| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 7 | Pages 1211-1236 | 2003
© Oxford University Press 2003
Temporal Evolution of Boron Flux in the NE Japan and Izu Arcs Measured by Ion Microprobe from the Forearc Tephra Record
1 DEPARTMENT OF GEOLOGY AND GEOPHYSICS, WOODS HOLE OCEANOGRAPHIC INSTITUTION, WOODS HOLE, MA 02543, USA
2 DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF EDINBURGH, WEST MAINS ROAD, EDINBURGH EH9 3JW, UK
3 SCRIPPS INSTITUTION OF OCEANOGRAPHY, UNIVERSITY OF CALIFORNIA, SAN DIEGO, 9500 GILMAN DRIVE, LA JOLLA, CA 92093-0244, USA
4 GEOGRAPHY AND ENVIRONMENTAL MANAGEMENT RESEARCH UNIT, UNIVERSITY OF GLOUCESTERSHIRE, CHELTENHAM GL50 4AZ, UK
* Corresponding author. E-mail: pclift{at}whoi.edu
RECEIVED JUNE 6, 2002; ACCEPTED FEBRUARY 14, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The enrichment of boron relative to similarly incompatible elements, such as Be, in arc volcanic rocks has been used as a proxy for the involvement of slab flux in petrogenesis. New ion microprobe analyses of single glass shards in tephra layers recovered by the Ocean Drilling Program (ODP) in the Izu and NE Japan forearc basins now allow the temporal variation in slab flux to be charted since 7 and 5 Ma, respectively. B/Be ratios are typically <70 in NE Japan and <130 in Izu, with no single grain exceeding 200. Although moderate to high for modern arcs, these values are much less than those recorded in the Marianas and Tonga at 3–4 Ma, shortly after the start of rifting of their back-arc basins. This observation suggests that the peak B/Be values seen in Tonga and the Marianas are related to the tectonics of slab roll-back and basin opening, rather than changes in the dynamics of the Pacific Plate. There is no temporal trend to enrichment in the high field strength elements (HFSE) or rare earth elements (REE) in either Izu or NE Japan since 7 Ma, although the two elemental groups do show clear positive correlation. A lack of correlation between REE, HFSE and B/Be suggests that slab flux is not the only control on melting in these arcs.
11B values measured in the NE Japan glasses ranged from +4·1 to -8·3
, with significant scatter over short periods of time and even between different grains within single tephra layers. Such variations are interpreted to reflect short-term changes in the degree of sediment subduction, as well as heterogeneity in the magma chamber prior to eruption. Greater than 80% of the NE Japan boron budget appears to derive from altered oceanic crust, with additional involvement from continental trench sediments in NE Japan and carbonates in Izu. There is no evidence to support the idea of slab melting in either Izu or NE Japan arcs. KEY WORDS: boron; Pacific; NE Japan; Izu; tephra; subduction
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Quantifying the influence of fluid flux from a subducting oceanic plate on the volume and chemistry of the associated arc magmatism is central to an understanding of how melts are generated in active margin settings. Although alternative models that invoke a dominantly arc lithospheric influence (e.g. Plank & Langmuir, 1988), or that focus on the chemical composition of the mantle wedge (e.g. Ewart & Hawkesworth, 1987) have been proposed, dewatering of the subducting plate has long been recognized as the root cause of subduction magmatism. Several petrogenetic models for subduction magmatic systems invoke fluid flux from the subducting plate as the dominant control on melt generation and magmatic chemistry (e.g. Stolper & Newman, 1994; Eiler et al., 1998). However, debate continues as to whether the fluids released from the oceanic sedimentary cover or from the hydrothermally altered oceanic crust (AOC) are dominant, and what controls variability in the balance between these two sources. Debate continues as to whether melting of the AOC occurs below the volcanic front of active margins (e.g. Yogodzinski & Kelemen, 1998), or if dehydration alone is responsible for the character of the magmatism observed (e.g. Tatsumi et al., 1983). A third alternative envisages the sedimentary cover melting, while the AOC experiences dehydration (Johnson & Plank, 1999).
Sources of fluid flux
Boron has been used as a proxy for the influence of fluids derived from the subducting plate, because it is much more abundant in the subducting plate than in the upper mantle. Although it is clear that fluid is stored and released from both the AOC and the sedimentary cover during subduction, argument still continues as to whether it is the AOC (Stolper & Newman, 1994; Clift et al., 2001) or the sedimentary cover that dominates (Hole et al., 1984; Elliott et al., 1997; Plank & Langmuir, 1998).
Different geochemical methods have been used to measure the involvement of slab-derived fluids in arc magmatism. The high concentrations of large ion lithophile elements (LILE) relative to high field strength elements (HFSE) and rare earth elements (REE) in arc lavas, compared with mid-ocean ridge basalts (MORB), show that the LILE budget of arc lavas is dominated by subducted materials (e.g. Pearce, 1983). Other water-mobile elements, such as Ba, U, Rb, Sr, as well as B, show elevated concentrations in arc lavas compared with mid-ocean ridge lavas, and are inferred to be derived from the subducted slab.
Resolving between fluids derived from the sedimentary cover and the AOC is not always straightforward, but can be achieved because the boron concentrations and isotopic compositions of fluid from each source are typically very different (Morris et al., 1990; Palmer, 1991; You et al., 1993; Leeman et al., 1994; Smith et al., 1995). Although sediment 11B is variable, mostly depending on sediment composition (-17·0 to +9·2 for non-carbonate lithologies; Ishikawa & Nakamura, 1993), the mantle has relatively homogeneous negative values of 11B (-10 ± 2; Chaussidon & Marty, 1995). In contrast, AOC shows positive values of 11B (between +0·1 and +9·2 ± 0·4; Spivack & Edmond, 1987; ranging up to +24·9; Smith et al., 1995).
