Origin of Silicic Magmas at Spreading

Centres—an Example from the South East Rift,

Manus Basin

Christoph Beier1*, Wolfgang Bach2, Simon Turner3, Dominik

Niedermeier2, Jon Woodhead4, Jorg Erzinger5 and Stefan Krumm1

1GeoZentrum Nordbayern, Friedrich-Alexander-Universitat Erlangen-Nurnberg, Erlangen, Germany, 2Institut fur

Geoscience Department and MARUM, University of Bremen, Germany, 3Department of Earth and Planetary Sciences,

Macquarie University, Sydney, NSW 2109, Australia, 4School of Earth Sciences, University of Melbourne, VIC 3010,

Australia and 5Helmholtz-Zentrum Potsdam Deutsches GeoForschungsZentrum (GFZ), Telegrafenberg, 14473 Potsdam,

Germany

*Corresponding author. Telephone: þ49/9131/85-26064. E-mail: christoph.beier@fau.de

Received July 27, 2014; Accepted December 22, 2014

ABSTRACT

There has been much recent interest in the origin of silicic magmas at spreading centres away from

any possible influence of continental crust. Here we present major and trace element data for 29

glasses (and 55 whole-rocks) sampled from a 40 km segment of the South East Rift in the Manus

Basin that span the full compositional continuum from basalt to rhyolite (50–75 wt % SiO2). The glass

data are accompanied by Sr–Nd–Pb, O and U–Th–Ra isotope data for selected samples. These overlap

the ranges for published data from this part of the Manus Basin. Limited increases in Cl/K ratios with

increasing SiO2, La–SiO2 and Yb–SiO2 relationships, and the oxygen isotope data rule out models inwhich the more silicic lavas result from partial melting of altered oceanic crust or altered oceanic gab-

bros. Rather, the data form a coherent array that is suggestive of closed-system fractional crystalliza-

tion and this is well simulated by MELTS models run at 0�2 GPa and QFM (quartz–fayalite–magnetite

buffer) with 1 wt % H2O, using a parental magma chosen from the basaltic glasses. Although some as-

similation of altered oceanic crust or gabbro cannot be completely ruled out, there is no evidence that

this plays an important role in the origin of the silicic lavas. The U-series disequilibria are dominatedby 238U and 226Ra excesses that limit the timescale of differentiation to less than a few millennia.

Overall, the data point to rapid evolution in relatively small magma lenses located near the base of

thick oceanic crust; we speculate that this was coupled with relatively low rates of basaltic recharge.

A similar model may be applicable to the generation of silicic magmas elsewhere in the ocean basins.

Key words: silicic lavas; differentiation; oceanic spreading centre; South East Rift; Manus Basin

INTRODUCTION

The origin of silicic magma is of interest in any tectonic

setting but such compositions are rare at spreading

centres and so their evolution here is particularly enig-

matic. Mid-ocean ridges are dominated by basalts

(MORB) and proposed petrogenetic models for the oc-casional, more evolved magmas include fractional crys-

tallization (e.g. Juster et al., 1989), partial melting of

altered oceanic crust (e.g. Coogan et al., 2003) or altered

gabbro (e.g. Koepke et al., 2007), and combined

assimilation and fractional crystallization scenarios that

incorporate a combination of these processes and com-ponents (Le Roux et al., 2006). Significant changes in Cl/

K2O and d18O ratios have often been regarded as diag-

nostic tracers in such studies. For example, a recent

study by Wanless et al. (2010) investigated the origins

of dacites at the Juan de Fuca Ridge, eastern Galapagos

spreading centre and the East Pacific Rise. Theseauthors used increases in Cl/K2O ratios in conjunction

with other geochemical modelling to conclude that

VC The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com 1

J O U R N A L O F

P E T R O L O G Y

Journal of Petrology, 2015, 1–17

doi: 10.1093/petrology/egu077

Article

Journal of Petrology Advance Access published February 13, 2015

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

assimilation of partial melts of altered oceanic crust

was crucial to facilitating silica enrichment in magmas

at these localities.

Back-arc basin basalts (BABB) dominate back-arc

spreading centres, although the occurrence of andesiticrocks is not uncommon (e.g. Kent et al., 2002; Sinton

et al., 2003). This might be perceived as being in keep-

ing with this tectonic setting, yet more evolved dacites

and rhyolites are equally rare at these spreading centres

(e.g. Pearce et al., 1994; Stolper & Newman, 1994;

Fretzdorff et al., 2002; Caulfield et al., 2012) and their ori-

gins have not been studied in great detail. Here we pre-sent major and trace element and isotopic data for a

suite of rocks sampled from a restricted segment (40

km in length) along the South East Rift in the Manus

Basin, Papua New Guinea, that encompass the entire

compositional continuum from basalt through andesite

to dacite and rhyolite. Our interest in this study lies pri-marily in the causes of melt composition evolution ra-

ther than source region characteristics and melting

dynamics. The latter have recently been explored in de-

tail by Sinton et al. (2003) and Beier et al. (2010), for the

Manus Basin in general, and by Park et al. (2010) for the

South East Rift in particular.

GEOLOGICAL SETTING AND SAMPLE DETAILS

The Manus Basin (Fig. 1) is a rapidly opening (�10 cm

a�1), magmatically active back-arc basin associated

with northward subduction of the Solomon Sea Plate(�15�4 cm a�1) at the New Britain Trench (Fig. 1;

Tregoning, 2002; Lee & Ruellan, 2006). The region has

experienced a complex tectonic history, commencing

with southward subduction of the Pacific Plate along

the Manus Trench until the late Miocene (Martinez &

Taylor, 1996). At this time, the West Melanesian palaeo-

island arc formed the islands of New Britain, NewIreland and the Huon Peninsula of mainland Papua New

Guinea (Johnson et al., 1979).

In Miocene times, subduction along the Manus

Trench stopped as a result of the collision of the

Ontong–Java Plateau with the Manus Trench (Martinez

& Taylor, 1996). This led to a reversal of subduction po-larity and northward subduction of the Solomon Sea

Plate along the New Britain Trench. Back-arc opening in

the Bismarck Sea started at around 3�5–4 Ma, when the

Huon Peninsula and New Britain rotated after colliding

with New Guinea (Taylor, 1979). The eastward propa-

gating arc–continent collision of the Huon–New Britain

block with New Guinea led to the formation of threemajor left-lateral transform faults, framing a clockwise

rotating Manus microplate between the NW–SE-trend-

ing Willaumez, Djaul and Weitin transform faults

(Taylor et al., 1994; Martinez & Taylor, 1996; Wallace

et al., 2005). At the northern end of the Manus micro-

plate, seafloor spreading started at <0�78 Ma along theExtensional Transform Zone (ETZ) and the Manus

Spreading Centre (MSC) with maximum spreading

rates of 92 mm a�1 (Taylor et al., 1994; Martinez &

Taylor, 1996; Lee & Ruellan, 2006).

The South Rift (SR) and South East Rift (SER) in the

Eastern Manus Basin (EMB) mark the southern border

of the Manus microplate (Fig. 1). The SER is character-ized by pull-apart tectonics between the Djaul and

Weitin transforms. Here, we focus on the neovolcanic

centers along the SER, along with previously published

data (Sinton et al., 2003); these range from sinusoidal,

commonly forked ridges to solitary volcanoes or short

chains of volcanoes. The ridges spread en echelon on

tectonically stretched arc crust between the Djaul andWeitin transform faults. Water depths along the SER are

generally shallower than 2000 m and these are given in

Supplementary Data Table S1 (supplementary data are

available for downloading at http://www.petrology.

oxfordjournals.org).

The samples analysed were collected during R.V.Melville cruise MAGELLAN-06 in 2006 using the re-

motely operated vehicle Jason-2 (Woods Hole

Oceanographic Institution). Sampling by Jason-2 was

accompanied by detailed seafloor mapping by autono-

mous benthic explorer. Samples were collected in vari-

ous areas of the central and eastern Manus Basin,including (from west to east) Pual Ridge, Desmos,

Umbo, and SuSu Knolls (Fig. 1). In addition, we provide

data for some new samples from the Manus Spreading

Centre (Vienna Woods, Fig. 1), although these will not

be discussed in the text.

WHOLE-ROCK ANALYTICAL TECHNIQUES

A total of 55 whole-rock samples were selected for

major and trace element analysis; only fresh volcanic

samples with no visible signs of alteration were used

(Supplementary Data Table S1). For whole-rock ana-

lyses any potentially altered rims were removed prior to

sample preparation. Many samples from the SuSuKnolls area contain sulfide veins along fine cracks

which are commonly attributed to incipient hydrother-

mal alteration processes and thus samples with ele-

vated sulfur contents were not included in this study.

The samples were crushed in a steel-beaker anvil and

powdered in an agate planetary ball mill.For the mineral analyses provided in Supplementary

Data Table S2, major elements (SiO2, Al2O3, TiO2, MgO,

FeO*, MnO, CaO, K2O, Na2O) and Ni and Cr in olivine

xenocrysts and their spinel inclusions were measured

using a JEOL JXA-8900 electron microprobe at the

Institut fur Geowissenschaften, Universitat Kiel. The in-

strument was operated with a fully focused electronbeam at 15 kV acceleration voltage and a beam current

of 12 nA. Wollastonite, anorthite rutile, forsterite, faya-

lite, rhodonite, microcline, apatite, NiO and chromite

were used as calibration standards.

Whole-rock major and trace elements (Cr, Ga, Ni, Sc,

V, Zn) were determined by X-ray fluorescence (XRF)spectrometry using a PANalyticalVR AXIOS Advanced

system at the Helmholtz-Zentrum Potsdam Deutsches

2 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

GeoForschungsZentrum, Potsdam. The total iron con-

tent was measured by XRF as Fe2O3 and for selected

representative samples from each locality the concen-

trations of ferrous iron was determined by titration

using K2Cr2O7 (Wilson, 1960). Sulfur contents were

determined on dried powders with a carbon sulfurLECO CS-225 instrument.

Trace element concentrations were determined by

inductively coupled plasma mass spectrometry

(ICP-MS) on a ThermoFinnigan Element2 at the Geo

ForschungsZentrum, Potsdam and at the Institute for

Geosciences, University of Bremen, respectively.

Samples analysed in Bremen were digested using anacid digestion technique and an MLS Ethos microwave

heating system. The sample powder (�40 mg) was

mixed with a mixture of sub-boiled acids (9 ml HNO3

65%, 2 ml HCl 32%, 3 ml HF 48%) in a Teflon beaker for

the acid digestion. Final solutions were spiked with

1 ng ml�1 indium as an internal standard. As an externalreference material the USGS standard BCR-2 was dis-

solved and measured together with the sample set. The

instrument in Bremen was calibrated using a mixture of

pure element standards in various dilutions. In Bremen

all trace elements and La, Ce, Pr and Nd were measured

in low-resolution mode; all other rare earth elements

(REE) were measured in high-resolution mode. At the

GeoForschungsZentrum, Potsdam 250 mg of sample

powder was dissolved in 4 ml concentrated HF and 4 ml

HCl–HNO3 (3:1) using SavillexVR vials. The closed

SavillexVR beakers were stored for 14 h at 160�C in aheating block. After cooling, 1 ml HClO4 was added and

the solution was dried at 180�C. Then 2 ml H2O and 1 ml

concentrated HNO3 were added to the residue and the

solution was dried again; then 1 ml concentrated HNO3

and 5 ml H2O were added and the SavillexVR vials were

closed and stored at <100�C overnight. Finally the sam-

ple solutions were decanted into volumetric flasks andfilled up to 50 ml with Millipore H2O for the ICP-MS ana-

lyses. At the GeoForschungsZentrum, Potsdam the bas-

alt standard JB-3 was used as external reference

material.

Results for international rock standards at both labo-

ratories were within error of the preferred values(Supplementary Data Table S3). Duplicate samples dis-

solved and measured in Bremen and Potsdam are in

agreement for all elements. The trace element concen-

trations agree within error and allow the datasets to be

merged.

