6 5 Todd G3

Geochemistry 3 Article
G
Volume 11, Number 4
Geophysics 29 April 2010
Geosystems Q04009, doi:10.1029/2009GC002888

AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES ISSN: 1525‐2027
Published by AGU and the Geochemical Society
Click
Here
for
Full
Article
Sources of constructional cross‐chain volcanism in the

southern Havre Trough: New insights from HFSE
and REE concentration and isotope systematics
E. Todd and J. B. Gill
Department of Earth and Planetary Sciences, University of California, Santa Cruz, California 95064, USA
([email protected])
R. J. Wysoczanski
National Institute of Water and Atmospheric Research, Wellington 6021, New Zealand
Monica R. Handler
Institute for Research on Earth Evolution, Japan Agency for Marine‐Earth Science and Technology, Yokosuka 237‐0061,
Japan
Now at School of Geography, Environment and Earth Sciences, Victoria University of Wellington, PO Box 600,
Wellington, New Zealand
I. C. Wright
National Oceanography Centre, Southampton, Empress Dock, European Way, Southampton SO14 3ZH, UK
J. A. Gamble
Department of Geology, University College Cork, Cork, Ireland
[1] We report new compositional and isotopic data for submarine lavas erupted on the Rumble V Ridge
cross chain behind the Kermadec Arc at ∼36°S and for locally subducting sediment. In order to constrain
cross‐arc changes in the melt source, Havre Trough ambient mantle wedge isotope and trace element char-
acteristics are interpreted from regional back‐arc basalts that are relatively free of slab‐derived components.
They have MORB‐like trace element concentrations and are isotopically “Pacific” but define greater het-
erogeneity in 206Pb/204Pb and 176Hf/177Hf than previously known within the Havre Trough. In contrast with
the ambient mantle, all Rumble V Ridge lavas have trace element and isotopic characteristics consistent
with subduction zone contributions despite their rear‐arc setting but are less fluid‐enriched than at the
Kermadec volcanic front. A broad trend in Nd‐Hf isotopic space and the correlation between 176Hf/177Hf
and Hf concentration anomaly for Rumble V Ridge lavas is explained by across‐arc changes in (1) the
mantle wedge component, (2) the nature of the subduction component, and (3) the mass fraction of sub-
duction components added. Samples from the distal cross chain tend to have lower 176Hf/177Hf at similar
143
Nd/144Nd compared with samples closer to the arc, suggesting that a low‐176Hf/177Hf component is pref-
erentially removed from the mantle wedge during trenchward advection. Isotope trends suggest that locally
subducting sediment is the primary slab component for Rumble V Ridge magmas, but bulk mixing of
ambient mantle with sediment or slab‐derived fluids cannot account for cross‐arc trace element ratios.
Instead, cross‐chain isotope and trace element characteristics are explained by the addition of 0.05%–
2.0% sediment melts where trace zircon, monazite, and rutile are residual. The cross chain tracks an east
Copyright 2010 by the American Geophysical Union 1 of 31

Geochemistry 3 10.1029/2009GC002888
Geophysics
Geosystems G TODD ET AL.: HAVRE CROSS CHAIN
to west increase in the mass fraction of a common subduction component. A particularly enriched subset of
eastern basalts is interpreted as being derived from the addition of an even higher mass fraction of a dif-
ferent subduction component with greater apparent stability of refractory trace phases during sediment
melting. An implication of this study is that both Nd and Hf can be mobile in sediment‐rich subduction
zones, but the relative mobility depends on the sediment composition and depth of melting, and the abso-
lute mobility is small.
Components: 16,000 words, 14 figures, 5 tables.

Keywords: subduction; hafnium isotopes; sediment partial melting; cross chain; hafnium concentration anomaly.
Index Terms: 3613 Mineralogy and Petrology: Subduction zone processes (1031); 1040 Geochemistry: Radiogenic isotope
geochemistry (1009); 3610 Mineralogy and Petrology: Geochemical modeling (1009).
Received 5 October 2009; Revised 19 January 2010; Accepted 26 January 2010; Published 29 April 2010.
Todd, E., J. B. Gill, R. J. Wysoczanski, M. R. Handler, I. C. Wright, and J. A. Gamble (2010), Sources of constructional
cross‐chain volcanism in the southern Havre Trough: New insights from HFSE and REE concentration and isotope
systematics, Geochem. Geophys. Geosyst., 11, Q04009, doi:10.1029/2009GC002888.
1. Introduction components in arc magmas: aqueous fluids versus

hydrous silicate melts. Near‐ubiquitous enrichments
[2] Back‐arc magmas are chemically diverse, rang- in arc magmas of large ion lithophile elements
ing from MORB‐like to arc‐like in composition, (LILE) that have high fluid/rock distribution coef-
reflecting the juxtaposition of decompression and ficients [Brenan et al., 1995; Keppler, 1996; Brenan
flux melting of variably subduction‐modified man- et al., 1998; Johnson and Plank, 1999] indicate
tle [Stolper and Newman, 1994; Hochstaedter et al., widespread fluid enrichment of the mantle wedge
2000; Taylor and Martinez, 2003; Pearce et al., beneath arcs [Perfit et al., 1980; Gill, 1981; Pearce
2005; Martinez and Taylor, 2006]. An important and Peate, 1995; Ryan et al., 1995; Tatsumi and
consideration in examining the interaction between Eggins, 1995]. Aqueous fluids released from the
subduction and the convecting mantle is how the slab during dehydration processes may be derived
interaction changes across an extending back arc. from sediment, altered oceanic crust, serpentinized
The proportion and distribution of MORB‐like to mantle lithosphere, or a combination thereof. Ele-
arc‐like magmatism in a given back arc vary, and ments that are less soluble in fluids (Th, LREE) are
may be affected by slab/wedge geometry and the also commonly enriched in volcanic arc magmas,
distance to the volcanic front. Mature back‐arc leading many to conclude that slab‐derived silicate
spreading centers have volcanism primarily focused melts also contribute to arc source enrichments
at a spreading center. Thus it becomes difficult to [Elliott et al., 1997; Hawkesworth et al., 1997;
interpret how the subduction contributions affect the Johnson and Plank, 1999; Elliott, 2003]. A contin-
mantle synchronously across the back arc unless the uum between fluid and melt may exist as supercrit-
spreading center is at an angle oblique to the strike of ical fluids [Bureau and Keppler, 1999; Kessel et al.,
the arc (e.g., Eastern Lau Spreading Center (ELSC)) 2005; Hermann et al., 2006]. Melting of subducted
[Escrig et al., 2009]. In more immature back arcs basalt is unlikely to play a significant role in mantle
such as Havre, magmatism is distributed across the wedge enrichment, because modeled temperature
back arc prior to focused melt extraction at a profiles at most subduction zones indicate mantle
spreading center. Consequently, magmas are gen- temperatures are below the water‐undersaturated
erated above the downgoing slab at variable depths basalt solidus [Peacock, 1991; Conder et al., 2002;
and temperatures. Such an immature system should van Keken, 2003; Wiens et al., 2006] (although the
show if subduction components change systemati- water‐saturated solidi of basalt and sediment are at
cally across the back arc and test the relative mobility about the same P and T [Hermann and Spandler,
of elements at different temperatures, pressures, and 2008]). Slab‐derived melts are thus more likely to
matrix conditions. be partial melts of the sediment carapace at the slab/
mantle interface.
[3] Many recent studies have attempted to identify
and characterize two end members of slab‐derived
2 of 31
Geophysics
et al., 2004; Marini et al., 2005; Chauvel et al.,

2009] despite refractory trace zircon [Tollstrup
and Gill, 2005; Duggen et al., 2007; Carpentier
et al., 2009].
[5] We explore the range of subduction zone
enrichment to the mantle wedge source beneath
a 100 km long chain of young (<2 Ma) densely
distributed arc‐like volcanoes extending across the
southern Havre Trough back arc. We examine trace
elements and isotopes to determine the role of
sediment melts versus aqueous fluids. We find that
slab‐derived fluids are present near the volcanic
front but that melts and Hf mobility are restricted to
the distal back arc. Exceptions to the trend are
observed in a suite of high‐Mg magmas where
extreme fluxing adds melt even near the volcanic
front.
2. Tectonic and Petrologic Setting

[6] The Kermadec Arc–Havre Trough (KAHT)
system is associated with westward subduction at the
Pacific‐Australian plate boundary south of ∼25°30′S
(Figure 1), where subduction of the Louisville
Ridge marks the boundary between the Kermadec
Arc (KA) and the Tonga Arc (TA) [Ruellan et al.,
2003; Smith and Price, 2006]. Slab surface tem-
Figure 1. Regional map showing the Kermadec (KE) peratures beneath the volcanic front are predicted to
and Tonga (TO) volcanic arcs, the Havre Trough (HT) be amongst the coldest on Earth beneath Tonga and
and Lau Basin (LB) back arcs, and the Colville (CR) to increase southward, but to remain atypically
and Lau (LR) remnant arcs. The Rumble V Ridge is out- chilly at the latitude of the subject of this paper
lined by the orange polygon and shown in greater detail (E. Syracuse et al., The global range of subduction
in Figure 2. The Taupo volcanic zone (TVZ) is shown zone thermal models, submitted to Physics of the
in grey on the North Island of New Zealand. Locations Earth and Planetary Interiors, 2009). Subduction
of some newly analyzed sediment samples are green of the Hikurangi Plateau south of 36°30′S, where the
crossed circle symbols. The purple crosses show the
overlying plate includes continental New Zealand,
location of basalts used to define the Ambient Mantle
Wedge (AMW; see section 5.1). The solid black lines marks a transition from oceanic to continental con-
indicate spreading axes, whereas dashed black indicate vergence (the Taupo volcanic zone) [Wright et al.,
back‐arc rifting [from Wysoczanski et al., 2010]. 1990; Gamble et al., 1990, 1993, 1995; Wright,
1992; Macpherson et al., 1998].
[7] The Lau Basin and Havre Trough back arcs
[4] Many radiogenic isotope tracers of slab com-
formed following the breakup of the Vitiaz arc
ponents are soluble in aqueous fluids (e.g., Pb, U,
∼5.5 Ma [Ballance et al., 1999] leaving the Lau‐
Li, B, Sr) and therefore track fluid enrichments.
Colville Ridge as a remnant arc. Faster extension
Conversely, Hf, a high field strength element
rates in the Lau Basin than Havre Trough [DeMets
(HFSE) has been interpreted to be conservative
et al., 1994], combined with the subduction of the
during subduction, consistent with characteristic
Louisville Ridge, has been interpreted to prevent
depletions of HFSE among arc magmas relative to
southward progression of seafloor spreading into
other elements having similar mantle/melt partition
the Havre Trough. Prior to Lau‐Havre extension,
coefficients [White and Patchett, 1984; Peate et al.,
rifting of the Eocene Vitiaz Arc and back‐arc
1997; Pearce et al., 1999]. This interpretation has
spreading formed the South Fiji Basin (18 to 23 Ma)
been challenged by evidence that Hf may be added
[Crawford et al., 2003; Sdrolias et al., 2003;
from the slab in certain arcs [Bizimis et al., 2000;
Schellart et al., 2006; Mortimer et al., 2007].
Woodhead et al., 2001; Jicha et al., 2004; Muenker
3 of 31
Geophysics
Figure 2. Three‐dimensional view of Rumble V Ridge, with dredge locations of analyzed samples. Volcanic front
samples are in green, “Western” cross‐chain samples are in blue, “Eastern” cross‐chain samples are in orange, and
“Eastern enriched” samples are in yellow (see section 4.2 for definitions of these cross‐chain groups). Purple
crosses in the Ngatoro Rift (X158/1 and X160) show the most proximal basalts used to define the Ambient Mantle
Wedge (AMW).
[8] Extension in back‐arc basins progresses from “slow” spreading centers (≥4000 m) and are appar-
youthful tectonically controlled rifting [Taylor, ently floored by pillow lavas, and in some cases axial
1992; Wright, 1993; Martinez et al., 1995; Parson pillow ridges [Wysoczanski et al., 2010]. Major rift
and Wright, 1996] to mature magmatically con- segments (typically <100 km long) are bounded by
trolled spreading [Parson et al., 1990; Taylor et al., shallow constructional arc edifices, or “cross chains”
1996]. The boundary between the Tonga and extending across the back arc that may reflect locally
Kermadec arcs is associated with such a change, high melt productivity [e.g., Tamura et al., 2002;
with a north to south transition from spreading in the Wysoczanski et al., 2010].
Lau Basin at the ELSC to mostly rifting in the Havre
[10] The subject of this paper, Rumble V Ridge
Trough [Parson and Wright, 1996; Ruellan et al.,
(∼36°S) is the most prominent cross chain in the
2003]. Seafloor spreading in mature back arcs, like
Havre Trough. It forms a continuous, high‐standing,
the Lau Basin, creates basaltic oceanic crust at
magmatic chain, extending from volcanic front
axial MOR‐like spreading centers. Extension at
seamounts (Rumble V and Rumble IV, ± Clark) to
more youthful back‐arc basins, prior to the organi-
the remnant Colville ridge (Figure 2). The cross
zation of active spreading centers, is characterized by
chain was previously interpreted to result from a
arc‐parallel normal faulting of foundered arc base-
monotonic progression of the arc source across
ment and intrusive rift magmatism [e.g., Gamble
an actively rifting back arc [Wright et al., 1996].
and Wright, 1995; Ruellan et al., 2003; Martinez
However, that conclusion is not supported by sub-
and Taylor, 2006].
sequent Ar/Ar dating [Gamble et al., 2009]. The slab
[9] Extension in the Southern Havre Trough (south lies 150 km beneath the volcanic front end of the
of 32°S) is transitional between primarily amagmatic chain [Syracuse and Abers, 2006], and ≥280 km
rifting and accretionary spreading, with both thinning beneath our westmost sampling site. The Rumble V
of the arc basement crust and accretion of new oce- cross chain has petrologic, volcanic, and tectonic
anic crust by diffuse intrusive volcanism in disorga- features different from the local volcanic front and
nized short‐segment spreading ridges [Wysoczanski bounding back‐arc rift segments to the north (Rumble
et al., 2010]. Rift grabens have depths similar to rift, 35°40′S) and south (Ngatoro rift; 36°20′S)
4 of 31
Geophysics
Table 1 (Sample). Major, Trace Element, and Isotope Data for Rumble V Ridge Lavas, Organized by Decreasing MgO Within
Suitesa [The full Table 1 is available in the HTML version of this article]
Suite and Sample ID
Eastern Group
X690A V432A X690B X689A X686B X691A X682 V441A
Latitude 35.96°S 35.72°S 35.96°S 36°S 36°S 35.9°S 35.97°S 35.82°S
Longitude 177.94°E 178.05°E 177.94°E 177.96°E 178.07°E 177.98°E 178.02°E 177.94°E
From front 30.7 46.0 30.7 26.9 19.5 33.1 24.9 57.0
(km)
Depth (mbsl) 1805 2075 1805 1880 1995 1999 1480 2308
SiO2 47.23 49.78b 48.98 49.16 49.56 50.47 51.13 50.54b