Studies of fluids vented from subduction accretionary prisms indicate that most boron is held in an exchangeable form in the sedimentary cover or AOC, but is not native to the mantle wedge (e.g. Spivack et al., 1987; Morris et al., 1990). Boron is lost mostly from the subducting plate at low temperatures (e.g. You et al., 1993, 1995; Deyhle & Kopf, 2002), but a minor amount is structurally bound into minerals and can be transferred to the mantle. A complete mass balance requires that some exchangeable boron is subducted to relatively deep levels, especially in colder, faster subduction zones (Bebout et al., 1999). If subduction is rapid then boron dissolved in sediment pore-waters may be subducted further than the normal shallow regions below forearcs.
Boron isotopes have been used successfully to trace the source of fluids from the subducting plate in a number of arc systems (Ishikawa & Nakamura, 1994; Ishikawa & Tera, 1999). Although the cosmogenic isotope 10Be has been found in several arcs, and is used to demonstrate the involvement of very young sediment in petrogenesis (e.g. Brown et al., 1982; Tera et al., 1986; Morris et al., 1990), this does not preclude the AOC as a major source of fluid. Indeed, the influence of the AOC, as well as sediments, as a source of fluid to the arc volcanic front has been highlighted using Pb isotopes (Miller et al., 1994), as well as the boron system (Seyfried et al., 1984; Spivack & Edmond, 1987; Clift et al., 2001).
Slab flux in the Western Pacific
The nature of flux from the subducting slabs into the volcanic fronts of the western Pacific island arcs has been investigated through direct measurement of modern arc volcanic rocks in the area, using water-mobile boron as a tracer of the fluid flux (e.g. Morris et al., 1990; Ishikawa & Tera, 1999). In addition, the temporal evolution of slab flux has been traced through B/Be measurements of volcaniclastic tephra particles in the Mariana (Clift & Lee, 1998) and Tonga forearc basins (Clift et al., 2001). B/Be is considered to be a fractionation-resistant proxy for the volume of fluid flux, because both elements are strongly incompatible in mantle phases. In both the Mariana and Tonga arcs a similar temporal pattern was observed, with B/Be at relatively low levels (<100) prior to 5 Ma, rising progressively to peak values (
500) at 3–4 Ma, then falling to lower present-day values. The peak values are also noticeably higher than any measurements made on active modern arcs worldwide, which are typically <100, and are not known to exceed 200 anywhere (Morris et al., 1990).
Clift & Lee (1998) interpreted the increase in B/Be at 3–4 Ma to reflect increased sediment subduction during an oceanward retreat of the trench during the initial opening of the Mariana Trough. However, Clift et al. (2001) noted that the peak values are too high, well above measured values for oceanic sediment (100; Morris et al., 1990), for this to be the dominant source of the fluid. Moreover, sediment boron isotopic compositions are typically negative and would not make a good source of the strongly positive compositions measured in the Tonga volcanic rocks (11B = -11·6 to +37·5; Clift et al., 2001). Consequently, Clift et al. (2001) inferred that the AOC was the dominant source to the Tonga Arc and that the peak B/Be values at 3–4 Ma must therefore reflect a period of more efficient dehydration of the slab, probably linked to the slab roll-back during Lau Basin rifting.
One possible method for simultaneously increasing fluid flux to the Mariana and Tonga arcs at 3–4 Ma would be if the flux was controlled by the regional motion of the Pacific Plate. Speeding up subduction might introduce more slab flux into the roots of the arc volcanic front. Indeed, there is evidence that the Pacific Plate accelerated its westward motion just prior to that time, at 5 Ma (Cande et al., 1995). If variation in Pacific Plate motion rather than local arc tectonics is controlling the fluid flux, and in turn the melt compositions, then a similar pulse in boron might be expected to be seen in all western Pacific arcs at this time. Furthermore, the close temporal coincidence of accelerating plate motion and arc rifting in the Marianas and Tonga may indicate a direct link between the dynamics of the Pacific and the rift tectonics of some, but not all, of the western Pacific marginal basins. It should be noted that the Sea of Japan rifted at 27 Ma and became inactive after 14 Ma (Cadet & Fujioka, 1980; Tamaki et al., 1990; Otofuji et al., 1991), well before any 5 Ma Pacific tectonic event, or deposition of the ash record considered here. Furlong et al. (1982) predicted that changes in subduction velocity may drive arc extension as a result of the increased viscous drag of the asthenosphere in the mantle wedge on the base of the arc lithosphere. Thus it is possible that changes in the behavior of the Pacific Plate govern not only arc chemistry but also the tectonic evolution of the western Pacific arcs. In this study we tested these hypotheses through analysis of the slab flux history in the Izu and NE Japan Arc systems.
|
GEOLOGY OF IZU AND NE JAPAN ARCS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
To demonstrate whether there is a regional slab dewatering event throughout the western Pacific at 3–4 Ma we examined volcaniclastic records of volcanic activity for two arcs in the region. The NE Japan and Izu arcs (Fig. 1) were selected because both are significantly different in terms of their tectonic evolution compared with the Marianas and Tonga, and are separated from these two arcs and each other, allowing a broad area of the western Pacific to be characterized. Both the NE Japan and Izu arcs are subducting thermally mature oceanic crust formed at >100 Ma (von Huene et al., 1982). At each arc the volcaniclastic sequences have been continuously sampled by the Ocean Drilling Program (ODP). ODP Site 1151 provides a continuous record for the NE Japan Arc since 5 Ma (Sacks et al., 2000), and in the Izu Arc ODP Sites 788 and 792 are used together to construct a complete record since 7 Ma. All drilled sections are rich in volcanic ash, allowing a high-resolution geochemical stratigraphy to be generated. The Izu ashes have been analyzed for a suite of trace elements and REE using whole-rock specimens, demonstrating that the material is of clear arc provenance and that there is temporal variability within the section (Hiscott & Gill, 1992; Gill et al., 1994).
|
The NE Japan and Izu arcs have important differences from the Marianas and Tonga. Most notably the NE Japan Arc is continental and has not been actively extending since
16 Ma (Tamaki et al., 1990). The Bonin Arc is a part of the intra-oceanic Izu–Bonin–Mariana Arc and is known to have begun rifting at 2 Ma along the Sumisu Rift in the vicinity of ODP Site 792 (e.g. Taylor, 1992). Therefore, if the fluid pulse seen in the Marianas and Tonga is regional and driven by changes in the Pacific Plate then peak B/Be should be visible in all four arcs at the same time.
