Processes controlling alongarc isotopic ... - Wiley Online Library

Geochemistry Geophysics Geosystems

3

G

AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society

Article Volume 8, Number 6 12 June 2007 Q06008, doi:10.1029/2006GC001475 ISSN: 1525-2027

Processes controlling along-arc isotopic variation of the southern Izu-Bonin arc Osamu Ishizuka Institute of Geoscience and Geoinformation, Geological Survey of Japan/AIST, Central 7, 1-1-1 Higashi, Tsukuba, Ibaraki, 305-8567 Japan ([email protected]) School of Ocean and Earth Science, National Oceanography Centre, Southampton, European Way, Southampton, SO14 3ZH, UK

Rex N. Taylor School of Ocean and Earth Science, National Oceanography Centre, Southampton, European Way, Southampton, SO14 3ZH, UK

Makoto Yuasa Institute of Geoscience and Geoinformation, Geological Survey of Japan/AIST, Central 7, 1-1-1 Higashi, Tsukuba, Ibaraki, 305-8567 Japan

J. Andy Milton and Robert W. Nesbitt School of Ocean and Earth Science, National Oceanography Centre, Southampton, European Way, Southampton, SO14 3ZH, UK

Kozo Uto Institute of Geoscience and Geoinformation, Geological Survey of Japan/AIST, Central 7, 1-1-1 Higashi, Tsukuba, Ibaraki, 305-8567 Japan

Izumi Sakamoto Center for Deep Earth Exploration, Japan Agency for Marine-Earth Science and Technology, 2-15 Natsushima, Yokosuka, Kanagawa, 237-0061, Japan Now at School of Marine Science and Technology, Tokai University, 3-20-1 Orito, Shimizu, Shizuoka, 424-8610, Japan

[1] The southern Izu-Bonin arc is the volcanic representation of Pacific Ocean lithosphere subduction beneath the oceanic crust of the Philippine Sea plate. We present new geochemical data including highprecision Pb isotopic measurements from the 700 km length of this intraoceanic arc. An aim of this study is to link the along-arc characteristics with variations in the subducting Pacific crust. Chemical variations have been previously recognized along the northern section of the arc as a southward decrease in 87Sr/86Sr and increase in 206Pb/204Pb. This trend continues into the southern arc as far as 27.5°N. An overall correlation between 87Sr/86Sr and fluid-mobile element enrichment between 35 and 27.5°N implies a contribution of a slab-derived fluid, mainly from altered oceanic crust and pelagic sediment. However, the observation that volcanoes plot systematically closer to the Northern Hemisphere Reference Line (NHRL) with increasing 206Pb/204Pb indicates that a component other than pelagic sediment and ocean crust plays a key role toward the south. South of 27.5°N, the along-arc isotopic trend changes dramatically, with 87 Sr/86Sr increasing southward from 27.5°N and the 206Pb/204Pb becoming highly radiogenic (19.6). This isotopic signature requires involvement of a component with high 206Pb/204Pb and low D8/4. Copyright 2007 by the American Geophysical Union

1 of 20


3

G

ishizuka et al.: izu-bonin arc isotopic variation

10.1029/2006GC001475

Volcaniclastic sediments originating from HIMU oceanic islands on the subducting Pacific Plate are possible candidates to introduce such a component into the mantle wedge. This is consistent with the sedimentary section of ODP Site 801, located outboard of the Mariana arc which has >40% mass of HIMU volcaniclastics. An aqueous fluid, not melt, has played a major role in the source magma compositions in the 27.5–25°N segment. South of 25°N the isotopic characteristics again change significantly. 143Nd/144Nd decreases down to 0.51280, but the high 206Pb/204Pb and low D8/4 signatures are retained. The remarkable Th enrichment associated with low 143Nd/144Nd, D8/4, and D7/4 suggests that melting a mixture of HIMU volcaniclastics and pelagic sediment is responsible for the geochemical characteristics of this most southerly section. Components: 11,672 words, 9 figures, 2 tables. Keywords: Izu-Bonin arc; slab component; HIMU; double-spike Pb. Index Terms: 8413 Volcanology: Subduction zone processes (1031, 3060, 3613, 8170); 8185 Tectonophysics: Volcanic arcs; 1115 Geochronology: Radioisotope geochronology. Received 10 September 2006; Revised 8 February 2007; Accepted 7 March 2007; Published 12 June 2007. Ishizuka, O., R. N. Taylor, M. Yuasa, J. A. Milton, R. W. Nesbitt, K. Uto, and I. Sakamoto (2007), Processes controlling along-arc isotopic variation of the southern Izu-Bonin arc, Geochem. Geophys. Geosyst., 8, Q06008, doi:10.1029/2006GC001475.

1. Introduction [2] Oceanic island arcs are a key location in understanding crust-mantle interaction and elemental recycling in the earth. As oceanic crust is subducted, elements are selectively removed though chemical reactions and phase transitions via hydrous fluids, siliceous melts and supercritical liquid [e.g., Tatsumi and Eggins, 1995; Elliott, 2003]. The primitive arc magma is produced by mantle melting facilitated by the hydrous flux from the subducting plate and consequently acquires the slab’s chemical signature. In this context, the oceanic island arc, where continental crust is absent, provides a unique opportunity to understand the processes of elemental transfer in a subduction zone. [3] The Izu-Bonin-Mariana arc is such an oceanic island arc formed on the eastern margin of the Philippine Sea Plate, where Pacific Ocean crust is subducted. Recent efforts to understand the chemical composition of the subducting Pacific crust have revealed a variation along the length of the arc [Plank and Langmuir, 1998; Kelley et al., 2003; Hauff et al., 2003; Plank, 2005]. The occurrence of oceanic island and seamount constructs outboard of the Mariana arc, a larger contribution of dust from the Asian continent and a lack of carbonaceous sediment outboard of the Izu-Bonin arc are examples of why this variation could exist. In addition to variations in slab composition, the angle of subduction also varies along the arc. It

changes from 40° to 80° southward in the IzuBonin arc and becomes nearly vertical beneath the Mariana arc [Engdahl et al., 1998; Stern et al., 2003]. [4] This study examines the effects of variations in subducted material on arc magmas by using radiogenic isotopes and trace element systematics including high precision Pb isotope data. The relatively well-characterized latitudinal variation in the subducting material provides a yardstick by which to judge compositional changes in the island magmas. Specifically, we wish to examine whether the concentrations of the HIMU-type seamounts [e.g., Staudigel et al., 1991] outboard of the subduction front are reflected in the composition of the adjacent Mariana arc, and correspondingly how pelagic sediments and altered oceanic crust have affected the more northern arcs. [5] In the northern Izu-Bonin arc volcanic front, Taylor and Nesbitt [1998] demonstrated that hydrous fluid released from altered oceanic crust and pelagic sediment is the chief contributor to the mantle wedge. In contrast, the northern Mariana arc bears the hallmark of sediment melt addition from the slab [e.g., Elliott et al., 1997; Sun and Stern, 2001]. Using the new isotopic data in conjunction with earlier data [e.g., Yuasa and Nohara, 1992; Taylor and Nesbitt, 1998; Ishizuka et al., 2003a] this study uses mixing models to examine the transition between hydrous fluid and 2 of 20


