Lava accretion system around mid-ocean ridges: Volcanic stratigraphy in the Wadi Fizh area, northern Oman ophiolite

Authors


Abstract

[1] Detailed lithological study combined with geochemical variations of lavas reveals the across-axis accretionary process at Wadi Fizh in the northern Oman ophiolite. The >900 m thick V1 sequence is divided into the lower V1 (LV1), middle V1 (MV1) and upper V1 (UV1) sequence by 0.4 m and 0.8 m thick umbers at 410 mab (meters above the base of the extrusive rocks) and 670 mab, respectively. The lowest part of the LV1 (LV1a) consists of lobate sheet and pillow lava flows extruded on the relatively flat ridge crest. Elongate pillows at 230 mab are flows draping downslope from the ridge crest and characterize the lithofacies on the ridge flank. Just above a jasper layer at 270 mab, 130 m thick evolved lavas were transported from the crest and emplaced on the ridge flank (LV1b). Off-axial accretionary processes recorded in the MV1 resulted in alternating flows of less evolved, depleted lava and evolved lava, suggesting that the MV1 off-axial lava sequence comprises flows emanated from both on- and off-axis source vents. The less evolved and depleted UV1 flows suggest independent sources distinct from the axial lavas. The Lasail Unit is regarded as a subunit of the V1 because it is comparable to the UV1 in the geological, petrological, and geochemical characteristics. The broad compositional range of the V1 sequence endorses a view that the Wadi Fizh area corresponds to a segment end of the Oman paleospreading system accompanied by off-axis volcanism as in segment boundaries of the present East Pacific Rise.

1. Introduction

[2] Recent ocean-floor explorations based on high-resolution bathymetry surveys, submersible and deep-tow camera observations and sampling as well as dredging and ocean-floor drilling have revealed construction processes of the present oceanic crust. A number of observations of on- and off-axis regions of fast spreading ridges have confirmed the occurrence of pahoehoe and lobate sheet flows on shallow slopes (<5°) on the ridge summit and off-axial plains, whereas steeper axial slopes are dominantly covered with pillow lavas [e.g.,Auzende et al., 1996; Gregg and Smith, 2003; Umino et al., 2002]. Generally, extrusive rocks emplaced on the spreading axis show a temporal variation through the crustal accretion process on the ridge axis. However, Hooft et al. [1996]reported that oceanic crustal layer 2A thickens double or more off axis compared to the thickness on axis. Such off-axis lavas are considered to have been emplaced by transportation through lava channels and tubes downslope from the sources on the ridge crest or by extrusion from off-axial vents [Geshi et al., 2007; Perfit et al., 1994; Reynolds and Langmuir, 2000; White et al., 2000, 2002, 2006a]. Detailed geological mapping of the axial terrain of fast spreading ridges has depicted narrow and shallow axial troughs accompanied by small pillow mounds, with some flows filling out of the troughs and draping downslope on the ridge flanks [Fornari et al., 1998]. In addition to these, recent accurate acoustic imagery shows many submarine volcanoes erupted off axis around the East Pacific Rise (EPR) [Scheirer and Macdonald, 1995; White et al., 2006a, 2006b]. Although a large number of off-ridge volcanoes exist in the vicinity (∼3 km) of fast spreading ridge axes, their distribution is clearly biased and clustered at around second- and third-order segment boundaries [Haymon et al., 1991; Macdonald et al., 1991; White et al., 2000, 2002, 2006a, 2006b]. On the EPR, lava tubes and channels have been found to extend several kilometers in length from the ridge crest to the flanks [Soule et al., 2005]. Off-ridge lavas from the EPR 9°–12°N show a larger compositional variation ranging from highly depleted to enriched basalts compared to axial normal mid ocean ridge basalt (MORB) s [Perfit et al., 1994; Niu and Batiza, 1997; Reynolds and Langmuir, 2000; Sims et al., 2003]. This diversity may have inherited from the heterogeneity of the mantle source because they are free from mixing in the axial magma chambers [Niu and Batiza, 1997].

[3] It is apparent that extrusive layers thicken almost continuously as the oceanic crust moves away from the ridge crest to fault-bounded basins. Eruptive style and lava flow morphology may change along with the varying basement topography and distance from the source to the site of emplacement. From lab-based fluid-dynamics experiments, lava flow morphology varies in response to the change in density, viscosity and flow rate of lava, and slope of the basement [Griffiths and Fink, 1992; Gregg and Fink, 1995, 2000]. From seafloor observations, steep flanks (>5° in gradient) of submarine volcanoes off Hawaii Islands and rise slopes of the EPR are covered predominantly with elongate pillows, but subhorizontal seamount summits and rise crests are overwhelmingly underlain by lobate sheets and pahoehoe flows [Auzende et al., 1996; Gregg and Smith, 2003; Umino et al., 2002]. The formation microscanning imagery of in situ crust at Ocean Drilling Program (ODP) Hole 1256D, located at a super-fast spread, 15 Ma EPR seafloor, has demonstrated that almost half of the extrusive layers were emplaced off axis and the uppermost 100 m thick ponded lava layer was emplaced by off-axis volcanism [Crispini et al., 2006; Tominaga and Umino, 2010].

