Journal of Petrology Advance Access originally published online on August 17, 2006
Journal of Petrology 2006 47(11):2185-2232; doi:10.1093/petrology/egl041
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Contributions of Slab Fluid, Mantle Wedge and Crust to the Origin of Quaternary Lavas in the NE Japan Arc
1 DEPARTMENT OF GEOSCIENCE, SHIMANE UNIVERSITY MATSUE 690-8504, JAPAN
2 INSTITUTE OF MINERALOGY, PETROLOGY AND ECONOMIC GEOLOGY, TOHOKU UNIVERSITY AOBAKU, SENDAI 980-8578, JAPAN
RECEIVED NOVEMBER 10, 2004; ACCEPTED JULY 25, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGY OF THE NE...SAMPLES AND ANALYTICAL METHODSCHEMICAL VARIATIONS IN THE...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Quaternary lavas from the NE Japan arc show geochemical evidence of mixing between mantle-derived basalts and crustal melts at the magmatic front, whereas significant crustal signals are not detected in the rear-arc lavas. The along-arc chemical variations in lavas from the magmatic front are attributable almost entirely to geochemical variations in the crustal melts that were mixed with a common mantle-derived basalt. The mantle-derived basalts have slightly enriched Sr–Pb and depleted Nd isotopic compositions relative to the rear-arc lavas, but the variation is less pronounced if crustal contributions are eliminated. Therefore, the source mantle compositions and slab-derived fluxes are relatively uniform, both across and along the arc. Despite this, incompatible element concentrations are significantly higher in the rear-arc basalts. We examine an open-system, fluid-fluxed melting model, assuming that depleted mid-ocean ridge basalt (MORB)-source mantle melted by the addition of fluids derived from subducted oceanic crust (MORB) and sediment (SED) hybrids at mixing proportions of 7% and 3% SED in the frontal- and rear-arc sources, respectively. The results reproduce the chemical variations found across the NE Japan arc with the conditions: 0·2% fluid flux with degree of melting F = 3% at 2 GPa in the garnet peridotite field for the rear arc, and 0·7% fluid flux with F = 20% at 1 GPa in the spinel peridotite field beneath the magmatic front. The chemical process operating in the mantle wedge requires: (1) various SED–MORB hybrid slab fluid sources; (2) variable amounts of fluid; (3) a common depleted mantle source; (4) different melting parameters to explain across-arc chemical variations.
KEY WORDS: arc magma; crustal melt; depleted mantle; NE Japan; Quaternary; slab fluid
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF THE NE...SAMPLES AND ANALYTICAL METHODSCHEMICAL VARIATIONS IN THE...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Island arcs develop because of subduction of oceanic lithosphere. The fluids released from the subducting oceanic plate and their reaction with the overlying mantle wedge provide the prime control on magma genesis (Gill, 1981
; Hasegawa et al., 1991; Hawkesworth et al., 1993a; Pearce et al., 1995; Tatsumi & Eggins, 1995; Davidson, 1996; Peacock, 1996; Peacock & Wang, 1999; Stern, 2002). Controlling factors for magma genesis along convergent plate boundaries include: (1) adiabatic upwelling of asthenospheric mantle induced by slab penetration (Hasegawa et al., 1991; Peacock, 1996; Peacock & Wang, 1999); (2) partial melting of the mantle wedge as a result of the addition of slab-derived fluids (Arculus & Powell, 1986; Hawkesworth et al., 1993a; Pearce & Parkinson, 1993; Poli & Schmidt, 1995; Tatsumi & Eggins, 1995; Davidson, 1996; Iwamori, 1998; Schmidt & Poli, 1998; Hochstaedter et al., 2001); (3) melting of the subducted slab and addition of the resultant melts to the mantle wedge (Defant & Drummond, 1990; Drummond & Defant, 1990; Yogodzinski et al., 1994; Kelemen et al., 1998; Tatsumi & Hanyu, 2003).
Experimental petrology has provided important constraints for understanding melting and dehydration of the subducted slab, including phase relationships and elemental behavior in the slab (Defant & Drummond, 1990; Ayers & Watson, 1993; Poli & Schmidt, 1995; Keppler, 1996; Kogiso et al., 1997; Schmidt & Poli, 1998; Johnson & Plank, 1999). Interactions between mantle peridotite and slab-derived fluid or melt have also been studied both experimentally (Wyllie, 1982; Brenan et al., 1995; Ayers et al., 1997; Gaetani et al., 2003; McDade et al., 2003) and theoretically (Kelemen et al., 1998). Geochemical studies of marine sediments, oceanic crust and their metamorphic equivalents have revealed considerable variety in the chemical reservoirs of subduction zones (Staudigel et al., 1996; Plank & Langmuir, 1998; Bebout et al., 1999). Although we now have a much better understanding of the chemical variations of the sources of subduction-related magmas and the nature of subduction zone metamorphism and mantle metasomatic processes, quantitative treatment of the major and trace element geochemistry is still difficult because of the complex phase relationships under hydrous conditions (Poli & Schmidt, 1995; Keppler, 1996; Spandler et al., 2004). Nevertheless, the behavior of trace elements under hydrous conditions has been investigated, and more realistic trace element distribution coefficients have been determined for slab dehydration and hydrous mantle melting processes (Brenan et al., 1995; Ayers et al., 1997; Kogiso et al., 1997; Aizawa et al., 1999; Johnson & Plank, 1999; Green et al., 2000; Gaetani et al., 2003; McDade et al., 2003; Kessel et al., 2005).
