We report results of detailed petrological analyses of volcanic rocks from the Enpo seamount chain that reveal the characteristics of volcanism in a backarc en echelon seamount chain in an oceanic island arc setting. We identified two types volcanic rock suites in the Enpo seamount chain based on bulk rock trace element chemistry and compositions of chrome spinels included in olivine phenocrysts of primitive basalts. “More Enriched Suites” (MES) have enriched HFSE compositions (such as Nb, Ta), higher Nb/Zr values, and low Cr# in spinel. “Less Enriched Suites” (LES) have depleted HFSE compositions, lower Nb/Zr values, and high Cr# in spinel. These results require distinct mantle sources with different for the volcanic rocks constituting the Enpo seamount chain. Mineralogical and geochemical analyses show that petrological variations within LES lavas are explained by fractionation with open system magma mixing. In contrast, variations between MES lavas require different mechanisms of magma genesis because petrological variation within this suite cannot be explained by fractionation. MES basalts also exhibit mineralogical evidence for magma mixing. Volcanism on the Enpo seamount chain is characterized by complex spatial and temporal relationships between MES lavas consisting of subalkaline basalt, alkaline andesite, and hornblende (water rich) andesite, and LES lavas consisting of subalkaline basalt and fractionated andesite. We suggest that these features are characteristic of variability within magma systems that feed backarc seamount chains in oceanic island arcs.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The island arc “Subduction Factory” is a window into the earth's mass flux and recycling system. Examining the origins of backarc volcanism and its relationships to backarc basins expands this view, providing a broader perspective on the processes controlling the recycling of oceanic lithosphere and the generation of continental lithosphere along convergent plate margins. In the Izu-Ogasawara (Bonin) island arc, en echelon seamounts located in the backarc region form several ENE-WSW-trending seamount chains (Figure 1). The individual volcanic edifices comprising the en echelon chains are excellent candidates for geological, geophysical and volcanological investigations, and lead us to propose that these are key components of backarc volcanism. However, although there are several studies that discuss the origin and tectonics of the en echelon seamounts based on geophysical, geological and age data [e.g., Karig and Moore, 1975; Kaizuka, 1975; Bandy and Hilde, 1983; Ishizuka, 1998; Morita, 1998], petrological and geochemical data are limited [e.g., Ikeda and Yuasa, 1989; Tatsumi et al., 1992; Hochstaedter et al., 2000, 2001]. The details of the magmatic system and volcanic characteristics of the backarc en echelon seamounts are not well known, preventing the adaptation of the conventional island arc magma production model [e.g., Kuno, 1966; Tatsumi et al., 1983] to the backarc region. Detailed petrological and geochemical examinations will contribute to understanding of geochemical recycling and mass flux in the backarc “en echelon seamounts,” and will expand our understanding of the Izu-Ogasawara “subduction factory.”
 This study is the first approach to defining the magma characteristics and magma systems of the backarc en echelon seamounts using mineralogical and geochemical analyses (especially the concentration and behavior of incompatible elements). The Enpo seamount chain (the Enpo chain) was chosen for this case study, and we focus on the petrological variations and relationships of lavas sampled along the Enpo chain.
2. Geological Background
 The Izu-Ogasawara Arc, situated on the eastern edge of the Philippine Sea Plate, has grown due to subduction of the Pacific Plate (Figure 1a). It is separated into two segments (northern and southern) by the Sofugan Tectonic Line. This fault, proposed by Yuasa , marks a significant change in geological and geophysical parameters between the northern and southern part of the Izu-Ogasawara Arc. The topography of the northern segment is characterized by shallow water depths, the existence of an intra-arc rift zone, and the backarc en echelon seamounts (Figure 1b). The Intra-arc rift zone consists of the back arc knolls zone [Honza and Tamaki, 1985] and a backarc depression [Tamaki et al., 1981] which is an active backarc basin located just behind the Quaternary volcanic front (VF). The Sumisu depression, one of the largest backarc depressions, was investigated in detail using SeaMARC II sidescan sonar system and ALVIN submersible [Taylor et al., 1990; Fryer et al., 1990; Hochstaedter et al., 1990a, 1990b]. Each volcanic edifice from the forearc to the backarc depression (the Sumisu depression) were also studied by ODP drilling (Leg. 126) [e.g., Gill et al., 1992; Taylor, 1992]. Hochstaedter et al. [2000, 2001] reported geochemical characteristics of the backarc knolls. Volcanological features of lavas from the intra-arc rift zone are characterized by a bimodal distribution of basaltic and rhyolitic rock types. Back arc spreading between 30 and 15 Ma formed the Shikoku Basin [Okino et al., 1994].