[4] In order to understand the constructional processes of upper oceanic crust, it is critical to pursue detailed analyses of stratigraphic variations of the lava accretion styles. However, it has been challenging to directly observe the sequence of in situ upper oceanic crust because of its rare exposure [e.g., Francheteau et al., 1992] as well as the difficulty in drilling through the intact upper crust [e.g., Teagle et al., 2006]. On the other hand, ophiolites provide superb exposures on which we can reconstruct detailed three dimensional architecture of oceanic crust.

[5] The Oman ophiolite is the largest and best exposed ophiolite in the world that preserves the original structure of oceanic lithosphere formed at a fast-spread Neotethys ridge system [e.g.,Nicolas, 1989]. Based on the geological and petrological lines of evidence on the gabbros and sheeted dike complex, detailed structures of ridge segments have been identified in the northern Oman ophiolite [Adachi and Miyashita, 2003; Miyashita et al., 2003; Umino et al., 2003]. Wadi Fizh area is considered to be a second- or third-order discontinuity of a paleoridge segment, as is evidenced by the presence of dolerite dikes intruded into the plutonic complexes (blocks) consisting of the lower crustal layered gabbro, upper gabbro and wehrlitic intrusions of MORB-like affinities [Adachi and Miyashita, 2003]. From the segment center at Wadi ath Thuqbah toward the segment end at Wadi Fizh, the dike thickness increases and the number of dikes decreases [Umino et al., 2003]. Comparatively uniform and moderately evolved dikes occur in the segment center while those from the segment end show a larger variation from highly evolved to primitive compositions [Miyashita et al., 2003; Umino et al., 2003].

[6] The volcanic formations in the Oman ophiolite are divided into three units: V1 (Geotimes Unit), V2 (Alley Unit) and V3 (Salahi Unit) [Alabaster et al., 1980, 1982; Lippard et al., 1986; Ernewein et al., 1988; Umino et al., 1990]. Despite of different ideas on the genesis of magmatism, the stratigraphic division of V1 = Geotimes, V2 = Alley and V3 = Salahi Unit is broadly accepted. However, there is a debate on the lava stratigraphy in particular for the Lasail Unit. Alabaster and Pearce [1980], Alabaster et al. [1982] and Lippard et al. [1986] defined the Lasail Unit between the Geotimes and Alley Unit. Umino et al. [1990] and A'Shaikh et al. [2005] regarded the Lasail Unit as a part of the Geotimes Unit, while Ernewein et al. [1988] and Godard et al. [2003] regarded it as a subdivision of the V2 Unit. Detailed stratigraphic analyses are indispensable to settle this controversy, but only a part of the V1 extrusive rocks exposed in the Wadi Shaffan area has been studied yet [Einaudi et al., 2000, 2003].

[7] This paper presents the first thorough description of the whole V1 sequence in the Wadi Fizh area and detailed stratigraphic variations of eruptive products and lava geochemistry, and discusses the accretionary processes of the upper crust at a paleoridge segment boundary (Figure 1).

Figure 1.

(a) Simplified geological map of the Oman ophiolite after Lippard et al. [1986]. (b) Geological route map along the Wadi Fizh.

2. Geology

2.1. Geology and Description of Lithofacies

[8] The V1 Unit was generated in a mid-ocean ridge setting [Nicolas, 1989; Umino et al., 1990; Godard et al., 2003], or in a supra-subduction zone setting [Pearce, 1980; Alabaster et al., 1982; Lippard et al., 1986], whereas the V2 Unit was formed by subduction-related [Pearce, 1980; Alabaster et al., 1982; Lippard et al., 1986] or intraoceanic detachment magmatism [Boudier et al., 1988; Ernewein et al., 1988]. The collision event by which the Oman ophiolite was emplaced on the Arabian Continent occurred after the V2 magmatism. The V3 magmatism would have been originated during the 85 Ma collision event [Alabaster et al., 1982; Lippard et al., 1986; Ernewein et al., 1988; Umino et al., 1990].

[9] The extrusive sequence in Wadi Fizh area is more than 900 m in thickness, striking N-S to NWN-SES and dipping 25° to 40° east (Figure 1). The extrusive rocks gradually change into the underlying sheeted dike complex, through the 20 m thick transition zone with an increasing number of dikes 0.8–1.5 m in thickness downward. The sheeted dike complex trends N-S to NW-SE and inclines 20°–45° west, generally perpendicular to the extrusive rocks.

[10] Although the V2 lavas appear at the eastern end of the mapped area (Figure 1), the direct relationship between the V1 and V2 flows is not observed because of lack of exposures. A 0.3 m thick umber layer intercalated with the V2 pillow lavas trends N50°W and dips 20° east, which is concordant to the general structure of the underlying V1 extrusive rocks.

[11] Based on a series of observations on the modern-day submarine lava formation described below, we have identified four lithofacies in the V1 sequence; pillow, pahoehoe, lobate sheet and massive lava flows.

[12] Pillow flows comprise 31% of the V1 section. Pillow flows consist mainly of “bulbous” pillows (Figure 2b) and only locally of “elongate” pillows (Figure 2c). Characteristic features indicative of pillow budding are sometimes observed such as transverse spreading cracks, but none of corrugations on pillow surfaces are preserved due to weathering. However, flow directions of pillow lobes could be determined by the alignment of elongate, cylindrical pillow lobes, which sometimes bifurcate downslope [Auzende et al., 1996; Umino et al., 2002; Gregg and Smith, 2003].