Another approach is to investigate the range of chemical variation in the lavas erupted in island arcs. Chemical variations between arcs may originate from the kinematics of plate subduction (e.g. different rates or slab angles) that produce variations in thermal structure beneath arcs (Peacock, 1996; Peacock & Wang, 1999). However, the compositions of arc lavas can also vary across and along individual arcs. This probably results from: (1) differences in subducted slab materials (Plank & Langmuir, 1993); (2) differences in the dehydration or melting conditions of slab materials (Tatsumi, 1986; Defant & Drummond, 1990; Pearce & Parkinson, 1993; Hochstaedter et al., 2001); (3) differences in melting conditions in the mantle wedge (Sakuyama & Nesbitt, 1983; Pearce & Parkinson, 1993); (4) differences in the amounts of slab-derived components added to the overlying mantle wedge (Ishikawa & Nakamura, 1994; Shibata & Nakamura, 1997); (5) pre-existing mantle heterogeneity (Hochstaedter et al., 2001). Identification of the particular causes of chemical diversity in an arc system is difficult because apparently similar effects can have different causes, including modification of the primary magmas during their passage through the arc crust (Davidson, 1996; Davidson et al., 2005).
Northeastern Japan is one of the best-documented volcanic arcs (Kuno, 1966; Sakuyama & Nesbitt, 1983; Tatsumi et al., 1983; Kushiro, 1987; Nakagawa et al., 1988; Nohda et al., 1988; Togashi et al., 1992; Ohki et al., 1994; Kersting et al., 1996; Gust et al., 1997; Yoshida, 2001). The chemical variations within the NE Japan arc lavas have been discussed in terms of their major element, trace element and isotopic compositions, although there are few comprehensive datasets. Proposed geochemical models for the origin of across-arc chemical variations include: (1) different degrees of partial melting (Sakuyama & Nesbitt, 1983; Nakagawa et al., 1988); (2) the presence of two separate dehydration regions (at 100 and 150 km depths) along the slab surface, with hydrated mantle rising through the mantle wedge as diapirs (Tatsumi, 1986; Tatsumi & Eggins, 1995); (3) mixing of slab-derived fluids with the mantle wedge in variable proportions (Shibata & Nakamura, 1997); (4) mantle heterogeneity (Notsu, 1983; Togashi et al., 1992); (5) crustal contamination of mantle-derived magmas (Kersting et al., 1996; Gust et al., 1997; Kobayashi & Nakamura, 2001); (6) remelting of underplated gabbroic rocks at Moho depths in the arc crust to yield andesites (Takahashi, 1986a; Kimura et al., 2001b). Based on numerical modeling and experimentally determined phase relationships, Iwamori (1998) first simulated fluid transport and partial melting beneath the NE Japan arc. His model assumed that dehydration occurs along the slab surface and that the fluids released are captured by hydration of the overlying mantle peridotite, which is then dragged down to be released at a depth of around 150 km. The fluids released are transported back into the high-temperature mantle wedge via channels and induce melting of the mantle wedge.
Detailed tomographic images have been published for the mantle beneath the NE Japan arc (Hasegawa et al., 1991). These have been revised based on additional seismic data (Zhao et al., 1992; Nakajima & Hasegawa, 2003), and NE Japan is now the best-studied arc in terms of seismology. The tomography results suggest that a region characterized by low P- and S-wave velocities, interpreted to be a zone of relatively high temperatures (and even partial melting), exists in the middle of the mantle wedge. Tamura et al. (2002) noted a spatial correlation between volcano distribution and low-velocity regions in the underlying mantle wedge, and proposed that finger-shaped hot regions (‘hot fingers’) characterize the mantle wedge beneath NE Japan (Tamura et al., 2002). Seismologists argue that the anomalous mantle wedge region forms an inclined sheet rather than fingers, but regions of lower seismic velocity do seem to be concentrated beneath the volcanic front as well as beneath rear-arc volcanoes (Nakajima & Hasegawa, 2003). The origin of heterogeneity in the mantle wedge has become a matter of debate that might stimulate further discussion regarding the origin of the arc magmas.
Because of the large geological, geochemical, and geophysical databases available, the NE Japan arc is one of the best places for examining magma genesis beneath island arcs, although a multiplicity of ideas must be recognized and evaluated. In this paper, we focus on the geochemical variations of the Quaternary lavas along and across the NE Japan arc. We selected 61 lava samples along the volcanic front and rear arc. Of these, 57 were analyzed for trace elements, 54 were further analyzed for Sr isotopes, 43 samples for Nd isotopes and 29 samples for Pb isotopes. Combined with published geochemical data, more than 1100 major element, 198 trace element, and 168 Sr, 138 Nd, and 86 Pb isotope ratio analyses are available for the arc. Based on these data, we discuss the origin of the geochemical variations in the lavas both along and across the NE Japan arc. An important result is the identification of crustal melt contributions to the lava chemistry, particularly in the frontal-arc lavas. By removing the effects of this crustal filter, we conclude that the composition of the parental magma and its source (mantle wedge plus subducted components) is relatively uniform along and across the arc. Different degrees of melting of the common mantle source induced by a relatively constant flux of slab fluids best explain the chemical variations of the basalts within the arc. We propose here a new mantle wedge mass balance model to explain the origin of the Quaternary magmas of the NE Japan arc.
|
GEOLOGY OF THE NE JAPAN ARC |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF THE NE...SAMPLES AND ANALYTICAL METHODSCHEMICAL VARIATIONS IN THE...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
The NE Japan volcanic arc is located about 500 km west of the Japan Trench, where the Pacific Plate is being subducted at a rate of about 10 cm per year; the arc is about 100 km wide (Fig. 1a and b). Pre-Tertiary basement rocks beneath the NE Japan arc consist of Carboniferous to Jurassic metamorphic rocks of the Tamba–Ashio–Mino Terrane (MZ in Fig. 1b), Cretaceous sedimentary rocks and granitoids of the Abukuma Terrane (AZ), and an Ordovician to Cretaceous complex known as the Kitakami Terrane (KZ). The Tamba–Ashio–Mino and Abukuma Terranes are in contact along the Tanakura Tectonic Line, which is the main boundary fault between NE and SW Honshu. The Abukuma and Kitakami Terranes are separated by the Hatagawa Tectonic Line (Ichikawa, 1990
). A chain of 55 volcanoes obliquely crosses the terrane boundaries. Volcanoes in this arc define two rows, the frontal row and the rear-arc row (Kawano et al., 1961; Fujinawa, 1988; Tatsumi & Eggins, 1995). These have been called the Nasu (front) and Chokai (rear) Volcanic Zones (Kawano et al., 1961). A few volcanoes are located between the chains, and these second-order alignments of volcanoes are regarded as cross chains (Tamura et al., 2002). Quaternary intra-arc basins are always located between the second-order volcanic alignments. Here we follow the classical definition of the two volcanic rows and refer to them as the frontal arc (Nasu Volcanic Zone) and the rear arc (Chokai Volcanic Zone) (Fig. 1b). The two arcs are also geochemically distinct (Kawano et al., 1961). Frontal-arc volcanoes are located parallel to the 110 km depth contour of the Wadati–Benioff zone, whereas the rear-arc volcanoes are located above the 150–190 km depth contours (Fig. 1a). The modern arcs consist mostly of stratocone volcanoes. Establishment of the present volcanic arc dates back to about 1·7 or 2·5 Ma, and preceding stages include large-scale calderas, rather than stratocones (Yoshida, 2001).