The NE Japan Arc has been a classic continental volcanic arc active since at least the Cretaceous, as shown by the development of Mesozoic–Cenozoic accretionary complexes (Taira et al., 1988). The NE Japan Arc appears to have been in steady-state activity since 5 Ma. This period followed a hiatus in volcanism after the cessation of back-arc spreading in the Sea of Japan (Cadet & Fujioka, 1980; Clift, 1995). Since 5 Ma subduction erosion of the plate margin appears to have been continuous (von Huene & Lallemand, 1990; von Huene et al., 1994). Although the forearc is not known to have collided with any major oceanic plateaux or ridges, the scattering of smaller seamounts on the subducting Pacific Plate and the evidence of mass wasting on the trench slope during the Late Pliocene (Arthur et al., 1980) suggests that the outer forearc has probably suffered some tectonic disruption during this period. The temporal evolution of at least the explosive portion of the NE Japan Arc's magmatism is recorded in the tephra record of the forearc basin, sampled at ODP Site 1151.
Subduction in the Izu Arc is estimated to have started at around 45 Ma in an intra-oceanic setting (Karig, 1975). Since that time the arc has been split twice by rifting; first in the Oligocene during formation of the Shikoku Basin, and in the late Pliocene to form the Sumisu Rift (Taylor, 1992). The Sumisu Rift represents a continuation of the northward propagating Mariana Trough Spreading Center (e.g. Martinez et al., 1995), but has not yet progressed to full seafloor spreading. The Izu forearc basin overlies boninitic and depleted tholeiitic crust formed shortly after 45 Ma (mid-Eocene: Hickey & Frey, 1982; Bloomer et al., 1995), and has been split from the arc volcanic front during each back-arc rifting event.
|
THE NATURE OF THE ASH RECORD |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Study of the marine record has advantages and disadvantages compared with work on the arc volcanoes themselves. The provenance of recent lava sequences exposed on land is typically obvious, although determining the age of extrusion often requires radioactive dating methods. In contrast, marine airfall tephra are always in the correct stratigraphic order and can be dated using biostratigraphy with reference to an established time scale (e.g. Berggren et al., 1995). Determining the source of the tephra can sometimes be a major problem in interpreting their chemistry, because tephra can be blown far down-wind, or along strike of an arc system into a given part of the forearc basin.
Prevailing regional winds in the western Pacific favor transport of tephra from west to east, making northern Honshu the most likely source of tephra at ODP Site 1151 and the Izu Arc the principal source for ODP Sites 788 and 792. Natland (1993) used major element analyses of tephra to show an eastward transport of material from the NE Japan Arc on to the Shatsky Rise. Using the relationship between particle size and distance from source described by Walker (1971), i.e. that even for the most powerful eruptions particles >50 µm are not typically found more than 1000 km from the source volcano, it is possible to rule out influence of volcanism from central Asia. Tephra derived from the Ryukyu Arc might be deposited in the Izu forearc, but would be readily distinguishable by its continental crustal influence on the trace element chemistry. By limiting sampling to the thickest, coarsest ashes we further reduce the chance of analyzing Ryukyu tephra within the Izu sections.
Use of the distal ashes to reconstruct arc evolution is likely to result in a record biased in favor of the more explosive eruptions, typically the more silicic compositions. Thus ashes may provide information on periods of arc volcanism poorly or completely unrepresented elsewhere, as explosive eruptions may not be accompanied by a voluminous extrusive sequence located close to the center. In any case proximal lavas and pyroclastic sediments located close to the volcanic vents may no longer be exposed for sampling because of burial by more recent flows. As such the marine record provides a rather different record from that exposed on land.
Sample collection and preparation
Samples were chosen from discrete volcaniclastic layers and were preferentially taken from the coarser-grained layers where this was possible. All samples from ODP Site 1151 are considered to be primary airfall deposits (Shipboard Scientific Party, 2000), whereas those at ODP Sites 788 and 792 are dominantly redeposited debris flows and turbidites with minor primary airfall deposits (Shipboard Scientific Party, 1990; Hiscott et al., 1992). Sediments were disaggregated by being mixed with water and placed in an ultrasonic bath. After this the sediments were sieved through a 63 µm mesh and washed by a high-pressure water jet. A selection of grains from the >63 µm fraction was then mounted using epoxy in 1 inch round mounts and polished using a combination of alumina and diamond pastes. The mounts were coated in graphite prior to electron probe analysis.
|
ANALYTICAL PROCEDURE |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The tephra from ODP Site 1151 were analyzed using a Cameca Camebax Microbeam electron microprobe at Edinburgh University, UK, for a series of 10 major elements: Si, Al, Na, K, Ca, Mg, Fe, P, Ti, and Mn. A beam current of 10 nA was employed, operating at a voltage of 20 kV. A count time of 30 s was used for each element. Several grains were analyzed from each sediment sample, with the full dataset published separately by Hunt & Najman (2003). The tephra from Izu (ODP Sites 788 and 792) were analyzed for the same elements using the JEOL 733 Superprobe at the Massachusetts Institute of Technology (MIT), USA, using a 10 nA beam with a voltage offset of 15 kV.
