Download evidence from Th, Sr and Pb isotope data for Tonga

EPSL
ELSEVIER
Earth and Planetary
Science Letters 150 ( 1997) 29 l-302
Trace element transport rates in subduction zones: evidence from
Th, Sr and Pb isotope data for Tonga-Kermadec arc lavas
M. Regelous
Department
*,
K.D. Collerson, A. Ewart, J.I. Wendt
of Earth Sciences. The lh~i~~er.~it~of Queensland, Brisbane, Qld. 4072. Austmlin
Received 3 February
1997; revised 26 May 1997; accepted
12 June 1997
Abstract
Trace element and Th, Sr and Pb isotope data for young lavas from the Tonga-Kermadec
arc in the southwest Pacific
suggest that geochemical variations in the lavas along the arc are linked to differences in the material being subducted
beneath the arc. Lavas from the southern (Kermadec) segment of the arc have relatively radiogenic Pb isotope compositions,
which reflects a contribution from subducted sediment. In contrast, much of the Pb in Tonga lavas is derived from the
altered oceanic crust in the subducting Pacific Plate, and lavas from the northernmost Tonga islands of Tafahi and
Niuatoputapu contain Pb and Sr derived from the subducted part of the Louisville Seamount Chain. The origin of the Pb in
the lavas from these two islands can thus be traced to a point on the subducting slab, and this observation is used to estimate
the rate at which trace elements are transported beneath the arc. Our calculations suggest that fluid-soluble elements such as
U, Sr and Pb are transported from the subducted slab, across the mantle wedge and back to the surface in lavas over a period
of approximately 2-3 Ma, and that magmas are erupted at the surface less than 350 ka after the melts are generated in the
mantle wedge. 6 1997 Elsevier Science B.V.
Keywords:
island arcs; subduction:
isotopes; trace elements;
magmas;
1. Introduction
Subduction-related
magmatism is thought to occur
in response to the addition of fluids derived from
dehydrating sediments and/or altered oceanic crust
in the subducting slab to the sub-arc mantle wedge
[l-4]. Fluid-soluble
trace elements are transported
from the slab into the mantle wedge, and magmas
are produced when these fluids reach mantle that is
hot enough to melt. However, the rate at which these
processes occur are not well understood.
Recent
* Corresponding
author. Fax:
[email protected]
+61
7
33651277.
e-mail:
geochemistry
studies of subduction-related
lavas have used shortlived radionuclides,
which have half-lives comparable to the timescales of magmatic processes such as
partial melting and magma or fluid transport, to
investigate the rates of processes occurring in subduction zones. The presence of cosmic ray-produced
“Be in some lavas [5-71 indicates that Be may be
transported
from the trench, through the mantle
wedge and back to the surface in lavas, over a period
of less than about 7 Ma. Disequilibrium
between
U-series isotopes [g-10] implies that magmas are
rapidly transferred to the surface; in some cases, less
than 8 ka after melt generation. Combined “Be and
U-series studies [ 1 1- 131 have yielded conflicting
0012-821X/97/$17.00
0 1997 Elsevier Science B.V. All rights reserved.
PII SO0 12-82 I X(97)00
107.6
292
M. Regelous et al. /Earth
and Planetary Science Letters 150 (1997) 291-302
estimates of between 20 ka and 6 Ma for the residence time of U, Th and Be in the mantle wedge.
These various estimates may reflect real differences
in transport times between arcs, or variable transport
rates for different elements.
Recent studies have shown that although crustallevel contamination may have an important influence
on the chemistry of arc lavas (e.g. [14]), variations in
the composition
of the material being subducted
beneath the arc can also play a part in generating
chemical variations in the lavas [ 15,161. In this paper, we show that geochemical
variations in the
chemistry of lavas from the Tonga-Kermadec
arc
can be linked to differences in the material being
subducted beneath the arc. This observation is used
to argue that beneath the Tonga-Kermadec
arc,
fluid-soluble
elements such as Pb, U and Sr are
transported from the subducted slab to the surface in
lavas over a period of about 2-3 Ma.
2. Samples
The Tonga-Kermadec
arc is an approximately
3000 km long chain of active volcanoes situated in
the southwest Pacific to the north of New Zealand
(Fig. 1). Volcanism is related to the subduction of
the Pacific plate westwards
beneath
the IndoAustralian plate, at a rate of between about 5 cm/yr
in the south of the arc and about 20 cm/yr in the
north [ 17,181. The Tonga and Kermadec volcanic
chains are separated by a non-volcanic
zone, corresponding to the intersection of the trench with the
subducting Louisville Seamount Chain, an aseismic
ridge on the Pacific plate. Behind the arc, back-arc
extension is taking place in the Lau Basin-Havre
Trough, and true seafloor spreading initiated in the
northern and eastern Lau Basin at about 6 Ma 1191.
The intraplate volcanic islands of Samoa are situated
on the Pacific plate N 200 km to the north of the
northern end of the Tonga trench.
We have measured trace element concentrations
and Sr, Pb and Th isotope ratios for young volcanic
rocks from the arc and the back-arc volcano of Niua
fo’ou, and for sediments cored from the subducting
Pacific plate. The volcanic rocks analysed include
low-K tholeiites,
basaltic
andesites
and dacites
[20,21] from the youngest flows on thirteen islands
Fig. I. Simplified tectonic map of the Tonga-Kermadec
region,
showing sample sites, and (inset) location of DSDP/ODP
sites
from which sediment samples were collected. Islands from which
lavas were analysed are: I Niua fo’ou; 2 Tafahi; 3 Niuatoputapu;
4 Fonualei; 5 Late; 6 Metis Shoal; 7 Kao; 8 Tofua; 9 Hunga
Ha’apai; 10 Raoul Group; II Macauley; 12 Curtis; 13 L’Esperante. CL.SC, ILSC and ELSC are the Central, Intermediate and
Eastern Lau Basin Spreading Centers.
along the length of the arc (Table l>, and are representative samples from a larger selection of TongaKermadec lavas analysed by Ewart et al. (submitted).
The sediment samples are from DSDP/ODP
sites
596, 204 and 275 in the southwest Pacific (Fig. 1). A
weighted average for the sediment column at Site
204 was calculated using seven individual analyses
(Table 1).