3

G


10.1029/2006GC001475

Figure 1. Bathymetric map of the southern Izu-Bonin arc showing locations of dredge stations where the studied samples were collected. (left) Bathymetric data from Smith and Sandwell [1997]. (right) Bathymetric data from JTOPO30 published by Japanese Coast Guard. Volcanoes studied here are divided into 3 groups: Sofugan Island and the Shichiyo Seamounts (pink symbols) located between 29.5 and 27.5°N, volcanoes between 27.5 and 25°N (green symbols), and the Iojima Island and volcanoes south of the Iojima Island between 24.5 and 24°N (blue symbols).

melt dominated slab inputs and explores some of the potential causes of the latitudinal change.

2. Geological Background [6] The N-S aligned Izu-Bonin arc is bounded by the Izu-Bonin Trench to the east and the Shikoku Basin to the west (Figure 1), and forms a broad volcanic zone up to 400 km wide. North of 30°N, this arc is characterized by en-echelon cross-arc seamount chains and rifting [e.g., Taylor, 1992;

Ishizuka et al., 2002, 2003a, 2003b]. South of this point the arc is characterized by thinner crust [Ishihara, 1985; Kodaira et al., 2007], a closer volcano spacing, a lack of silicic volcanism and the intersection of the Sofugan Tectonic Line [Yuasa, 1985], which is the major tectonic boundary between a thick arc crust in the north and stretched, or possibly rifted, thinner crust on the northern tip of the Parece Vela back-arc basin [Yuasa, 1985]. [7] The Shichiyo Seamount chain is the name given to the volcanic front from south of Sofugan 3 of 20

5.62 154 29.3 50.6 0.69 0.50 114 2.49 7.25 1.19 7.35 2.71 0.94 3.59 0.68

Rb Sr Y Zr Nb Cs Ba La Ce Pr Nd Sm Eu Gd Tb

6.90 172 33.4 54.1 0.75 0.53 120 2.56 7.41 1.19 7.41 2.48 0.93 3.68 0.68

55.55 0.85 17.37 9.94 0.17 3.25 8.44 3.12 0.52 0.11 99.32

29.793 140.342 0

4.90 209 24.3 44.1 0.73 0.11 90.1 2.55 6.98 1.14 6.53 2.28 0.82 2.98 0.55

53.58 0.71 19.28 9.48 0.17 3.44 10.15 2.59 0.46 0.10 99.96

29.767 140.380 558

GH669-1

a

JB2 compiled value from Imai et al. [1995].

5.12 199 27.6 46.6 0.64 0.45 95.2 2.31 6.57 1.06 6.50 2.35 0.83 3.20 0.57

53.21 0.78 18.84 10.48 0.18 3.22 9.97 2.64 0.45 0.10 99.87

29.767 140.380 558

GH669-2

a

1.44 293 10.3 13.0 0.24 0.08 14.9 0.90 2.43 0.42 2.55 0.97 0.43 1.25 0.25

47.58 0.4 22.05 8.74 0.13 5.01 13.85 1.47 0.16 0.04 99.43

29.449 140.370 1289

KT97-8 D5-1

KT97-8 D5-3

51.28 0.54 17.74 10.27 0.18 6.3 11.78 1.88 0.34 0.05 100.34 4.63 230 15.0 25.9 0.32 0.28 50.6 1.43 4.06 0.63 3.93 1.47 0.60 2.05 0.38

Trace Element, ppm 6.79 5.83 301 295 21.9 20.0 42.6 42.2 0.91 0.77 0.37 0.33 61.2 56.3 3.02 3.04 8.18 8.13 1.27 1.17 6.96 6.68 2.27 2.22 0.86 0.81 2.70 2.76 0.49 0.47

29.449 140.370 1289

KT97-8 D5-4

wt% 50.67 0.77 18.89 10.52 0.16 4.27 10.83 2.41 0.43 0.14 99.09

29.449 140.370 1289

Major Element, 51.26 0.8 19.21 10.58 0.17 3.66 10.4 2.46 0.51 0.11 99.15

29.449 140.370 1289

KT97-8 D5-2

2.21 284 10.8 15.0 0.27 0.13 24.0 0.97 2.69 0.42 2.56 1.04 0.45 1.41 0.27

48.23 0.42 21.56 9.15 0.14 4.98 13.24 1.49 0.2 0.03 99.44

29.449 140.370 1289

KT97-8 D5-5

11.0 231 31.6 55.1 1.14 0.50 107 4.16 10.7 1.61 9.21 2.98 1.02 3.88 0.70

52.86 0.96 15.55 12.09 0.2 4.58 9.43 2.72 0.7 0.34 99.44

29.449 140.370 1289

KT97-8 D5-8

Nichiyo Smt.

2.78 280 11.4 15.3 0.34 0.16 24.7 1.02 2.74 0.46 2.65 1.01 0.43 1.39 0.26

48.32 0.42 21.97 9.09 0.14 4.84 13.41 1.51 0.22 0.04 99.96

29.449 140.370 1289

KT97-8 D5-22

12.5 232 24.1 65.2 1.33 0.58 108 4.18 10.8 1.51 8.26 2.46 0.79 2.95 0.51

56.50 0.61 16.23 8.55 0.16 4.35 8.69 2.78 0.89 0.13 98.89

29.454 140.377 1455

GH673-5a

7.33 170 22.5 43.2 0.56 0.44 75.7 2.05 5.92 0.94 5.60 1.92 0.69 2.57 0.48

53.86 0.75 15.95 10.78 0.18 4.93 9.58 2.19 0.58 0.09 98.89

29.550 140.451 2350

GH796-5a


a Major element data are from Yuasa and Nohara [1992]. b Analyzed at National Oceanography Centre, Southampton. c Analyzed at Geological Survey of Japan/AIST. d

55.85 0.85 17.39 9.98 0.18 3.23 8.50 3.04 0.52 0.11 99.65

29.793 140.342 0

SOF2

a

Sofugan Island

G

3

SiO2 TiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O P2O5 Total

Latitude, °N Longitude, °E Altitude, m

Sample

SOF1

a

Table 1 (Representative Sample). Whole Rock Chemical and Isotopic Compositions of Lavas From the Southern Part of the Izu-Bonin Arc [The full Table 1 is available in the HTML version of this article at http://www.g-cubed.org] Geochemistry Geophysics Geosystems