After electron probing the mounts were cleaned, coated in gold and analyzed using the Cameca ims 3f ion microprobe at Woods Hole Oceanographic Institution (WHOI), USA, for a suite of trace elements and REE. Electron backscattered images of glass obtained on the electron microprobe often showed the presence of microphenocrysts of plagioclase within the shards. Where possible, glass grains analyzed by ion microprobe were aphyric or contained the minimum number of phenocrysts within the available sample set. There is no way of establishing whether phenocryst phases were actually absent from the host magma of an aphyric glass fragment on eruption. Results of the major and trace element analyses are shown in Tables 1 (NE Japan) and 2 (Izu). Uncertainties in the trace element analyses are 2–5% of most elements.
|
|
Boron isotopes were measured for the NE Japan tephra using a Cameca ims 1270 ion microprobe of the Northeast National Ion Microprobe Facility (NENIMF) at WHOI. Analysis followed the procedures developed by Chaussidon et al. (1997). The primary beam is accelerated to 10 keV, with a secondary accelerating voltage of 4·5 keV, and produces a beam size of 20–40 µm on the sample. Background interference levels are low,
0·05 ppm, and matrix effects are negligible (Chaussidon et al., 1997). Replicate analyses of a silicate glass standard (GB4) have produced 11B = -12·7 ± 0·8 (2
standard error), consistent with the accepted value of 11B = 12·80 (Chaussidon et al., 1997). The precision of individual analyses, involving 160 cycles of measurements (counting times were 15 s for 10B and 8 s for 11B for each cycle) are shown in Table 3.
|
Variations in boron isotope ratios are described by the notation
11B where
 |
|
MAJOR ELEMENT CHEMISTRY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Analytical totals for tephra grains were typically 4–6% less than 100%, indicating the presence of volatiles in the glass. Some of the low analytical totals are related to the loss of Na and K during analysis. Loss of Na during electron microprobe analysis of volcanic glass is well known (e.g. Neilson & Sigurdsson, 1981) and can be significant. To assess the possible influence of Na loss we plot Na2O against silica (Fig. 2a). The positively correlated trend and the total values of the Na2O contents indicate that Na2O loss was probably not significant in this study. This is because Na loss tends to affect high-silica glasses more than low-silica ones, resulting in negative relationships between Na2O and silica that are not observed here.
|
Low analytical major element totals can arise from an indigenous volatile content in the melt or from subsequent hydration, possibly without accompanying visual evidence. Distinguishing between primary magmatic and alteration-related volatile content is difficult but important to the interpretation of water-mobile element analyses, such as boron. Primary water contents in the melt may be expected to rise during crystal fractionation, as this is an incompatible component in most igneous phases (e.g. Burnham & Jahns, 1962). Consequently, it is possible to define a trend of increasing volatile content above which any given analysis may be considered to be possibly altered (see Clift & Vroon, 1996). Primary volatile contents are much higher in arcs than in other tectonic settings. Gill et al. (1994) reported 2·25% H2O in basaltic glass from the Sumisu Rift in the Izu Arc, similar to the 2% estimated for primitive Mariana Arc basalts by Newman et al. (2000), but rather less than the 6·2% measured in a mafic melt inclusion by Sisson & Layne (1993). Newman et al. (2000) further measured maximum original volatile contents of 4–6% in evolved glasses in the Marianas, compared with >13% measured by Anderson (1974) from dacitic melt inclusions from the Cascades. The work of Sisson & Layne (1993) and Anderson (1974) defines a theoretical upper limit to the volatile contents of arc melts, but because these values are derived from study of melt inclusions that have had no opportunity to devolatilize after eruption, we opt to compare our values with those taken from lavas. Consequently, we define a trend from a 2% H2O content in 49% SiO2 basalt to 6% H2O in 70% SiO2 rhyolite, as being a filter (Fig. 2b). Points from the Izu and NE Japan tephra considered here that lie above this trend probably contain a significant amount of burial-related H2O. Although the exact slope and trend of the filter is not unique, such an approach does allow those grains most likely to be altered to be distinguished from those that may be close to pristine. It does not guarantee the pristine character of the grains whose totals are high enough to pass this filter. It is noteworthy that at any given value of SiO2 there is a significant range of volatile contents, even within those glasses that fall below the filter. This probably reflects different degrees of degassing of the melt as it reached the surface, as this will depend on a series of variables, especially particle size determined by eruption volume and explosiveness, which are independent of the silica content.
The general character of the major element chemistry of the tephras is shown in Fig. 3. Figure 3a shows an alkali–iron–magnesium (AFM) triangular diagram with the Izu glasses falling on a consistent tholeiitic trend. NE Japan glasses are all very evolved and are not resolvable between tholeiitic and calc-alkaline trends on this diagram. By plotting K2O against SiO2 for all glasses (Fig. 3b) it may be seen that the Izu tephra are classified uniformly as low-K rocks according to the scheme of Peccerillo & Taylor (1976), whereas the NE Japan glasses span the full range from low- to high-K compositions. Compositions measured from samples taken from the volcanic centers themselves show that the NE Japan Arc is more alkalic than the Izu Arc, with the volcanic centers in Kyushu (SW Japan) being the most alkaline of all. Izu tephra show good overlap with the Izu–Bonin compilation of Tamura & Tatsumi (2002), consistent with their presumed provenance. Analyses from the Japan Arc tend to be from more primitive rocks from which the more evolved NE Japan tephra might be derived through fractional crystallization. Thus although the NE Japan tephra do not overlap with the onland analyses this does not preclude these centers from being the source.
|
Co-variation diagrams demonstrate that concentrations of MgO and CaO change with SiO2 along a well-defined trend for all glasses, regardless of their age (Fig. 3c and d). A typical tholeiitic fractionation series of olivine followed by clinopyroxene, magnetite and plagioclase fractionation is consistent with the pattern. The evolved tephra may reflect a basic fractionation process. However, Tamura & Tatsumi (2002) used major elements to propose an origin of the rhyolites in the Izu Arc by remelting andesitic mid-crust. Crustal assimilation is also believed to have affected the NE Japan Arc (Kimura et al., 2002). Our analyses are also compatible with this remelting model. We note that on the basis of the major element data alone it is not possible to constrain the petrogenesis of the evolved rocks to being simply the product of fractional crystallization.