|
Quaternary volcanoes in the NE Japan arc
Volcanism in the NE Japan arc has been thought to have been established in the Quaternary at 1·7 Ma (Yoshida, 2001
) or 1 Ma (Umeda et al., 1999). An age data compilation shows that several volcanoes have ages between 2·5 and 1·5 Ma, but these are relatively rare compared with those younger than 1·5 Ma (Fig. 2, based on Committee for Catalog of Quaternary Volcanoes in Japan, 1999). Because of this age gap, we define the Quaternary volcanic arc stage to be younger than 1·5 Ma. Volcanism between 2·5 and 1·5 Ma could be included in the Quaternary stage. We do not preclude this possibility, but focus on the volcanic arc that is younger than 1·5 Ma in this study, because this volcanic activity is typical of the NE Japan arc from that time to the present.
|
The geochemical definition of the Quaternary volcanic arc relates to the establishment of a strong across-arc variation in the K content of the lavas, which is lower along the volcanic front and gradually increases rearward (Nakagawa et al., 1988
; Yoshida, 2001). Medium-K lavas have been erupted along the volcanic front over the past 1·5 Myr, and have been accompanied by low-K lavas from as early as 1·2 Ma (Kimura et al., 2001b). The onset of low-K activity in the frontal arc occurred almost simultaneously over the entire arc, including central Japan. Extremely low-K suites occur in the very frontal arc at Osore, Adachi, and Aoso volcanoes (Nakagawa et al., 1988), and also behind the frontal-arc row at Nekoma (Kimura et al., 2001b, 2002). Therefore, we include them in the frontal-arc group in this paper. Both medium- and low-K suites were erupted at discrete eruptive centers, but they sometimes occur together in the same center, such as at Bandai, Adatara, Azuma, Zao, Funagata, and Hakkoda (Fujinawa, 1988; Kimura et al., 2001b; Fig. 2). The low-K lavas are almost always tholeiitic and the medium-K lavas calc-alkalic, which is the common type in many frontal-arc lavas (Fujinawa, 1988; Kimura et al., 2001b) in terms of iron enrichment (Miyashiro, 1974). The activity of rear-arc volcanoes began later than that along the magmatic front. Kimura (1996) first pointed out this temporal variation in the southern part of the NE Japan arc (Kimura, 1996). This observation holds over the entire NE Japan arc; rear-arc volcanoes are as old as 0·8 Ma (Umeda et al., 1999; see Fig. 2). The rear-arc lavas are almost all high-K. The high-K lavas are calc-alkalic, apart from a few exceptions at Chokai. The onset of low-K activity in the frontal arc and high-K activity in the rear arc occurred almost simultaneously over the NE Japan arc, and does not show any distinctive development of the different ‘fingers’ proposed by Tamura et al. (2002). The geochemical distinction between frontal- and rear-arc lavas is well illustrated by the variation of SiO2 vs K2O (Fig. 3) as pointed out in previous studies (Kawano et al., 1961; Nakagawa et al., 1988; Yoshida, 2001). There is some overlap between the two geochemical groups in Fig. 3, as some frontal-arc Moriyoshi lavas and high-silica Akagi lavas plot in the field of rear-arc lavas. Low-K to medium-K lavas also occur in the rear-arc Numazawa and Sunagohara volcanoes (Fig. 3). However, such cases are relatively rare.
|
The frontal and rear arcs also have different magma production rates. The erupted volumes are greater along the magmatic front (total lava volume of
4·6 km3/km arc length at a production rate of 1·5 km3/ky) than in the rear-arc chain (0·6 km3/km arc length at 0·2 km3/ky) (Committee for Catalog of Quaternary Volcanoes in Japan, 1999). Eruptions of frontal-arc volcanoes are frequent, whereas activity has been much less frequent in the rear-arc volcanoes, according to tephrochronological studies (Kimura, 1996). |
SAMPLES AND ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF THE NE...SAMPLES AND ANALYTICAL METHODSCHEMICAL VARIATIONS IN THE...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Samples
Our samples comprise basalt and intermediate lava pairs from 30 volcanoes. Several researchers have proposed that some intermediate lavas in NE Japan were produced by lower crustal melting (Takahashi, 1986a
; Fujinawa, 1991; Kobayashi & Nakamura, 2001; Kimura et al., 2002). Therefore, studying intermediate lavas together with the basalts is important for evaluation of the role of crustal filtering. Sample selection was designed to encompass all of the low-K volcanic centers along the volcanic front. Basalt to dacite samples examined range from low-K, through medium-K, to high-K. Two low-K andesite and two medium-K dacite samples are from Numazawa volcano, which has some unusual features for a rear-arc volcano (Fig. 3). Other samples selected from rear-arc high-K volcanoes include those from Kyurokujima, Chokai, Gassan, and Numazawa. We did not analyze frontal-arc medium-K samples because low-K lavas are characteristic of the frontal arc, and the trace element and isotope compositions of medium-K samples covering almost the entire frontal-arc region are available from other studies (Togashi et al., 1992; Kersting et al., 1996; Gust et al., 1997; Shibata & Nakamura, 1997). Analyses made as part of this study include 57 samples for trace elements, 54 for Sr isotopes, 43 for Nd isotopes and 29 for Pb isotopes. Interpretations in this study are based on about 1100 major element analyses from the GEOROC database (http://georoc.mpch-mainz.gwdg.de/georoc/) in addition to our new data. Data coverage for the samples discussed here is shown in Fig. 3.