Figure 2.

Representative lava morphology and lithofacies. (a) 0.4 m thick glossy metalliferous sedimentary layer at 410 mab dividing the LV1 and MV1 is traceable for as long as 3 km southward. (b) Bulbous pillow flows in the LV1 sequence (405 mab) just below the metalliferous sediment in 2a. A 0.2 m thick dike intruded in the LV1 penetrates through the metalliferous sediment into the overlying MV1. (c) Elongate pillows at 230 mab in the LV1 showing subparallel alignment, indicative of flowage on a slope. (d) Pahoehoe flows in the UV1 at 735 mab. Pahoehoe lobes are 1 × 1.5–3 m wide and 0.1–0.3 m thick have much smaller thickness/width ratios than pillow lobes. (e) Lobate sheet flows at 160 mab are 20 m in width and have hummocky, undulating tops. Small pillow lobes and breccias accompany the bottom of each flow lobe. (f) Thick lobate sheet flow underlain by pillow lava at 350 mab. Columnar joints develop in the basal part of the sheet flow. Flow boundaries commonly coincide with unit boundaries. (g) Thick pillow flows in the LV1b. Sharp change in color corresponds to the flow unit boundary within the pillow flows at 330 mab. Colors inherit from the difference in secondary mineralogy response to the different bulk compositions. The brownish pillows are evolved whereas the grayish pillows are less evolved (see text). (h) Pillowed ridge with a fissure vent at 854 mab. The fissure is filled with numerous dikes intruded into elongate pillows. Note the overturned and elongated pillow lobes around the fissure.

Figure 2.

(continued)

[13] Pahoehoe flows are abundant in the UV1. Individual pahoehoe flows have a minimum thickness of 30 m and attains to 20% of the entire V1 sequence. Subaqueous pahoehoe lobes have been mistaken as pillow lobes but the former has much smaller ratios of thickness to width than the latter [Walker, 1992; Umino and Nakano, 2007]. Pahoehoe lobes are approximately 1–3 × 2–5 m in extension and 0.2–1.0 m thick. They are characterized by their bun-like or amoeboid shapes and smooth surfaces without corrugations and are intimately associated with lobate sheet flows like their subaerial equivalents accompanying pahoehoe sheet (lava rise) (Figure 2d) [Hon et al., 1994; Umino et al., 2002]. Some pahoehoe lobes are interconnected upstream to a larger and thicker lobate sheet, to which coalesced adjacent pahoehoe lobes merge [Gregg and Chadwick, 1996]. Hollow lobes are present in pahoehoe flows, which are formed by drainage of molten lava within partially solidified lobes [Batiza and White, 2000; Umino et al., 2000].

[14] Lobate sheet flows occupy 41% of the V1 sequence and are the dominant lithofacies in this section (Figure 2e). A lobate sheet has a gently curved upper surface with well-developed columnar joints perpendicular to the surface. A lobate sheet flow is a compound flow consisting of numerous lobes about 2 m in thickness and a lateral extension of 20 m or more, which is piled up to form a domal structure as a whole (Figure 2f). Some lobate sheet flows include breccias 0.1–0.2 m thick and are underlain by lenses of pillows less than 0.5 m thick. Varioles less than 0.01 m in diameter are locally present in the chilled margin of a lobate sheet.

[15] Massive lava constitutes 7% of the V1 sequence, which is the least abundant among the four lithofacies. Unlike lobate sheet flows, massive lava has a flat surface with well-developed columnar joints perpendicular to the surface. A massive flow is 1.5–35 m thick and extends at least 100 m long.

[16] The 900 m thick sequential V1 unit is separated by a 0.4 m and a 0.8 m thick metalliferous sedimentary layers (umber) (Figure 2a), on which the V1 unit is divided into three subunits: the lower V1 (LV1), the middle V1 (MV1) and the upper V1 (UV1) (Figure 3). The uppermost UV1 with at least 200 m in thickness shows shallower dips of 25°–32° east than the LV1 and MV1 units dipping 30°–40° east. In the following sections, stratigraphic levels are denoted by the height in meters above the base (mab) of the V1 sequence resting on the sheeted dike complex (Table 1).

Figure 3.

Stratigraphic column of the V1 sequence along Wadi Fizh. Broken lines indicate unit boundaries.

Table 1. Locations, Mode of Occurrence and Phenocryst Assemblages of the Wadi Fizh Lava Flows and Dikesa
Sample NumberLatitudeLongitudeMode of OccurrenceMineral AssemblagesTexturemab
  • a

    Pl: plagioclase; Ol: olivine; Cpx: clinopyroxene; Opx: orthopyroxene.