 In the northern Izu-Ogasawara area, the Kan'ei, Manji, Enpo and Genroku seamount chains (from north to south) constitute four especially well documented and investigated examples of backarc en echelon seamounts (Figure 1b). Topographic and structural geological surveys were carried out since 1990 using the SeaBEAM, the IZANAGI sidescan sonar system, and single and multi channel seismic profiling systems. Volcanic rocks were recovered by the R/V Moana Wave (Univ. Hawaii, cruise MW9507) in 1995. Ishizuka  reported 40Ar/39Ar dates on these volcanic rocks and Morita  investigated the topography and geological structure in the backarc region. These diverse analyses show that volcanic activity along the backarc en echelon seamounts started around 17 Ma, slightly before the cessation of Shikoku Basin spreading (15 Ma [Okino et al., 1994]). Volcanic activity along the en echelon seamounts ceased coincident with the onset of intra-arc rifting volcanism that formed the backarc small knolls from 2.5 Ma [e.g., Hochstaedter et al., 2000; Morita, 1998]. The eastern sections of each en echelon seamount chain are overprinted by many small knolls.
3. Sample Collection and Analytical Methods
 All volcanic rock samples examined in this study were dredged by the R/V Moana Wave, University of Hawaii (cruise MW9507, 30 May–23 June 1995). About 1200 samples were described from 120 dredge hauls in the backarc region of the northern Izu-Ogasawara arc. Dredge hauls were selected using detailed IZANAGI and SeaMARC II side-scan sonar seafloor maps in combination with shipboard 3.5 kHz subbottom profiling [Morita, 1998]. In this study, the Enpo chain was chosen for the detailed petrological examination as it contains the most dredge sites and recovered samples suitable for analysis. The locations of dredge hauls are shown in Figure 1c.
 About 50 samples collected from the Enpo chain were analyzed for bulk rock composition (Table 1). An initial split of about 20 g of chips were separated from the dredged rock. These sliced samples were dipped (over three days) and ultrasonic washed in distilled water for removing of seawater contamination, and removing was confirmed by silver nitrate solution (AgNO3). The samples were pulverized using tungsten carbide mortar and agate ball mill, and H2O- and loss on ignition (LOI) were determined at 110°C and 950°C (over 6 hours), respectively. Major and minor elements were measured using a RIGAKU model 3270 X-ray fluorescence (XRF) analyzer at the Ocean Research Institute, the University of Tokyo (ORI). Major elements were determined using fused glass discs. A mixture of 0.5 g of dried and powdered sample and 5.0 g of anhydrous lithium tetraborate (Li2B4O7) was used for major element analysis, without matrix correction because of the high dilution. Minor elements were determined using pressed powder discs. Eight samples including one or two samples from each dredge site, were selected for trace element analyses using a VG model PQ3 inductively coupled plasma mass spectrometer (ICP-MS) at the Department of Geoscience, Faculty of science and Engineering, Shimane University (Table 2). Analytical procedures for trace elements including rare earth elements (REE) are the same as those used by Kimura et al. . Mineral compositions were analyzed using a JEOL JXA-733 electron probe micro analyzer (EPMA) at ORI. Silicate and oxide minerals were analyzed using a focused beam, an accelerating voltage of 15 kV, and a beam current of 12 nA. Compositional maps of 40 mm by 20 mm thin sections were collected using a JEOL JXA-8900R EPMA with five wavelength dispersive spectrometers (WDS) at ORI. Analysis was performed using an accelerating voltage 15 kV, beam current of 50 nA, and a beam diameter with less than 10 micrometers. Mineral compositions are listed in auxiliary material (see HTML version of this article).
Table 1. Bulk Rock Composition for Volcanic Rocks From the Enpo Seamount Chaina
Ten'na Seamount (N)
Ten'na Seamount (S)
Nishi-Jokyo Seamount (N)
Nishi-Jokyo Seamount (S)
Nishi-Jokyo Seamount (S)
Nishi-Jokyo Seamount (S)
All elemental concentrations were analyzed by X-ray fluorescence (XRF) analyzer.