Analytical methods
Samples were crushed manually in an iron pestle, rinsed with distilled water in an ultrasonic bath, and dried for 2 h at 110°C. Dried chips were then ground in an agate mortar for 30 min. The powder produced was ignited at 1100°C for 3 h. Glass disks with a 2:1 flux:sample ratio were prepared following the method of Kimura & Yamada (1996). The resulting glass disks were analyzed for major and 14 trace elements at Shimane University using a Rigaku RIX 2000 wavelength-dispersive X-ray fluorescence (XRF) spectrometer. Analytical precisions for major and trace elements are better than 1·5% and 8%, respectively.
Additional trace and ultra-trace element analyses were carried out on selected samples using solution inductively coupled plasma mass spectrometry (ICP-MS), following the methods described by Kimura et al. (1995, 2001a). Acid reagents used were EL-grade nitric (Kanto Chemicals) and hydrofluoric acid (Tama Chemicals), and analytical grade perchloric acid (Wako Chemicals). Experimental water was distilled and subsequently ion exchanged with a Milli Q filter (Millipore). Procedural blanks were all <1 ppt. The ICP-MS system used was a Thermo ELEMENTAL VG PQ3 at Shimane University, equipped with a normal concentric nebulizer and a water-chilled impact bead-type nebulizer. Instrument settings were fundamentally those of Kimura et al. (1995).
The analytical procedure for Sr and Nd isotope analyses follows Iizumi et al. (1994, 1995). Acid reagents were Supra grade hydrofluoric and nitric acids (Merck), and precise measurement grade hydrochloric acid (Wako Chemicals). Distilled-ion exchanged water and hydrochloric acid were simmered before use, and procedural blanks were <1 ppt for both Nd and Sr. Samples were analyzed by thermal ionization mass spectrometry (TIMS), using the Ta–Re double filament method, for both Nd and Sr isotopes, using a Finnigan MAT 262 system equipped with five collectors in static mode. The NIST SRM987 Sr standard and La Jolla Nd standard were analyzed before and after each 12 unknowns (Table 1). Standard values during the analyses were 87Sr/86Sr = 0·710245 ± 0·000010 (n = 3) and 143Nd/144Nd = 0·511847 ± 0·000010 (n = 3) for Nd (errors in 2 SD), respectively. Typical internal 2 SE for sample analyses are ±0·000010 for Sr and ±0·000015 for Nd.
|
Lead isotopes were analyzed following the conventional TIMS method (Kimura et al., 2003a
). Lead separation was made using DOWEX 1X8 anion exchange resin and the single column-single bead method (Manton, 1988). Acids used were Tama Chemicals TAMA PURE AA10 grade for HCl, HBr, and HF. Phosphoric acid was TAMA PURE AA100 grade. Procedural blanks for Pb were typically <100 pg. Single Ta filaments were used for analysis. Mass fractionation factors were determined by using NIST SRM981 and normalizing to the 207Pb/206Pb ratios reported by Todt et al. (1996). Mass fractionation correction factor during the analyses was 0·95 per mil per unit mass. Typical reproducibility of the analyses was within 1 per mil for each isotopic ratio. Analytical results are reported in Table 1. |
CHEMICAL VARIATIONS IN THE QUATERNARY LAVAS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF THE NE...SAMPLES AND ANALYTICAL METHODSCHEMICAL VARIATIONS IN THE...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Most of the lavas from the frontal arc are olivine basalt through olivine–pyroxene basaltic andesites to pyroxene andesites. Hydrous silicate minerals such as hornblende are occasionally found in dacites. Rear-arc mafic lavas are similar to frontal-arc equivalents. Felsic lavas and ignimbrites commonly contain hornblende and occasionally biotite (Sakuyama, 1983
). Plagioclase is ubiquitous in most of the rocks.
Major elements
The most notable chemical feature of the NE Japan arc is the spatial variation of K2O across strike (see Fig. 3). Based on the total alkalis–silica (TAS) classification (LeMaitre et al., 1989), all samples except for a few alkalic rear-arc lavas fall in the field of basalt–andesite–dacite. Like K2O, total alkali contents also differ systematically between frontal- and rear-arc lavas (Fig. 4). Iron enrichment features shown by the FeO*/MgO vs silica plot indicate that tholeiitic and calc-alkalic suites exist in both frontal- and rear-arc lavas (Tatsumi & Eggins, 1995); most of the data cluster around the discrimination boundary between high–medium Fe and medium–low Fe (Arculus, 2003). However, low-K lavas are typically tholeiitic, whereas medium-K lavas are calc-alkalic. This feature is common in frontal-arc lavas. Rear-arc high-K lavas are mostly calc-alkalic, as noted above. This across-arc variation is a general characteristic of volcanic rocks in subduction zones (Wilson, 1989). However, coexistence of low-K and medium-K lavas in the frontal arc or even within a single volcano is a notable feature of the NE Japan Quaternary volcanic arc. The occurrence of low- and medium-K lavas in individual volcanoes is not systematic. Low-K lavas tend to occur only in the earliest stages at Azuma and in the Nekoma–Bandai volcano complex, whereas they occur only in the middle stage at Adatara (see Fig. 2). The contents of MgO, TiO2, Al2O3 and P2O5 show no systematic differences between frontal- and rear-arc lavas groups at a given SiO2 content. In contrast, Fe2O3 and CaO contents in the rear-arc lavas are systematically lower than those in the frontal-arc lavas (Fig. 4). Such differences in major element compositions have been recognized previously (Sakuyama & Nesbitt, 1983; Nakagawa et al., 1988; Kaneko, 1995; Tatsumi & Eggins, 1995).
