Earth and Planetary Science Letters 335-336 (2012) 180–194

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters

journal homepage: www.elsevier.com/locate/epsl

A variably enriched mantle wedge and contrasting melt types during arc
stages following subduction initiation in Fiji and Tonga, southwest Pacific
Erin Todd a,b,n, James B. Gill a, Julian A. Pearce c
a
Department of Earth and Planetary Sciences, University of California, Santa Cruz, USA
b
Institut für Mineralogie, WWU—University of Münster, Germany
c
School of Earth, Ocean and Planetary Sciences, Cardiff University, Cardiff CF10 3YE, UK

a r t i c l e i n f o a b s t r a c t

Article history: The earliest subaerially exposed magmatic products of the Fiji–Tonga–Kermadec (FTK) arc are preserved
Received 29 December 2011 in the Yavuna Group of Viti Levu, Fiji, and cobbles from ‘Eua, Tonga. They are similar in age and magma
Received in revised form types to the earliest rocks of the Izu–Bonin–Mariana (IBM) arc. In Fiji they include typical island arc
4 May 2012
tholeiitic (IAT), boninitic (BON), and MORB-like early arc tholeiitic (EAT) pillow lavas that are interpreted
Accepted 6 May 2012
as products of flux- and decompression-melting which occurred simultaneously during subduction
Editor: T.M. Harrison
Available online 15 June 2012 initiation. Although the oldest rocks in the southwest Pacific (FTK) and the northwest Pacific (IBM) arcs
are generally similar, they differ in two important respects. First, all magma types erupted simulta-
Keywords: neously in the SW Pacific whereas a similar assemblage may have erupted sequentially in IBM. Second,
subduction initiation
the primary mantle wedge was ‘‘Pacific’’ in isotopic character in the SW Pacific, but ‘‘Indian’’ in the NW
Fiji
Pacific.
Tonga
hafnium isotopes & 2012 Elsevier Ltd All rights reserved.
tholeiitic magma
boninitic magma

1. Introduction 2. Regional setting and geological history

Studies of Fijian volcanic rocks have contributed to the under-
standing of subduction-zone petrology and the evolution of
oceanic arc stages, including infancy (Gill, 1970, 1987; Wharton
et al., 1995), break-up (Gill and Whelan, 1989b; Gill, 1976a;
Hathway, 1993; Rodda et al., 1984), and post-arc (Gill, 1976b,
1984; Gill and Whelan, 1989a). These studies have additional
significance because subduction initiation may have happened
simultaneously throughout the western Pacific (e.g., Koppers
et al., 2004; Whittaker et al., 2007). New high-resolution geo-
chemical and geochronological evidence for the Izu–Bonin–Mari-
ana (IBM) arcs has led to a detailed model of magmatism during
and following subduction initiation (Ishizuka et al., 2006; Reagan
et al., 2008, 2010). This study focuses on the initial arc volcanism
in Fiji and Tonga and compares it to IBM. Similarities and
contrasts between the two arcs, formed at the same plate margin
at around the same time, provide new insights into the associa-
tions of lava types produced during subduction initiation and arc
infancy.
n
Corresponding author. Now at: U.S. Geological Survey, 4210 University Drive,
Anchorage, Alaska 99508, USA.
E-mail address: todd.erin@gmail.com (E. Todd).
The modern Fiji–Tonga–Kermadec (FTK) arc extends for
 2500 km from New Zealand to Tonga due to westward sub-
duction of the Pacific Plate beneath the Australian Plate (Fig. 1).
This paper focuses on its initial arc volcanism, with special
attention to rocks from southwest Viti Levu that have the most
extensive stratigraphic record in the SW Pacific. We assume that
Fiji and Tonga were part of the same arc at its inception that this
early arc extended from north of Australia to south of Tonga, and
that it developed on intra-oceanic crust due to inception of west-
dipping subduction of the Pacific Plate. These assumptions are
shared by most regional tectonic syntheses (Hall, 2002; Sdrolias
et al., 2003; Crawford et al., 2003; Schellart et al., 2006), but there
are relatively few constraints. It is clear, however, that no pre-
Cenozoic continental crust underlies Fiji (Drewes et al., 2009) and
that Eocene arc basement extends to at least 241S (ODP Site 841),
so Fiji and Tonga share the same history prior to opening of the
Lau Basin. This initial arc, prior to breaking into western (New
Britain to Vanuatu) and eastern (FTK) arms, is called the Vitiaz Arc
(Gill and Gorton, 1973). The crust on which it was constructed is
sometimes inferred to be the North Loyalty Basin (e.g., Crawford
et al., 2003). However, we consider that inference unlikely
because mantle from which the Vitiaz Arc was drawn is isotopi-
cally ‘Pacific’ (see below) whereas the one available sample
of North Loyalty Basin crust is ‘Indian’ (Pearce et al., 2007).

0012-821X/$ - see front matter & 2012 Elsevier Ltd All rights reserved.
http://dx.doi.org/10.1016/j.epsl.2012.05.006
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194 181

Fig. 1. Regional map showing Fiji, the Kermadec (KA) and Tonga (TA) arcs, the Havre Trough (HT), Lau Basin (LB), and South Fiji Basin backarcs, and the Colville (CR) and
Lau (LR) remnant arcs. Also shown are New Caledonia (NC), Vanuatu (VA), Norfolk Ridge (NR), Three Kings Rise (TKR), and North Loyalty Basin (NLB). The location of ‘Eua,
Tonga is indicated by the red triangle. The locations of samples from Viti Levu, Fiji are shown in greater detail in Fig. 2, within the region indicated by the dashed white box.
Locations of South Fiji Basin and Havre Trough backarc samples discussed in this study are shown by pink symbols (squares, BABB-like; circles, OIB-like) (Todd et al., 2011).
The solid black lines indicate spreading axes, whereas dashed black indicate backarc rifting. Also shown are the locations of Lau Basin ODP holes 834 and 839 and locations
of sediments sampled by drilling at DSDP 204 (black circles). DSDP holes 595/596 are located east of this field of view (  1651310 W). Figure was modified from Todd et al.
(2010). (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.)

Therefore, although the pre-arc plate configuration is unknown,
subduction may have initiated along an intra-Pacific fracture zone.
Subduction in both the northwestern and southwestern Pacific
has long been thought to have begun around the same time
because the oldest volcanic rocks in the IBM and Vitiaz arcs have
Middle to Upper Eocene Ar–Ar ages (Ewart et al., 1977; Cosca
et al., 1998) and are associated with limestones that have similar
Middle to Upper Eocene fauna (e.g., Cole, 1960; Milner, 1992). The
east side of ‘Eua island, Tonga, has the oldest exposed igneous
rocks of the Tonga Ridge (Ewart and Bryan, 1972). They include
cobbles of gabbro and volcanic rocks in a basal conglomerate,
plus overlying altered lavas and tuffs cut by dikes. The clasts are
embedded in a Middle Eocene shallow water limestone (Letter
Stage Ta3; Plankton Zone P.14; Tappin and Ballance, 1994) that is
overlain by Upper Eocene limestone (Letter Stage Tb) similar
to that described below from Fiji and lacking igneous clasts.
182 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194

The widely quoted 46–47 Ma Ar–Ar ages of Ewart et al. (1977) for
these rocks are from unpublished, and now-unavailable, total
fusion results for a basalt and a gabbro cobble from Lokupo (Liku)
Beach (A. Ewart, personal communication, 2009). The ages are
from the UC Berkeley lab in 1973 before the era of plateau and
isochrons, and are heuristic at best. However, a U/Pb zircon age
of 43.67 0.6 Ma has been obtained for the same gabbro clast
(K. Tani, personal communication, 2011). In addition, Duncan
et al. (1985) reported a 40–43 Ma Ar–Ar result for another basalt
cobble, and McDougall (1994) reported a 44 72 Ma result for
glass from ODP Site 841 in the Tonga Ridge basement that is
overlain by the same Upper Eocene fauna (Letter Stage Tb), noted
above (Chaproniere, 1994). Two samples from the basal conglom-
erate in this study (T08-10 and -22; Table A1) yielded 35.270.5
and 30.671.4 Ma Ar–Ar plateaux for plagioclase and bulk
groundmass, respectively (O. Ishizuka, personal communication,
2010), but we consider them to be minimum ages for the same
reasons. Consequently, the best estimate of the age of volcanic
and gabbroic rocks of the Tongan basement exposed on land is
 44 Ma. Tani (in preparation) reports a cluster of 52 Ma Lower
Eocene gabbros from the Tonga forearc, and Ishizuka et al. (2011)
report the same for IBM, so subduction of the Pacific Plate may
have nucleated beneath at least two different upper plates in the
western Pacific at about the same time, just prior to the bend in
Hawaii-Emperor Seamount Chain (Sharp and Clague, 2006).
The Upper Eocene rocks exposed on Viti Levu are part
of the Yavuna Group (Hathway, 1992; Hathway and Colley,
1994; Rodda, 1994) (Fig. 2), which includes pillow and massive
basalts, volcaniclastic rudites, mafic dikes, and a tonalitic stock.
Extrusive rocks are largely aphyric basalts to andesites with
some acicular plagioclase7pyroxene microphenocrysts in a fine-
grained altered matrix of plagioclaseþmagnetiteþpyroxenes7
chlorite, epidote, quartz, and calcite. There are no consistent
petrographic differences between the four rock types discussed
in this study. All have been metamorphosed under zeolite to
greenschist conditions. Near the Yavuna pluton, in upper Namosi
Creek, they are pervasively intruded by mafic to felsic dikes,
comprising up to 60% of the crustal section (Wharton, 1993;
Wharton et al., 1995). Yavuna Group volcanic rocks have late
Eocene minimum ages based on interbedded shallow water
limestones that contain Letter Stage Tb foraminifera similar
to the upper limestone unit on ’Eua, Tonga described above
(Bartholomew, 1960; Cole, 1960; Colley et al., 1986; Hathway,
1993; C.G. Adam, in Hathway and Colley, 1994). Ar–Ar and K–Ar
ages of Yavuna Group volcanic rocks are younger (25–34 Ma:
Whelan et al., 1985; Wharton, 1993; O. Ishizuka, personal