Table 2. Trace Element Data for Volcanic Rocks From the Enpo Seamount Chaina
Ten'na (N) MWD21-2
Ten'na (S) MWD103-1
Nishi-Jokyo (N) MWD112-1
Nishi-Jokyo (S) MWD108-102
All elemental concentrations were analyzed by inductively coupled plasma mass spectrometry (ICP-MS).
4.1. Petrography, Distribution, and Age
 Nearly all volcanic rocks from the Enpo chain, except for one aphyric basalt taken from Jokyo seamount (D105b), exhibit porphyritic textures containing euhedral to subhedral phenocrysts. Detailed petrographic descriptions of volcanic rocks recovered from each seamount are presented in Table 3. Basalts were recovered from two seamounts on the backarc side of the Enpo chain, Jokyo seamount (D105) and the southern peak of Nishi-Jokyo seamount (D108) (Figure 1c). Both of these basalts are classified as ol-cpx basalt. Andesites were recovered from four volcanic edifices along the Enpo chain from the backarc side to the volcanic front side, respectively (Figure 1c): the northern peak of Nishi-Jokyo seamount (D112), the northern and southern peaks of Ten'na seamount (D21 and D103), and from Unnamed seamount (D70). Petrographically, andesites from the Enpo chain are classified into two groups. The first group of andesites, taken from D21, D103 and D70, contain clinopyroxene (augite), orthopyroxene (hypersthene) and plagioclase phenocrysts (Opx-cpx andesite); the second group, taken from D112, contains hornblende and plagioclase phenocrysts (Hbl andesite). 40Ar/39Ar ages of volcanic rocks from each dredge site (except for D21) in the Enpo chain were reported by Ishizuka , (Table 1), and range from 5.7 to 3.9 Ma.
Table 3. Petrographic Descriptions for Volcanic Rocks From the Enpo Seamount Chain
Additional descriptions and mineralogical features are given in Table 4.
Clinopyroxene: Some crystals exhibit sector zonation.
Groundmass: Intersertal textures consisting of olivine, clinopyroxene, plagioclase and glass.
Hornblende andesite, containing euhedral to subhedral plagioclase and orthopyroxene phenocrysts with Fe-Ti oxides and rare olivines as microphenocrysts.
Hornblende: Often containing Fe-Ti oxides and plagioclase (poikilitic) inclusions.
Plagioclase: Usually oscillatory-zoned, contain glass inclusions, and are surrounded by dusty zones. In contrast, plagioclase microphenocrysts are free from inclusions and dusty zone.
Groundmass: (1) hyalopilitic texture (Phase A) consisting of hornblende, plagioclase and Fe-Ti oxide, and (2) microcrystalline texture (Phase B) consisting of granular hornblende, plagioclase and Fe-Ti oxide.
Olivine-clinopyroxene basalt (D105a)
Olivine: Containing very small chromian spinel inclusions.
Clinopyroxene: Usually exhibiting oscillatory zonation.
Groundmass: Intersertal textures consisting of olivine, clinopyroxene, plagioclase and glass.
Aphyric basalt (D105b)
Intergranular or intersertal textures consisting of olivine, clinopyroxene and plagioclase, including small amounts of olivine, clinopyroxene and plagioclase microphenocrysts.
Orthopyroxene-clinopyroxene andesites with accessory plagioclase, Fe-Ti oxide and ±olivine
Orthopyroxene: Some grains exhibit clinopyroxene reaction rims.
Plagioclase: Commonly surrounded by dusty zoned rim and exhibit oscillatory zoning.
Groundmass: Intersertal textures consisting of plagioclase, Fe-Ti oxide and glass.
4.2. Bulk Rock Chemistry
 All volcanic rocks from the Enpo chain range between 47.12 and 58.57 weight percent SiO2. Basalts from Nishi-Jokyo seamount (D108) and andesites from Unnamed seamount (D70) exhibit respective linear trends in all variation diagrams (Figure 2). Most volcanic rocks from these sites are medium-K series based on Gill  (Figure 2a), or subalkaline series based on Kuno [1960, 1965] (Figure 2b), however andesites from site D21 fall into the high-K series (Figure 2a) or alkaline series (Figure 2b). Andesites, except for those from site D21 fall into high-alumina basalt series (Figure 2b). Basalts from site D105 and basaltic andesites from site D103 shows FeO* increasing trend within the tholeiite field of Miyashiro  (Figure 2c) and Kuno  (Figure 3). In contrast, basalts from site D108, opx-cpx andesites from site D70 and hlb andesite from site D112 plot in the calc-alkaline or intermediate fields. The most important aspect of these relationships is that opx-cpx andesites are further classified into subalkalic (medium-K) lavas (D103 and D70) and alkalic (high-K) lavas (D21) (Figure 2b) and each lava series is found on Ten'na seamount (D103 and D21, Figure 1c).
