Genesis of the Western Samoa seamount province: Age, geochemical fingerprint and tectonics

Abstract

The Samoan volcanic lineament has many features that are consistent with a plume-driven hotspot model, including the currently active submarine volcano Vailulu'u that anchors the eastern extremity. Proximity to the northern end of the Tonga trench, and the presence of voluminous young volcanism on what should be the oldest (∼5 my) western island (Savai'i) has induced controversy regarding a simple plume/hotspot model. In an effort to further constrain this debate, we have carried out geochronological, geochemical and isotopic studies of dredge basalts from four seamounts and submarine banks that extend the Samoan lineament 1300 km further west from Savai'i. 40Ar/39Ar plateau ages from Combe and Alexa Banks (11.1 my—940 km, and 23.4 my—1690 km from Vailulu'u, respectively) fit a Pacific age progression very well. The oldest volcanism (9.8 my) on Lalla Rookh (725 km from Vailulu'u) also fits this age progression, but a new age is much younger (1.6 my). Isotopically, these three seamounts, along with Pasco Bank (590 km from Vailulu'u), all lie within, or closely along extensions of, the Sr–Nd–Pb fields for shield basalts from the Eastern Samoan Province (Savai'i to Vailulu'u); this clearly establishes a Samoan pedigree for this western extension of the Samoan hotspot chain, and pushes the inception of Samoan volcanism back to at least 23 my. From geodetic reconstructions of the Fiji–Tonga–Samoa region, we show that the northern terminus of the Tonga arc was too far west of the Samoa hotspot up until 1–2 my ago to have been a factor in its volcanism. Young rejuvenated volcanism on Lalla Rookh and Savai'i may be related to the rapid eastward encroachment of the Trench corner. The Vitiaz Lineament, previously thought to mark a proto-Tongan subduction zone, was more likely created by the eastward propagation of the tear in the Pacific Plate at the northern end of the arc.

Full text

Genesis of the Western Samoa seamount province: age,
geochemical fingerprint and tectonics
S.R. Hart
a,
*, M. Coetzee
b
, R.K. Workman
a
, J. Blusztajn
a
, K.T.M. Johnson
c
,
J.M. Sinton
c
, B. Steinberger
d
, J.W. Hawkins
e
a
Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA
b
University of Cape Town, Cape Town, Rondebosch, 7701, South Africa
c
University of Hawaii at Manoa, Honolulu, HI 96822, USA
d
Japan Marine Science and Technology Center, Yokosuka 237-0061, Japan
e
Scripps Institution of Oceanography, La Jolla, CA 92093, USA
Received 11 March 2004; received in revised form 2 August 2004; accepted 5 August 2004
Editor: K. Farley
Abstract
The Samoan volcanic lineament has many features that are consistent with a plume-driven hotspot model, including the
currently active submarine volcano Vailulu’u that anchors the eastern extremity. Proximity to the northern end of the Tonga
trench, and the presence of voluminous young volcanism on what should be the oldest (~5 my) western island (Savai’i) has
induced controversy regarding a simple plume/hotspot model. In an effort to further constrain this debate, we have carried out
geochronological, geochemical and isotopic studies of dredge basalts from four seamounts and submarine banks that extend the
Samoan lineament 1300 km further west from Savai’i.
40
Ar/
39
Ar plateau ages from Combe and Alexa Banks (11.1 my—940
km, and 23.4 my—1690 km from Vailulu’u, respectively) fit a Pacific age progression very well. The oldest volcanism (9.8 my)
on Lalla Rookh (725 km from Vailulu’u) also fits this age progression, but a new age is much younger (1.6 my). Isotopically,
these three seamounts, along with Pasco Bank (590 km from Vailulu’u), all lie within, or closely along extensions of, the Sr–
Nd–Pb fields for shield basalts from the Eastern Samoan Province (Savai’i to Vailulu’u); this clearly establishes a Samoan
pedigree for this western extension of the Samoan hotspot chain, and pushes the inception of Samoan volcanism back to at least
23 my. From geodetic reconstructions of the Fiji–Tonga–Samoa region, we show that the northern terminus of the Tonga arc
was too far west of the Samoa hotspot up until 1–2 my ago to have been a factor in its volcanism. Young rejuvenated volcanism
on Lalla Rookh and Savai’i may be related to the rapid eastward encroachment of the Trench corner. The Vitiaz Lineament,
0012-821X/$ - see front matter D2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2004.08.005
* Corresponding author.
E-mail address: shart@whoi.edu (S.R. Hart).
Earth and Planetary Science Letters 227 (2004) 37 – 56
www.elsevier.com/locate/epsl

previously thought to mark a proto-Tongan subduction zone, was more likely created by the eastward propagation of the tear in
the Pacific Plate at the northern end of the arc.
D2004 Elsevier B.V. All rights reserved.
Keywords: Samoa; geochemistry; tectonics; 40/39 Ar ages; Vitiaz Lineament; Tonga Trench; hotspot volcanism
1. Introduction
The question of whether the Samoan volcanic
lineament is or is not a bplume-drivenQage-progressive
hotspot chain has been debated for decades. Hawkins
and Natland [1] and Natland [2] noted that the
voluminous young volcanism on Savai’i was incon-
sistent with a simple hotspot model, and suggested that
at least some of the volcanic timing was controlled by
tectonism related to proximity to the Tonga Trench, and
the transform zone bounding that subduction zone on
the north. Duncan [3] noted a reasonable age-pro-
gressive trend, when the ages of the seamounts to the
west of Savai’i are taken into account. His argument
implicitly invokes a Samoan bpedigreeQfor these
seamounts, and the available geochemical data were
consistent with this [4,5]. With the exception of
Savai’i, there is a consistent east to west aging of
volcano morphology, as witnessed by youthful
uneroded shield morphology at Ta’u, well-advanced
erosion of shield on Tutuila with the beginning of a
post-erosional veneer, and a major post-erosional
veneer on Upolu, covering deeply eroded shield
remnants. As noted by Natland [2], any possible shield
on Savai’i has been massively covered by post-
erosional volcanism. While accepting a byounging to
the eastQmodel, Natland and Turner [6] were unwilling
to accept this as proof of a hotspot model, arguing that
the volcanism could still be driven by thermo-
mechanical processes related to the corner of the
Tonga Trench. Natland [7] has presented a new model
for Samoa where the volcanism is all derived from
shallow sources, under control of plate fracture
mechanics. However, the discovery of the young active
submarine volcano Vailulu’u, anchoring the eastern
end of the chain, and well removed from the corner of
the Tonga trench, lend strong support to a basic plume
model. Furthermore, recent seismic tomography has
imaged a plume stem under Samoa, extending well into
the lower mantle [8]. This of course does not preclude a
strong brejuvenatedQstage of volcanism related to
proximity to the northern Tonga transform zone as
advocated by Hawkins and Natland [1].
The purpose of this paper is to firmly define a
Samoan chemical pedigree for the WESAM (Western
Samoan) seamounts, and to provide new age data
relevant to the question of age progression in the
WESAM province.
2. Physical setting
Conventionally, the Samoa hotspot lineament
stretches from the large subaerial island of Savai’i in
the west, to Ta’u Island in the east (see Fig. 1A).
Vailulu’u seamount (originally discovered and named
Rockne Volcano [9]) has recently been shown to be
volcanically active [10], and is thought to be the
current location of the Samoan hotspot. In addition,
the long ridge extending SE from Tutuila has been
swath-mapped, and culminates in a young volcano
named Malumalu (The Cathedral). Comprehensive
geochemical and isotopic data now exist for all of the
volcanoes of this bEastern Volcanic ProvinceQ[11–
13,2]. West of Savai’i, there are many seamounts and
submarine banks that may reflect continuation of the
Samoan lineament (Fig. 1B). However, these do not
define a single lineament, and it is uncertain which of
these may be part of the Samoan Chain, as opposed to
belonging to other older hotspot chains that may have
lineaments passing through this region (for example,
the Tuvalu lineament [14] or lineaments from the
Cook–Austral or Louisville hotspots [15–17]).
A number of these western seamounts were
dredged during the 1982 KK820316 cruise of the
R/V Kana Keoki, and initial petrographic, geo-
chronologic and major element analyses did support
a Samoan pedigree for some of these features [3–
5,18]. In addition, a nephelinite dredged from Pasco
Bank during the 1971 ANTIPODE cruise was shown
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5638

