlossary of Geology (Neuendorf et al., 2006) defines ankara-mite as an olivine-bearing basanite containing abundantpyroxene. Because of this confusion we wish to play downthe term ‘ankaramite’, and thus consider that our findingsare applicable to volcanic front lavas in other arcs.Discussion of the primitive and primary magmas of west-ern Epi by Barsdell & Berry (1990) and Green et al. (2004)can be applied to the genesis of the Pagan basalts, espe-cially COB1, which are characterized by high CaO andCaO/Al2O3.Green et al. (2004) concluded that a primary magma

with �16^17wt %MgO separates from residual harzburg-ite at �1^2GPa, 1300^13508C and contains dissolvedC^H^O fluid components as (OH)^ and (CO3)

2^ in themelt. The dissolved C^H^O components lower the liqui-dus temperature of the primary magma from above14008C to �13208C at 1·5GPa. Green et al. (2004) also sug-gested that depleted mantle sources may be fertilized byslab-derived carbonatite melts, which enrich CaO relativeto Al2O3 in the primary magma derived from the mantlesource. Evidence for carbonatite metasomatism in spinelperidotite mantle xenoliths has been reported fromwesternVictoria, Australia (Yaxley et al., 1998). The model ofGreen et al. (2004) could explain the enigmatic enrichmentof CaO in COB1 lavas compared with COB2 lavas(Fig. 7a), and thus the mantle source of COB1 might havebeen fertilized by a slab-derived hydrous carbonatite melt.Moreover, carbonatite melts and silicate melts are immis-cible, and the miscibility gap expands with increasing pres-sure and decreasing temperature (e.g. Mattey et al., 1990;Brooker & Kjarsgaard, 2011). Subduction componentsderived from the subducting slab below the volcanic front

at depths of �100 km could consist of hydrous carbonatemelt and silicate sediment melt. The expanded miscibilitygap between hydrous carbonatite and sediment silicatemelt at high pressure would result in the coexistence oftwo subduction components that could be added separatelyto the magma source mantle diapir.

Crustal contribution to arc magmas:‘Mission Inevitable’There are a number of well-documented instances of crus-tal contamination in oceanic arcs. For example, crustalcontamination has been recognized as an important pro-cess in the Lesser Antilles, based principally on (1) correl-ations between isotopic compositions and indices ofdifferentiation (e.g. Thirlwall & Graham, 1984; Davidson,1987; Davidson & Wilson, 2011), (2) large variations inoxygen isotope ratios (Davidson & Harmon, 1989) and (3)overabundance of the most incompatible elements com-pared with those expected from major element fraction-ation models (Davidson & Wilson, 2011).Figure 11c shows 176Hf/177Hf vs 143Nd/144Nd for Mariana

arc volcanoes and the Mariana Trough (MT), with thedashed line representing the mixing curve between the ap-propriate MT basalt (ambient mantle) and a bulk sub-ducting sediment comprising a mixture of 40% pelagicclay and 60% volcanogenic sediment (Woodhead et al.,2012). Lower 143Nd/144Nd in the Mariana CIP volcanoes,which include the subaerial Pagan lavas, results in anoffset from the mixing curve and is widely attributedto the involvement of fluids (Woodhead et al., 2012).However, the primitive basalts from Pagan, NWR1 andChaife plot nearer the mixing curve or between themixing curve and the MT mantle array (ambient mantle;Tamura et al., 2011; Woodhead et al., 2012). Thus the differ-ences between the primitive submarine lavas and theevolved (low-MgO) subaerial lavas might not simply re-flect different primary magmas and/or mantle sources.The basement of the Mariana CI consists of Mio-Pliocenevolcanic rocks with relatively low 143Nd/144Nd (0·5127^0·5129) at the same 176Hf/177Hf (�0·2832) compared withthe Quaternary volcanic rocks (S. Straub, personal com-munication). Thus, assimilation of this low 143Nd/144Ndbasement could change isotopic ratios as primitivemagmas fractionate; the latent heat of crystallization isthought to be the major heat source of assimilation. Aquantitative estimate of assimilation should be conductedonce data are published for the Miocene^Pliocene volcanicrocks, but primitive magmas are likely to escape storageand AFC in magma chambers existing in the low43Nd/144Nd basement. Assimilation might be inevitablefor arc magmas when they evolve, even in an oceanic arc.Thus the role of the crust could be coined as ‘MissionInevitable’.


94

Are refractory dunite bodies residues ofarc magmas?High-Fo olivines and high Cr-number spinels, whichextend to Fo94 and Cr# �0·8, respectively, suggest thatthe Pagan lavas formed from high degrees of mantle melt-ing (Figs 4 and 5). The study of ultramafic rocks is there-fore necessary to complement the study of volcanic rocks.