TiO2 0.59 0.94 0.85 1.07 1.08 0.95 1.03 1.1
Al2O3 12.88 15.25 15.22 15.68 16.56 16.43 16.64 16.46
FeOt 9.64 8.6 9.79 9.3 9.71 9.09 8.91 9.35
MnO 0.16 0.14 0.16 0.15 0.16 0.15 0.16 0.17
MgO 14.9 11.1 10.24 8.66 8.5 8.28 8.17 8.04
CaO 12.49 11.23 11.61 12.32 10.3 11.28 9.99 10.91
Na2O 1.68 2.54 2.54 3.16 3.41 2.83 3.39 2.86
K2O 0.32 0.32 0.44 0.35 0.54 0.4 0.42 0.43
P2O5 0.11 0.1 0.15 0.14 0.17 0.12 0.16 0.14
LOI 0.74 0.7
CaO/Al2O3 0.97 0.74 0.76 0.79 0.62 0.69 0.60 0.66

Mg #c 0.73 0.70 0.65 0.62 0.61 0.62 0.62 0.61
Sc 35 35 11 35d 29d 31d 28d 17

V 224 214 170 261 253 294 251 111
Cr 542 532 290 353 315 300 333 128
Co 46 44 39 32
Ni 280 212 265 121 138 127 138 53
Cu 48 57 73 57
Zn 59 66 68 65
Ga 15 16 17 14
Rb 6.2 4.5 4.0 4.1 8.2 5.8 5.5 2.4
Sr 323 209 195 244 268 241 248 114
Y 12.9 20.6 10.9 21.0 22.0 23.0 9.3
Zr 37 74 36 76 83 58 88 46
Nb 1.74 3.94 2.41 2.89 3.00 2.03 1.89 2.52
Cs 0.11 0.07 0.09 0.10 0.17 0.13 0.05
Ba 93 60 95 165 132 126 126 44
La 4.20 4.36 4.49 2.93 4.89 4.61 4.96 2.22
Ce 10.1 10.7 10.3 7.3 12.0 11.2 12.8 6.1
Pr 1.51 1.62 1.52 1.15 1.82 1.69 1.98 0.86
Nd 7.35 7.89 6.89 5.89 8.98 8.38 9.73 4.08
Sm 2.10 2.37 1.98 1.91 2.69 2.48 2.79 1.17
Eu 0.77 0.85 0.63 0.85 1.06 0.93 1.07 0.33
Gd 2.29 2.82 1.97 2.20 3.17 3.14 3.47 1.06
Tb 0.38 0.48 0.33 0.45 0.55 0.54 0.59 0.19
Dy 2.23 3.15 1.99 2.88 3.50 3.47 3.81 1.17
Ho 0.46 0.69 0.41 0.62 0.74 0.74 0.81 0.25
Er 1.26 1.99 1.13 1.83 2.14 2.11 2.41 0.73
Tm 0.18 0.29 0.17 0.27 0.31 0.31 0.35 0.11
Yb 1.16 1.81 1.07 1.76 2.02 2.04 2.29 0.69
Lu 0.18 0.27 0.16 0.27 0.32 0.31 0.35 0.10
a
Major element data are in wt %, and trace element data are in ppm. Suites are Eastern, Eastern enriched, and Western. Trace element (ICPMS/
HR) and isotope analyses (TIMS and MC‐ICPMS) done at UCSC except where otherwise indicated. Italic values indicate previously published
major element data.
b
Major element data were given previously by Wright et al. [1996].
c
Mg # is defined as 100 * Mg(Mg + Fe2+).
d
Trace element analyses done at JAMSTEC.
e
Samples X690B, X691A, V441A, X692A, X686A, X693A, and X699B were analyzed by MC‐ICP‐MS. All others were analyzed by TIMS.
f
Accuracy of 143Nd/144Nd < 0.00001, based on replicate analyses of BHVO2 (0.512987 ± 0.000006; n = 6).
g
Accuracy of 176Hf/177Hf ≤ 0.000005, based on replicate analyses of BHVO2 (0.283096 ± 0.000005; n = 6).
5 of 31
Geophysics
Table 2. New Sr, Nd, Hf, and Pb Isotope Data for One AMW of magma compositions, varying from MORB‐like
Havre Trough Back‐Arc Sample, One Rear Arc Sample, and to arc‐like primitive basalts, with rare rhyodacites
One Volcanic Front Sample From Clark Volcanoa [Gamble et al., 1990; Gamble and Wright, 1995;
Sample ID Gamble et al., 1996; Smith et al., 1997; Haase et al.,
X160 X161b C/8c (V445) 2002; Smith et al., 2003b; Wysoczanski et al., 2006,
2010]. Subduction contributions have been found in
Type back arc VF (rear) VF
Location Ngatoro Rift Rumble IV Clark
lavas well into the back arc, although others have
Latitude 32.12°S 36.04°S 36.4°S little or no subduction input.
Longitude 177.51°E 177.95°E 177.84°E
From front (km) 55.2 25.2
Depth (mbsl) 2300 2035 2010 3. Samples and Analytical Techniques
87
Sr/86Sr 0.703221 ± 12 0.703651 0.704274
143
Nd/144Nd 0.513079 ± 5 0.513042 0.512966 3.1. Samples
176
Hf/177Hf 0.283115 ± 3 0.283151 ± 2 0.283106 ± 2
206
Pb/204Pb 18.695 ± 1 18.706 18.776 ± 1 [12] A total of 21 samples from 16 dredge sites
207
Pb/204Pb 15.558 ± 1 15.573 15.605 ± 1 along the Rumble V cross‐arc chain were analyzed.
208
Pb/204Pb 38.367 ± 3 38.469 38.636 ± 1
The samples were collected during cruises by the
a
See section 3.1 for details. X160, AMW Havre Trough back‐arc Akademik M.A. Lavrentyev in July 1993 (samples
sample (Ngatoro Rift); X161, rear arc sample (Rumble IV seamount); V432–V441) and the Tangaroa in February 1996
VF, volcanic front. Isotopes in italics were previously published
(references given as sample ID footnotes). All other isotope analyses (samples X682–X707). These rocks were sampled
done at UCSC. Major and trace element data for these samples are over a distance of 80 km along the chain from
given by Gamble et al. [1995, 1996, 1997] and Wysoczanski et al. 178.07° to 177.55°W and 36° to 35.7°S (Figure 1).
[2006]. All previously published isotope data (italicized) have been
renormalized to accepted international rock standard values, as dis- Dredge depths ranged from 1480 to 2308 mbsl.
cussed in section 3.2.
b
Isotope data in italics were given previously by Gamble et al. [13] Off‐chain back‐arc lavas (interpreted as rep-
[1996].
c
resentative of the ambient mantle wedge), volcanic
Isotope data in italics were given previously by Gamble et al. front lavas, and Pacific Plate sediments, (Kermadec
[1997].
subducting sediments) that have previously been
reported were also analyzed for additional major,
[Gamble et al., 1995, 1996, 1997; Wright et al., trace and/or isotopic compositions. References
1996; Wysoczanski et al., 2006] (Figure 2). These for these samples can be found in footnotes of
authors have interpreted compositional trends in Tables 1–3.
the Kermadec arc and Havre Trough as reflecting
decreased subduction components with depth to 3.2. Analytical Techniques
the slab. The Rumble V cross chain at 36°S offers
a test of this hypothesis. [14] Major element analyses of most Rumble V
Ridge lava samples were performed by XRF at
[11] Kermadec volcanic front lavas are predomi- Auckland University. Major element data for sam-
nantly basalt to basaltic andesite [Wright, 1993; ples V432 and V441 are from Wright et al. [1996].
Gamble et al., 1995; Turner et al., 1997; Ewart et Concentrations of 30 trace elements were deter-
al., 1998; Haase et al., 2002; Smith and Price, mined at UCSC and JAMSTEC (Japan) (Table 1) by
2006]. Most volcanic front lavas are low‐K tholeiite inductively coupled plasma mass spectrometry
[Smith and Brothers, 1988; Smith et al., 1997], but (ICPMS) following sample preparation and analyt-
rare coeval medium‐ to high‐K basalts similar to ical procedures described by Ryder et al. [2006] and
some of those discussed in this paper coexist with Hanyu et al. [2006], respectively. Four samples
low‐K lavas at Clark and Sonne volcanoes [Gamble were analyzed at both labs (X685A, X690A,
et al., 1997; J. Gamble, personal communication, X693B, and X696A). Interlaboratory agreement for
2009]. Regional along‐arc chemical gradients have element concentrations is within 10% for most ele-
been attributed to both a decreasing role of conti- ments, including critical trace element ratios (e.g.,
nental crust interaction, as in the transition from Th/La, Ce/Pb, HFSE/REE). Precision for elemental
onshore to offshore Taupo volcanism [Gamble et al., analysis at UCSC is estimated from duplicate sam-
1995, 1996], and a northward decrease in the role for ple measurements to be within 1% or 0.1 ppm (for
Pacific sediments (and increasing role for altered transition metals, within 2% RSD or 1 ppm),
oceanic crust) in the Kermadec lava source [e.g., whichever is greater. Multiple digests of an in‐house
Gamble et al., 1996; Turner et al., 1997; Haase et rock standard and multiple analyses of the same
al., 2002]. Havre Trough lavas have a wide range digests of international rock standards DNC1 (n = 15)
6 of 31
Geophysics
Table 3. Major, Trace Element, and Isotope Data for Southern Kermadec Subducting Sediments on an Anhydrous Basisa
Sample ID Average
N17/30 S936 A305 E009 D593 D592 Sediment UCCb
Latitude 32.5°S 36.38°S 37.53°S 41°S 41.92°S 42.92°S
Longitude 177.03°W 178.95°W 179.83°E 178.42°E 177.77°E 178.93°E
Depth (mbsl) 2159 2076 2017 2084 2122 2108
SiO2 64.91 65.91 64.2 62.23 63.37 58.52 64.31

TiO2 0.69 0.65 0.59 0.76 0.71 0.68 0.67
Al2O3 16.82 16.08 14.59 17.2 16.38 15.58 16.17
FeOt 6.66 4.98 4.26 5.87 5.45 5.64 5.44
MnO 0.43 0.08 0.19 0.07 0.08 0.12 0.19
MgO 2.5 2.06 2.13 2.22 2.29 2.49 2.23
CaO 1.93 2.85 8.54 5.2 5.58 11.41 4.63
Na2O 2.93 3.92 2.78 2.88 2.9 2.49 3.13
K2O 2.92 3.32 2.59 3.41 3.04 2.92 3.06
P2O5 0.22 0.14 0.13 0.17 0.19 0.16 0.17
LOI 7.73 16.08 10.25 7.93 8.46 13.77
CaO/Al2O3 0.11 0.18 0.59 0.30 0.34 0.73