|
Radiogenic isotopes
Trace element and Sr–Nd–Pb isotope data are available from previous studies (Notsu, 1983
; Togashi et al., 1992; Kersting et al., 1996; Gust et al., 1997; Shibata & Nakamura, 1997; Kobayashi & Nakamura, 2001; Kimura et al., 2002) and this study. The compositional range of the samples studied for their isotopic composition is from basalt to dacite (SiO2 48–67 wt %). Only the basaltic lavas are likely to reflect the composition of their mantle source (Shibata & Nakamura, 1997). However, in many cases in the NE Japan arc, the isotopic compositions of lavas of the same K suite from an individual volcano do not differ greatly over the compositional range (Fujinawa, 1991; Kimura et al., 2001b). This is also true for most of the basalt–intermediate lava pairs examined in this study. Nevertheless, there can be complex variations in Nd–Sr–Pb isotope characteristics; for example: (1) Sr isotopic compositions differ between low-K tholeiitic (radiogenic) and medium-K calc-alkalic (less radiogenic) suite lavas at Adatara (Fujinawa, 1991); (2) Sr and Nd isotopic compositions of low-K tholeiitic and medium-K calc-alkalic lavas from the Nekoma–Bandai volcano complex are identical to each other (Kimura et al., 2002); (3), at Akagi volcano, Sr isotope compositions become more radiogenic with increasing silica (Kobayashi & Nakamura, 2001). Notwithstanding the complexity noted above, in general the NE Japan Quaternary lavas can reasonably be classified into frontal-arc low-K to medium-K and rear-arc high-K groups based on major element compositions, as noted above. Further subdivision is possible based on Sr–Nd isotopic characteristics (Fig. 5). In contrast to the small isotopic variations observed in the rear-arc high-K lavas, which generally appear to have been derived from a more depleted source, the more radiogenic frontal-arc lavas show large variations in Nd–Sr composition that are uncorrelated with K suite affinity. The key feature is that the Nd–Sr isotopic compositions of frontal-arc lavas are similar to those of the neighboring basement granitoids.
|
Cretaceous to Paleogene granitoids in NE Japan can be divided into three groups based on their Sr–Nd isotope characteristics (Kagami, 2005
): the Kitakami (KZ), North (NZ), and South Zones (SZ), in order of increasing Sr isotopic enrichment with Nd isotopic depletion, and from NE to SW (Fig. 5a). The North and South Zones are extensions of the isotopic zones identified in the SW Japan arc (Kagami, 2005). A Transitional Zone (TR) is found between the North and South Zones (Fig. 5a), but is here included in the North Zone for convenience. Such geochemical zoning of the NE Japan basement granitoids was previously recognized for Sr isotopes (Shibata & Ishizaka, 1979), but the proposal by Kagami (2005) extends and reinforces the observation. The zones generally correlate with the basement terranes bordered by major transform faults. However, slight offsets are seen between some of the basement tectonic and isotopic boundaries, particularly between the KZ and NZ (Fig. 5a).
Isotopic variation is rather large in NE Japan lavas of Quaternary age. A southward increase in Sr isotope ratios has been identified along the frontal arc (Notsu, 1983). Kersting et al. (1996) first noticed the spatial correlation between Nd–Sr isotopic groups of frontal-arc lavas and their underlying basement. The chemical boundaries do not correlate with terrane boundaries as well as they do with the basement granitoid zones (Fig. 5a). Clear spatial–chemical correlations are, however, found between basement granitoids and Quaternary lavas. Frontal-arc Quaternary lavas cluster tightly in three distinct areas in Nd–Sr isotope space; these are classified as QVF-I, -II, and -III (Fig. 5b). These three clusters overlap with those of the basement granitoid zones, namely QVF-I with the Kitakami Zone, QVF-II with the North Zone, and QVF-III with the South Zone (Fig. 5b). The spatial distributions of the three lava groups also correlate well with the basement granitoid zones (Fig. 5a). This striking correlation might reflect strong contributions of the underlying crust and/or mantle to the petrogenesis of the Quaternary lavas, as suggested by previous studies (Kersting et al., 1996; Kobayashi & Nakamura, 2001; Kimura et al., 2002). Hereafter, we use the lava group classification specified above. In contrast, rear-arc Quaternary lavas (QRA), always have a depleted source and plot close to the mid-ocean ridge basalt (MORB) field, notwithstanding the isotopically enriched nature of the granitoid zone (NZ) in which they were erupted. Low- to medium-K lavas from Numazawa and Sunagohara volcanoes are exceptions in the QRA, as they fall in the QVF-I field (the two open circles in Fig. 5b).
Conventional Pb isotope plots indicate that all rear-arc lavas are relatively unradiogenic, and are more similar to Indian MORB than to Pacific MORB (Fig. 6). The isotopic compositions of QRA also overlap those of frontal-arc lavas from the Izu arc (Hochstaedter et al., 2001; Ishizuka et al., 2003), and some Pb isotope compositions are similar to those of Japan Sea floor basalts (Cousens & Allan, 1992; Pouclet et al., 1995). In contrast, lavas from the magmatic front are more radiogenic than the rear-arc lavas. Volcanic front lavas plot between the field for rear-arc lavas and Pacific sediments or between the rear-arc lavas and NE Japan crustal rocks. Among the frontal-arc lavas, the QVF-I group plots closest to the Izu frontal-arc field, although they are more radiogenic. In contrast, QVF-III has the lowest 206Pb/204Pb but high 207Pb/204Pb and 208Pb/204Pb and is close in composition to some Japan Sea oceanic island basalts with EM-I signatures (Tatsumoto & Nakamura, 1991) or global lower crustal compositions (Kobayashi & Nakamura, 2001). QVF-II lavas plot between the other two QVF groups (Fig. 6a and b), as do their Nd–Sr isotopic systematics.