Fig. 1. (A, B) Location map for the WESAM (Western Samoa) seamounts relative to the subaerial islands (Savai’i, Upolu, Tutuila and Ta’u) and
submarine volcanoes (Muli, Malumalu and Vailulu’u) of Eastern Samoa. Papatua (PPT) and Uo Mamae (Machias) are isolated seamounts that
may or may not be related to the Samoa hotspot. Alexa Bank is not shown in panel B, but lies almost 78further WNW from Combe. The Vitiaz
Lineament is shown as a continuation of the northern termination of the Tonga Trench.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 39

to be geochemically similar to Samoan basalts [1].
We have fully re-inspected the dredge collection
from the Kana Keoki cruise, along with the existing
thin section collection, and selected the best avail-
able material from Lalla Rookh, Combe and Alexa
for further geochemical and geochronologic work.
The bathymetry and dredge locations for these
seamounts are given in Table 1, and more fully
described in Brocher [14] and Sinton et al. [4]. The
Pasco Bank setting is described in Hawkins and
Natland [1].
3. Techniques
Cut slabs were rough-crushed in plastic, and the
cleanest possible chips were handpicked and pow-
dered in agate. Major and trace elements were done on
these powders by combined XRF/ICP-MS at the
GeoAnalytical Lab, Washington State University.
Powders for isotopic analysis were leached in warm
6 N HCl for 1 h; Sr and Nd chemistry was done with
conventional ion chromatography, using DOWEX 50
cation resin, and HDEHP-treated teflon for Nd
separation [19]. Pb chemistry utilized the HBr–
HNO
3
procedure of Galer [20] and Abouchami et al.
[21], with a single column pass. Sr and Nd analyses
were done on the WHOI VG354 TIMS multi-
collector; these analyses carry internal precisions of
5–10 ppm; external precision, after adjusting to
0.710240 and 0.511847 for the SRM987 and La Jolla
Nd standards, respectively, is estimated to be 15–25
ppm. Pb isotopic analysis was done on the WHOI
NEPTUNE [64]. Pb analyses carry internal precisions
on XXX/204 (where XXX=206, 207 or 208) ratios of
15–30 ppm; SRM997 Tl was used as an internal
standard, and external reproducibility (including full
chemistry) ranges from 17 ppm for
207
Pb/
206
Pb, to
117 ppm for
208
Pb/
204
Pb [22]. Pb ratios were adjusted
to the SRM981 values of Todt et al. [23]; two separate
lots of the SRM981 standard were inter-compared and
any possible isotopic heterogeneity between these lots
was b10 ppm for XXX/204 ratios, and b1.5 ppm for
208
Pb/
206
Pb ratios.
4. Age–distance relationships
New high quality
40
Ar/
39
Ar step-release plateau
ages for Lalla Rookh, Combe and Alexa seamounts
are given in Table 2, along with earlier 40/39 total
fusion ages for other samples from these same dredges
[3]. Ages were measured on hand-picked holocrystal-
line fragments from the whole rock material. No glass
was encountered during the picking, and though the
dredge depths were all relatively deep (Table 1), we
do not see any evidence in the 40/39 plateau data for
excess argon. A summary of all Samoa age data as a
function of distance from Vailulu’u is shown in Fig. 2.
For Lalla Rookh, the new age is 8 my younger
than the prior age, and establishes a very long
eruptive history for this seamount; while the older
age is quite consistent with the expected age
progression, the young age clearly is not. In this
respect, it is similar to the young (0.08–0.8 my)
volcanism on Wallis Island, some 70 km SW of
Lalla Rookh [3,26] Unfortunately, there are no
geochemical data for the old sample (3-16), to test
if the large age gap is accompanied by significant
geochemical differences. The young sample has
similar major element characteristics to four other
basalts from the same dredge [4].
Table 1
Western Samoa seamounts, dredge locations
Dredge no. Seamount Setting of
dredge
Latitude Longitude Relief
(top–bottom)
Dredge
depth
(m)
Distance from
Vailulu’u
(km)
239 Pasco Bank SE flank 13.1438S 174.3008W 13–4800 m 1679–1567 579
3 Lalla Rookh Bank S slope 12.9858S 175.6358W 18–4400 m 2800–2400 723
7 Combe Bank SW ridge 12.7028S 177.6858W 25–3500 m 2800–2550 948
14 Alexa Bank W ridge 11.6858S 184.9538W 180–3300 m 3500–2800 1745
239: SIO Antipode, leg 16 [1].
3, 7, 14: Cruise KK820316, leg 2, R/V Kana Keoki [4].
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5640

For Combe, the new age is in reasonable agree-
ment with the earlier data (11.1 versus 14.1 my), and
together they straddle the expected age line.
For Alexa bank, the two new ages agree very
well (22.9 and 23.9 my), though the total fusion
ages are significantly older (as was the total fusion
age of 36.9 my reported by Duncan [3]). This
appears to remove the inconsistency in the older
data with respect to the age progression; the two new
ages fall very close to the 7.1 cm/year age progression
(Fig. 2).
In summary, insofar as Lalla Rookh, Combe and
Alexa can be shown to be of Samoan pedigree (see
below), the new data for Combe and Alexa strongly
support a simple age-progressive nature for the
Samoan lineament. At the same time, the young age
for Lalla Rookh (and Wallis) provides evidence for a
late stage of rejuvenation, perhaps related to tectonism
along the northern Tongan transform zone, as advo-
cated by Hawkins and Natland [1], and Natland and
Turner [6].
5. Geochemical characteristics
5.1. Alteration effects
Given that all of the WESAM basalts are
submarine dredge samples, and exposed to seawater
for times up to 24 my, the likelihood of weathering
and alteration effects must be considered. The
alkalis are among the most mobile elements under
these conditions [27], so that ratios such as Rb/Cs
and Rb/Ba may be used as balteration indicatorsQ.
Fig. 3 compares these ratios in WESAM basalts
with values from both submarine and subaerial
Samoan volcanoes, and with the bcanonicalQvalues
established for fresh oceanic basalts by Hofmann
and White [28], the 2rbounds of which are shown
in Fig. 3 by the rectangle. Only four of the
WESAM samples fall within the field of other
(younger and fresher?) Samoan basalts, and only
three fall close to the canonical field; the remainder
are well outside the rectangle and the Samoan field.
Submarine weathering typically involves addition of
alkalis, in the order CsNRbNK; Ba is less mobile
and more erratic [27,29]. This would cause trajecto-
ries down and to the right in Fig. 3 (note
bsubmarineQarrow); only two WESAM samples
have moved in that direction. Subaerial basalt
weathering typically leads to leaching of alkalis
[30], though the relative leaching behavior of Rb
and Cs has not been established (note bsubaerialQ
arrow). Curiously, seven of the WESAM samples
are in fact closely aligned along a Rb-mobility
trajectory (solid line in Fig. 3); four of the samples
exhibit marked Rb depletion. Location along this
line implies a constant Ba/Cs ratio during alteration,
and we know of no reason why Rb would be
mobile and not Cs. Comparing data for 58 subaerial
and 38 submarine basalts from the Eastern Volcanic
Province [11] shows no statistically significant
difference between basalts erupted above water or
underwater (Rb/Cs=96F9and117F12; Rb/
Table 2
40/39
Ar plateau ages, Western Samoa seamounts
Sample
number
Location Steps used/
total steps
39
Ar fraction
used
40/39 Total fusion
(my)
Weighted plateau
(my)
3–26 LALLA ROOKH 6/6 0–100% 1.63F0.06 1.62F0.05
3–16
a
LALLA ROOKH 9.8F0.3
7–100 COMBE 8/9 4.6–100% 11.03F0.07 11.12F0.06
7–11
a
COMBE 14.1F1.1
14–19 ALEXA 5/9 20.0–95.9% 34.01F0.55 23.94F0.36
14–100 ALEXA 4/8 8.2–67.4% 27.80F0.24 22.91F0.20
14–23
a
ALEXA 36.9F0.5
!Step-release heating from 600 to 1400 8C.
!2rerrors include measurement uncertainties, and uncertainty in J-value (flux gradient from FCT-3 biotite monitor), but not uncertainty in
monitor age.
!New ages are from the lab of Robert A. Duncan, Oregon State University (full details may be found in Supplementary materials).
a
These samples were reported by Duncan [3].
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 41

Ba=0.117F0.006 and 0.105F0.007, submarine ver-
sus subaerial, respectively, 2jstandard errors). We
have no ready explanation for the type of mobility
expressed in the WESAM basalts; though the tops
of these seamounts were certainly subaerially
exposed in their younger days, the samples were
dredged from depths well-below possible subaerial
exposure (Table 1).
In any event, the alkalis in the WESAM samples
have clearly been mobile, and this suggests caution in
using the alkali data, or that for other potentially
mobile elements such as U. Since Th is generally
insensitive to alteration, and Th/U is a relatively
constant ratio in OIBs, this ratio can provide added
information about the trace element reliability in the
WESAM basalts. U behaves much like the alkalis
during weathering, being added in submarine domains
[31] and leached in subaerial domains. Surprisingly,
the four samples from the oldest seamount, Alexa,
have normal and nearly constant Th/U (3.50–3.76,
Table 3). One sample from Lalla Rookh has very high
Th/U (8.9; sample 3-43), and one from Combe is very
low (2.1; sample 7-100); the other samples from these
seamounts are normal. This suggests that spidergram
patterns will in general be useful for petrogenetic
considerations.
The isotope data appear quite robust with respect to
alteration effects (note that this may in part be due to
the strong leaching that sample powders undergo prior
to analysis). Again, the four samples from the oldest
seamount, Alexa, while wildly dispersed in the Rb/
Fig. 3. Rb/Cs–Rb/Ba relationships of WESAM basalts, in compa-
rison to other Samoan basalts. The field encloses both shield and
post-erosional basalts from Vailulu’u, Ta’u, Muli, Malumalu and
Upolu, and post-erosional basalts from Upolu and Savai’i (four
widely scattered point are not enclosed; data from Workman et al.
[11]). The rectangle represents F2 standard deviations from the
mean (circled cross) of the canonical Rb/Ba and Rb/Cs values
recommended for oceanic basalts by Hofmann and White [28]. The
arrows represent general tendencies for subaerial and submarine
weathering (see text). The solid line represents a Rb-bmobilityQ
trajectory, drawn through the canonical value. Note that only four of
the WESAM basalts fall in the Samoan field (3-26, 3-36, 14-19,
239-1) and close to the canonical field; seven lie close to the Rb
mobility line.
Fig. 2. (A) Age–Distance relationships for bshieldQlavas from the
subaerial Samoan islands. Ta’u, Tutuila and Eastern Upolu data
(small lightly shaded squares) are K–Ar ages, from McDougall [24]
and Natland and Turner [6]. The Tutuila field shows the range of
data available, not all of the individual ages. Western Upolu and
Savai’i data (large darkly shaded squares) are
40/39
Ar plateau ages
[11]. (B) Age–Distance relationships for dredge basalts from the
WESAM Seamount province (the subaerial data from panel A is
also shown, small filled squares). Small lightly shaded squares are
40/39
Ar total fusion ages, from Duncan [3]; large darkly shaded
squares are
40/39
Ar bplateauQages, from Table 2. The dashed line is
for the 7 cm/year local Pacific plate velocity derived from the
REVEL and NUVEL-1A plate models [25].
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5642