148˚E

148˚E

149˚E

149˚E

150˚E

150˚E

151˚E

151˚E

152˚E

152˚E

153˚E

153˚E

154˚E

154˚E

6˚S 6˚S

5˚S 5˚S

4˚S 4˚S

3˚S 3˚S

−9000 −8000 −7000 −6000 −5000 −4000 −3000 −2000 −1000 0 1000 2000

Depth

m

New Britain trench

New Britain arc−front

Solomon SeaHuon Peninsula

New IrelandWeitin Transform

Manus trenchTabar−Lihir−Tanga−Feni arc

St. And

rew S

trait

Witu Islands

MANUS BASIN

South East RiftSouth Rift

ETZWillaumez Transform Djaul Transform

600 km

500 km400 km

300 km

200 km

100 km

50 km

MSC

Pual Ridge DesmosUmbo

SuSu Knolls

Vienna Woods

Manus Basin published

MSC this study

SER Pual Ridge

SER Desmos

SER SuSu Knolls

SER Umbo

Fig. 1. Map of the Manus Basin using GMT (Wessel & Smith, 1991, 1998) (modified after Martinez & Taylor, 2003; Pearce & Stern,2006). Major tectonics and the depths of Benioff zones (dashed lines) are from Cooper & Taylor (1987). ETZ, Eastern TransformZone; MSC, Manus Spreading Center; SER, South East Rift. Grey labels mark major structures in the Manus Basin. Localities of pub-lished MORB and BABB data are from Sinton et al. (2003).

Journal of Petrology, 2015, Vol. 0, No. 0 3

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

GLASS ANALYTICAL TECHNIQUES

Glassy fragments or pillow-rim glasses were available

for 35 samples and were handpicked under a binocular

microscope to select the freshest material. The glasseswere analysed for major and trace elements and a sub-

set was also analysed for Sr–Nd–Pb, O and U–Th–Ra

isotopes (Tables 1 and 2). Chips handpicked from the

crushed glasses were mounted in epoxy and diamond

polished for microprobe analysis. Major elements

(SiO2, Al2O3, TiO2, MgO, FeO*, MnO, CaO, K2O, Na2O)

and Cl and S in volcanic glasses were measured using aJEOL JXA-8900 electron microprobe at the Institut fur

Geowissenschaften, Universitat Kiel. Glasses were ana-

lysed with a defocused beam diameter of 10 mm at 15

kV acceleration voltage and a beam current of 12 nA.

Trace elements were analysed by laser ablation (LA)-

ICP-MS at the University of Bremen using a NewWaveUP 193 solid-state laser combined with a Thermo

Finnigan Element2 ICP-MS system. Laser spot sizes

were adjusted between 50 and 100mm, adapted to the

available amount of glassy material. NIST 610, NIST

612 and BCR-2 glasses were used as reference mater-

ials. The trace element concentrations were calculated

with the GeoProVR software, using the Si contents deter-mined by electron microprobe as an internal standard.

Fe2þ was determined on sample powders prepared

from handpicked glass chips by titration at the GFZ

Potsdam (Wilson, 1960).

Strontium and Nd isotopes were analysed atMacquarie University, Sydney using �100 mg hand-picked glass chips that were leached with a 1þ 1 mix-ture of 2�5N HCl and 30% H2O2 for 10 min. Chemicalpurification followed the methods described byHeyworth et al. (2007) and Le Roux et al. (2009), and iso-tope ratios were obtained by thermal ionization massspectrometry (TIMS) using a ThermoFisher Triton sys-tem in static mode. Instrumental bias was correctedassuming 86Sr/88Sr¼ 0�1194 and 146Nd/144Nd¼0�7219.Analyses of NIST SRM-987 gave 0�710210 6 37 (2SD,n¼ 14) and BHVO2 gave 0�703490 6 52 (2 SD, n¼ 15),whereas the JMC Nd standard gave 0�511106 6 2 (2SD,n¼ 8) and BHVO-2 yielded 0�512967 6 6 (2SD, n¼ 17),respectively. To be able to compare the new data withthose published by Sinton et al. (2003) and Beier et al.(2010), the new data were normalized to 0�710250 for Srand all Nd isotope data were normalized to 0�511489.

Lead isotope ratios were determined at MelbourneUniversity employing the methods described in detail

by Woodhead (2002). Hand-picked glass chips were

leached for 30 min with hot 6N HCl, then washed re-

peatedly in ultrapure water, before digestion in Savillex

beakers with HF þ HNO3 on a hotplate. Lead was then

extracted using conventional anion exchange methods

in HBr–HCl media. Total procedural blanks are less than20 pg in all cases and are considered negligible. Isotope

ratios were measured by multi-collector (MC)-ICP-MS

using a Nu Instruments system coupled to a CETAC

Aridus desolvating nebulizer operating at a sample up-

take rate of �30mm min�1. Instrumental mass bias was

corrected using an admixed Tl solution and all analyses

were normalized to the Woodhead et al. (1995) values

for the NIST SRM 981 reference material. A highly

diluted Broken Hill galena solution was employed as a

secondary reference material and provided the follow-ing results over the period encompassing this study

(mean, 62SD): 206Pb/204Pb¼ 16�004 6 0�006, 207Pb/204Pb¼15�388 6 0�009, 208Pb/204Pb¼35�660 6 0�028, n¼ 66.

These results compare well with those obtained by both

TIMS and MC-ICP-MS in other laboratories.

d18O values were determined on cleaned glass chips

by laser fluorination using a 25 W Synrad CO2-laser and

F2 as reagent at GeoZentrum Nordbayern (Table 2). The

laser was operated in continuous mode and the energy

was manually adjusted to allow a reaction process that

was as smooth as possible and to avoid any sputtering.

The general setup follows that of Sharp (1990). Samples

were prefluorinated overnight at 0�02 bar F2. F2 pressure

during fluorination was 0�1 bar. The released oxygen

was cleaned using one heated NaCl and two LN2 traps.

The isotopic composition of oxygen was analysed on a

ThermoFisher Delta Plus mass spectrometer. During

each measurement day, four standard samples (UWG-

2, NBS-30) were measured together with the unknowns.

The d18O raw values were corrected by the mean differ-

ence of the reference values from the standards (5�8and 5�1%, respectively). All oxygen isotope values are

given in per mil relative to V-SMOW. Reproducibility of

10 replicate samples measured in the same time period,

which would include analytical error as well as sample

heterogeneity and impurity, varied between 0�02 and

0�07% (1r; mean 0�06% 1r). Long-term reproducibility

of the UWG-2 garnet standard obtained during this

study was 5�84 6 0�07% (1r, n¼ 29). The mean 1r of

our glass replicates varies between 0�06 and 0�28; the

mean 2r is 0�28. Thus the lowest and highest d18O val-

ues of our samples do not overlap within their errors.

Uranium, Th and Ra concentrations and isotope

ratios were determined at Macquarie University at thesame time as, and following the same methods as for

other Manus Basin samples analysed for U-series iso-

topes by Beier et al. (2010). Suitable glasses were hand-

picked under a binocular microscope to avoid pieces

with visible evidence for crystals, vesicles, hydrother-

mal alteration and/or Mn-coatings. Selected chips were

then ultrasonicated in cold 2�5N HCl and 30% H2O2 (1:1mix) for 10–20 min before being washed and ultrasoni-

cated (20 min) in deionized water. These were subse-

quently spiked with 226U–229Th and 228Ra tracers and

dissolved using an HF–HNO3–HCl mix in heated Teflon

pressure bombs. The product was converted to chloride

using 6N HCl and then 6N HCl saturated with H3BO3 todrive off residual fluorides. The final product was then

converted to nitrate using 14N HNO3 and finally taken

up in 7N HNO3. Uranium and Th purification was

achieved via a single pass through a 4 ml anionic resin

column using 7N HNO3, 6N HCl and 0�2N HNO3 as elu-

tants. We purposefully avoided the use of ElChromVR

resins for the U–Th chemistry as these bleed organic

4 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

Table 1: Major (wt %) and trace element (mg g–1) analyses of South East Rift glasses plus six new samples from the ManusSpreading Centre

Sample no. Locality Longitude(�W)

Latitude(�N)

Waterdepth (m)

SiO2 TiO2 Al2O3 FeOtot FeO Fe2O3 MnO MgO CaO Na2O K2O P2O5 S Cl Total

J2-208-5-R1 Pual 151�6741 –3�7214 1670�00 71�39 0�53 13�37 4�03 3�36 0�74 0�11 0�59 2�21 4�56 2�17 0�11 36 4367 99�52J2-209-9-R1 Pual 151�6744 –3�7212 1689�00 64�37 0�83 15�27 5�75 4�23 1�69 0�18 1�77 4�48 4�55 1�39 0�32 98 3063 99�94J2-211-2-R2 Pual 151�6854 –3�7136 1660�00 65�05 0�88 14�92 5�81 4�45 1�51 0�14 1�49 4�11 4�32 1�60 0�34 28 3368 99�80J2-211-3-R1 Pual 151�6724 –3�7220 1626�00 71�79 0�65 13�48 4�09 3�37 0�80 0�14 0�49 2�23 4�45 2�16 0�17 8 4590 100�51J2-211-8-R1 Pual 151�6714 –3�7238 1649�00 71�33 0�67 13�75 3�86 3�39 0�53 0�14 0�45 2�35 4�65 2�16 0�14 24 4481 99�70J2-213-4-R1 Pual 151�6745 –3�7212 1683�00 73�29 0�71 13�11 3�76 3�25 0�57 0�13 0�34 1�83 4�44 2�35 0�16 32 4853 99�22J2-214-8-R1 Pual 151�6718 –3�7274 1660�00 67�94 0�65 14�67 4�86 3�44 1�58 0�15 1�43 3�48 4�52 1�74 0�16 40 3380 99�73J2-214-9-R1 Pual 151�6689 –3�7283 1660�00 67�70 0�64 14�58 4�95 3�58 1�52 0�16 1�49 3�50 4�45 1�72 0�17 24 3370 99�38J2-215-2-R1 Pual 151�6691 –3�7294 1663�00 70�15 0�65 13�37 4�63 3�24 1�54 0�19 1�02 2�52 4�48 2�09 0�21 20 4250 98�99J2-216-11-R1 Pual 151�6692 –3�7289 1724�00 67�24 0�63 14�51 4�87 3�23 1�82 0�15 1�51 3�59 4�68 1�67 0�18 34 3460 98�52J2-218-4-R1 Pual 151�6665 –3�7298 1875�00 67�41 0�66 14�63 4�98 3�44 1�71 0�15 1�64 3�60 4�50 1�75 0�16 28 3310 99�04J2-218-5-R1 Pual 151�6662 –3�7279 1875�00 72�47 0�65 13�44 3�86 3�09 0�86 0�12 0�46 2�14 4�56 2�23 0�14 12 4473 98�75J2-218-7-R1 Pual 151�6724 –3�7287 1868�00 64�14 0�92 14�88 5�90 4�03 2�08 0�16 1�61 4�30 4�44 1�53 0�31 49 3166 99�29J2-218-9-R1 Pual 151�7275 –3�6751 1915�00 62�95 1�05 14�45 7�78 5�43 2�61 0�19 1�62 4�60 4�12 1�47 0�51 48 3180 99�69J2-220-10-R1 Desmos 151�7274 –3�6751 1914�00 61�66 0�95 15�26 7�39 5�37 2�25 0�16 1�92 5�10 4�25 1�26 0�50 144 2864 97�83J2-220-14-R1 Desmos 151�7286 –3�6753 1903�00 64�27 0�94 14�85 6�71 5�13 1�76 0�19 1�43 4�13 4�37 1�58 0�47 60 3450 97�76J2-220-14-R2 Desmos 151�7332 –3�6673 1903�00 69�27 0�87 13�32 5�44 4�38 1�18 0�15 0�73 2�66 4�38 2�13 0�29 36 4429 98�71J2-220-2-R1 Desmos 151�8794 –3�6913 1995�00 55�98 0�60 15�50 8�01 5�49 2�80 0�14 4�74 8�55 2�94 0�97 0�19 68 2100 97�92J2-222-11-R1 Pual 151�8787 –3�6920 1671�00 55�42 0�61 15�84 8�04 5�35 2�99 0�14 5�47 9�05 2�85 0�89 0�19 40 2010 100�11J2-222-12-R1 Pual 151�8825 –3�6945 1651�00 55�11 0�55 15�82 7�73 5�60 2�37 0�15 5�44 9�10 2�79 0�86 0�18 48 1940 99�95J2-222-9-R2 Pual 151�8787 –3�6920 1678�00 54�94 0�59 15�68 7�77 5�68 2�32 0�15 5�36 9�08 2�93 0�87 0�18 66 2012 100�61J2-223-10-R1 SuSu 152�1014 –3�7992 1198�00 74�57 0�50 12�18 3�57 0�13 0�39 1�83 4�09 2�05 0�14 12 2722 96�16J2-223-5-R1 SuSu 152�1017 –3�8002 1198�00 74�57 0�45 12�55 3�67 3�66 0�02 0�12 0�40 1�99 4�17 2�09 0�13 36 2742 100�42J2-226-8-R2 SuSu 152�1047 –3�8104 1591�00 74�78 0�41 12�59 2�37 0�10 0�12 1�29 4�22 2�30 0�15 48 4030 99�72J2-227-1-R1 SuSu 152�0917 –3�7918 1549�00 68�45 0�86 14�11 5�49 4�39 1�22 0�13 0�63 3�07 4�25 2�05 0�36 44 3260 98�56J2-227-2-R2 SuSu 152�0991 –3�8152 1642�00 61�39 0�77 15�26 7�25 4�83 2�69 0�17 2�47 5�76 3�97 1�11 0�23 76 1760 99�13J2-227-4-R1 SuSu 152�0959 –3�8120 1365�00 67�17 0�81 14�31 5�36 4�79 0�63 0�15 0�83 3�76 4�35 1�82 0�33 12 2520 96�36J2-228-1-R1 Umbo 151�9421 –3�7123 1818�00 49�99 0�51 16�00 9�08 5�45 4�03 0�21 6�87 12�33 1�86 0�62 0�13 68 990 97�69J2-228-2-R1 Umbo 151�9559 –3�7170 2100�00 52�52 0�45 15�78 7�48 5�52 2�18 0�16 6�79 11�49 2�11 0�51 0�09 116 920 97�49J2-202-13-R1 MSC 150�2795 –3�1642 2485�00 51�25 1�19 14�13 11�89 9�87 2�25 0�20 6�84 11�09 2�39 0�05 0�09 839 350 99�25J2-203-2-R1 MSC 150�3005 –3�1408 2400�00 51�35 1�46 13�53 13�30 10�62 2�98 0�25 6�18 10�28 2�45 0�06 0�11 947 150 99�11J2-203-7-R2 MSC 150�2972 –3�1455 2468�00 51�37 1�44 13�46 13�46 10�39 3�41 0�27 6�24 10�20 2�43 0�06 0�10 955 160 99�14J2-205-1-R1 MSC 150�3595 –3�1135 2520�00 51�30 1�42 13�50 13�47 10�65 3�13 0�22 6�18 10�27 2�46 0�06 0�10 963 180 99�09J2-206-4-R1 MSC 150�4406 –3�0719 2594�00 51�39 1�29 13�81 12�58 9�66 3�16 0�24 6�39 10�65 2�37 0�06 0�10 923 290 99�00J2-207-6-R1 MSC 150�2783 –3�1636 2487�00 63�51 1�14 13�07 9�92 8�84 1�18 0�18 1�34 5�07 3�77 0�38 0�32 348 3470 99�14