Lower V1 Sequence
LV1a      
    07Fizh524.2911856.21450Massive lavasaphyricDoleritic15
    07Fizh624.2911856.21478Pillow lavasPl-Ol-CpxVariolitic45
    07Fizh724.2911656.21484Pillow lavasaphyricIntersertal56
    07Fizh824.2911656.21484Pillow lavasPl-Ol-CpxIntersertal56
    07Fizh924.2911756.21492Pillow lavasPl-CpxHyaloophitic69
    07Fizh1024.2911756.21493DikePl-CpxHyaloophitic75
    07Fizh1124.2911556.21495Pillow lavasPl-CpxVariolitic73
    07Fizh1224.2911456.21498Lobate sheet flowsaphyricHyaloophitic80
    07Fizh1324.2910956.21500Lobate sheet flowsPlHyaloophitic90
    07Fizh1424.2910856.21509Pillow lavasPlVariolitic99
    07Fizh1524.2910056.21514Pillow lavasPl-CpxVariolitic108
    07Fizh1624.2909256.21526Pillow lavasPl-Ol-CpxVariolitic123
    07Fizh1724.2908756.21532Lobate sheet flowsaphyricIntersertal129
    07Fizh1824.2907556.21533Lobate sheet flowsPl-CpxVariolitic131
    07Fizh1924.2906856.21535Lobate sheet flowsaphyricIntersertal143
    07Fizh2024.2905756.21534Lobate sheet flowsPl-CpxIntersertal142
    07Fizh2124.2904956.21542Pillow lavasaphyricIntersertal165
    09vFizh124.2904056.21567Lobate sheet flowsaphyricIntersertal183
    07Fizh2224.2903956.21584Lobate sheet flowsPl-Ol-CpxIntersertal213
    07Fizh2324.2905956.22000Pillow lavasaphyricVariolitic246
    09vFizh224.2904556.22006Lobate sheet flowsPl-Ol-CpxHyaloophitic248
    09vFizh324.2904656.22009Lobate sheet flowsPl-Ol-CpxHyaloophitic249
    07Fizh2424.2905456.22006Lobate sheet flowsPl-CpxIntersertal250
LV1b      
    07Fizh2624.2906156.22018Lobate sheet flowsPl-Ol-CpxIntersertal275
    07Fizh2724.2906056.22028Pillow lavasaphyricHyaloophitic290
    07Fizh2824.2906056.22028Pillow lavasPl-CpxHyaloophitic290
    07Fizh2924.2906156.22033Massive lavasaphyricIntersertal293
    07Fizh3024.2908356.22041Pillow lavasaphyricHyaloophitic325
    09vFizh424.2909256.22030Pillow lavasPl-CpxVariolitic323
    09vFizh524.2909456.22038Pillow lavasCpxHyaloophitic335
    07Fizh3124.2909556.22042Pillow lavasPl-CpxHyaloophitic335
    07Fizh3224.2910456.22051Pillow lavasaphyricHyaloophitic343
    07Fizh3324.2910956.22055Lobate sheet flowsPl-CpxIntersertal358
    07Fizh3424.2911656.22055Lobate sheet flowsaphyricIntersertal370
    07Fizh3524.2912356.22068Lobate sheet flowsaphyricHyaloophitic378
    07Fizh3624.2913056.22074Pillow lavasaphyricVariolitic409
 
Middle V1 Sequence
07Fizh3724.2913256.22076Lobate sheet flowsaphyricHyaloophitic414
07Fizh3824.2912356.22081Pillow lavasaphyricHyaloophitic416
07Fizh3924.2912356.22109Lobate sheet flowsaphyricIntersertal430
07Fizh4024.2912256.22096Lobate sheet flowsPl-CpxHyaloophitic449
07Fizh4124.2912456.22100Pillow lavasaphyricHyaloophitic467
07Fizh4324.2914056.22131Pahoehoe flowsaphyricHyaloophitic504
07Fizh4424.2914556.22143Pahoehoe flowsPl-CpxHyaloophitic516
07Fizh4524.2914956.22156Lobate sheet flowsPl-CpxHyaloophitic532
07Fizh4624.2915156.22162Pillow lavasPl-CpxVariolitic553
07Fizh4724.2915356.22171Pahoehoe flowsPl-CpxHyaloophitic567
07Fizh4824.2915556.22190Pillow lavasPl-CpxHyaloophitic606
07Fizh4924.2915556.22203Pillow lavasaphyricHyaloophitic623
07Fizh5024.2917056.22207Lobate sheet flowsPl-CpxIntersertal645
07Fizh5124.2917456.22219Pillow lavasPl-CpxHyaloophitic662
 
Upper V1 Sequence
07Fizh5224.2917456.22227Lobate sheet flowsPl-Ol(-Opx)Intersertal679
07Fizh5324.2917656.22233Pahoehoe flowsOlHyaloophitic695
07Fizh5424.2922756.22191Pillow lavasPlHyaloophitic693
07Fizh5524.2923356.22196Pillow lavasPl-CpxVariolitic712
07Fizh5624.2923856.22206Lobate sheet flowsPl-CpxHyaloophitic742
08Fizh224.2918556.22269Pahoehoe flowsPlHyaloophitic731
09vFizh624.2918256.22245Pahoehoe flowsPl-Ol-CpxHyaloophitic736
08Fizh324.2918556.22277Pahoehoe flowsPl-Ol-CpxVariolitic742
09vFizh724.2918456.22283Pahoehoe flowsPl-Ol-CpxHyaloophitic753
08Fizh524.2920256.22306HyaloclastitesPlHyaloophitic782
08Fizh624.2920456.22313Lobate sheet flowsPl-CpxHyaloophitic815
08Fizh724.2920656.22314Lobate sheet flowsPl-Ol-Cpx(-Opx)Hyaloophitic820
08Fizh824.2921656.22332Pillow lavasPl-Ol-CpxHyaloophitic854
09vFizh924.2924656.22389calcite vein hostaphyricIntersertal 
 