Cratonic mantle xenoliths

Bernstein et al. (2007) compiled data for many suites ofshallow cratonic mantle xenoliths worldwide, and demon-strated that their olivines have a remarkably narrowrange of Mg#, with an average of �92·8. The olivineMg# generally cannot increase via igneous processesafter orthopyroxene exhaustion, thus Bernstein et al.(2007) explained the consistent Mg# as the result ofmantle melting and melt extraction to the point of ortho-pyroxene exhaustion (e.g. �40% melt extraction from apyrolite mantle), leaving a nearly monomineralic olivineresidue. Similarly, magnesian olivines from Pagan (Fig. 4)and the higher degrees of melting (�36%) (Fig. 12) maysuggest that the COB1mantle source is dunitic.

Pagan olivine and spinel compared with dunites fromIwanaidake peridotite

Olivine and spinel in the HPD1147 Pagan lavas are com-pared with those in a more depleted peridotite body,including dunite and harzburgite from the Iwanaidakeperidotite in central Hokkaido, Japan (Kubo, 2002;Fig. 13). The Iwanaidake peridotite (c. 2 km� 0·5 km inarea) is mainly composed of spinel harzburgite anddunite with a small amount of pyroxenite and chromitite.The dunite commonly forms tabular layers or veins thatrange in thickness from 10 cm to 10m within massive harz-burgite (Kubo, 2002; Tamura & Arai, 2005). Kubo (2002)showed that although the harzburgite is depleted in bas-altic components, the dunite is even more depleted; thedunites have higher olivine Fo, ranging from 92·5 to 94;higher spinel Cr-number, 60^85; and lower olivine NiOcontent, from 0·35 to 0·37wt % (Fig. 13). This highlydepleted dunite may have formed by extensive harzburgitemelting, with incongruent melting of orthopyroxene trig-gered by injection of hydrous melts into the host harzburg-ite (Kubo, 2002). The Iwanaidake peridotite is suggestedto have been emplaced in the mantle wedge above a sub-duction zone when the dunite and chromitite formed(Kubo, 2002; Tamura & Arai, 2005).There is a similarity between the Pagan lavas and the

refractory Iwanaidake peridotite body (Kubo, 2002), interms of olivine and spinel compositions. Figure 13a showsolivine^Cr-spinel pairs from Pagan primitive basalts andthose in the dunite and harzburgite from Iwanaidake. Inthe Iwanaidake peridotites, olivine Fo content correlateswith spinel Cr-number and increases from harzburgite to

Fig. 13. (a) Relationships between olivine and spinel compositions inPagan primitive basalts, and dunite and harzburgite from Iwanaidakeperidotite, Hokkaido, Japan (Kubo, 2002). The range of Cr# of spinelin abyssal peridotite (Dick & Bullen, 1984) is shown by the double-pointed arrow. In the Iwanaidake peridotites, olivine Fo contents correl-ate with spinel Cr# and increase from harzburgite to dunite (Kubo,2002). (b) Fo^NiO variation diagram for olivines in Pagan basalts, anddunites and harzburgites from Iwanaidake (Kubo, 2002). Olivine frac-tionation trends are indicated (c) Calculated primary olivines in equi-librium with the primary magmas of COB1 and COB2.


95

dunite (Fig. 13a; Kubo, 2002). Most magnesian olivinesfrom Pagan reach Fo94, but spinel inclusions are notobserved. However, spinels within magnesian olivines(Fo91^93) have very high Cr-numbers (0·7^0·8), similar todunite spinels from Iwanaidake (Fig. 13a). Moreover,Pagan spinels within olivines (Fo84^91) have Cr-numbers of0·5^0·8 (Fig. 5b), similar to spinels in Iwanaidake harz-burgite (Fig. 13a). These high Cr-numbers suggest that theresidues of Pagan lavas in the mantle wedge are dunite^harzburgite.Figure 13b shows the variation of NiO wt % and Fo mol

% for olivine phenocrysts in Pagan basalts compared witholivines from dunites and harzburgites from Iwanaidake(Kubo, 2002). Olivines are in the range of Fo91·5^93 andFo92·5^94 in harzburgite and dunite, respectively. Generalolivine fractionation trends show strong decreases in NiOcontent and thus have steep trends in this diagram(Fig. 13b). The compositional variation of the olivinephenocrysts (Fig. 13b) also suggests that the residue of thePagan primary magmas could be a mixture of dunite andharzburgite.Calculated primary olivines, which are in equilibrium

with the calculated COB1 and COB2 primary magmas,are shown in Fig. 13c. Calculated primary olivines fromCOB1 plot in the dunite field and those from COB2 plotnear the Iwanaidake harzburgite field. COB1 olivinephenocrysts have lower NiO contents than those of COB2at the same Fo content, Fo90 (Fig. 4), suggesting thatCOB1 might have crystallized more Fo-rich (Fo490) oliv-ines than those of COB2. In summary, the primary Paganmagmas could have been in equilibrium with mantle resi-dues ranging from harzburgite to dunite (Fig. 13c).These observations suggest that the residual mantle of

the primary Pagan magmas might consist of harzburgiteand dunite, similar to the Iwanaidake peridotite and therefractory peridotite bodies of the Bay of IslandsOphiolite (Suhr et al., 2003). However, the mode of duniteformation, suggested by both Kubo (2002) and Suhr et al.(2003), is different from the crystal^liquid mush (diapir)model for arc magma generation (e.g. Green et al., 1967;Tamura et al., 2011). The latter involves simply batch melt-ing; that is, the magma remains at the site of melting andis in chemical equilibrium with the solid residue untilmechanical conditions allow it to escape as a single ‘batch’of primary magma.We propose that COB1 and COB2 represent batches of

primary magma derived from an originally similar peri-dotite source, but reflect different degrees of melting.Assuming that COB1 and COB2 are counterparts ofdunite and harzburgite residues, respectively, we canmake the following deductions.