Sample no. V Cr Co Cu Zn Ga P K Ti Rb Sr Y Zr Nb Cs Ba La Ce Pr

J2-208-5-R1 17�8 1�30 4�56 15�9 79�4 34�0 477 17986 3149 24�1 255 29�5 106 1�52 0�642 319 10�2 23�4 3�24J2-209-9-R1 86�5 13�5 10�9 27�2 99�2 25�6 1389 11547 4966 21�4 367 29�7 92�9 1�41 0�566 284 9�67 22�4 3�17J2-211-2-R2 84�7 14�5 10�5 25�8 88�6 30�8 1484 13250 5260 19�3 325 26�7 85�3 1�32 0�535 257 8�80 20�6 2�87J2-211-3-R1 19�4 0�956 4�91 20�4 88�6 34�0 745 17899 3913 27�7 247 32�4 117 1�74 0�708 336 10�7 25�2 3�53J2-211-8-R1 19�1 0�719 4�80 16�9 83�2 32�5 609 17942 4001 26�0 261 31�4 114 1�65 0�671 328 10�6 24�2 3�41J2-213-4-R1 19�2 0�767 5�09 17�1 86�9 33�3 696 19518 4231 27�2 263 30�7 110 1�77 0�728 341 10�6 25�7 3�57J2-214-8-R1 36�3 26�9 7�37 18�4 92�2 36�4 692 14402 3924 25�6 313 30�6 107 1�61 0�663 327 10�5 24�5 3�40J2-214-9-R1 37�7 30�2 7�62 18�3 90�7 34�4 748 14303 3812 23�9 294 29�1 101 1�51 0�628 309 9�93 23�3 3�26J2-215-2-R1 20�0 0�224 4�96 16�7 97�8 40�0 934 17377 3920 29�5 299 33�4 120 1�90 0�765 363 12�1 28�0 3�79J2-216-11-R1 38�3 30�3 7�52 18�3 91�9 25�1 796 13863 3799 24�3 296 29�6 102 1�52 0�628 310 10�0 23�0 3�18J2-218-4-R1 40�4 34�7 7�93 18�4 84�9 34�1 714 14518 3968 24�0 296 29�3 101 1�50 0�595 291 9�19 21�4 2�94J2-218-5-R1 19�5 1�37 4�91 18�9 86�0 36�4 620 18488 3871 26�9 267 32�3 115 1�69 0�714 336 10�8 25�4 3�55J2-218-7-R1 88�1 15�6 10�9 26�0 96�0 24�6 1360 12667 5543 20�6 363 29�1 90�9 1�35 0�562 276 9�51 21�9 3�08J2-218-9-R1 183 0�348 14�4 10�8 101 30�6 2223 12169 6264 16�3 402 25�6 73�4 1�10 0�451 225 8�04 18�8 2�72J2-220-10-R1 183 0�393 14�4 8�28 104 32�5 2201 10418 5676 16�3 413 24�6 68�8 1�08 0�451 217 7�78 18�7 2�68J2-220-14-R1 100 0�381 11�0 19�4 97�0 32�1 2038 13104 5650 22�8 419 29�0 85�2 1�44 0�579 282 10�1 23�8 3�36J2-220-14-R2 40�1 0�566 7�07 12�9 94�1 30�2 1275 17702 5190 25�8 289 29�7 98�7 1�54 0�638 296 10�4 23�8 3�34J2-220-2-R1 297 81�9 29�2 93�2 68�2 23�1 818 8066 3622 11�2 426 13�0 44�0 0�668 0�351 173 4�95 11�4 1�64J2-222-11-R1 316 19�2 27�5 106 73�5 24�3 836 7357 3664 12�3 420 13�1 45�7 0�707 0�353 192 5�30 12�6 1�77J2-222-12-R1 305 42�1 29�5 97�6 74�4 24�8 767 7143 3275 12�0 443 13�6 47�0 0�736 0�381 193 5�47 12�7 1�80J2-222-9-R2 361 46�9 34�3 113 87�1 23�7 778 7246 3538 13�9 507 14�7 49�4 0�833 0�416 213 5�97 14�2 1�99J2-223-10-R1 64�5 0�719 7�86 53�1 69�7 34�9 632 17054 2982 20�0 340 23�6 89�2 1�32 0�935 402 9�23 22�4 3�17J2-223-5-R1 168 0�063 12�5 120 89�9 37�8 557 17382 2725 21�7 308 24�3 94�2 1�38 0�962 412 9�19 22�0 3�12J2-226-8-R2 99�5 0�279 13�7 94�6 87�4 35�6 637 19092 2464 19�5 403 23�6 81�9 1�14 0�764 363 8�39 20�1 2�84J2-227-1-R1 313 2�59 21�7 153 87�5 32�9 1588 17045 5135 16�6 455 19�0 68�6 1�01 0�690 340 7�49 17�7 2�51J2-227-2-R2 255 0�250 19�4 148 88�9 30�3 1002 9246 4600 13�5 477 18�7 61�5 0�914 0�568 288 6�97 16�7 2�42J2-227-4-R1 256 0�404 19�7 111 94�7 34�0 1442 15087 4836 14�8 520 19�7 65�7 0�963 0�623 323 7�98 18�8 2�70J2-228-1-R1 281 29�1 40�5 99�6 67�7 18�0 551 5160 3064 8�54 383 10�7 22�0 0�298 0�265 108 2�52 6�02 0�894J2-228-2-R1 246 102 32�6 82�2 62�1 18�5 396 4193 2715 6�15 332 10�4 28�3 0�381 0�215 112 3�07 7�70 1�15J2-202-13-R1 387 117 49�1 61�0 96�3 15�3 414 435 7152 0�734 73�4 32�2 63�4 1�03 0�009 7�65 1�91 6�67 1�19J2-203-2-R1 458 49�3 52�0 44�2 112 16�1 469 487 8727 0�785 72�6 37�7 73�3 1�18 0�020 8�58 2�26 7�76 1�41J2-203-7-R2 459 45�9 52�5 45�8 113 16�3 419 468 8657 0�820 72�5 36�7 71�9 1�17 0�009 8�54 2�25 7�70 1�38J2-205-1-R1 455 46�1 50�5 132 109 15�2 451 468 8501 0�790 67�5 33�5 67�0 1�16 0�015 8�44 2�22 7�76 1�45J2-206-4-R1 420 40�0 48�0 127 95�9 14�7 421 510 7710 0�707 71�9 29�0 56�1 0�937 0�013 8�08 1�88 6�66 1�22J2-207-6-R1 74�9 0�419 17�2 27�9 133 22�4 1379 3165 6813 4�42 97�9 91�1 263 4�62 0�102 57�2 9�76 30�0 5�08

(continued)

Journal of Petrology, 2015, Vol. 0, No. 0 5

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

compounds that can lead to memory effects and inter-

ferences during MC-ICP-MS analysis. Concentrations

and isotope ratios were measured in dynamic mode on

a Nu Instruments MC-ICP-MS system. 238U and 235U

were analysed on Faraday cups, using the 238U/235U ratio

to determine the U mass bias, assuming238U/235U¼137�88, whereas 236U and 234U were alter-

nately collected in the IC0 ion counter, which is preceded

by an energy filter. The IC0 gain was determined during

interspersed dynamic analyses of CRM145 assuming a234U/238U ratio of 5�286� 10–5 (Cheng et al., 2000).

Methods for Th isotope measurements employed a dy-namic routine with 232Th in Faraday cups and 230Th and229Th alternating on IC0 and using bracketing measure-

ments of the Th‘U’ standard (Turner et al., 2000) to obtain

the Th mass bias, which is different from that for U.

Measurements at masses 230�5 and 229�5 were used to

derive a correction for residual 232Th tail interference as

described in detail in appendix A of Sims et al. (2008).Multiple analyses of the secular equilibrium rock stand-

ard TML-3 (n¼ 5) performed at the same time as the

sample analyses yielded the following results:

U¼ 10�315 ppm, Th¼ 29�034 ppm, (234U/238U)¼1�002,

(230Th/232Th)¼1�082, (230Th/238U)¼1�004. These are

within error of secular equilibrium and published valuesfor this rock (Sims et al., 2008). The results for other

standards and a full discussion of precision and accuracy

in this laboratory have been given by Beier et al. (2010).