Cs–Rb/Ba plot, are virtually constant in
87
Sr/
86
Sr,
with a maximum spread of only 0.02%; the same is
true of Combe. As Nd and Pb are generally less
affected by weathering and alteration than Sr, we
argue that all of the isotope data are reliable for the
purposes of establishing a Samoan signature.
Table 3
Geochemical data for Western Samoa Seamounts
Sample Pasco
a
ANT 239-1
Lalla Rookh Combe Alexa
3-26 3-36 3-43 7-17 7-100 7-102 14-15 14-19 14-23 14-100
SiO
2
39.75 41.61 42.93 48.33 46.62 45.56 46.96 50.03 50.15 48.82 49.07
Al
2
O
3
12.61 10.66 11.37 15.14 13.25 13.78 13.57 14.40 14.57 14.87 14.92
TiO
2
3.77 3.855 4.332 4.023 3.759 3.880 4.020 2.712 2.934 3.084 3.224
FeO* 12.19 13.49 13.66 11.39 13.71 14.18 12.14 11.34 12.74 12.47 12.36
MnO 0.15 0.193 0.185 0.129 0.305 0.220 0.169 0.139 0.151 0.274 0.341
MgO 11.95 12.67 9.58 5.75 6.96 6.00 8.93 6.84 6.61 6.95 6.56
CaO 13.97 11.13 12.58 9.46 10.08 12.10 8.18 10.66 8.98 8.80 8.79
Na
2
O 2.64 3.129 2.627 3.384 3.103 2.716 3.816 2.911 3.033 3.915 3.832
K
2
O 1.82 2.636 2.008 1.543 1.433 0.754 1.728 0.675 0.538 0.436 0.518
P
2
O
5
1.14 0.638 0.729 0.850 0.783 0.805 0.490 0.293 0.298 0.371 0.387
Pre-total 95.88 99.40 98.60 97.23 97.01 96.82 97.22 99.27 98.57 98.58 98.38
Mg# 0.673 0.663 0.595 0.514 0.516 0.470 0.607 0.559 0.521 0.539 0.527
Ni 280 265 246 69 319 171 248 65 52 96 98
Cr 463 373 373 74 502 586 270 103 32 142 129
V 250 292 344 303 356 356 302 318 349 333 347
Ga 20 20 24 23 21 17 25 22 20 21
Cu 65 59 114 64 101 148 92 117 78 92 96
Zn 150 131 136 162 136 123 115 103 123 92 84
Cs 0.802 0.795 0.702 0.406 0.221 0.043 1.208 2.068 0.056 0.143 0.124
Rb 48.7 76.2 50.6 21.50 12.68 5.21 27.70 18.08 6.79 2.86 3.61
Ba 644 707 604 499 209.1 164.2 221.7 102.8 85.3 129.9 127.5
Th 6.49 6.01 7.92 10.22 2.562 2.574 3.087 1.468 1.398 1.930 1.940
U 1.106 1.207 1.606 1.150 0.617 1.233 0.725 0.420 0.372 0.538 0.534
Nb 83.9 65.9 87.6 72.5 28.92 30.76 35.8 16.67 15.36 20.66 20.35
Ta 5.16 4.42 5.68 4.90 2.084 2.205 2.547 1.184 1.094 1.439 1.443
La 70.3 43.5 66.7 65.2 24.16 26.51 28.54 14.16 12.59 18.25 17.56
Ce 132.1 77.5 123.2 117.0 51.4 51.1 61.7 33.10 27.66 40.4 38.9
Pb 5.42 5.67 4.78 6.29 6.69 2.822 2.931 1.485 1.281 0.957 0.763
Pr 15.98 8.74 13.74 13.53 6.56 7.31 7.79 4.37 3.67 5.25 5.06
Nd 62.5 36.1 55.3 54.4 30.3 33.5 34.6 20.65 17.42 24.41 23.07
Sr 739 663 794 817 406 473 628 346 316 360 359
Zr 252.5 173.9 263.0 300.5 203.9 213.6 236.1 158.8 150.1 179.8 177.2
Hf 5.75 4.75 6.84 7.98 5.47 5.55 6.13 4.30 4.08 4.72 4.74
Sm 12.55 8.39 11.61 11.94 8.27 8.98 8.84 5.82 5.08 6.69 6.42
Eu 3.534 2.722 3.64 3.76 2.599 2.942 2.824 2.099 1.986 2.339 2.301
Gd 10.25 7.40 9.93 10.21 8.10 8.62 8.27 6.25 5.52 6.91 6.59
Tb 1.313 1.121 1.369 1.445 1.253 1.322 1.247 1.008 0.940 1.097 1.064
Dy 6.52 5.79 7.05 7.66 6.97 7.20 6.81 5.89 5.61 6.30 6.17
Ho 1.083 0.977 1.199 1.320 1.258 1.295 1.237 1.093 1.074 1.174 1.150
Y 34.20 25.12 30.06 37.16 32.13 33.20 31.14 27.88 26.99 30.01 29.77
Er 2.533 2.150 2.655 3.132 3.001 3.097 2.916 2.687 2.643 2.894 2.857
Tm 0.316 0.254 0.329 0.391 0.394 0.401 0.384 0.363 0.370 0.387 0.384
Yb 1.759 1.326 1.741 2.145 2.199 2.209 2.169 2.105 2.128 2.232 2.203
Lu 0.254 0.173 0.249 0.306 0.318 0.318 0.300 0.304 0.327 0.319 0.327
Sc 24.92 20.79 26.72 20.48 28.35 35.74 25.88 35.33 33.52 30.82 29.18
87
Sr/
86
Sr 0.705916 0.704430 0.704467 0.704931 0.704828 0.704698 0.704816 0.703489 0.703498 0.703591 0.703622
143
Nd/
144
Nd 0.512715 0.512795 0.512802 0.512812 0.512855 0.512819 0.512823 0.512964 0.512945 0.512920 0.512921
206
Pb/
204
Pb 18.9479 19.1674 19.6570 19.2688 18.9591 19.1730 18.9724 18.7431 18.7171 18.7727 18.7763
207
Pb/
204
Pb 15.5978 15.6189 15.6252 15.6143 15.5857 15.5878 15.5760 15.5273 15.5189 15.5302 15.5387
208
Pb/
204
Pb 39.2874 39.2042 40.0460 39.6267 39.0212 39.2519 39.0617 38.5409 38.4937 38.5771 38.5876
Major elements are calculated volatile-free; pre-total without volatiles; Mg# is calculated with FeO=0.85 FeO*.
a
Major element data for 239-1 is from Hawkins and Natland [1].
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 43

5.2. Classification and major elements
The alkali–silica–MgO variations are shown in
Fig. 4A,B. The single sample from Pasco Bank is a
nephelinite [1], and the most under-saturated of the
whole WESAM sample suite. The Lalla Rookh
samples vary widely in composition, ranging from
undersaturated high-alkali basanites (3-26) to mod-
erately differentiated alkali basalt (3-43). Three
samples analyzed by Sinton et al. [4] from the same
dredge are ankaramitic basanites; a fourth is a
basanite similar to our sample 3-26. Note that this
most alkalic sample is the one with the young plateau
age (1.62 my); the sample dated at 9.8 my by Duncan
[3] was not analyzed for major elements, but its K
2
O
content (1.2%) is clearly lower than that of sample 3-
26, possibly suggesting a typical (for Samoa) early
alkali basalt shield-building stage, followed after a
long hiatus by a highly undersaturated brejuvenatedQ
stage.
Our Combe samples are alkali basalts, though one
is differentiated and slightly hawaiitic. They continue
the unusual positive trend on the K
2
O–MgO plot (Fig.
4B) defined by our Lalla Rookh samples. They tend to
be more alkalic than the Sinton et al. samples from the
same dredge; one of these latter (7-11) was classified
as tholeiitic, but we believe this is a cpx accumulation
signature, and that this sample is innately alkalic, like
our (aphyric) samples. While we do not have cpx
analyses for any of these rocks, phenocrysts from
several E. Province basalts have very low Na
2
O+K
2
O
(b0.5%), and would pull whole rock compositions into
the tholeiitic field. At face value, the total fusion age
for the 7-11 btholeiiteQis older than our plateau age on
7-100 (14.1 versus 11.1 my), but we do not feel this
represents a Hawaiian-type bage transitionQfrom
tholeiite to alkali basalt.
Our four Alexa samples are all very similar in major
elements, and straddle the alkali basalt–tholeiite divid-
ing line; they are also all relatively low in MgO. The
two tholeiitic samples have the same low MgO as the
others, so have not simply been pulled into the tholeiitic
field by cpx accumulation. Three of our Alexa samples
are cut from the same samples as those analyzed by
Sinton et al. [4], and agree very well in composition.
The data for these WESAM province basalts can
be compared in Fig. 4 to data from both subaerial and
submarine samples from the Eastern Volcanic Prov-
ince (EVP) of Samoa (small filled symbols); only the
Pasco and Lalla Rookh samples plot well outside the
field of these EVP basalts. While the Alexa tholeiites
appear to be higher in SiO
2
than most of the EVP
tholeiites, it should be noted that many of these latter
tholeiites, especially those that are low in SiO
2
, are
Fig. 4. (A) Alkali–silica classification plot for new basalt data
(shaded symbols) from Pasco, Lalla Rookh, Combe and Alexa
seamounts (see legend in panel B). For comparison, data for other
samples from three of the same dredges are shown as unfilled
symbols [4], and all available data for Eastern Province basalts is
shown as small dots [2,11]. Note that our three Lalla Rookh samples
(shaded diamonds) are portions of the same three rocks reported by
Sinton et al. [4]; other than this, there is no overlap between the
sample suites. The alkali basalt/tholeiite dividing line is that of
Macdonald and Katsura [32]. The Eastern Province samples that
plot in the tholeiite field with less than 47% silica are all strongly
picritic. Dashed lines indicate the basalt classification of Le Bas et
al. [33]. (B) K
2
O–MgO variation diagram for basalts from the
WESAM and Eastern Province, as described in A. The high MgO
samples are all either picritic or ankaramitic.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5644