 Basalts from site D108 and phyric basalts from site D105a have very high MgO contents (10.61–13.48 wt%) (Figure 4a). Other samples exhibit linearly decreasing trends with lower MgO content (4.93–1.93 wt%) in MgO versus SiO2 diagrams. CaO/Al2O3 decreases as MgO decreases in all these samples (Figure 4b). The decrease in CaO/Al2O3 is consistent with fractionation of phenocrystic minerals such as olivine, pyroxene and plagioclase. Incompatible elements such as K2O, Rb, Zr, Nb inclease as MgO decreases at each dredge site (Figures 4c–4f), however basalts from site D108 (2.8–5.5 ppm), and andesites from sites D112 (13.1–14.0 ppm) and D21 (12.1–13.4 ppm) have higher Nb content than basalts from site D105 (1.2–3.7 ppm), and andesites from site D103 (2.5–5.0 ppm) and D70 (3.4–4.2 ppm) respectively for the same MgO content (Figure 4f).
 The trace element concentrations of samples with highest MgO contents from the Enpo backarc seamounts are shown in Figure 5. Basalt from site D105, basaltic andesite from site D103 and andesite from site D70 show enrichiment of large ion lithophile elements (LILE) such as K, Rb and Ba, and depletion in Nb and Ta relative to Th and Ce. Basalts from site D105 also show depletion in Zr and Hf. Basalt from site D108, andesite from site D112 and D21 exhibit no Nb depletion. Rare earth element (REE) concentrations (Figure 5) indicate that all basalts (D108, D105) and andesites, except for andesites from Nishi-Jokyo seamount (D112), are slightly enriched in light rare earth elements (LREE) relative to heavy rare earth elements (HREE). Andesites from Nishi-Jokyo seamount (D112) are more depleted with respect to HREE.
5. Petrology of Volcanic Rocks From the Enpo Seamount Chain
5.1. Petrological Differences and Relationship
 Normalized trace element spidergrams indicate that basalt from Jokyo seamount (D108) is more enriched in high-field-strength elements (HFSE), such as Nb, Ta, Zr and Hf, than basalts from Nishi-Jokyo seamount (D105), and shows similar composition to enriched mid ocean ridge basalt (E-MORB [Sun and McDonough, 1989]). In contrast, the two textural types of basalt from Jokyo seamount (phyric: D105a, aphyric: D105b) have stronger HFSE negative anomalies (Figure 5a). Furthermore, opx-cpx andesites (D30 and D70) exhibit relatively low trace element compositons in trace element versus MgO diagrams (Figures 4c–4f), and exhibit Nb, Ta, Zr and Hf negative anomalies similar to that of basalt from D105 (Figure 5b). Other andesites (D21 and D112) have higher HFSE compositions, with very high values for Nb (Figure 4f and 5c). We recognize two suites of volcanic rocks in Enpo chain: the less enriched suite (LES) comprises basalt from D105 and andesites from D103 and D70 with lower HFSE contents, while the more enriched suite (MES) comprises basalt from D108 and andesites from D112 and D21 with elevated HFSE contents.