Mg #c 0.40 0.42 0.47 0.40 0.43 0.44
Sc 14 11 9 11 11 9 12 14
V 92 90 72 97 98 85 88 97
Cr 40 46 35 52 55 54 43 92
Co 25 10 7 9 9 8 13 17.3
Ni 40 25 20 22 24 25 27 47
Cu 28
Zn 67
Ga 17.5
Rb 105.8 118.5 64.5 138.3 118.2 109.1 106.8 84
Sr 169 229 304 314 342 441 254 320
Y 30.46 25.20 17.10 21.89 21.20 19.09 23.66 21
Zr 129 99 127 94 105 95 112 193
Nb 10.72 10.60 6.67 11.70 10.49 9.22 9.92 12
Cs 6.86 7.18 3.17 8.56 7.19 7.01 6.44 4.9
Ba 701 926 526 645 624 782 699 624
La 24.22 26.13 15.24 25.71 25.78 20.47 22.82 31
Ce 56.66 55.09 33.45 54.43 53.18 43.23 49.91 63
Pr 5.98 6.43 3.85 6.27 6.25 4.89 5.63 7.1
Nd 23.41 24.55 14.74 23.37 23.60 18.82 21.52 27
Sm 5.64 5.39 3.15 5.16 5.05 4.10 4.84 4.7
Eu 1.34 1.27 0.67 1.18 1.10 0.97 1.12 1
Gd 5.50 4.73 2.77 4.39 4.43 3.83 4.35 4
Tb 0.83 0.74 0.42 0.65 0.62 0.55 0.66 0.7
Dy 5.13 4.17 2.57 3.58 3.63 3.09 3.86 3.9
Ho 1.06 0.83 0.55 0.74 0.74 0.63 0.80 0.83
Er 2.91 2.37 1.56 2.11 1.95 1.90 2.24 2.3
Tm 0.44 0.38 0.24 0.31 0.30 0.28 0.34 0.3
Yb 3.07 2.41 1.66 2.01 2.13 1.88 2.29 2
Lu 0.44 0.37 0.24 0.30 0.29 0.28 0.34 0.31
Hf 3.55 2.99 3.11 2.82 3.10 2.77 3.12 5.3
Ta 0.75 0.76 0.48 0.85 0.79 0.68 0.71 0.9
Pb 25.84 21.98 10.14 18.72 21.75 19.17 17
Th 9.24 10.26 6.17 11.48 10.20 8.71 9.29 10.5
U 1.87 3.50 1.49 3.00 2.90 2.80 2.47 2.7
87
Sr/86Sr 0.708999 ± 13 0.709784d 0.709143d 0.710222 ± 12 0.709723 ± 12 0.709621 ± 6 0.709537
143
Nd/144Nd 0.512508 ± 4 0.512457d 0.512543 ± 22 0.512412 ± 8 0.512424 ± 7 0.512451 ± 17 0.512480
176
Hf/177Hf 0.282867 ± 3 0.282754e 0.282758 ± 2 0.282718 ± 6 0.282747 ± 8 0.282742 ± 4 0.282774
206
Pb/204Pb 18.824 ± 1 18.774d 18.800 ± 1 18.906 ± 1 18.826
207
Pb/204Pb 15.645 ± 1 15.619d 15.640 ± 1 15.654 ± 1 15.640
208
Pb/204Pb 38.784 ± 1 38.647d 38.715 ± 1 38.856 ± 2 38.750
7 of 31
Geophysics
and AGV2 (n = 10) run as unknowns gave similar JMC 475‐normalized values for isotope standard
external errors (Table 1). data during the analytical period were BHVO2 =
0.283096 ± 0.000005 (2s, n = 6) and JB2 =
[15] Most Sr isotope ratios were measured at UCSC
0.283267 ± 0.000001 (2s, n = 2).
by thermal ionization mass spectrometry (TIMS)
following methods described by Ryder et al. [18] Pb isotopes were measured using internal
[2006]. Results for NBS 987 during this study spiking with NBS 997 Tl for mass fractionation
averaged 0.710263 ± 0.000013 (2s, n = 40). Results correction, assuming 203Tl/205Tl = 0.418911. Aver-
were normalized to NBS 987 = 0.710250. Typ- age results for NBS 981 over a 4 year period (n = 41,
ical internal error (2s) was <0.00001. Average 8 unique analytical sessions) were 206Pb/204Pb =
values for two international isotope standards 16.931 ± 0.001 (2s), 207Pb/204Pb = 15.483 ± 0.002,
data during the analytical period were AGV2 = and 208Pb/204Pb = 36.675 ± 0.005. Results for each
0.703982, and BHVO2 = 0.703475. Seven sam- analytical session were normalized to 206Pb/204Pb =
ples were analyzed by multicollector inductively 16.941 ± 0.001 (2s), 207Pb/204Pb = 15.496 ± 0.001,
coupled plasma mass spectrometry (MC‐ICPMS) and 208Pb/204Pb = 36.722 ± 0.003 [Abouchami et al.,
using a ThermoFinnigan Neptune(TM) at UCSC 2000]. NBS 981‐normalized values for BCR2
following methods described by Tollstrup et al. (n = 7) during the analytical period were
206
[2010]. Pb/204Pb = 18.761 ± 0.002, 207Pb/204Pb =
15.628 ± 0.002, 208Pb/204Pb = 36.746 ± 0.004.
[16] Nd, Hf, and Pb isotope ratios were measured
by MC‐ICPMS at UCSC following methods in
Ryder et al. [2006] for Pb, Tollstrup et al. [2010] 4. Results
for Nd, and Tollstrup and Gill [2005] for Hf. Nd
isotope standard data acquired during the analytical 4.1. General Characteristics
period were JNdi‐1 = 0.512102 ± 0.000013 (2s, n =
[19] All analyzed lavas from the Rumble V Ridge
59). Results were reproducible within 1 day to
<0.000004 (2s). Results for each analytical session cross chain are mafic silica‐undersaturated basalts,
were normalized to JNdi‐1 = 0.512115 (La Jolla = either weakly hypersthene (<10%) or strongly
0.511858) [Tanaka et al., 2000]. JNdi‐1‐normalized nepheline normative (5%–13%), ranging from
143
Nd/144Nd values for multiply analyzed unknown ∼47 to ∼51.5 wt % SiO2, ∼7.0 to 13.2 wt % MgO
solutions agree between sessions to within <0.000015, (anhydrous), with high Mg # (60 to 73; only two
which we take as our external reproducibility. are <60) (Table 1 and Figure 3). Most plot near the
Multiple digests of the same sample agree within low‐ to medium‐K boundary [Gill, 1981], ranging
0.000010. JNdi‐1‐normalized values for isotope from 0.28 to 0.61 wt % K2O (Figure 3). Elements
standards during the analytical period were La compatible in olivine or clinopyroxene, such as
Jolla = 0.511851 ± 0.000013 (2s, n = 24), and Ca, Fe, Cr and Ni, correlate positively with
BHVO2 = 0.512987 ± 0.000006 (2s, n = 6). MgO, indicating a significant role for fractional
crystallization or accumulation of these mineral
[17] Results for the Hf isotope standard JMC 475 phases.
were 0.282154 ± 0.000007 (2s, n = 74). Results
[20] Compared to N‐MORB, Rumble V Ridge
were reproducible within 1 day to <0.000004 (2s).
Results for each analytical session were normalized basalts are enriched in fluid‐soluble large ion
to JMC 475 = 0.282160 [Chauvel and Blichert‐ lithophile elements (LILE) (e.g., K, Cs, Rb, Ba, Sr,
Toft, 2001]. JMC‐normalized 176Hf/177Hf values Pb, and U) and light rare earth elements (LREE),
for multiply analyzed unknown solutions agree but have lower concentrations of more immobile
between sessions to within <0.000007 which we high field strength elements (HFSE) (e.g., Ti, Y,
take as our external reproducibility. Multiple di- Zr, Nb, Ta, and Hf) and heavy rare earth elements
gests of the same sample agree within 0.000005. (HREE) (Figure 4), typical of convergent plate
Notes to Table 3:
a
Major element data are in wt %, and trace element data are in ppm. Original LOI provided for information only. Average sediment is estimated
as the mean composition and isotopic ratios of N17/30, S936, A305, and E009. Upper continental crust (UCC) estimate of Rudnick and Gao [2003]
is also shown for comparison.
b
Upper continental crust (UCC) composition from Rudnick and Gao [2003].
c
Mg # is deﬁned as 100 * Mg(Mg + Fe2+).
d
Major element data and isotope data in italics were given previously by Gamble et al. [1996].
e
Hafnium isotopic analysis is given by Woodhead et al. [2001].
8 of 31
Geophysics
Figure 3. (a) K2O and (b) FeO*/MgO versus SiO2 for Rumble V Ridge, Clark, and Rumble IV lavas. Discrimination
lines (a) between low‐, medium‐, and high‐K fields from Gill [1981] and (b) between low‐, medium‐, and high‐Fe
fields from Arculus [2003]. Sample X704 has similar trace element ratios to Eastern enriched samples and is thus
plotted with a unique symbol (yellow enclosed in white) here and in Figures 4–14.
boundary magmas. Ratios of fluid mobile to less al., 1993, 1996; Haase et al., 2002]. They resem-
mobile elements (e.g., Ba/Th, Sr/Nd) are gener- ble Kermadec arc compositions except that most
ally higher and more variable than in N‐MORB Rumble V Ridge rocks have less radiogenic Sr at
(Figure 5). Similarly, ratios of the less mobile to a given 143Nd/144Nd. The highest‐K basalts have
the immobile elements (e.g., Th/Yb, Hf/Yb) also the least radiogenic Nd, similar to rocks from the
differ from N‐MORB (Figure 5). Taupo volcanic zone (continental and offshore) and
K‐rich magmas from Clark Volcano [Gamble et
[21] The Sr and Nd isotope ratios for Rumble
al., 1993; Gamble and Wright, 1995; Gamble et
V Ridge basalts are shown in Figure 6, along with
al., 1995, 1996, 1997], though Clark has higher
other back‐arc and arc‐front data from the region. 87
Sr/86Sr relative to143Nd/144Nd (Table 2).
Rumble V Ridge rocks have higher 87Sr/86Sr and
lower 143Nd/144Nd than MORB and most previ- [22] Like Kermadec arc and southern Havre Trough
ously published Havre Trough basalts [Gamble et samples, Rumble V Ridge basalts are displaced
9 of 31
Geophysics
Figure 4. N‐MORB‐normalized multielement variation diagrams displaying the trace element composition of
(a) representative Kermadec subducting sediment and AMW samples, (b) samples from the Eastern group, (c) Eastern
enriched group, (d) and Western group. Normalizing values from Sun and McDonough [1989]. Note the larger y axis
scale for insets of Figures 4a and 4c. Rumble V Ridge symbols as in Figure 3. The AMW group is composed of four
northern Havre Trough samples (gray), two southern Havre Trough samples from the Ngatoro Rift (orange) one
Havre Trough sample from 33°30′S (PPTUW/5, in red), and two samples from the South Fiji Basin (SFB, in purple).
See section 5.1 for individual AMW sample ID and data sources.
above the Northern Hemisphere Reference Line the field for Taupo volcanic zone and locally
(NHRL) in Pb isotopes within a triangle defined by subducting sediment, and are more radiogenic than
South Fiji Basin Hole 205, Havre sample PPTUW/5 other Rumble V Ridge rocks [Gamble et al., 1996,
(both on or near NHRL), and local sediment 1997].
(Figure 7). Most Rumble V Ridge basalts lie
[23] Rumble V Ridge basalts have lower 176Hf/177Hf
near a chord between sediment and PPTUW/5 on
ratios (0.283004 to 0.283178) than previously
a plot of 206Pb/204Pb versus 208Pb/204Pb (Figure 7).
known for Havre Trough lavas and straddle the
In contrast, some northern Havre Trough and
Terrestrial Array [Vervoort et al., 1999] in 176Hf/177Hf
South Fiji Basin rift basalts plot on or just below
versus 143Nd/144Nd space (Figure 8). Volcanic front
the NHRL. The highest K Rumble V Ridge basalts,
magmas from Clark volcano and Rumble IV sea-
like K‐rich magmas from Clark Volcano, overlap
10 of 31
Geophysics
Figure 5. Diagrams showing (a) Pb isotope ratios, (b) Ba/Th, (c) Sr/Nd, (d) Hf/Hf *, (e) Hf/Yb, and (f) Th/Yb versus
LaN/YbN for Rumble V Ridge, Clark and Rumble IV lavas, and Kermadec subducting sediment. Hf/Hf * is defined as
Hf/Hf * = HfNch/(Nd1/4 3/4
Nch * SmNch), after Duggen et al. [2007]. CI chondrite values used were 0.474 ppm Nd, 0.154 ppm
Sm, and 0.107 ppm Hf, from Palme and O’Neill [2003]. The dashed line at Hf/Hf * = 1.0 indicates no Hf anomaly.
Symbols for Rumble V Ridge and Kermadec volcanic front as in Figure 3 and sediments as in Figure 4 (same symbol
for D592, D593, and E009). AMW values are indicated by purple crosses. N‐MORB and E‐MORB values from Niu
et al. [1999] and Regelous et al. [1999]. The range of compositions for southern Kermadec volcanic front rocks is
shown by gray shaded region, compiled from Haase et al. [2002], Gamble et al. [1993, 1995], and Wysoczanski et al.
[2006].
mount lie at the isotopically depleted end of chemistry. Samples from the first of the eastern
Rumble V Ridge basalts, whereas high‐K Clark Rumble V Ridge groups, “Eastern enriched,” are
basalts have low 176Hf/177Hf and 143Nd/144Nd high‐Mg (10.24 to 14.90 wt % MgO) and medium‐
(Table 2 and Figure 8). to high‐K (∼0.4 to 1.6 wt % K2O) lavas from the
eastern cross chain. These samples have high Ni
and Cr (212 to 374 ppm, and 290 to 1003 ppm) and
4.2. Group Distinctions low Al2O3 and TiO2 (12.13 to 17.68 wt % and
[24] Rumble V Ridge basalts are divided into three 0.64 to 0.83 wt %, respectively). They have steep
groups based on their spatial distribution and geo- chondrite‐normalized REE patterns (LaN/YbN =
chemistry (Figures 2 and 9). The first two groups 2.16 to 4.90), and pronounced negative Zr and Hf
include basalts to the east of the Havre Trough rift anomalies (Hf/Hf * = 0.48 to 0.88) (Hf/Hf * =
axis (<40 km behind volcanic front) and are dis- HfNch/(Nd1/4 3/4
Nch * SmNch) [after Duggen et al., 2007],
tinguished based on major and trace element calculated using 0.474 ppm Nd, 0.154 ppm Sm,
11 of 31
Geophysics
Figure 6. Sr‐Nd isotope diagram for Rumble V Ridge, Clark and Rumble IV lavas, and Kermadec subducting sed-
iment. Symbols for N‐MORB and E‐MORB as in Figure 5. Data field and sources for southern Kermadec volcanic
front are the same as Figure 5, plus Gamble et al. [1996]. Data for Taupo volcanic zone (TVZ) from Gamble et al.
[1993] are shown by red dots. Mixing curves show addition of average sediment (gray) and sediment partial melt
(orange) (PM‐2 (Table 5)) to isotopically enriched and depleted AMW, respectively. Tick marks on mixing lines
indicate the fraction of subduction component added at 0.25%, 0.5%, 1.0%, 1.5%, and 2.0%. If the sediment pref-
erentially lost Sr relative to Nd due to dehydration beneath the fore arc, then the mixing lines would become concave‐
up, which might explain some Eastern enriched and Taupo volcanic zone samples.
and 0.107 ppm Hf for CI chondrites [Palme and shallower REE patterns (LaN/YbN = 1.12 to 2.01)
O’Neill, 2003]) (Figures 5 and 9). They have and most have positive Hf anomalies (Hf/Hf* = 0.93
the lowest Nd and Hf isotopic ratios, are closest to to 1.47; X690A/B ∼0.7) (Figures 5 and 9). Com-
the sediment apex of all Rumble V Ridge basalts, pared with Eastern enriched cross‐chain basalts, the
and overlap Taupo volcanic zone in Pb, Sr, and Eastern group has higher TiO2, Al2O3, and Na2O,
Nd isotopes. Sample X704 is less mafic (6.88 wt % and lower FeO*, K2O, CaO/Al2O3, K2O/TiO2, and
MgO), but is included within the Eastern enriched Ni, with little compositional overlap (Table 1). The
group because it shares similar trace element and Eastern group tends to define a shallow trend of
isotope ratios. uranogenic and thorogenic Pb extending to the
lowest 206 Pb/204Pb, similar to previously analyzed
[25] The other eastern Rumble V Ridge basalt group,
Havre Trough back‐arc samples (e.g., Ngatoro
“Eastern,” has lower MgO (7.93 to 8.66 wt % MgO),
Rift). Compared with the Eastern enriched group,
12 of 31
Geophysics
Figure 7. Values of (a) 207Pb/204Pb and (b) 208Pb/204Pb versus 206Pb/204Pb isotope diagrams for Rumble V Ridge,
Clark, and Rumble IV lavas and Kermadec subducting sediment. Mixing lines are shown between average sediment