COB1^COB2 pairs vs harzburgite^dunite pairs

One interesting aspect of the differences between COB1and COB2 is that the NiO contents are systematically

lower in COB1 olivines compared with COB2 olivines(Fig. 4) and that Ni contents in COB1 whole-rocks(60^80 ppm) are much lower than in COB2 whole-rocks(120^150 ppm) at the same MgO content (10^11wt %MgO) (Fig. 7b). This consistency (Fig. 4) suggests that theolivine phenocrysts are nearly in equilibrium with theirhost basalts.Kubo (2002) and Suhr et al. (2003) made similar obser-

vations in the Iwanaidake peridotite and the Bay ofIslands Ophiolite bodies, in which a decrease in NiO(�0·05wt %) in olivine is associated with an increase inFo content (�1mol %) along the harzburgite^duniteboundary (Fig. 13). Internally, the dunite bodies arenearly homogeneous (Kubo, 2002; Suhr et al., 2003). A pri-mary magma in equilibrium with such dunite olivinewould crystallize olivine that is more magnesian butlower in NiO than a magma in equilibrium with harzburg-ite olivine. Olivine fractionation trends on NiO^Fo dia-grams are so steep (Figs 4 and 13) that olivines inmagmas segregated from dunite residues will have lowerNiO contents at the same Fo content compared with oliv-ines in magmas segregated from harzburgite residues.Thus the differences between the COB1and COB2 olivines(Fig. 4) are consistent with the idea that COB1 magmaswere segregated from a dunite residue and COB2magmas from a harzburgite residue.In the case of batch melting of the same original perido-

tite, the bulk partition coefficient plays an important rolein determining the Ni content of primary magmas,whereas the extent of melting has a minimal effect. A dis-tribution coefficient of Ni between olivine and melt ismuch larger than that between orthopyroxene and melt.Thus, the bulk distribution coefficient of Ni betweendunite and COB1 should be a few per cent to 10% largerthan that between harzburgite and COB2. As a result it isexpected that the primary COB1 magma should contain�10% less Ni than COB2. The observed difference in Nicontents between COB1 and COB2 is much larger than�10% (Fig. 4b), which reflects �15% of olivine fraction-ation from the primary magmas. Therefore the differencein Ni contents between primary COB1 and COB2magmas can be generally explained by the two magmasoriginating from different sources; dunite and harzburgiteresidues, respectively (Fig. 13c,Table 3).

Invisible phase in the mantle wedge;phlogopite?Arc basalts often appear to be highly depleted in HFSE,such as Nb,Ta, Zr and Hf. However, many researchers be-lieve that the marked depletions in these elements maynot reflect real depletions, but are an artifact of trace elem-ent abundance diagrams, and actually reflect marked en-richments of the adjacent trace elements as a result ofsubduction addition. In the normalized abundance dia-gram shown in Fig. 12, which illustrates four types of


96

primary magma in the Mariana Arc, it is worth notingthat NWR1 POB, the driest and least depleted primarymagma type among the four, is not as depleted in Nb, Ta,Zr and Hf as the other three, containing about four timestheir concentration in primitive mantle, comparable withthe HREE. Tamura et al. (2011) suggested that the differ-ences in Nb and Ta contents between NWR1 POB andCOB are much larger (�2) than those expected if only dif-ferent degrees of melting are considered. This is also thecase for comparisons between Pagan COB1 and COB2,and Pagan and NWR1 (Fig. 12). Accordingly, there is ananomalous decrease in Nb and Ta when the fraction ofmelting increases or when the primary magmas becomemore hydrous from POB through COB and COB2 toCOB1. The behavior of Zr and Hf is similar to that of Nband Ta to a lesser extent. Moreover, Zr fractionates fromHf (Fig. 14a) and Nb possibly also fractionates from Ta(Fig. 14b), neither of which can be explained by differencesin the degree of melting. However, Zr/Hf and Nb/Ta cor-relate with the degree of melting, ranging from 18 to 36%(Fig. 12). Nb and Ta abundances in Pagan COB1 andCOB2 are much lower than in primitive mantle, suggest-ing that Nb and Ta are not incompatible during mantlemelting and the production of COB1 and COB2, or thatthe mantle was already depleted in these elements whenthe melting to generate COB began. The latter scenariomight not be possible because the POB source was notdepleted. In the former scenario, however, the possible re-sidual phases in harzburgite or dunite, which are olivine,spinel and orthopyroxene, cannot host Nb and Ta and,moreover, cannot fractionate Zr from Hf, and Nb fromTa.