3. Analytical
methods
Sediment samples were rinsed several times with
water in an ultrasonic bath before digestion. For Th
analysis of lavas, between 0.1 and 1.0 g of handpicked rock chips were washed in 1 M HCl in
ultrasound
prior to HF/HNO,
digestion
and
M. Regelous et al./Earth
HNO,/HCl dissolution. When larger sample sizes
were used (0.8-1.0 g), an initial concentration of Th
was made by co-precipitation with iron hydroxide.
Th was separated from the rock matrix using conventional anion exchange methods and loaded on the
side filament of a triple filament assembly of zoneTable 1
Trace element and isotope data for Tonga-Kermadec
lavas and South Pacific sediments
208Pb/
204Pb
U
Th
Zr
Pb
204 pb
k 11
18.413
15.602
38493
0.110
0.410
128.1
1.19
0.704342
k 10
18.410
15.585
38.459
O.OY8 0.306
95.8
0.71
1.607
0.704300,
15.619
38.934
0.108
0.703992
0.703833
0.703510+
0.703919
0.703822
0.703643
13
+ 10
k 10
11
+ 11
rt 10
* 13
19.306
1.828
2.726
2.852
3.115
2.611
2.373
19.103
18.580
18.505
18.542
18.571
18.562
15.612
15.549
15.516
15.535
15.559
15.567
38.782
38.189
38.090
38.131
38.199
38.214
0.092
0.393
0.273
0.344
0.141
0.251
0.210
0.158
0.452
0.300
0.347
0.169
0.332
11.4
10.5
38.7
29.5
29.8
20.9
36.2
1.17
1.10
3.85
2.18
3.82
2.33
4.53
0.023
0.013
0.015
0.066
0.024
0.021
0.043
0.012
0.022
2.345
2.166
2.628
2.638
1.279
1.360
1.263
1.226
1.122
0.703263 +
0.703337 f
0.703503 +
0.703527 +
0.704122 +
0.703605 +
0.703612 +
0.703384&
0.703469 +
8
11
11
10
9
14
13
13
11
18.622
18.651
18.640
18.649
18.709
18.590
18.633
18.719
18.528
15.570
15.574
15.582
15.584
15.586
15.594
15.578
15.586
15.564
38.236
38.293
38.291
38.308
38.479
38.380
38.373
38.428
38.232
0.149
0.143
0.182
0.128
0.072
0.129
0.103
0.140
0.153
0.199
0.207
0.217
0.153
0.176
0.297
0.254
0.359
0.428
46.0
36.6
33.2
21.5
27.5
39.2
21.1
38.2
24.6
1.96
1.67
2.70
1.80
1.40
2.05
1.75
1.60
1.69
1.127 + 0.008
1.242
0.703705
f 10
18.699
15.589
38.452
0.681
1.720
130
7.12
1.393 + 0.024
1.375
0.703504 + 14
18.671
15.587
38.406
0.061
0.139
15.2
2.66
1.277 + 0.016
1.142 f 0.012
1.027 + 0.016
1.605
2.509
1.089
0.704097 + 10
0.704077 +9
0.704041 f 9
18.739
18.724
38.597
38.615
38.493
0.338
0.574
0.659
68.6
18.696
15.619
15.620
15.587
0.207
0.561
41.8
2.85
0.894 f 0.011
I.156
0.704393 f 10
18.716
15.616
38.575
0.195
0.561
36.0
2.40
0.881
0.707517
18.836
15.679
38.849
1.38
4.90
111.4
18.1
232Th)
(238U/
*‘Sr/ 86Sr *
Rocktype
TlO7
Niua fo’ou
Basalt
1.189 + 0.031 a 0.844
0.704181
Tlll
(1)
Niua fo’ou
Basalt
1.140 & 0.021
1.006
T113CP
(1)
Tafahi (2)
B. andesite
1.174 & 0.028
+
f
+
i
+
1.824+
0.014
0.014
0.024
0.018
0.014
0.030
+
+
+
*
f
f
+
+
f
‘3’Th)
Tafahi (2)
B. andesite
Fonualei (4) Dacite
Fonualei (4) Andesite
Fonualei (4) Andesite
Late (5)
B. andesite
Metis Shoal
Dacite
(6)
Kao (7)
Andesite
64T4C
Kao (7)
T103C
B. andesite
Tofua 32 Tofua (8)
Dacite
Hunga (9)
HHTF
B. andesite
14775
Raoul (10)
B. andesite
14790
Raoul (10)
B. andesite
Raoul (10)
14796
B. andesite
Raoul(10)
23376
B. andesite
10379
Macauley
Basalt
(11)
Macauley
Dacite
10384
(11)
Macauley
10415
Basalt
(11)
Curtis (12)
B. andesite
14849
Curtis (12)
14864
Andesite
L’Esperance
B. andesite
14835-l
(13)
L’Esperance
14840
B. andesite
(13)
14840R b L’Esperance
B. andesite
T114
Fon 8
Fon 30
Fon 31
L21
MG
refined Re. Pb was separated using standard chemical techniques, blanks for the entire procedure were
< 0.2 ng. All isotope measurements were carried out
at the University of Queensland, using a VG Sector
54-30 thermal ionisation mass spectrometer equipped
with a Daly ion-counting system and electrostatic
204pb
(a3’Th/
Island (see
Fig. 1)
Sample
293
and Planetary Science Letters 150 (1997) 291-302
1.255
1.756
1.825
2.471
1.878
1.501
I.479
1.705
2.002
0.843
1.609
1.435
1.160
1.087
ao6Pb/
*O’Pb/
0.921 + 0.022
(13)
Site 204 ’ Sediment
0.881 *
a Errors are 2~. For Sr isotope analyses,
b Analysis of a separate dissolution.
error refers to final digits.
’ Weighted mean of the sediment column at ODP Site 204 was calculated from analysis of seven representative sediment samples. This
average includes topmost 50 m of sediment, which contains ash from the Tonga arc, and Louisville ash layer at IO+150 m.b.s.f.
d ?‘Th/
‘32Th) for Site 204 sediment calculated using mean U/Th ratio, and assuming ‘30Th-238U equilibium.
Brackets indicate activity ratios; activities were calculated using decay constants of 9.1952 x 10e6 yr-‘, 4.9475 x 10-r’
yr-‘, and
1.5513 X lo-”
yrr’ for 230Th, *a2Th and ‘38U respectively.
Trace element concentrations
in ppm. U and Th measured by isotope dilution, Zr and Pb by ICPMS.