Fig. 2. Map of southwestern Viti Levu, Fiji. The early arc basement comprises the Yavuna Group (dashed ellipse) and encloses the Yavuna tonalite stock. Sample locations
are shown by black symbols—those in this study are shown with diamonds and those in previous studies (Wharton, 1993) by X’s. Inset shows the localized Yavuna Group
stratigraphic section exposed in Namosi Creek, modified from Wharton (1993) and supplemented with our new samples. Y -axis is elevation above sea level. Symbols
indicate melt types, as discussed in Section 4.1 and Fig. 3 caption.
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194
communication, 2010), but heating and alteration associated with
the pluton has likely reset Ar chronometers.
Silicic volcanic rocks are not known from the Yavuna Group,
though high-Si dacite dikes intrude its western exposures
(Wharton, 1993; Hathway, 1995) and, in at least one instance, cut
through Tertiary Letter Stage Tc (early Oligocene, 37–34 Ma) lime-
stone. These intrusives may be related to the 40 km2 trondhjemitic
Yavuna Pluton (Wharton, 1993; Hathway, 1995) that has been
dated by K–Ar and Ar–Ar (hornblende) (31.5–34.9 Ma; McDougall,
1963; Whelan et al., 1985) and U–Pb (zircon) (36.670.63 Ma,
Williams, 1991, in Wharton et al., 1995; confirmed by K. Tani and
C. Allen, personal communication, 2010).
Details of the early arc volcanic stratigraphy in Fiji and Tonga
remain sparse because subaerial exposures are limited. However,
recent submarine studies of the IBM arc have contributed to a
well-developed model of its earliest magmatic history. This
sequence, from subduction initiation to the mature-arc phase, is
similar to the record of some ophiolite sequences (Whattam and
Stern, 2011) and provides a testable hypothesis for early stages of
the Vitiaz Arc too. Volcanism associated with subduction initia-
tion in the IBM arc began with MORB-like basalts (52–51 Ma)
(Reagan et al., 2010; Ishizuka et al., 2011) and was characterized
by melt production rates much higher than in mature arcs. These
are termed ‘‘fore-arc basalts’’ (FAB) because that is where they
were found in the IBM arc. They are depleted in light rare earth
elements (LREE), consistent with fluid-poor decompression melt-
ing of lherzolite as mantle rises to fill space voided by the sinking
of Pacific Plate lithosphere (Gurnis et al., 2004; Reagan et al.,
2010). Later boninitic volcanism (48–44 Ma) (Cosca et al., 1998;
Ishizuka et al., 2006) is characterized by high MgO and SiO2, low
high field-strength element (HFSE) concentrations, variable
enrichments in large ion lithophile elements (LILE), and flat to
U-shaped REE patterns. Until recently, these were considered the
initial IBM lavas formed during ‘‘forearc spreading’’ across a zone
at least 300 km wide along the entire length of the IBM arc (Stern
and Bloomer, 1992; Bloomer et al., 1995). These features are
consistent with flux melting of harzburgitic mantle that was
depleted by the generation of earlier FAB volcanism. More typical
low- to medium-K arc tholeiitic to calcalkaline magmas appear by
45 Ma, continue to  27 Ma, and represent establishment of the
mature arc (Ishizuka et al., 2006; Reagan et al., 2008). This phase
was followed by arc quiescence and backarc spreading primarily
during  15225 Ma (Taylor, 1992; Ishizuka et al., 2009).
3. Samples, methods, and data
We present new major-element, trace-element, and/or isotope
data for 43 early arc samples from western Viti Levu, Fiji, and
‘Eua, Tonga. Most of them were collected in 2006 and 2008.
Others come from previous investigations (Wharton et al., 1995;
Gill, 1987). The Fijian rocks come from the Yavuna Group and
include both pillow and massive lavas from northern Masi,
Namosi, Nawaka, and Kawa Creeks (west to east; Fig. 2). Special
care was taken in the field to sample only the least altered clearly
extrusive lavas. We also present data for a new suite of cobbles
from ‘Eua, Tonga, collected at South Liku beach. In addition, fresh
clinopyroxene micro-phenocrysts from Fiji, plus one FAB from
Mariana, were analyzed from representative samples in order to
evaluate similarities between arcs and the degree to which Fiji
whole rock analyses were affected by low-grade metamorphism.
Sample preparation and analytical methods are presented in
Supplement A. Data are presented in supplemental Tables A1
(bulk rock) and A2 (clinopyroxene). All oxides (wt.%) are given an
anhydrous basis.
183
4. Major and trace element geochemistry
4.1. Chemical classification of the Yavuna group, Fiji
Few early FTK arc samples reflect primary mantle melts (i.e.,
Mg# 4 0:70, Cr 4 500 ppm, and Ni 4250 ppm). However, high
MgO ( 4 10:0 wt:%) in some samples indicate minimal fractiona-
tion because olivine phenocrysts are absent or minor. Our new
samples were the freshest available in the field, and were
collected with the advantage of knowing what others had found
previously at about the same location. Nevertheless, they are
altered and lack the fresh glass that is found occasionally in some
other early arcs. This alteration is evident in their groundmass
mineralogy, high LOI (predominantly 4 2 wt:%), and scatter on
conventional major element classification diagrams (not shown).
However, predicted melt compositions calculated using average
clinopyroxene trace-element concentrations and Dcpx=melt values
agree with whole-rock host compositions in the least-differen-
tiated rocks (Fig. A1), providing independent evidence that the
high-MgO lavas approximate liquid compositions. Because iden-
tifying primary magmatic traits presents inherent challenges for
these rocks, we used proxy diagrams based on immobile elements
for classification (Fig. 3).
As proxy for the IUGS-recommended total alkali-silica (TAS)
diagram, we use the Zr/Ti–Nb/Y figure of Winchester and Floyd
(1977), as updated and extended to all magma types by Pearce
(1996), in which Zr/Ti monitors the degree of differentiation
( C SiO2 ) and Nb/Y tracks enrichment ( C Salkalis ) (Fig. 3A). All
of our samples plot in the low Nb/Y (sub-alkaline) field (i.e.,
Nb=Y o 0:9) with a spread of Zr/Ti that places them within the
BADR (Basalt–Andesite–Dacite–Rhyolite) spectrum, with basalts
dominant (  2=3 of samples).
The tectonic affinity of basalts in Fig. 3A is then examined
using the Ti-V plot (Fig. 3B). Because V crystal-melt partitioning
during mantle melting varies as a function of f O2 , Shervais (1982)
notes that Ti/V can be used to distinguish between boninites
(BON), island arc tholeiites (IAT), MORB, and OIB tectonic settings.
Our samples range from nearly BON to MORB-like. Those in the
MORB field are bimodal, with one group extending to high-Ti
concentrations. The MORB-like group is accordingly subdivided
into low-Ti and high-Ti subgroups (1.29 wt.% and 2.11 wt.%
average TiO2, respectively).
Those near the lower edge of the IAT field (Ti/V¼10) may be
boninitic, a key category in the subduction initiation setting. This
designation is ambiguous using the main criterion of Si8 4 52
(Crawford, 1989; adapted by Pearce and Robinson, 2010 to
incorporate fractionation) because of the potential mobility of Si
and Mg in altered rocks. However, the second criterion (Ti8 o 0:5)
can be made alteration-insensitive by substituting Cr¼300 ppm
for MgO¼8 wt.%. The inset diagram shows the boninite–IAT
boundary for samples with 4 300 ppm Cr and Ti=V o20. A subset
clearly plots within the boninite field (Ti8 o 0:5); they are termed
‘boninites’ or ‘BON’ in the remainder of this paper.
Another important question is whether the samples plotting as
MORB on the Ti–Cr and Ti–V plots are true MORB or whether they
also contain a subduction component. This can be assessed using
a Th/Yb–Nb/Yb plot of Pearce (2008) in which about half of the
High-Ti group basalts lie within the field of MORB (Fig. 3C). All the
rest have some Th enrichment. IAT, as defined on the Ti–V
diagram, have as little Th enrichment as the active Tonga arc.
More formally, these samples are tholeiitic rather than calcalka-
line on a Th–Co proxy for the K2O–SiO2 diagram (Hastie et al.,
2007; inset of Fig. 3C), an inference supported by the incompatible
behavior of Ti and V in the Ti–V diagram. We accordingly refer to
these as IAT. We did not use Fe-enrichment as a criterion for
tholeiitic versus calcalkaline magmas because of their potential
184 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194

TiO2

.1

.2
.3

.5

2
.1

Cr (ppm)
yolite

BON
dacite/rh 1,000

IAT
10
.05 ,

/=
andesite 600
basaltic

V
Ti
.03 it e IAT 300
ande s
.02
boninite 20
Ti / V=

V (ppm)
Cent. MORB
Zr/Ti

.01 basalt 400 Tonga

.005 'Eua
200 Low-Ti EAT
.003
High-Ti EAT
.002 'Eua
IAT
Fiji Early Arc BON
.001 0
.10 .02 .03 .05 .1 .2 0 5,000 10,000 15,000 20,000
Nb/Y Ti (ppm)

3
2 1
CA

Th (ppm)
1
.5
IAT .1
.3 Cent.
Th/Yb

Tonga

BA/A
.2

D/R
B
B
.1 OR .01
M

20
60

40
.05
Co (ppm)
.03
.02
.01
.1 .2 .3 .5 1 2 3 5 10 20

Nb/Yb

Fig. 3. Diagrams used for the classification of early arc lavas from Fiji and ‘Eua, Tonga. Panel A shows Zr/Ti versus Nb/Y for all samples in this study, with lines distinguishing
between rock types, after Pearce (1996). The different symbols distinguish between samples from Fiji and ‘Eua. Data from this study are supplemented with previously
published Fiji data from Wharton (1993) and ‘Eua data from Cunningham and Anscombe (1985), Wharton (1993), Hergt and Woodhead (2007), Pearce et al. (2007). Panel B
shows Ti versus V for only those samples plotting within the basaltic field in Panel A, with diagonal lines indicating constant Ti/V ratios (¼ 10 and 20) corresponding to
discrimination boundaries between different tectonic settings (boninite, IAT, and MORB), after Shervais (1982). Gray lozenge shows the range of compositions for central
Tonga basalts, from Ewart et al. (1994b), Turner et al. (1997), Ewart et al. (1998), Hergt and Woodhead (2007), and Pearce et al. (2007). Inset shows TiO2 versus Cr for the least
fractionated samples (Cr 4 300), with vertical red line distinguishing boninite-like samples with TiO2 o 0:5 wt:%. For classification of boninitic traits, Cr¼ 300 ppm is
substituted for MgO¼ 8.0 wt.% because a linear regression of global MORB data (PetDB, 2010) gives Cr¼ MgO  80 344 (at MgO¼ 8.0 wt.%, Cr¼ 296 ppm). Panel C shows
Th/Yb versus Nb/Yb for basaltic samples with Ti=V 410 and TiO2 4 0:5 wt:%. Diagonal band shows the range of MORB compositions using data compiled from PetDB. Field for
Tonga as in Panel B. Inset shows these Fiji samples on a plot of Th versus Co. The shallower line distinguishes between island arc tholeiite (IAT) and calcalkaline (CA) rocks, and
the steeper lines distinguish between basalt (B), basaltic andesite/andesite (BA/A), and dacite/rhyolite (D/R), after Hastie et al. (2007). Different symbols are used in Panel B
and Panel C to show magma type classifications based on the rationale presented in the text: blue crosses, boninitic (BON); green diamonds, IAT; colored circles, low-Ti EAT;
black circles, high-Ti EAT. (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.)