and western (blue) versus eastern (orange) AMW compositions. Tick mark spacing as in Figure 6. Data field and
sources for southern Kermadec volcanic front are the same as in Figures 5 and 6.
Eastern cross‐chain basalts tend to have lower Western basalts have lower Hf isotope ratios than
LREE/HFSE, Th/HFSE, and 87Sr/86Sr, and higher Eastern, and define a different vector than Eastern
143
Nd/144Nd (Figures 6, 9, and 10). enriched (±Eastern) in Hf‐Nd isotope space.
Western basalts extend to lower 143Nd/144Nd and
[26] The third group, “Western,” includes samples
higher 87Sr/86Sr than most Eastern. For both
from the western back arc (>40 km behind volcanic 207
Pb/204Pb and 208Pb/204Pb versus 206Pb/204Pb,
front). Western basalts resemble Eastern basalts
Eastern enriched basalts are more radiogenic than
with respect to MgO and other major element
all but one Western (Figure 7). Rumble V Ridge
oxides, but have more negative HFSE anomalies
values of 143Nd/144Nd are strongly correlated
and lower 176Hf/177Hf. Eastern and Western basalts
with 206Pb/204Pb, with Eastern enriched basalts
are indistinguishable based on ratios of fluid mobile
being the most like sediment, Eastern being the
to immobile elements, but Western lavas have
most like MORB, and Western intermediate
higher Th/HFSE and REE/HFSE (Figure 10).
(Figure 11).
13 of 31
Geophysics
Figure 8. Nd‐Hf isotope diagrams for Rumble V Ridge, Clark, and Rumble IV lavas and Kermadec subducting sed-
iment. Gray mixing lines show addition of bulk sediment to mantle with isotopically “eastern” (high 176Hf/177Hf) and
“western” (low 176Hf/177Hf) AMW. Colored mixing lines show addition of sediment partial melts (PM‐1, PM‐2, and
PM‐3 (Table 5)) to AMW compositions. Blue shows PM‐1 plus “western” AMW. Orange shows PM‐2 plus “eastern”
AMW. Green shows PM‐3 plus “eastern” AMW. Tick mark spacing as in Figure 5. The isotopic range observed for
northern Kermadec volcanic front (KA) is shown for comparison by gray field for data from Pearce et al. [2007] and
Woodhead et al. [2001]. Unpublished data for one Rumble IV basalt (X162) and for Clark volcano are shown by
green hexagons (Rumble IV), green triangles (Clark), and green stars (Clark high‐K suite) (from M. Handler, personal
communication, 2006). The Terrestrial Array is from Vervoort et al. [1999]. Inset distinguishes between “Indian” and
“Pacific” mantle [Pearce et al., 2007].
4.3. Locally Subducting Sediment ducting sediment of Plank and Langmuir [1998].
Although our study of local sediments is more
[27] Five newly analyzed Kermadec subducting
thorough than their early reconnaissance, it is also
sediments (Table 3) have trace element concentra-
exploratory. XRF major and trace element analyses
tions generally similar to the Kermadec bulk sub-
for these sediments given by Gamble et al. [1996]
14 of 31
Geophysics
Figure 9. Diagram of the relationship between (a) Nd isotopes, (b) Hf/Hf * (see Figure 5 for definition), (c) Nb/Yb,
and (d) Th/Yb among Rumble V Ridge and Rumble IV lavas with increasing distance from the Kermadec volcanic
front. Rumble V Ridge symbols as in Figure 3. Sample V441B, at 55 km from the volcanic front, is classified as
Eastern (rather than Western) because it resembles those in Th/HFSE, Hf/Hf *, and 176Hf/177Hf ratios. Data sources
and symbols for Clark volcano are the same as in Figure 8.
and Woodhead et al. [2001]. REE concentrations (7.3 to 23.5) than previous sediment estimates
are lower than those given by Plank and Langmuir [Plank and Langmuir, 1998]. The sediments are
[1998] and, therefore, the sediments do not have isotopically enriched with 87Sr/86Sr from 0.70916
significant negative Zr and Hf concentration to 0.71022, 143Nd/144Nd from 0.51251 to 0.51241,
anomalies with respect to MREE (Hf/Hf * = 0.7 to and 176Hf/177Hf from 0.28287 to 0.28271, and
0.9, versus 0.3 for estimated Kermadec bulk sedi- straddle the Terrestrial Array of Vervoort et al.
ment) (Figure 5). The lower REE concentrations [1999] (Figure 8), although one sediment with
also affect ratios of radiogenic elements used to high Mn (N17/30 (Table 3)) lies above it. Pb iso-
track sediment addition to a mantle source (e.g., topes for Kermadec sediments are also enriched
Sr/Nd and Nd/Hf). All Kermadec sediments have (206Pb/204Pb ∼18.8 to 18.9 and 208Pb/204Pb ∼38.7
much lower Nd/Hf (8.8 to 6.6) and high Sr/Nd to 38.9).
15 of 31
Geophysics
Figure 10
16 of 31
Geophysics
Figure 11. Values of 206Pb/204Pb versus 143Nd/144Nd for Rumble V Ridge, Clark, and Rumble IV lavas; AMW; and
Kermadec subducting sediment. Symbols as in Figure 5. Mixing lines as in Figure 10. The northern Kermadec
volcanic front is shown for comparison by the gray field, from data sources listed in Figure 10, plus Gamble et al.
[1993, 1995, 1996], Haase et al. [2002], and Smith et al. [2003a].
[28] Given the general similarity in Kermadec sedi- average sediment has an average composition sim-
ment samples, an average sediment was estimated ilar to upper continental crust [Rudnick and Gao,
using the mean composition and isotopic ratios of 2003] except that it is not as enriched in Zr and
representative samples A305, N17/30, S936, and LREE (Table 3), and hence has a strongly terrige-
E009 and this composite sediment is used for mixing nous versus pelagic character. Local terrigenous
and melting calculations discussed below. The sediments are well represented in ODP Holes 1124
Figure 10. Diagram of 176Hf/177Hf versus (a) Hf/Hf * and (b)Th/Hf for Rumble V Ridge, Clark, and Rumble IV
lavas; AMW; and Kermadec subducting sediment. Symbols as in Figure 5. Mixing between components listed in
Table 5 are shown by fan‐shaped symbols in blue (PM‐1), orange (PM‐2), and green (PM‐1), with corresponding
solid mixing lines as in Figure 8. Dashed lines in Figure 10b show range of 1% to 25% partial melts of east (orange)
and west (blue) modified mantle wedge to which 0.1% and 2% sediment melt has been added. Note that y axis in
Figure 10b is logarithmic, so mixing arrays are curved, rather than straight lines. Sample X689A with unusually high
Hf/Hf * is omitted to emphasize differences in other Rumble V Ridge samples. The northern Kermadec volcanic front
is shown for comparison by the gray field, from data sources listed in Figure 8, plus Ewart et al. [1998] and Turner et
al. [1997].
17 of 31
Geophysics
and 1123 (600 km and 1200 km east of the North and Western basalts support the general trend of
Island of New Zealand) [Joseph et al., 2004; increasing depletion toward the trench, but the
Weedon and Hall, 2004], and may sample the inte- overlap suggests that melt extraction is heteroge-
grated subducting Kermadec sediment column. While neous and nonlinear in extent perpendicular to the
key incompatible elements have yet to be analyzed strike of arc and trench. Notably, nowhere is the
for these core samples, LREE/HREE ratios resemble back arc as depleted in Nb/Yb as at the volcanic front
average sediment, supporting the assumption that (Figure 12).
this average serves as a reasonable proxy for the
[31] The AMW suite also has considerable Hf and
Kermadec subducting sediment column.
Pb isotopic heterogeneity (Figures 7, 8, and 11).
The southern samples have higher 207Pb/204Pb and
208
5. Discussion Pb/204Pb at a given 206Pb/204Pb and define an
array steeper than the NHRL. This suggests that
5.1. Ambient Mantle Wedge mantle in the southern Havre Trough has been
more enriched by a crustally derived component,
[29] The Havre Trough mantle to which subduction consistent with previous interpretations [Haase et
components are added is variably enriched with al., 2002]. However, the sample with the highest
respect to the uniform DMM source that has been 206
Pb/204Pb ratio, South Fiji Basin Hole 205, lies
assumed in previous studies that used a singe on the NHRL and exceeds anything observed
depleted Havre Trough basalt (PPTUW/5) as rep- within the Havre Trough, so a FOZO‐like [Stracke
resentative of local mantle [Gamble et al., 1993; et al., 2005] component is present regionally but
Gamble and Wright, 1995; Gamble et al., 1996; has been dilute in Havre Trough basalts for the
Woodhead et al., 2001]. In contrast, we combine past 5 Myr.
that basalt with eight others that lack subduction
influences to define the ambient mantle wedge [32] Hf and Nd isotopic ratios for AMW basalts
(AMW). These basalts include two from the South and the most MORB‐like Rumble V Ridge basalts
Fiji Basin (Hole 205 and Hole 285 [Pearce et al., plot within Pearce et al.’s [2007] “Pacific” space
2007]), four from the Havre Trough from north of (Figure 8), so we infer that isotopically Pacific
30°S (samples 18DR‐2, 18DR‐3, 19DR‐1, and mantle underlies the Havre Trough back arc. How-
19DR2 [Haase et al., 2002]) and two from the Havre ever, 176Hf/177Hf is variable for AMW samples
Trough Ngatoro Rift south of 36°S (samples X158/1 within a small range of 143Nd/144Nd, so some of the
and X160/1 [Gamble et al., 1995; Wysoczanski et heterogeneity in the Havre Trough mantle existed
al., 2006], and new isotope data in Table 2). These prior to recent modification by slab‐derived com-
nine samples plot close to the Th/Yb versus Nb/Yb ponents and defines a trend which is steeper than the
and Hf/Yb versus Nb/Yb arrays defined by N‐MORB Terrestrial Array (Figure 8). In our models we will
and E‐MORB [Pearce et al., 2005] (e.g., Figure 12a) refer to the isotopically more depleted AMW com-
though all have higher Nb/Yb and Th/Yb than ponent as “eastern,” and the more enriched compo-
DMM (Figures 9 and 12a) [Workman and Hart, nent as “western,” because the mantle sources for
2005], indicating a slightly enriched source. AMW Eastern and Western basalts seem to differ in this
samples appear not to have been significantly way. If the Havre mantle initially had variable
176
modified by subduction, with Hf/Hf* ≥ 1.0, Ce/Pb ≥ Hf/177Hf, then the lower 176Hf/177Hf in the
13.0, 143Nd/144Nd > 0.51305, and LaN/YbN ≤ 1.3 western Rumble V Ridge may be derived from a
(Figures 5, 12, and 13). preexisting enriched component (a “plum”) in
trenchward advecting mantle. Progressive melt
[30] Among Rumble V Ridge basalts, the average extraction may preferentially remove this isotopi-
and extremity of Nb/Yb is lower in the east than the cally enriched component, leaving behind more
west, even though the ranges of all three groups refractory and isotopically depleted mantle even in
overlap (∼1 to 3.5) (Figures 9 and 12). The range the absence of back‐arc spreading [Hochstaedter
in Nb/Yb observed across the back arc might be et al., 2001].
explained by variable depletion of the mantle under-
lying the Australian plate by melt extraction during [33] Sample PPTUW/5 is the most isotopically
progressive trenchward advection, consistent with depleted member of the AMW suite. The difference
what has been observed at other back‐arc–arc sys- between it and Kermadec arc lavas figured promi-
tems [Hochstaedter et al., 2000]. If Nb/Yb reflects nently in Woodhead et al.’s [2001] proposal that Hf
prior melt depletion, the differences between Eastern can be mobile in sediment‐poor arcs. However,
18 of 31
Geophysics
Figure 12
19 of 31
Geophysics
their argument fails in the case of KAHT. Most The positive correlation often attributed to addition
Kermadec volcanic front samples overlap the of fluid from altered oceanic crust is absent.
AMW field in Figure 8. The shallow vector within
[37] Ce/Pb ratios in Rumble V Ridge basalts are
the volcanic front parallels, but is shorter than, that
lower than in MORB, indicating addition of Pb
of Rumble V Ridge basalts, and we infer that the
from the slab (Figure 13b). However, samples with
required extent of Hf mobility at the Kermadec
Ce/Pb ≥ 13 have as wide a range in 206Pb/204Pb as
volcanic front is small.
in AMW, indicating that they may preserve a pri-
marily mantle Pb isotopic signature. In contrast, the
5.2. Effects of Crustal Level Contamination cross‐chain samples with Ce/Pb < 13 make a “V”
pointing to sediment, indicating that these have a
[34] Rumble V Ridge basalts are too mafic for their
primarily sediment Pb signature. Positive correla-
characteristics to be attributed to crustal level tion between 206Pb/204Pb and Th/Nb (Figure 13a)
contamination, nor is there seismic evidence for and LaN/YbN (Figure 5), and negative correlation
continental crust beneath the ridge [Scherwath et between 206Pb/204Pb and 143Nd/144Nd (Figure 11)
al., 2008]. Although onshore and offshore Taupo for most Rumble V Ridge samples also show that
volcanic zone isotopic compositions suggest crustal sediment is the primary source of Pb in Rumble V
contamination, primitive KAHT samples north of Ridge basalts. With the exception of our new
∼36°30′S do not [Gamble et al., 1995, 1996]. Kermadec rear‐arc analyses, nothing in the back
Therefore, we conclude therefore that crustal con- arc resembles the northern Kermadec/Tonga trend
tamination has minimal effect on Rumble V Ridge at low 206Pb/204Pb and Ce/Pb, which has been
magma compositions. attributed to contributions from altered oceanic crust
[e.g., Regelous et al., 1997; Turner et al., 1997;
5.3. Slab‐Derived Contributions to the Haase et al., 2002]. The cross‐chain trend shown
Mantle Wedge Beneath Cross Chains in the plot of 207Pb/204Pb versus 206Pb/204Pb may
suggest a need for some altered oceanic crust to
[35] Rumble V Ridge magmas, from a 60 km wide explain the radiogenic mixing component, but not
swath of back‐arc terrane, show more composi- so for 208Pb/204Pb (Figure 7). Therefore, we con-
tional diversity than can be duplicated by simple clude that altered oceanic crust is not the primary
modeling. However, we can constrain the primary source of Pb in Rumble V Ridge basalts.
end‐member sources and processes. All cross‐
chain basalts are enriched in Ba, La, Th, and [38] Although no samples of subducting altered