10.1029/2006GC001475

4 of 20


3

G


Island at 29.5°N to the Doyo Seamount at 27.5°N and straddles the NE-SW aligned Sofugan Tectonic Line (Figure 1). The volcanoes forming this chain are large submarine stratocones with basal diameters of 20–30 km and reach heights of 2200–2800 m above their local basement [Yuasa et al., 1991]. Some of the seamounts are hydrothermally active in the summit area [e.g., Urabe et al., 1987]. For later discussions we regard Sofugan Island and the Shichiyo Seamount chain as a single group which we refer to as the Shichiyo Seamounts. South of the Shichiyo Seamounts (27.5 – 25°N) the volcanic front includes the Kaikata and Kaitoku Seamounts and the volcanic islands of Nishinoshima and Kitaiojima. The last eruption of Nishinoshima occurred in 1974, and formed the new small island. The Kaikata Seamount (26.7°N) has a small caldera at its summit with a nested active hydrothermal system. The Kaitoku seamount (26.1°N) has a recorded submarine eruption in 1984. Iojima and Minamiiojima Islands form the southernmost part of the Izu-Bonin arc. This area is volcanically active and repeated volcanism has been recorded (e.g., Fukutokuokanoba erupted in July, 2005).

10.1029/2006GC001475

Figure 2. SiO2-K2O plot for the volcanic rocks from the southern Izu-Bonin arc. Data for Quaternary volcanic front in the northern Izu arc are from Taylor and Nesbitt [1998]. Black solid circle, volcanoes between 35 and 33°N; black open circle, volcanoes between 33 and 30°N. Other symbols as for Figure 1. The subdivision of rocks is adopted from Gill [1981].

3. Geochemistry of Volcanic Rocks [8] New Major and trace element analyses along with Sr, Nd and Pb isotope ratios for the Southern IzuBonin arc are presented in Table 1. Details of the analytical techniques are provided in Appendix A and information on the samples studied is in Appendix B.

3.1. Along-Arc Variation of Major Element Composition (K2O) [9] Lavas from the Shichiyo Seamounts are dominated by low-K tholeiite and have similar K2O contents to the northern volcanoes (Figure 2) [Yuasa and Nohara, 1992]. To the south between 27.5 and 25°N the volcanoes have slightly higher K2O and then reach a maximum of 3.8–4.2 wt% K2O south of 25°N around Iojima Island [Yuasa and Nohara, 1992; Sun and Stern, 2001]. Low to medium K andesite is dominant at Kitaiojima Island [Yuasa and Nohara, 1992] whereas Iojima is characterized by the presence of shoshonitic rocks. Shoshonitic rocks also flank the submarine edifices around Minamiiojima Island, while medium K andesite appears again south of Iojima Island [Yuasa and Nohara, 1992].

3.2. Along-Arc Isotopic Variation [10] 87Sr/86Sr has been observed to decrease steadily from north to south in the Izu-Bonin arc between

35°N and 31°N [Taylor and Nesbitt, 1998]. The volcanoes examined in this study, which lie to the south of 31°N, form a continuation of this trend with 87Sr/86Sr reducing from 0.7035 in the north to 0.7032 at the southern most of the Shichiyo Seamounts (Figure 3 and Table 1). However, south of 27.5°N the isotopic trend is reversed with 87Sr/86Sr beginning to increase. This increase continues to the south of Iojima Island where 87Sr/86Sr reaches 0.7038. [11] Nd isotopic compositions are relatively unchanged between 30°N and 25°N with 143 Nd/ 144 Nd = 0.513084 ± 0.000064 along the Shichiyo Seamounts (Figure 3). This means that the Nd isotopic compositions are not significantly different from the arc north of 30°N [Taylor and Nesbitt, 1998]. However, at around 25°N 143Nd/144Nd starts decreasing significantly southward to Iojima Island from approximately 0.513060 to 0.512820. [12] 206Pb/204Pb increases continuously toward the south but increases sharply south of 27.5°N to reach a maximum of 19.6 at Kitaiojima Island, before decreasing southward to 19.2 around Iojima Island (Figure 3). D7/4 and D8/4, which are measures of the vertical distance of data points from the Northern Hemisphere Reference Line of 5 of 20


3

G


10.1029/2006GC001475

Figure 3. Along-arc variation of 87Sr/86Sr, 143Nd/144Nd, and Pb isotopic compositions. Data for the northern Izu arc (35 – 30°N) are from Taylor and Nesbitt [1998]. Symbols as for Figures 1 and 2. Dashed lines indicate the boundaries between the three groups of volcanoes defined in this study.

Hart [1984], do not show a systematic variation in the northern Izu-Bonin arc (Figure 3: 5 and 40, respectively, between 35 and 30°N), whereas the Shichiyo Seamounts show slightly lower D7/4 and 8/4 (4 and 30, respectively). However, south of the Shichiyo Seamounts both D values appear to decrease southward. These trends continue until north of Iojima Island at 25°N where D7/4 and D8/4 reach minimum values ( 1 and 18, respectively).

3.3. Trace Element–Isotope Systematics [13] Trace element compositions also show systematic along-arc variations (Figure 4 and Table 1). As recognized in the isotopes, there is a significant change at 25°N, but some trace element ratios change south of Sofugan (29°N). Light REE/heavy REE ratios like Ce/Yb are relatively constant in the north of 30°N (Ce/Yb = 1.5–3.5); however, south

of this point Ce/Yb increases to 7 through the Shichiyo Seamounts, and becomes higher still (Ce/Yb = 17 –32) south of 25°N. Other highly incompatible/moderately incompatible trace element ratios such as Nb/Zr and Zr/Y show a comparable variation southward. Fluid mobile/REE element ratios such as Ba/La, Cs/La and Pb/Ce show a limited variation south of 30°N, although these ratios are generally lower than north of 30°N. Th/Ce is constant at 0.02 southward to 29°N then shows slightly higher and scattered values (0.02– 0.05) between 29 and 25°N. Further south the Th enrichment relative to Ce becomes more significant reaching >0.10 at Iojima. [14] The systematics of isotope and trace element ratio variations with latitude appear to be decoupled in this section of the arc. An example of this decoupling is the behavior of 87Sr/86Sr with Ba/La 6 of 20


3

G


10.1029/2006GC001475

Figure 4. Along-arc variation of trace element parameters. Data for the northern Izu arc (35 –30°N) are from Taylor and Nesbitt [1998]. Symbols as for Figures 1 and 2.

which are shown in Figures 5a and 5b. These plots show a broad positive correlation for volcanoes between 35 and 25°N, but lavas between 25 and 24°N are offset from this trend toward high 87Sr/86Sr. Pb isotope relationships with fluid mobile elements show slightly different characteristics to those of Sr isotopes and are shown in Figures 5c to 5f.