294
M. Regeious et al./ Earth and Planetary Science Letters I50 (1997) 291-302
filter. Sr was analysed using a dynamic procedure,
and an exponential
fractionation
correction
to
86Sr/ **Sr = 0.1194. Mean values (2~) for the
NBS987 Sr and UCSC Th standards over the period
of analysis were *?Sr/ 86Sr 0.710244 k 17 (n = 10)
and 232Th/ 230Th 170,823 + 1032 (n = 8); laboratory averages for these standards during 1995 were
0.710249 + 24 and 170,345 f 938, respectively. Zr
and Pb concentrations
were measured by ICPMS
using a Fisons PlasmaQuad II instrument, and U and
Th concentrations
were determined by isotope dilution using a 233U- 229Th spike.
4. Results
I-
/’
Equilhe
/
q
,_
IL
>_
4.1. Th isotope data
Selected trace element and isotope data for those
samples analysed for Th isotopes are presented in
Table 1, and the Th isotope data are illustrated on a
conventional equiline diagram in Fig. 2. Pacific sediments from the boreholes studied have U/Th values
of between 0.18 and 0.78, similar to those previously
reported for Pacific sediments [8], and corresponding
to equilibrium (230Th/ 232Th) values of 0.5-I 5 (Fig.
2).
Tonga
lavas
have
among
the
highest
(238U/232Th)
and (230Th/232Th)
values yet reported for subduction-related
rocks (Fig. 21, and our
new analyses extend the existing database for Tonga
lavas [8,9,22,23]. All Tonga lavas analysed have
excess U, with a range in (230Th/ 238U) between
0.54 and 0.79. Lavas from the Tonga islands of
Tafahi and Kao have lower (230Th/ 23’Th) and
(‘38U/ 232Th) than other Tonga samples. The Tonga
lavas define an array in Fig. 2 corresponding to an
“age” of N 70 ka. However, a best-lit line through
this array intersects
the equiline
at negative
( 230Th/ 232Th), and is therefore unlikely to have any
age significance. Instead, the alignment of the data in
Fig. 2 may reflect mixing processes, as discussed
later in this paper. Kermadec lavas have lower
(238U/ 232Th) th an T onga lavas, are generally closer
to equilibrium,
and have U excess or Th excess.
Lavas from Niua fo’ou volcano in the back-arc Lau
Basin have larger 230Th excess, and (230Th/ 232Th)
values similar to oceanic basalts (Fig. 2).
Unlike lavas from other tectonic environments,
I -
0
1
2
3
4
( 23aU/232Th)
Fig. 2. A. Equiline diagram illustrating Th isotope data for lavas
from the Tonga-Kermadec
arc, and from the back-arc volcano of
Niua fo’ou. Field is for existing analyses of Tonga lavas [8,9]. B.
Data for MORB, OIB and other subduction-related
rocks
[8,9,12,13] are shown for comparison. Abbreviations;
P Philippines; J Japan/Indonesia;
C Costa Rica; L Lesser Antilles; M
Marianas; T Tonga; A Aleutians; Ae Aeolian; V Valu Fa; N
Nicaragua.
many arc lavas, including Tonga lavas, are characterised by excess 238U, which is due to addition of
the fluid-soluble element U to the source of the arc
lavas < 350 ka prior to mantle melting [8-lo].
Subduction-related
lavas also tend to have more
variable (230Th/ 232Th) than oceanic basalts, and
this is generally ascribed to subduction of materials
(pelagic sediments, carbonates, altered oceanic crust)
with variable U/Th [S-lo]. The significance of the
Th isotope compositions of Tonga-Kermadec
lavas
is discussed further in Section 5.5.
Most oceanic basalts are characterised by excess
230Th due largely to the greater incompatibility
of Th
compared to U during mantle melting [24,25]. The
Th excess observed in Niua fo’ou lavas from the Lau
Basin is therefore interpreted to reflect a much reduced input from slab-derived fluids in the back-arc
h4. Regelous et al./Earth
[20,21]. In detail, the U-Th systematics of oceanic
basalts are sensitive to the process and timescale of
melting, and the subsequent behaviour of the melts
(e.g. [26]). The data for the back-arc lavas are therefore not discussed in detail here, except to note that
the large degree of disequilibrium (230Th/ 238U 1. l1.4), and the low (‘30Th/232Th) values (1.1-1.2) of
the Niua fo’ou lavas are more characteristic of ocean
island basalts (OIB) than of mid-ocean ridge basalts
(MoRB).
4.2. Pb isotope data
Pacific sediments from the three boreholes studied
have relatively radiogenic Pb isotope compositions
(Fig. 31, similar to previously reported analyses of
western Pacific sediments [8,14,27,28]. Two samples
from depths of between 100 and 150 m.b.s.f. at Site
204 have significantly higher 206Pb/ *04Pb and lower
“‘Pb/ 2”4Pb. These samples are composed largely
15.e
295
and Planetary Science Letters 150 (1997) 291-302
of volcanic material derived from the nearby
Louisville Seamount Chain (LSC) [29], and have
similar Pb isotope compositions to basalts dredged
from the LSC (Fig. 3). The least radiogenic sediment
samples are from the uppermost 50 m of hole 204;
these sediments contain a large arc-derived ash component, deposited in the youngest units as the Pacific
Plate has drifted towards the Tonga arc.
Lavas from Tonga islands to the south of Niuatoputapu define an array between the fields of
Pacific MORB and Pacific sediments in Fig. 3.
Lavas from the northernmost Tonga islands of Niuatoputapu and Tafahi are more radiogenic than those
of other Tonga islands (Fig. 3). In order to highlight
the difference between these two Tonga arrays, a
suite of thirteen lava flows from the latter two
islands were analysed [30]. Pb isotope compositions
for these samples extend to high ‘06Pb/ ‘04Pb in Fig.
3, overlapping the field of basalts from the Louisville
Seamount Chain. Kermadec lavas have generally
’
A
f’
+
Sediment
+
++
8 15.7
B
N
+
+
$
$
15.6
15.5
B
%
LSC
Samoa
39
*+
++
0
8
@
2
‘5.
3
:g$$
Pactfic
MOW
.
-I
39
18
19
206Pb1204Pb
20
19
20
206Pb/204Pb
Fig. 3. (A) and (B); Pb isotope data for lavas and sediments analysed in this study, and for lavas from Samoa [33,51,52], the Louisville
Seamount Chain [39], and Indian and Pacific MORB [37j. Crosses represent Pacific sediments (this study); other symbols as in Fig. 2. The
two sediment samples with high *06Pb/ 204Pb are from depths of between 100 and 150 m.b.s.f. at ODP site 204, and consist largely of
volcanic material from the nearby Louisville Seamount Chain [29]. The 13 Tonga samples with ‘06Pb/ *04Pb > 18.8 (diamond symbols) are
from the two northernmost Tonga islands of Niuatoputapu and Tafahi: these have similar Pb isotope compositions to lavas from the LX.