radiogenic Sr and Pb compared with the AMW oceanic crust are available for the Pacific plate in
(Figures 4–7). Thus, significant slab‐derived com- the region of the Tonga and Kermadec arcs, the
ponents are added to the Rumble V Ridge source composition previously assumed for it in Tonga is
up to ∼80 km behind the active volcanic front. isotopically similar to AMW [Staudigel et al.,
The nature of these slab‐derived components is 1995; Regelous et al., 1997; Turner et al., 1997;
described in sections 5.3.1–5.3.3. Haase et al., 2002; Hergt and Woodhead, 2007]
which makes altered oceanic crust difficult to iden-
tify absent correlations with trace element ratios like
5.3.1. Major Role of Sediment and Minor Role Ce/Pb. Whereas the mass fraction of altered oceanic
of Altered Oceanic Crust crust‐derived components may be large in Tonga
and even the Kermadec Islands, we conclude that it
[36] Sr and Nd isotope ratios correlate negatively is small and swamped by sediment‐derived com-
in cross‐chain basalts, extend beyond the field ponents at the latitude of the Rumble V Ridge cross
of AMW and require a component with higher chain. The general increase in sediment‐like Nd and
87
Sr/86Sr at low 143Nd/144Nd (Figure 6). Sediment Pb isotope ratios (Figure 11) and steepness of REE
is a suitable mixing component for these isotopes.
Figure 12. (a) Th/Yb versus Nb/Yb diagram and (b) Hf/Hf* versus Th/Yb diagram for Rumble V Ridge, Clark, and
Rumble IV lavas; AMW; and Kermadec subducting sediment. Symbols as in Figure 5. Solid mixing curves, dashed
partial melting curves, and partial melt components are shown by same colors, patterns, and symbols as in Figure 10.
Note that axes in Figure 12a are logarithmic, so mixing arrays are curved, rather than straight lines. DMM com-
position [Workman and Hart, 2005] shown by black circle. The northern Kermadec volcanic front is shown for
comparison by the gray field, from data sources listed in Figure 11. Sample X689A (unusually high Hf/Hf *) omitted
to emphasize differences among other Rumble V Ridge samples.
20 of 31
Geophysics
Figure 13. (a) Th/Nb versus 206Pb/204Pb diagram and (b) Ce/Pb versus 206Pb/204Pb diagram for Rumble V Ridge,
Clark, and Rumble IV lavas; AMW; and Kermadec subducting sediment. Symbols as in Figure 5. Solid mixing curves
and dashed partial melting lines (Figure 13b only) as in Figure 10. The northern (light gray) and southern (dark gray)
Kermadec volcanic front are shown for comparison, from data sources listed in Figures 8 and 11. Basalts unaffected
by subduction have Ce/Pb ∼25 ± 5 [Hofmann et al., 1986].
21 of 31
Geophysics
Figure 14. Rumble V Ridge, Clark, and Rumble IV lavas; AMW; and Kermadec subducting sediment. Symbols as
in Figure 5. Mixing lines and dashed partial melting curves as in Figure 10. The northern (light gray) and southern
(dark gray) Kermadec volcanic front are shown for comparison, from data sources listed in Figures 8 and 11.
143
patterns from Eastern to Western (Figure 4) shows Nd/144Nd may require that average sediment is
that the sediment contribution increases away from previously dehydrated to preferentially remove
the arc front. more Sr than Nd so that mixing with AMW pro-
duces a concave‐up mixing curve. However, other
5.3.2. How Is Sediment Transferred to the considerations show that addition of bulk sediment,
or even dehydrated sediment, to a variably depleted
Mantle Wedge?
mantle cannot account for the range of Rumble V
Ridge basalt compositions.
5.3.2.1. Bulk Addition? No
[40] Mixtures of depleted mantle and bulk sediment
[39] Bulk mixing of ∼5% sediment with a MORB will define a straight line in Figure 14. All bulk
source has been proposed to explain southern mixing lines to analyzed sediments (Th/La = 0.43
KAHT Sr, Nd and Pb isotopes and trace elements to 0.49) plot above the compositions for Rumble V
[Gamble et al., 1996]. As noted in section 5.3.1, Ridge basalts. If the mantle has the Sm/La of
mixing the most isotopically depleted AMW AMW (0.6 to 0.9), then the most Th‐enriched
(assuming DMM‐like Sr and Nd concentrations) Rumble V Ridge basalts mix to a subduction
and ≤2% average sediment potentially explains the component with Th/La ∼0.3, nearly a factor of two
range of 87Sr/86Sr versus 143Nd/144Nd in Rumble V lower than local sediments. Tongan sediment has
Ridge basalts (Figure 6). Samples with the lowest
22 of 31
Geophysics
lower Th/La (∼0.13) [Plank and Langmuir, 1998], would also fractionate Sr from Nd such that
but Kermadec sediment is more Th‐enriched than mixing would produce a prominent concave‐down
87
farther north. Thus, enrichment of the source of Sr/86Sr‐143Nd/144Nd curve, well above the slope
Rumble V Ridge melts must include fractionation defined by Rumble V Ridge and regional KAHT
of the bulk sediment component. samples (Figure 6).
[41] A similar conclusion is implied by Hf and Nd
isotopic systematics. Bulk sediment addition to the 5.3.2.3. Melt Addition? Yes
enriched end of the AMW field melts may account [43] A thick sediment column (∼500 m) subducting
for some of the Western trend, but cannot account beneath the Kermadec arc at relatively slow velocity
for most Eastern samples which require a shallower (<7 cm/year) may allow for heating of the uppermost
concave‐down mixing line (Figure 8). The same slab above the water‐saturated sediment solidus at
conclusion is reached when comparing 176Hf/177Hf relatively shallow depths (<120 km). Slab melting
to Th. The negative correlation indicates that the has previously been interpreted to occur at shallower
metasomatizing agent which adds Th also adds depths beneath the volcanic front [Haase et al.,
sediment Hf, but in detail the Rumble V Ridge 2002]. As shown above, bulk sediment addition
samples define a fanning trend that converges near cannot explain the Th/La‐Sm/La, Hf‐Nd isotope,
the AMW array. Bulk mixing between AMW and and 176Hf/177Hf to Th/Hf systematics within the
average sediment cannot explain why most Eastern Rumble V Ridge basalts. Instead, we propose that
samples have a range of Th/Hf above a simple sediment partial melting is required to explain
sediment‐DMM mixing line, so the process that the HFSE/REE fractionation in Rumble V Ridge
adds sediment must also increase its Th/Hf ratio. magmas.
Finally, the magnitude of negative Hf‐Zr concen-
tration anomalies for most Rumble V Ridge basalts [44] We use Hf concentration anomalies (Hf/Hf*)
(Figure 4) cannot be explained by simple bulk addi- [Tollstrup and Gill, 2005] to quantify the effects of
tion because the sediment does not have prominent Hf fractionation during sediment partial melting.
negative Hf‐Zr concentration anomalies. Thus, we The low Hf/Hf* in Rumble V Ridge basalts cannot
propose that while sediment‐derived Nd and Hf be explained by crystal fractionation because the
are added to the mantle wedge during subduction, basalts are so mafic, with no fractionating phase
the mechanism of addition is accompanied by Th/ capable of depleting the melt in Hf. Bulk addition of
HFSE/REE fractionation. local sediment would not significantly fractionate
Hf from LREE because its Hf/Hf* is ∼0.8. However,
5.3.2.2. Fluid Addition? No
sediment melting with refractory zircon would
efficiently fractionate these elements (Figures 10
[42] The calculated trace element concentrations of and 12).
a 2% subsolidus fluid [e.g., Hochstaedter et al.,
[45] The observed negative correlation between
2001; Ishizuka et al., 2006] equilibrated with eclo- 176
Hf/177Hf and Th/Hf (Figure 10b) can be
gitic metasediment (rock/fluid partition coefficients
explained if zircon is a residual phase during
at 700°C, 4GPa [Kessel et al., 2005]) are too dilute to
sediment partial melting. This would also explain
account for Rumble V Ridge basalt compositions.
why Nd and Hf isotopes for most Eastern and
For example, the 143Nd/144Nd for Western, Eastern,
all Eastern enriched samples describe a more
and Eastern enriched basalts would require unreal-
concave‐down mixing line than for bulk sediment
istic mass fractions of aqueous fluid in the mantle of
(Figure 8). Residual monazite can explain the
5%–15%, 10%–20%, and 15%–40%, respectively.
shallower mixing trend between Th/La and Sm/La
Moreover, trace element fractionation by fluids
(Figure 14). The similar range of Nb/Yb for
cannot produce the trace element enrichment patterns
Rumble V Ridge and AMW samples, with some
of Rumble V Ridge basalts. Pb enrichment (e.g., low
volcanic front and Rumble V Ridge samples
Ce/Pb) is less than would be expected for fluid‐only
extending to Nb/Yb ratios less than in N‐MORB
material transport (Figure 13). Calculated 2% fluid
(Figure 12a), suggests that sediment addition
has Ba/Th ∼50 times greater than average sediment
buffers this ratio due to the presence of trace rutile
(∼3,400 and ∼75), but Rumble V Ridge basalts lie
during sediment melting. Thus, the sediment‐
along a trend with Ba/Th nearly an order magnitude
related mixing component is more consistent with
lower than in the predicted fluid. Additionally, there
a melt leaving residual zircon, monazite, and ±
is little overlap in volcanic front and Rumble V
rutile than with bulk sediment. A role for these
Ridge Ce/Pb or Ba/Th. Sediment fluid addition
23 of 31
Geophysics
Table 4. Mineral/Melt Partition Coefficients Used to Model phase proportions of rutile: zircon: monazite of
Sediment Partial Meltinga 0.005: (0.001 to 0.0002): (0.000004 to 0.0000002)
Element (Table 5) have Hf/Hf* and Nb/Yb lower than bulk
cpx b
Garnet b
Rutile c
Zircond Monazitee Micaf sediment ratios and are thus possible mixing
components for observed Rumble V Ridge trends
Sr 0.095 0.0025 0.048 0 0 0.173 (Figures 10a and 12a). Sediment partial melts are
Zr 0.119 0.27 3.79 2005 0 0.011
Nb 0.004 0.0031 540 0 0 0.07 mixed in 0.05% increments with DMM mantle
La 0.049 0.0016 0.237 0 2376.93 0.0004 [Workman and Hart, 2005] having Pb, Nd, and Hf
Ce 0.07 0.005 0.296 0 2883.84 0.00006 isotopic compositions of the radiogenically enriched
Nd 0.178 0.052 0.684 8.5 3518.77 0.0002 and depleted ends of the AMW array (Table 5). This
Sm 0.293 0.25 2.4 16 5543.94 0.0007
Yb 0.4 6.6 0.0158 345 0 0.035
mixture was subsequently batch melted using Ds
Hf 0.2 0.24 4.61 3476 0 0.016 from McDade et al. [2003] (cpx, opx and olivine),
Pb 0 0 0.0154 0 0 0.096 and Adam and Green [2006] (garnet). Garnet lher-
Th 0 0 0.54 41 3919.74 0.0002 zolite starting mantle compositions and modal
a
Coefficients for which no data are available are set to “0.” melting assumed mineral proportions olivine:
b
c
Johnson [1998]. orthopyroxene: clinopyroxene: garnet of 0.6: 0.2:
Foley et al. [2000]. 0.1: 0.1 [Johnson, 1998] because melting is inferred
d
Rubatto and Hermann [2007].
e
Bea and Montero [1999]. to start tens of km above the slab beneath the
f
Adam and Green [2006]. Rumble V Ridge at 3–4 PGa.
[48] Sediment melting in the presence of trace
residual phases during partial melting of sediment (0.02%) residual zircon is required to produce
would explain the negative correlation between the shallow Hf‐Nd isotope mixing array from
Hf/Hf* and Th/Yb (Figure 12b) and positive corre- low‐176Hf/177Hf “western” AMW (PM‐1 (Table 5))
lation between Hf/Hf* and 176Hf/177Hf (Figure 10a). through the trend of Western cross‐chain samples.
The even shallower trend defined by Eastern sam-
ples from slightly higher 176Hf/177Hf “eastern”
5.3.3. Modeling Sediment Melting Leaving AMW requires even more residual zircon and
Residual Accessory Minerals monazite (PM‐2 (Table 5)). This model also appears
to account for trace element ratios for these two
[46] Detailed interpretation of Nd‐Hf isotope and suites. Mixing with average sediment partial melts in
HFSE‐Th‐REE ratios often requires appeal to the presence of 0.02% to 0.03% residual zircon can
sediments melts that leave accessory minerals in explain the negative correlation between 176Hf/177Hf
the residue [e.g., Tollstrup and Gill, 2005; Duggen versus Hf/Hf* and Th/Hf (Figure 10), and Th/Yb
et al., 2007; Carpentier et al., 2009]. In effect, arc versus Hf/Hf* (Figure 12b). Figure 12a shows the
magmas, especially rear‐arc magmas, require the role for residual rutile. Most Rumble V Ridge basalts
addition of minor S‐type granite to their mantle have higher Th/Yb and a greater range in Nb/Yb
source. In the KAHT, the requisite accessory than AMW. Adding a sediment component without
minerals (zircon, rutile, and monazite) are actually residual rutile increases Nb/Yb too much with
present in the subducting Torlesse‐type sediments respect to Th/Yb so some (0.5%) residual rutile is
derived from North Island, New Zealand [Wood required. In the case of both Eastern and Western
and Davy, 1994; Carter et al., 1996; Graham et samples, trace monazite (Table 4) is required to
al., 1992; Gamble et al., 1996; Joseph et al., buffer the effect of zircon on Th/REE (Figure 14).
2004; Weedon and Hall, 2004]. In order to model
[49] Eastern enriched basalts have a wide range of
the behavior of these minerals during water satu- 176
Hf/177Hf which could result from a larger range
rated sediment melting at 3–5 GPa, we calculated
of sediment enrichment, greater variability in iso-
the effect of ≤1% of them in a representative
topic enrichment of the initial AMW, or some
eclogitic metapelite melt (clinopyroxene: garnet:
combination of both. The high Th/Hf and K2O and
mica = 0.4: 0.3: 0.3; ±coesite, kyanite) [Hermann
low 143Nd/144Nd and Hf/Hf* for Eastern enriched
and Spandler, 2008; Li et al., 2008] using parti-
samples compared with most other Rumble V
tion coefficients published for these phases at
Ridge basalts is consistent with more sediment
eclogitic conditions (Table 4).
enrichment. The Eastern enriched samples with the
[47] Calculated partial melts of average sediment lowest 176Hf/177Hf (X686A, X693A, and X693B)
(f = 50%; modal batch melting) with residual trace
24 of 31
Geophysics
Table 5. End‐Member Compositions Used in Mixing Modelsa