Volcanoes between 35 and 27.5°N have elevated Ba/La and Pb/Ce (low Ce/Pb) associated with moderate 206Pb/204Pb. On the other hand, volcanoes between 27.5 and 25°N show much higher 206 Pb/204Pb (19.6) and are associated with similar Ba/La and Pb/Ce to the Shichiyo Seamounts. Volcanoes south of 25°N, including Iojima Island, also 7 of 20


3

G


10.1029/2006GC001475

Figure 5. Relationship between the enrichment of fluid-mobile elements and Sr and Pb isotope ratios for the southern Izu-Bonin arc lavas. Volcanic front data are from Taylor and Nesbitt [1998]. Fields for Philippine Sea MORB (PSP) [Hickey-Vargas, 1998] are shown as reference to mantle wedge composition. Mixing lines between the assumed mantle wedge composition and slab-derived end-members are shown in each figure. Dashed lines in Figures 5a, 5c, and 5e are mixing lines between mantle wedge and fluid, where the fluid is derived from altered oceanic crust (AOC) and subducted pelagic sediment in the proportion 49:1 (in weight). Dotted lines in Figures 5a, 5c, and 5e are mixing lines between mantle wedge and fluid, where the fluid is derived from a combination of mixed fluid from AOC and pelagic sediment mentioned above and a fluid from volcaniclastics. The 30:1 mixture and 7:1 mixture lines indicate the respective proportions of AOC-sediment mixed fluid to volcaniclastics fluid that contribute to the final threecomponent fluid (Table 2). Dash-dot lines in Figures 5b, 5d, and 5f are mixing lines between mantle wedge and a melt, where the melt consists of a 1:4 mixture of pelagic sediment and subducted volcaniclastics. End-member compositions are listed in Table 2.

have high 206Pb/204Pb (19.09–19.31) but are associated with lower Ba/La and Pb/Ce relative to the northern volcanoes.

[15] The relationships between isotopes and fluidimmobile element ratios are shown in Figure 6. These plots show the contrast between the 35– 27.5°N region, which has no clear correlation 8 of 20


3

G


10.1029/2006GC001475

Figure 6. (a) Th/Ce-143Nd/144Nd, (b) Th/Ce-D7/4, and (c) Ce/Yb-143Nd/144Nd relationships for the southern Izu-Bonin arc lavas. Symbols as for Figures 1 and 2. Assumed slab-derived end-member compositions are designated. Mixing lines between mantle wedge and slab-derived fluid end-members (the same composition as for Figures 5a, 5c, and 5e) are shown. A dashed line indicates a mixing line between mantle wedge and melt, where the melt consists of a 1:4 mixture of pelagic sediment and subducted volcaniclastics (the same component as in Figures 5b, 5d, and 5f).

between Th/Ce and Nd isotopes or D7/4, and the 25–24°N section which shows strong Th enrichment (i.e., high Th/Ce) in conjunction with lower 143Nd/144Nd and higher D7/4. The lowest 143 Nd/144Nd is found in the 25 to 24°N section around Iojima and is associated with high LREE/ HREE ratios such as Ce/Yb (Figure 6c).

4. Discussion 4.1. Origins of Along-Arc Isotopic Variation 4.1.1. Shichiyo Seamounts (30–27.5°N) [16] We present the Pb isotopes in this paper in the form of classical 207Pb/204Pb-206Pb/204Pb, 208 Pb/204Pb-206Pb/204Pb and 208Pb/206Pb-206Pb/204Pb plots and also in the form of D7/4 or D8/4 versus 206 Pb/204Pb. The latter style allows significant detail and structure to be recognized in the data;

in particular the trends of individual volcanoes or groups of volcanoes relative to the NHRL. [17] Lavas from the Shichiyo Seamounts form a roughly subparallel trend to the volcanoes north of 30°N (Figures 7a and 7b), and generally lie closer to the NHRL (Figures 7d and 7e). This would appear to confirm the observation that the Pb isotope ratios of both the volcanic front and back-arc seamounts in the northern Izu-Bonin arc plot systematically closer to NHRL from north to south [Ishizuka et al., 2003a]. These Pb isotope changes, combined with a lack of correlation between the Pb isotopes and enrichment in fluidmobile elements or elements enriched in subducted sediment, led us to propose an along-arc mantle wedge heterogeneity prior to slab addition. The along-arc variation in Pb isotopes between 30– 27.5°N appears to be a continuation of the northern Izu-Bonin arc trend and could imply that the mantle wedge composition prior to the addition of slab-derived component steadily changes from 9 of 20


3

G


10.1029/2006GC001475

Figure 7. Pb isotopic plots for the lavas from the southern Izu-Bonin arc. Symbols as for Figures 1 and 2. Northern Hemisphere Reference Line (NHRL) and lines of equal D7/4 and D8/4 are shown for comparison in Figures 7a and 7b. Figures 7a and 7c show isotopic range of the possible slab-derived components for the southern Izu-Bonin arc. Site 1149 sediment and basaltic crust: Hauff et al. [2003]. Subducting HIMU seamounts: Staudigel et al. [1991] and Koppers et al. [2003]. Star symbols show assumed slab-derived end-member compositions used in this study. Both external (2 s.d.) and internal errors (2 s.e.) of double-spike Pb isotopic analyses are smaller than the symbols for 20x Pb/204Pb ratios and D8/4. External errors for D7/4 by double-spike analyses (black solid line) and conventional analyses [Taylor and Nesbitt, 1998] (dotted line) are shown in Figure 7d.

north to south as far as 27.5°N [Ishizuka et al., 2003a]. In turn, this would suggest that the slabderived component and its effect on the mantle wedge remain relatively constant. [18] Overall, the observed Pb-Pb isotope trends in the 35 and 27.5°N section suggest a contribution

from a component with high D7/4 relative to MORB-like mantle; similar to the characteristics of pelagic sediment [Ishizuka et al., 2003a] and one with low D7/4; similar to subducted altered oceanic crust. The lack of correlation between Th/Ce and Nd isotopes or D7/4 is compatible with this con10 of 20


3

G


tribution being a hydrous fluid released from subducted oceanic crust and pelagic sediment. This fluid is expected to have a high Pb content with variable D7/4, but low concentrations of Nd, Th and Ce (Figure 6 and Table 2). In contrast to hydrous fluid the introduction of sediment melt is likely to lower 143Nd/144Nd and increase Th/Ce. This implies that sediment melt is not a significant contributor to the magma source beneath the 35°N to 27.5°N section of the arc. [19] The Shichiyo Seamounts as a group show a much larger scatter in the high-precision Pb isotope data than could be reasonably ascribed to mixing between two components such as a MORB-like mantle wedge and pelagic sediment. However, the Pb data from a single seamount, such as Suiyo, form a trend with a similar or slightly shallower slope than the NHRL. Among the possible reasons for this are (1) a heterogeneity in the mantle wedge beneath the seamount and (2) contribution of a small yet variable amount of a slab-derived component other than pelagic sediment and altered oceanic crust, with lower D7/4. As we will examine in the next section, volcanoes south of the Shichiyo Seamounts have high 206Pb/204Pb, moderate D7/4 and low D8/4. It is therefore possible that a component with these Pb isotope characteristics might contribute to some extent to the source of the Shichiyo Seamounts and thus perturb their Pb-Pb isotopic trends from pure two-component mixing.