Figure 4 shows the temporal variability in SiO2 and K2O concentrations of the Izu and NE Japan tephra since 7 and 5 Ma, respectively. There is no apparent temporal progression seen in the NE Japan Arc, although there is a good deal of short-term variability in terms of K2O. The Izu Arc shows the highest values of SiO2 content between 4·2 and 1·0 Ma, prior to and immediately following the time of initiation of the Sumisu Rift (Taylor, 1992). This trend was previously noted by Hiscott & Gill (1992), and is commonly associated with arc rifting events elsewhere in the western Pacific (Clift, 1995).
|
|
TRACE ELEMENT CHEMISTRY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The trace element chemistry of the tephras can be rapidly assessed using multi-element MORB-normalized trace element variation diagrams (Fig. 5). The plots are arranged so that the compatibility of the elements in mantle phases increases in either direction away from Nb. Water-mobile elements are placed on the left of the diagram, and immobile elements on the right (modified after Pearce, 1983). The element concentrations are all normalized to N-MORB (Sun & McDonough, 1989). We plot multi-element diagrams for both basaltic and more evolved glasses from the Izu tephra to compare their form and assess the origin of the evolved liquids. All the NE Japan tephra are silica-rich glasses.
|
The plots share several common features, including the strong enrichment in water-mobile, incompatible elements for all glasses, as a result of flux from the subducting slab. The fact that the analyses appear similar to patterns derived from subaerial volcanic rocks from the associated arc volcanoes is consistent with limited remobilization of water-mobile elements during burial diagenesis. The tephra also all share the relative Nb depletion characteristic of arc rocks, except for two NE Japan tephra, 1151C-7H-1, 125–128 cm and 1151A-22R-1, 35–37 cm, in which the Nb depletion is almost non-existent (Fig. 5c). We refer to these tephra as ‘Japan exotic’. Their enriched water-mobile elements still indicate an arc source, but one with a much more enriched mantle source than the normal NE Japan tephra.
The Izu tephra show relatively flat water-immobile element patterns, close to N-MORB values, whereas the water-mobile elements are enriched, but not as strongly as in the NE Japan tephras. This is true even for the minority of rhyolitic Izu tephra, which otherwise show greater concentrations than their basaltic equivalents for all incompatible trace elements. The enrichment, especially in the most incompatible elements, may be a simple product of fractional crystallization. The trace element characteristics of the Izu tephra are consistent with the basaltic and rhyolitic glasses being derived by fractional crystallization from a common parental magma with arc characteristics. It is noteworthy that the NE Japan tephra show similar element concentrations to the Izu tephra for the most compatible elements, but are much more enriched in the incompatible elements, even compared with the rhyolitic Izu glasses. This suggests that much of that relative enrichment is not merely due to fractional crystallization, but must have been present in the parental liquid, as a result of differences in the source mantle, the nature of sediment subducted at the trench or crustal contamination prior to eruption. Tamura & Tatsumi (2002) demonstrated using major elements that rhyolitic magmas in the Izu Arc may be formed by remelting of pre-existing andesitic crust. Similarly, Kimura et al. (2002) used major and trace element data, together with Nd and Sr isotopes, to propose that low-K andesites from Nekoma volcano (Fig. 1) were derived from melting of the lower crust. Thus, crustal melting might be expected to be a common influence throughout the NE Japan Arc. As the continental crust is typically relatively enriched in incompatible trace elements this process could be important in the formation of some of the more enriched compositions seen in NE Japan.
The petrogenesis of the tephra can be further assessed through examination of the REE. NE Japan tephra show uniformly higher enrichment in the light rare earth elements (LREE) than the Izu tephra (Fig. 6). None the less, each set of tephra shows a significant degree of variability. The most rhyolitic Izu ashes are slightly more LREE enriched than the most basaltic Izu ashes, although the effect of fractional crystallization mostly increases the total concentration of REE rather than altering the relative enrichment of LREE over heavy REE (HREE). It is noteworthy that the Izu rhyolitic tephra are not as enriched as even the least enriched NE Japan tephra, implying that their patterns are also controlled by either contamination from enriched continental crust and/or a more enriched mantle source. The ‘Japan exotic’ tephra show a wide variation in LREE enrichment, although one unit (1151C-7H-1, 125–128 cm) shows extreme LREE enrichment.
|
The provenance and petrogenesis of the tephras can be partially constrained by comparison with the REE patterns of possible source volcanoes onland, where comparable data exist. In NE Japan some of the rhyolitic tephra show similar degrees of LREE enrichment to basalts analyzed at Ueno (Fig. 6a: Ujike & Stix, 2000) and Akagi volcanoes (Kobayashi & Nakamura, 2001). This implies that they could be produced by fractional crystallization of a similar basaltic parent. The same is true of the Izu tephra (Fig. 6b). However, more enriched NE Japan tephra do not provide good matches to the basalts analyzed from onshore and indicate a different petrogenesis, possibly linked to the preferential assimilation of enriched continental crust.