|
The QRA and QVF-I groups have somewhat similar and distinctive trends on Sr–Pb and Nd–Pb isotope diagrams (Fig. 7). QRA and QVF-I lavas have flatter trends on both the Sr–Pb and Nd–Pb isotope plots, overlapping the fields of Philippine Sea Plate or Indian MORB. The QRA lavas have relatively high Nd isotope ratios coupled with low Sr isotope ratios (Fig. 7a and b). The QRA lavas almost exactly overlap the Izu rear-arc lava field in both the Sr–Pb and Nd–Pb diagrams, whereas the QVF-I lavas are identical with Izu frontal-arc lavas only for Sr–Pb isotopes. The Izu frontal-arc lavas have higher Nd isotopic ratios than any other lava either from NE Japan or from the Izu rear arc (Fig. 7a and b). In contrast to the rear-arc lavas, QVF-II and -III lavas together define near-vertical linear trends that have end-members lying close to QRA and QVF-I. QVF-III lavas have the highest Sr and lowest Nd isotope ratios, and QVF-II are intermediate between QVF-III and -I. Isotopic data for the NE Japan crust are limited, but those available plot on the vertical QVF-II and -III trends. Isotopic data for QVF-II and QVF-III lavas do not trend towards the isotopic compositions of Pacific seafloor sediments (Fig. 7). QVF-III lavas are from Akagi and Nikko–Shirane volcanoes. It has been argued previously that the intermediate lavas at Akagi are crustal melts (Kobayashi & Nakamura, 2001
). QVF-II lavas are from Nekoma volcano. Kimura et al. (2002) suggested that the Nekoma lavas represent melts of lower crustal amphibolite. According to these previous studies, the isotopic trends defined by the QVF-II and QVF-III lavas can be explained by mixing of mantle-derived basalt similar to QRA or QVF-I with crustal melts. This is consistent with the isotopic compositions of the crustal rocks.
|
QVF-I volcanoes are spread over almost three-quarters of the northern frontal arc. QVF-III occurs only at Akagi and Nikko–Shirane (see Fig. 5b). The remaining volcanoes are classified as intermediate QVF-II. Two basaltic andesites from Nikko–Shirane (QVF-III) fall in the field of QVF-II, and a few dacites from Nanashigure (QVF-I) overlap the QVF-II field. A dacite from Nasu (QVF-II) plots in the QVF-III field (Fig. 5b; Table 1). Medium-K basalts in the QVF-I group have been reported from Kayou and Moriyoshi volcanoes (Shibata & Nakamura, 1997
). Isotopically, all are classified into the QVF-I group here. However, their Sr–Nd–Pb isotopic compositions fall between those of frontal-arc low-K basalts from Iwate volcano (QVF-I; radiogenic) and rear-arc high-K basalts from Kampu (QRA; less radiogenic) (Shibata & Nakamura, 1997). Low-K to medium-K lavas from Numazawa and Sunagohara (the two most radiogenic QRA lavas in Figs 5 and 6) occur in the rear-arc QRA group. In terms of Nd–Sr–Pb isotope and K suite characteristics, Numazawa–Sunagohara lavas have some affinity with QVF-I (see Fig. 5b and Table 1), which is the reverse case for Kayou and Moriyoshi in the QVF-I group.
Major element classification (e.g. low-K tholeiitic and medium-K calc-alkalic in frontal-arc lavas) seems to correlate with more radiogenic and less radiogenic isotopic compositions (respectively) in the cases of Iwate (radiogenic low-K tholeiitic), Kayou and Moriyoshi (less radiogenic medium-K calc-alkalic). The same is true at Adatara volcano for the two K suites, which are radiogenic low-K tholeiitic and less-radiogenic medium-K calc-alkalic (Fujinawa, 1991). Kayou and Moriyoshi medium-K suite lavas have features transitional between low-K Iwate (QVF-I) and high-K Kampu (QRA). However, medium- and low-K Adatara lavas are all in the more radiogenic QVF-II group. Notwithstanding the complexity and systematic variations between major element and isotopic suites, the general isotopic classification of the Quaternary lavas adopted in this study is tenable throughout the NE Japan arc, indicating a genetic link between the crust and the lavas, particularly in the frontal arc.
Trace elements
Primitive mantle normalized multi-element patterns for NE Japan lavas are shown in Fig. 8. The trace element analyses used are from recent publications (Gust et al., 1997; Shibata & Nakamura, 1997; Kobayashi & Nakamura, 2001; Kimura et al., 2002); all data were determined by ICP-MS. Some elemental data that were questionable (e.g. excessive Ta suspected to be caused by contamination) are not plotted.
|
Rear-arc basalts from six high-K centers have almost identical trace element patterns, replicating the homogeneity shown in their Nd–Sr–Pb isotopic compositions. Low-K basaltic andesite from Numazawa also has a similar trace element pattern, but with lower concentrations of large ion lithophile elements (LILE) (Fig. 8a). The QRA patterns are generally enriched in LILE (Rb, Ba, Th, and U) and light rare earth elements (LREE) and are relatively depleted in high field strength elements (HFSE) (Nb and Ta). Strong enrichments are also seen for K, Pb, and Sr, whereas Zr and Hf are slightly depleted relative to Sm. Similar patterns of relative enrichment and depletion are typical for many island arcs (Pearce & Parkinson, 1993
; Davidson, 1996).
QVF-I lavas are subdivided into basalts (SiO2 <53 wt %: QVF-IB; Fig. 8b) and intermediate lavas (QVF-IA; Fig. 8c). The QVF-I basalts examined here are all low-K and have the most depleted REE abundances. Except for spikes in Ba, Pb, and Sr, QVF-I basalts have flatter normalized patterns than QRA equivalents. Strong depletion in Nb and Ta is also evident, along with minor but clear depletion in Zr–Hf. QVF-I intermediate lavas (QVF-IA) have patterns parallel to QVF-IB, although Sr enrichment diminishes as negative Eu anomalies develop, suggesting control by fractionation of plagioclase. Depletions in Zr–Hf relative to Sm are seen in some samples, but half show no Zr–Hf depletion. Medium-K basalts from Kayou and Moriyoshi of the QVF-I group have incompatible element abundances intermediate between QVF-I low-K and QRA high-K basalts (Shibata & Nakamura, 1997). They retain strong positive LILE anomalies (Ba, Sr, Pb) and negative HFSE anomalies (Nb, Zr, Hf). Positive Ba and Pb spikes are not as prominent compared with the low-K types and slopes in the REE region are intermediate. Medium-K QVF-I basalts have incompatible element abundances and isotopic characteristics intermediate between QVF-I and QRA. This also relates to their geographical distribution across-arc (see Fig. 1).