4.1.2. South of the Shichiyo Seamounts (27.5–25°N) [20] The Pb isotopic characteristics of the Izu Bonin arc change markedly south of the Shichiyo seamounts (Figure 7). Here compositions have characteristically elevated 206Pb/204Pb (>18.6) and form a trend extending to high 206Pb/204Pb and decreasing D8/4. The D7/4 values of these southern volcanoes remain relatively constant at 1–2 over the range of 206 Pb/204Pb. These Pb isotope trends are indicative of the involvement of a HIMU-like component in the petrogenesis of the 27.5 to 25°N arc. Even though the Pb isotopic trends for volcanoes south of 27.5°N differ from those of the northern arc, both can be traced back to a similar compositions at low 206 Pb/204Pb (Figures 7b and 7c). This could be the mantle wedge composition beneath the Shichiyo Seamounts as it is in the isotopic range of the reported Philippine Sea MORB [Hickey-Vargas, 1991, 1998]. A further implication is that the transformation in Pb isotopes at 27.5°N and 25°N is caused by a change in slab-derived component

10.1029/2006GC001475

rather than a change in the mantle wedge. Accordingly, we use the same mantle wedge composition in isotope modeling of the Shichiyo Seamounts and volcanoes south of 27.5°N.

4.1.3. Volcanoes Between 25 and 24°N [21] The volcanoes south of 25°N plot in a similar position to those between 27.5 and 25°N in 206Pb/204Pb versus 208Pb/204Pb space, but they are not coincident in the 206Pb/204Pb versus 207Pb/204Pb plot (Figure 7). South of 25°N the volcanoes have higher D7/4 at a given 206Pb/204Pb, and form a trend between the Pb isotope compositions of subducted pelagic sediment and HIMU, suggesting major role for these two components in the material released from the slab beneath these volcanoes. [22] The arc north of 25°N produces a correlation between Sr isotopes and Ba/La (Figure 5); however, the volcanoes between 25 and 24°N are offset from this trend. This implies a different type of flux in the 25–24°N section, perhaps one with weaker fluid-mobile element enrichment relative to the arc further north. Another possibility is that the mantle wedge in the south has had a time integrated incompatible element enriched signature, and this will be discussed later. In addition to the elevated D7/4, the 25–24°N volcanoes are distinguished by their enrichment in Th relative to Ce and LREE relative to HREE; both of which correlate with decreasing 143Nd/144Nd (Figure 6). These features suggest melts as opposed to aqueous fluids are the primary flux from the slab, and this melt is generated from pelagic and/or volcanogenic sediment. [23] A recent study by Kessel et al. [2005] has demonstrated the possibility of a chemical transfer from slab to mantle wedge by supercritical fluids at slab depths greater than 160 km. Experiments involving supercritical fluids indicate that they preferentially take Th, Nd and Ce from the original solid and that solid/supercritical fluid kD Th < Ce. This implies that such liquids could give the arc similar geochemical characteristics to those imparted by a slab melt, i.e., higher Th/Ce, associated with lower 143 Nd/144Nd. It is difficult to distinguish a supercritical fluid from a siliceous melt using these parameters, and hence we assume the two phases are equivalent in the following discussion.

4.2. Origins of the HIMU-Like Component [24] The isotopic characteristics of the arc south of 27.5°N require that one component involved in the 11 of 20

122 34.7 101 3.52 19.1 3.78 11.4 10.4 3.54 0.511 0.210 0.186 0.70397 0.513149 18.40 15.45 37.65

224 0.33 4.07 0.66 178 3.99 6.40 1.578 0.238 5.6 0.095 0.190 0.70397 0.513149 18.40 15.45 37.65

AOC Fluidc

158 26.9 6.53 830 38.0 47.3 44.5 9.70 16.7 5.67 1.53 0.7095 0.51234 18.61 15.60 38.67

Pelagic Sedimentd

295 1.20 9360 40.1 26.3 6.6 0.64 233 2.55 1.56 0.7095 0.51234 18.61 15.60 38.67

Fluid From Pelagic Sedimente

393 4.96 1374 22.4 27.2 17.5 3.21 13.3 7.44 2.08 0.7095 0.51234 18.61 15.60 38.67

Pelagic Sediment Meltf

212 17.3 125 20.6 92 20.6 38.6 18.9 4.13 2.90 2.64 0.50 0.70337 0.51289 20.76 15.71 40.21

Volcaniclasticsg

396 0.161 4.97 3.78 1038 21.8 21.4 2.81 0.273 40.5 1.18 0.510 0.70337 0.51289 20.76 15.71 40.21

Fluid From Volcaniclasticsh

528 17.3 125 15.7 152 12.2 22.2 7.45 1.37 2.3 3.46 0.678 0.70337 0.51289 20.76 15.71 40.21