Hofmann (1988) discussed Nb and Ta anomaliesobserved in average continental crust and MORB. He con-cluded that the anomalies could be explained if Nband Ta have relatively large partition coefficients duringcontinental crust production and smaller coefficientsduring oceanic crust production. Partition coefficients forNb and Ta are apparently higher than unity in someamphiboles and phlogopite; thus, if a partial melt separatesfrom its residue within the stability field of such amphi-boles or phlogopite, the negative Nb^Ta anomaly couldbe produced by magmatic processes (Hofmann, 1988).Ionov & Hofmann (1995) have reported the existence ofNb^Ta-rich mantle amphiboles and micas. They also sug-gested that the amphibole and phlogopite have a strikinglybroad range of Nb/Ta; in particular, Nb/Ta values in themantle phlogopite range from 19 to 54 (Ionov &Hofmann, 1995). Gregoire et al. (2000) studied Type Imantle xenoliths from the Kerguelen Islands and showedthat phlogopite contributes Ba, Nb, Ta and Ti to the bulk-rock, in which Nb and Ta contents range from 38 to98 ppm and from 1·4 to 6·2 ppm, respectively. Phlogopiteis commonly found in harzburgite and spinel lherzoliteand less commonly in plagioclase lherzolite in theHoroman complex (Arai & Takahashi, 1989). The modalabundance of phlogopite in the Horoman peridotites is ex-tremely variable; however, it is not the product of verylow-temperature alteration because it never coexists withserpentine. Phlogopite in the Horoman complex peridotitehas a replacement texture; it mainly replaces olivine andorthopyroxene (Arai & Takahashi, 1989). Moreover, thespinel lherzolite and harzburgite of the Horoman complex

Fig. 14. (a) Zr/Hf vsYb for primary Pagan COB1and COB2, and NW Rota-1COB and POB. Zr fractionates from Hf with increased degree ofmelting. (b) Zr/Hf vs Nb/Ta.


97

are more depleted in HREE compared with the mostprimitive Horoman lherzolite, consistent with different de-grees of partial melting, but are not depleted in Nb(Takazawa et al., 2000). Phlogopite can be produced abovea subducting slab by reactions between a hydrous K-richfluid and peridotite (Sato et al., 1997; Konzett & Ulmer,1999). Such K-rich melts and fluids originate from the melt-ing of phengite (Vielzeuf & Schmidt, 2001; Tamura et al.,2007). Phlogopite could coexist with olivine� orthopyrox-ene and melt at temperatures of 1300^13508C and pres-sures of 2^3GPa in the absence of a gas phase (Yoder &Kushiro, 1969; Sato et al., 1997). Enggist et al. (2012) studiedthe phase relations of phlogopite with magnesite from 4 to8GPa. Below the solidus, phlogopite coexists with magne-site, pyrope and a fluid. At the solidus, magnesite is thefirst phase to react out, and enstatite and olivine appear.Phlogopite melts over a temperature range of �1508C. At13508C and 4GPa, olivine, enstatite, pyrope and melt co-exist with phlogopite (Enggist et al., 2012).When the melt-ing conditions are ‘dry’ as is the case for POB, phlogopite

would not be present and Nb and Ta would becomehighly incompatible. It is possible that the highly depletedresidual mantle wedge of the subduction zone could com-prise phlogopite-bearing dunite and harzburgite.

CONCLUSIONSSubduction zone magmas that erupt at the surface are pro-duced, processed and modified in the so-called ‘subductionfactory’, which consists of the subducting slab, mantlewedge and the overlying arc crust.We have recovered andstudied primitive basalt lavas from Pagan Volcano in theMariana oceanic arc, and our findings have been attribu-ted to each of the three parts of the subduction factory,which we describe as ‘Mission Immiscible’, ‘MissionInvisible’ and ‘Mission Inevitable’, respectively (Fig. 15).Mission Immiscible involves the subduction components,derived from the subducting slab below the volcanic frontat a depth of �100 km, which consist of hydrous carbonatemelt (carbonatite) and silicate melt (sediment melt). The

Fig. 15. Schematic cross-section through the Mariana arc^back-arc basin system, showing upwelling of mantle diapirs. Pagan is on theMariana Arc volcanic front, whereas NW Rota-1 is 40 km behind the volcanic front, resulting in a significant difference between Wadati^Benioff zone depths beneath the volcanoes. Labeled boxes denote locations of different processes described in the text: a, Mission Immiscible;a’, Mission Miscible; b, Mission Invisible; c, Mission Inevitable.