strongly picritic; in aphyric bstateQthey would plot
much closer to the Alexa tholeiites.
5.3. Isotopic signatures
It is well established that Samoan volcanism is of
EM2 character, and indeed is the most extreme
example of EM2, with
87
Sr/
86
Sr values as high as
0.7089 [10,11]). In Sr–Nd–Pb isotope bspaceQ, Samoa
occupies a unique domain relative to other mantle
end-members [34] and all other oceanic island basalts
(OIBs). It is even distinctive relative to the Society
hotspot, which is also strongly EM2 in character. For
example, Workman et al. [11] show that every
analyzed Samoan basalt has a higher
D8/4
Pb [35] than
any basalt from the Society hotspot. Isotopic finger-
prints then should allow us to ascertain the bpedigreeQ
of the WESAM seamounts with some certainty.
The WESAM isotope data are compared with fields
for the subaerial and submarine volcanoes of Samoa in
Fig. 5. Note first that post-erosional and shield basalts
in Samoa are isotopically distinct from each other, as
first pointed out by Wright and White [12]; the PE
basalts are consistently lower in
206
Pb/
204
Pb. The
nephelinite from Pasco Bank appears to be transitional
between shield and PE, in terms of Sr and Pb isotopes.
None of the other WESAM samples show any PE
tendencies in this isotope space. It is difficult to say
whether this argues against a rejuvenated stage of
volcanism in the WESAM province, or is just a
reflection of sampling limitations; we are dealing here
with samples from only one dredge for each of the four
WESAM seamounts.
In Sr and Pb isotopes, the four basalts from Alexa
are very tightly clustered. Two of the three Combe
basalts are similar, whereas a third sample resembles
sample 3-26 from Lalla Rookh in having higher
206
Pb/
204
Pb. The Lalla Rookh samples show a limited
range in Sr and Nd ratios, but are quite variable in Pb,
with sample 3-36 having the highest
206
Pb/
204
Pb yet
found for Samoa basalts. In the Sr–Pb plot (Fig. 5A),
this sample and all of the Alexa samples fall outside
any of the Samoan basalt fields, and on this basis
might be argued to be bnon-SamoanQ. However, in the
Pb–Pb plots (Fig. 5B,C), all of these samples are on
extensions of the tightly aligned Samoan shield
arrays. The Alexa samples are very close to the
Upolu and Tutuila Pago shield array, while the Combe
Fig. 5.
87
Sr/
86
Sr–
206
Pb/
204
Pb–
207
Pb/
204
Pb–
208
Pb/
204
Pb plot for the
WESAM seamounts, in comparison with Samoa shield and post-
erosional basalts. The Samoa data are divided into an Eastern
Province (Ta’u Island and Vailulu’u, Muli, and Malumalu
seamounts, all shield lavas [10–12]), Tutuila shields (southern and
younger Pago shield, and northern, older Masefau shield [12,11]),
the Upolu shield (both eastern and western shields [12,11]), and
post-erosional basalts from both Upolu and Savai’i [12,11]. The
WESAM seamounts appear closely affiliated with Samoa (and very
distinct from fields for EM1 and HIMU basalts [34], not shown).
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 45

samples and two of the three Lalla Rookh samples are
within the Samoan shield arrays. In general terms, the
Samoan shield isotope data require at least three
bmixingQcomponents. Workman et al. [11] argued for
four: a strongly enriched, high
87
Sr/
86
Sr end-member
(EM2), a depleted low
87
Sr/
86
Sr–low
206
Pb/
204
Pb
component (MORB mantle?), a high
3
He component
(FOZO), and a radiogenic Pb component (mild
HIMU). The PE lavas, with their high
D7/4
Pb, require
yet one more component, probably bsedimentaryQin
character. In this context, the isotopic signature of the
WESAM basalts is easily identifiable as Samoan in
general character. The Alexa basalts are extended
beyond the Samoan arrays precisely in the direction of
NMORB (upper) mantle, while the high
206
Pb/
204
Pb
sample from Lalla Rookh is extended toward HIMU.
The remaining WESAM samples fall within the
confines of the existing Samoa data. The principal
characteristic of the WESAM data, vis-a`-vis eastern
Samoa, is the general paucity of the enriched (high
87
Sr/
86
Sr) component; the highest
87
Sr/
86
Sr value is
0.7049 (Lalla Rookh 3-43). While this may be just a
reflection of the small sample set here, Ta’u Island, in
the Eastern Province, is similarly restricted in its
evidence for the enriched component (total range for
Ta’u basalts is 0.7044–0.7051). We conclude that the
WESAM basalts are indeed of Samoan pedigree,
differing only in the somewhat larger range of
206
Pb/
204
Pb, and the relative absence of the high
87
Sr/
86
Sr enriched component. These differences are
not unexpected in what is obviously a heterogeneous
mantle source and considering the much larger age
range embraced by the WESAM seamounts (22 my
versus only 3 my for the EVP).
5.4. Trace element signatures
The WESAM basalts can be similarly ‘‘finger-
printedQwith trace elements. Based on Workman et al.
[11], we have chosen a plot of Ba/Nb–Zr/Hf, Fig. 6,
Fig. 6. Ba/Nb–Zr/Hf plot for the WESAM seamounts, in comparison with data for Samoan shield and post-erosional basalts, and end-member
EM1 (Pitcairn) and HIMU (Tubuai and Mangaia) OIBs. The Pitcairn (EM1) data include both the Tedside (shield) series on the island, as well as
samples from the seamounts east of the island [36–38]. Two Pitcairn samples, with Zr/Hf of 56.7 and 65.1, have been omitted from this plot. The
HIMU data are from Tubuai and Mangaia Islands (Cook–Austral chain [39–41]). Average N-MORB [42] and PUM (primitive upper mantle;
[43]) are also shown. Note the clear distinction between shield and post-erosional basalts from Samoa, and the distinction between the Samoan
(EM2) basalts and the EM1 and HIMU fields. One Upolu sample (Ba/Nb~11.4), collected from the western (A‘ana) shield, plots well outside
the bshieldQfield on this plot, but is consistent with Upolu shield Pb data (Fig. 5). Again, the WESAM seamounts show clear kinship with
Samoa.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5646

as usefully distinguishing between Samoa and EM1
and HIMU basalts. This plot contrasts the slope of the
left, more incompatible, end of a spidergram (see Fig.
7) with that of the right, less incompatible, end.
Relative to the location of bulk earth (PUM), virtually
all of the OIBs on this plot have positive (depleted)
slopes for the left side of the spidergram, and
negative (enriched) slopes for the right side of the
spidergram (this bhumpedQshape spidergram is quite
evident in Fig. 7, see below). Fig. 6 also cleanly
separates Samoan shield and PE basalts, while
showing little distinction between the Upolu shield,
and shields of the Eastern Province. Relative to
Samoa, the basalts from Pitcairn (EM1) are higher in
Zr/Hf, while those from Mangaia and Tubuai (HIMU)
are lower in Ba/Nb; in this plot, there is relatively
little overlap between Samoa and the OIBs from end-
member EM1 and HIMU.
With only two exceptions, all of the WESAM
basalts are tightly clustered and plot within the field
for Eastern Province Samoan shield basalts. The
exceptions are samples 239-1 from Pasco and 3-26
from Lalla Rookh, which plot close to the Samoa PE
field. Both of these samples are under-saturated
basanites that plot as extremes in the alkali–silica
plot (Fig. 4A). Sample 3-26 has a
40
Ar/
39
Ar age of 1.6
my, which is significantly younger than the ~10 my
age expected from plate motion considerations, and
might suggest a younger rejuvenated stage of volcan-
ism on Lalla Rookh (though, as discussed above, this
sample was well within the shield basalt fields on the
Sr–Pb isotope plots).
The trace element patterns (spidergrams) for the
averaged Lalla Rookh, Combe and Alexa data are
compared to each other, and to average Ta’u Island
basalt, in Fig. 7. In trace elements, Ta’u is typical of the
Eastern Province basalts [11]; we use it here because it
has Sr and Nd isotopic signatures most like the
WESAM seamounts. In concert with their major
element characteristics, which range from basanites
Fig. 7. Trace element patterns (spidergrams) for WESAM basalts, normalized to primitive mantle. Each location is the average of the data given
in Table 3 for that location. Also shown for comparison is an average of 23 subaerial and submarine basalts from Ta’u Island, corrected to Mg#
73 by olivine addition [11]. Note that negative anomalies at U, K and Pb are typical of almost all OIB; the lack of a U anomaly for Combe and
Alexa may represent addition of U during seawater interaction.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 47