(CJ and (D); comparison of Tonga-Kermadec
lavas with data for other arcs from the western Pacific. Most Tonga-Kermadec
lavas are
similar to lavas from the Marianas 1531, New Britain [54], N. Japan [14] and.Philippines
[55] arcs, but Tafahi and Niuatoputapu lavas have
significantly different Pb isotope compositions, in particular higher ‘06Pb/ 2”4Pb , suggesting a unique (Louisville) source for the Pb.
296
IU. Regelous et al./ Earth and Planetary Science L.etters I50 (1997) 291-302
more radiogenic Pb isotope compositions than Tonga
lavas, and extend towards the field defined by Pacific sediments in Fig. 3. Lavas from the southern
Kermadec islands of Curtis and L’Esperance are
more radiogenic than lavas from Macauley and Raoul
[28,31].
Most Tonga-Kermadec lavas have similar Pb isotope compositions to lavas from the Marianas, New
Britain, Philippines and Japanese arcs (Fig. 3C, D).
However, samples from Tafahi and Niuatoputapu
islands have very different Pb isotope compositions
to lavas from other arcs in the western Pacific, in
particular higher 206Pb/ *04Pb.
5. Discussion
5.1. Geochemical
variations
in Tonga-Kennadec
lavas
Geochemical studies of subduction-related lavas
have shown that arc magmatism occurs in response
to the addition of fluids and/or melts derived from
sediments and altered oceanic crust in the subducting
slab, to the sub-arc mantle wedge [I -3,321. This
model can account for the characteristic enrichment
in arc lavas of large-ion lithophile elements (LILE)
relative to high field strength elements (HFSE) compared to MORB. In this section, we briefly evaluate
the relative contributions of subducted sediment,
subducted oceanic crust, and mantle wedge to the
trace element budget of Tonga-Kermadec lavas.
Quantitative modelling of these data is presented
elsewhere (Ewart et al., submitted). The purpose of
this discussion is to demonstrate that along-arc variations in geochemistry of the lavas can be linked to
differences in the material being subducted along the
arc, and that this can be used to estimate the
timescales of processes beneath the arc.
Previous geochemical studies of Tonga-Kermadec lavas [20,21] have shown that broad variations in
lava geochemistry along the arc can be linked to
variable depletion of the mantle wedge beneath the
arc. Increasing depletion of the wedge towards the
northern end of the arc, linked to greater extension in
the northern Lau Basin, coupled with a relatively
uniform flux from the subducting slab result in more
extreme LILE/LREE enrichment in northern Tonga
lavas. However, differences in the material being
subducted along the arc also play a part in generating
along-arc variations in lavas chemistry, as outlined
below.
The mantle wedge beneath the northern end of the
Tonga arc contains an OIB component, derived from
the Samoa mantle plume [23,30]. As a result, many
commonly-used indicators of the subduction input,
such as the U/Nb ratio [33], do not accurately
reflect the relative roles of wedge and subduction
component. OIB typically have high concentrations
of Nb relative to Zr compared to MORB and islandarc lavas. Zr is therefore less sensitive than Nb to the
presence of an OIB component in the mantle wedge,
and here we use U/Zr as a measure of the subduction input. Both U and Zr are incompatible in mantle
materials, but U is mobile in aqueous fluids, whereas
Zr is not [2,34], and so higher U/Zr ratios may
indicate a larger contribution from the subducted
slab to the trace element budget of the lavas. Data
for lavas from the volcanic island of Niua fo’ou
(Table 1) in the northern Lau Basin behind the
Tonga arc (Fig. l), which contain a much reduced
subduction input [20,21] are shown for comparison.
Analyses of sixteen lavas from the Intermediate and
Central Lau Basin active spreading centers [35] are
also shown in Fig. 4A. The high and variable U/Zr
values of Tonga-Kermadec lavas compared to Lau
Basin lavas are interpreted to reflect addition of U
from the subducted slab, rather than partial melting
or fractional crystallisation processes.
5.2. Source of trace elements in the lavas
For Tonga lavas, excluding those from Tafahi,
lavas with the greatest slab contribution (high U/Zr)
have high 87Sr/ 86Sr, but tend to have lower
*06Pb/ 204Pb (Fig. 4). The isotope variations cannot
be explained by dehydration of subducted Pacific
sediments, which have high 206Pb/ *04Pb and high
Pb concentrations (Table 1, Fig. 3). However, altered
oceanic crust in the subductin Pacific plate would
have the required high 89Sr
LT;;; *G;
‘,“,“d
206Pb/204Pb [36]. The low
*‘*Pb/ *04Pb inferred for the slab component (Figs.
3 and 4D) is consistent with a source from subducted
Pacific oceanic crust, whereas the mantle underlying
much of the Lau Basin is Indian type MORB mantle
[37]. The array defined by Tonga lavas extending
between the fields for Pacific sediment and Pacific
M. Regelous et al./Earth
and Planetary Science Letters 150 (19971291-302
MORB in Fig. 3 is thus interpreted to reflect addition
of variable amounts of slab-derived
fluid to the
mantle wedge. The Tonga samples with the least
0.7032
0.7030
-D
0°$
19.2 -
%
4
4
I
19.0 -
18.4d
0.000
I
8 '8
0.004
'I"
O.OfJ8
.'L
0.012
URr
Fig. 4. Variation of (A) Th/Zr, (B) cz3’Th/ 232Th), (C) “Sr,/ 86Sr
and (D) lo6Pb/ ‘“Pb with U/Zr for the volcanic rocks analysed
in this study. U and Th concentrations were measured by ICPMS.
2o external errors are shown for the isotope measurements.
Symbols as in Fig. 2; Tonga lavas (excluding those from Niuatoputapu and Tafahi) are highlighted. Shaded field is for data for
young lavas from Lau Basin spreading centers [35], and the star
represents an average N-MORB composition [50]. Higher U/Zr
ratios indicate a larger contribution from the subducted slab to the
trace element budget of the lavas (see text), showing that the slab
component beneath the Tonga arc has high “‘Sr/ a6 Sr but low
‘06Pb/ *04Pb compared to the mantle wedge.