98

expanded miscibility gap between hydrous carbonatite andsediment silicate melt at high pressure results in the coex-istence of two subduction components that are added sep-arately to the magma source mantle diapir, resulting intwo different primary magmas. Mission Invisible involvesthe highly depleted residual mantle wedge of the subduc-tion zone, composed of phlogopite-bearing dunite andharzburgite, which depletes Nb, Ta, Zr and Hf, and frac-tionates Nb from Ta, and Zr from Hf in arc magmas.Mission Inevitable describes the assimilation of crust byarc magmas when they evolve; this is inevitable even inan oceanic arc.

ACKNOWLEDGEMENTSWe thank Jon Davidson and Shoji Arai for careful andinstructive reviews, which improved the paper. S. Turneris thanked for his editorial help.

FUNDINGThis work was supported in part by the JSPS Grant-in-Aidfor Scientific Research (B) (23340166) and Grant-in-Aidfor Creative Scientific Research (19GS0211). Samples werecollected during JAMSTEC cruise NT10-12. Participationof US scientists in this cruise was made possible by NSFgrant 1026150 to R.J.S.

REFERENCESArai, S. (1994). Characterization of spinel peridotite by olivine^spinel

compositional relationships: review and interpretation. Chemical

Geology 113, 191^204.Arai, S. & Takahashi, N. (1989). Formation and compositional vari-

ation of phlogopites in the Horoman peridotite complex,Hokkaido, northern Japan: implications for origin and fraction-ation of metasomatic fluids in the upper mantle. Contributions toMineralogy and Petrology 101, 165^175.

Banks, N. G., Koyanagi, R. Y., Sinton, J. M. & Honma, K. T. (1984).The eruption of Mount Pagan volcano, Mariana islands, 15 May1981. Journal of Volcanology and Geothermal Research 22, 225^269.

Barsdell, M. & Berry, R. F. (1990). Origin and evolution of primitiveisland arc ankaramites from Western Epi, Vanuatu. Journal of

Petrology 31, 747^777.Bernstein, S., Kelemen, P. B. & Hangh�j, K. (2007). Consistent olivine

Mg# in cratonic mantle reflects Archean mantle melting to the ex-haustion of orthopyroxene. Geology 35, 459^462.

Bloomer, S. H., Stern, R. J. & Smoot, N. C. (1989). Physical volcan-ology of the submarine Mariana and Volcano arcs. Bulletin of

Volcanology 51, 210^224.Brooker, R. A. & Kjarsgaard, B. A. (2011). Silicate^carbonate liquid

immiscibility and phase relations in the system SiO2^Na2O^Al2O3^CaO^CO2 at 0·1^2·5 GPa with applications to carbonatitegenesis. Journal of Petrology 52, 1281^1305.

Davidson, J. P. (1987). Crustal contamination versus subduction zoneenrichment: Examples from the Lesser Antilles and implicationsfor mantle source compositions of island arc volcanic rocks.Geochimica et Cosmochimica Acta 51, 2185^2198.

Davidson, J. P. & Harmon, R. S. (1989). Oxygen isotope constraintson the origin of volcanic arc magmas from Martinique, LesserAntilles. Earth and Planetary Science Letters 95, 255^270.

Davidson, J. P. & Wilson, M. (2011). Differentiation and source pro-cesses at Mt Pele¤ e and the Quill; active volcanoes in the LesserAntilles arc. Journal of Petrology 52, 1493^1531.

Dick, H. J. B. & Bullen,T. (1984). Chromian spinel as petrogenetic in-dicator in abyssal and alpine-type peridotites and spatially asso-ciated lavas. Contributions to Mineralogy and Petrology 86, 54^76.

Dixon, T. H. & Stern, R. J. (1983). Petrology, chemistry, and isotopiccomposition of submarine volcanoes in the southern Mariana arc.Geological Society of America Bulletin 94, 1159^1172.

Elliott, T. (2003). Tracers of the slab. In: Eiler, J. (ed.) Inside the

Subduction Factory. Geophysical Monograph, American Geophysical Union

138, 23^45.Elliott, T., Plank, T., Zindler, A., White, W. & Bourdon, B. (1997).

Element transport from slab to volcanic front at the Mariana arc.Journal of Geophysical Research 102, 14991^15019.

Enggist, A., Chu, L. & Luth, R.W. (2012). Phase relations of phlogo-pite with magnesite from 4 to 8 GPa. Contributions to Mineralogy and

Petrology 163, 467^481.Gill, J. B. (1981). Orogenic Andesites and PlateTectonics. Berlin: Springer,

390 p.Govindaraju, K. (1994). 1994 compilation of working values and

sample description for 383 geostandards. Geostandards Newsletter 18,1^158.

Green, D. H., Schmidt, M.W. & Hibberson,W. O. (2004). Island-arcankaramites: primitive melts from fluxed refractory lherzoliticmantle. Journal of Petrology 45, 391^403.

Green, T. H., Green, D. H. & Ringwood, A. E. (1967). The originof high-alumina basalts and their relationships to quartztholeiites and alkali basalts. Earth and Planetary Science Letters 2,41^51.

Gregoire, M., Moine, B. N., O’Reilly, S. Y., Cottin, J. Y. & Giret, A.(2000). Trace element residence and partitioning in mantle xeno-liths metasomatized by highly alkaline, silicate- and carbonate-rich melts (Kerguelen Islands, Indian Ocean). Journal of Petrology41, 477^509.