to tholeiites (Fig. 4A), Lalla Rookh has the most
enriched spidergram and Alexa the least enriched. The
Combe pattern is curiously smooth, without the
negative U, K and Pb anomalies shared by Lalla
Rookh and Ta’u; Alexa shows the K and Pb anomalies,
but not the U. The negative K anomalies are not related
to lithospheric interactions, as argued more generally
for OIBs [65], as virtually every Eastern Province
basalt, including those with high
3
He/
4
He, show the
same anomaly [11]. Since these are all elements that are
potentially mobile during alteration, some caution must
underpin these comparisons. Note also the presence of
a small negative Ba anomaly in the Ta’u pattern (and
indeed in virtually every basalt from Vailulu’u and
Malumalu as well), compared to all of the WESAM
basalts. It seems unlikely that this is an alteration effect,
unless alteration has coincidentally stopped just short
of generating any hint of a positive Ba anomaly.
5.5. Pb isotope versus distance correlations
Because the Eastern Province shield basalts form a
very regular increasing
206
Pb/
204
Pb trend with dis-
tance (younging) to the east [22,11], it is useful to
compare the WESAM basalts to this trend, Fig. 8.
With the exception of the Pasco sample, the other
three localities do show increasing
206
Pb/
204
Pb to the
east, though with considerable scatter. Pasco may
represent a bresettingQof this trend, or it may provide
an early start of the post-erosional trend shown in the
eastern volcanoes. Note that PE basalts from Tutuila,
Upolu and Savai’i form an array almost orthogonal to
the shield array, and Pasco would lie near an extension
of this array. We showed earlier that in Pb isotope
space, Pasco was transitional between the shield and
PE fields (Fig. 5); it also lies near the post-erosional
field on the trace element discriminant diagram, Fig.
6. The expected age of ~8 my for Pasco is also the
time when the Samoa plume may have changed its
bdriftQdirection (see below), and this could drive a
different bsampling mixQfrom what is clearly an
isotopically heterogeneous plume.
6. Tectonic setting during WESAM province time
6.1. Regional plate motions
Because of the controversy regarding possible
plume–trench interactions at present, we show in
Fig. 9 our best estimate of the tectonic situation during
the time of volcanic activity in the WESAM Province.
Because of the enormous present-day complexity in
this area, plate reconstructions in past times give only
qualitative information [44,45]. Consequently, we
have relied on published geodetic measurements to
backtrack several key features, such as the Tonga Arc,
the Fiji Platform and the Samoa Chain. Plate motion
models (e.g. NUVEL-1) over time scales of 2–3
million years have been shown to be in excellent
agreement [46] with current geodetic motions for the
major plates in this region (Pacific and Australia). On
the other hand, it is unlikely that the geodetically
determined motion of sub-domains (the Tonga Arc,
for example) can be confidently projected very far
back in time. Zellmer and Taylor [47] have shown a
persistence of Australia–Tonga vectors back to at least
0.8 my, but there are also indications that the opening
rate of the northern Lau basin may have slowed
significantly during subduction ~4 to ~2 my ago of
the Louisville Seamount Chain (LSC; [17]). A
summary of the geodetic data used in Fig. 9 is given
in Table 4.
Fig. 8.
206
Pb/
204
Pb of Samoan basalts as a function of distance from
Vailulu’u volcano (the presumed present location of the Samoa
hotspot). Shield basalts from the Eastern Province are shown as
solid diamonds; the error bars represent 2rstandard errors for the
data set given in Ref. [11]. The open circles are for post-erosional
basalts from Savai’i, Upolu and Tutuila; the error bars are as above.
Note that only one analysis of a PE basalt from Tutuila exists, and
that all of the Savai’i data are classified as PE despite many samples
analyzed from areas mapped as shield [11]. The open squares are
the individual analyses from Table 3, this paper. For reference, at the
accepted Pacific Plate motion velocity of 71.3 mm/year, the bageQat
Pasco would be 8.2 my.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5648

Currently, the Pacific Plate is moving N63W at 71
mm/year; the Australian Plate, New Caledonia and the
Fiji Platform are all moving ~N30E, but at different
rates (Australia and New Caledonia are converging on
Fiji); see Fig. 9B. The three stations on the Fiji
Platform are moving coherently, within the errors of
the GPS survey [49]. The Tonga Arc is rotating
rapidly to the east [49,63], with the transcurrent
motion between Tonga and the Pacific Plate occurring
somewhere in the vicinity of the Vitiaz Lineament.
Various Samoan hotspot features, including Savai’i
(S) and Upolu (U), are located on the Pacific Plate
north of the Vitiaz. Rotuma Island (R) and Futuna-Nui
Island (F; not to be confused with Futuna-Iti in
Vanuatu) lie ~50 km and ~150 km south, respectively,
of the Vitiaz, but are currently moving with the Pacific
Plate [48]. The locus of the northern terminus of the
Tonga Arc (NT), as imprinted on the Pacific Plate
over the age span 4 my to present, is shown in Fig. 9B
as a dotted line.
Going back in time, Fig. 9C and D, the Fiji
Platform has shifted mildly to the SW, and the
Samoan volcanoes have shifted markedly eastwards.
The Tonga Arc rotates back to the west, docking with
the Lau Ridge (an extinct arc) at ~4 my. Note that
this bdockingQage is somewhat younger than that
constrained by geological evidence, as the initial
rifting of the Lau Basin was underway at ~6.5 my,
and fully active seafloor spreading was underway by
4.5 my [50]. This discrepancy likely arises from our
extrapolation of current Tonga arc geodetic motions
back in time as being constant; if the rate between 5
and 3 my was taken as half of current rates, the
docking would occur at ~5 my, in better agreement
with the geologic evidence. As noted above, the
northern Lau Basin opening may also have slowed
during the encounter of the LSC with the northern
Tonga subduction zone. An increase of bdockingQage
only shifts the docking location further west (due to
Fiji–Lau Ridge motion to the SW), and exacerbates
the problem with the Vitiaz lineament, to be
discussed below.
The bback-trackedQnorthern terminus (NT) of the
Tonga arc (Fig. 9) passes very close to Futuna Island
at 2 my, and to Rotuma Island at 4 my. A single K–Ar
age of 4.9 my has been published for Futuna Island
[3], so it apparently existed prior to the bdrive-byQof
the Tonga arc. Tholeiites from Futuna may have arc-
like chemistry [4]; if substantiated, this would be
consistent with their formation above a south-dipping
Vitiaz arc (now bfossil arcQ). This would require that
Futuna has been transferred to the Pacific plate within
the past several million years. The inception age of
Rotuma Island is unknown; it does exhibit b1my
rejuvenated volcanism [26], but with no arc-like
geochemical characteristics.
This reconstruction appears to us to reveal several
infidelities in current models for this region. First, the
Vitiaz Lineament has been accepted as a proto-
Tongan or bfossilQsubduction zone, marking
Pacific–Australian convergence prior to the inception
(N7 my) of New Hebrides subduction and opening of
the Lau Basin [51, 44]. However, a comparison of
Fig. 9A and D shows that the Vitiaz Lineament is
already well east of the early Tonga subduction zone
when Tonga bundocksQfrom the Lau Ridge, and thus
it cannot represent a fossil subduction signature from
pre-Tongan subduction (though the segment west of
180 8W can be smoothly connected to the Tonga arc
at 4–5 my, Fig. 9D, and thus this Vitiaz segment is
consistent with identification as a bfossilQarc). On the
other hand, the Vitiaz segment east of 1808closely
mimics the trace of the northern terminus of the
Tonga Arc as it sweeps eastward (compare Fig. 9A
and B), and we propose that this is not a fossil
subduction zone but rather marks the locus of the
hinge of Pacific Plate tearing over the past 4–5 my.
This would be consistent with the plate-bending
model of Hawkins and Natland [1], for the interaction
between the Pacific Plate and the Tonga subduction
zone. Clear seismic evidence exists for present-day
tearing at the northern terminus of the Tonga Arc
[52], and the bathymetry in this area is not dissimilar
to that marking the contiguous Vitiaz Lineament to
the west.
The second problem concerns the NE motion of
the Fiji Platform and the bwedgeQof Australian Plate
between the Conway–Kandavu Lineament (CKL in
Fig. 9A; also called Hunter fracture zone) and the
Lau Ridge. This NE motion requires convergence
with the Pacific Plate, yet no obvious convergence
zone has ever been delineated in the North Fiji
Basin, or anywhere in the region between the Fiji
Platform and the Vitiaz Lineament (now attached to
the Pacific Plate). There is copious evidence for
divergent motions (small spreading centers) through-
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 49

out this region [50,53,45,47], and we are left to
somehow understand that the NE convergence
between Fiji–Australia and the Pacific plate has
been brectifiedQthrough a series of small spreading
centers, into an overall E–W divergence.
The Samoan plume center (Vailulu’u seamount
[10]) has apparently migrated NE over the past 4
my, as its location now lies well north of the
backtracked volcanic chain (Fig. 9D). Tutuila (Tu)
was the active center at 2 my (the oldest basalts on
Tutuila are ~1.5 my [6]), and Upolu (U) was
probably just becoming active at 4 my (the oldest
basalts on Upolu are 2.8 my [6,11]).
6.2. Plume–trench interactions
From Fig. 9, it seems very clear that the Tonga
Arc-Subduction zone likely had no impact on
Samoan volcanism in the WESAM Province. Five
million years ago, when Savai’i would have been
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5650