modification by alteration. The low-Ti and high-Ti MORB-like
basalts from the Ti–V diagram are even less Th enriched and lie
between the IAT and the MORB array, in the same field as backarc
basin basalts (BABB) (Fig. 5). Therefore, the early Fijian tholeiites
could be termed ‘BABB’ geochemically. The ‘Eua samples that are
transitional between IAT and low-Ti MORB on the Ti–V diagram
plot within or slightly above the MORB field. Although different
from Fiji, they are again transitional between MORB and IAT and
could be called ‘BABB’ using these criteria despite their more
differentiated major element compositions.
However, the recent discovery of basalts from the IBM forearc
with MORB geochemical affinity and a small subduction influ-
ence has led to the term Forearc Basalts, or FAB by Reagan et al.
(2010) and Ishizuka et al. (2011). They are MORB from a REE-
HFSE perspective (Fig. 4) but not in Ti–V (Ishizuka et al., 2011;
Fig. 8) are intermediate between MORB and IAT in alteration-
resistant subduction characteristics (e.g., Th) occur in a fore-arc
rather than backarc setting, and are the earliest arc-related
rocks. Because no arc preceded our samples, we could refer to
the MORB-like group as low-Ti and high-Ti FAB rather than IAT,
MORB, or BABB even though they share some traits with all of
these categories. They have higher Ti than IAT (Fig. 3B) and their
LREE/HREE, Th (Fig. 3C), and Sr (see Discussion in Supplemental
Fig. A2) are close to MORB values. However, they differ from IBM
FAB in having higher HFSE concentrations and differently shaped
REE patterns (Fig. 4), both of which are more like MORB, and
they have negative Nb anomalies that are more like IAT than
MORB or IBM FAB. Using ‘‘FAB’’ also implies (falsely) that they
were erupted in a forearc setting (see Section 6) during the
earliest phase of arc initiation. Therefore, any of the possible
names (IAT, MORB, BABB, or FAB) has drawbacks. Because they
are found near the FTK arc axis, rather than its forearc, are
associated with shallow-water limestones, and extend from the
Middle to Upper Eocene, we call these lavas ’’early arc tholeiites’’
(EAT). On the basis of Fig. 3, we subdivide our samples into four
groups: boninites (BON), island arc tholeiites (IAT), low-Ti EAT,
and high-Ti EAT. Their comparative trace element systematics
are shown in Figs. 4 and 5. A more detailed comparison between
FTK EAT and IBM FAB will be made in a subsequent section. One
sample (T08-13) plots with low-Ti EAT in Fig. 3B, but is classified
as IAT because REE and Th enrichments suggest affinity with this
group (e.g., Figs. 4 and 3B).
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194 185

40
30
IAT 20
100

rock/CI
IAT
BON 10
BON
4
3
rock/N-MORB
2 IBM-BON
10
1

Nb
Th
La
Ce
Nd

Sm
Eu
Gd
Tb
Dy
Ho
Yb
Lu
Pr
Zr

Er
Hf
1

.1
Ba
Th
U
Nb
Ta
La
Ce
Pb

Nd
P

Sm
Eu
Gd
Tb
Ti
Dy
Ho
Y

Yb
Lu
Sr

Zr

Er
Hf
40 EAT (High-Ti)
30
100 Low-Ti EAT 20
rock/CI

High-Ti EAT 10
B
-FA
IBM EAT (Low-Ti)
4
3
rock/N-MORB

2
10
1

Gd
Sm
Ce

Eu

Yb
Ho
Dy
Nb

Nd
La
Th

Tb

Lu
Er
Pr
Zr
Hf

.1
Ba
Th
U
Nb
Ta
La
Ce
Pb

Nd
P

Sm
Eu
Gd
Tb
Ti
Dy
Ho
Y

Yb
Lu
Sr

Zr

Er
Hf

Fig. 4. Normalized trace element patterns for selected early arc Viti Levu rocks. Panels are divided into emphasize distinctions between more LREE-enriched and LREE-
depleted samples. Panel A shows boninitic and IAT Yavuna samples. Panel B shows EAT Yavuna samples (orange is high-Ti; black is low-Ti). The magnitude of positive Ba,
U, and Pb spikes, when present, could reflect metasomatism, but the differences between rock types seems independent of the degree of metamorphism. Black outlines in
insets of panels A and B show ranges of IBM-BON and IBM-FAB, respectively, from Pearce et al. (1999), Reagan et al. (2010), Ishizuka et al. (2011). Only Yavuna Group
samples with Cr 4100 ppm are shown to more clearly emphasize differences between least differentiated magmas. No ‘Eua samples meet this criterion, but generally they
have normalized trace element patterns similar to Viti Levu low-Ti EAT. Normalization values for N-MORB from Sun and McDonough (1989) and for CI chondrite from
Palme and O’Neill (2003). (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.)

4.2. Clinopyroxene geochemistry
In order to evaluate whether the differences between rock types
discussed above are primary or related to alteration, we compare
lavas with melts calculated from coexisting clinopyroxene micro-
phenocryst compositions for the Yavuna Group basement of Fiji
(see Supplement A). Most of their augites lie just below the
diopside–hedenbergite join on the pyroxene quadrilateral. Their
degree of Fe-enrichment varies with rock type. Fiji BON clinopyr-
oxenes have Mg# 0.90–0.80, similar to clinopyroxene from IBM –
FAB, though they extend to slightly lower Wo values. Clinopyrox-
enes from mafic low-Ti EAT and IAT samples have Mg# 0.85–0.75,
extending to 0.65 in more evolved rocks. Clinopyroxenes from
high-Ti EAT basalts have Mg# 0.80–0.70. Clinopyroxene trace-
element systematics differ between rock types. Mg#’s correlate
with Cr, Y, and TiO2, with the lowest Ti and Y concentrations in
clinopyroxenes from BON and the highest concentrations in high-
Ti EAT, just as in whole rocks. Bimodal Sr contents (  10 ppm and
 30 ppm) are not correlated with melt type or degree of differ-
entiation, but correlate positively with whole rock Sr contents
which, therefore, seem to be a magmatic feature despite the
alteration. In general, variability in whole rock major and trace-
element compositions is reflected in the mean composition of
hosted clinopyroxenes (Fig. A1), confirming that compositional
variability among magma types (BON, IAT, high- and low-Ti EAT)
reflects magmatic rather than diagenetic processes.
186 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194

10 50%
25%
6 “early” Lau BABB
rray
5 tle A
Man
Nd/Yb
4
3
Low-Ti EAT
2 High-Ti EAT
IAT
BON
'Eua
1

2 90% 80%
1 50%

Central
.3 rray
Tonga le A
Th/Yb

.2 nt
Ma
.1

.03 BON
.02 “early” Lau BABB IBM:
FAB
.01

90% 80%
400
300
200 50%
B
” La u BAB
100 “early
Sr/Yb

Array
Mantle
40
30
20

10
.7
.2

.4
.5
.6

7
.3

5
4

6
1
.1

10

Nb/Yb
Fig. 5. Ratios of variably fluid-mobile incompatible elements (X) as X/Yb ratios plotted versus Nb/Yb. (A) Nd/Yb, (B) Th/Yb, and (C) Sr/Yb. Dark-red lines delineate the approximate
center of the ‘mantle array’ of MORB, shown in pink, using data from PetDB (2010). Deviations above that array are attributed to slab-derived components; black lines indicate the
percent added by subduction, after Pearce et al. (1995). The differences between rock types in Sr/Yb in panel C are considered to be minimally affected by metasomatism because
Sr/Yb and Zr/Yb ratios in whole rocks approximate those calculated from clinopyroxene microphenocrysts (Fig. A2). Symbols and data sources for central Tonga as in Fig. 3, plus
Escrig et al. (2009). Field for ‘‘early’’ Lau BABB represents the MORB-like samples erupted early during Lau Basin extension at ODP Site 834 (Hergt and Farley, 1994; Ewart et al.,
1994a). Modern Lau Basin basalts plot mostly within the EPR N-MORB field. Dashed fields show the range of typical IBM FAB (black) and boninites (blue), from Reagan et al. (2010)
and Ishizuka et al. (2011). (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.)

4.3. ‘Eua, Tonga
Most of this paper deals with the Yavuna Group in Fiji.
However, the classification applies equally to the basement of
Eua, Tonga, that is transitional between IAT and low-Ti EAT. Our
data supplement and agree with other studies of ‘Eua by Ewart and
Bryan (1972), Hergt and Woodhead (2007), and Pearce et al.
(2007). Samples from all sources are basalt to andesite and gabbro.
All samples with o15 ppm Y from Ewart’s collection and ours are
gabbros, not BON, and are not discussed further because of effects
of crystal accumulation and uralitization. Despite being clasts in
conglomerate, the ‘Eua volcanics are more uniform than Yavuna
lavas. They are quite evolved (Cro 100 ppm, Ni r 10 ppm), so
classification based on the same criteria as for Fiji is more
speculative. However, they most resemble evolved Yavuna low-Ti
EAT, with even greater LREE-depletion. They have intermediate
LILE enrichment (e.g., Ba, Pb) relative to Yavuna EAT (more than in
high-Ti but less than in low-Ti). All three studies (Hergt and
Woodhead, 2007; Pearce et al., 2007; this study) found bimodal
Th/La ratios in ‘Eua; even the higher ratios (  0:08) are lower than
in the modern Tonga arc, and the lower ratios (  0:03) are similar
to N-MORB (Fig. 3C). Compared to modern Tonga, ‘Eua rocks are
more LREE-depleted and have lower LILE contents (Rb, Sr, Ba, Pb,
Th, U) at any MgO despite minor alteration. Their HFSE and HREE
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194 187

contents are similar to, or slightly higher than, in modern Tonga.
Taken together, these features suggest lower degree melts of flux
melting early in Tonga history than now. Hergt and Woodhead
(2007) report large negative Hf (7Zr) concentration anomalies for
‘Eua whereas they are absent in data from Pearce et al. (2007) and
this study and may be analytical artifacts.
5. Isotopes
5.1. Viti Levu, Fiji
Pb isotopes for most Yavuna Group samples plot on or near the
NHRL at higher 206Pb/204Pb than most IBM FAB (Fig. 6). Fiji EAT
are the most depleted; all others extend to 206Pb/204Pb  19:0
15.68
t
Sed gic
imen
Pela
with an average  18:85. Notably, Yavuna Pb isotopes overlie
regional BABB and OIB-like South Fiji Basin basalts (SFB-OIB)
(Todd et al., 2011), with little evidence of subducted pelagic or
terrigenous sediment or continental crust. On the plot of Nd
versus Hf isotopes (Fig. 7), most Yavuna Group samples lie below
the Pacific–Indian discrimination boundary of Pearce et al. (2007).
They define an array between depleted BABB-like and enriched
OIB-like magmas from SW Pacific backarc basins (Todd et al.,
2011). As in 206Pb/204Pb isotope space, their relative isotopic
enrichment is positively correlated with Ti and incompatible
element enrichment. Fiji BON and IAT samples lie at the more
enriched end of the array whereas EAT samples are isotopically
more depleted. Overall, therefore, EHf in Fiji early arc samples
correlates negatively with LREE/HREE ratios. 87Sr/86Sr ratios were
not measured in most samples because of alteration might
complicate interpretation. Nevertheless, 87Sr/86Sr extends down
Hikurangi to  0.7030 at high 206Pb/204Pb.
Plateau