The Ra analysis procedure followed that used by

Turner et al. (2000). Radium was taken from the first

elution from the anionic column and converted to chlor-

ide using 6N HCl. This was then loaded in 3N HCl onto

an 8 ml cationic column and Ra was eluted using 3�75M

HNO3, and the process was repeated on a scaled-down,

0�6 ml column. The REE were then removed using a

150ml column of ElChrom Ln-spec resin and 0�1N HNO3.Radium and Ba were finally chromatographically sepa-

rated using ElChrom Sr-spec resin and 3N HNO3 as the

elutant in a 150ml procedure. Samples were loaded

onto degassed Re filaments using a Ta–HF–H3PO4 acti-

vator solution (Birck, 1986) and 228Ra/226Ra ratios were

measured to a precision typically �0�5% in dynamic ioncounting mode on a ThermoFinnigan Triton TIMS sys-

tem. Organic interferences are often noted at low tem-

peratures during TIMS analysis for Ra but were limited

here by fitting a dry scroll pump instead of the standard

rotary pump. This restricts leakage of organic mol-

ecules into the source during venting. Accuracy was as-

sessed via replicate analyses (n¼5) of TML-3 thatyielded 226Ra¼ 3532 fg g�1 and (226Ra/230Th)¼1�02,

which is within error of secular equilibrium (see also

Beier et al., 2010). Eruption ages are unknown so no

age correction was applied to the U-series data.

PETROGRAPHY

The rocks recovered from the MSC mostly have glassy

pillow rims with cryptocrystalline to very fine-grainedcentres. Phenocryst abundances in the glasses vary

from only a few phenocrysts in mafic samples from the

Table 1: Continued

Sample no. Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U

J2-208-5-R1 14�9 3�92 1�17 4�40 0�711 4�99 1�09 3�33 0�519 3�65 0�573 3�15 0�098 4�79 1�09 0�634J2-209-9-R1 15�3 4�10 1�31 4�54 0�729 5�09 1�09 3�26 0�496 3�51 0�536 2�70 0�089 4�41 0�937 0�544J2-211-2-R2 14�0 3�73 1�13 4�00 0�635 4�38 0�95 2�79 0�436 3�13 0�481 2�28 0�074 3�86 0�882 0�516J2-211-3-R1 16�5 4�22 1�26 4�78 0�801 5�69 1�19 3�61 0�563 3�95 0�598 3�32 0�105 5�09 1�15 0�684J2-211-8-R1 16�0 4�18 1�23 4�58 0�742 5�25 1�13 3�38 0�531 3�76 0�581 3�24 0�105 4�93 1�13 0�681J2-213-4-R1 16�8 4�26 1�27 4�63 0�722 5�26 1�14 3�35 0�535 3�82 0�560 3�16 0�108 5�11 1�14 0�722J2-214-8-R1 16�0 4�01 1�23 4�45 0�712 5�16 1�12 3�37 0�522 3�68 0�546 2�94 0�100 4�95 1�06 0�629J2-214-9-R1 15�1 4�04 1�19 4�30 0�687 4�95 1�06 3�21 0�486 3�47 0�531 2�90 0�093 4�76 1�05 0�630J2-215-2-R1 17�6 4�36 1�28 4�64 0�769 5�36 1�15 3�50 0�536 3�81 0�593 3�25 0�112 5�29 1�14 0�674J2-216-11-R1 15�1 3�96 1�26 4�31 0�724 5�05 1�09 3�32 0�514 3�70 0�574 2�99 0�097 4�81 1�09 0�634J2-218-4-R1 13�9 3�59 1�15 4�13 0�655 4�67 1�02 3�10 0�473 3�33 0�506 2�73 0�089 4�43 0�978 0�586J2-218-5-R1 16�4 4�33 1�23 4�50 0�748 5�30 1�16 3�58 0�544 3�94 0�577 3�24 0�102 5�05 1�13 0�672J2-218-7-R1 15�0 3�96 1�33 4�58 0�729 5�09 1�09 3�26 0�489 3�50 0�534 2�70 0�087 4�33 0�920 0�537J2-218-9-R1 13�4 3�58 1�25 3�99 0�643 4�55 0�967 2�78 0�434 2�95 0�455 2�14 0�075 3�57 0�729 0�428J2-220-10-R1 13�5 3�46 1�20 3�85 0�619 4�23 0�884 2�63 0�393 2�77 0�405 1�91 0�065 3�48 0�668 0�416J2-220-14-R1 15�9 4�15 1�38 4�62 0�741 4�92 1�04 3�10 0�478 3�35 0�511 2�40 0�087 4�23 0�848 0�527J2-220-14-R2 15�6 4�09 1�24 4�57 0�735 5�08 1�09 3�26 0�504 3�59 0�546 2�79 0�097 4�70 1�04 0�625J2-220-2-R1 7�77 2�07 0�680 2�14 0�337 2�23 0�479 1�39 0�207 1�43 0�227 1�25 0�044 2�53 0�488 0�302J2-222-11-R1 8�54 2�13 0�721 2�19 0�339 2�33 0�476 1�44 0�216 1�52 0�236 1�29 0�041 2�78 0�520 0�334J2-222-12-R1 8�29 2�07 0�707 2�21 0�336 2�35 0�506 1�44 0�217 1�54 0�227 1�31 0�048 2�64 0�471 0�294J2-222-9-R2 9�36 2�44 0�857 2�56 0�395 2�67 0�539 1�62 0�239 1�74 0�255 1�44 0�049 3�29 0�568 0�371J2-223-10-R1 15�0 3�79 1�04 3�94 0�598 4�21 0�891 2�58 0�395 2�78 0�423 2�61 0�078 5�59 0�923 0�554J2-223-5-R1 14�7 3�87 1�09 4�01 0�622 4�11 0�872 2�63 0�403 2�87 0�427 2�80 0�088 6�38 0�977 0�577J2-226-8-R2 13�6 3�64 1�09 3�85 0�586 4�01 0�860 2�47 0�377 2�77 0�417 2�31 0�075 5�33 0�856 0�511J2-227-1-R1 12�1 3�01 0�952 3�20 0�488 3�31 0�698 2�02 0�309 2�14 0�329 2�00 0�062 4�90 0�737 0�420J2-227-2-R2 11�5 2�95 0�985 3�15 0�486 3�31 0�674 1�97 0�299 2�10 0�320 1�78 0�057 4�55 0�705 0�422J2-227-4-R1 12�8 3�28 1�09 3�37 0�505 3�45 0�731 2�10 0�323 2�22 0�338 1�95 0�067 5�22 0�738 0�437J2-228-1-R1 4�42 1�33 0�492 1�61 0�266 1�84 0�403 1�19 0�173 1�25 0�190 0�651 0�019 1�84 0�241 0�151J2-228-2-R1 5�59 1�61 0�545 1�66 0�262 1�79 0�392 1�14 0�165 1�19 0�182 0�847 0�020 1�68 0�257 0�171J2-202-13-R1 7�02 2�67 1�04 4�16 0�748 5�45 1�21 3�59 0�540 3�65 0�551 1�96 0�080 0�346 0�098 0�031J2-203-2-R1 8�49 3�19 1�18 4�94 0�887 6�47 1�42 4�27 0�636 4�43 0�653 2�34 0�086 0�405 0�115 0�040J2-203-7-R2 8�17 3�26 1�19 4�82 0�859 6�33 1�36 4�15 0�630 4�35 0�654 2�28 0�087 0�418 0�119 0�038J2-205-1-R1 8�67 3�27 1�18 4�79 0�864 6�43 1�39 4�04 0�615 4�25 0�643 2�22 0�086 0�384 0�114 0�052J2-206-4-R1 7�32 2�79 1�04 3�93 0�702 5�25 1�16 3�37 0�514 3�53 0�545 1�81 0�064 0�338 0�087 0�029J2-207-6-R1 27�7 9�41 2�56 12�6 2�21 16�5 3�59 10�7 1�63 11�3 1�69 8�10 0�320 1�67 0�654 0�247

6 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

Tab

le2

:R

ad

iog

en

ic,ra

dio

act

ive

an

dst

ab

leis

oto

pe

da

tafo

rS

ou

thE

ast

Rif

tsa

mp

les

plu

ssi

xn

ew

Ma

nu

sS

pre

ad

ing

Ce

ntr

esa

mp

les

Sa

mp

len

o.

Lo

cality

Lo

ng

itu

de

(�W

)L

ati

tud

e(�

N)

Wa

ter

de

pth

(m)

U[I

D]

(mg

g–1)

Th

[ID

](m

gg

–1)

87S

r/8

6S

r14

3N

d/1

44N

d2

06P

b/2

04P

b2

07P

b/2

04P

b2

08P

b/2

04P

b(2

34U

/23

8U

)(2

30T

h/2

32T

h)

(238U

/23

2T

h)

(23

0T

h/2

38U

)2

26R

a(f

gg

–1)

(22

6R

a/2

30T

h)

d18O

(SM

OW

)