Fig. 9. Tectonic setting of the WESAM Seamounts in relation to the Fiji Plateau, Vitiaz Lineament and the Tonga Arc (map modified from Ruellan
et al. [17]). Location abbreviations: A—Alexa Bank; C—Combe Bank; F—Futuna Island; L—Lalla Rookh Bank; N—Niue Island; P—Pasco
Bank; R—Rotuma Island; S—Savai’i; T—Ta’u Island; Tu—Tutuila; U—Upolu; V—Vailulu’u Seamount; CKL—Conway–Kandavu lineament
(also called Hunter fracture zone; LLAU—Lakemba; NT—northern terminus, Tonga arc; NTPT—Niuataputapu; TGPU—Tongatapu; VANU—
Vanua Levu; VAVA—Vavaa`; VITI—Vitu Levu. Panel A shows the present-day setting, with extant geodetic stations labeled; see Table 4 for
derived plate motions. In panel B, we have connected the Tonga Arc locations and the Samoa chain features that appear to have a Samoan pedigree
(Alexa Bank is assigned to the chain, but is just off the figure to the west, and connected by a dashed line). The location picked for the northern
terminus of the Tonga Arc (NT) is the shallowest northernmost feature before the steep drop into the trench; a geodetic vector was derived for this
location by extrapolation from the three stations further south along the arc. Futuna and Rotuma Islands, though lying to the south of the Vitiaz
Lineament, are considered to move with the Pacific Plate [48]. Vectors are shown for the present-day motions of the Pacific Plate (71.3 mm/year),
the Australian Plate (61.2 mm/year) and the Fiji Plateau (34.1 mm/year), see Table 4 for references. Note that the Samoa chain does not lie along a
Pacific Plate motion direction, but is more westerly; this implies a slow northward component of migration of the hotspot over the past 20 million
years. The dotted line running WNW from the northern terminus of the Tonga Arc (NT) is the locus of NT positions over the past 4 millionyears, as
they might have been registered on the Pacific Plate. Panels C and D show the locations of the various features at 2 and 4 my before present, as
derived by backtracking the localities along the present geodetic vectors. The Vitiaz Lineament from panel A has been backtracked with the Pacific
Plate. Note that the northern terminus of the Tonga Arc passes Futuna Island at 2 my, and Rotuma Island at 4 my; the arc itself docks with the Lau
Ridge (Fiji Plateau) at close to 4 my. Note also that the present location of the hotspot, V, is well NE of the active part of the Samoa volcanic
lineament at 2 my (Tu, Tutuila) and 4 my (U, Upolu). At bUpolu TimeQ, the northern Tonga subduction zone was over 1200 km west of active
Samoan volcanism, so is unlikely to be implicated in the volcanism.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 51

at the hotspot location, the northern terminus (NT)
of the arc was more than 1400 km west of the
hotspot. Combe Bank, with its age of 11 my
(Table 2), had been in existence for more than 9
my when the NT passed by it ~1.8 my ago (Fig.
9C). Alexa Bank, 23 my old, was not adjacent to
the NT until ~4.5 my ago (Fig. 9D). However,
because of its very rapid eastward motion, the NT
is quickly bclosingQon the hotspot, and the
voluminous rejuvenated volcanism on Savai’i may
be the first witness to this interaction [1]. The
apparent drift of the hotspot to the NE between 2
my and the present may also signal this interaction
(the 100 km NE migration from Tu to V in 2 my
(Fig. 9C) represents a velocity of 50 mm/year).
As can be seen in Fig. 1A, the young end of the
Samoan chain exhibits a series of en echelon
lineaments, stepping off to the NE (one runs from
Savai’i through the western shield on Tutuila, one
runs from the eastern shield on Tutuila to Malu-
malu seamount and one runs from Muli seamount
through Ofu and Olosega islands to Ta’u Island;
Vailulu’u appears to represent the beginning of a
new lineament [11]. Assuming the hotspot is in fact
bplume-drivenQ, then a NE motion of the plume
conduit would be a natural dynamical consequence
Table 4
Geodetic plate motions in the SW Pacific
Locality Station
name
Latitude
(8S)
Longitude
(8W)
Velocity
(mm/year)
Azimuth Reference
Western Samoa (Faleolo, Upolu) WSAM 13.832 172.015 71.9 N63.8W [61]
Faleolo, W. Upolu FALE 13.832 172.000 71.8 N63.4W [61]
Faleolo, W. Upolu FALE 13.83 172.00 70.2 N64.2W [25]
Niue Island (outboard of Tonga trench) NIUC 19.062 169.932 71.3 N62.1W [61]
Pacific Plate average 71.3 N63.38W
Tonga Arc, N. terminus NT 14.95 173.50 192 E8.8S Extrapolated
Niuataputapu, N. Tonga Arc NTPT 15.947 173.764 170.5 E8.8S [49]
Vava’u, central Tonga Arc VAVA 18.585 173.960 134.3 E8.8S [49]
Tongatapu, central Tonga Arc TGPU 21.174 175.309 89.2F0.2 E5.7S [63]
Lakemba, Lau Ridge, Fiji LLAU 18.167 178.817 36.1 N22.2E [49]
Vanua Levu, Fiji (east coast) VANU 16.50 180.50 32.6 N26.5E [49]
Vitu Levu, Fiji (west coast) VITI 17.80 182.50 33.7 N39.5E [49]
Fiji average 34.1 N29.4E
Noumea, New Caledonia NOUM 22.270 193.590 49.9 N24.1E [61]
Noumea, New Caledonia NOUM 22.268 193.593 48.8 N28.9E [62]
Noumea, New Caledonia NOUM 22.27 193.59 48.3 N26.8E [25]
New Caledonia average 49.0 N26.6E
Townsville, Australia TOW2 19.270 212.945 61.4 N28.0E [61]
Townsville, Australia TOW2 19.268 212.943 60.7 N30.8E [62]
Townsville, Australia TOW2 19.27 212.94 61.6 N29.7E [25]
Townsville average 61.2 N29.5E
Pacific Plate locations
Vailulu’u Seamount V 14.215 169.058
Ta’u Island T 14.25 169.47
Tutuila Island Tu 14.30 170.70
Upolu Island (center) U 13.90 171.77
Savai’i Island (center) S 13.62 172.43
Pasco Bank (center) P 13.08 174.37
Lalla Rookh Seamount (center) L 12.93 175.65
Wallis Island W 13.28 176.18
Combe Seamount C 12.38 177.52
Futuna-Nui Island [48] FTNA 14.31 178.12
Rotuma Island R 12.50 182.90
Alexa Bank A 11.53 184.42
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5652

of the eastward roll-back motion of the trench and
subduction zone (and rapid opening of the Lau
backarc basin). Laboratory experiments clearly
show horizontal flow underneath and around the
edge of a plate during roll-back, with concomitant
down-flow in the mantle-wedge [54]. If there is any
vertical component to this flow during escape from
underneath the subducting plate, the decompression
may induce minor melting that could explain some
of the anomalous boff-chainQvolcanism in Samoa
(Uo Mamae and Papatua Seamounts, Fig. 1A). This
would be an adaptation of the badakiteQmodel of
Yogodzinski et al. [55] that posits melting whenever
mantle is forced to flow around the edge of a torn
subducting slab (the difference being that Uo
Mamae seamount, at least, does not have barcQ
chemistry [1,56,57]).
It is interesting to note that the plume motion
models of Steinberger [58,59] provide some support
for a NE motion for the Samoa hotspot over the past
25–30 my. These models calculate advective motions
of plumes in a realistic mantle flow field, driven by
surface plate motions and internal density heteroge-
neities inferred from seismic tomography. Fig. 10
shows the surface motion for the Samoan plume,
derived for a plume initiation age of 40 my, and using
the tomographic and plate models described in [60].
The plume appears to have made a large clockwise
bhookQ, drifting roughly along the plate motion
direction until 17 my ago, then reversing direction
and drifting counter to plate motion. Overall, from 30
to 3 my, there has been a persistent NE drift of ~100
km, at a rate of about 5 mm/year.
While this motion would not explain the en
echelon trends discussed above that have occurred
over the past 2–3 my, it accounts for part of the misfit
between the trend of the Samoa Chain and the
direction of Pacific plate motion. In Fig. 9B, it is
clear that the Samoan chain, as drawn from Savai’i
through Pasco and Lalla Rookh to Combe, delineates
a shallower azimuth (N77W) than is specified for the
Pacific Plate (N63.4W; see vector marked 71.3). This
misfit problem was also noted by Brocher [14]. There
are several possible explanations. First, we have
chosen, based on the geochemistry discussed above,
to assign Pasco, Lalla Rookh and Combe a Samoan
pedigree (i.e. they belong to the Samoan chain). There
are other banks and seamounts in this region (e.g.
Field and Horseshoe, see Fig. 1A) which might better
fit the actual Pacific plate motion vector, if all were to
show a Samoan pedigree. Alternatively, we have both
the evidence from the dynamical plume-drift model
(Fig. 10), and the en echelon arrangement of the active
end of the chain, which suggest a north to northeast
drift of the plume. Such a drift would generate a
volcanic chain or lineament that was shallower in
azimuth than the actual plate motion vector.
Resolution of this question must await both
better dynamic modeling of the Samoa plume
conduit (with full definition of the slab roll-back),
and the isotopic characterization of all of the
seamounts and banks bdownstreamQfrom the
hotspot, to delimit those that are clearly Samoan
and that validly may be used to define the chain. It
must be recognized that several other hotspots may
have left btracksQthrough this region (e.g. Cook–
Austral, Louisville [15–17]), and a robust geo-
chemical fingerprinting of the myriad seamounts in
this region is crucial to fully understanding the
history of the Samoan hotspot.
Finally, the persistent E to SE drift shown in Fig.
10 over the past 15 my is counter to plate motion, and
would have the interesting effect of increasing the
bapparentQvelocity of the Samoan volcanic age
Fig. 10. Surface motion of the Samoa plume over the past 40 my,
derived from a mantle dynamics model driven by surface plate
motions and internal density heterogeneities. The details of this
model can be found in Steinberger [59] and Steinberger et al. [60].
The initiation age assumed for the Samoan plume is 40 my;
numbers at various points along the curve are ages in millions of
years. The filled circle marked Vailulu’u is the present position of
the plume [10]. For geographic reference, the locations of Tutuila
and Ta’u Islands are also shown. The arrow marks the azimuth and
distance for 1 my of present Pacific plate motion (see Table 4).
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 53