15.64
5.2. ‘Eua, Tonga
207Pb/204Pb

15.60 Cent. Tonga

tic
as
icl t There are now three separate sample collections with isotope
l c an imen
15.56 Vo Sed
u data from the same conglomerate in ‘Eua. All agree for EHf but
ern La there is about 1ENd unit difference between Hergt and Woodhead
15.52 Mod
(2007) and Pearce et al. (2007) (higher in the former), even after
15.48 normalization to 0.511860 for La Jolla. Our results lie in between,
MORB “early”
(EPR)
Lau BABB and we replicated Pearce et al.’s ENd for two samples from Yavuna

7.6 8.0 8.4 8.8 9.2 9.6

39.0 14
18 Lau
dim gic
t
en

ern
Mod
Se ela
208Pb/204Pb

“Indian”
P

13
38.5 Low-Ti EAT
Cent. Tonga
12 MORB
High-Ti EAT (EPR)
au
ern L IAT 14
38.0 Mod L BON 11
R
NH 'Eua “early”
“Pacific” SFB/HT - BABB 10 Lau BABB
SFB - OIB
Volcaniclastic
10 Sediment
Hf

20
Modern Lau Cent. Tonga
+”early” Lau
ary

16
nd

Pelagic
ou

MORB
6
tB

(EPR) Volcaniclastic
12 Sediment
an

Sediment
rim
isc
Hf

8
D

tea ig

Low-Ti EAT
ian

Plakuran
u
nd

High-Ti EAT
c-I

4 2 IAT
cifi

Pelagic
Hi
Pa

Sediment Hikurangi BON

0 Plateau 'Eua
SFB/HT - BABB
-4 SFB - OIB
-2
18.2

18.4

18.6

18.8

19.0

19.2

19.4

206Pb/204Pb
Fig. 6. 206Pb/204Pb versus (A) 207Pb/204Pb, (B) 208Pb/204Pb, and (C) 176Hf/177Hf. Also
shown are previously published ‘Eua data from Pearce et al. (2007), Ewart and
Bryan (1972), and Hergt and Woodhead (2007). South Fiji Basin and Havre Trough
backarc samples discussed in this study are shown by pink squares (BABB-like)
and circles (OIB-like) (Todd et al., 2011). Fields for central Tonga lavas and Pacific
crust (igneous and sediment components) from Ewart et al. (1998), Pearce et al.
(1999), Hauff et al. (2003), Hergt and Woodhead (2007), Pearce et al. (2007),
Castillo et al. (2009), Chauvel et al. (2009) and Escrig et al. (2009). Symbols and
sources for average Louisville and Hikurangi seamount basalt are as in Fig. 5. NHRL
in A and B is from Hart (1984). (For interpretation of the references to color in this
figure caption, the reader is referred to the web version of this article.)
-6 -2 2 6 10
Nd
Fig. 7. 143Nd/144Nd versus 176Hf/177Hf. The presumed range of early FTK arc
magmas least modified by slab-derived Hf and Nd is shown by tan field. Symbols
and data sources as in Fig. 6, plus Turner et al. (1997). Red line is the mantle array
of Pearce et al. (2007), which discriminates between isotopically ‘‘Indian’’ and
‘‘Pacific’’ regimes. Black line is the trend of time-integrated recycled MORB
compositions with tick-marks for 0.5, 1.0 and 1.5 Ga, from Geldmacher et al.
(2011). Inset shows detail of early arc samples within the range indicated by the
dashed box, highlighting the contrasting ranges of more LREE-depleted EAT (red
field) and more LREE-enriched IAT and BON (blue field) samples. (For interpreta-
tion of the references to color in this figure caption, the reader is referred to the
web version of this article.)
188 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194
(Table A1). There is also a difference in Pb isotopes with, in
general, Pearce et al. (2007) o Hergt and Woodhead (2007) o this
study for 206Pb/204Pb. TIMS data in Pearce et al. (2007) have lower
207
Pb/204Pb than the other two that are by Tl-spiked MC-ICMPS.
Our ‘Eua data have 206Pb/204Pb similar to Yavuna but with slightly
higher 208Pb/204Pb. ‘Eua has higher ENd and EHf than even Yavuna
Hi–Ti EAT, consistent with ‘Eua being more depleted with greater
LREE-depletion. Pb in ‘Eua rocks is more radiogenic than in the
modern central Tongan arc, and is more like the Pb in the
northernmost Tongan arc. Specifically, it is similar to the recent
spiked analyses of Niuatoputapu (Regelous et al., 2010) and older
unspiked ones of Tafahi (Turner et al., 1997; Ewart et al., 1998;
Pearce et al., 2007). In Pb–Nd–Hf isotope space, ‘Eua is more
‘‘Pacific’’ than modern central Tonga, and resembles the oldest
Lau Basin BABB (see Discussion in Todd et al., 2011).
6. Discussion
6.1. Nature of slab-derived subduction components
Identifying slab-derived contributions in these rocks is com-
plicated by alteration and the diversity of magma types. However,
we have tried to apply the principles of Pearce et al. (1995).
Specifically, the background mantle is presumed to lie some-
where along the array of MORB and OIB in Fig. 5. If little Nb is
added from the slab, than the relative slab contributions of other
trace elements can be estimated from displacement above the
mantle arrays. Our clinopyroxene data show that whole rock Sr
concentrations are little affected by alteration and enriched only
in IAT and BON (Figs. 5, A1, and A2). Ba seems most enriched in
low-Ti EAT (Fig. 4). Nd is not enriched in Fiji BON or ‘Eua, and Th
is only modestly enriched in EAT and ‘Eua (Fig. 5A and B). Because
Sr and Th7Nd are most enriched in IAT whereas Ba is most
enriched in EAT, the former are most likely to reflect addition of a
more melt-like slab component versus the latter with a more
fluid-like component. Neither component seems to have carried
Hf. Therefore, slab surface temperatures during early arc history
were rarely high enough to melt subducted sediment beneath the
site of magmatism. Dilute fluids were the norm.
6.2. Nature of mantle components
Because there is little evidence of added Hf or, in most cases,
even Nd from slab components (Fig. 5A), and because early FTK
Hf–Nd isotope ratios completely overlap the ratios for regional
backarc basins (Fig. 7), we infer that the isotopic character of the
Hf and Nd in the early arc is enriched relative to MORB and is a
primary feature of the ambient mantle wedge. The enriched
mantle is most prominent in Fiji BON and IAT rocks in which
LREE-enrichment also is greatest. Therefore, we believe that the
mantle source early in arc history included old enrichments. The
same may even be true for Pb and Sr because Ce/Pb and Sr/Nd
ratios in some EAT approach values for MORB. The average
206
Pb/204Pb is  18:9, and the lowest 87Sr/86Sr is  0:7030.
Indeed, the samples with the most enriched Nd and Hf isotopes
have Pb isotopes below the NHRL at high 206Pb/204Pb (Fig. 6).
This general kind of isotopic enrichment has been found before
in Fiji and Tonga, especially in northernmost Tonga where it has
been attributed to subducted volcaniclastic sediment shed from
the Louisville Seamount Chain (e.g., Turner et al., 1997; Regelous
et al., 2010). However, we show here that isotopically similar
enriched mantle was present in the mantle wedge throughout the
early arc stage and is present in the regional backarc basins far
removed from current subduction (Todd et al., 2010, 2011).
Although it is possible that volcaniclastic sediment with just the
right isotopic ratios was universally subducted beneath the Vitiaz
Arc from inception to the present day and became well mixed
throughout the mantle wedge, it is simpler to infer that the mantle
wedge contains a persistent indigenous enriched component. We
will return to the tectonic significance of this inference later.
One of us has argued before for an old enriched mantle
component in the origin of Fijian magma on the basis of isotope
systematics (Gill, 1984), as have others elsewhere (e.g., Morris
and Hart, 1983; Reagan and Gill, 1989). This idea has not gained
traction because the arc rocks with isotopic enrichments usually
have HFSE depletions. For example, the early arc rocks considered
here have Nb/Yb ratios within the range of N-MORB and less than
in SFB-OIB (Fig. 5). Resolution of this topic is beyond the scope of
this paper, but we emphasize the long-lived character of this
isotopic feature in Fiji–Tonga and its presence in rocks with
minimal evidence of a slab melt component.
The specific combination of Sr–Nd–Hf–Pb isotopes in the
enriched Fijian component does not yet match any known OIB
signature in either the eastern Australian or western Pacific
Plates. The closest analogues are the Louisville Seamount Chain
(i.e., its volcaniclastic sediments; Pearce et al., 2007) or Ravaivai
Island in the Cook-Austral Seamount Chain (e.g., Lassiter et al.,
2003), both of which can be related to the South Pacific Super-
swell and to the large Cretaceous oceanic plateaux on the Pacific
Plate (e.g., Koppers et al., 2003; Hoernle et al., 2010). If similar
enrichments also have been present in the mantle wedge beneath
the arcs and backarc basins of the easternmost Indian Plate for the
last 50 Myr, then (a) subduction may have initiated along an
intra-Pacific fracture zone thereby explaining similar mantle on
both sides of the new plate boundary, and (b) seemingly ‘‘anom-
alous’’ isotopic enrichments in the FTK regime, such as in the
modern Quaternary North Tongan arc, may reflect old indigenous
mantle rather than recently subducted volcaniclastic sediment
(cf. Turner et al., 1997; Regelous et al., 2010).
6.3. Comparison of IBM and FTK early arc rock suites
Early IBM and FTK magmas have similar relative Hf–Nd
isotope, Ti/V, and Th/Nb ratio systematics (e.g., Fig. 8). Boninitic
magmas in both systems have the lowest Nd and Hf isotope ratios
and lowest Ti and HREE contents, whereas EAT/FAB have the
opposite. The differences in enrichment and depletion between
tholeiitic and boninitic rocks in the early IBM arc stages have been
interpreted as reflecting initial decompression-dominated melt-
ing of a source with higher HFSE and HREE contents to form
EAT/FAB, followed by flux-dominated melting of an isotopically
more enriched source with lower HFSE and HREE contents to
form the boninites (Reagan et al., 2010).
Although Fijian early arc rocks lack the fresh glass found by
diving on the IBM trench wall, we can compare the clinopyroxene
and whole rock compositions of the two early arcs. The simila-
rities between them are greater than the differences. Fijian-EAT
and IBM-FAB clinopyroxenes have similar Sr concentrations. The
Fijian clinopyroxenes have more MORB-like Ti and Y contents
relative to Mg# and Cr contents, as do bulk compositions. Sr in
Fijian BON and IAT clinopyroxene is bimodal, with the low-Sr
group overlapping EAT/FAB (Fig. A2). All rock types have bulk-
rock Nb/Yb ratios within the range of N-MORB and, in both arcs,
the EAT/FAB are similarly depleted in LREE and more depleted in
Nb than the boninites (Fig. 4). For bulk compositions, Fijian EAT
have higher Ti/V and MREE/HREE ratios, similarly low Ba, and
negative Nb anomalies. Boninitic rocks from both arcs have flat
REE patterns, high Mg and Cr contents, and similarly low Ti and
REE relative to Mg# or Cr. However, IBM boninites have positive
Zr–Hf anomalies whereas the Fijian ones do not (Fig. 4). Fewer
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194 189

Fig. 8. (A) Tectonic reconstruction of the Pacific region at  40 Ma (Middle Eocene), after Hall (2002), and (B) age-corrected E Hf(t) versus E Nd(t) for early arc lavas from
IBM (red) and FTK (black), assuming a 45 Ma age for Fijian and Tongan rocks. For both arcs, EAT/FAB lavas are shown with squares and BON (and IAT) lavas with diamonds.
Large arrows in A indicate the isotopic character (‘‘Indian’’ versus ‘‘Pacific’’) of the mantle wedge beneath the early Izu–Bonin and Vitiaz Arcs. The dashed gray line shows
our inferred boundary between ‘‘Indian’’ and ‘‘Pacific’’ mantle domains. Its location is poorly constrained due to few available samples from this time, though it was likely
at, or just east of, the NW Pacific plate margin because IBM lavas plot in ‘‘Indian’’ space in B. Its location in the SW Pacific is loosely constrained by \ 55 Ma isotopically
depleted ‘‘Indian’’ lavas from Cape Vogel, Papua New Guinea (‘‘PNG’’) (König et al., 2010) and North Loyalty Basin (Pearce et al., 2007), and ‘‘Pacific’’ magmas in the Solomon
Sea (thought to be Oligocene; see Woodhead et al., 1998, 2001) and Vitiaz Arc (this study). FTK data shown in B are from the same sources as in Fig. 7. IBM data are from
Pearce et al. (1999) and Reagan et al. (2010). (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.)