Mantle Component Calculated Slab Components
AMW‐West AMW (∼DMM) AMW‐East PM‐1 PM‐2 PM‐3
Zircon 0.0002 0.0003 0.001
Monazite 0.000001 0.000002 0.000004
Sr 7.664 445.2 445.2 445.25

Zr 5.082 119.03 98.28 44.28
Nb 0.1485 3.09 3.09 3.09
La 0.192 44.66 44.66 44.66
Ce 0.55 96.2 95.55 94.29
Nd 0.581 39.83 39.77 39.34
Sm 0.239 8.45 8.42 8.27
Yb 0.365 3.53 3.36 2.49
Hf 0.157 2.47 1.94 0.77
Pb 0.018 36.25 36.25 36.24
Th 0.0079 11.58 8.46 5.44
87
Sr/86Sr 0.702490 0.702490 0.709537
143
Nd/144Nd 0.513100 0.513125 0.512480
176
Hf/177Hf 0.283100 0.283165 0.282774
206
Pb/204Pb 18.673 18.550 18.826
207
Pb/204Pb 15.520 15.520 15.640
208
Pb/204Pb 38.192 38.192 38.750
a
Mantle components are defined as an isotopically “Eastern” and isotopically “Western” AMW, with DMM trace element concentrations
[Workman and Hart, 2005]. Calculated slab components components are 50% partial melts of average sediment (Table 3), with variable resid-
ual zircon and monazite as indicated for each. All calculated sediment partial melts are assumed to have the same isotopic composition, assumed
from average sediment (Table 3).
also have the lowest Th/La and Hf/Hf* and highest and a different residual trace mineral assemblage
Th/Yb and La/Yb, indicating the most extreme than for Eastern and Western samples.
fractionation of [HFSE + Th]/REE and LREE/
[50] Rumble V Ridge rocks define a positive
HREE. Particularly low Hf/Hf* and high Th/Hf for
Hf/Hf*‐176Hf/177Hf slope, and the Kermadec volcanic
these samples require a partial melt with the most
front, Eastern cross‐chain, and Western cross‐chain
residual zircon (0.1%) (PM‐3 (Table 5)) (Figures 10a,
samples each define shallow 176Hf/177Hf‐143Nd/144Nd
10b, and 12b), low Th/La is consistent with more
and Th/La‐Sm/La slopes. All therefore require some
residual monazite (Figure 14), and high Nb/Yb
residual zircon and, to a lesser extent, monazite.
may indicate less residual rutile (Figure 12b).
How much residual zircon (±rutile, monazite) is
However, Eastern enriched samples define a
required in our model depends on which samples
steeper 176Hf/177Hf‐143Nd/144Nd trend than may
are considered (Western versus Eastern, ±which
be explained by addition of either partially melted
Eastern samples) and what AMW is assumed (high
sediment (PM‐2 or PM‐3) or bulk average sedi-
versus low 176Hf/177Hf). For example, some residual
ment (gray) to a homogeneous mantle (Figure 8).
zircon is required to produce a low‐Hf/Hf* slab
High‐176Hf/177Hf Eastern enriched samples fall
component, but too much will restrict the rate of
closer to mixing trends for “western” AMW plus
Th increase and 176Hf/177Hf decrease. Similarly,
partial melt (PM‐1, in dark blue (Table 5)) and
using less residual monazite will produce more
AMW plus bulk average sediment (gray) mixing
negative Hf anomalies, but increase Th/La in the
lines, but require greater Hf/REE fractionation
partial melt. Therefore, zircon and monazite con-
than can be produced by these models (Figures 10a
tents are somewhat subjective.
and 12b). Therefore, we propose that Eastern
enriched (±Clark high‐K) rocks are produced by [51] Although the assumption of a single averaged
adding sediment melt to a heterogeneous mantle sediment composition might not be that realistic,
that includes both low‐solidus isotopically enriched using high‐ or low‐176Hf/177Hf sediments instead
“plums” as well as more refractory, isotopically of average sediment has only a minor effect.
depleted “pudding.” Thus, we consider that Eastern Adding 0.2% and 0.65% of sediment melt with
enriched basalts reflect an isotopically variable higher (N17–30) or lower 176Hf/177Hf (A305)
source mantle, variable amounts of sediment melt, than average sediment produces mixtures with a
25 of 31
Geophysics
176
Hf/177Hf range of only ±0.000002 and ±0.000005, changes in the percent and composition of sedi-
respectively, bracketing mixes produced by adding ment melt added, and secondarily to differences in
the same fraction of average sediment partial melts. the relative fertility and isotopic composition of the
Therefore, we presume that a single averaged sedi- mantle wedge. Those changes are summarized here.
ment composition reasonably constrains an initial
[55] Previous work on KAHT magmas has shown
parameter from which to model the effects on sedi-
that subduction results in the addition of sediment
ment partial melting.
and fluid to the source of Kermadec volcanic front
[52] Though the KAHT is a sediment‐rich system, magmas, with the altered oceanic crust:sediment
the range of basalt compositions requires only a ratio decreasing to the south, from 30°S to ∼36°S
modest addition of sediment melt to the mantle [Gamble et al., 1996; Haase et al., 2002; Turner et
source. Eastern and Western basalts require 0.15% al., 1997]. Hf behaves conservatively in the source
to 0.6%, and Eastern basalts require even less of volcanic front magmas (excluding high‐K Clark
(∼0.05% to 0.5%) if the mantle wedge has been magmas (Table 2)) as indicated by their shallow
previously depleted by prior melt removal. Eastern slope on the Hf‐Nd isotope diagram at the iso-
enriched basalts require the most (0.5% to 2.0%). topically depleted end of the KAHT array
In all, the least to most slab‐affected Rumble V [Woodhead et al., 2001; Pearce et al., 2007]. Low
Ridge samples require only 0.05% to 2.0% sediment Nb/Yb in the Kermadec volcanic front indicates
melt, and most require <0.1% to 0.5% (Figure 8). that the mantle component is more depleted than
This is equivalent to recycling 0.2% to 4% bulk DMM.
sediment. For the least subduction‐affected rocks
[56] The Eastern group has the lowest Th/Yb and
(e.g., Eastern), only 2%–5% of HFSE and HREE
LaN/YbN and the highest Hf/Hf* of the cross‐chain
are sediment‐derived, compared with 10% of Sr,
basalts, plotting close to AMW (Figures 5 and 12),
35% of LREE and 80% of Pb and Th. These per-
and hence this suite has the least sediment melt.
centages increase to 6%–10% of HFSE and HREE,
Additionally, Eastern basalts are derived from the
∼30% of Sr, ∼65% of LREE, and ∼90% of Pb and
most isotopically depleted ambient mantle, indi-
Th for the most sediment‐affected rocks (Eastern
cated by higher 176Hf/177Hf and 143Nd/144Nd. This
enriched ± Western). Only 2%–7% Zr and 1%–3%
is the “pudding” matrix within the advecting
Hf are sediment‐derived when zircon is 0.1% of
mantle, as discussed above. Eastern samples more
the residue (Eastern enriched). For all Rumble V
closely resemble BABB from the Havre Trough at
Ridge samples sediment‐derived Nd is 4–10 times
non‐cross‐chain latitudes. In contrast, Western
greater than Hf.
samples, with higher Th/Yb and Nb/Yb, and lower
[53] Slab surface temperatures at 150 km depth Hf/Hf* and 176Hf/177Hf require both more sediment
along the Tonga‐Kermadec‐New Zealand subduc- melt and an isotopically more enriched ambient
tion system are predicted to be amongst the coldest mantle.
on Earth, to increase about 100°C north to south,
[57] Eastern enriched samples require the most
and to be near the water‐saturated sediment solidus
sediment melt of all, and potentially a different
at the latitude of the Rumble V Ridge (Syracuse
sediment melt, given the offset to high Th/Yb and
et al., submitted manuscript, 2009). Therefore,
[Cs, Rb, Ba, Sr]/Th and low Th/La. The range of
degrees of sediment melting small enough to 176
Hf/177Hf for most Eastern enriched samples
leave residual accessory minerals beneath the
overlaps the range for Eastern basalts, but not
Rumble V Ridge are plausible, but require an
Western (Figure 8). This supports our interpretation
external source of fluid. Although isotope and
that the ambient mantle to the east is isotopically
trace element evidence shows the principle sedi-
more depleted than to the west. Conversely, East-
ment signal in cross‐chain basalts is sediment
ern enriched samples are unlike the Eastern suite
melt, there is evidence even in the back arc for
in that they have high Th/Yb and 143Nd/144Nd,
supplemental fluid‐based enrichment of Ba, Sr,
overlapping and exceeding the Western suite. This
and Pb. However, it is less than for the volcanic
indicates that the source of Eastern enriched mag-
front and Rumble seamounts.
mas has had the most sediment melt enrichment.
Hf/Hf* in Eastern enriched basalts are generally
5.4. Spatial Patterns lower at a given 176Hf/177Hf than for Eastern
or Western samples (Figure 10), requiring more
[54] We attribute the differences amongst Rumble
residual zircon. The large variation in 176Hf/177Hf
V Ridge basalts, as well as between them and ad-
among Eastern enriched samples, overlapping both
jacent volcanic front rocks at similar latitudes, to
26 of 31
Geophysics
the most and least radiogenic Western and Eastern insight into how and where slab components are
basalts, contrasts with them being the most sediment‐ added to the mantle wedge in a sediment‐rich
like in other isotopic tracers (e.g., high 208Pb/204Pb, oceanic arc setting. The mass fraction of sediment
87
Sr/86Sr; low 143Nd/144Nd). This suggests that while melt increases with depth. The cross‐arc changes in
all Eastern enriched sources are enriched by sedi- slab components are most simply explained if the
ment‐derived Sr, Nd, and Pb, the sediment melts are slab component is added vertically. If so, the slab
variable which results in subtle changes in the component is related to the temperature of the
“conservative” behavior of Hf. uppermost slab beneath the cross chain. Others
have shown that the sediment solidus is reached
[58] Thus, we recognize three spatial trends in the
beneath the Kermadec volcanic front [Haase et al.,
addition of sediment‐derived melts across the arc.
2002]. We have shown that the amount of sediment
First, there is a continuum in the mass fraction of
melt in the mantle wedge increases downdip, and
sediment melt added (Eastern enriched > Western >
that residual accessory minerals are less abundant
Eastern). This may reflect an increase in tempera-
(i.e., dissolve) with greater depth. Both reflect
ture with depth at the slab surface (i.e., more sed-
increasing slab surface temperatures. We further
iment melt generation or migration into the mantle
identify differences in the slab component among
at greater slab depth). Second, the apparent relative
the two most trenchward Rumble V Ridge groups,
mobility of Hf increases to the west. This trend is
and show that a large and similar sediment melt
consistent with less residual zircon during sediment
component can occur even in the easternmost
melting, which may also indicate higher tempera-
Havre Trough and at the volcanic front (Clark
tures at the slab surface. Third, the ambient mantle
High‐K suite).
to which sediment melts are added is more depleted
closer to the volcanic front while remaining isoto-
pically “Pacific” throughout. 6. Conclusions
[61] 1. Cross chains of large volcanoes between the
5.5. Tectonic Implications volcanic front and remnant arc are a prominent
[59] The Rumble V Ridge cross chain is similar to feature of the Havre Trough. The Rumble V Ridge
some others. For example, Nd and Hf isotope ratios is a representative example. Its basalts are too
are lower in the Guguan and Kasuga cross chains in mafic for their characteristics to be attributed to
the Mariana arc than in the adjacent volcanic front, crustal level contamination. Therefore, they sample
which has been interpreted as reflecting a more the Havre Trough mantle wedge for 80 km behind
OIB‐like enriched mantle for Guguan [Stern et al., the volcanic front. Three basalt suites are distin-
2006] or more sediment melt for Kasuga [Tollstrup guished geographically and geochemically: East-
and Gill, 2005]. Tamura et al. [2002] and Tamura ern, Western, and Eastern enriched. All contain a
[2003] proposed trenchward advection of a hot subduction component despite their rear‐arc set-
fertile mantle “finger” to explain similar patterns of ting, but are less fluid‐enriched than at the volcanic
isotope ratios, volcanic chains, and mantle low‐ front.
velocity regions perpendicular to the strike of the
[62] 2. The least subduction‐modified Havre Trough
NE Japan arc. Some of the chemical variability at
back‐arc basalts indicate significant chemical and
the 36°S cross chain might be explained by the
isotopic heterogeneity in the ambient mantle wedge,
removal of partial melts from a mantle “hot finger”
all of which is isotopically “Pacific.” The west‐to‐
which could explain the greater depletion of the
east increase in Hf isotopes for the Rumble V Ridge
mantle near the arc front [e.g., Hochstaedter et al.,
basalts with the least slab‐derived component
2001]. Unlike Guguan and NE Japan, however, we
suggests progressive melting of low 176Hf/177Hf
find strong evidence for sediment recycling in the
“plums” during trenchward mantle convection.
rear arc. The evidence for sediment melt in the
source of Rumble V Ridge rocks, plus the high [63] 3. Rumble V Ridge cross‐chain basalt sources
melt productivity required to form the cross chain are variably modified by a common subduction
suggests a relationship between the addition of component that increases in mass fraction with
sediment melt and the presence of hot fingers. This depth to the slab. The subduction component is
will be the topic of a subsequent paper. derived primarily from locally subducting sediment
rather than from altered oceanic crust as is in the