291
subduction input are more radiogenic than depleted
upper mantle (Fig. 31, suggesting that the mantle
beneath the arc was contaminated by sediment prior
to addition
of U enriched fluid, and hinting that
sediments and fluids may be added to the sub-arc
mantle by different processes [381.
Lavas from the northern Tonga islands of Tafahi
and Niuatoputapu have more radiogenic Pb and Sr
isotope compositions than other Tonga lavas (Fig. 4).
Pb isotope compositions of the most radiogenic samples are similar to those of basalts from the Louisville
Seamount Chain (LSC) [39] (Fig. 3), the northern
part of which has been subducted beneath the Tonga
arc. Our ICPMS analyses show that the Louisville
volcanic ash and tuff material in the Site 204 core
have Pb concentrations
of N 1 ppm, in contrast to
the overlying Fe-rich clays which contain 20-30
ppm Pb. A 200 m thick layer of Pb-rich clay is being
subducted at the Tonga Trench, however Fig. 3
clearly shows that the Pb in Tafahi and Niuatoputapu
lavas is dominated by the Louisville-derived
component. The Louisville Pb is therefore most likely
derived from the altered basaltic crust of the subducted LSC, in which Pb has been concentrated by
hydrothermal fluids. This is further supported by the
fact that Pb in other Tonga lavas is derived from
altered Pacific MORB crust, as discussed above. The
new Pb isotope data for Tonga lavas is convincing
evidence that subducted oceanic crust may make a
significant contribution to the trace element budget
of arc magmas [40].
Kermadec lavas have generally lower U/Zr values than Tonga lavas (Fig. 4), consistent with a less
depleted mantle beneath the Kermadec arc [20,21],
and thus a larger contribution from the mantle wedge
to the trace element budget of Kermadec lavas.
Kermadec lavas have more variable trace element
and isotope compositions,
and unlike Tonga lavas,
do not lie on well-defined arrays in Fig. 4. Lavas
from the southern Kermadec islands of L’Esperance
and Curtis tend to contain more radiogenic Pb and Sr
than lavas from Raoul and Macauley to the north,
and Pb isotope compositions extend towards the field
defined by western Pacific sediments in Fig. 3A,
which suggests a small (2-3%) Pb contribution from
subducted sediment [28].
The geochemical differences between Tonga and
Kermadec lavas discussed above can be linked to
298
M. Regelous et al./ Earth and Planetary Science Letters I50 (1997) 291-302
variations in the material being subducted along the
arc. East of the Tonga trench, the Pacific plate
carries a relatively thin sediment cover (N 100 m at
site 595; N 200 m at site 204), and thus the contribution from altered oceanic crust in the subducting slab
can be detected in the Sr and Pb isotope compositions of Tonga lavas. At the southern end of the arc,
a layer of sediment approximately 1 km thick is
being subducted at the Hikurangi trench [41], and the
greater input of sediment, derived from the continental shelf of New Zealand, into the trench at the
southern end of the arc [42] can explain the more
radiogenic Pb isotope compositions of Kermadec
lavas, particularly those from the southern Kermadec
islands of Curtis and L’Esperance.
I
>.
,\
\
\\\Tsfahi
5.3. Element transport rates in the Tonga subduction
zone
As discussed above, the Pb in lavas from the
northern Tonga islands of Tafahi and Niuatoputapu
was derived from the subducted portion of the LSC
on the downgoing Pacific plate. The LSC signature
is not seen in young ( < 350 ka) lavas from islands
further to the south along the arc. This observation
can potentially be used to estimate the time taken for
elements such as Pb to travel from the subducting
slab, through the mantle wedge, and back to the
surface in lavas (the mantle residence time). This
Ido-Australian
T&hi
Nluatoputapu
Pacific Plate
hydration (-100 km)
Fig. 5. Schematic section through the northern part of the Tonga
arc. The time taken for water-soluble elements to travel from the
subducted part of the LSC to the surface in lavas, is equal to the
time taken for the northern part of the LSC to be subducted from
the depth of dehydration to its inferred present depth beneath the
arc.
Fig. 6. Projection of the northern (subducted) portion of the LSC
onto the surface of the subducted Pacific plate beneath the northern end of the Tonga arc, modified from [43]. The present depth
of the LSC beneath Tafahi is estimated at 400-500 km, using data
from [43-481. Taking into account the geometry of the subducted
slab, the inferred depth of dehydration and the subduction rate in
northern Tonga, this gives an estimate of 2-3 Ma for the residence time of Pb in the mantle wedge, as discussed in the text.
approach is illustrated schematically in Fig. 5. The
residence time of Pb is equivalent to the time taken
for the LX to be subducted from the depth at which
Pb was released from the slab, to its present depth
beneath the northern end of the Tonga arc.
The geodynamics of the Tonga region has been
extensively studied, and the structure and orientation
of the subducting slab, and regional plate motions
are relatively well constrained. The geometry of the
subducted Pacific Plate [43,44] and the LSC [45-481,
were used to estimate the depth of the subducted
seamounts beneath the northern end of the Tonga arc
(Fig. 6) at 400-500 km. The present-day convergence rate between the Pacific and Indo-Australian
plates at the northern end of the Tonga trench is
extremely rapid, at around 20 cm/yr [17,18]. This
high convergence rate in part reflects rapid Lau
Basin spreading since _ 6 Ma.
Fluids and fluid-soluble elements are released
from the subducting material by complex dehydration reactions occurring at depths of between 50 and
150 km [3]. Assuming that the slab dehydrates at a
M. Regelous et al./ Earth and Planetary Science Letters 150 C19971 291-302
depth of 100 km, and taking into account the shape
of the subducting slab [43,44], then for a subduction
rate of 20 cm/yr, the time taken for Pb and Sr to
travel from slab to surface is between 2 and 3 Ma
(Fig. 5). If the slab dehydrates at depths < 100 km,
or if subduction rates were lower in the past, this is a
minimum estimate of the residence time. The possible errors involved in this estimate are difficult to
evaluate, but this approach does at least provide an
alternative estimate to geophysical
modelling and
“Be data, and the results do not support very brief
( < 350 ka) mantle residence times.
This estimated mantle residence time, together
with the measured subduction rate in northern Tonga
[ 17,181, yield transport times from the trench, through
the mantle wedge, and back to the surface in arc
lavas, of 3-4 Ma. This is therefore consistent with
the < 7 Ma constraint obtained from “Be data for
other arc lavas [5-71.