Fijian boninitic rocks have SiO2 455 wt:%. Both have Nd/Yb near
the mantle array but are enriched in Th (Fig. 5).
The chief difference between the arcs is that everything in FTK
is isotopically ‘‘Pacific’’ whereas, apart from a few FAB, the IBM is
‘‘Indian’’ (Fig. 8). Therefore, the mantle wedge of the two subduc-
tion zones differed even though the timing of subduction initia-
tion is similar, and the depleted and enriched components within
the two embryo mantle wedges also differed. That is, the modern
isotopic differences between northwestern and southwestern
Pacific arcs (Hickey-Vargas et al., 1995) reflect differences in
pre-existing mantle domains that were inherited at subduction
initiation (Fig. 8).
The other major difference between the two arcs is strati-
graphic. In Fiji, where one can sample the units on land, the
different rock types are inter-bedded (Fig. 2). Fig. 9 shows the
relative proportion of the four rock types in a 375-m-thick
composite section. The pattern demonstrates their contempor-
aneity and cyclicity. We also interpret the presence of both
boninitic (high-Sr,Cr) and EAT (low-Sr,Cr) clinopyroxenes in one
sample as evidence of mixing between these two magma types
and, therefore, their contemporaneity (Fig. A2). However, because
none of the lavas reported here could be reliably dated by Ar–Ar
or U–Pb (zircon) methods, we cannot determine when during
‘‘early arc’’ history they were erupted. Tholeiites both pre-date
and post-date boninites in IBM and distinguishing between the
tholeiites on purely geochemical grounds hinges on the magni-
tude of Nb and LILE anomalies (Reagan et al., 2010) the latter of
which can be modified by low grade metamorphism. Likewise,
both boninitic and tholeiitic rocks occur at 41–43 Ma in the IBM
and Mariana forearcs at ODP sites and on Saipan and Guam
(Reagan et al., 2010; Ishizuka et al., 2011). In both settings, they
are associated with sediments that have the same Middle Eocene
fauna that are associated with our samples from ‘Eua and that
overlie the rhyolite in Site 841 in Tonga (Tappin and Ballance,
1994; Chaproniere, 1994). Therefore, our ‘‘early arc’’ may extend
from the Upper to Middle Eocene and represent the end of
boninitic magmatism rather than the beginning (Fig. 9). Although
future discoveries from the Tonga Trench wall may show that our
EAT rock type extends even earlier, this paper demonstrates that
all magma types can be contemporaneous and, therefore, that
millions of years between them are not necessary.
6.4. Implications of Fijian early arc stratigraphy
Samples from the Fiji early arc figured in the original definition
of the somewhat MORB-like ‘‘Island Arc Tholeiitic Series’’ of Jakes̆
and Gill (1970). Gill (1970) argued that this low-K, Fe-enriched,
LREE-depleted group was succeeded in Fiji by medium-K, LREE-
enriched Upper Miocene calcalkaline rocks lacking Fe enrichment.
This interpretation contributed to the long-held idea that tholeii-
tic rocks make up the majority of arc basements and are followed
by calcalkaline volcanism (e.g., Jakes̆ and White, 1972; Gill, 1981).
Subsequent discovery of boninites below arc tholeiites in the IBM
arc and some ophiolites, and more recently of even more MORB-
like ‘‘Forearc Arc [tholeiitic] Basalts’’ below the boninites (Reagan
et al., 2010; Ishizuka et al., 2011) underscores the complexity of
temporal patterns, especially early in arc history.
Thus far, the Middle Eocene basement of Tonga is known only
from cobbles in the conglomerate on ‘Eua and basal rhyolite at
ODP Site 841. All are depleted and tholeiitic but none are as free
190 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194

Fiji IBM
325-375
IAT (n = 11)
(Ti/V 10 - 20)

BON 300-325
(Ti/V < 10) (n = 10)
arc tholeiites
Low-Ti EAT
(Ti/V > 10)
? and calc-
alkaline rocks
37-44 Ma

Hi-Ti EAT 225-300

(Ti/V > 10) (n = 9)

meters

high-Mg
175-225 andesite 44-45 Ma
(n = 7)

150-175
boninite andesite,
(n = 18)
differentiates 44-48 Ma
125-150
(n = 20)

Namosi Ck.
100-125
(n = 19)

75-100
(n = 15)
basalt (FAB) 50-52 Ma
40-75
(n = 17)

0% 50% 100%
% of strat. horizon

Fig. 9. (A) Stratigraphic distribution of different lava types for all early arc samples in this and previous studies and (B) generalized stratigraphic sequence for IBM forearc,
after Ishizuka et al. (2011). Note different units for Y axes in A and B, showing elevation and time, respectively. Column A shows proportions of magma types (IAT, BON,
high- and low-Ti EAT) within a given stratigraphic horizon, using 126 samples from this study plus Wharton (1993). The proportions are shown as normalized percentages
of samples ð ¼ nÞ. Stratigraphic arrangement of samples is after Wharton (1993) with the Y -axis representing column thickness in meters after removal of the effects of
faulting and later intrusion. Bracket shows range within the Namosi Creek sequence shown in inset of Fig. 2. Not all samples have high-resolution trace element data, so
samples with XRF data are classified using the Ti/V discriminant described in the text and Fig. 3 caption. Dashed lines show possible stratigraphic correlations between the
Fiji and IBM sequences. Our favored interpretation is that the base of the Fiji sequence is analogous to the late FAB stage of the IBM subduction initiation volcanic sequence
(lower arrow). The Fijian case implies contemporaneity of BON and EAT rather than a sequential change from one to another. We predict an older age for the Fiji section
because it is easier to imagine complicated mixing of flux- and decompression-melting regimes early, but the presence of shallow water forums in the Fiji section supports
thicker crust and a more mature arc.

of subduction components as IBM FAB. However, what we call
EAT and BON pillow lavas are interleaved in outcrop in the
Yavuna Group. Lacking reliable radiometric dates, we cannot say
when these magma types erupted or how much time is repre-
sented by the sequence.
Fig. 9 shows that EAT lavas are the most widely distributed
Yavuna Group lava type and comprise the majority of the exposed
Fijian basement. However, IAT and BON are present together, early
and late. The inset of Fig. 2 shows the stratigraphic relationships of
different Yavuna Group magma types within the most densely
sampled area, along Namosi Creek where a  53 m-thick vertical
section is exposed. All rock types are represented by pillow lavas
and associated volcaniclastic sediments, and are inter-bedded.
Therefore, all erupted at the same place within a short time interval
early in arc history. There is no systematic shift over time from one
rock type to another. This pattern is present throughout the Yavuna
Group (Fig. 9), with all types present in most stratigraphic horizons,
and low-Ti EAT and IAT alternatively dominant. Thus, while early
arc volcanism was primarily tholeiitic, there was rapid and repeated
alternation of magma types.
6.5. Significance of variable, yet closely associated magma types
The most MORB-like tholeiites (high-Ti EAT) in the Yavuna
Group cannot be pre-existing oceanic crust because there is no
evidence of thrusting or intrusion between the rock types. Nor can
the differences between rock types be attributed to open system
differentiation because the differences in high-Mg,Cr rocks are so
great. In principle, ubiquitous mixing of high-Ti EAT and BON
magmas could explain the more common rock types between them,
but most rocks are nearly aphyric and, anyway, this leaves the need
to produce the two extremes throughout the time period involved.
Therefore, we follow others including Reagan et al. (2010) and
Pearce and Robinson (2010) in attributing the differences between
early arc rock types to variations in mantle sources and melting
circumstances. For example, popular models for the formation of
boninite predict melting shallower than 50 km under high P–H2O and
temperature conditions of harzburgite that was depleted by prior
melting (Kushiro, 1974; Green, 1976; Kuroda et al., 1978). Although
few high-Mg,Cr Yavuna boninitic and transitional lavas have SiO2
4 55%, they have the other traits that support such inferences.
Conversely, it is commonly inferred that arc tholeiites with minimal
evidence of slab components reflect the same kind of decompression
melting as at mid-ocean ridges, and a similar range of mantle
potential temperatures and Mg#’s. More typical IAT with various
kinds of slab signatures reflect conditions between these extremes.
Our discovery that all magma types erupted at the same time
and place requires that these different conditions of magma
genesis coincided early in arc history. It is easy to imagine but
difficult to illustrate such a coincidence that might vary across-
strike, along-strike, and temporally during subduction initiation.
Fig. 10 is one heuristic example. By stressing close juxtaposition
of decompression and flux-melting, we consciously evoke a
model created to explain magma genesis during the initiation of
a backarc basin (Langmuir et al., 2006, Fig. 47). Its key features are
shallow flux melting closer to the trench, deeper decompression
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194 191

Lithospheric weakness
Crust

LM
AM

Zone of “forearc extension”

producing FAB lava

upwelling
fertile mantle

Future volcanic front Future forearc Legend:

active volcano slab-derived flux
decompression
extinct volcano melting regime
More depleted mantle;
interacting with FAB/EAT crust dry melts
slab-derived fluids
BON crust
hydrous melts