V2 Lavas
08Fizh1124.2931856.23074Massive lavasOl-CpxHyaloophitic 
08Fizh1324.2932156.23072Massive lavasPl-Ol-CpxHyaloophitic 
08Fizh1424.2931756.23068Massive lavasaphyricHyaloophitic 
08Fizh1524.2931656.23066Massive lavasaphyricHyaloophitic 
08Fizh1624.2931456.23055Pillow lavasPlVariolitic 
 
Dikes
07Fizh4224.2913256.22116Massive lavasPl-CpxIntersertal486
08Fizh124.2918356.22259SillPl-Ol-Cpx(-Opx)Hyaloophitic731
08Fizh1724.2913056.22074DikeOlHyaloophitic410
 
Boninite Dike
08Fizh424.2920056.22302DikePl-Ol-Cpx(-Opx)Hyaloophitic770

2.2. Lower V1 Sequence (LV1)

[17] The LV1 is 410 m thick and consists mainly of pillow and lobate sheet flows with a subordinate amount of massive lava. Massive lava occurs in the lowermost V1 sequence directly overlying the sheeted dike complex. Pillow flows comprise 54% in the LV1. Most pillow lobes are bulbous pillows about 1.0 m across, while elongate pillows are locally present. Distinction of individual lithofacies is relatively easy at outcrops as shown in Figure 2f and is usually identified by the occurrence of different lava flow type and/or structure. At 330 mab, a sharp color change in thick pillow flows marks a boundary of superposed two flow units with a thickness of 45 and 25 m, respectively, which shows different flow directions as indicated by the difference in elongated pillow alignment (Figure 2g). At 230 mab, a 20 m thick elongate pillow flow is underlain by a 40 m thick lobate sheet flow associated with bulbous pillows (Figure 2c). At 270 mab, a jasper layer is interbedded between the lower LV1 (LV1a) and the upper LV1 (LV1b) (Figure 3). The jasper forms a thin wavy layer with a variable thickness about 0.05–0.2 m and a lateral extension about 50 m, filling interpillow spaces and radial cracks of the underlying pillow lava. The jasper layer is overlain by lobate sheet flows, which are in turn overlain by a 70 m thick pillow flow unit interlayering 5 m thick massive lavas. The top of the LV1 is conformably overlain by a 0.4 m thick metalliferous sedimentary layer extending approximately 3 km to the south (Figure 2a) where no erosion is observed along the boundary. At 410 mab, a 0.2 m thick dike trending NW-SE and dipping 60° west intrudes into the LV1b and the MV1 through the metalliferous sedimentary layer in between (Figure 2c).

2.3. Middle V1 Sequence (MV1)

[18] The MV1 has a thickness of 260 m and consists of bulbous pillows, pahoehoe and lobate sheet flows. The basal part of MV1 is shown in Figures 2a and 2b, where thin hyaloclastite conformably overlies an umber bed. Pillow and pahoehoe flows constitute 30% and 25% of the MV1, respectively. The thickness of individual flows is 10–50 m, which is similar to that of the LV1. The lower part of the MV1 is dominated by lobate sheet and pahoehoe flows, while the upper part is dominantly composed of pillow flows. A dike >5 m in thickness intrudes into pahoehoe flows at 480 mab. The MV1 is terminated by the appearance of a 0.8 m thick metalliferous sedimentary layer extending about 4 km to the south.

2.4. Upper V1 Sequence (UV1)

[19] The UV1 is more than 200 m thick in total, and consists of 46% pahoehoe flows and 30% variolitic lobate sheet flows. While the upper sequence is poorly exposed, the UV1 sequence begins with thick pahoehoe flows resting on a metalliferous sedimentary layer. Between 730 and 880 mab, several 1.0 m thick basaltic dikes intrude into pahoehoe and lobate sheet flows. These dikes strike WNW and dip 22°–78° south. A 0.75 m thick boninite dike trending 72° west and dipping 74° south is present in the UV1 sequence at 770 mab (Figures 1 and 3). At 854 mab, near the upper end of continuous exposures of the extrusive section, a zone of elongate pillows 6–7 m in width and >3 m in height is intruded by an intense swarm of thin dikes (0.1–0.25 m thick), interbedded with the host pahoehoe lava which strikes N50°E and dips 28° east (Figure 2h). The dikes concentrate within the 2 m wide central portion of the pillowed zone that extends laterally a few tens of meters along N70°W and dips 70° south, almost perpendicular to the surrounding pahoehoe lava structure. The dikes are intruded into and sided by elongate pillows 0.6 m in diameter which are concordant to the attitude of dikes as shown in Figure 2h. These dikes and pillows are interpreted as a fissure vent opened through a pillowed ridge. Lava spilt out of the vent draped on the steep flanks of the ridge as elongate pillows bifurcating downslope.

3. Petrography

[20] Based on abundance and assemblage of phenocrysts, the V1 volcanic rocks are divided into aphyric, plagioclase (Pl)-phyric, plagioclase-olivine-clinopyroxene (Pl-Ol-Cpx)-phyric and Pl-Cpx-phyric types. Hyaloophitic and intersertal textures are well observed in the groundmass and a variolitic texture commonly develops near the surface of pillows (Figure 4a). Lava samples were undergone intensive low-temperature alteration. Most plagioclases are altered to albite, saussurite or calcite. Chlorite and clay minerals replace olivine phenocrysts. Only clinopyroxene, Fe-Ti oxide and chrome-spinel remain as primary minerals. Clinopyroxene occurs as discrete crystals or glomerocrysts. Summary of petrography is shown inTable 1.