J2

-20

8-5

-R1

Pu

al

15

1�6

70

0–3�7

20

01

67

01�1

20�6

81�0

02

1�5

94

1�8

40

0�8

66

56

52�8

26�3

0J

2-2

09

-9-R

1P

ua

l1

51�6

71

8–3�7

27

41

68

91�1

70�5

70�7

03

66

70�5

13

03

51�0

05

1�5

71

1�4

95

1�0

60

52

22�5

26�0

7J

2-2

11

-2-R

2P

ua

l1

51�6

66

5–3�7

29

81

66

01�2

00�6

10�7

03

66

10�5

13

01

41�0

04

1�5

62

1�5

40

1�0

24

51

22�4

26�3

2J

2-2

11

-3-R

1P

ua

l1

51�6

66

2–3�7

27

91

62

60�7

03

70

20�5

13

04

36�2

0J

2-2

11

-8-R

1P

ua

l1

51�6

68

9–3�7

28

31

64

91�0

30�6

31�0

05

1�5

92

1�8

34

0�8

68

50

72�7

56�2

0J

2-2

13

-4-R

1P

ua

l1

51�6

74

4–3�7

21

21

68

3J

2-2

14

-8-R

1P

ua

l1

51�6

69

1–3�7

29

41

66

00�7

03

66

00�5

13

01

3J

2-2

14

-9-R

1P

ua

l1

51�6

69

2–3�7

28

91

66

0J

2-2

15

-2-R

1P

ua

l1

51�6

85

4–3�7

13

61

66

30�8

70�5

20�7

03

71

61�0

03

1�5

83

1�8

13

0�8

74

42

42�7

6J

2-2

16

-11

-R1

Pu

al

15

1�6

72

4–3�7

28

71

72

4J

2-2

18

-4-R

1P

ua

l1

51�7

27

5–3�6

75

11

87

5J

2-2

18

-5-R

1P

ua

l1

51�7

27

4–3�6

75

11

87

50�7

20�4

30�7

03

58

81

8�7

66

15�5

32

38�3

60

1�0

02

1�5

64

1�8

34

0�8

60

6�2

6J

2-2

18

-7-R

1P

ua

l1

51�7

28

6–3�6

75

31

86

80�8

30�5

11�0

03

1�5

95

1�8

41

0�8

67

42

02�8

26�0

5J

2-2

18

-9-R

1P

ua

l1

51�7

33

2–3�6

67

31

91

51�0

20�6

20�7

03

68

00�5

13

02

41�0

02

1�5

83

1�8

46

0�8

65

39

42�1

66�2

3J

2-2

20

-10

-R1

De

smo

s1

51�8

79

4–3�6

91

31

91

40�4

30�2

50�7

03

73

00�5

13

03

61�0

06

1�5

40

1�8

07

0�8

52

45

16�1

55�7

8J

2-2

20

-14

-R1

De

smo

s1

51�8

78

7–3�6

92

01

90

3J

2-2

20

-14

-R2

De

smo

s1

51�8

78

7–3�6

92

01

90

3J

2-2

20

-2-R

1D

esm

os

15

1�8

82

5–3�6

94

51

99

50�3

80�2

80�7

03

72

11

8�7

56

15�5

31

38�3

49

1�0

01

1�7

02

2�2

40

0�7

66

5�9

6J

2-2

22

-11

-R1

Pu

al

15

1�6

72

4–3�7

22

01

67

1J

2-2

22

-12

-R1

Pu

al

15

1�6

71

4–3�7

23

81

65

1J

2-2

22

-9-R

2P

ua

l1

51�6

74

5–3�7

21

21

67

81�1

10�5

70�7

04

66

51�0

01

1�5

72

1�5

64

1�0

05

61

23�1

36�1

0J

2-2

23

-10

-R1

Su

Su

15

2�1

01

4–3�7

99

21

19

80�7

30�4

41�0

04

1�5

69

1�8

55

0�8

46

32

92�5

76�2

8J

2-2

23

-5-R

1S

uS

u1

52�1

01

7–3�8

00

21

19

80�7

80�4

90�7

03

89

50�5

13

02

91

8�7

61

15�5

32

38�3

52

1�0

11

1�5

77

1�9

18

0�8

30

44

73�2

36�1

9J

2-2

26

-8-R

2S

uS

u1

52�0

91

7–3�7

91

81

59

10�6

40�3

70�7

03

68

60�5

13

01

80�9

95

1�5

59

1�7

72

0�8

80

12

21�1

0J

2-2

27

-1-R

1S

uS

u1

52�0

99

1–3�8

15

21

54

90�6

90�4

10�7

03

69

70�9

91

1�6

39

1�7

80

0�9

21

17

21�3

5J

2-2

27

-2-R

2S

uS

u1

52�0

95

9–3�8

12

01

64

20�6

50�4

10�7

03

72

10�5

13

04

01�0

30

1�6

70

1�9

03

0�8

78

16

61�3

66�4

7J

2-2

27

-4-R

1S

uS

u1

52�1

04

7–3�8

10

41

36

51�1

00�6

50�7

03

69

00�5

13

03

01�0

01

1�5

85

1�7

97

0�8

83

56

12�8

76�2

5J

2-2

28

-1-R

1U

mb

o1

51�9

42

1–3�7

12

31

81

80�2

00�1

20�7

03

83

50�5

13

01

11�0

21

1�6

96

1�8

24

0�9

38

17

74�6

45�8

2J

2-2

28

-2-R

1U

mb

o1

51�9

55

9–3�7

17

02

10

00�2

20�1

41�0

22

1�7

85

1�8

88

0�9

54

5�7

2J

2-2

02

-13

-R1

MS

C1

50�2

80

0–3�1

60

02

48

5J

2-2

03

-2-R

1M

SC

15

0�3

00

0–3�1

00

02

40

00�7

03

33

90�5

13

07

51

8�1

00

15�5

00

37�9

00

0�9

83

0�9

98

1�4

95�5

9J

2-2

03

-7-R

2M

SC

15

0�3

00

0–3�1

50

02

46

8J

2-2

05

-1-R

1M

SC

15

0�3

60

0–3�1

10

02

52

0J

2-2

06

-4-R

1M

SC

15

0�4

00

0–3�0

70

02

59

40�7

03

26

90�5

13

06

60�9

98

0�9

98

5�6

3J

2-2

07

-6-R

1M

SC

15

0�2

80

0–3�1

60

02

48

70�7

03

38

10�5

13

08

01

8�2

82

15�4

86

38�0

45

1�1

22

1�0

42

1�0

35�8

6

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

MSC and in aphyric dacites from Pual Ridge (SER) to

high phenocryst abundances in the andesitic samples

from SuSu Knolls (SER).

Most of the basalts are aphyric with <2%

phenocrysts, dominated by euhedral plagioclase. A fewolivine and clinopyroxene phenocrysts occur in a

microlitic groundmass of plagioclase and clinopyrox-

ene. Olivine phenocryst compositions are in the

range of Fo76–80 (Supplementary Data Table S2). The

plagioclase phenocrysts have anorthite contents (An)

varying from 60 to 85 mol %. The larger plagioclase

crystals are often normally zoned with albite contentsincreasing from 15 to 45 mol % from core to rim.

Dacitic samples from the MSC are porphyritic and

contain 10–20% of plagioclase, clinopyroxene,

orthopyroxene, and Fe–Ti oxides. The dacites have

plagioclase phenocrysts with An contents between 40

and 60 mol %.Samples from Pual Ridge at the SER are mostly

aphyric, aphanitic to microcrystalline, andesites to da-

cites, and frequently have glassy rims. They are charac-

terized by a moderate vesicularity (5–20% vesicles) and

elongated vesicles. The samples contain sparse pheno-

crysts (<5%) of plagioclase, clinopyroxene, orthopyrox-ene and Fe–Ti oxides.

Samples from SuSu Knolls (SER) are highly porphy-

ritic (15–20%) andesites to dacites and contain numer-

ous large plagioclase and pyroxene phenocrysts

embedded in a fine matrix of glass with pyroxene and

feldspar microlites. Large glomeroporphyritic aggre-

gates of plagioclase, pyroxene and Fe–Ti oxides areabundant in these samples. Some samples from

SuSu Knolls contain olivine (Fo92–94) phenocrysts

(Supplementary Data Table S2), which are separated

from the microlite-rich rhyolitic glass by narrow

(<10 mm) hypersthene rims.

Samples from Desmos (SER) are highly vesicular

(>30%), mostly aphyric basaltic andesites with a few

larger phenocrysts (<2% up to 2 mm in size) of plagio-

clase, clinopyroxene and olivine. Euhedral plagioclase

and anhedral pyroxene are embedded in a glassy ma-

trix containing microlites of clinopyroxene and lath-like

plagioclase. Pillow lavas sampled at Umbo in the SER

are highly vesicular (>40% vesicles by volume) and por-

phyritic (>20%), containing numerous large olivine

phenocrysts (>500mm) and minor clinopyroxene and

rare plagioclase in a glassy matrix. Olivine phenocrysts

are zoned, with Fo-rich cores (Fo94) and more ferroan

rims (Fo87). Chromium-spinel inclusions (Cr# up to 86)

are abundant in the Fo-rich olivine cores but are absent

in the rims (Supplementary Data Table S2). The pillow

rim glasses contain smaller phenocrysts (<100mm) of

olivine (Fo87) and clinopyroxene. In agreement with pre-

vious studies (Kamenetsky et al., 2001; Sinton et al.,

2003; Park et al., 2010), we did not observe any evidence

for amphibole or mica in any of the Manus Basin lavas.

Rhyolite compositions are exclusively glasses and no

whole-rock with a rhyolitic composition has been

sampled.

GEOCHEMICAL RESULTS

Major and trace element concentrations for the glasses

are presented in Table 1 and the isotope data in Table 2.

Because our main interest here concerns the liquid lineof descent we restrict the majority of the remaining dis-

cussion to the data obtained from the glasses where the

effects of crystal accumulation can be assumed to be

negligible. However, the major and trace element data

for the whole-rocks are included in Supplementary Data

Table S1 for comparison with other studies using

whole-rock data.

Major elementsIn our SER glasses, SiO2 ranges from 50 to 75 wt %,over which range MgO decreases from 6�9 to 0�1 wt %.

Mg# values in the most mafic basalts are �0�7. Figure 2

shows that these changes are accompanied by de-

creases in Al2O3 and CaO from 16 to 12 wt % and from

12 to 1 wt %, respectively, whereas total alkalis increase

from 2�5 to 6�5 wt %. On a total alkalis versus silica dia-gram there is a continuum from basalt through basaltic

andesite, andesite and dacite to rhyolite (Fig. 2a). TiO2

shows a marked inflection, increasing from 0�4 to 1 wt

% across the range 50–63 wt % SiO2 and then decreases

again to 0�4 wt % by the time SiO2¼ 75 wt % (Fig. 2b).

On the basis of this observation it is probable that

FeOtot undergoes a similar break in slope, but this is notwell captured by the available glass data (see Fig. 2d).

However, the whole-rock data show that FeOtot contents

begin to decrease after �60 wt % SiO2 (Supplementary

Data Table S1). As illustrated by comparison with the

grey fields outlined in Fig. 2, these major element

trends are consistent with published data from theEMB.

Trace elementsCompatible trace elements such as Cr have relativelylow concentrations in the glasses, ranging from 100 to

<1 ppm. This highlights that Cr and Ni concentrations

of 1240 and 550 ppm, respectively, in the whole-rock

analyses of the same samples (Supplementary Data

Table S1) primarily reflect the presence of olivine and

pyroxene phenocrysts. As shown in Fig. 3, incompatibletrace element concentrations all increase with increas-

ing SiO2, whereas Sr undergoes an initial increase in

concentration from 380 to 500 ppm followed by a de-

crease to �300 ppm. Although these trends broadly mir-

ror those of Park et al. (2010), there are subtle

differences between our glass data and their whole-rock data. For example, the increase in the Cl and Nb

contents of our SER glasses is less pronounced than in

the Park et al. (2010) whole-rock data. In detail, our Susu

Knolls glasses also show less pronounced increases in

Cl, Nb and Yb versus SiO2 than do the Pual Ridge and

Susu Knolls samples (Fig. 3). Figure 4 is a MORB-nor-

malized incompatible trace element diagram showingthat the patterns of a mafic and a silicic SER sample are

essentially parallel and have strong BABB signatures

8 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

with elevated concentrations of large ion lithophileelements (LILE) and depleted high field strength elem-

ent (HFSE) concentrations relative to the REE. Indeed,

all of the SER samples are classified as BABB using the

criteria of Sinton et al. (2003) and Beier et al. (2010), and

are markedly different from MORB samples found

along the Manus Spreading Centre further to the north(see Fig. 4).

Isotope dataSelected glasses were analysed for radiogenic, radio-

active and stable isotopes (Figs 5–7). 87Sr/86Sr and143Nd/144Nd isotope ratios are in the range of

0�7036–0�7039 and 0�51301–0�51304, respectively. One

sample (J2-222-9-R2) with an anomalously high87Sr/86Sr ratio of 0�7047 has probably undergone sea-

water contamination and was therefore not analysed

for Nd or Pb isotopes. Only three samples were ana-

lysed for Pb isotopes and for these 206Pb/204Pb,207Pb/204Pb and 208Pb/204Pb are in the range of

18�756–18�766, 15�531–15�532 and 38�349–38�360, re-spectively. These data are plotted in Fig. 5, which shows

that they essentially overlap the fields for published

data (Park et al., 2010). The most data were obtained for

the Sr isotope system but there is no clear correlation

between 87Sr/86Sr and SiO2 (Fig. 7a).

The new U–Th–Ra disequilibria data are shown in

Fig. 6, where they are also compared with Manus Basindata published by Beier et al. (2010). On the U–Th

8

10

12

14

16

18

20

Al 2O

3 [w

t.%]

45 50 55 60 65 70 75SiO2 [wt.%]

0

5

10

MgO

[wt.%

]Basalt

basalt.A

ndesite

Dacite

Andesite

Rhyolite

EMBpublished

(a)

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

EMB published

EMB published

(c)

(e)

EMBpublished

EMBpublished

EMB published

(b)

(d)

(f)

0.2 GPa, dry0.2 GPa, 1 wt.% H2O0.2 GPa, 4 wt.% H2O0.4 GPa, 0.7 wt.% H2O

45 50 55 60 65 70 75SiO2 [wt.%]

0.0

0.4

0.8

1.2

1.6

TiO2 [w

t.%]

2

4

6

8

10

12

14

FeOT [w

t.%]

0

5

10

15

CaO

[wt.%

]

0

2

4

6

8

Na 2O

+K2O

[wt.%

]

Fig. 2. Major element Harker diagrams for South East Rift glasses. Grey field shows the range of published data (Sun et al., 2007;Park et al., 2010) from the Eastern Manus Basin (EMB). Superimposed are curves for the results of rhyolite-MELTS models (Gualdaet al., 2012) run at a range of pressures and H2O contents (see text for details). Tick marks indicate 10% increments of crystallizationfrom 10 to 80%.