 Basalts recovered from Nishi-Jokyo seamount (D108) and Jokyo seamount (D105) have FeO*/MgO ratios less than unity. Furthermore, magnesian olivines with high Ni contents (up to 0.4 wt.%) are found in both basalts, and clinopyroxenes with high Cr contents (up to 1.0 wt.%) are also common. These results strongly suggest that basalts from the Enpo chain formed in equilibrium with the mantle and were affected by minimal degrees of crystal differentiation. Arai  suggested that the composition of Cr-spinel in primitive basalt is indicative of parental magma character. Cr# (=Cr/(Cr + Al)) of Cr-spinels included in core of olivine phenocrysts of basalts from both seamounts fall into two different clusters (D108; 0.5–0.6, D105; 0.7) (Figure 6). MES basalts with weaker Nb and Ta depletions (D108) have lower Cr# of Cr-spinel than LES basalt with stronger Nb and Ta depletions (D105). A similar correlation between Cr# and HFSE exists in the southernmost cascade arc [Clynne and Borg, 1997]. Clynne and Borg  indicate that the range of Cr# of Cr-spinel cannot explain by differentiation or variable pressure, variations in f(O2), subsolidus equilibration or variations in degree of partial melting of a single source, and conclude that geochemical diversity reflects the relative fertility of source mantle. Furthermore, LES lavas have low Nb/Zr ratios (0.028–0.054) similar to N-MORB value (=0.03). In contrast, Nb/Zr ratios in MES lavas (0.052–0.113) are higher than in LES lavas and close to E-MORB values (=0.11) (Figure 7). These relationships indicate that petrological differences between LES and MES are explained by differences in source mantle fertility. We conclude that MES lavas are derived from a more fertile (enriched) mantle source (E-MORB-source-like) than that of LES lavas.
 MES lavas erupted between 5.78 and 4.81 Ma and LES lavas between 4.57 and 3.91 Ma (Table 1; 40Ar/39Ar ages from [Ishizuka, 1998]). These data show that these two kinds of magmas erupted over a relatively short time-span along the Enpo chain, and indicate that the source mantle beneath the Enpo chain is heterogeneous. Nd isotope values that range from +2.9 to +6.1 [Hochstaedter et al., 2001] support our observations. We suggest that similar heterogeneity occurs throughout the Izu-Ogasawara backarc region.
 Petrologically, LES samples exhibit the following features: (1) similar phenocrystic mineral assemblages (ol-cpx basalt and opx-cpx andesites), (2) similar alkalinity (high-alkali tholeiite series), (3) andesites have similar CaO/Al2O3 concentrations and fall on a linear trend from depleted basalt (D105) in CaO/Al2O3 versus MgO diagrams (Figure 4b), (4) relatively consistent Rb/Zr ratios of ≈0.2 and Rb/Nb ratios of ≈5 in variation diagrams (Figure 8). These correlations indicate that LES volcanic rocks are produced by fractionation of basaltic magma represented by basalts from the Jokyo seamount (D105), which were produced from depleted (N-MORB-source-like) mantle. In contrast, MES have much more complex petrological features: (1) different phenocrystic mineral assemblages (ol-cpx basalt, opx-cpx andesite, and hbl andesite), (2) opx-cpx andesites from D21 are classified into alkaline andesite (different alkalinity), and (3) different LILE/HFSE ratios, that is, enriched basalts (D108) have lower Rb/Zr ratios (0.04–0.22) and Rb/Nb ratios (0.81–4.30) than LES (Rb/Zr ≈0.2 and Rb/Nb ≈5). Rb/Nb ratios of alkaline opx-cpx andesites (D21) and subalkaline hbl andesites (D112) (2.38–3.68) are similar to enriched basalts (D108), however Rb/Zr ratios of andesites (0.19–0.26) are higher than LES (Figure 8). Therefore petrological differences among MES lavas cannot be explained by fractional crystallization model. We suggest that differing degrees of partial melting and/or inclusion of slab components control the petrogenesis of MES lavas.
5.2. Magma Mixing
5.2.1. Unnamed Seamount (LES)
 The geochemical trend of LES samples cannot be extended from depleted basalt (D105) to fractionation-related andesite (D70) in LILE-HFSE diagrams (Figure 8). The fractionation trend exhibits a clear inflection point between low and high LILEs on these diagrams. This suggests that a different magma system, rather than simply fractionation, is involved in the formation of the various rock types sampled in the LES suite. Figure 9 and Table 4 shows variations of mineralogical features and descriptions accompanying the progress of fractionation from basaltic andesites (D103) to andesites (D70). Whole thin section compositional maps of andesites from D70 are shown in Figure 10. Andesites from D70 have a wide compositional range of SiO2, and they can be divided into four stages by groups of samples that have similar SiO2 compositions (Figures 2 and 9). We observe disequilibrium textures such as reverse zoning of orthopyroxenes and dusty zones in plagioclase, and/or disequilibrium phenocrystic mineral assemblages such as magnesian olivine and clinopyroxene or calcic plagioclase in each stage. Figures 9 and 10 show that the frequency of disequilibrated crystals reaches a peak during more felsic stage III. Furthermore, while there are disequilibrated crystals that occur during all stages (I–IV), the overall composition of mafic or felsic minerals change to ferric or sodic, respectively. Phenocrystic mineral compositions, especially plagioclase, show bimodal distributions after stage II. This indicates that in each stage most phenocrysts reflect compositional shifts by fractionation, such as Mg decreasing in pyroxene and Ca decreasing in plagioclase. These observations strongly suggest that continuous mafic magma input and magma mixing occurred during extraction and transport of D70 andesite magma, in particular because disequilibrated crystals have chemical compositions consistent with formation from primitive magma. The magma chamber that fed the D70 andesitic magma was an open system, and fractionated with mafic magma input from outside the system. Rb shows positive correlation with SiO2 (Table 4) and SiO2 content increases with each successive stage. Disequilibrium crystals are included and/or inherited in andesites that have higher Rb content (up to 29.08 ppm). As a result, the external magma input drove the change in trend on the trace element diagram (Figure 8).