5.4. Model for arc magmatism
The calculations above suggest that fluid-soluble
elements such as Pb are transported from the subducting slab to the surface in lavas, over a period of
around 2-3 Ma. On the other hand, excess 238U
observed in all young Tonga lavas indicates that
addition of U to the magma source, and subsequent
mantle melting occurred < 350 ka before the lavas
were erupted at the surface.
Recent numerical modelling studies of the thermal
structure of subduction zones [4,49] have proposed
that fluid-soluble
elements are transported through
the mantle wedge by a series of hydration-dehydration reactions. Fluids transport elements such as Pb
vertically until stabilised in hydrous minerals in the
overlying mantle wedge, these elements are then
“fixed”
in the mantle until continued subduction
brings the minerals back down to a depth at which
they dehydrate. In this way, fluid-soluble
elements
are transported across the mantle wedge over a period of time, and arc magmas with (‘38U/ “30Th) > 1
are generated when the last such dehydration event
released fluid into the region of melting.
The timescales estimated above imply a rate of
magma transport to the surface of at least 40 cm/yr,
and that elements are transported laterally within the
wedge by fluids at an average rate of about 2-3
cm/yr. This is similar to estimates of average lateral
199
transport rates of fluids within the mantle wedge
based on thermal-mechanical
modelling (centimeters
per year) [4].
5.5. Mantle residence times and Th isotope systematits
Th isotope compositions of lavas erupted at convergent plate margins reflect the derivation of the
Th, and/or the timing of U/Th fractionation events
occurring
within
the last 350 ka. The high
( 230Th/ 23’Th) va 1ues observed in many arc lavas
(Fig. 2) may reflect recent addition of Th from a
source with time-integrated high U/Th, for example
carbonate sediments or altered subducted oceanic
crust; alternatively, high cz3’Th/ ‘3’Th) in arc lavas
may be due to ingrowth of ‘30Th in a U-enriched
component of the mantle wedge over a period of
time [8- 101.
Tonga-Kermadec
arc lavas have higher U/Zr
and higher Th/Zr than depleted upper mantle in the
region as recorded by Lau Basin lavas (Fig. 4A). In
Fig. 4A, Tonga and Kermadec lavas define two
arrays with positive slope, indicating that both U and
Th are added from the slab to the mantle wedge.
This is consistent with the results of recent experimental studies which suggest that Th is as soluble as
Sr in Cl-rich fluids [2,34]. Assuming that Zr is
completely immobile in slab-derived fluids, then the
U/Zr and Th/Zr values of Tonga-Kennadec
lavas
compared to those of average MORB [50] indicate
that 70-90% of the total U and Th in the arc lavas is
derived from the slab (Fig. 4A). These are underestimates if Zr is also transferred from the slab to the
mantle wedge.
About 80% of the Th in Tonga lavas is thus
derived from the slab. However, our estimate that U
and Th take 2-3 Ma to travel from the subducted
slab to the surface in lavas indicates that the high
cz3’Th/ ‘j*Th) va 1ues of Tonga lavas were not inherited
directly
from the slab. Instead,
high
cz3’Th/ 23’Th) must be due to ingrowth of 230Th in
the mantle wedge. The fact that many arc lavas are
close to 130Th- ‘38U equilibrium has previously been
interpreted as due to ingrowth of ‘3”Th in a U-enriched, easily melted component of the mantle wedge
[lo]. Earlier studies have also shown that a positive
and
208pb ~,?06p,,*
correlation
exists between
300
hf. Regelous et al. /Earth
and Planetary Science Letters I.50 (1997) 291-302
Th/U for arc lavas worldwide [8,9], which suggests
that much of the Th/U variation in arc lavas is
generated during addition of old (> 100 Ma) sediment and/or altered oceanic crust to the mantle
wedge. Thus, although hydration-dehydration reactions in the mantle wedge will fractionate U and Th,
it appears that the Th isotope compositions of arc
lavas broadly reflects the material being subducted,
via the U/Th ratio of the slab component. Fig. 4A
indicates that the U/Th ratio of the slab component
in Tonga and Kermadec lavas is different, having
higher U/Th beneath the Tonga arc. We interpret
high U/Th and high cz3’Th/ 232Th) in Tonga lavas
to reflect derivation from U-enriched, altered oceanic
crust in the subducting slab. Lower cz3’Th/ 232Th)
in Kermadec lavas, and small 230Th excess in some
samples are the result of a larger contribution from
the less depleted mantle wedge, and from subducted
Pacific sediment with low U/Th. Tafahi samples
have lower (230Th/ 232Th) than other Tonga lavas,
and the inferred U/Th of the subduction component
is also intermediate between Tonga and Kermadec
lavas, perhaps reflecting the younger, presumably
less U-enriched crust of the LSC.
6. Conclusions
Geochemical variations in young volcanic rocks
from the Tonga-Kermadec arc can be linked to
differences in the material being subducted beneath
the arc. Lavas from the Kermadec segment of the arc
contain a contribution from subducted sediment
whereas the Sr and Pb in Tonga lavas is derived
from altered oceanic crust in the subducting Pacific
plate. Lavas from the northernmost Tonga islands of
Tafahi and Niuatoputapu contain Pb and Sr derived
from the subucted portion of the Louisville Seamount
Chain, and this is used to estimate the rate at which
water-soluble trace elements are transported beneath
the arc. These elements have residence times in the
mantle wedge of approximately 2-3 Ma, and so the
variable Th isotope compositions of Tonga and Kermadec lavas reflect ingrowth of 230Th within a variably U- and Th-enriched mantle wedge prior to 350
ka. Magmas are erupted within 350 ka of their
generation in the mantle wedge, indicating magma
ascent rates of > 40 cm/yr, whereas element trans-
port by fluids within the wedge takes place at an
average rate of 2-3 cm/yr.
Acknowledgements
This research was supported by ARC grants to
K.D.C. and A.E. We thank Y. Niu, J.-X. Zhao, R.
Frankland and M. Bruce for their help with the
ICPMS analyses, and Julie Morris, Terry Plank and
Brad Singer for very helpful reviews which improved the manuscript. [FAI
References
01 J.B.
Gill, Orogenic Andesites and Plate Tectonics, Springer,
New York, 1981.
Dl H. Keppler, Constraints from partitioning experiments on the
composition of subduction-zone
fluids, Nature 380 (1996)
237-240.
131 Y. Tatsumi, Migration of fluid phases and genesis of basalt
magmas in subduction zones, J. Geophys. Res. 94 (1989)
4697-4707.