The QVF-II Azuma and Adatara volcanoes contain both low-K and medium-K lavas. Low-K lavas are identified with filled circle symbols in Fig. 8d and e. The low-K QVF-II lavas are similar to QVF-I basalts in both absolute elemental abundances and patterns. In contrast, medium-K lavas from Azuma and Adatara have chemical abundances similar to QRA basalts. However, substantial differences are also evident, such as the modest Sr and Ba spikes and positive Zr–Hf anomalies shown by Azuma and Adatara medium-K lavas. Similar features are shown by intermediate medium-K QVF-II samples from Bandai (Fig. 8f). Intermediate Nekoma lavas are classified as QVF-II on the basis of isotopic composition. This volcano contains both low-K and medium-K lavas (Kimura et al., 2001b, 2002), and the medium-K lavas have similar trace element patterns to other QVF-II lavas (Fig. 8g). Low-K lavas have similar trace element patterns to the medium-K lavas, except for slight positive Ba and Sr spikes with almost no Zr–Hf anomalies. These characteristics are intermediate between those of low-K and medium-K suites (Fig. 8g). Other QVF-II volcanoes (Nasu to Nantai) are all medium-K and show trace element patterns intermediate between those of low-K and medium-K lavas from Adatara (Fig. 8h).
The QVF-III intermediate medium-K calc-alkalic lavas from Akagi have positive Sr spikes (Fig. 8i). Other features such as positive Ba spikes and positive Zr–Hf anomalies are similar to those in the Adatara and Bandai medium-K suite. Akagi low-K tholeiitic basalts, in contrast, are almost identical to low-K lavas from Adatara or to QVF-I basalts (Fig. 8i; patterns with filled circles). The low-K basalts alone plot in the field of QVF-II, and thus are similar to Adatara low-K tholeiitic lavas in terms of trace element and isotope compositions.
Based on incompatible trace element characteristics, the QRA and QVF-I basalts are distinctive in terms of element abundances and slopes of trace element patterns. Frontal-arc low- to medium-K suite intermediate lavas in the QVF-II and -III groups are similar in terms of positive Zr–Hf anomalies and less prominent (or no) Ba and Sr spikes. Akagi lavas differ slightly owing to higher heavy REE (HREE)/LREE. Low-K basaltic lavas from the QVF-II Adatara and Azuma and QVF-III Akagi volcanoes are similar to QVF-I basalts.
Chemical characteristics of QVF-I basalts are commonly found in all basalts of the NE Japan arc, whereas intermediate and evolved lavas from QVF-I volcanoes are distinct from QVF-I or QRA basalts. Combined with the wide isotopic variations observed for frontal-arc intermediate lavas, these geochemical characteristics might reflect significant crustal inputs. The distinctive characteristics of the isotopically depleted QRA and QVF-I basalts, and perhaps some low-K basalts from QVF-II and -III, might thus reflect mantle processes.
Correlations between incompatible element pairs are shown in Fig. 9. Co-variations are seen between K2O and Rb for all geochemical groups, with decreasing element abundances in the order QRA > QVF-III > QVF-II > QVF-I (Fig. 9a). The same is true for U and Th, with slight relative enrichment of U in QRA, as shown by its differing slope in Fig. 9b. The element pair Zr–Nb also shows strong correlation, but overall enrichments for the groups do not follow the order shown by Rb–K and U–Th. The greatest enrichments in Zr and Nb are seen in QVF-II, whereas QVF-III intermediate lavas and QVF-I basalts are the most depleted (Fig. 9c). A similar enrichment trend is true for Nb–Ta (Fig. 9e).
|
Characteristic element fractionation between middle REE (MREE) and HFSE is represented by the Sm–Zr diagram (Fig. 9d). QRA and QVF-I basalts are aligned along a line (Zr/Sm
20: thick grey line in Fig. 9d). In contrast, intermediate QVF-II and -III lavas depart from this trend. Elemental fractionation between Ba and Rb (or Ba and Th, not shown) is also apparent in the multi-element plots. The element relationship is shown on a Rb vs Ba/Rb plot (Fig. 9f). Ba/Rb is most elevated in QVF-I basalts, and decreases in response to increasing Rb through QVF-I intermediate lavas, QVF-II and QVF-III. QVF-II and -III intermediate lavas and QVF-I basalts again have contrasting features, but all still lie in a continuum, suggesting that these characteristics result from mixing.
Geochemical summary of NE Japan Quaternary lavas
The characteristics of the NE Japan Quaternary lavas are summarized in Table 2. Significant geochemical variations include: (1) low-, medium- and high-K suites with tholeiitic and calc-alkalic characteristics; (2) variable Sr–Nd–Pb isotopic compositions in frontal-arc lavas; (3) positive and negative Zr–Hf anomalies in primitive mantle normalized multi-element plots; (4) varied Ba/Rb (Ba/Th) ratios, as manifest by the Ba spikes in the multi-element plots.
|
The QRA lavas commonly have high-K calc-alkalic characteristics, non-radiogenic Sr and Pb and radiogenic Nd isotopic compositions, and negative Zr–Hf and positive Ba anomalies. Tholeiitic lavas occur at Chokai, but these are similar to the calc-alkalic lavas in other respects. Numazawa and Sunagohara lavas have medium- to low-K characteristics with no Zr–Hf and positive Ba anomalies, somewhat similar to those in some frontal-arc low-K tholeiitic lavas.