Fig. 10. Depiction of the generation of different magma types during subduction initiation and nascent arc phases, adapted from the subduction-zone infancy model of
Stern and Bloomer (1992). Panels to the left show the progression from (A) presubduction, to (B) early fore-arc extension, producing EAT/FAB lavas, and (C) the later
boninitic stage. See Section 2 for a description of the petrogenetic sequence proposed by Reagan et al. (2010) and Ishizuka et al. (2011) to explain these stages at Izu–
Bonin–Mariana. Note that we have drawn the range of boninitic volcanism to be smaller than that of earlier EAT/FAB, since BON was not found trenchward of FAB (Reagan
et al., 2010; Ishizuka et al., 2011). Panel D shows a schematic of our interpretation of the petrogenesis of Fijian magmas in this study, formed during the stage represented
in panel C (i.e., the BON stage) in the area within the dashed box. Contemporaneous decompression-melting (to the left) and flux-melting (to the right) indicated by the
wavy blue lines. Both types of melt are focused to a narrow eruption site by the structure of overlying lithosphere, indicated by the dashed brown triangle. The focusing
resembles the trench-perpendicular juxtaposition of ‘‘drier’’ and ‘‘wetter’’ magmas that was proposed to explain the range of melt types in near-arc backarc-basin settings
(Langmuir et al., 2006). These authors interpret transitional basalts as resulting from focusing of drier and wetter melts, consistent with our observation that at least one
sample has bimodal clinopyroxene populations, with high- and low-Sr contents (Fig. A2). The occurrence of both decompression- and flux-melt types in the same place and
time at Fiji, but not in the IBM fore-arc, may be explained by Fiji’s position farther from the trench. The simultaneous eruption of different types at this time might not
occur closer to the trench because the forearc zone is dominated by magmas produced by flux melting (i.e., BON). The proximity of flux melting to the trench is limited by a
critical threshold of slab dehydration (i.e., the crust is composed of predominantly older FAB closest to the trench because there is not enough slab-derived fluid released
there to produce BON). Therefore, the contrasts observed at IBM by Reagan et al. (2010) and Ishizuka et al. (2011) reflect the progression in time (earlier FAB to later BON)
and space (within the bracket labeled ‘‘Future forearc’’ in panel (C), whereas the contrasts we observe at Fiji occur at the same time and place (i.e., within the bracket
labeled ‘‘Future volcanic front’’ in panel (C).