[41 J.H. Davies, D.J. Stevenson, Physical model of source region
of subduction zone volcanics, J. Geophys. Res. 97 (1992)
2037-2070.
151 L. Brown, J. Klein, R. Middleton, I.S. Sacks, F. Tera, “Be
in island arc volcanoes and implications
for subduction,
Nature 299 (1982) 718-720.
161F. Tera, L. Brown, J. Morris, IS. Sacks, J. Klein, R.
Middleton, Sediment incorporation
in island-arc magmas:
inferences from “Be, Geochim. Cosmochim. Acta 50 (1986)
535-550.
[71 J. Morris, W.P. Leeman, F. Tera, The subducted component
in island arc magmas: constraints from Be isotopes and
B-Be systematics, Nature 344 (1990) 31-36.
181F. McDermott, C.J. Hawkesworth, Th, Pb, and Sr isotope
variations in young island arc volcanics and oceanic sediments, Earth Planet. Sci. Lett. 104 (1991) 1-15.
[91 J.B. Gill, R.W. Williams, Th isotopes and U-series studies of
subduction-related
volcanic rocks, Geochim. Cosmochim.
Acta 54 (1990) 1427-1442.
[lOI M. Condomines, 0. Sigmarsson, Why are so many arc
radioactive
equilibrium?,
magmas
close to 23sU-230Th
Geochim. Cosmochim. Acta 57 (1993) 4491-4497.
[Ill 0. Sigmarsson, M. Condomines, J.D. Morris, R.S. Harmon,
Uranium and “Be enrichments by fluids in Andean arc
magmas, Nature 346 (1990) 163-165.
WI J.B. Gill, J.D. Morris, R.W. Johnson, Timescale for producing the geochemical signature of island arc magmas: U-ThPO and Be-B systematics in recent Papua New Guinea lava%
Geochim. Cosmochim. Acta 57 (1993) 4269-4283.
[I31 M.K. Reagan, J.D. Morris, E.A. Herrstrom, M.T. Murrell,
M. Regelous et al. /Earth and Planetary Science Letters 150 (1997) 291-302
Uranium series and beryllium isotope evidence for an extended history of subduction
modification
of the mantle
below Nicaragua,
Geochim. Cosmochim.
Acta 58 (1994)
4199-4212.
[14] A.B. Kersting, R.J. Arculus, D.A. Gust, Lithospheric contributions to arc magmatism: isotope variations along strike in
volcanoes of Honshu, Japan, Science 272 (1996) 1464-1468.
[15] T. Plank, C.H. Langmuir. Tracing trace elements from sediment input to volcanic output at subduction zones. Nature
362 (1993) 739-743.
[ 161 B.S. Singer, W.P. Leeman, M.F. Thirlwall, N.W. Rogers,
Does fracture zone subduction increase sediment flux and
mantle melting in subduction zones? Trace element evidence
from Aleutian arc basalt. Am. Geophys. Union Monogr. 96
(1996) 285-291.
[17] B. Pelletier. R. Louat, Seismotectonics
and present day relative plate motions in the Tonga-Lau
and Kermadec-Havre
region, Tectonophysics
165 (1989) 237-250.
[I81 M. Bevis. F.W. Taylor, B.E. Schtz, I. Recy. B.L. Hacks, 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)
249-25 1.
1191 J.W. Hawkins, Evolution of the Lau Basin - insights from
ODP Leg 135, in: Active Margins and Marginal Basins of
the Western Pacific, Am. Geophys. Union Geophys. Monogr.
88 (1995) 125- 173.
[20] A. Ewart,
C.J. Hawkesworth,
The Pleistocene-Recent
Tonga-Kermadec
Arc lavas: interpretation of new isotopic
and rare earth data in terms of a depleted mantle source
model. J. Petrol. 28 (1987) 495-530.
[21] A. Ewart, K.D. Collerson, J.I. Wendt, Y. Niu and M.
Regelous, Isotopic and geochemical
evolution within the
western Pacific Tonga-Kermadec-Lau
arc-back-arc
system.
Trans. Am. Geophys. Union 76 (22), Western Pacific Geophys. Meeting Suppl. W127 (abstract). 1996.
1221 S.P. Turner, C.J. Hawkesworth,
J. Bartlett, N.W. Rogers, I.
Smith and T. Worthington.
U-Th disequilibria
along the
Tonga-Kermadec
island arc. Goldschmidt
Conference.
p.
630 (abstract), 1996.
[23] M. Regelous. K.D. Collerson, A. Ewatt and J.I. Wendt, Two
mantle plumes contribute to island arc magmatism in northem Tonga. EOS Trans. Am. Geophys. Union 77, Fall Meeting Suppl. F842 (abstract), 1996.
]24] P.D. Beattie. Uranium-thorium
disequilibria and partitioning
on melting of garnet peridotite. Nature 363 (1993) 63-65.
[25] T.Z.
LaTourette,
A.K.
Kennedy.
G.J. Wasserburg,
Thorium-uranium
fractionation
by garnet: evidence for a
deep source and rapid rise of oceanic basal&, Science 261
(1993) 739-742.
1261 M. Spiegelman, T. Elliott, Consequences
of melt transport
for uranium series disequilibrium
in young lavas. Earth
Planet. Sci. Lett. 118 (1993) l-20.
]27] D. Ben Othman, W.M. White. J. Patchett, The geochemistry
of marine sediments, island arc magma genesis, and crustmantle recycling, Earth Planet. Sci. Lett. 94 (1989) 1-21.
1281 J. Gamble, J. Woodhead, I. Wright, I. Smith, Basalt and
301
sediment geochemistry and magma petrogenesis in a transect
from oceanic island arc to rifted continental margin arc: the
Kermadec-Hikurangi
Margin, SW Pacific. J. Petrol. 37
(1996) 1523-1546.
[29] Shipboard Scientific Party, Init. Rep. Deep Sea Drilling
Project XXI. 33-56, 1973.
[30] J.I. Wendt, M. Regelous, K.D. Collerson and A. Ewart,
Evidence for a contribution
from two mantle plumes to
island-arc lavas from northern Tonga, Geology (in press.
1997).
[31] J.A. Gamble. I.E.M. Smith, M.T. McCulloch, I.J. Graham.