Melt From Volcaniclasticsi

0.526 109 25.7 5.55 0.070 0.946 1.81 6.85 15.4 0.053 2.25 0.977

AOCFluidj

0.526 109 25.7 5.55 0.070 0.946 1.81 6.85 15.4 0.053 2.25 0.977

SedimentFluidj

0.526 109 25.7 5.55 0.070 0.946 1.81 6.85 15.4 0.053 2.25 0.977

Fluid From Volcaniclasticsj

0.32 1.36 0.55 1.79 1.84 2.75 3.3 1.29 0.73 0.7

SedimentMeltk

Partition Coefficient

0.32 1.36 0.55 1.79 1.84 2.75 3.3 1.29 0.73 0.7

VolcaniclasticsMeltk

Trace element concentrations were adopted from composition of Depleted Mantle by Salters and Stracke [2004]. Pb concentration is adjusted to get Ce/Pb = 21.55 (value for depleted MORB compiled by Salters and Stracke [2004]). Isotopic compositions are estimated from the compositions of Shikoku and Parece Vela Basin lavas [Hickey-Vargas, 1991, 1998]. b Composition of the altered oceanic crust is a simple average of compositions of igneous crust at ODP Site1149 (outboard of the Izu arc) reported by Kelley et al. [2003]. Isotopic compositions of Sr, Nd and Pb are from Hauff et al. [2003]. c Fluid released from altered oceanic crust. Fluid composition was calculated by assuming 4% fluid equilibrated with altered oceanic crust. Rock/fluid partition coefficients for eclogite with rutile (garnet:cpx:rutile = 53.72:46.04:0.24) at 800°C and 4 GPa are from Kessel et al. [2005]. d Composition of bulk sediment is as reported from ODP Site 1149 by Straub et al. [2004]. Isotopic compositions of Sr, Nd and Pb are from Hauff et al. [2003]. e Fluid released from subducting sediment. 2% fluid is equilibrated with sediment using the rock/fluid partition coefficients for eclogite with rutile (garnet:cpx:rutile = 53.72:46.04:0.24) at 800°C and 4 GPa from Kessel et al. [2005]. f Partial melt of subducting sediment. Composition calculated as a 12% melt of sediment using the sediment/melt partition coefficients at 900°C and 2 GPa from Johnson and Plank [1999]. g Chemical composition of volcaniclastics with ‘‘HIMU-like’’ character is an average composition of volcanic turbidites from ODP Sites 800 and 801 reported by Plank and Langmuir [1998]. Nd and Pb isotopic compositions are average of leached whole rock data and Sr isotopic composition is an unleached whole rock datum from HIMU seamount in the South Wake seamount chain reported by Koppers et al. [2003]. h Fluid released from volcaniclastics. Fluid composition was calculated by assuming 2% fluid equilibrated with volcaniclastics. Rock/fluid partition coefficients for eclogite with rutile (garnet:cpx:rutile = 53.72:46.04:0.24) at 800°C and 4 GPa are from Kessel et al. [2005]. i Partial melt of volcaniclastics. Composition calculated as a 12% melt of volcaniclastics using the sediment/melt partition coefficients at 900°C and 2 GPa from Johnson and Plank [1999]. j Rock/fluid partition coefficients used for fluid compositions derived from altered oceanic crust, pelagic sediment and volcaniclastics. These coefficients are for eclogite with rutile (garnet:cpx:rutile = 53.72:46.04:0.24) at 800°C and 4 GPa published by Kessel et al. [2005]. k Sediment/melt and volcaniclastics/melt partition coefficients used for calculation of composition of melt of pelagic sediment and volcaniclastics. These coefficients are for pelagic sediment at 900°C and 2 GPa from Johnson and Plank [1999].

7.94 0.21 1.20 0.234 0.772 0.713 0.270 0.0358 0.0137 0.0047 0.7028 0.51309 18.00 15.47 37.80

9.80

Altered Oceanic Crust (AOC)b

G

3

a

Trace element, ppm Sr Y Zr Nb Ba La Ce Nd Sm Pb Th U 87 Sr/86Sr 143 Nd/144Nd 206 Pb/204Pb 207 Pb/204Pb 208 Pb/204Pb

Philippine Sea MORB-Type Source (PSP)a

Table 2. End-Member Components and Partition Coefficients Used in Isotopic Modeling Geochemistry Geophysics Geosystems

ishizuka et al.: izu-bonin arc isotopic variation 10.1029/2006GC001475

12 of 20


3

G


magma generation has HIMU-like characteristics. This component could either be present in the mantle wedge or be a constituent of the subducting material. Whether the HIMU-like contribution is mantle or slab derived can potentially be determined using high field strength element (HFSE) characteristics. OIB including those with a HIMU character have higher Nb/Y than MORB at similar Zr/Y [e.g., Fitton et al., 1997]. This difference between OIB and MORB holds true even though Nb/Y and Zr/Y can change extensively with degree of partial melting. The cause of elevated Nb/Y in OIB can itself be explained by subduction processing. If Nb is retained in the slab relative to Y and Zr [Ayers and Watson, 1993; Brenan et al., 1995] and potentially U relative to Pb, then when the slab is recycled into the mantle there will be a net increase in the Nb/Y and U/Pb of the resulting mixed mantle. [25] If the HIMU component in the southern IzuBonin arc is derived from the slab then its Pb isotope signature would reflect a HIMU volcaniclastics-equilibrated fluid. However, the original HFSE characteristics of the HIMU slab would not be transferred as readily as Nb, Zr and Y are not significantly mobile in aqueous fluids. This scenario would result in the arc having MORB-like (or lower) Nb/Y at a given Zr/Y and yet elevated 206 Pb/ 204 Pb. On the other hand if the HIMU signature is an inherent feature of the mantle wedge then the HFSE signature of the arc would be dominated by the mantle wedge, while the Pb could be a mixture of subduction Pb and wedge Pb. [26] The origins of the HIMU signature are tested on the Nb/Y-Zr/Y diagram in Figure 8a. In this plot the arc south of 30°N forms a trend almost identical to MORB and also to the MORB-like backarc basin basalts from the Philippine Sea region which are assumed to be close to the mantle wedge composition beneath the Izu-Bonin arc (Figure 8a) The same is true for the Izu-Bonin arc north of 30°N which is thought to have Philippine Sea MORB-like mantle wedge and fluid contribution from AOC and pelagic sediment [Taylor and Nesbitt, 1998]. This trend is significantly different from that of ocean island basalts, including presentday HIMU lavas. The arc also contrasts with the composition of HIMU seamounts that are present on the subducting Pacific Plate [Staudigel et al., 1991; Koppers et al., 2003] which have similar trace element characteristics to present-day HIMU (Figure 8a). [27] As the Nb/Y-Zr/Y relationships of the arc volcanoes are likely to reflect the characteristics

10.1029/2006GC001475

of the mantle involved in the melting, the IzuBonin wedge does not appear to have any inherent HIMU characteristics. Even the volcano with the highest 206Pb/204Pb, i.e., the strongest HIMU Pb signature, does not show any offset from MORB compositions toward OIB and HIMU (Figure 8a). [28] 206Pb/204Pb-DNb relationships reinforce this interpretation (Figure 8b). DNb is a measure of a sample’s deviation from the OIB trend on a Nb/Y-Zr/ Y plot [e.g., Fitton et al., 1997]. On this diagram, addition of HIMU mantle should increase DNb together with 206Pb/204Pb. However, addition of 2– 6% fluid (with a composition estimated in Figures 5 and 9) containing a fluid liberated from volcaniclastics with HIMU character does not change DNb significantly (Figure 8b). Since the arc does not have a positive correlation in Figure 8b, fluid addition seems to explain the data better than the involvement of HIMU mantle. These characteristics, in conjunction with the Pb and Sr isotope correlations with fluid-mobile element enrichment, imply that the HIMU Pb isotopic signature of the entire arc is imparted from the slab-derived component. [29] The origin of the HIMU signature in the subducted component is likely to be the seamounts and their associated volcaniclastic aprons on the subducting plate. South of 27.5°N there is an abundance of seamounts distributed on the Pacific outboard of the Izu-Bonin-Mariana arcs and many of these seamount chains are known to have HIMU characteristics [e.g., Staudigel et al., 1991]. Substantial HIMU volcaniclastic deposits are known to be present in the sedimentary sequence. For example ODP Site 801 outboard of the northern Mariana arc has a 190 m thick volcaniclastic sequence (42% by volume of the sediments [Plank and Langmuir, 1998]). These sediments have 206Pb/204Pb of 19.76, D7/4 of 0.2 and D8/4 of 8.1 [Plank and Langmuir, 1998] and are characterized by moderate enrichment in Th and low 143Nd/144Nd [Plank and Langmuir, 1998; Kelley et al., 2003; Hauff et al., 2003]. Fluid or melt in equilibrium with volcaniclastics or indeed the HIMU volcanic basement when mixed with a MORB-like mantle would provide the most reasonable source for the southern Izu-Bonin arc volcanoes.