Figure 4.

Microphotographs of flows and dikes. (a) Altered plagioclase and clinopyroxene microphenocrysts in the variolitic groundmass (LV1). (b) Aphyric lava with doleritic texture (LV1). (c) Plagioclase and clinopyroxene phenocrysts in a hyaloophitic matrix (MV1). (d) Olivine, plagioclase and clinopyroxene phenocrysts in a hyaloophitic matrix (UV1). (e) Boninite dike at 770 mab containing with clinopyroxene phenocrysts and clinopyroxene and orthopyroxene microphenocrysts. Ol: olivine; Pl: plagioclase; Cpx: clinopyroxene; Ves: vesicle.

[21] The LV1 consists mainly of aphyric and sparsely Pl-Cpx-phyric basalt. A few Pl-phyric and Pl-Ol-Cpx-phyric lavas are also present. Aphyric lava sparsely contains plagioclase and clinopyroxene microphenocrysts (Figure 4b) embedded in a doleritic groundmass. Clinopyroxene phenocrysts and microphenocrysts show either a sector zoning or a normal zoning. Vesicles are filled with quartz, calcite and clay minerals. Pl-Cpx basalt dominates in the MV1 interbedded with a few aphyric flows (Figure 4c). Clinopyroxene shows either a sector zoning or a normal zoning as those in the LV1. In the MV1 lava, clinopyroxene occurs both as a phenocryst and a microphenocryst. Amygdules filled with calcite, zeolite and clay minerals are common. The UV1 lava consists of Pl-Ol-Cpx-phyric and a few amounts of Pl- and Pl-Cpx-phyric basalts. Pl-Ol-Cpx-phyric lavas contain either discrete or glomerocrystic olivine (Figure 4d). Clinopyroxenes have no distinct zoning. Vesicles are filled with calcite and clay minerals. A 0.2 m thick Ol-phyric basalt dike at 410 mab has chrome spinel (Cr/(Cr + Al) = 0.76–0.81) as microphenocrysts and inclusions in altered olivine. A 1.0 m thick dike at 730 mab has orthopyroxene as a phenocryst in addition to plagioclase, olivine and clinopyroxene. A 0.75 m thick boninite dike at 770 mab is Ol-Opx-Cpx-phyric with olivine and orthopyroxene phenocrysts completely replaced by clay minerals (Figure 4e).

4. Bulk Chemistry

[22] Whole-rock major elements, Ni, Y, Zr, Cr and V contents of 55 samples were analyzed by X-ray fluorescence (XRF) (RIX3000, Rigaku Denki) at Niigata University. The analytical method is described inTakahashi and Shuto [1997]. Other trace and rare earth elements of 41 samples were analyzed by inductively coupled plasma mass spectrometry (ICP-MS) (Agilent 7500a ICP-MS) afterRoser et al. [2000]at Niigata University. When acid-resistant minerals such as zircon and titanite are present, the analyses of ICP-MS by acid digestion may give lower Zr, Hf, Th and U contents. However, rare earth element (REE) concentrations by acid digestion show similar values to those by alkali fusion [Neo et al., 2009]. Therefore, we used Zr values by XRF method and also used other trace element contents by ICP-MS method for this study. The results of XRF and ICP-MS analyses are given inTables 2 and 3.

Table 2. Bulk-Rock Major and Trace Element Compositions Analyzed by XRF
 SiO2 (wt%)TiO2Al2O3Fe2O3MnOMgOCaONa2OK2OP2O5LOITotalNi (ppm)YZrCrV
Lower V1 Sequence
LV1a                 
    07fizh548.631.7914.5810.570.316.018.504.700.630.134.40100.232.341.2110.851324
    07fizh652.461.6115.529.740.165.396.424.750.300.173.75100.321.739.8112.839321
    07fizh 749.301.3914.119.670.186.098.785.420.220.135.02100.432.231.687.341308
    07fizh849.281.6913.9410.200.196.108.415.270.240.164.78100.332.542.6118.366346
    07fizh 950.111.4513.749.740.165.629.045.000.810.144.41100.131.332.593.238296
    07fizh1248.641.7314.6211.170.236.548.804.290.300.123.73100.025.038.4107.231348
    07fizh1349.211.9613.6111.850.226.856.695.010.270.194.38100.324.249.1141.217379
    07fizh1552.191.4114.589.770.165.148.404.690.170.153.37100.227.333.895.334293
    07fizh1750.581.4314.119.500.245.737.765.350.810.124.62100.222.032.186.317342
    07fizh1949.681.6413.3210.590.215.128.925.450.420.164.75100.314.940.5114.819329
    07fizh2050.241.8014.509.310.264.618.535.740.130.144.97100.138.938.5106.730350
    09vfz155.191.7714.6110.140.122.447.876.770.050.180.84100.04.840.3104.810345
    07fizh2251.741.4015.119.840.187.644.495.490.370.123.77100.128.337.7105.834275
    07fizh2358.871.4112.809.110.177.201.814.920.070.163.63100.212.839.1118.311227
    09vfz251.730.7515.608.630.165.976.355.520.640.064.79100.241.418.647.045225
    09vfz351.670.6515.617.590.184.858.895.370.390.064.97100.431.617.142.144214
    07fizh2452.770.7115.148.360.166.336.055.570.820.064.28100.331.617.344.739222
LV1b                 
    07fizh2652.652.0013.9011.790.245.864.145.880.170.183.32100.122.944.4127.113331
    07fizh2752.741.9214.0312.120.195.334.865.860.060.192.81100.16.245.2119.019390
    07fizh2853.571.8414.2910.980.145.434.166.540.080.182.90100.111.645.4127.515370
    07fizh2953.161.9713.9511.640.175.245.165.440.660.182.50100.214.341.5118.821377
    09vfz452.450.7316.008.340.164.907.535.190.420.054.37100.125.319.046.936250
    09vfz551.811.9314.1113.290.224.615.035.620.720.192.5399.911.846.3129.114395
    07fizh3151.101.9213.7113.040.246.784.735.180.340.182.88100.014.345.1122.118401
    07fizh3352.051.9514.5211.670.195.804.934.921.090.212.72100.016.747.6152.619369
    07fizh3450.532.1514.8311.760.606.624.325.240.280.243.59100.011.351.8155.418374
    07fizh3553.481.9814.0910.050.154.415.926.690.070.173.06100.216.846.3134.619345
    07fizh3652.141.7714.5910.260.154.187.474.540.100.184.84100.214.246.2137.717402
 