Journal of Petrology, 2015, Vol. 0, No. 0 9

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

isochron diagram the majority of the samples lie to theright of the equiline with 5–23% 238U excess. All sam-

ples analysed for 226Ra have excesses with (226Ra/230Th)

ratios that range from 1�1 to 6�1; these indicate that no

age correction is required for the 230Th–238U data.

However, because the age of the samples is not known,

an unknown amount of 226Ra decay may have occurredand the (226Ra/230Th) ratios must be considered as min-

ima. One Pual sample that is within error of 230Th–238U

secular equilibrium and two that lie to the other side of

the equiline with small (2–6%) 230Th excesses fall out-

side the fields for the published data in Fig. 6. There is

no correlation of either (230Th/238U) or (226Ra/230Th) with

SiO2 content (Fig. 7b and c).Oxygen isotopes are highly sensitive to assimilation

of materials that have undergone low-temperature frac-

tionation but far less sensitive to the small amounts

of subducted components added to the source of

arc-related magmas (e.g. Eiler et al., 2000a, 2000b). Forthe SER basalts, the O isotope data exhibit a moderately

large range in d18O from 5�72 to 6�47% (Table 2) and

these ratios systematically increase between 50 and

70 wt % SiO2 (Fig. 7d), after which they remain relatively

constant. The range is smaller than observed by

Macpherson et al. (2000) for samples mainly locatedfurther north in the Manus Basin (d18O¼ 5�28–6�68%)

where there may be a plume influence.

PETROGENESIS

As stated in the introduction, it is not our purpose here

to discuss source composition variations in any detail

and we accept previously published arguments that the

entirety of the Manus Basin is polluted by a single com-ponent containing contributions from both subducted

sediments and fluids from altered oceanic crust (Sinton

amphibolitemelting

EMB published

EMB published

EMB published

EMB publishedEMB published

EMB published

amphibolitemelting

0

2,000

4,000

6,000

8,000

Cl [

ppm

]

0

200

400

600

800

Sr [

ppm

]

45 50 55 60 65 70 75SiO2 [wt.%]

0

5

10

15

La [p

pm]

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

0

100

200

300

400

500

Ba [ppm

]

0

1

2

3

Nb [ppm

]

45 50 55 60 65 70 75SiO2 [wt.%]

0

1

2

3

4

5

Yb [ppm

]

(a)

(c)

(e)

(b)

(d)

(f)

Fig. 3. Selected trace element Harker diagrams for South East Rift glasses. Grey field shows the range of published data (Sun et al.,2007; Park et al., 2010) from the Eastern Manus Basin (EMB). Superimposed are fractional crystallization curves that combine theSiO2 results of the 0�2 GPa, 1% H2O MELTS model with Rayleigh crystal fractionation calculations (see text for details). Tick marksindicate 10% increments of crystallization from 10 to 80%. The amphibolite melting curves on the plots for La and Yb are fromBrophy (2009).

10 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

et al., 2003; Sun et al., 2007; Beier et al., 2010; Park et al.,

2010). The overlap of our new Sr–Nd–Pb data with those

from the previous studies (Fig. 5) is entirely consistent

with this and we assume that similar H2O contents to

those measured by Sinton et al. (2003) and Shaw et al.(2004) are applicable to the SER BABB.

One feature of our present dataset is that there are at

least two distinct trends on a plot of Ba versus Yb

(Fig. 8) and that the main distinction between these

groups of samples is in Yb not Ba, as evidenced by two

populations distinct in their Yb concentrations (Fig. 3f).This permits models in which there is variable addition

of the slab-derived components to a mantle that was

variably depleted. From a U-series perspective there is

no reason arising from our new data not to adopt the

model of Beier et al. (2010) in which the 238U excesses

are attributed largely to decompression melting of oxi-

dized mantle (approximately quartz–fayalite–magnetite;QFM) in which U is significantly more incompatible

than Th.

In summary, although the trace element and isotope

data require variations in relative contributions of

source components, the magnitudes of those variations

are insufficient to demand that there was much vari-ation in melting conditions and, by implication, the

major element compositions of the parental magmas.

Accordingly, in the following sections, we explore mod-

els to explain the extended compositional range of the

SER samples assuming a broadly similar parental

magma in terms of major and, to a lesser extent, trace

element concentrations. We begin with closed-systemfractional crystallization and then assess evidence for

partial melting of either altered oceanic crust or gab-

bros in the crust the magmas traversed. Thermally

linked models that combine fractional crystallization

with assimilation of either of those components are

also explored along with the timescale implications

from the U-series isotopes.

Fractional crystallizationThe combination of the restricted sampling region and

the coherent, curvilinear major and trace elementarrays (Figs 2–4) supports the likelihood of a broadly

similar liquid line of descent for the SER glasses. Under

0.01

0.1

1

10

100

1,000 S

ampl

e/N

-MO

RB

CsRb

BaTh

UKNb

TaLa

CePb

PrSr

NdPSm

ZrHf

EuGd

TbTi

HoY

ErTm

DyYb

Lu

J2-223-10-R174.57 wt.% SiO2 SuSu Knolls

MSC

J2-202-13-R151.24 wt.% SiO2 MSC

SER

J2-228-1-R1, 49.99 wt.% SiO2 Umbo

Fig. 4. MORB-normalized incompatible trace element diagramcomparing the patterns for a mafic and a silicic glass from theSouth East Rift with a typical MORB from the Manus SpreadingCentre (MSC). Normalization values used here are from Sun &McDonough (1989).

0.7035 0.7037 0.7039 0.704187Sr/86Sr

0.51295

0.51300

0.51305

0.51310

0.51315

0.51320

143 N

d/14

4 Nd

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

15.53

15.54

15.55

207 P

b/20

4 Pb

18.75 18.77 18.79206Pb/204Pb

38.33

38.35

38.37

38.39

208 P

b/20

4 Pb

(a)

(b)

(c)

EMB published

EMB published

EMB published

Fig. 5. Variation of 87Sr/86Sr vs 143Nd/144Nd, 207Pb/204Pb vs206Pb/204Pb, and 208Pb/204Pb vs 206Pb/204Pb showing the newSouth East Rift glass data relative to fields for published data(Sinton et al., 2003; Beier et al., 2010; Park et al., 2010). It shouldbe noted that the anomalous sample with the highest 87Sr/86Srratio (0�7047) has probably experienced seawater contamin-ation and was not analysed for 143Nd/144Nd and Pb isotopes.

Journal of Petrology, 2015, Vol. 0, No. 0 11

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

these circumstances, the simplest mechanism to ex-

plain the extended compositional range of the glasses

is closed-system fractional crystallization. We have

used 5�C steps in rhyolite-MELTS (Gualda et al., 2012)

to simulate this using one of the basalts (J2-228-2-R1)as the parental melt (see Table 3 for summary details).

The measured Fe2þ/Fetot ratio was used to constrain the

oxygen fugacity and this corresponds to �QFM, con-

sistent with the back-arc setting and previous work

(Sinton et al., 2003; Beier et al., 2010). The remaining

key variables are pressure and H2O content, and we

conducted multiple runs varying these two parameters,the results of which are shown in Fig. 2. As can be seen,

pressures around 0�2 GPa and relatively low H2O con-

tents of �1 wt % provide the best simulation of the com-

bined major element data. This pressure corresponds

to the base of the crust at a spreading centre (see

Wanless & Shaw, 2012) and is consistent with recentgeophysical observations for the location of magma

chambers (e.g. Crawford et al., 1999). The H2O contents

are consistent with measured H2O contents in basalts

elsewhere in the Manus Basin (Sinton et al., 2003; Shaw

et al., 2004; Sun et al., 2007).

In detail, rhyolite-MELTS models under these condi-tions result in small initial increases in FeO and Al2O3,

followed by decreases with increasing SiO2 that are not

observed in the available glass data (Fig. 2c and d).

0.7035

0.7037

0.7039

0.7041

87S

r/86S

r0.70

0.80

0.90

1.00

1.10

(230 Th

/238 U

)

1234567

(226 R

a/23

0 Th)

45 50 55 60 65 70 75SiO2 [wt.%]

5.0

5.4

5.8

6.2

6.6

18O

(SM

OW

)

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

Th excess U excess

Ra excess

fractional crystallisation

EMBpublished

EMB published

EMB published

N-MORB range

(a)

(b)

(c)

(d)

Fig. 7. Variation of 87Sr/86Sr, (230Th/238U), (226Ra/230Th) andd18O vs SiO2. Published data (grey fields) from Beier et al.(2010). N-MORB range in Fig. 7d from Eiler et al. (2000b).

0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2(238U/232Th)

0.8

1.0

1.2

1.4

1.6

1.8(23

0 Th/23

2 Th)

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

0.7 0.8 0.9 1.0 1.1(230Th/238U)

0

1

2

3

4

5

6

7

(226 R

a/23

0 Th)

Manus Basin published

Manus Basinpublished

230Th-excess

238U-excess

230Th-excess226Ra-excess

238U-excess226Ra-excess

(a)

(b)

Fig. 6. Variation of (230Th/232Th) vs (238U/232Th) and(226Ra/230Th) vs (230Th/238U) showing the new South East Riftglass data relative to fields for published data (Beier et al.,2010).

12 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

However, this is probably a reflection of the limited

number of glasses analysed across this SiO2 range and

the whole-rock data suggest that such inflections do infact occur in these lavas (Supplementary Data Table

S1). The other major elements are well modelled and it

is clear from the FeO, Al2O3 and CaO trends that high or

lower H2O contents result in major inflections for these

elements that are not evident in the data (see Fig. 2).

After 70 wt % SiO2 was reached rhyolite-MELTS pre-

dicted subsequent decreases in SiO2 that seem highlyimprobable and so we display only the MELTS results

up to 70 wt % SiO2.

To assess further the efficacy of closed-system frac-

tional crystallization to explain the data we also per-

formed trace element modelling. We assumed a

Rayleigh fractionation model and the phase proportionsfrom the 0�2 GPa, 1 wt % H2O rhyolite-MELTS model in

5 wt % SiO2 increments (see Table 3). As detailed in

Table 3, the extracted mineral assemblage is dominated

by plagioclase and clinopyroxene in different propor-

tions throughout and for these we used partition coeffi-

cients from the compilation of Halliday et al. (1995). Allother mineral modes (i.e. olivine, orthopyroxene and

magnetite) were <2–6% and thus insignificant for the

trace elements we chose to model. The results are plot-

ted in Fig. 3 and show that the model simulates the data

reasonably well for most incompatible elements. That

these results are not as satisfactory as for the major

elements (see Fig. 2) probably reflects subtle variationsin the trace element composition of the parental mag-

mas consistent with the variations in radiogenic iso-

topes (Fig. 5). It should be noted that the Cl model is

from rhyolite-MELTS, which provides a close match to

the data (Fig. 3a). An inflection owing to an increase in

the proportion of plagioclase in the fractionating assem-blage at 57 wt % SiO2 is most apparent in the models

for Sr, Nb, Ba and Yb (see Fig. 3). This does a good job

of simulating the Nb data for the Umbo glasses but is

not apparent in the glass data for Sr and Yb; this may

reflect the choice of partition coefficients used.

Partial melting of altered oceanic crust or gabbroAlthough the fractional crystallization modelling pro-

vides a reasonable fit to the glass data, there are some

mismatches and a number of workers have used obser-

vational and/or experimental data to suggest that silicic

lavas at spreading centres derive from melting of

altered oceanic crust or gabbro (e.g. Coogan et al.,2003; Koepke et al., 2007). In the case of our SER

glasses, the d18O (5�72–6�47%) data overlap other data

from the Manus Basin, being similar to or slightly

higher than values for pristine MORB glasses (Eiler

et al., 2000b; Eiler, 2001). Because altered gabbros have

much lower d18O of 0–5% (Eiler et al., 2000b; Staudigel,2003) this would appear to preclude an origin via partial

melting of altered gabbros.