progression on the Pacific Plate. This drift rate,
averaged over the past 15 my, is 12 mm/year and
would increase the bage progressionQfrom 71 to 83
mm/year (creating a shallower slope in Fig. 2A, and a
better fit to the Tutuila and Upolu age data).
7. Summary
!Basalts dredged from four submarine volcanic
constructs west of Savai’i (Pasco, Lalla Rookh,
Combe and Alexa seamounts), in the Samoan
hotspot chain, are shown to have trace element and
Sr, Nd and Pb isotopic signatures consistent with
derivation from the Samoan hotspot.
!New
40
Ar/
39
Ar plateau ages on Combe and Alexa
basalts (11.1 and 22.9–23.9 my, respectively) fit an
age progression model for Samoa, with the
observed Pacific Plate velocity of 71.3 mm/year.
!A previous age for Lalla Rookh (9.8 my, Duncan
[3]) also fits this age progression; we find a
younger age of 1.6 my for a highly under-saturated
basanite from the same dredge, suggesting a history
of volcanic rejuvenation on Lalla Rookh.
!Geodetic reconstructions of the northern terminus
of the Tonga arc/trench suggest that the young
rejuvenated volcanism on Lalla Rookh and Savai’i
may relate to interaction with the eastward migra-
tion of the trench corner (currently almost due
south of Savai’i, and migrating rapidly eastward at
N190 mm/year). We suggest that the Vitiaz Linea-
ment is the trace of the eastward motion of this tear
in the Pacific Plate.
!The volcanoes of the Eastern Samoan Province
show an en echelon alignment, with the volcanic
activity stepping to the northeast over the past 1–2
my. We suggest this is due to interaction of the
Samoan plume with northward flow of mantle that
is escaping from beneath the subducting Tonga
plate, as it rolls-back to the east.
Acknowledgements
We are grateful to Jim Natland for preprints of
several unpublished papers, and for his unfailing
support of the anti-plume model. Matt Jackson’s
passion for the Samoa plume model has been a
continuous source of stimulation. Samoa is also
special because of time spent in Tisa’s with Hubert
Staudigel and Anthony Koppers. We thank Steve
Galer and Wafa Abouchami for help in implement-
ing the Mainz Pb chemistry at WHOI. The output
of high-precision Pb data from the WHOI NEP-
TUNE is due largely to Lary Ball’s skill and
tenacity; our many thanks. This research was
supported by NSF EAR-0125917 (to SRH). This
is SOEST contribution number 6484. We recovered
from the fire of October 2002, and the coating of
Pb it left in the clean labs, thanks to the unceasing
help of Ernie Charette and the many other
professionals at WHOI. We thank Bill White and
Dave Graham for their reviews, and Hiroshi
Munekane, from the Geographical Survey Institute
of Japan, for the unpublished geodetic data for
Tongatapu.
Appendix A. Supplementary Material
Supplementary data associated with this article can
be found, in the online version, at doi:10.1016/
j.epsl.2004.08.005.
References
[1] J.W. Hawkins, J.H. Natland, Nephelinites and basanites of
the Samoan linear volcanic chain: their possible tectonic
significance, Earth Planet. Sci. Lett. 24 (1975) 427 – 439.
[2] J. Natland, The progression of volcanism in the Samoan linear
volcanic chain, Am. J. Sci. 280-A (1980) 709 – 735.
[3] R.A. Duncan, Radiometric ages from volcanic rocks along
the New Hebrides–Samoa lineament, in: T.M. Brocher
(Ed.), Investigations of the Northern Melanesian Border-
land, Circum-Pacific Council for Energy and Mineral
Resources Earth Science Series, vol. 3, 1985, pp. 67– 75.
[4] J.M. Sinton, K.T.M. Johnson, R.C. Price, Petrology and
geochemistry of volcanic rocks from the Northern Melanesian
Borderland, in: T.M. Brocher (Ed.), Investigations of the
Northern Melanesian Borderland, Circum-Pacific Council for
Energy and Mineral Resources Earth Science Series, vol. 3,
1985, pp. 35 – 65.
[5] K.T.M. Johnson, J.M. Sinton, R.C. Price, Petrology of
seamounts northwest of Samoa and their relation to
Samoan volcanism, Bull. Volcanol. 48 (1986) 225 – 235.
[6] J.H. Natland, D.L. Turner, Age progression and petrolog-
ical development of Samoan shield volcanoes: evidence
from K–Ar ages, lava compositions and mineral studies,
in: T.M. Brocher (Ed.), Investigations of the Northern
Melanesian Borderland, Circum-Pacific Council for Energy
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5654

and Mineral Resources Earth Science Series, vol. 3, 1985,
pp. 139 – 172.
[7] J.H. Natland, The Samoan Chain: A Shallow Lithospheric
Fracture System, www.mantleplumes.org, 2004.
[8] R. Montelli, G. Nolet, F.A. Dahlen, G. Masters, E.R. Engdahl,
S.-H. Hung, Finite-frequency tomography reveals a variety of
plumes in the mantle, Science 303 (2004) 338 – 343.
[9] R.H. Johnson, Exploration of three submarine volcanoes in
the South Pacific, Res. Rep.-Natl. Geogr. Soc. 16 (1984)
405 – 420.
[10] S.R. Hart, H. Staudigel, A.A.P. Koppers, J. Blusztajn, E.T.
Baker, R. Workman, M. Jackson, E. Hauri, M. Kurz, K. Sims,
D. Fornari, A. Saal, S. Lyons, Vailulu’u undersea volcano: the
new Samoa, Geochem. Geophys. Geosyst. GC000108 (2000)
1 – 13.
[11] R.K. Workman, S.R. Hart, M. Jackson, M. Regelous, K.
Farley, J. Blusztajn, M. Kurz, H. Staudigel, Recycled
Metasomatized Lithosphere as the Origin of the Enriched
Mantle II (EM2) End-member: evidence from the Samoan
Volcanic Chain, Geochem. Geophys. Geosyst. 5 (2004)
DOI:2003GC000623.
[12] E. Wright, W.M. White, The origin of Samoa: new evidence
from Sr, Nd and Pb isotopes, Earth Planet. Sci. Lett. 81 (1986/
87) 151 – 162.
[13] K.A. Farley, J.H. Natland, H. Craig, Binary mixing and
enriched and undegassed (primitive?) mantle components (He,
Sr, Nd, Pb) in Samoan lavas, Earth Planet. Sci. Lett. 111
(1992) 193 – 199.
[14] T.M. Brocher, On the age progression of the seamounts west of
the Samoan islands, S.W. Pacific, in: T.M. Brocher (Ed.),
Investigations of the Northern Melanesian Borderland, Circum-
Pacific Council for Energy and Mineral Resources Earth
Science Series, vol. 3, 1985, pp. 173 – 185.
[15] C. Gaina, R.D. Mqller, S.C. Cande, Absolute plate motion,
mantle flow and volcanism at the boundary between the
Pacific and Indian Ocean mantle domains since 90 Ma, in:
M.A. Richards, R.G. Gordon, R.D. van de Hilst (Eds.), The
History and Dynamics of Global Plate Motions, Geo-
physical Monograph, vol. 121, Am. Geophys. Union,
2000, pp. 189 – 210.
[16] R.D. Mqller, C. Gaina, A. Tikku, D. Mihut, S.C. Cande, J.M.
Stock, Mesozoic/Cenozoic tectonic events around Australia,
in: M.A. Richards, R.G. Gordon, R.D. van de Hilst (Eds.),
The History and Dynamics of Global Plate Motions, Geo-
physical Monograph, vol. 121, Am. Geophys. Union, 2000,
pp. 161 – 188.
[17] E. Ruellan, J. Delteil, I. Wright, T. Matsumoto, From rifting to
active spreading in the Lau Basin–Havre Trough backarc
system (SW Pacific): locking/unlocking induced by seamount
chain subduction, Geochem. Geophys. Geosyst. 4 (2003)
GC000261.
[18] K.T.M. Johnson, The Petrology and Tectonic Evolution of
Seamounts and Banks of the Northern Melanesian Borderland,
Southwest Pacific, Unpubl. M.Sc, University of Hawaii, 1983,
116 pp.
[19] B.D. Taras, S.R. Hart, Geochemical evolution of the New
England seamount chain: isotopic and trace element con-
straints, Chem. Geol. 64 (1987) 35 – 54.
[20] S.J.H. Galer, Chemical and isotopic studies of crust–mantle
differentiation and the generation of mantle heterogeneity,
Unpubl. PhD, University of Cambridge, 1986.
[21] W. Abouchami, S.J.G. Galer, A. Koschinsky, Pb and Nd
isotopes in NE Atlantic Fe–Mn crusts: proxies for trace metal
paleosources and paleocean circulation, Geochim. Cosmo-
chim. Acta 63 (1999) 1489 – 1505.
[22] S.R. Hart, R.K. Workman, M. Coetzee, J. Blusztajn, L. Ball,
K.T.M. Johnson, The Pb isotope pedigree of Western Samoan
volcanics: new insights from high-precision analysis by
NEPTUNE ICP/MS, EOS 83 (2002) F20.
[23] W. Todt, R.A. Cliff, A. Hanser, A.W. Hofmann, Evaluation of
a
202
Pb–
205
Pb double spike for high-precision lead isotope
analysis, in: A. Basu, S.R. Hart (Eds.), Earth Processes:
Reading the Isotopic Code, Geophysical Monograph, vol. 95,
1996, pp. 429 – 437.
[24] I. McDougall, Age and evolution of the volcanoes of Tutuila,
American Samoa, Pac. Sci. 39 (1987) 311– 320.
[25] G.F. Sella, T.H. Dixon, A. Mao, l. REVEL: a model for recent
plate velocities from space geodesy, J. Geophys. Res. 107
(ETG 11) (2002) 1 – 32.
[26] R.C. Price, L.E. Johnson, A.J. Crawford, Basalts of the North
Fiji Basin: the generation of back arc basin magmas by mixing
of depleted and enriched mantle sources, Contrib. Mineral.
Petrol. 105 (1990) 106 – 121.
[27] S.R. Hart, K, Rb, Cs contents and K/Rb, K/Cs ratios of fresh
and altered submarine basalts, Earth Planet. Sci. Lett. 6 (1969)
295 – 303.
[28] A.W. Hofmann, W.M. White, Ba, Rb, Cs in the Earth’s mantle,
Z. Naturforsch. 38a (1983) 256 – 266.
[29] J.A. Philpotts, C.C. Schnetzler, S.R. Hart, Submarine basalts:
some K, Rb, Sr, Ba, rare earth, H
2
O and CO
2
data bearing on
their alteration, modification by plagioclase, and possible
source materials, Earth Planet. Sci. Lett. 7 (1969) 293 – 299.
[30] M.D. Feigenson, A.W. Hofmann, F.J. Spera, Case studies on
the origin of basalt: II. The transition from tholeiitic to alkalic
volcanism on Kohala volcano, Hawaii, Contrib. Mineral.
Petrol. 84 (1983) 390 – 405.
[31] S.R. Hart, H. Staudigel, Isotopic characterization and identi-
fication of recycled components, 15–28, in: S.R. Hart, L.
Gulen (Eds.), Crust/Mantle Recycling at Convergence Zones,
NATO ASI Series C, vol. 258, Kluwer Academic Publishers,
Dordrecht, The Netherlands, 1989.
[32] G.A. Macdonald, T. Katsura, Chemical composition of the
Hawaiian lavas, J. Petrol. 5 (1964) 83 – 133.
[33] M. LeBas, R.W. Maitre, A. Streckeisen, B. Zanettin, A
chemical classification of volcanic rocks based on the total
alkali–silica diagram, J. Petrol. 27 (1986) 745 – 750.
[34] A. Zindler, S.R. Hart, Chemical geodynamics, Annu. Rev.
Earth Planet. Sci. 14 (1986) 493 – 571.
[35] S.R. Hart, A large-scale isotopic anomaly in the southern
hemisphere mantle, Nature 309 (1984) 753 – 757.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–56 55