Gribble, R. F., Stern, R. J., Bloomer, S. H., Stu« ben, D., O’Hearn,T. &Newman, S. (1996). MORB mantle and subduction componentsinteract to generate basalts in the southern MarianaTrough back-arc basin. Geochimica et Cosmochimica Acta 60, 2153^2166.

Hart, S. R. (1984). A large-scale isotope anomaly in the SouthernHemisphere mantle. Nature 309, 753^757.

Hauff, F., Hoernle, K. & Schmidt, A. (2003). Sr^Nd^Pb compositionof Mesozoic Pacific oceanic crust (Site 1149 and 801, ODP Leg185): Implications for alteration of ocean crust and the input intothe Izu^Bonin^Mariana subduction system. Geochemistry,

Geophysics, Geosystems 4, doi:10.1029/2002GC000421.Hofmann, A.W. (1988). Chemical differentiation of the Earth: the re-

lationship between mantle, continental crust, and oceanic crust.Earth and Planetary Science Letters 90, 297^314.

Ionov, D. A. (2010). Petrology of mantle wedge lithosphere: new dataon supra-subduction zone peridotite xenoliths from the andesiticAvacha volcano, Kamchatka. Journal of Petrology 51, 327^361.

Ionov, D. A. & Hofmann, A. W. (1995). Nb^Ta-rich mantle amphi-boles and micas: Implications for subduction-related metasomatictrace element fractionations. Earth and Planetary Science Letters 131,341^356.

Ishii,T., Robinson, P.T., Maekawa, H. & Fiske, R. (1992). Petrologicalstudies of peridotites from diapiric serpentinite seamounts in theIzu^Ogasawara^Mariana forearc, Leg 125. In: Fryer, P., Pearce, J.A. & Stokking, L. B. et al. (eds) Proceedings of the Ocean Drilling


99

Program, Scientific Results, 125. College Station, TX: Ocean DrillingProgram, pp. 445^485.

Ishimaru, S., Arai, S., Ishida, Y., Shirasaka, M. & Okrugin, V. M.(2007). Melting and multi-stage metasomatism in the mantlewedge beneath a frontal arc inferred from highly depleted perido-tite xenoliths from the Avacha volcano, southern Kamchatka.Journal of Petrology 48, 395^433.

Ishizuka, O., Taylor, R. N., Milton, J. A. & Nesbitt, R. W. (2003).Fluid^mantle interaction in an intra-oceanic arc: Constraintsfrom high-precision Pb isotopes. Earth and Planetary Science Letters

211, 221^236, doi:10.1016/S0012-821X(03)00201-2.Ishizuka, O., Geshi, N., Itoh, J., Kawanabe, Y. & TuZino, T. (2008).

The magmatic plumbing of the submarine Hachijo NW volcanicchain, Hachijojima, Japan: Long-distance magma transport?Journal of Geophysical Research 113, doi:10.1029/2007JB005325.

Johnson, C. M. & Beard, B. L. (1999). Correction of instrumentallyproduced mass fractionation during isotopic analysis of Fe by ther-mal ionization mass spectrometry. International Journal of Mass

Spectrometry 193, 87^99.Kawamoto, T., Kanzaki, M., Mibe, K., Matsukage, K. N. & Ono, S.

(2012). Separation of supercritical slab-fluids to form aqueous fluidand melt components in subduction zone magmatism. Proceedingsof the National Academy of Sciences, USA 109, 18695^18700.

Kelley, K. A. & Cottrell, E. (2009). Water and the oxidation state ofsubduction zone magmas. Science 325, 605^607.

Kelley, K. A., Plank, T., Newman, S., Stolper, E. M., Grove, T. L.,Parman, S. & Hauri, E. H. (2010). Mantle melting as a function ofwater content beneath the Mariana arc. Journal of Petrology 51,1711^1738.

Konzett, J. & Ulmer, P. (1999). The stability of hydrous potassicphases in lherzolitic mantleçan experimental study to 9·5 GPa insimplified and natural bulk compositions. Journal of Petrology 40,629^652.

Kubo, K. (2002). Dunite formation processes in highly depleted peri-dotite: case study of the Iwanaidake peridotite, Hokkaido, Japan.Journal of Petrology 43, 423^448.

Kushiro, I. (2001). Partial melting experiments on peridotite andorigin of mid-ocean ridge basalt. Annual Review of Earth and

Planetary Sciences 29, 71^107.Li, C. & Ripley, E. M. (2010). The relative effects of composition and

temperature on olivine^liquid Ni partitioning: Statistical deconvo-lution and implications for petrologic modeling. Chemical Geology275, 99^104.

Marske, J. P., Pietruszka, A. J., Trusdell, F. A. & Garcia, M. O. (2011).Geochemistry of southern Pagan Island lavas, Mariana arc: therole of subduction zone processes. Contributions to Mineralogy and

Petrology 162, 231^252.Mattey, D. P., Taylor, W. R., Green, D. H. & Pillinger, C. T. (1990).