The relationship between REE and high field strength elements (HFSE) can be seen in Fig. 7. There is a broad positive correlation between these elemental groups, suggesting that both groups are largely acting as incompatible elements, with the most incompatible member of each group being most concentrated in the lowest degree partial melts. Although the REE may be affected by the involvement of subducted sediments in petrogenesis (e.g. Elliott et al., 1997), the HFSE in arc volcanic rocks are typically interpreted to reflect melting of the mantle wedge alone (Pearce, 1983), and thus reflect both the degree of mantle enrichment and the degree of partial melting. The correlation between the two groups seen in Fig. 7 suggests that the REE are mostly controlled by partial melting processes, source character and the degree of fractional crystallization, rather than sediment subduction. The figure shows the clear separation of the ‘Japan exotic’ tephra from the other tephra sampled at ODP Site 1151. The ‘exotic’ tephra have much higher Nb/Zr values for a given La/Sm value, implying a much more enriched mantle source.
|
Tephra provenance
Comparison of the measured trace element characteristics of the tephra particles with those measured from terrestrial exposures allows some provenance information to be derived. Many of the NE Japan grains lie close to the measured range of trace element enrichment in the Japan Arc volcanoes of Ueno, Adatara, Osore and Akagi (Figs 1, 6 and 7; Togashi et al., 1992; Ujike & Stix, 2000; Kobayashi & Nakamura, 2001). As noted above, the more incompatible element enriched compositions seen in NE Japan cannot be simply explained by simple fractionation of the basalts analyzed onland in the Japan Arc. Instead, we suggest that the enriched grains are either derived from the more enriched volcanic province in Kyushu (SW Japan; Arakawa et al., 1998; Kita et al., 2001), or are the product of a different petrogenesis than the NE Japan Arc. In this latter scenario at least some of the high-silica volcanic rocks must have a different petrogenesis than the basalts extruded and preserved close to the volcanic centers. Exactly why Kyushu is more enriched than NE Japan is controversial. Kita et al. (2001) suggested that the presence of plume-type material in the mantle wedge of Kyushu can best explain the anomalously enriched chemical compositions found in this area. Because the degree of relative enrichment of some of the ‘Japan exotic’ grains does not exceed the value of upper continental crust (Rudnick & Fountain, 1995), assimilation of crust alone can account for the variability seen in the NE Japan tephra with no need to invoke plume contamination.
Temporal evolution
The temporal evolution of the trace element chemistry of the Izu and NE Japan arcs since 7 Ma is shown in Fig. 8. Neither the Izu Arc nor the NE Japan Arc shows coherent changes in LREE or HFSE enrichment since 7 Ma. The lack of change in the Izu Arc prior to, and immediately after, the initiation of the Sumisu Rift is especially noteworthy, as it implies that the tectonics of arc extension do not dominate the Izu petrogenetic process. At all times the NE Japan tephra show higher La/Sm and Nb/Zr values, consistent with their more evolved character, but also with the fact that the NE Japan Arc is emplaced through enriched continental crust.
|
Two popular proxies for slab flux, B/Be and Li/Y, are shown in Fig. 8. Each ratio represents an element that is enriched in the sediment or AOC relative to the mantle wedge (B and Li) and another that is not (Be and Y). B/Be is especially useful because the two elements have similar compatibility in igneous phases and consequently the ratio is not disturbed by fractional crystallization. Be is less fluid mobile than boron and is thus subducted to deeper levels, behaving much like a LREE (Tatsumi & Isoyama, 1988). Although partition coefficients of B and Be can differ by two orders of magnitude under certain conditions (Chaussidon & Libourel, 1993), the correlation of B/Be with other ratios indicative of slab-derived fluids (e.g. Ba/Ce, Rb/La) demonstrates that these elements have very similar mineral–melt partition coefficients in most subduction zone environments (Ryan, 1989; Ryan & Langmuir, 1993). Partial melting and fractional crystallization processes do not therefore significantly fractionate B from Be, and variations in the B/Be ratio in arc lavas are controlled primarily by differences in the slab input to the mantle sources of the lavas.
Most arc lavas have significantly higher B/Be ratios than mid-ocean ridge and oceanic island basalts, because boron is added to the source of the arc lavas by fluids derived from the subducting slab. The B/Be ratio is thus a useful indicator of the amount of slab-derived boron in arc lavas (Morris et al., 1990; Edwards et al., 1993; Gill et al., 1993; Hochstaedter et al., 1996; Clift & Lee, 1998). Correlations of the B/Be ratio with 10Be/9Be in some arc lavas (Morris et al., 1990; Leeman et al., 1994) suggest that boron is derived at least in part from subducted sediment, and is rapidly transferred from the subducting slab to the surface in lavas (within about 5 half-lives of 10Be, or 7·5 Myr). Morris et al. (1990) noted that modern oceanic sediments have B/Be ratios of 50–60, and occasionally close to 100, so that values exceeding that level require input from an additional boron reservoir, probably the AOC, and/or preferential transfer of boron by aqueous fluid from the slab, rather than by melting. In a similar fashion, because Li is concentrated in crustal material relative to the mantle and is highly soluble in aqueous fluids, it has the potential to trace the movement of fluids and slab components in the subduction zone (Chan & Kastner, 2000). However, although Li is a fluid-mobile element, its partitioning into Mg-silicates may cause it to be effectively removed during equilibration with subarc mantle peridotite, i.e. the mantle buffers the degree of Li involvement in the arc volcanism (Tomascak et al., 2002). Elements with stronger fluid–mantle partitioning behavior, such as B, are not so affected. This makes Li a less useful proxy of flux from the subducting plate than B.