[60] Differences in the composition of basalts from
case at most Kermadec and Tongan volcanic fronts.
the Rumble V Ridge and volcanic front track the
Our new analyses of these sediments show them to
slab component with high fidelity, providing new
27 of 31
Geophysics
be similar to upper continental crust and poorer in Bea, F., and P. Montero (1999), Behavior of accessory phases
LREE, with less negative Zr‐Hf concentration and redistribution of Zr, REE, Y, Th, and U during metamor-
anomalies, than previously estimated. phism and partial melting of metapelites in the lower crust:
An example from the Kinzigite Formation of Ivrea‐Verbano,
[64] 4. The sediment‐derived component cannot be NW Italy, Geochim. Cosmochim. Acta, 63(7–8), 1133–
bulk sediment because of the need for element 1153.
Bizimis, M., V. J. M. Salters, and E. Bonatti (2000), Trace and
fractionation within it. A sediment‐derived fluid is REE content of clinopyroxenes from supra‐subduction zone
excluded due to its predicted low REE and Th peridotites: Implications for melting and enrichment processes
concentrations and high LILE/[Th, REE] ratios. in island arcs, Chem. Geol., 165(1–2), 67–85.
Therefore, the slab‐derived component is a sedi- Brenan, J. M., H. F. Shaw, F. J. Ryerson, and D. L. Phinney
ment melt with refractory accessory zircon, mon- (1995), Mineral‐aqueous fluid partitioning of trace elements
at 900 degrees C and 2.0 GPa: Constraints on the trace ele-
azite, and rutile. The proportion of these minerals ment chemistry of mantle and deep crustal fluids, Geochim.
decreases along the cross chain, reflecting higher Cosmochim. Acta, 59(16), 3331–3350.
temperatures at the slab surface. Brenan, J. M., F. J. Ryerson, and H. F. Shaw (1998), The role
of aqueous fluids in the slab‐to‐mantle transfer of boron,
[65] 5. A subset of high–Mg # and high‐K near‐front beryllium, and lithium during subduction: Experiments
basalts (Eastern enriched group) has the highest and models, Geochim. Cosmochim. Acta, 62(19–20),
mass fraction, and a slightly different kind, of sedi- 3337–3347.
ment melt. The same component also appears in Bureau, H., and H. Keppler (1999), Complete miscibility
between silicate melts and hydrous fluids in the upper man-
high‐K basalts at Clark volcano at the nearby vol- tle: Experimental evidence and geochemical implications,
canic front. Earth Planet. Sci. Lett., 165(2), 187–196.
Carpentier, M., C. Chauvel, R. C. Maury, and N. Mattielli
(2009), The “zircon effect” as recorded by the chemical
Acknowledgments and Hf isotopic compositions of Lesser Antilles forearc sedi-
ments, Earth Planet. Sci. Lett., 287, 83–99.
Carter, L., R. M. Carter, I. N. McCave, and J. Gamble (1996),
[66] Funding for this project came from NSF award 0751600. Regional sediment recycling in the abyssal southwest Pacific
I. C. Wright, R. J. Wysoczanski, and J. A. Gamble helped in Ocean, Geology, 24(8), 735–738.
obtaining the samples used for this project and in their XRF Chauvel, C., and J. Blichert‐Toft (2001), A hafnium isotope
analyses. M. Handler made the ICPMS analyses at JAMSTEC, and trace element perspective on melting of the depleted
and E. Todd made all other analyses at UCSC. Discussions mantle, Earth Planet. Sci. Lett., 190(3–4), 137–151.
between Yoshi Tamura, Janet Hergt, Darren Tollstrup, and Chauvel, C., J.‐C. Marini, T. Plank, and J. N. Ludden (2009),
the first author helped develop some of the concepts presented Hf‐Nd input flux in the Izu‐Mariana subduction zone and
recycling of subducted material in the mantle, Geochem. Geo-
in this paper. The final paper significantly benefited from the
phys. Geosyst., 10, Q01001, doi:10.1029/2008GC002101.
constructive reviews of Mary Reid, Brian Jicha, and Catherine Conder, J. A., D. A. Wiens, and J. Morris (2002), On the
Chauvel. We are grateful to Dan Sampson, Rob Franks, Darren decompression melting structure at volcanic arcs and back‐
Tollstrup, Caroline Harris, and Emily Peterman for their assis- arc spreading centers, Geophys. Res. Lett., 29(15), 1727,
tance at the UCSC labs and to Qing Chang for his assistance doi:10.1029/2002GL015390.
at JAMSTEC. We thank E.T.’s dogs, Kodos and Yerba, for Crawford, A. J., S. Meffre, and P. A. Symonds (2003), 120 to
keeping him at home long enough to complete the project. 0 Ma tectonic evolution of the southwest Pacific and analo-
gous geological evolution of the 600 to 220 Ma Tasman fold
belt system, Spec. Pap. Geol. Soc. Am., 372, 383–403.
DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein (1994),
References Effect of recent revisions to the geomagnetic reversal time
scale on estimates of current plate motions, Geophys. Res.
Abouchami, W., S. J. G. Galer, and A. W. Hofmann (2000), Lett., 21(20), 2191–2194.
High precision lead isotope systematics of lavas from the Duggen, S., M. Portnyagin, J. Baker, D. Ulfbeck, K. Hoernle,
Hawaiian Scientific Drilling Project, Chem. Geol., 169(1–2), D. Garbe‐Schonberg, and N. Grassineau (2007), Drastic shift
187–209. in lava geochemistry in the volcanic‐front to rear‐arc region
Adam, J., and T. Green (2006), Trace element partitioning of the Southern Kamchatkan subduction zone: Evidence for
between mica‐ and amphibole‐bearing garnet lherzolite the transition from slab surface dehydration to sediment
and hydrous basanitic melt: 1. Experimental results and the melting, Geochim. Cosmochim. Acta, 71(2), 452–480.
investigation of controls on partitioning behaviour, Contrib. Elliott, T. (2003), Tracers of the slab, in Inside the Subduction
Mineral. Petrol., 152(1), 1–17. Factory, Geophys. Monogr. Ser., vol. 138, edited by J. Eiler,
Arculus, R. J. (2003), Use and abuse of the terms calcalkaline pp. 23–45, AGU, Washington, D. C.
and calcalkalic, J. Petrol., 44(5), 929–935. Elliott, T., T. Plank, A. Zindler, W. White, and B. Bourdon
Ballance, P. F., A. G. Ablaev, I. K. Pushchin, S. P. Pletnev, (1997), Element transport from slab to volcanic front at the
M. G. Birylina, T. Itaya, H. A. Follas, and G. W. Gibson Mariana Arc, J. Geophys. Res., 102(7), 14,991–15,019.
(1999), Morphology and history of the Kermadec Trench‐ Escrig, S., A. Bézos, S. L. Goldstein, C. H. Langmuir, and P. J.
arc‐backarc basin‐remnant arc system at 30 to 32°S: Geo- Michael (2009), Mantle source variations beneath the East-
physical profile, microfossil and K‐Ar data, Mar. Geol., ern Lau Spreading Center and the nature of subduction com-
159(1–4), 35–62.
28 of 31
Geophysics
ponents in the Lau basin‐Tonga arc system, Geochem. Geophys. geochemistry, Geochem. Geophys. Geosyst., 7, Q08002,
Geosyst., 10, Q04014, doi:10.1029/2008GC002281. doi:10.1029/2005GC001220.
Ewart, A., K. D. Collerson, M. Regelous, J. I. Wendt, and Hawkesworth, C. J., S. P. Turner, F. McDermott, D. W. Peate,
Y. Niu (1998), Geochemical evolution within the Tonga‐ and P. van Calsteren (1997), U‐Th isotopes in arc magmas:
Kermadec‐Lau arc‐back‐arc systems: The role of varying Implications for element transfer from the subducted crust,
mantle wedge composition in space and time, J. Petrol., Science, 276(5312), 551–555.
39(3), 331–368. Hergt, J., and J. D. Woodhead (2007), A critical evaluation of
Foley, S. F., M. G. Barth, and G. A. Jenner (2000), Rutile/melt recent models for Lau‐Tonga arc‐backarc basin magmatic
partition coefficients for trace elements and an assessment of evolution, Chem. Geol., 245(1–2), 9–44.
the influence of rutile on the trace element characteristics of Hermann, J., and C. J. Spandler (2008), Sediment melts at sub‐
subduction zone magmas, Geochim. Cosmochim. Acta, 64(5), arc depths: An experimental study, J. Petrol., 49(4), 717–740.
933–938. Hermann, J., C. Spandler, A. Hack, and A. V. Korsakov
Gamble, J. A., and I. C. Wright (1995), The Southern Havre (2006), Aqueous fluids and hydrous melts in high‐pressure
Trough: Geological structure and magma petrogenesis of and ultra‐high pressure rocks: Implications for element trans-
an active backarc rift complex, in Backarc Basins: Tectonics fer in subduction zones, Lithos, 92(3–4), 399–417.
and Magmatism, edited by B. Taylor, pp. 193–212, Plenum, Hochstaedter, A. G., J. B. Gill, B. Taylor, O. Ishizuka,
New York. M. Yuasa, and S. Morita (2000), Across‐arc geochemical
Gamble, J. A., I. E. M. Smith, I. J. Graham, B. P. Kokelaar, trends in the Izu‐Bonin Arc: Constraints on source composi-
J. W. Cole, B. F. Houghton, and C. J. N. Wilson (1990), tion and mantle melting, J. Geophys. Res., 105(1), 495–512.
The petrology, phase relations and tectonic setting of basalts Hochstaedter, A., J. Gill, R. Peters, P. Broughton, P. Holden,
from the Taupo volcanic zone, New Zealand and the Kermadec and B. Taylor (2001), Across‐arc geochemical trends
island arc‐Havre Trough, SW Pacific, J. Volcanol. Geotherm. in the Izu‐Bonin arc: Contributions from the subducting
Res., 43(1–4), 253–270. slab, Geochem. Geophys. Geosyst., 2(7), 1019, doi:10.1029/
Gamble, J. A., I. E. M. Smith, M. T. McCulloch, I. J. Graham, 2000GC000105.
and B. P. Kokelaar (1993), The geochemistry and petrogen- Hofmann, A. W., K. P. Jochum, M. Seufert, and W. M. White
esis of basalts from the Taupo volcanic zone and Kermadec (1986), Nb and Pb in oceanic basalts: New constraints on
island arc, S.W. Pacific, J. Volcanol. Geotherm. Res., 54(3–4), mantle evolution, Earth Planet. Sci. Lett., 79(1–2), 33–45.
265–290. Ishizuka, O., R. N. Taylor, J. A. Milton, R. W. Nesbitt, M. Yuasa,
Gamble, J. A., I. C. Wright, J. D. Woodhead, and M. T. and I. Sakamoto (2006), Variation in the mantle sources of
McCulloch (1995), Arc and back‐arc geochemistry in the the northern Izu arc with time and space—Constraints from
southern Kermadec arc‐Ngatoro Basin and offshore Taupo high‐precision Pb isotopes, J. Volcanol. Geotherm. Res.,
Volcanic Zone, SW Pacific, in Volcanism Associated With 156(3–4), 266–290.
Extension at Consuming Plate Margins, edited by J. L. Jicha, B. R., B. S. Singer, J. G. Brophy, J. H. Fournelle, C. M.
Smellie, Geol. Soc. Spec. Publ., 81, 193–212. Johnson, B. L. Beard, T. J. Lapen, and N. J. Mahlen (2004),
Gamble, J. A., J. D. Woodhead, I. Wright, and I. Smith (1996), Variable impact of the subducted slab on Aleutian island arc
Basalt and sediment geochemistry and magma petrogenesis magma sources: Evidence from Sr, Nd, Pb, and Hf isotopes
in a transect from oceanic island arc to rifted continental and trace element abundances, J. Petrol., 45(9), 1845–1875.
margin arc: The Kermadec‐Hikurangi margin, SW Pacific, Johnson, K. T. M. (1998), Experimental determination of par-
J. Petrol., 37(6), 1523–1546. tition coefficients for rare earth and high‐field‐strength ele-
Gamble, J. A., R. H. K. Christie, I. C. Wright, and R. J. ments between clinopyroxene, garnet, and basaltic melt at
Wysoczanski (1997), Primitive K‐rich magmas from high pressure, Contrib. Mineral. Petrol., 133(1–2), 60–68.
Clark Volcano, southern Kermadec Arc: A paradox in Johnson, M. C., and T. Plank (1999), Dehydration and melting
the K‐depth relationship, Can. Mineral., 35, 275–290. experiments constrain the fate of subducted sediments,
Gamble, J., I. Wright, R. Wysoczanski, E. Todd, and W. McIntosh Geochem. Geophys. Geosyst., 1(12), 1007, doi:10.1029/
(2009), Geochemistry, geochronology and tectonic implica- 1999GC000014.
tions of cross‐arc ridge volcanism in the Southern Kermadec Joseph, L. H., D. K. Rea, and B. A. van der Pluijm (2004),
Arc, SW Pacific, Geochim. Cosmochim. Acta, 73(13), suppl. 1, Neogene history of the Deep Western Boundary Current at
A523. Rekohu sediment drift, southwest Pacific (ODP Site 1124),
Gill, J. B. (1981), Orogenic Andesites and Plate Tectonics, Mar. Geol., 205(1–4), 185–206.
Springer, Berlin. Keppler, H. (1996), Constraints from partitioning experiments
Graham, I. J., B. L. Gulson, J. W. Hedenquist, and K. Mizon on the composition of subduction‐zone fluids, Nature,
(1992), Petrogenesis of late Cenozoic volcanic rocks from 380(6571), 237–240.
the Taupo volcanic zone, New Zealand, in the light of new Kessel, R., M. W. Schmidt, P. Ulmer, and T. Pettke (2005),
lead isotope data, Geochim. Cosmochim. Acta, 56(7), Trace element signature of subduction‐zone fluids, melts
2797–2819. and supercritical liquids at 120–180 km depth, Nature,
Haase, K. M., T. J. Worthington, P. Stoffers, D. Garbe‐Schönberg, 437(7059), 724–727.
and I. Wright (2002), Mantle dynamics, element recycling, Li, Y., H.‐J. Massonne, A. Willner, H.‐F. Tang, and C.‐Q. Liu
and magma genesis beneath the Kermadec Arc‐Havre (2008), Dehydration of clastic sediments in subduction
Trough, Geochem. Geophys. Geosyst., 3(11), 1071, zones: Theoretical study using thermodynamic data of
doi:10.1029/2002GC000335. minerals, Isl. Arc, 17(4), 577–590.
Hanyu, T., Y. Tatsumi, S. Nakai, Q. Chang, T. Miyazaki, Macpherson, C. G., J. A. Gamble, and D. P. Mattey (1998),
K. Sato, K. Tani, T. Shibata, and T. Yoshida (2006), Contri- Oxygen isotope geochemistry of lavas from an oceanic to
bution of slab melting and slab dehydration to magmatism continental arc transition, Kermadec‐Hikurangi margin,
in the NE Japan arc for the last 25 Myr: Constraints from SW Pacific, Earth Planet. Sci. Lett., 160(3–4), 609–621.
29 of 31
Geophysics
Marini, J.‐C., C. Chauvel, and R. C. Maury (2005), Hf isotope Perfit, M. R., D. A. Gust, A. E. Bence, R. J. Arculus, and S. R.
compositions of northern Luzon arc lavas suggest involve- Taylor (1980), Chemical characteristics of island‐arc basalts:
ment of pelagic sediments in their source, Contrib. Mineral. Implications for mantle sources, Chem. Geol., 30(3),
Petrol., 149, 216–232. 227–256.
Martinez, F., and B. Taylor (2006), Modes of crustal accretion in Plank, T., and C. H. Langmuir (1998), The chemical composi-
back‐arc basins: Inferences from the Lau Basin, in Back‐Arc tion of subducting sediment and its consequences for the
Spreading Systems: Geological, Biological, Chemical, crust and mantle, Chem. Geol., 145(3–4), 325–394.
and Physical Interactions, Geophys. Monogr. Ser., vol. 166, Regelous, M., K. D. Collerson, A. Ewart, and J. I. Wendt
edited by D. Christie et al., pp. 5–30, AGU, Washington, D. C. (1997), Trace element transport rates in subduction zones:
Martinez, F., P. Fryer, N. A. Baker, and T. Yamazaki (1995), Evidence from Th, Sr and Pb isotope data for Tonga‐Kermadec
Evolution of backarc rifting: Mariana Trough, 20°–24°N, Arc lavas, Earth Planet. Sci. Lett., 150(3–4), 291–302.
J. Geophys. Res., 100(B3), 3807–3827. Regelous, M., Y. Niu, J. I. Wendt, R. Batiza, A. Greig, and
McDade, P., J. D. Blundy, and B. J. Wood (2003), Trace ele- K. D. Collerson (1999), An 800 Ka record of the geochem-
ment partitioning between mantle wedge peridotite and istry of magmatism on the East Pacific Rise at 10°30′N:
hydrous MgO‐rich melt, Am. Mineral., Part 1, 88(11–12), Insights into magma chamber processes beneath a fast‐
1825–1831. spreading ocean ridge, Earth Planet. Sci. Lett., 168, 45–63.
Mortimer, N., R. H. Herzer, P. B. Gans, C. Laporte‐Magoni, Rubatto, D., and J. Hermann (2007), Experimental zircon/melt
A. T. Calvert, and D. Bosch (2007), Oligocene‐Miocene and zircon/garnet trace element partitioning and implications
tectonic evolution of the South Fiji Basin and Northland for the geochronology of crustal rocks, Chem. Geol., 241(1–2),
Plateau, SW Pacific Ocean: Evidence from petrology and 38–61.
dating of dredged rocks, Mar. Geol., 237(1–2), 1–24. Rudnick, R. L., and S. Gao (2003), Composition of the conti-
Muenker, C., G. Woerner, G. Yogodzinski, and T. Churikova nental crust, in Treatise on Geochemistry, vol. 3, The Crust,
(2004), Behaviour of high‐field‐strength elements in sub- edited by D. H. Heinrich and K. T. Karl, pp. 1–64, Pergamon,
duction zones: Constraints from Kamchatka‐Aleutian Arc Oxford, U. K.
lavas, Earth Planet. Sci. Lett., 224(3–4), 275–293. Ruellan, E., J. Delteil, I. Wright, and T. Matsumoto (2003),
Niu, Y., K. D. Collerson, R. Batiza, J. I. Wendt, and M. Regelous From rifting to active spreading in the Lau Basin–Havre
(1999), Origin of enriched‐type mid‐ocean ridge basalt Trough backarc system (SW Pacific): Locking/unlocking
at ridges far from mantle plumes: The East Pacific Rise at induced by seamount chain subduction, Geochem. Geophys.
11°20′N, J. Geophys. Res., 104(B4), 7067–7087. Geosyst., 4(5), 8909, doi:10.1029/2001GC000261.
Palme, H., and H. S. C. O’Neill (2003), Cosmochemical esti- Ryan, J. G., J. Morris, F. Tera, W. P. Leeman, and A. Tsvetkov
mates of mantle composition, in Treatise on Geochemistry, (1995), Cross‐arc geochemical variations in the Kurile Arc
vol. 2, The Mantle and Core, pp. 1–38, Pergamon, Oxford, as a function of slab depth, Science, 270(5236), 625–627.
U. K. Ryder, C., J. B. Gill, F. J. I. Tepley, F. Ramos, and M. Reagan
Parson, L. M., and I. C. Wright (1996), The Lau‐Havre‐Taupo (2006), Closed‐ to open‐system differentiation at Arenal vol-
back‐arc basin: A southward‐propagating, multi‐stage evo- cano, J. Volcanol. Geotherm. Res., 157, 75–93.
lution from rifting to spreading, Tectonophysics, 263(1–4), Schellart, W. P., G. S. Lister, and V. G. Toy (2006), A Late
1–22. Cretaceous and Cenozoic reconstruction of the southwest Pa-
Parson, L. M., et al. (1990), Role of ridge jumps and ridge cific region: Tectonics controlled by subduction and slab
propagation in the tectonic evolution of the Lau back‐arc rollback processes, Earth Sci. Rev., 76(3–4), 191–233.
basin, southwest Pacific, Geology, 18(5), 470–473. Scherwath, M., H. Kopp, E. Flueh, S. Henrys, and R. Sutherland
Peacock, S. M. (1991), Numerical simulation of subduction (2008), Structure and deformation of the Hikurangi‐Kerma-
zone pressure‐temperature‐time paths: Constraints on fluid dec Subduction Zone: Transitions revealed by seismic
production and arc magmatism, Philos. Trans. R. Soc. London, wide‐angle data, Eos Trans. AGU, 89(53), Fall Meet.
Ser. A, 335, 341–353. Suppl., Abstract T23A‐1997.
Pearce, J. A., and D. W. Peate (1995), Tectonic implications of Sdrolias, M., R. D. Mueller, and C. Gaina (2003), Tectonic
the composition of volcanic arc magmas, Annu. Rev. Earth evolution of the southwest Pacific using constraints from
Planet. Sci., 23, 251–285. backarc basins, Spec. Pap. Geol. Soc. Am., 372, 343–359.
Pearce, J. A., P. D. Kempton, G. M. Nowell, and S. R. Noble Smith, I. E. M., and R. N. Brothers (1988), Petrology of the
(1999), Hf‐Nd element and isotope perspective on the nature Rumble seamounts, southern Kermadec Ridge, southwest
and provenance of mantle and subduction components in Pacific, Bull. Volcanol., 50(3), 139–147.
western Pacific arc‐basin systems, J. Petrol., 40(11), Smith, I. E. M., and R. C. Price (2006), The Tonga‐Kermadec
1579–1611. arc and Havre‐Lau back‐arc system: Their role in the devel-
Pearce, J. A., R. J. Stern, S. H. Bloomer, and P. Fryer (2005), opment of tectonic and magmatic models for the western
Geochemical mapping of the Mariana arc‐basin system: Pacific, J. Volcanol. Geotherm. Res., 156(3–4), 315–331.
Implications for the nature and distribution of subduction Smith, I. E. M., T. J. Worthington, R. C. Price, and J. A. Gamble
components, Geochem. Geophys. Geosyst., 6, Q07006, (1997), Primitive magmas in arc‐type volcanic associations:
doi:10.1029/2004GC000895. Examples from the southwest Pacific, Can. Mineral., 35(2),
Pearce, J. A., P. D. Kempton, and J. B. Gill (2007), Hf‐Nd 257–273.
evidence for the origin and distribution of mantle domains Smith, I. E. M., R. B. Stewart, and R. C. Price (2003a), The
in the SW Pacific, Earth Planet. Sci. Lett., 260(1–2), 98–114. petrology of a large intra‐oceanic silicic eruption: The Sandy
Peate, D. W., J. A. Pearce, C. J. Hawkesworth, H. Colley, Bay Tephra, Kermadec Arc, southwest Pacific, J. Volcanol.
C. M. H. Edwards, and K. Hirose (1997), Geochemical var- Geotherm. Res., 124(3–4), 173–194.
iations in Vanuatu Arc lavas: The role of subducted material Smith, I. E. M., T. J. Worthington, R. B. Stewart, R. C. Price,
and a variable mantle wedge composition, J. Petrol., 38(10), and J. A. Gamble (2003b), Felsic volcanism in the Kermadec
1331–1358. Arc, SW Pacific: Crustal recycling in an oceanic setting, in
30 of 31
Geophysics
Intra‐Oceanic Subduction Systems: Tectonic and Magmatic Turner, S., C. Hawkesworth, N. Rogers, J. Bartlett, T. Worthington,
Processes, edited by R. D. Larter and P. T. Leat, Geol. J. Hergt, J. Pearce, and I. Smith (1997), 238U‐230Th disequili-
Soc. Spec. Publ., 219, 99–118. bria, magma petrogenesis, and flux rates beneath the depleted
Staudigel, H., G. R. Davies, S. R. Hart, K. M. Marchant, and Tonga‐Kermadec island arc, Geochim. Cosmochim. Acta,
B. M. Smith (1995), Large scale isotopic Sr, Nd and O iso- 61(22), 4855–4884.
topic anatomy of altered oceanic crust: DSDP/ODP sites van Keken, P. E. (2003), The structure and dynamics of the
417/418, Earth Planet. Sci. Lett., 130(1–4), 169–185. mantle wedge, Earth Planet. Sci. Lett., 215(3–4), 323–338.
Stern, R. J., E. Kohut, S. H. Bloomer, M. Leybourne, M. Fouch, Vervoort, J. D., P. J. Patchett, J. Blichert‐Toft, and F. Albarede
and J. Vervoort (2006), Subduction factory processes beneath (1999), Relationships between Lu‐Hf and Sm‐Nd isotopic
the Guguan cross‐chain, Mariana Arc: No role for sediments, systems in the global sedimentary system, Earth Planet.
are serpentinites important?, Contribut. Mineral. Petrol., Sci. Lett., 168(1–2), 79–99.
151(2), 202–221. Weedon, G. P., and I. R. Hall (2004), Neogene palaeoceano-
Stolper, E., and S. Newman (1994), The role of water in the graphy of Chatham Rise (southwest Pacific) based on sedi-
petrogenesis of Mariana Trough magmas, Earth Planet. ment geochemistry, Mar. Geol., 205(1–4), 207–225.
Sci. Lett., 121(3–4), 293–325. White, W. M., and J. Patchett (1984), Hf‐Nd‐Sr isotopes and
Stracke, A., A. W. Hofmann, and S. R. Hart (2005), FOZO, incompatible element abundances in island arcs: Implica-
HIMU, and the rest of the mantle zoo, Geochem. Geophys. tions for magma origins and crust‐mantle evolution, Earth
Geosyst., 6, Q05007, doi:10.1029/2004GC000824. Planet. Sci. Lett., 67(2), 167–185.
Sun, S. S., and W. F. McDonough (1989), Chemical and iso- Wiens, D. A., K. A. Kelley, and T. Plank (2006), Mantle tem-
topic systematics of oceanic basalts: Implications for mantle perature variations beneath back‐arc spreading centers in-
composition and processes, Geol. Soc. Spec. Publ., 42, ferred from seismology, petrology, and bathymetry, Earth
313–345. Planet. Sci. Lett., 248(1–2), 30–42.
Syracuse, E. M., and G. A. Abers (2006), Global compilation Wood, R., and B. Davy (1994), The Hikurangi Plateau, Mar.
of variations in slab depth beneath arc volcanoes and impli- Geol., 118(1–2), 153–173.
cations, Geochem. Geophys. Geosyst., 7, Q05017, Woodhead, J. D., J. M. Hergt, J. P. Davidson, and S. M. Eggins
doi:10.1029/2005GC001045. (2001), Hafnium isotope evidence for “conservative” element
Tamura, Y. (2003), Some geochemical constraints on hot mobility during subduction zone processes, Earth Planet. Sci.
fingers in the mantle wedge: Evidence from NE Japan, in Lett., 192(3), 331–346.
Intra‐Oceanic Subduction Systems: Tectonic and Magmatic Workman, R. K., and S. R. Hart (2005), Major and trace ele-
Processes, edited by R. D. Larter and P. T. Leat, Geol. ment composition of the depleted MORB mantle (DMM),
Soc. Spec. Publ., 219, 221–237. Earth Planet. Sci. Lett., 231(1–2), 53–72.
Tamura, Y., Y. Tatsumi, D. Zhao, Y. Kido, and H. Shukuno Wright, I. C. (1992), Shallow structure and active tectonism of
(2002), Hot fingers in the mantle wedge: New insights into an offshore continental back‐arc spreading system: The Taupo
magma genesis in subduction zones, Earth Planet. Sci. Lett., volcanic zone, New Zealand, Mar. Geol., 103(1–3), 287–309.
197(1–2), 105–116. Wright, I. C. (1993), Pre‐spread rifting and heterogeneous vol-
Tanaka, T., et al. (2000), JNdi‐1: A neodymium isotopic refer- canism in the southern Havre Trough back‐arc basin, Mar.
ence in consistency with La Jolla neodymium, Chem. Geol., Geol., 113(3–4), 179–200.
168(3–4), 279–281. Wright, I., L. Carter, and K. Lewis (1990), GLORIA survey of
Tatsumi, Y., and S. Eggins (1995), Subduction Zone Magma- the oceanic‐continental transition of the Havre‐Taupo back‐
tism, Blackwell Sci., Cambridge, U. K. arc basin, Geo Mar. Lett., 10(2), 59–67.
Taylor, B. (1992), Rifting and the volcanic‐tectonic evolution Wright, I. C., L. M. Parson, and J. A. Gamble (1996), Evolu-
of the Izu‐Bonin‐Mariana Arc, Proc. Ocean Drill. Program tion and interaction of migrating cross‐arc volcanism and
Sci. Results, 126, 627–651. backarc rifting: An example from the southern Havre Trough
Taylor, B., and F. Martinez (2003), Back‐arc basin basalt sys- (35°20′–37°S), J. Geophys. Res., 101(B10), 22,071–22,086.
tematics, Earth Planet. Sci. Lett., 210(3–4), 481–497. Wysoczanski, R. J., I. C. Wright, J. A. Gamble, E. H. Hauri,
Taylor, B., K. Zellmer, F. Martinez, and A. M. Goodliffe J. F. Luhr, S. M. Eggins, and M. R. Handler (2006), Volatile
(1996), Sea‐floor spreading in the Lau back‐arc basin, Earth contents of Kermadec Arc‐Havre Trough pillow glasses:
Planet. Sci. Lett., 144(1–2), 35–40. Fingerprinting slab‐derived aqueous fluids in the mantle
Tollstrup, D., and J. Gill (2005), Hafnium systematics of the sources of arc and back‐arc lavas, J. Volcanol. Geotherm.
Mariana Arc: Evidence for sediment melt and residual Res., 152(1–2), 51–73.
phases, Geology, 33(9), 737–740. Wysoczanski, R. J., E. Todd, I. C. Wright, M. I. Leybourne,
Tollstrup, D., J. Gill, A. Kent, D. Prinkey, R. Williams, Y. Tamura, J. M. Hergt, C. Adam, and K. Mackay (2010), Backarc rift-
and O. Ishizuka (2010), Across‐arc geochemical trends in the ing, constructional volcanism and nascent disorganised
Izu‐Bonin arc: Contributions from the subducting slab, revis- spreading in the southern Havre Trough backarc rifts
ited, Geochem. Geophys. Geosyst., 11, Q01X10, doi:10.1029/ (SW Pacific), J. Volcanol. Geotherm. Res., 190(1–2), 39–57.
2009GC002847.
31 of 31

6 5 Todd G3

Uploaded by

Copyright:

Available Formats

6 5 Todd G3

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

6 5 Todd G3

Uploaded by

Copyright:

Available Formats

Geochemistry 3 Article

Geosystems Q04009, doi:10.1029/2009GC002888

Sources of constructional cross‐chain volcanism in the

Copyright 2010 by the American Geophysical Union 1 of 31

Components: 16,000 words, 14 figures, 5 tables.

1. Introduction components in arc magmas: aqueous fluids versus

et al., 2004; Marini et al., 2005; Chauvel et al.,

2. Tectonic and Petrologic Setting

SiO2 47.23 49.78b 48.98 49.16 49.56 50.47 51.13 50.54b

CaO/Al2O3 0.97 0.74 0.76 0.79 0.62 0.69 0.60 0.66

Sc 35 35 11 35d 29d 31d 28d 17

SiO2 64.91 65.91 64.2 62.23 63.37 58.52 64.31

CaO/Al2O3 0.11 0.18 0.59 0.30 0.34 0.73

Table 5. End‐Member Compositions Used in Mixing Modelsa

Sr 7.664 445.2 445.2 445.25

You might also like