Carbon isotopic fractionation between CO2 vapour, silicate andcarbonate melts: an experimental study to 30 kbar. Contributions toMineralogy and Petrology 104, 492^505.

McDonough,W. F. & Frey, F. A. (1989). REE in upper mantle rocks.In: Lipin, B. & McKay, G. R. (eds) Geochemistry and Mineralogy of

Rare Earth Elements. Mineralogical Society of America, Reviews in

Mineralogy 21, 99^145.McDonough,W. F. & Sun, S.-s. (1995). The composition of the Earth.

Chemical Geology 120, 223^253.Mibe, K., Kawamoto, T., Matsukage, K. N., Fei, Y. & Ono, S. (2011).

Slab melting versus slab dehydration in subduction-zone magma-tism. Proceedings of the National Academy of Sciences, USA 108,8177^8182.

Mu« nker, C.,Weyer, S., Scherer, E. & Mezger, K. (2001). Separation ofhigh field strength elements (Nb, Ta, Ar, Hf) and Lu from rock

samples form MC-ICPMS measurements. Geochemistry, Geophysics,Geosystems 2, doi:10.1029/2001GC000183.

Neuendorf, K. K. E., Mehl, J. P., Jr & Jackson, J. A. (2006). Glossary ofGeology, 5th edn. Alexandria,VA: American Geological Institute.

Pearce, J. A., Stern, R. J., Bloomer, S. H. & Fryer, P. (2005).Geochemical mapping of the Mariana arc^basin system: implica-tions for the nature and distribution of subduction components.Geochemistry, Geophysics, Geosystems 6, doi:10.1029/2004GC000895.

Plank,T. & Langmuir, C. H. (1998).The chemical composition of sub-ducting sediment and its consequences for the crust and mantle.Chemical Geology 145, 325^394.

Pozgay, S. H., Wiens, D. A., Conder, J. A., Shiobara, H. &Sugioka, H. (2009). Seismic attenuation tomography of theMariana subduction system: Implications for thermal structure,volatile distribution, and slow spreading dynamics. Geochemistry,

Geophysics, Geosystems 10, doi:10.1029/2008/GC002313.Sato, K., Katsura, T. & Ito, E. (1997). Phase relations of natural

phlogopite with and without enstatite up to 8 GPa: implication formantle metasomatism. Earth and Planetary Science Letters 146,511^526.

Seno,T., Stein, S. & Gripp, A. E. (1993). A model for the motion of thePhilippine Sea Plate consistent with NUVEL-1 and geologicaldata. Journal of Geophysical Research 98, 17941^17948.

Stern, R. J. (2010). The anatomy and ontogeny of modern intra-ocea-nic arc systems. In: Kusky, T. M., Zhai, M.-G. & Xiao, W. (eds)The Evolving Continents: Understanding Processes of Continental Growth.

Geological Society, London, Special Publications 338, 7^34.Stern, R. J., Bloomer, S. H., Lin, P-N., Ito, E. & Morris, J. (1988).

Shoshonitic magmas in nascent arcs: New evidence from submar-ine volcanoes in the northern Marianas. Geology 16, 426^430.

Stern, R. J., Fouch, M. J. & Klemperer, S. L. (2003). An overview ofthe Izu^Bonin^Mariana subduction factory. In: Eiler, J. (ed.)Inside the Subduction Factory. Geophysical Monograph, American

Geophysical Union 138, 175^222.Stern, R. J., Ed, Kohut, Bloomer, S. H., Leybourne, M., Fouch, M. &

Vervoort, J. (2006). Subduction factory processes beneath theGuguan cross-chain, Mariana Arc: no role for sediments, are ser-pentinites important? Contributions to Mineralogy and Petrology 151,202^221.

Stern, R. J., Tamura, Y., Masuda, H., Fryer, P., Martinez, F.,Ishizuka, O. & Bloomer, S. H. (2013). How does the Mariana Arcend in the south? Island Arc 22, 133^148.

Suhr, G., Hellebrand, H., Snow, J. E., Seck, H. A. & Hofmann, A.W.(2003). Significance of large, refractory dunite bodies in the uppermantle of the Bay of Islands Ophiolite. Geochemistry, Geophysics,

Geosystems 4, doi:10.1029/2001GC000277.Sun, S.-s. & McDonough,W. F. (1989). Chemical and isotopic system-

atics of oceanic basalts: implications for mantle composition andprocesses. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in

the Ocean Basins. Geological Society, London, Special Publications 42,313^345.

Takazawa, E., Frey, F. A., Shimizu, N. & Obata, M. (2000).Whole rock compositional variations in an upper mantleperidotite (Horoman, Hokkaido, Japan): Are they consistent witha partial melting process? Geochimica et Cosmochimica Acta 64,695^716.