B. P Kokelaar. The geochemistry and petrogenesis of basal&
from the Taupo Volcanic Zone and Kermadec Volcanic Arc,
J. Volcanol. Geotherm. Res. 54 (1993) 265-290.
[32] J.A. Pearce and I.J. Parkinson, Trace element models for
mantle melting: application to volcanic arc petrogenesis. in:
H.M. Pritchard, T. Alabaster, N.B.W. Harris and C.R. Neary
(Eds), Magmatic Processes and Plate Tectonics, Geol. Sot.
Spec. Publ. 76, 373-403.
[33] H.E. Newsom. W.M. White, K.P. Jochum, A.W. Hofmann,
Siderophile and chalcophile element abundances in oceanic
basalts, Pb isotope evolution and growth of the Earth’s core.
Earth Planet. Sci. Lett. 80 (1986) 299-3 13.
1341 J.M. Brenan, H.F. Shaw, F.J. Ryerson, D.L. Phinney, Mineral-aqueous
fluid partitioning of trace elements at 900°C
and 2.0 GPa: Constraints on the trace element chemistry of
mantle and deep crustal fluids, Geochim. Cosmoshim. Acta
59 (1995) 3331-3350.
[35] J.A. Pearce, M. Emewein, S.H. Bloomer, L.M. Parson, B.J.
Mutton, L.E. Johnson, Geochemistry of Lau Basin volcanic
rocks: influence of ridge segmentation and arc proximity. in:
J.L. Smellie (Ed.) Volcanism Associated with Extension at
Consuming Plate Margins, Geol. Sot. Spec. Pub]. 81 (1995)
53-75.
1361 H. Staudigel. G.R. Davies, S.R. Hart. K.M. Marchant, B.M.
Smith. Large scale Sr, Nd and 0 isotopic anatomy of altered
oceanic crust, DSDP/ODP
sites 417/418. Earth Planet. SC;.
Lett. 130 (1995) 169-185.
1371 J.M. Hergt and C.J. Hawkesworth, Pb-, Sr-. and Nd-isotopic
evolution of the Lau Basin: implications for mantle dynamics
during back-arc opening, in: J. Hawkins et al. (Eds.), Proc.
ODP Sci. Results 135, pp. 505-518,
Ocean Drilling Program. College Station. TX, 1994.
1381 T. Plank, C.H. Langmuir. Sediments melt and basaltic crust
dehydrates at subduction zones, EOS 73 ( 1992) 637.
[39] Q. Cheng. K.-H. Park, J.D. Macdougall. A. Zindler, C.W.
Lugmair. H. Staudigel, J. Hawkins, P. Lonsdale. Isotopic
evidence for a hotspot origin of the Louisville Seamount
Chain, Am. Geophys. Union Geophys. Monogr. 43 (1987)
283-296.
1401 D.M. Miller, S.L. Goldstein, C.H. Langmuir. Cerium/lead
and lead isotope ratios in arc magmas and the enrichement of
lead in the continents. Nature 368 (I 994) 5 14-520.
[41] F.J. Davey, M. Hampton, J. Childs. M.A. Fisher, K.B.
Lewis, J.R. Pettinga. Structure of a growing accretionary
prism, Hikurangi margin. New Zealand, Geology 14 (1986)
663-666.
302
M. Regelous et al. /Earth
and Planetary Science Letters 150 (I 997) 291-302
[42] L. Carter, R.M. Carter, IN. McCave, 3. Gamble, Regional
sediment recycling in the abyssal southwest Pacific Ocean.
Geology 24 (1996) 735-738.
(431 S. Billington, The Morphology
and Tectonics of the Subducted Lithosphere
in the Tonga-Kermadec-Fiji
Region
from Seismicity and Focal Mechanism Solutions, PhD Thesis, Cornell Univ., Ithaca, NY. 1980.
[44] K.M. Fischer, K.C. Creager, T.H. Jordan. Mapping the Tonga
slab, J. Geophys. Res. 96 (1991) 14103-14427.
[45] A.B. Watts, J.K. Weissel, R.A. Duncan, R.L. Larson, Origin
of the Louisville Ridge and its relationship to the Eltanin
Fracture Zone system, J. Geophys. Res. 93 (1988) 30513077.
[46] J. Hawkins, P. Lonsdale, R. Batiza, Petrologic evolution of
the Louisville Seamount Chain, in: B.H. Keating, P. Fryer.
R. Batiza and G.W. Boehlert (Eds.), Seamounts, Islands and
Atolls, Geophys. Monogr. Ser. 43, AGU Washington.
pp.
235-254,
1987.
[47] J. DuPont and R.H. Herzer, Effect of subduction of the
Louisville Ridge on the structure and morphology
of the
Tonga arc, in: D.W. Scholl and T.L. Vallier (Eds.), Geology
and Offshore Resources of Pacific Island Arcs-Tonga
Region. Circum-Pacific
Council for Energy and Mineral Resources Earth Sciences Series 2. Houston, TX, 1985.
[48] P. Lonsdale, Geography and history of the Louisville hotspot
chain in the southwest Pacific, J. Geophys. Res. 93 (1988)
3078-3 104.
1491 J.H. Davies, M.J. Bickle. A physical model for the volume
and composition of melt produced by hydrous fluxing above
subduction zones, Philos. Trans. R. Sot. London A335 (1991)
355-364.
[50] S.-s. Sun, W.F. McDonough, Chemical and isotopic systematics of oceanic basalts: implications for mantle composition
and process, in: A.D. Saunders and M.J. Norry (I%.), Magmatism in the Ocean Basins, Geol. Sot. Spec. Publ. 42
(1989) 313-345.
[51] E. Wright, W.M. White. The origin of Samoa: new evidence
from Sr, Nd, and Pb isotopes, Earth Planet. Sci. Len. 81
(1987) 151-162.
[52] Z.A. Palacz, A.D. Saunders, Coupled trace element and
isotope enrichment
in the Cook-Austral-Samoa
islands,
southwest Pacific, Earth Planet. Sci. Lett. 79 (1986) 270-280.
[53] J.D. Woodhead. Geochemistry
of the Mariana arc (western
Pacific): source composition and processes, Chem. Geol. 76
(1989) l-24.
[54] J.D. Woodhead. R.W. Johnson, Isotopic and trace element
profiles across the New Britain island arc, Papua New Guinea,
Contrib. Mineral. Petrol. 113 (1993) 479-491.
[55] F. McDermott, M.J. Defant, C.J. Hawkesworth, R.C. Maury.
J.L. Joron, Isotope and trace element evidence for three
component mixing in the genesis of the North Luzon arc
lavas (Philippines),
Contrib. Mineral. Petrol. 113 (1993)
9-23.