Middle V1 Sequence
07fizh3750.720.5913.377.750.397.168.175.460.330.056.52100.748.616.536.263257
07fizh3947.950.6415.668.770.158.458.684.090.180.045.75100.339.718.143.070257
07fizh4051.860.6014.698.970.177.397.624.610.500.053.61100.334.416.635.348245
07fizh4354.061.5814.249.900.164.965.186.230.640.163.00100.216.137.1112.217321
07fizh4549.320.5615.927.120.195.6510.914.490.310.055.80100.364.315.039.5160209
07fizh4649.260.6415.548.220.176.648.674.520.480.066.23100.467.217.542.1136226
07fizh4750.901.2415.629.940.154.897.254.070.490.125.66100.224.331.688.816282
07fizh4847.300.6515.948.710.198.438.614.080.210.056.25100.357.717.444.5146235
07fizh4957.471.9713.6710.200.134.363.436.000.540.192.10100.12.244.5131.45279
07fizh5051.752.0413.9511.190.206.804.925.180.840.163.04100.029.044.8126.118342
07fizh5154.511.8314.1710.580.215.014.665.790.570.192.59100.215.641.6118.415353
 
Upper V1 Sequence
07fizh5354.640.5316.846.000.063.528.245.310.210.044.86100.159.413.236.5199243
07fizh5548.580.5916.528.620.156.709.124.340.130.055.49100.132.119.234.970296
08fizh352.300.6714.708.930.157.046.834.640.270.064.63100.146.918.242.034257
08fizh552.920.7214.899.080.156.007.165.510.190.073.45100.238.319.339.042312
08fizh652.970.5914.187.631.414.209.975.290.230.083.59100.247.820.546.328241
08fizh751.350.7016.128.540.345.878.094.010.850.064.27100.138.419.148.533262
08fizh853.560.6114.097.480.134.558.284.890.780.095.92100.432.118.042.932261
09vfz749.460.7516.119.370.186.387.864.680.260.065.11100.537.020.250.131281
09vfz951.360.7215.487.840.155.517.805.120.640.065.62100.271.319.442.773231
 
V2 Lavas
08fizh1154.860.6813.658.770.158.283.515.110.060.055.19100.330.619.538.643268
08fizh1356.320.7613.638.660.128.022.404.980.040.055.29100.324.022.442.336301
08fizh1461.151.0012.999.710.124.422.445.980.070.082.08100.17.224.853.58393
 
Dikes
07fizh4251.300.4614.748.080.179.646.894.070.760.034.02100.282.714.324.3298249
08fizh150.690.4915.957.870.126.349.194.640.420.034.48100.243.616.726.5181280
08fizh1746.520.4812.807.540.426.5612.425.060.110.038.89100.8184.114.525.8628242
 
Boninite Dike
08fizh454.250.2311.739.070.1412.418.860.530.230.012.6599.9270.37.89.0972231
Table 3. Bulk-Rock Trace Element Compositions Analyzed by ICP-MSa
 ScCoGaRbSrNbCsBaLaCePrNdSmEuGdTbDyHoErTmYbLuHfTaThU
  • a

    Duplicate analyses of USGS W-2 (diabase) and the recommended values byEggins et al. [1997] are shown along with relative deviations between the measured and recommended values.