[36] J.D. Woodhead, C.W. Devey, Geochemistry of the Pitcairn
seamounts: I. Source character and temporal trends, Earth
Planet. Sci. Lett. 116 (1993) 81 – 99.
[37] J. Eisele, M. Sharma, S.J.G. Galer, J. Blichert-Toft, C.W.
Devey, A.W. Hofmann, The role of sediment recycling in EM-
I inferred from Os, Pb, Hf, Nd, Sr isotope and trace element
systematics of the Pitcairn hotspot, Earth Planet. Sci. Lett. 196
(2002) 197 – 212.
[38] Hart, Hauri and Farley, unpublished.
[39] J.D. Woodhead, Extreme HIMU in an oceanic setting: the
geochemistry of Mangaia Island (Polynesia), and temporal
evolution of the Cook–Austral hotspot, J. Volcanol. Geoth.
Res. 72 (1996) 1 – 19.
[40] E.H. Hauri, S.R. Hart, Re–Os isotope systematics of HIMU
and EMII oceanic island basalts from the south Pacific Ocean,
Earth Planet. Sci. Lett. 114 (1993) 353 – 371.
[41] E.H. Hauri, S.R. Hart, Rhenium abundances and systematics
in oceanic basalts, Chem. Geol. 139 (1997) 185 – 205.
[42] A.W. Hofmann, Chemical differentiation of the Earth: the
relationship between mantle, continental crust and oceanic
crust, Earth Planet. Sci. Lett. 90 (1988) 297 – 314.
[43] W.F. McDonough, S.S. Sun, The composition of the earth,
Chem. Geol. 120 (1995) 223 – 253.
[44] C.Y. Yan, L.W. Kroenke, A plate tectonic reconstruction of the
Southwest Pacific, 0–100 Ma, in: W.H. Berger, L.W. Kroenke,
L.A. Mayer, et al. (Eds.), Proceedings of the Ocean Drilling
Program. Scientific Results, vol. 130, 1993, pp. 697 – 709.
[45] B. Pelletier, S. Calmant, R. Pillet, Current tectonics of the
Tonga–New Hebrides region, Earth Planet. Sci. Lett. 164
(1998) 263 – 276.
[46] C. DeMets, R.G. Gordon, D.F. Argus, S. Stein, Effect of
recent revisions to the geomagnetic reversal time scale on
estimates of current plate motions, Geophys. Res. Lett. 21
(1994) 2191 – 2194.
[47] K.E. Zellmer, B. Taylor, A three-plate kinematic model for
Lau Basin opening, Geochem. Geophys. Geosyst. 2 (2001)
DOI:2000GC000106.
[48] S. Calmant, B. Pelletier, P. Lebellegard, M. Bevis, F.W. Taylor,
D.A. Phillips, New insights on the tectonics along the New
Hebrides subduction zone based on GPS results, J. Geophys.
Res. 108 (2003) 1 – 22 (ETG 17).
[49] M. Bevis, F.W. Taylor, B.E. Schutz, J. Recy, B.L. Isacks, S.
Helu, R. Singh, E. Kendrick, J. Stowell, B. Taylor, S. Calmant,
Geodetic observations of very rapid convergence and back-arc
extension at the Tonga arc, Nature 374 (1995) 249 – 251.
[50] J.W. Hawkins, Evolution of the Lau Basin: insights from ODP
135, in: B. Taylor, J. Natland (Eds.), Active Margins and
Marginal Basins of the Western Pacific, Geophys. Monog.
Ser., AGU, Washington, DC, 1995, pp. 125 – 173.
[51] T.M. Brocher, On the formation of the Vitiaz trench lineament
and North Fiji basin, in: T.M. Brocher (Ed.), Investigations of
the Northern Melanesian Borderland, Circum-Pacific Council
for Energy and Mineral Resources Earth Science Series, vol. 3,
1985, pp. 13 – 33.
[52] D.W. Millen, M.W. Hamburger, Seismological evidence for
tearing of the Pacific plate at the northern termination of the
Tonga subduction zone, Geology 26 (1998) 659– 662.
[53] B. Pelletier, Y. Lagabreille, M. Menoıˆt, G. Cabioch, S.
Calmant, E. Garel, C. Guivel, J. Perrier, Newly discovered
active spreading centers along the North Fiji transform zone
(Pacific–Australia Plate boundary): preliminary results of the
R/V l’Atalante Alaufi Cruise (February–March 2000), Ridge
Events, 2001, pp. 7 – 9 Feb.
[54] C. Kincaid, R.W. Griffiths, Laboratory models of the thermal
evolution of the mantle during rollback subduction, Nature
425 (2003) 58 – 62.
[55] G.M. Yogodzinski, J.M. Lees, T.G. Churikova, F. Dorendorf,
G. Wferner, O.N. Volynets, Geochemical evidence for the
melting of subducting oceanic lithosphere at plate edges,
Nature 409 (2001) 500 – 504.
[56] A.Y. Sharaskin, I.K. Pustchin, S.K. Zlobin, G.M. Kolesov,
Two ophiolite sequences from the basement of the Northern
Tonga arc, Ofioliti 8 (1983) 411– 430.
[57] S.K. Zlobin, G.M. Kolesov, N.N. Kononkova, Development of
within-plate magmatism on the landward and offshore slopes
of the Tonga Trench, Ofioliti 16 (1991) 17 – 35.
[58] B. Steinberger, Plumes in a convecting mantle: models and
observations for individual hotspots, J. Geophys. Res. 105
(2000) 11127 – 11152.
[59] B. Steinberger, Motion of the Easter hot spot relative to
Hawaii and Louisville hotspots, G-Cubed 3 (2002) GC000334.
[60] B. Steinberger, R. Sutherland, R.J. O’Connell, Prediction of
Emperor-Hawaii seamount locations from a revised model of
plate motion and mantle flow, Nature 430 (2004) 167– 173.
[61] J. Beavan, P. Tregoning, M. Bevis, T. Kato, C. Meertens,
Motion and rigidity of the Pacific Plate and implications for
plate boundary deformation, J. Geophys. Res. 107 (2002)
1 – 15 (ETG 19).
[62] P. Tregoning, Plate kinematics in the western Pacific derived
from geodetic observations, J. Geophys. Res. 107 (2002) 1– 8
ECV 7.
[63] H. Munekane, Geographical Survey Institute of Japan,
personal communication, June 2004.
[64] S.R. Hart, R.K. Workman, L. Ball, J. Blusztajn, High Precision
Pb Isotope Techniques from the WHOI NEPTUNE PIMMS
WHOI Plasma Facility Open File Technical Report, 10, 2004
bhttp://www.whoi.edu/science/GG/people/shart/Open_File/
open_file.htmN.
[65] C. Class, S.L. Goldstein, Plume–lithosphere interactions in the
ocean basins: constraints from the source mineralogy, Earth
Planet. Sci. Lett. 150 (1997) 245 – 260.
S.R. Hart et al. / Earth and Planetary Science Letters 227 (2004) 37–5656