Table 4. Mineralogical Descriptions for Less Enriched Suite (LES) Andesites From the Enpo Seamount Chain
Phenocrystic crystals have normal zonning. The core of large clinopyroxene phenocrysts (=megacrysts) have 0.84 in Mg#
Clinopyroxenes have similar composition to those of found in MWD103-1basaltic andesite, but orhopyroxenes have slightly lower Mg contnet and orthopyroxenes comonly exhibit clinopyroxene reaction rims. Plagioclase crystals are uncommon, exhibit disequilibrium textures and have high Ca (An ∼ 94) cores.
In stage II the mineral assemblage and bimodal distribution of An compositions in plagioclase are evidence for disequilibrium. Disequilibrium textured plagioclases are more common than in stage I. There are rare occurrences of clear, unzoned calcic plagioclase with high Ca content (An92–90).
Stage III is characterized by the clearest evidence for disequilibrium mineral assemblages. Olivine (0.79–0.75 in Mg#) occur adjacent to magnesian clinopyroxenes (0.86–0.75 in Mg#) and there are many clear calcic and dusty zoned plagioclase crystals. Clear calcic plagioclase crystals have very narrow sodic rims (An66).
Phenocrysts are uncommon, and glomerocrysts constitute the most prominent mineral habit, with dusty rims of plagioclase on the margins of these clusters. This stage includes uncommon magnesian clinopyroxenes (0.86 to 0.75 in Mg#) and reverse zoned orthopyroxenes (rim = 0.72, core = 0.65 in Mg#). Disequilibrium textured plagioclases also are seen in this stage.
5.2.2. Nishi-Jokyo Seamount (MES)
 Enriched basalts (D108) exhibit high concentrations of LILE with increasing HFSE, especially in the Rb versus Zr diagram (Figure 8a). This trend also cannot be explained by simple fractionation. Figure 11 shows chemical characteristics of a typical olivine phenocryst. Mg# of olivine phenocrysts in sample MWD108-1 change from 0.86 in the core through 0.89 in intermediate regions to 0.82 at the rim. Other chemical components (such as NiO, Al2O3 and CaO) behave in similar fashion. These disequilibrium characteristics in olivines can be explained by mixing of different magnesian magmas over the course of magma chamber evolution and crystallization. Furthermore, olivine phenocrysts in the MWD108-7 samples with low Rb/Zr ratio (0.05) have a wide areas of low Mg#s (0.84–0.86) in their cores and have a narrow outer areas of high Mg#s (∼0.89) near their rims. These relationships indicate mixing effects between phenocrysts and newly introduced magnesian melts. Olivine phenocrysts with resorptive rims are common in samples with high Rb/Zr ratios (0.20 in sample MWD108-206) and high Mg#s (∼0.9) in their cores. We suggest that the geochemical trend (solid arrow in Figure 11) in basalts from Nishi-Jokyo seamount (D108) is related to magma mixing.