The analyses conducted here show no clear temporal trend in either B/Be or Li/Y for either the Izu or NE Japan arc. The Izu tephra show slightly more scatter in their range of B/Be, reaching a peak of 200 towards the base of the section. None the less, both sets of tephra appear to mostly range <100, less than the maximum known from Pacific pelagic sediments, measured by Morris et al. (1990). At no time since 7 Ma is there any evidence of the very high B/Be ratios noted in the Marianas and Tonga at 3–4 Ma.
The NE Japan Arc is systematically higher in Li/Y than the Izu Arc, despite the overlap in B/Be. This reflects the greater incompatibility of Li compared with Y in igneous phases, meaning that in the evolved compositions found in the NE Japan tephra there is an increase in Li/Y compared with Izu that is not driven by a difference in slab flux. Figure 9 shows the relationship between Li/Y and B/Be, and the lack of a clear positive correlation between the two for all tephra compositions, or even within the dominantly basaltic Izu tephra. The highest B/Be ratios are limited to the Izu Arc and are not accompanied by exceptionally high Li/Y.
|
The trace element data are suggestive of the melts in both arcs being significantly altered by contamination from the slab flux, and indicate that this varies strongly over short time spans. These data would be consistent with the model of Ishikawa & Tera (1999) in which boron flux to the Mariana Arc is highly variable as a result of the collision of seamounts on the subducting plate with the trench, introducing different amounts of sediment into the wedge over short time spans. Although the range of B/Be is explicable in terms of simple sediment–mantle melt mixing, the influence of boron from the AOC cannot be ruled out, as this ranges to much higher possible B/Be values. We thus employ boron isotope analyses to constrain the provenance of the fluids.
|
BORON ISOTOPES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The temporal variation in boron isotope composition,
11B, in the NE Japan Arc is shown in Fig. 10. No clear progression is apparent since 5 Ma, and considerable scatter in the range of values is noted over short time periods, and even between grains in a single layer. The total range of 11B = +4·1 to -8·3 is similar to the range measured in subaerial volcanic rocks in this region (Ishikawa & Nakamura, 1994; Ishikawa & Tera, 1999), and in other active margins using standard mass spectrometry (e.g. Palmer, 1991; Smith et al., 1997). The boron isotopic character of the volcanic rocks is controlled not only by the source of the boron but also by the degree of isotopic fractionation in the subduction zone. The degree of isotopic fractionation of the fluid expelled from the slab is a function of the temperature (Oi et al., 1989), with the greatest fractionation occurring at low temperatures.
|
Boron isotopic analyses of minerals in subduction-zone metamorphic rocks yield
11B values of -11 to -3, suggesting that slab dehydration reactions significantly lower the 11B values of subducted oceanic crust and sediments (e.g. Moran et al., 1992; Peacock & Hervig, 1997). In practice, this means that fluids with positive 11B values are lost from the slab during subduction, mostly under the forearc (You et al., 1993, 1995), but also at depth under the arc volcanic roots. Measurements of pore fluids at ODP Site 1151 by Deyhle & Kopf (2002) show very positive 11B values (up to +45), which are interpreted to be partially derived via fault conduits from the subducting plate. These data are consistent with isotopic fractionation and loss of 11B from the slab under the NE Japan Forearc, resulting in lower 11B values in the boron that reaches the arc volcanic roots.
The source and fractionation of boron from the subducting slab can be further investigated by plotting 11B values against boron concentrations, as shown in Fig. 11. In this figure the volcanic compositions are compared with the known ranges from the mantle, the altered oceanic crust and a range of possible sediment sources. The NE Japan tephra are compared with boron isotope analyzes of Izu tephra from Straub & Layne (2002). Both arcs show strong boron contamination of the mantle signature. The Izu Arc tephra have systematically more positive 11B values than their NE Japan equivalents. The range of NE Japan tephra values could be explained entirely as a mixing between boron from the AOC, or from pelagic clays, and from the mantle. In contrast, the more positive 11B Izu values require either involvement of carbonate sediments or stronger isotopic fractionation to displace them to higher 11B values. This different source for boron is consistent with the drilling data from the Pacific Plate, summarized by Plank & Langmuir (1998), showing a large thickness of pelagic carbonates offshore Izu and the Marianas, whereas further north the pelagic sedimentation is dominated by siliceous oozes and diatomites.
|
It is likely that the observed range of NE Japan tephra reflects input for all possible major end-members, i.e. pelagic clay, continental sediments, carbonates, AOC and the mantle. Consequently, Fig. 11 does not allow the degree of sediment involvement in the NE Japan boron budget to be uniquely assessed. However, because both NE Japan and Izu arcs have similar rates of convergence and similar age oceanic crust in the slab, it might be predicted that the thermal state of both subduction zones is similar. As a result we favor differences in the source of boron, rather than different degrees of isotope fractionation during subduction to explain the different magmatic values observed between the NE Japan and Izu arcs. Figure 11 suggests more influence from continental sediments or pelagic clays in NE Japan and more influence from carbonates in Izu.
Tectonic erosion of the NE Japan forearc might represent a method by which significant volumes of pelagic clays (with high boron concentration and more positive 11B values than continental sediments) deposited in the forearc basin could be incorporated into the arc magmas. The seismic section of von Huene & Lallemand (1990) shows 1·2 km of sedimentary cover to the Pacific Plate immediately offshore NE Japan and 2·0 km in the trench axis itself. Therefore, the rate of delivery of sediment to trench is 108–180 km3/Myr per km of trench axis, using an average convergence rate of 9·0 cm/yr (DeMets et al., 1990). This compares with tectonic erosion rates for the NE Japan forearc of 55 km3/Myr per km of trench axis estimated by von Huene & Lallemand (1990). Although this suggests that tectonically eroded pelagic clays could represent a major proportion of the subducted sediment load it should be remembered that the forearc basin fill probably represents a small proportion of the total eroded forearc crustal section, and thus that the contribution of pelagic clay to the arc magma is probably small. None the less, the evidence of Fig. 11 alone does not allow the relative influence of AOC and pelagic clay to be resolved uniquely.