melting farther away, advection of increasingly depleted mantle
through the melting zone, and focusing and mixing of melts
beneath young lithosphere under tension. Such a dynamical
system during subduction initiation was suggested by Hall and
Spakman (2003). It is more consistent with what we describe here
than a more gradual evolution of magma types over millions of
years (Reagan et al., 2010; Ishizuka et al., 2011), but the two need
not be mutually exclusive. Indeed, the two may differ more in
tectonic setting than age, with Fiji and Eua lying closer to the
eventual mature volcanic arc versus the IBM sites lying closer to
the plate boundary (Fig. 10). In addition, each subduction initia-
tion might be different (e.g., Whattam and Stern, 2011).
6.6. Isotopic components of the early arc mantle
In this section we focus on Hf and Nd because they show little
or no evidence of slab-derived enrichments. IBM and FTK magmas
of similar arc stages occupy similar relative positions along Hf–Nd
isotope arrays (Fig. 7; Reagan et al., 2010, Fig. 6). Boninitic magmas
in both systems tend to have the lowest isotopic compositions. The
IBM array was explained by Pearce et al. (1999) as due to adding
a slab-derived component from volcaniclastic sediment and/or
Pacific Plate altered oceanic crust to depleted ‘‘Indian’’ mantle.
The most isotopically depleted arc and backarc rocks in the SW
Pacific lie below the Mantle Array, so FTK mantle source is more
‘‘Pacific’’ than in IBM. Mixing depleted Pacific mantle with the
same slab component as in IBM cannot account for FTK composi-
tions because their EHf lie below such a mixing array.
The Louisville Seamount Chain, the potential source of OIB-like
volcaniclastic sediment added to the mantle wedge of the modern
FTK arc by subduction, is a possible source of unradiogenic Nd
and Hf (and radiogenic Pb) that could explain SW Pacific basalt
arrays. However, a similar enriched component is variably pre-
sent in arc rocks from 4 40 Ma, and in subduction unmodified
backarc basin rocks from 33 to o1:0 Ma beneath which the
Louisville Chain cannot have subducted (Todd et al., 2010, 2011).
Therefore, it is difficult to imagine a scenario whereby this source
of volcaniclastic sediment can contribute a continual supply of
volcanogenic sediment to the entire arc and backarc from the
Eocene to present. In addition, the wide range of EHf at relatively
constant ENd does not correlate with the magnitude of Hf
concentration anomalies in FTK early arc rocks, as it should if
explained by recent the addition of sediment or melt thereof.
Therefore, we contend that the similarity in Hf–Nd isotopes
between Neogene SW Pacific backarc samples without slab-derived
material on the one hand (Todd et al., 2010, 2011), and the FTK
basement samples of this paper on the other, reflects the composi-
tional diversity of the ambient mantle wedge throughout this 50 Myr.
period. One implication of this idea is that the anomalous isotopes of
the modern northern Tonga arc reflect the presence of trapped old
mantle lithosphere rather than delayed melting of recently subducted
Louisville sediment.
192 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194
Because the more boninitic rocks in both the IBM and FTK arcs
have lower Hf and Nd isotope ratios, they may be derived from
old mantle enrichments present at the initiation of subduction
(Reagan et al., 2010). Based on our experience studying how
similar enriched components melt in the adjacent backarc basins
(Todd et al., 2011), we infer that FTK mantle enrichments are
pyroxenitic veins with lower solidi than their peridotitic hosts.
The difference between the two environments is that enrich-
ments in the backarc occur in HFSE-rich basalts (OIB) whereas in
the early arc they occur in HFSE-poor boninitic rocks and IAT. If
decompression-dominated melting of enriched mantle compo-
nents leaves harzburgite residues that retain their original iso-
topic composition (as proposed for boninites in the NE Lau Basin
by Falloon et al., 2007), then subsequent melting of that residue
fluxed by shallow slab-derived fluid poor in Hf and Nd might
produce the contrasts observed.
7. Conclusions
New major element, trace element, and Sr–Nd–Pb–Hf isotope
data are presented for lavas and their associated clinopyroxenes
that erupted soon after subduction initiation in the Fiji–Tonga–
Kermadec (FTK) arc. They define four rock types that are not
related by differentiation of a common parent. They are typical
island arc tholeiitic (IAT), boninitic (BON), and early arc tholeiitic
(EAT) pillow lavas that are interbedded in outcrop. They are
interpreted as products of flux- and decompression-melting that
occurred simultaneously during subduction initiation (Fig. 10).
How long after initial slab foundering is unknown because we
could not date the Fijian lavas, but some from ‘Eua, Tonga, are at
least  44 Ma. The Fijian rocks may reflect the dynamical transi-
tion from slab foundering (characterized mostly by EAT) to true
slab-parallel subduction, the start of more typical arc volcanism
(characterized by IAT and BON), and the true root of the Vitiaz arc.
The rock types span a range of REE and isotopic enrichment
that overlaps the range in Neogene basalts from the adjacent
backarc basins. The boninitic rocks have the greatest LREE and
Hf–Nd–Pb isotopic enrichment whereas EAT are the opposite. We
attribute this diversity to variations within the primary mantle
wedge prior to subduction initiation. We argue that the slab
components during subduction initiation were dilute fluids rather
than sediment melts.
Although the magmatic products accompanying subduction
initiation in the SW Pacific (FTK) and the northwest Pacific (IBM)
arcs are generally similar, they differ in two important respects.
First, all four rocks erupted simultaneously at an unknown time in
the SW Pacific whereas they have been interpreted as erupting
sequentially in IBM. Second, the primary mantle wedge was
‘‘Pacific’’ in isotopic character in the SW Pacific, but ‘‘Indian’’ in
the NW Pacific. A fundamental mantle domain boundary sepa-
rated the two sites of roughly contemporaneous subduction
initiation (Fig. 8).
Acknowledgments
This work was funded by NSF MARGINS grant OCE0751600.
Additional funding came from small grants in 2006 from the UCSC
Center for the Study of Imaging and Dynamics of the Earth (CSIDE)
and the Geological Society of America. We thank O. Ishizuka,
K. Tani, and C. Allen for allowing us to present their unpublished
and challenging Ar–Ar and U–Pb age results. Six Yavuna Group
samples are from the 1993 Ph.D. Thesis of M. Wharton, Durham
University. New unpublished trace element and isotope data for
these samples are from J. A. Pearce (Cardiff) and P. Kempton
(NIGL). One IBM FAB sample (1093R6) was donated by M. Reagan
for clinopyroxene analysis. We thank two anonymous reviewers
for comments that helped to improve the paper. This work greatly
benefitted from the technical assistance of C. Harris, D. Sampson,
R. Franks, D. Tollstrup, and E. Peterman at UCSC, and F. Tepley,
D. Burns, and A. Koleszar at OSU. It also celebrates JBG’s return to
the topic 44 yr after beginning work on it.
Appendix A. Supplementary materials
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org.10.1016/j.epsl.2012.05.006.
References
Bartholomew, R.W., 1960. Geology of the Nandi Area, Western Viti Levu. Bulletin—
Geological Survey of Fiji 7, 27 pp.
Bloomer, S.H., Taylor, B., MacLeod, C.J., Stern, R.J., Fryer, P., Hawkins, J.W., Johnson,
L., 1995. Early arc volcanism and the ophiolite problem; a perspective from
drilling in the western Pacific. In: Taylor, B., Natland, J. (Eds.), Active Margins
and Marginal Basins of the Western Pacific, Geophysical Monograph, vol. 88.
American Geophysical Union, Washington, DC, pp. 1–30.
Castillo, P.R., Lonsdale, P.F., Moran, C.L., Hawkins, J.W., 2009. Geochemistry of mid-
Cretaceous Pacific crust being subducted along the Tonga–Kermadec Trench:
implications for the generation of arc lavas. Lithos 112, 87–102.
Chaproniere, G., 1994. Middle and late Eocene larger foraminifers from Site 841
(Tongan Platform). Proc. Ocean Drilling Program Sci. Results 135, 231–243.
Chauvel, C., Marini, J.-C., Plank, T., Ludden, J.N., 2009. Hf–Nd input flux in the Izu–
Mariana subduction zone and recycling of subducted material in the mantle.
Geochem. Geophys. Geosyst. 10, 23.
Cole, W.S., 1960. Upper Eocene and Oligocene Larger Foraminifera from Viti Levu,
Fiji. U.S. Geological Survey Professional Paper P 0374-A, pp. 1–7.
Colley, H., Hindle, W.H., Hathway, B., 1986. Early arc and basinal activity on Viti
Levu, Fiji. In: International Volcanological Congress, New Zealand, 1–9 Feb-
ruary 1986, abstracts, p. 140.
Cosca, M., Arculus, R., Pearce, J., Mitchell, J., 1998. 40Ar/39Ar and K–Ar geochrono-
logical age constraints for the inception and early evolution of the Izu–Bonin–
Mariana arc system. Isl. Arc 7 (3), 579–595.
Crawford, A.J., 1989. Classification petrogenesis and tectonic setting of boninites.
In: Crawford, A.J. (Ed.), Boninitesa. Unwin Hyman, London, pp. 1–49.
Crawford, A.J., Meffre, S., Symonds, P.A., 2003. 120 to 0 Ma tectonic evolution of
the Southwest Pacific and analogous geological evolution of the 600 to 220 Ma
Tasman fold belt system. In: Hills, R.R., Mueller, R.D. (Eds.), Evolution and
Dynamics of the Australian Plate, vol. 372. Geological Society of America
Special Paper, Boulder, pp. 383–403.
Cunningham, J.K., Anscombe, K.J., 1985. Geology of ’Eua and other Islands, King-
dom of Tonga. In: Scholl, D.W., Vallier, T.L. (Eds.), Geology and Offshore
Resources of Pacific Island Arcs; Tonga Region, vol. 2. Circum-Pac. Counc.
Energy and Miner. Resour., Houston, pp. 221–257.
Drewes, E., Gill, J.B., Allen, C.M., Campbell, I.H., Todd, E., 2009. Fijian tonalites and
crustal evolution. In: Eos: Transactions of the American Geophysical Union,
vol. 90(52), Fall Meeting Supplement, Abstract T21A-1787.
Duncan, R.A., Vallier, T.L., Falvey, D.A., 1985. Volcanic episodes at ’Eua, Tonga
Islands. In: Scholl, D.W., Vallier, T.L. (Eds.), Geology and Offshore Resources
of Pacific Island Arcs; Tonga Region, vol. 2. Circum-Pac. Counc. Energy and
Miner. Resour., Houston, pp. 281–290.
Escrig, S., Bézos, A., Goldstein, S.L., Langmuir, C.H.H., Michael, P.J.J., 2009. Mantle
source variations beneath the Eastern Lau Spreading Center and the nature of
subduction components in the Lau basin–Tonga arc system. Geochem.
Geophys. Geosyst. 10 (4), 33.
Ewart, A., Brothers, R.N., Mateen, A., 1977. An outline of the geology and
geochemistry, and the possible petrogenetic evolution of the volcanic rocks
of the Tonga–Kermadec–New Zealand island arc. J. Volcanol. Geotherm. Res. 2
(3), 205–250.
Ewart, A., Bryan, W.B., 1972. Petrography and geochemistry of the igneous rocks
from Eua, Tongan islands. Geol. Soc. Am. Bull. 83 (11), 3281–3298.
Ewart, A., Bryan, W.B., Chappell, B.W., Rudnick, R.L., 1994a. Regional geochemistry
of the Lau–Tonga arc and backarc systems. Proc. Ocean Drill Program Sci.
Results 135, 385–425.
Ewart, A., Collerson, K.D., Regelous, M., Wendt, J.I., Niu, Y., 1998. Geochemical
evolution within the Tonga–Kermadec–Lau arc-back-arc systems; the role of
varying mantle wedge composition in space and time. J. Petrol. 39 (3), 331–368.
Ewart, A., Hergt, J.M., Hawkins, J.W., 1994b. Major element, trace element, and
isotope (Pb, Sr, and Nd) geochemistry of Site 839 basalts and basaltic
andesites; implications for arc volcanism. Proc. Ocean Drilling Program Sci.
Results 135, 519–531.
Falloon, T.J., Danyushevsky, L., Crawford, A.J., Maas, R., Woodhead, J.D., Eggins, S.,
Bloomer, S., Wright, D.J., Zlobin, S.K., Stacey, A.R., 2007. Multiple mantle plume
components involved in the petrogenesis of subduction-related lavas from the
 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194
northern termination of the Tonga Arc and northern Lau Basin: evidence from
the geochemistry of arc and backarc submarine volcanics. Geochem. Geophys.
Geosyst. 8 (Q09003), 45.
Geldmacher, J., Hoernle, K., Hanan, B., Blichert-Toft, J., Hauff, F., Gill, J., Schmincke,
H., 2011. Hafnium isotopic variations in East Atlantic intraplate volcanism.
Contrib. Mineral. Petrol. 162 (1), 21–36.
Gill, J., Gorton, M., 1973. A proposed geological and geochemical history of eastern
Melanesia. In: Coleman, P.J. (Ed.), The Western Pacific: Island Arcs, Marginal
Seas, Geochemistry. University of Western Australia Press, pp. 543–566.
Gill, J.B., 1970. Geochemistry of Viti Levu, Fiji, and its evolution as an island arc.
Contrib. Mineral. Petrol. 27 (3), 179–203.
Gill, J.B., 1976a. Composition and age of Lau basin and ridge volcanic rocks;
implications for evolution of an interarc basin and remnant arc. Geol. Soc. Am.
Bull. 87 (10), 1384–1395.
Gill, J.B., 1976b. From island arc to oceanic islands; Fiji, southwestern Pacific.
Geology 4 (2), 123–126.
Gill, J.B., 1981. Orogenic Andesites and Plate Tectonics. Springer Verlag, Berlin.
Gill, J.B., 1984. Sr–Pb–Nd isotopic evidence that both MORB and OIB sources
contribute to oceanic island arc magmas in Fiji. Earth Planet. Sci. Lett. 68 (3),
443–458.
Gill, J.B., 1987. Early geochemical evolution of an oceanic island arc and backarc;
Fiji and the South Fiji Basin. J. Geol. 95 (5), 589–615.
Gill, J.B., Whelan, P., 1989a. Early rifting of an oceanic island arc (Fiji) produced
shoshonitic to tholeiitic basalts. J. Geophys. Res. B Solid Earth Planet. 94,
4561–4578.
Gill, J.B., Whelan, P., 1989b. Postsubduction ocean island alkali basalts in Fiji.
J. Geophys. Res. B Solid Earth Planet. 94, 4579–4588.
Green, D.H., 1976. Experimental testing of ‘‘equilibrium’’ partial melting of peridotite
under water-saturated, high-pressure conditions. Can. Mineral. 14, 255–268.