Tamura, A. & Arai, S. (2005). Unmixed spinel in chromitite from theIwanai-dake peridotite complex, Hokkaido, Japan: A reaction be-tween peridotite and highly oxidized magma in the mantle wedge.American Mineralogist 90, 473^480.

Tamura, Y. & Tatsumi, Y. (2002). Remelting of an andesitic crust as apossible origin for rhyolitic magma in oceanic arc: an examplefrom the Izu^Bonin arc. Journal of Petrology 43, 1029^1047.


100

Tamura, Y., Yuhara, M. & Ishii, T. (2000). Primary arc basalts fromDaisen volcano, Japan: equilibrium crystal fractionation versus dis-equilibrium fractionation during supercooling. Journal of Petrology41, 431^448.

Tamura, Y., Tani, K., Chang, Q., Shukuno, H., Kawabata, H.,Ishizuka, O. & Fiske, R. S. (2007).Wet and dry basalt magma evo-lution at Torishima volcano, Izu^Bonin arc, Japan: the possiblerole of phengite in the downgoing slab. Journal of Petrology 48,1999^2031.

Tamura, Y., Ishizuka, O., Aoike, K., Kawate, S., Kawabata, H.,Chang, Q., Saito, S., Tatsumi, Y., Arima, M., Takahashi, M.,Kanamaru, T., Kodaira, S. & Fiske, R. S. (2010). MissingOligocene crust of the Izu^Bonin arc: consumed or rejuvenatedduring collision. Journal of Petrology 51, 823^846.

Tamura, Y., Ishizuka, O., Stern, R. J., Shukuno, H., Kawabata, H.,Embley, R. W., Hirahara, Y., Chang, Q., Kimura, J.-I.,Tatsumi, Y., Nunokawa, A. & Bloomer, S. H. (2011). Two primarybasalt magma types from Northwest Rota-1 volcano, Mariana arcand its mantle diapir or mantle wedge plume. Journal of Petrology52, 1143^1183.

Tanaka,T.,Togashi, S., Kamioka, H. et al. (2000). JNdi-1: A neodymiumisotopic reference in consistency with LaJolla neodymium. ChemicalGeology168,279^281, doi:10.1016/S0009-2541(00)00198-4.

Taylor, R. N. & Nesbitt, R. W. (1998). Isotopic characteristics of sub-duction fluids in an intra-oceanic setting, Izu^Bonin arc, Japan.Earth and Planetary Science Letters 164, 79^98.

Thirlwall, M. F. & Graham, A. M. (1984). Evolution of high-Ca, high-Sr C-series basalts from Grenada, Lesser Antilles: Contaminationin the arc crust. Journal of the Geological Society, London 141, 427^445.

Turner, S., Bourdon, B. & Gill, J. (2003). Insights into magma genesisat convergent margins from U-series isotopes. In: Bourdon, B.,

Henderson, G. M., Lundstrom, C. C. & Turner, S. P. (eds) U-seriesGeochemistry. Mineralogical Society of America and Geochemical Society,

Reviews in Mineralogy and Geochemistry 52, 255^315.Vielzeuf, D. & Schmidt, M. W. (2001). Melting relations in hydrous

systems revisited: application to metapelites, metagreywackes andmetabasalts. Contributions to Mineralogy and Petrology 141, 251^267.

Wade, J. A., Plank, T., Stern, R. J., Tollstrup, D. L., Gill, J. B.,O’Leary, J. C., Eiler, J. M., Moore, R. B., Woodhead, J. D.,Trusdell, F., Fischer, T. P. & Hilton, D. R. (2005). The May 2003eruption of Anatahan volcano, Mariana Islands: Geochemical evo-lution of a silicic island-arc volcano. Journal of Volcanology and

Geothermal Research 146, 139^170.Walker, D., Shibata,T. & DeLong, S. E. (1979). Abyssal tholeiites from

the Oceanographer Fracture Zone III. Phase equilibria and mixing.

Contributions to Mineralogy and Petrology 70, 111^125.Woodhead, J. D. (1989). Geochemistry of the Mariana arc (western

Pacific): Source composition and processes. Chemical Geology 76, 1^24.Woodhead, J. D., Hergt, J. M., Davidson, J. P. & Eggins, S. M. (2001).

Hafnium isotope evidence for ‘conservative’ element mobilityduring subduction zone processes. Earth and Planetary Science Letters192, 331^346.

Woodhead, J., Stern, R. J., Pearce, J., Hergt, J. & Vervoort, J. (2012).Hf^Nd isotope variation in Mariana Trough basalts: The import-ance of ‘ambient mantle’ in the interpretation of subduction zonemagmas. Geology 40, 539^542.

Yaxley, G. M., Green, D. H. & Kamenetsky, V. (1998). Carbonatitemetasomatism in the southeastern Australian lithosphere. Journalof Petrology 39, 1917^1930.

Yoder, H. S. & Kushiro, I. (1969). Melting of a hydrous phase: phlogo-pite. AmericanJournal of Science 267, 558^582.


101

mission immiscible: distinct subduction components ...rjstern/pdfs/tamura... · mission immiscible:...

Documents