6. Volcanism in Backarc Seamounts Along Oceanic Island Arc Systems
 LES magmas with N-MORB like Nb/Zr values were erupted at Jokyo seamount (D105), Ten'na seamount (D103) and Unnamed seamount (D70), which are distributed along the constructional edifice of the Enpo seamount chain from the backarc region to the volcanic front, respectively (Figure 7). We infer that LES magmas are the major volcanic components of the Enpo chain. In contrast, there is an asystematic spatial distribution of MES magmas along the Enpo chain. Nishi-Jokyo seamount, situated furthest from the volcanic front, is dominated by enriched basalts (D108) and hornblende bearing andesites (D112) derived from more fertile mantle. Furthermore, Ten'na seamount, closer to the volcanic front, is dominated by alkaline series andesites (D21) classified into MES on the northern peak, and by fractionation-related basaltic andesites (D103), classified into LES, on the southern peak (Figures 1c and 7). We conclude that volcanism along the Enpo chain reveals complex coexistence between enriched lavas consisting of subalkaline basalts, alkaline andesites, and hornblende (water rich) andesites from a fertile mantle source, and less enriched lavas consisting of subalkaline basalts and fractionation-related andesites from more depleted mantle source. Moreover, the existence of (open system) magma mixing further complicates magma genesis along the Enpo chain. The most important relationship is that MES lavas are generally distributed in more backarc regions. Further, we emphasize the coexistence of LES and MES (or subalkaline and alkaline) andesites in Ten'na seamount.
 Similar patterns of backarc volcanism are recognized in the Mariana arc, for example “cross-chains” [e.g., Fryer and Hussong, 1982; Hussong and Fryer, 1983; Bloomer et al., 1989, 2001], backarc seamount chains extending from the volcanic front to the backarc region. These are associated with lithosphere-scale fractures or transforms in the backarc crust [Hussong and Fryer, 1983; Fryer, 1985]. The Kasuga cross-chain (Figure 1a), located in the Northern Seamount Province (NSP) [Bloomer et al., 1989], is well studied in terms of petrology and geochemistry, and is the best example for comparing the volcanic character and magmatic systems of backarc seamounts in the oceanic island arcs. Basalts of the Kasuga cross-chain recovered from two volcanic edifices (Kasuga 2 and Kasuga 3), range from medium-K to shoshonitic series, and erupted sporadically on each backarc volcano [e.g., Fryer et al., 1997]. There is no spatial correlation between alkaline and subalkaline series lavas. Stern et al.  reported that isotopic compositions of backarc basaltic lavas are remarkably heterogeneous with lower 87Sr/86Sr ratios (0.70327–0.70410) than the volcanic front and εNd ranging from +2.9 to +6.1, and that high-K lavas have lower εNd. They concluded that enriched mantle sources (OIB-source-like) exist in the mantle wedge below the Kasuga cross-chain. Furthermore, Meen et al.  showed that alkaline (low εNd: ∼3) and subalkaline (high εNd: ∼6) basaltic magmas mixed in an open system. These correspondences strongly suggest that mantle heterogeneity, indicated by coexistence of lavas from source mantle with various fertilities and open system magma mixing are common features of backarc seamount chains in the island arc system.
 Three volcanic provinces (olivine tholeiite, high-alumina basalt and alkali olivine basalt series), which are progressively more distant from the volcanic front are recognized in the quaternary Northeast Japan arc [Kuno, 1959, 1966]. Petrological models for the origin of these rocks series are discussed based on experimental results [e.g., Tatsumi et al., 1983]. On the quaternary volcanoes of the Izu-Ogasawara arc, lateral and across-arc variations similar to the Northeast Japan arc were proposed [Kuno, 1959; Onuma et al., 1983; Yuasa, 1992; Yuasa and Nohara, 1992; Ishikawa and Nakamura, 1994]. Most significantly, however, the observed complex distribution of different magma types in the Enpo and Kasuga seamount chains does not follow previous models for geochemical zonation. We cannot apply conventional island arc magma production models to backarc volcanism along the island arc system.
 Critical reviews and courteously comments by Jim Gill, an anonymous reviewer, Gray Bebout and William M. White are gratefully acknowledged. Reviews by Hidetsugu Taniguchi and Daniel Curewitz of a previous version also grateful helped to improve the manuscript. Jun-ichi Kimura is gratefully thanked for invaluable assistance with inductively coupled plasma mass spectrometry analyses and helpful comments. Mayumi Otsuki is also sincerely thanked for support of electron probe micro analyses. Detailed discussions with Kantaro Fujioka, Osamu Ishizuka, and Sumito Morita are gratefully thanked for understanding about geology of the Izu-Ogasawara arc system. Fruitful suggestion and discussions with Hiroshi Sato and Naoto Hirano are gratefully thanked.