Altered oceanic basalts have d18O>5�8 (Eiler et al.,

2000b) and on this basis could provide a source for the

silicic SER glasses. However, altered oceanic crust also

has 87Sr/86Sr�0�7045 (e.g. Bach et al., 2003), which,

with the exception of sample J2-222-9-R2, is signifi-cantly higher than values for any of the SER glasses

(Table 2), and there is no correlation between SiO2 and87Sr/86Sr (Fig. 7a). Brophy (2009) has shown that partial

melting of altered (i.e. amphibole-bearing) mafic litholo-

gies results in silicic melts with low La and Yb concen-

trations and trends that decrease with increasing SiO2

owing to the formation of clinopyroxene 6 garnet by

peritectic reactions during the breakdown of amphibole.

Calculated partial melting curves from Brophy (2009)

for La and Yb are shown in Fig. 3e and f and clearly do

not replicate either the elevated concentrations of these

elements or their positively sloped trends against SiO2.Therefore, we discount partial melting models for the

origin of the silicic SER lavas.

Assimilation and fractional crystallizationAlthough we have ruled out direct partial melting of

altered crust as an origin for the silicic SER lavas, it is

possible that fractional crystallization was accompaniedby assimilation of these lithologies (i.e. AFC; DePaolo,

1981) Indeed, this was the preferred model of Wanless

et al. (2010) for silicic lavas found at several mid-ocean

spreading centres. The latter authors used marked in-

creases in Cl/K along with oxygen isotopes as their pri-

mary evidence for AFC of altered crust; however, in the

case of the SER glasses closed-system fractional crys-tallization models can readily simulate both K (not

shown but implicit in Fig. 2a) and Cl (Fig. 3a) and thus

Cl/K (Fig. 9).

In principle, the increase in d18O with increasing de-

gree of differentiation in the SER glasses could be

attributed to an AFC process involving altered crust.However, as shown by the fractionation curve in

Fig. 10a, this evolution can equally be explained by

0 1 2 3 4 5Yb [ppm]

0

100

200

300

400

500B

a [p

pm]

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

EMB published

Fig. 8. Variation of Ba vs Yb showing the split of the South EastRift glass data into two groups with different slopes suggestingvariable addition of Ba from a subduction component to man-tle that was variably depleted in Yb. Published data from Sunet al. (2007) and Park et al. (2010) (grey field)

Journal of Petrology, 2015, Vol. 0, No. 0 13

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

closed-system fractionation (see Eiler, 2001).

Nevertheless, to test open-system processes further we

conducted energy-constrained AFC calculations using

the formulation of Spera & Bohrson (2001). For com-

pleteness, both altered crust and gabbro were con-sidered using the partition coefficients from Table 3

(plus DCl¼0�045) and the following end-member com-

positions: parental magma Cl¼ 0�013 wt %, K2O¼ 0�47

wt %, Sr¼ 383 ppm, d18O¼ 5�6%, 87Sr/86Sr¼0�7036;

altered oceanic crust Cl¼ 0�1 wt %, K2O¼ 1 wt %,

Sr¼ 200 ppm, d18O¼ 10%, 87Sr/86Sr¼0�7045; altered

gabbro Cl¼ 0�05 wt %, K2O¼ 0�1 wt %, Sr¼ 400 ppm,d18O¼3%, 87Sr/86Sr¼ 0�703 (based on Eiler et al.,

2000b; Kent et al., 2002; Faure & Mensing, 2004; Sun

et al., 2007). The results of these models are shown in

Fig. 10b and c. As can be seen, both models lead to

strongly coupled variations between d18O and Cl/K and87Sr/86Sr whereas the SER glasses show no coupling ofthese indices and are not simulated by the model

trends. Accordingly, we conclude that there is no com-

pelling evidence for a significant role for AFC in our

data.

U-series isotope implicationsThe U-series data largely overlap the previously pub-

lished data of Beier et al. (2010) and include both 230Th

and 238U excesses. Following Beier et al. (2010), we ex-

plain this by melting of mantle having variable oxida-

tion states. This changes the oxidation state of U, such

that U can be either more or less compatible than Th

during melting (Lundstrom et al., 1994) resulting in 238Uor 230Th excesses, respectively.

The age of the samples is unknown; the 226Ra

excesses represent minima and this limits the amount

of information that can be derived from these data.

Nevertheless, most models for Ra–Th disequilibria in

mantle-derived magmas attribute the disequilibria to

melting processes and/or fluid addition [see Turneret al. (2003) for a review]. Therefore, the preservation of226Ra excess in every SER glass analysed, irrespective

of its silica content (see Fig. 7c), suggests that magmatic

evolution, whether by closed-system fractionation or

AFC, must have occurred in less than a few millennia.

This requires relatively rapid cooling, which is mostlikely to occur in relatively small and shallow magma

bodies (e.g. Dosseto & Turner, 2010), consistent with

the evidence from the MELTS modelling for a low-

pressure liquid line of descent.

CONCLUDING REMARKS

A number of hypotheses have been advanced to ex-

plain the (rare) occurrence of silicic lavas at spreading

centres. Although the back-arc setting has greater com-

plexity (especially in incompatible trace elements,radiogenic isotopes and H2O content) than mid-ocean

ridges, many aspects pertaining to magmatic evolution

should be similar. In our case of the South East Rift in

the Manus Basin it would appear that closed-system

fractional crystallization is adequate to explain the ma-

jority of the geochemical observations. If some assimi-lation of altered oceanic crust does accompany this

fractional crystallization then it would appear that the

amount of assimilant added is very small and unlikely

to be the main driver of evolution to high-silica compos-

itions. Moreover, because there are Ra disequilibria in

all samples analysed this must have occurred in less

than a few millennia. The outstanding question is whythere is such an extended range of compositions here,

unlike other spreading centres in the Manus Basin or

elsewhere in general. The bathymetry in the SER is rela-

tively shallow (1200–2100 m), suggesting relatively thick

crust (e.g. Klein & Langmuir, 1987); we hypothesize that

this might be conducive to rapid cooling, in which casea lower magma supply rate would favour evolution to

highly differentiated compositions (Dosseto & Turner,

45 50 55 60 65 70 75SiO2 [wt.%]

0.10

0.20

0.30

0.40

0.50

0.60

Cl/K

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

EMB published

Fig. 9. Variation of Cl/K vs SiO2 along with results from the pre-ferred 0�2 GPa, 1 wt % H2O MELTS model. Tick marks indicate10% increments of crystallization from 10 to 80%. EMB,Eastern Manus Basin.

Table 3: Phase proportions and partition coefficients used in Rayleigh fractionation modelling

SiO2 range (wt %) DBa DK DNb DSr DLa DYb

50–55 55–60 60–65 65–70

Liquid fraction 60 11 7 6Clinopyroxene 50 36 22 30 0�0003 0�001 0�0089 0�091 0�054 0�43Plagioclase 50 64 78 70 0�63 0�09 0�02 6�8 0�32 0�04

14 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

2010; O’Neill & Jenner, 2013). Such a model may have

more general application to the generation of silicic

magmas elsewhere in the ocean basins, including off-

axis and ocean island settings.

ACKNOWLEDGEMENTS

We thank Heike Anders from the Universitat Bremen,

and Heike Rothe, Andrea Gottsche, Sabine Schumann,

Rudolph Naumann and Knut Hahne, all of GFZ

Potsdam, for their assistance and support during la-boratory work. We thank Peter Appel and Barbara

Mader from the Universitat Kiel for their help during

microprobe analyses, Richard Wysoczanski, Bruce

Schaefer and an anonymous reviewer for their con-

structive comments, H. and K. MacAskill for inspiration,

and Richard Price for editorial guidance. We acknow-ledge help by Roland Maas, Peter Wieland and Norman

Pearson during isotope analyses.

FUNDING

This research was funded by the DFG (German

Research Foundation) project BA 1605/4-1. C.B. was

funded by a Feodor Lynen fellowship of the Alexander

von Humboldt-Foundation. S.T. acknowledges the sup-

port of a Humboldt Research Award.

SUPPLEMENTARY DATA

Supplementary data for this paper are available at

Journal of Petrology online.

REFERENCES

Bach, W., Peucker-Ehrenbrink, B., Hart, S. R. & Blusztajn, J. S.(2003). Geochemistry of hydrothermally altered oceanic crust:DSDP/ODP Hole 504B—implications for seawater–crustexchange budgets and Sr- and Pb-isotopic evolution of themantle. Geochemistry, Geophysics, Geosystems 4, 8904.

Beier, C., Turner, S. P., Sinton, J. M. & Gill, J. B. (2010).Influence of subducted components on back-arc melting dy-namics in the Manus Basin. Geochemistry, Geophysics,Geosystems 11, Q0AC03.

Birck, J. L. (1986). Precision K–Rb–Sr isotopic analysis—application to Rb–Sr chronology. Chemical Geology 56,73–83.

Brophy, J. G. (2009). La–SiO2 and Yb–SiO2 systematics in mid-ocean ridge magmas: implications for the origin of oceanicplagiogranite. Contributions to Mineralogy and Petrology158, 99–111.

Caulfield, J., Turner, S., Arculus, R., Dale, C., Jenner, F. E.,Pearce, J., Macpherson, C. G. & Handley, H. K. (2012).Mantle flow, volatiles, slab-surface temperatures and melt-ing dynamics in the north Tonga arc–Lau back-arc basin.Journal of Geophysical Research 117.

Cheng, H., Edwards, R. L., Hoff, J., Gallup, C. D., Richards, D. A.& Asmerom, Y. (2000). The half-lives of uranium-234 andthorium-230. Chemical Geology 169, 17–33.

Coogan, L. A., Mitchell, N. C. & O’Hara, M. J. (2003). Roof as-similation at fast spreading ridges: An investigation combin-ing geophysical, geochemical, and field evidence. Journalof Geophysical Research—Solid Earth 108, 1–14.

Cooper, P. & Taylor, B. (1987). Seismotectonics of New Guinea:a model for arc reversal following arc–continent collision.Tectonics 6, 53–67.

Crawford, W. C., Webb, S. C. & Hildebrand, J. A. (1999).Constraints on melt in the lower crust and Moho at theEast Pacific Rise, 9�48’N, using seafloor compliance

0 1 2 3 4 5 6 7 8MgO [wt.%]

5.0

5.4

5.8

6.2

6.618

O (S

MO

W)

SER Pual RidgeSER DesmosSER SuSu KnollsSER Umbo

0.0 0.2 0.4 0.6 0.8 1.0Cl/K

5.0

5.4

5.8

6.2

6.6

18O

(SM

OW

)

N-MORB range

0.7030 0.7035 0.7040 0.704587Sr/86Sr

5.0

5.4

5.8

6.2

6.6

18O

(SM

OW

)

EC-AFC of altered oceanic crust

EC-AFC of altered

oceanic gabbro

EC-AFC of altered oceanic gabbro

EC-AFC of altered oceanic crust

(a)

(b)

(c)

Manus Basinpublished

crystallisation

fractional

Fig. 10. Variation of d18O vs (a) MgO, (b) Cl/K and (c) 87Sr/86Sr.Also shown in (a) is the range of fresh MORB glasses (Eileret al., 2000b) and a calculated, closed-system fractionationcurve (87% crystallization in total) from Juster et al. (1989).Shown in (b) and (c) are the results of energy-constrained as-similation and fractional crystallization (EC-AFC) models fol-lowing Spera & Bohrson (2001) using input parameters asdetailed in the text.

Journal of Petrology, 2015, Vol. 0, No. 0 15

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

measurements. Journal of Geophysical Research—SolidEarth 104, 2923–2939.

DePaolo, D. J. (1981). Trace element and isotopic effects ofcombined wallrock assimilation and fractional crystalliza-tion. Earth and Planetary Science Letters 53, 189–202.

Dosseto, A. & Turner, S. P. (2010). Magma cooling and differen-tiation—Uranium series isotopes. In: Dosseto, A., Turner, S.P. & van Orman, J. (eds) Timescales of Magmatic Processes:from Core to Atmosphere. Chichester: Wiley–Blackwell, pp.160–180.