4.3. What, and How Much, Is Added to the Mantle Wedge? [30] Since the constituents of the slab are expected to have different Sr and Pb isotope ratios and concentrations, Sr-Pb relations should provide us with an estimate of their relative proportions. 13 of 20


3

G


10.1029/2006GC001475

Figure 8. Zr/Y-Nb/Y and 206Pb/204Pb-DNb [Fitton et al., 1997] relationship for lavas from the southern Izu-Bonin arc. Symbols as for Figures 1 and 2. Volcanic front data: Taylor and Nesbitt [1998]. Philippine Sea MORB (PSP): Hickey-Vargas [1998]. MORB data: Fitton et al. [1997]. HIMU data: GEOROC database. Volcaniclastics: Plank and Langmuir [1998]. S. Wake smts.: HIMU basalts from Southern Wake seamount chain from Staudigel et al. [1991] and Koppers et al. [2003]. Two solid lines: reference lines mark the limits of Icelandic basalt data by Fitton et al. [1997], effectively showing OIB field. DNb in Figure 8b expresses deficiency or excess of Nb relative to the OIB field and is defined as DNb = 1.74 + log(Nb/Y) 1.92log(Zr/Y) [Fitton et al., 1997]. Black dashed curve in each plot represents mixing lines between Philippine Sea MORB and the composition of HIMU basalts from the Southern Wake seamount chain. Composition of the Philippine Sea MORB is an average composition of MORB from the Philippine Sea region from Hickey-Vargas [1991, 1998]: Nb = 2.16 ppm, Y = 27.9 ppm, Zr = 76.0 ppm, Pb = 0.60 ppm. Composition of the HIMU end-member is an average composition of HIMU basalts from Southern Wake seamount chain from Staudigel et al. [1991]: Nb = 102 ppm, Y = 13.6 ppm, Zr = 176 ppm, Pb = 13.3 ppm. The trend of this mixing line is expected to approximate the effect of contribution of HIMU-like mantle component to the mantle wedge. Solid curves in each plot represent mixing lines between MORB-like mantle wedge and slab-derived fluid end-members. These mixing lines demonstrate that fluid addition does not significantly change Nb/Y or DNb, but involvement of HIMU mantle significantly increases these parameters. This is compatible with the interpretation that the high 206Pb/204Pb characteristics of volcanoes between 27.5° and 25°N are not related to HIMU mantle but are due to addition of fluid from HIMU volcaniclastics. Capability of discrimination between MORB mantle wedge and HIMU mantle wedge is tested using Central American arc data in Figure 8a [Feigenson et al., 2004]. Fields from El Salvador (with MORB-type mantle wedge) and Costa Rica (with HIMU mantle wedge) are shown. Data sources: Carr et al. [1990]; Patino et al. [1997, 2000]; Clark et al. [1998]; Rotolo and Castorina [1998]; Feigenson et al. [2004].

Figure 9. 87Sr/86Sr versus 206Pb/204Pb for the southern Izu-Bonin arc: (a) component modeling for the lavas from the area between 30 and 25°N and (b) component modeling for the lavas from the area south of 25°N. Symbols as for Figures 1 and 2. Compositions of the end-member components are listed in Table 2. Mixing lines are drawn between the same components as in Figure 5. The 30:1 mixture and 7:1 mixture lines in Figure 9a indicate the respective proportions (in weight) of mixed fluid (AOC-sediment) and volcaniclastics fluid contributing to the fluid. The 1:4 mixture line in Figure 9b indicates the respective proportions (in weight) of melt of pelagic sediment and volcaniclastics. External errors (2 s.d.) of Sr and Pb isotopic data are smaller than the symbols. 14 of 20


3

G


Assuming a Philippine Sea MORB-type mantle, the lavas from the Shichiyo Seamounts can be roughly explained by the addition of a 2–5% slab-derived fluid composed of mixture of fluid from altered oceanic crust and pelagic sediment (Figure 9a). However, it is difficult to fully explain the isotopic range of the data especially from the southern part of the Shichiyo Seamounts using slab-derived endmembers with 206Pb/204Pb of around 18.6 or lower. Like the Pb-Pb isotope relationships, this suggests that a third component is present in this region. This is most likely to be a small contribution (much less than 10% of the slab-derived component) of fluid from volcaniclastics or lavas from HIMU seamounts. [31] In contrast to Shichiyo, the seamounts between 27.5 and 25°N form a relatively flat Sr-Pb isotope trend in Figure 9a. The increasing 206Pb/204Pb and relatively constant 87Sr/86Sr requires a much larger contribution from HIMU volcaniclastics. In general, these compositions can be explained by mixing of a fluid equilibrated with HIMU volcaniclastics with a fluid similar to the one calculated for the Shichiyo Seamounts (a fluid equilibrated with mixture of pelagic sediment and AOC). The mixing relationships shown in Figure 9a require that the contribution from the volcaniclastic material in the slab-derived component increases toward south, i.e., from 3% at 27.5°N to 12–15% at 25°N. [32] Volcanoes south of 25°N also require a contribution from HIMU volcaniclastics. However, since these volcanoes form a different Pb-Pb isotope trend relative to the 27.5–25°N volcanoes (Figure 7), an alternative mixture of the available components must be present. In particular, the mixture must involve a component with a higher D7/4 Pb and lower 206Pb/204Pb than HIMU volcaniclastics, such as pelagic sediment. As the relationship between Nd or Pb isotopes and Th/Ce suggests that sediment melting is present in this section of the arc, we have quantified the contribution of a sediment melt to the arc magma (Figures 6a and 6b and Table 2). This calculation implies a contribution of 2 – 5% of melt comprising a 1:4 mixture of pelagic sediment and HIMU volcaniclastics for the lavas south of 25°N. Addition of sediment melt in this section can also explain the low 143 Nd/144Nd and high 206Pb/204Pb found in these lavas (Figures 6 and 9).