QVF-I group lavas are commonly low-K tholeiites, are isotopically more radiogenic in Pb and Sr and less radiogenic in Nd than QRA lavas, and have clear negative Zr–Hf and extremely strong positive Ba anomalies. Medium-K calc-alkalic lavas from the Moriyoshi and Kayou centers are intermediate between QVF-I and QRA for all factors.
Most QVF-II lavas are medium-K and calc-alkalic. Isotopic compositions are more radiogenic than in QVF-I, elemental patterns lack Ba anomalies, and Zr–Hf anomalies are zero to positive. Adatara and Azuma low-K tholeiitic basalts are exceptions, and are more radiogenic in Pb and Sr isotope composition than coexisting medium-K basalts. The negative Zr–Hf and strong positive Ba anomalies of these tholeiitic lavas are similar to those in QVF-I tholeiitic lavas. Nekoma intermediate low-K tholeiitic lavas have slight positive Ba anomalies but lack Zr–Hf anomalies. These features are exceptions for low-K tholeiitic suite lavas.
QVF-III lavas are also characteristically medium-K and calc-alkalic, with trace element characteristics similar to QVF-II medium-K, but have extremely radiogenic isotopic compositions. Akagi low-K tholeiitic basalt is an exception, with characteristics broadly similar to those of Adatara and Azuma low-K tholeiitic lavas.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGY OF THE NE...SAMPLES AND ANALYTICAL METHODSCHEMICAL VARIATIONS IN THE...DISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Mantle–crust components for NE Japan lavas
Mantle component in rear-arc lavas
In contrast to the frontal-arc lavas, the rear-arc lavas show no clear characteristics attributable to contamination from basement granitoids, even though they were erupted over the differing NZ and SZ granitoid zones (see Fig. 5). All QRA lavas have the most depleted source compositions and are similar to regional asthenospheric melts such as Japan Sea basalts, Philippine Sea MORB or Indian Ocean MORB (see Figs 5–7). The NE Japan mantle, together with that beneath the Japan Sea, is part of the Indian MORB domain (Hickey-Vargas, 1991
, 1998; Tatsumoto & Nakamura, 1991; Jolivet & Tamaki, 1992; Pouclet & Bellon, 1992; Hickey-Vargas et al., 1995; Pouclet et al., 1995; Flower et al., 2001), and QRA lavas indicate derivation from a similar asthenospheric source. In addition, the QRA isotopic compositions are almost identical to those in the Izu arc (Hochstaedter et al., 2001; Ishizuka et al., 2003) suggestive of a common rear-arc asthenosphere beneath the Izu and NE Japan arcs.
Some Japan Sea submarine lavas, however, have a distinctly enriched Nd–Sr isotopic source (see Fig. 7a). This led Cousens & Allan (1992) to infer addition of a subducted Pacific sediment component to the mantle source. However, multiple isotope plots indicate that mixing of NE Japan crust can also generate such enrichments (see Figs 6 and 7) (Tatsumoto & Nakamura, 1991; Pouclet et al., 1995). The Japan Sea was formed by stretching of the Eurasia continental crust, and fragments of the continental crust remain in the basin (see Fig. 1) (Tamaki et al., 1992). Regional upwelling of depleted asthenosphere during opening of the Japan Sea has been proposed by several workers (Nohda et al., 1988; Tatsumi et al., 1989; Cousens & Allan, 1992; Ohki et al., 1994; Pouclet et al., 1995; Okamura et al., 1998, 2005; Kimura et al., 2003b, 2005). Consequently, contamination of mafic melts derived from the asthenosphere by remnant crust or lithosphere may cause the isotopic enrichment in the Japan Sea floor basalts. The isotopic trends of the QRA lavas approach those of the most depleted Japan Sea source and are almost identical to Philippine Sea MORB (Fig. 7a and b). The QRA lavas thus appear to represent the isotopic composition of the asthenosphere beneath the rear-arc region, which is comparable with the Sea of Japan back-arc basin mantle asthenosphere (Nohda et al., 1988; Tatsumi et al., 1989; Cousens & Allan, 1992; Ohki et al., 1994; Pouclet et al., 1995; Okamura et al., 1998, 2005; Kimura et al., 2003b, 2005).
Mantle component in frontal-arc lavas
Although QVF-I basalts have Sr–Nd isotopic compositions identical to those in the present-day Kitakami Zone granitoids (see Fig. 5), the isotopic characteristics are relatively non-radiogenic Sr and Pb and radiogenic Nd, but not as much as in the QRA basalts (see Fig. 7). Izu frontal-arc basalts have relatively radiogenic Sr and Pb isotope compositions compared with the rear-arc basalts (Hochstaedter et al., 2001; Ishizuka et al., 2003). However, these isotopes in the QVF-I basalts are more radiogenic than those in the Izu arc (see Figs 5–7). Zr–Hf depletions relative to Sm and strong Ba enrichment relative to Th are also similar in QVF-I and QRA basalts, and this differs from other frontal-arc chemical groups (see Fig. 9d). Such trace element characteristics are common in mantle-derived arc basalts (e.g. Arculus, 1981; Tatsumi & Eggins, 1995; Plank & Langmuir, 1998; Hochstaedter et al., 2001; Ishizuka et al., 2003). Therefore, the QVF-I basalts may preserve mantle characteristics, even after modification by crustal processes. Evidence of crustal assimilation would be minimal, because of the less enriched nature of potential crustal assimilants from Kitakami Zone basement (see Fig. 5). Low-K tholeiitic basalt lavas from QVF-II (Adatara and Azuma) and QVF-III (Akagi) also have weaker but similar trace element characteristics to QVF-I, suggesting some preservation of mantle characteristics. The radiogenic Sr and Pb characteristics of these basalts reflect contamination by more radiogenic crustal materials than in QVF-I.
Crustal component in frontal-arc lavas
Quaternary lavas from the volcanic front of the NE Japan arc have isotopic compositions that are almost identical to those of the present-day basement Cretaceous to Paleogene granitoids (see Fig. 5). This suggests four options for QVF magmas: (1) melting of the granitic upper crust; (2) melting of lower crustal rock residues left after yielding the granitoids; (3) melting