Eiler, J. M. (2001). Oxygen isotope variations of basaltic lavasand upper mantle rocks. In: Valley, J. W. & Cole, D. R. (eds)Stable Isotope Geochemistry. Mineralogical Society ofAmerica and Geochemical Society, Reviews in Mineralogyand Geochemistry 43, 319–364.

Eiler, J. M., Crawford, A., Elliot, T., Farley, K., A., Valley, J. W., &Stolper, E. M. (2000a). Oxygen isotope geochemistry of oce-anic-arc lavas. Journal of Petrology 41, 229–256.

Eiler, J. M., Schiano, P., Kitchen, N. & Stolper, E. M. (2000b).Oxygen-isotype evidence for recycled crust in the sources ofmid-ocean-ridge basalts. Nature 403, 530–534.

Faure, G. & Mensing, T. M. (2004). Isotopes: Principles andApplications. New York: John Wiley.

Fretzdorff, S., Livermore, R. A., Devey, C. W., Leat, P. T. &Stoffers, P. (2002). Petrogenesis of the back-arc East ScotiaRidge, South Atlantic Ocean. Journal of Petrology 43,1435–1467.

Gualda, G. A. R., Ghiorso, M. S., Lemons, R. V. & Carley, T. L.(2012). Rhyolite-MELTS: a modified calibration of MELTSoptimized for silica-rich, fluid-bearing magmatic systems.Journal of Petrology 53, 875–890.

Halliday, A. N., Lee, D.-C., Tommasini, S., Davies, G. R., Paslick,C. R., Fitton, J. G. & James, D. E. (1995). Incompatible traceelements in OIB and MORB and source enrichment in thesub-oceanic mantle. Earth and Planetary Science Letters133, 379–395.

Heyworth, Z., Turner, S., Schaefer, B., Wood, B., George, R.,Berlo, K., Cunningham, H., Price, R., Cook, C. & Gamble, J.(2007). 238U–230Th–226Ra–210Pb constraints on the genesis ofhigh-Mg andesites at White Island, New Zealand. ChemicalGeology 243, 105–121.

Johnson, R. W., Mutter, J. C. & Arculus, R. J. (1979). Origin ofthe Willaumez–Manus Rise, Papua New Guinea. Earth andPlanetary Science Letters 44, 247–260.

Juster, T., Grove, T. L. & Perfit, M. R. (1989). Experimentalconstraints on the generation of FeTi basalts, andesitesand rhyodacites at the Galapagos spreading center,85�W and 95�W. Journal of Geophysical Research 94,9251–9274.

Kamenetsky, V. S., Binns, R. A., Gemmell, J. B., Crawford, A. J.,Mernagh, T. P., Maas, R. & Steele, D. (2001). Parental bas-altic melts and fluids in eastern Manus backarc basin: impli-cations for hydrothermal mineralisation. Earth andPlanetary Science Letters 184, 685–702.

Kent, A. J. R., Peate, D. W., Newman, S., Stolper, E. M. &Pearce, J. A. (2002). Chlorine in submarine glasses from theLau Basin: seawater contamination and constraints on thecomposition of slab-derived fluid. Earth and PlanetaryScience Letters 202, 361–377.

Klein, E. M. & Langmuir, C. H. (1987). Global correlations ofocean ridge basalt chemistry with axial depth and crustalthickness. Journal of Geophysical Research, B, Solid Earthand Planets 92, 8089–8115.

Koepke, J., Berndt, J., Feig, S. T. & Holtz, F. (2007). The forma-tion of SiO2-rich melts within the deep oceanic crust by hy-drous partial melting of gabbros. Contributions toMineralogy and Petrology 153, 67–84, doi:10.1007/S00410-006-0135-Y.

Lee, S.-M. & Ruellan, E. (2006). Tectonic and magmatic evolu-tion of the Bismarck Sea, Papua New Guinea: Review andnew synthesis. In: Christie, D. M., Fisher, C. R., Lee, S.-M. &Givens, S. (eds) Back-arc Spreading Systems, Geological,Biological, Chemical and Physical Interactions. AmericanGeophysical Union, Geophysical Monographs 166, 263–286.

Le Roux, P. J., Shirey, S. B., Hauri, E. H., Perfit, M. R. & Bender,J. F. (2006). The effects of variable sources, processes andcontaminants on the composition of northern EPR MORB(8–10�N and 12–14�N): Evidence from volatiles (H2O, CO2, S)and halogens (F, Cl). Earth and Planetary Science Letters251, 209–231.

Le Roux, V., Bodinier, J. L., Alard, O., O’Reilly, S. Y. & Griffin, W.L. (2009). Isotopic decoupling during porous melt flow: Acase-study in the Lherz peridotite. Earth and PlanetaryScience Letters 279, 76–85.

Lundstrom, C. C., Shaw, H. F., Ryerson, F. J., Phinney, D. L., Gill,J. B. & Williams, Q. (1994). Compositional controls on thepartitioning of U, Th, Ba, Pb, Sr and Zr between clinopyrox-ene and haplobasaltic melts: implications for uranium seriesdisequilibria in basalts. Earth and Planetary Science Letters128, 407–423.

Macpherson, C. G., Hilton, D. R., Mattey, D. P. & Sinton, J. M.(2000). Evidence for an 18O-depleted mantle plume fromcontrasting 18O/16O ratios of back-arc lavas from the ManusBasin and Mariana Trough. Earth and Planetary ScienceLetters 176, 171–183.

Martinez, F. & Taylor, B. (1996). Backarc spreading, rifting, andmicroplate rotation, between transform faults in the ManusBasin. Marine Geophysical Researches 18, 203–224.

Martinez, F. & Taylor, B. (2003). Controls on back-arc crustal ac-cretion: insights from the Lau, Manus and Mariana basins.In: Larter, R. D. & Leat, P. T. (eds) Intra-oceanic SubductionSystems: Tectonic and Magmatic Processes. GeologicalSociety, London, Special Publications 219, 19–54.

O’Neill, H. S. C. & Jenner, F. E. (2013). The global pattern oftrace-element distributions in ocean floor basalts. Nature491, 698–704.

Park, S.-H., Lee, S.-M., Kamenov, G. D., Kwon, S.-T. & Lee, K.-Y.(2010). Tracing the origin of subduction components be-neath the South East rift in the Manus Basin, Papua NewGuinea. Chemical Geology 269, 339–349.

Pearce, J. A. & Stern, R. J. (2006). The origin of back-arc basinmagmas: trace element and isotope perspectives. In:Christie, D. M. (ed.) Back-Arc Spreading Systems:Geological, Biological, Chemical, and Physical Interactions,Washington D.C., pp. 63–86.

Pearce, J. A., Ernewein, M., Bloomer, S. H., Parson, L. M.,Murton, B. J. & Johnson, L. E. (1994). Geochemistryof the Lau Basin volcanic rocks: Influence of ridgesegmentation and arc proximity. In: Smellie, J. L. (ed.)Volcanism Associated with Extension at Consuming PlateMargins. Geological Society, London, Special Publications81, 53–75.

Sharp, Z. D. (1990). A laser-based microanalytical method forthe in situ determination of oxygen isotope ratios of silicatesand oxides. Geochimica et Cosmochimica Acta 54,1353–1357.

Shaw, A. M., Hilton, D. R., Macpherson, C. G. & Sinton, J. M.(2004). The CO2–He–Ar–H2O systematics of the Manus back-arc basin: resolving source composition from degassingand contamination effects. Geochimica et CosmochimicaActa 68, 1837–1855.

Sims, K. W. W., Gill, J. B., Dosseto, A., Hoffmann, D. L.,Lundstrom, C. C., Williams, R. W., Ball, L., Tollstrup, D.,Turner, S., Prytulak, J., Glessner, J. J. G., Standish, J. J. &Elliott, T. (2008). An inter-laboratory assessment of the thor-ium isotopic composition of synthetic and rock reference

16 Journal of Petrology, 2015, Vol. 0, No. 0

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from

materials. Geostandards and Geoanalytical Research 32,65–91.

Sinton, J. M., Ford, L. L., Chappell, B. & McCulloch, M. T. (2003).Magma genesis and mantle heterogeneity in the Manusback-arc basin, Papua New Guinea. Journal of Petrology 44,159–195.

Spera, F. J. & Bohrson, W. A. (2001). Energy-constrained open-system magmatic processes I: General model and energy-constrained assimilation and fractional crystallization(EC-AFC) formulation. Journal of Petrology 42, 999–1018.

Staudigel, H. (2003). Hydrothermal alteration processes in theoceanic crust. In: Rudnick, R. L. (ed.) Treatise onGeochemistry. Elsevier 3, 511–535.

Stolper, E. & Newman, S. (1994). The role of water in the petro-genesis of Mariana trough magmas. Earth and PlanetaryScience Letters 121, 293–325.

Sun, S.-s. & McDonough, W. F. (1989). Chemical and isotopicsystematics of oceanic basalts: implications for mantle com-position and processes. In: Saunders, A. D. & Norry, M. J.(eds) Magmatism in the Ocean Basins. Geological Society,London, Special Publications 42, 313–345.

Sun, W. D., Binns, R. A., Fan, A. C., Kamenetsky, V. S., Wysoczanski,R., Wei, G. J., Hu, Y. H. & Arculus, R. J. (2007). Chlorine in sub-marine volcanic glasses from the eastern Manus Basin.Geochimica et Cosmochimica Acta 71, 1542–1552.

Taylor, B. (1979). Bismarck Sea: Evolution of a back-arc basin.Geology 7, 171–174.

Taylor, B., Crook, K. & Sinton, J. (1994). Extensional transformzones and oblique spreading centers. Journal ofGeophysical Research 99, 19707–19718.

Tregoning, P. (2002). Plate kinematics in the western Pacificderived from geodetic observations. Journal of GeophysicalResearch—Solid Earth 107, ECV 7-1–ECV 7-8.

Turner, S., Bourdon, B., Hawkesworth, C. & Evans, P. (2000).226Ra–230Th evidence for multiple dehydration events, rapid

melt ascent and the time scales of differentiation beneaththe Tonga–Kermadec island arc. Earth and PlanetaryScience Letters 179, 581–593.

Turner, S., Bourdon, B. & Gill, J. B. (2003). Insights into magmagenesis at convergent margins from U-series isotopes. In:Bourdon, B., Henderson, G. M., Lundstrom, C. & Turner, S.P. (eds) Uranium-series Geochemistry. MineralogicalSociety of America and Geochemical Society, Reviews inMineralogy and Geochemistry 52, 255–310.

Wallace, L. M., McCaffrey, R., Beavan, J. & Ellis, S. (2005). Rapidmicroplate rotations and backarc rifting at the transition be-tween collision and subduction. Geology 33, 857–860.

Wanless, V. D. & Shaw, A. M. (2012). Lower crustal crystalliza-tion and melt evolution at mid-ocean ridges. NatureGeoscience 5, 651–655.

Wanless, V. D., Perfit, M. R., Ridley, W. I. & Klein, E. (2010).Dacite petrogenesis on mid-ocean ridges: evidence for oce-anic crustal melting and assimilation. Journal of Petrology51, 2377–2410.

Wessel, P. & Smith, W. H. F. (1991). Free software helps mapand display data. EOS, Transactions, American GeophysicalUnion 72, 441.

Wessel, P. & Smith, W. H. F. (1998). New, improved version ofthe Generic Mapping Tools released. EOS, Transactions,American Geophysical Union 79, 579.

Wilson, A. D. (1960). The micro-determination of ferrous iron insilicate minerals by a volumetric and a colorimetric method.Analyst 85, 823–827.

Woodhead, J. (2002). A simple method for obtaining highlyaccurate Pb isotope data by MC-ICP-MS. Journal ofAnalytical Atomic Spectrometry 17, 1381–1385.

Woodhead, J. D., Volker, F. & McCulloch, M. T. (1995). Routinelead-isotope determinations using a Pb-207–Pb-204 doublespike—a long-term assessment of analytical precision andaccuracy. Analyst 120, 35–39.

Journal of Petrology, 2015, Vol. 0, No. 0 17

by guest on February 14, 2015http://petrology.oxfordjournals.org/

Dow

nloaded from