4.4. Recycling of the HIMU Component [33] Many different subduction components have been proposed as being present beneath oceanic island arcs [e.g., Vroon et al., 1995; Elliott et al.,

10.1029/2006GC001475

1997; Class et al., 2000], but most workers have adopted altered oceanic mafic crust and sediments in their isotopic models. One reason for this is that arc Pb isotope compositions lie generally in a region bounded by mixing lines between MORBlike mantle wedge, altered oceanic crust and sediment (pelagic, terrigeneous, etc.). The sediment composition clearly varies according to its provenance [e.g., Plank and Langmuir, 1998], although most sediments have a much higher D7/4 (>+5) compared to the MORB-like mantle source and altered oceanic crust (typically D7/4 < +2 in high precision studies [e.g., Thirlwall et al., 2004; Meyzen et al., 2005]). As such the positive D7/4 values of arc lavas (+2 to +7) are compatible with the involvement of sedimentary Pb in their magma genesis. In the case of the southern Izu-Bonin arc, an end-member with low or negative D7/4 combined with high 206Pb/204Pb is required. [34] Our high-precision Pb isotope data for the volcanoes between 27.5 and 25°N forms a trend toward the typical isotopic compositions of the Southern Wake HIMU seamounts [Staudigel et al., 1991; Koppers et al., 2003] (Figures 7a and 7c). The emergence of this Pb isotope signature in the arc is closely correlated with a substantial increase in the number of seamounts on the subducting Pacific Plate, some of which are known to have HIMU character. The southward increase in the number of seamounts combined with the gravity anomaly in this region implies that the thickness of the ocean crust also increases, presumably because of the additional magmatic activity at these seamounts [Kasuga et al., 1995]. [35] This HIMU volcanic activity is associated with the south Pacific superswell region [McNutt and Fischer, 1987] and appears to be long-lasting feature which can be traced back to at least the Cretaceous [Davis et al., 1989; Staudigel et al., 1991; Koppers et al., 1998]. It is this long-lived HIMU subduction that points to recycling of HIMU being a significant contributor to mantle evolution. [36] HIMU mantle signature is normally regarded as a recycled subducted material initially with high U/Pb and Th/Pb, but low Rb/Sr. The character of the HIMU seamounts on the old Pacific Plate is slightly different. These seamounts were emplaced on the ocean crust and remained on the seafloor for between 100 and 120 Ma [Koppers et al., 2003], before being subducted, dehydrated and/or melted, and re-recycled into the mantle to generate the arc magmatism. Even though the isotopic character15 of 20


3

G


istics of the Izu-Bonin arc lavas between 27.5 and 25°N are similar to the OIB lavas with a HIMU signature, trace element characteristics have a clear ‘‘arc signature’’ such as enrichment of LILE and depletion of HFSE. [37] From this it would seem likely that the reprocessing of HIMU crust by the dehydration, melting and mixing with other subduction components might have a significant effect on the evolution of the mantle. After eruption the HIMU volcaniclastics are expected to have high U/Pb ratio relative to typical MORB and also relative to their HIMU mantle source (as kDU < kDPb). This effect is compounded by the interaction of the volcanics with seawater and/or hydrothermal solutions, which tends to increase the U content without changing the bulk Pb content [e.g., Kelley et al., 2005].

4.5. Other Causes of the Along-Arc Chemical Variation [38] The evidence above indicates that input from the subducting slab is dominated by hydrous fluid in the Izu-Bonin arc between 35–25°N, whereas south of 25°N around Iojima siliceous melt becomes a significant addition. Melt also appears to contribute to the magmatism further south into the Northern and Central Mariana arc [Elliott et al., 1997; Sun and Stern, 2001]. The sediment type transfers from pelagic dominated in the north to a volcaniclasticpelagic mix further south, although the change is not coincident with the change from hydrous fluid to melt dominated transfer to the mantle wedge. So apart from the composition and nature of the subducting material, what other factors might be controlling along arc chemical changes? [39] Seismological surveys have revealed that the subducting slab increases its subduction angle significantly in the southern Izu-Bonin arc [e.g., van der Hilst and Seno, 1993; Miller et al., 2004]. Miller et al. [2004] also suggested that the subduction of the Ogasawara Plateau causes a distortion and almost a ‘‘tear’’ in the subducting slab. This tear is located at shallow depth at 22°N [Castle and Creager, 1998] and at 350–400 km depth and between 28–31°N [Miller et al., 2004]. Furthermore the sudden deepening of the slab depth beneath the volcanic front by 20 km at 22°N has been recognized [Syracuse and Abers, 2006]. These changes in morphology and physical properties of the subducting plate appear to be correlated with our geochemical results which imply that slab-derived melt becomes an important component of magma at 25°N toward south. The

10.1029/2006GC001475

predicted tear or partial tear of a subducting slab might provide a pathway for upwelling asthenosphere. This in turn may cause a higher temperature at the slab surface or raise the slab temperature by increasing its surface area in contact with the surrounding mantle. A similar mechanism for slab melting has been proposed by Yogodzinski et al. [2001] who suggested that asthenospheric flow at the edge of the torn subducting Pacific Plate could explain the occurrence of adakite in the Aleutian arc. [40] Raising slab temperatures beneath the southern Izu-Bonin arc (25°N) relative to the northern arc could also be related to the angle of subduction. More effective convection in the wedge at higher slab angles found beneath the southern arc should raise the temperature of wedge and the surface of the slab. Furthermore, the increased convection could more effectively incorporate slab material into the mantle wedge and promote melting of subducted sediment [Peacock, 2003]. [41] Another implication from plate movement simulation is that subducted material from the Izu-Bonin-Mariana trench heads to the northwest as it descends. Castle and Creager [1999] predicted that material subducted at 22–24°N reaches 26–27°N beneath the volcanic front. This could potentially explain why there is a contribution from subducting HIMU seamounts as north as 27.5°N.

5. Conclusions [42] 1. Significant and systematic isotopic variations are recognized along the entire Izu-Bonin arc and represent one of the largest variations along an intraoceanic island arc. [43] 2. The along-arc isotopic variation between 30–27.5°N (Shichiyo Seamounts) appears to be a continuation of the trend from the northern IzuBonin arc. This implies that the fluid derived from altered oceanic crust and subducting pelagic sediment may affect the mantle wedge in a similar way to the northern arc. [44] 3. South of 27.5°N, the isotopic trends change dramatically, requiring a high 206Pb/204Pb but low D8/4 Pb component. This matches a gradual but significant southward increase in the quantity of HIMU volcaniclastics and HIMU seamounts on the Pacific Plate. [45] 4. The volcanoes south of 25°N also have an important HIMU contribution from the slab. However, the isotopic and trace element characteristics imply that melt (or supercritical fluid) is the likely 16 of 20


3

G


carrier, and that pelagic sediment returns as an important contributor to the slab flux.

Appendix A: Analytical Procedures [46] Major elements were measured on a Philips PW1404 XRF spectrometer using glass beads prepared by fusion of 0.6 g rock with 6 g of lithium tetraborate [Togashi and Terashima, 1997]. Analytical error is generally