Comparison of the range of tephra B/Be values and B concentrations confirms this general scenario and allows comparison with other arc systems (Fig. 12). The 11B values from NE Japan are generally more negative (+4·1 to -8·3) than others previously measured in Izu (+1·2 to +7·3; Ishikawa & Nakamura, 1994; +4·5 to +12·0; Straub & Layne, 2002), in Halmahera (-2 to +4; Palmer, 1991), in Martinique (-6 to +2; Smith et al., 1997), in the Kuriles (-4 to +6; Ishikawa & Tera, 1999), and in Tonga (-11·6 to +37·5; Clift et al., 2001). Given the range of possible compositions from the sedimentary cover reaching the trench, it seems likely that the reason for the boron isotopic differences between Izu and NE Japan is greater sediment influence on the boron budget resulting from greater degrees of sediment subduction in NE Japan. Certainly, there is long-term subduction erosion at both margins (von Huene & Lallemand, 1990), and more sediment flux to the trench in NE Japan.
|
Quantifying the relative influence of the AOC versus the sediments in the boron budget to the NE Japan Arc is possible using the model of Sano et al. (2001), based on trace element compositions (Fig. 13). This model was constructed for the basalts of Iwate volcano (Fig. 1), and was used to estimate that
90% of the fluid flux to Iwate was from the AOC (Sano et al., 2001). The high-silica tephra from NE Japan show significant overlap with the Iwate model, but also range to even higher degrees of AOC involvement in fluid generation. This model argues against pelagic clays, which are tectonically eroded from the NE Japan forearc, being major sources of boron to the arc volcanic front. Care must be taken, however, to understand the possible influences of extreme fractional crystallization and boron mobilization in interpreting these tephra data.
|
The Izu tephra, which are also basaltic, show a range of AOC involvement, spanning from the 95–90% seen at Iwate to much lower levels. The implication from this plot that the Izu Arc is processing more sediment than the NE Japan Arc is, however, counter-intuitive and also not supported by the higher
11B values measured from Izu (Straub & Layne, 2002). If the Izu subduction zone was colder than that under NE Japan then more positive 11B values in the Izu tephra could reflect loss of 11B deeper under the forearc than in NE Japan. However, the tectonics of the two subduction zones are similar, and so we suggest that the Sano et al. (2001) mixing model is inappropriate for Izu because it uses a Japan trench sediment end-member in the mixing calculation and not the carbonates suggested by the boron data (Fig. 11). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF IZU AND...THE NATURE OF THE...ANALYTICAL PROCEDUREMAJOR ELEMENT CHEMISTRYTRACE ELEMENT CHEMISTRYBORON ISOTOPESDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Magmatic boron and diagenesis
Any interpretation of boron isotopic data from tiny vesicular glass shards in an ash deposit surrounded by sea-water and subsequently diagenetic pore fluids must ask if the burial process itself may have affected the compositions measured. This seems unlikely in the case of the Honshu tephra because sea-water, which has a
11B value of +39·5 (Spivack & Edmond, 1987), is so much more positive than the values measured in the glasses.
As discussed above, more evolved compositions might be expected to have higher magmatic water contents than their mafic parents, and Fig. 2 shows the proposed cut-off between clearly altered and possibly fresh glasses in this dataset based largely on the volatile contents of Mariana lavas measured by Newman et al. (2000). Figure 14a shows how the proportion of volatiles in excess of the maximum magmatic levels predicted from the Newman et al. (2000) data varies with age for both Izu and NE Japan tephras. There is no clear long-term variability, although the most altered NE Japan tephras are all older than 2 Ma. At all ages, however, there is a wide spectrum of volatile contents, showing that there are tephras with essentially pristine compositions in all layers analyzed. Figure 14b and c demonstrates that there is a general negative relationship between the bulk of excess volatile contents of tephra shards and the 11B values. It is noteworthy that much of this pattern reflects the differences between Izu and Honshu grains, with little correlation visible in a single group. The negative relationship between 11B values and volatile contents is not consistent with the idea that the volatile contents are largely a function of diagenesis. As sea-water has a 11B value of +39·5 a positive correlation would be expected if alteration was governing the isotopic character. These plots do not support the idea of the measured boron isotopic character of the glasses being significantly controlled by hydration during burial diagenesis, at least back to 5 Ma. This conclusion is in accord with the study of Tonga tephra by Clift et al. (2001), which showed significant excess volatiles and a change in boron isotopes in tephra only starting in shards older than 5 Ma.
|
Slab flux as a petrogenetic control
The relationship between slab flux and the trace element chemistry of the arcs can be assessed by plotting B/Be, as a slab flux proxy, against La/Sm and Nb/Zr, as indicators of enrichment in the REE and HFSE, respectively. Figure 15 shows the lack of clear correlation between slab flux and the enrichment of these element groups. Within the Izu glasses there is a weak negative correlation between B/Be and enrichment, in that the most enriched grains always have low values of B/Be, although depleted grains are found at all B/Be values. This pattern is similar to that seen in the Tonga system (Clift et al., 2001), and can likewise be interpreted as reflecting greater degrees of partial melting of the mantle wedge when the flux is high. The NE Japan tephra do not show this sort of relationship for either Nb/Zr or La/Sm, indicating that partial melting alone does not control these element groups in NE Japan. The hypothesis that sediment subduction might play an important role in governing the REE in the arc volcanic output (e.g. Hole et al., 1984; Elliott et al., 1997) is a candidate to explain why the REE and HFSE do not behave coherently in the NE Japan Arc. In addition,