Gurnis, M., Hall, C., Lavier, L., 2004. Evolving force balance during incipient
subduction. Geochem. Geophys. Geosyst. 5 (7), 31.
Hall, R., 2002. Cenozoic geological and plate tectonic evolution of SE Asia and the
SW Pacific; computer-based reconstructions, model and animations. J. Asian
Earth Sci. 20 (4), 353–431.
Hall, R., Spakman, W., 2003. Mantle structure and tectonic evolution of the region
north and east of Australia. In: Hills, R.R., Mueller, R.D. (Eds.), Evolution and
dynamics of the Australian plate, vol. 372. Geological Society of America
Special Paper, pp. 361–381.
Hart, S.R., 1984. A large-scale isotope anomaly in the Southern Hemisphere
mantle. Nature 309 (5971), 753–757.
Hastie, A.R., Kerr, A.C., Pearce, J.A., Mitchell, S.F., 2007. Classification of altered
volcanic island arc rocks using immobile trace elements: development of the
Th–Co discrimination diagram. J. Petrol. 48 (12), 2341–2357.
Hathway, B., 1992. Definitions of formations in southwestern Viti Levu. In:
Rahiman, A. (Ed.), Parliament of Fiji, Paper 15 of 1991. Mineral Resources
Department: Report for the Year 1988, pp. 83–89.
Hathway, B., 1993. The Nadi Basin; Neogene strike-slip faulting and sedimentation in a
fragmented arc, western Viti Levu, Fiji. J. Geol. Soc. London 150 (3), 563–581.
Hathway, B., 1995. Deposition and diagenesis of Miocene arc-fringing platform
and debris-apron carbonates, southwestern Viti Levu, Fiji. Sediment. Geol. 94
(3–4), 187–208.
Hathway, B., Colley, H., 1994. Eocene to Miocene geology of Southwest Viti Levu,
Fiji. In: Stevenson, A.J., Herzer, R.H., Ballance, P.F. (Eds.), Geology and
Submarine Resources of the Tonga–Lau–Fiji Region. SOPAC Technical Bulletin,
vol. 8, pp. 153–169.
Hauff, F., Hoernle, K., Schmidt, A., 2003. Sr–Nd–Pb composition of Mesozoic Pacific
oceanic crust (Site 1149 and 801, ODP Leg 185); implications for alteration of
ocean crust and the input into the Izu–Bonin–Mariana subduction system.
Geochem. Geophys. Geosyst. 4 (8), 30.
Hergt, J.M., Farley, K.N., 1994. Major element, trace element, and isotope (Pb, Sr,
and Nd) variations in Site 834 basalts; implications for the initiation of backarc
opening. Proc. Ocean Drilling Program Sci. Results 135, 471–485.
Hergt, J.M., Woodhead, J.D., 2007. A critical evaluation of recent models for Lau-
Tonga arc-backarc basin magmatic evolution. Chem. Geol. 245 (1–2), 9–44.
Hickey-Vargas, R., Hergt, J.M., Spadea, P., 1995. The Indian Ocean-type isotopic
signature in western Pacific marginal basins; origin and significance. In:
Taylor, B., Natland, J. (Eds.), Active Margins and Marginal Basins of the Western
Pacific, Geophysical Monograph, vol. 88. American Geophysical Union,
Washington, DC, pp. 175–197.
Hoernle, K., Hauff, F., Werner, R., van den Bogaard, P., Mortimer, N., Geldmacher, J.,
Garbe-Schönberg, D., Davy, B., 2010. Age and Geochemistry of Volcanic Rocks
from the Hikurangi and Manihiki Oceanic Plateaus. Geochim. Cosmochim. Acta
74 (24), 7196–7219.
Ishizuka, O., Kimura, J.-I., Li, Y.B., Stern, R.J., Reagan, M.K., Taylor, R.N., Ohara, Y.,
Bloomer, S.H., Ishii, T., Hargrove Iii, U.S., Haraguchi, S., 2006. Early stages in the
evolution of Izu–Bonin arc volcanism: new age, chemical, and isotopic
constraints. Earth Planet. Sci. Lett. 250 (1–2), 385–401.
Ishizuka, O., Tani, K., Reagan, M., Kanayama, K., Umino, S., Harigane, Y., Sakamoto,
I., Miyajima, Y., Yuasa, M., Dunkley, D., 2011. The timescales of subduction
initiation and subsequent evolution of an oceanic island arc. Earth Planet. Sci.
Lett. 306 (3–4), 229–240.
Ishizuka, O., Yuasa, M., Taylor, R.N., Sakamoto, I., 2009. Two contrasting magmatic
types coexist after the cessation of back-arc spreading. Chem. Geol. 266 (3–4),
274–296.
Jakes̆, P., Gill, J., 1970. Rare earth elements and the island arc tholeiitic series. Earth
Planet. Sci. Lett. 9, 17–28.
193
Jakes̆, P., White, A., 1972. Major and trace element abundances in volcanic rocks of
orogenic areas. Bull. Geol. Soc. Am. 83 (1), 29–40.
König, S., Münker, C., Schuth, S., Luguet, A., Hoffmann, J.E., Kuduon, J., 2010.
Boninites as windows into trace element mobility in subduction zones.
Geochim. Cosmochim. Acta 74 (2), 684–704.
Koppers, A.A.P., Duncan, R.A., Steinberger, B., 2004. Implications of a nonlinear
(super 40) Ar/(super 39) Ar age progression along the Louisville seamount trail
for models of fixed and moving hot spots. Geochem. Geophys. Geosyst. 5 (6), 22.
Koppers, A.A.P., Staudigel, H., Pringle, M., Wijbrans, J.R., 2003. Short-lived and
discontinuous intraplate volcanism in the South Pacific: hot spots or exten-
sional volcanism? Geochem. Geophys. Geosyst. 4 (10), 49.
Kuroda, N., Shiraki, K., Urano, H., 1978. Boninites as a possible calc-alkaline
primary magma. Bull. Volcanol. 41, 563–575.
Kushiro, I., 1974. Melting of hydrous upper mantle and possible generation of
andesitic magma: an approach from synthetic systems. Earth Planet. Sci. Lett.
22 (4), 294–299.
Langmuir, C.H., Bezos, S.E., Parman, S.W., 2006. Chemical systematics and hydrous
melting of the mantle in back-arc basins. In: Christie, D., Fisher, C.R., Lee, S.-M.,
Givens, S. (Eds.), Back-Arc Spreading Systems: Geological, Biological, Chemical,
and Physical Interactions, Geophysical Monograph Series, vol. 166. American
Geophysical Union, Washington, DC, pp. 87–146.
Lassiter, J.C., Blichert-Toft, J., Hauri, E.H., Barsczus, H.G., 2003. Isotope and trace
element variations in lavas from Raivavae and Rapa, Cook-Austral islands:
constraints on the nature of HIMU- and EM-mantle and the origin of mid-plate
volcanism in French Polynesia. Chem. Geol. 202 (1–2), 115–138.
McDougall, I., 1963. Potassium–argon ages of some rocks from Viti Levu, Fiji.
Nature 198 (4881). 677–677.
McDougall, I., 1994. Data report; dating of rhyolitic glass in the Tonga forearc
(Hole 841B). In: Hawkins, J., Parson, L., Allan, J. (Eds.), Proceedings of the Ocean
Drilling Program Scientific Results, Lau Basin, vol. 135. Texas A & M University,
Ocean Drilling Program, College Station, TX, pp. 923.
Milner, G.J., 1992. Middle eocene to early oligocene foraminifers from the Izu–
Bonin forearc, hole 786a. In: Fryer, P., Pearce, J.A., Stokking, L.B. (Eds.),
Proceedings of the Ocean Drilling Program Scientific Results, vol. 125. Texas
A & M University, Ocean Drilling Program, pp. 71–90.
Morris, J., Hart, S., 1983. Isotopic and incompatible element constraints on the
genesis of island arc volcanics from Cold Bay and Amak Island, Aleutians, and
implications for mantle structure. Geochim. Cosmochim. Acta 47 (11),
2015–2030.
Palme, H., O’Neill, H.S.C., 200,. Cosmochemical estimates of mantle composition.
In: Treatise on Geochemistry. Pergamon, Oxford, pp. 1–38.
Pearce, J.A., 1996. A user’s guide to basalt discrimination diagrams. In: Wyman,
D.A. (Ed.), Trace Element Geochemistry of Volcanic Rocks: Applications for
Massive Sulphide Exploration, vol. 12. Geological Association of Canada, Short
Course Notes, pp. 79–113.
Pearce, J.A., 2008. Geochemical fingerprinting of oceanic basalts with applications
to ophiolite classification and the search for Archean oceanic crust. Lithos 100
(1–4), 14–48.
Pearce, J.A., Baker, P.E., Harvey, P.K., Luff, I.W., 1995. Geochemical evidence for
subduction fluxes, mantle melting and fractional crystallization beneath the
South Sandwich island arc. J. Petrol. 36 (4), 1073–1109.
Pearce, J.A., Kempton, P.D., Gill, J.B., 2007. Hf–Nd evidence for the origin and
distribution of mantle domains in the SW Pacific. Earth Planet. Sci. Lett. 260
(1–2), 98–114.
Pearce, J.A., Kempton, P.D., Nowell, G.M., Noble, S.R., 1999. Hf–Nd element and
isotope perspective on the nature and provenance of mantle and subduction
components in Western Pacific arc-basin systems. J. Petrol. 40 (11),
1579–1611.
Pearce, J.A., Robinson, P.T., 2010. The Troodos ophiolitic complex probably formed
in a subduction initiation, slab edge setting. Gondwana Res. 18 (1), 60–81.
PetDB, 2010. Petrological Database of the Ocean Floor /http://www.petdb.orgS.
Reagan, M., Gill, J., 1989. Coexisting calcalkaline and high-niobium basalts from
Turrialba volcano, Costa Rica: implications for residual titanates in arc magma
sources. J. Geophys. Res. 94 (B4), 4619–4633.
Reagan, M.K., Hanan, B.B., Heizler, M.T., Hartman, B.S., Hickey-Vargas, R., 2008.
Petrogenesis of volcanic rocks from Saipan and Rota, Mariana Islands, and
implications for the evolution of Nascent Island Arcs. J. Petrol. 49 (3), 441–464.
Reagan, M.K., Ishizuka, O., Stern, R.J., Kelley, K.A., Ohara, Y., Blichert-Toft, J.,
Bloomer, S.H., Cash, J., Fryer, P., Hanan, B.B., Hickey-Vargas, R., Ishii, T., Kimura,
J.-I., Peate, D.W., Rowe, M.C., Woods, M., 2010. Fore-arc basalts and subduction
initiation in the Izu–Bonin–Mariana system. Geochem. Geophys. Geosyst. 11
(3), 17.
Regelous, M., Gamble, J., Turner, S., 2010. Mechanism and timing of Pb transport
from subducted oceanic crust and sediment to the mantle source of arc lavas.
Chem. Geol. 273 (1–2), 46–54.
Rodda, P., 1994. Geology of Fiji. In: Stevenson, A.J., Herzer, R.H., Ballance, P.F. (Eds.),
Geology and Submarine Resources of the Tonga–Lau–Fiji Region. SOPAC
Technical Bulletin, vol. 8, pp. 131–151.
Rodda, P., Kroenke, L.W., Kroenke, L.W., 1984. Fiji; a fragmented arc. In: Kroenke,
L.W. (Ed.), Cenozoic tectonic development of the Southwest Pacific, vol. 6.
Technical Bulletin, United Nations, Economic and Social Commision for Asia
and the Pacific, Committee for Co-ordination of Joint Prospecting for Mineral
Resources in Asian Offshore Areas (CCOP), pp. 87–109.
Schellart, W.P., Lister, G.S., Toy, V.G., 2006. A late Cretaceous and Cenozoic
reconstruction of the Southwest Pacific region: tectonics controlled by sub-
duction and slab rollback processes. Earth-Sci. Rev. 76 (3–4), 191–233.
194 E. Todd et al. / Earth and Planetary Science Letters 335-336 (2012) 180–194
Sdrolias, M., Mueller, R.D., Gaina, C., 2003. Tectonic evolution of the Southwest
Pacific using constraints from backarc basins. In: Hills, R.R., Mueller, R.D.
(Eds.), Evolution and Dynamics of the Australian plate, vol. 372. Geological
Society of America Special Paper, Boulder, pp. 343–359.
Sharp, W.D., Clague, D.A., 2006. 50-Ma initiation of Hawaiian-Emperor bend
records major change in Pacific Plate motion. Science 313 (5791), 1281–1284.
Shervais, J., 1982. Ti-V plots and the petrogenesis of modern and ophiolitic lavas.
Earth Planet. Sci. Lett. 59 (1), 101–118.
Stern, R.J., Bloomer, S.H., 1992. Subduction zone infancy; examples from the
Eocene Izu–Bonin–Mariana and Jurassic California arcs. Geol. Soc. Am. Bull.
104 (12), 1621–1636.
Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic systematics of oceanic
basalts; implications for mantle composition and processes. In: Saunders, A.D.,
Norry, M.J. (Eds.), Magmatism in the Ocean Basins, vol. 42. Geological Society
Special Publications, London, pp. 313–345.
Tappin, D., Ballance, P., 1994. Contributions to the sedimentary geology of’Eua
Island, Kingdom of Tonga: reworking in an oceanic forearc. Geology and
submarine resources of the Tonga–Lau–Fiji region. SOPAC Tech. Bull. 8, 1–20.
Taylor, B., 1992. Rifting and the volcanic-tectonic evolution of the Izu–Bonin–
Mariana Arc. Proc. Ocean Drilling Program Sci. Results 126, 627–651.
Todd, E., Gill, J.B., Wysoczanski, R.J., Handler, M.R., Wright, I.C., Gamble, J.A., 2010.
Sources of constructional cross-chain volcanism in the southern Havre
Trough: new insights from HFSE and REE concentration and isotope systema-
tics. Geochem. Geophys. Geosyst. 11 (Q04009), 31.
Todd, E., Gill, J.B., Wysoczanski, R.J., Hergt, J.M., Wright, I.C., Leybourne, M.I.,
Mortimer, N., 2011. Hf isotopic evidence for small-scale heterogeneity in the
mode of mantle wedge enrichment: Southern Havre Trough and South
Fiji Basin back-arcs. Geochem. Geophys. Geosyst. Q09011 http://dx.doi.org/
10.1029/2011GC003683.
Turner, S., Hawkesworth, C., Rogers, N., Bartlett, J., Worthington, T., Hergt, J.,
Pearce, J., Smith, I., 1997. 238U–230Th disequilibria, magma petrogenesis, and
flux rates beneath the depleted Tonga–Kermadec island arc. Geochim. Cos-
mochim. Acta 61 (22), 4855–4884.
Wharton, M.R., 1993. Crustal Accretion During the Earliest Stages of Intra-Oceanic
Arc Volcanism: Examples from Fiji and Tonga, SW Pacific. Ph.D. Thesis,
Durham University.
Wharton, M.R., Hathway, B., Colley, H., 1995. Volcanism associated with extension
in an Oligocene–Miocene arc, southwestern Viti Levu, Fiji. In: Smellie, J.L. (Ed.),
Volcanism Associated with Extension at Consuming Plate Margins, vol. 81.
Geological Society Special Publications, London, pp. 95–114.
Whattam, S., Stern, R., 2011. The ‘subduction initiation rule’: a key for linking
ophiolites, intra-oceanic forearcs, and subduction initiation. Contrib. Mineral.
Petrol. 11, 1–15.
Whelan, P.M., Gill, J.B., Kollman, E., Duncan, R.A., Drake, R.E., 1985. Radiometric
dating of magmatic stages in Fiji. In: Scholl, D.W., Vallier, T.L. (Eds.), Geology
and Offshore Resources of Pacific Island Arcs; Tonga region, vol. 2. Circum-Pac.
Counc. Energy and Miner. Resour., Houston, pp. 415–440.
Whittaker, J.M., Muller, R.D., Leitchenkov, G., Stagg, H., Sdrolias, M., Gaina, C.,
Goncharov, A., 2007. Major Australian–Antarctic Plate reorganization at
Hawaiian-Emperor bend time. Science 318 (5847), 83–86.
Winchester, J.A., Floyd, P.A., 1977. Geochemical discrimination of different magma
series and their differentiation products using immobile elements. Chem. Geol.
20, 325–343.
Woodhead, J., Eggins, S., Johnson, R., 1998. Magma genesis in the New Britain
island arc: further insights into melting and mass transfer processes. J. Petrol.
39 (9), 1641–1668.
Woodhead, J.D., Hergt, J.M., Davidson, J.P., Eggins, S.M., 2001. Hafnium isotope
evidence for ‘‘conservative’’ element mobility during subduction zone pro-
cesses. Earth Planet. Sci. Lett. 192 (3), 331–346.