W-2
Measured (ppm) Mean (n = 6)36.844.4617.4720.662017.760.8517110.7723.193.0513.033.281.083.670.623.760.802.260.342.030.312.380.462.240.51
Recommended values (ppm)36.246.0017.4020.101927.760.9217110.5923.083.0312.953.311.093.690.623.790.802.260.332.030.302.300.482.210.50
Relative deviation (%)1.5−3.450.382.724.40.05−8.330.091.690.470.710.62−0.95−1.65−0.49−0.07−0.880.14−0.133.220.102.023.55−4.091.302.24
 
Lower V1 Sequence (ppm)
LV1a                          
    07fizh534.239.918.17.931972.990.12965.1514.632.3012.714.261.545.821.026.641.524.340.683.990.613.120.210.100.10
    07fizh630.533.616.57.401602.790.14264.9514.252.3812.514.141.425.531.006.451.424.240.633.890.583.010.200.290.12
    07fizh 7--------------------------
    07fizh837.740.116.62.851983.000.09464.9714.492.4512.564.261.535.641.056.761.554.470.674.100.613.210.210.270.15
    07fizh 9--------------------------
    07fizh1238.641.818.13.242212.660.041504.5613.312.1512.084.131.505.250.986.411.494.010.643.990.592.910.190.150.15
    07fizh1336.235.920.42.931754.160.03536.4917.832.9815.315.061.776.851.227.811.764.900.774.770.723.680.270.390.14
    07fizh1535.234.814.12.491752.650.06154.0812.182.0711.203.501.304.940.915.771.313.680.563.510.542.670.170.260.20
    07fizh1734.339.614.110.162232.120.291263.8311.361.9110.453.331.224.690.845.431.233.430.563.510.522.420.130.070.06
    07fizh19--------------------------
    07fizh2031.552.621.11.421372.640.11144.3913.062.2211.373.891.435.300.996.341.434.240.634.020.603.000.180.280.12
    09vfz128.416.715.20.78652.270.0564.2712.442.2011.123.551.365.060.946.021.333.940.613.590.542.710.150.230.14
    07fizh2233.235.620.24.562012.530.051174.3212.972.1611.643.951.335.240.946.011.374.010.613.860.572.920.160.100.10
    07fizh2323.227.417.70.62752.700.01114.8714.612.3112.114.011.415.420.966.361.404.140.643.990.603.130.180.320.09
    09vfz228.530.211.711.061750.600.07561.624.760.864.391.580.612.200.422.660.601.830.271.620.251.220.040.140.12
    09vfz325.925.211.87.51790.530.0661.564.030.754.121.420.581.910.372.380.541.580.221.450.231.070.040.140.08
    07fizh2432.133.414.621.331390.640.16141.774.970.894.711.750.682.410.452.930.661.910.301.800.271.310.050.210.10
LV1b                          
    07fizh2633.133.118.82.071163.480.03165.7416.972.9914.545.031.696.441.177.491.765.030.804.880.743.700.220.100.09
    07fizh27--------------------------
    07fizh2830.632.918.20.811112.930.0395.3815.222.6013.364.461.556.021.106.861.564.650.704.290.653.450.200.290.11
    07fizh29--------------------------
    09vfz425.627.112.67.081860.610.07681.794.840.834.681.600.662.200.422.690.611.870.261.660.241.170.040.140.05
    09vfz528.333.117.315.69892.950.12415.1615.112.5813.374.241.515.921.076.831.564.910.724.230.663.220.200.310.08
    07fizh3132.539.219.67.911323.260.08345.7217.052.8715.084.831.676.461.207.441.705.060.784.680.713.630.230.360.10
    07fizh3334.335.520.917.521843.600.041225.9818.903.1717.045.491.837.211.298.471.785.400.805.040.754.070.230.120.11
    07fizh3436.935.327.33.861174.450.041026.3618.803.2716.885.431.927.281.378.781.955.590.885.500.834.460.290.460.18
    07fizh3535.534.518.00.75563.550.0395.4416.292.7514.134.701.656.301.157.371.664.780.734.660.703.580.230.350.28
    07fizh3632.038.220.12.10913.510.0685.2315.822.6614.474.391.606.111.097.471.634.690.734.500.683.630.240.100.10
 
Middle V1 Sequence (ppm)
07fizh3735.738.57.68.671250.530.10271.744.030.754.121.400.552.140.382.610.591.740.281.790.261.010.030.060.06
07fizh3945.141.617.31.563510.610.04431.384.070.723.941.490.592.200.442.970.671.870.311.910.301.080.040.130.09
07fizh4038.436.012.56.992140.470.15381.243.550.603.521.330.522.000.382.460.591.710.261.680.250.940.030.130.13
07fizh4328.434.024.415.23822.420.13274.4613.222.2211.643.921.495.200.935.911.323.870.593.640.543.040.160.320.14
07fizh4536.334.715.55.273530.490.19371.453.950.603.661.330.622.020.392.510.561.520.251.520.220.990.030.050.05
07fizh4634.739.614.18.362810.540.10321.764.770.834.301.610.622.310.412.740.631.790.291.710.261.140.040.140.04
07fizh4728.333.318.76.401221.960.11243.8511.021.849.413.191.204.390.815.011.133.230.493.010.442.440.130.270.05
07fizh4836.739.415.93.253750.580.10321.534.670.774.291.630.632.180.402.760.601.680.261.660.241.140.040.000.00
07fizh4928.521.914.94.511103.240.06845.8216.852.7613.934.621.615.981.097.081.604.640.714.350.653.460.230.350.15
07fizh5036.046.820.520.181923.480.32955.4316.162.7314.154.771.596.231.157.311.644.770.724.480.663.530.240.330.18
07fizh5131.129.914.614.961463.350.20366.5217.302.8514.134.631.606.191.096.931.534.480.664.130.613.260.220.320.16
 
Upper V1 Sequence (ppm)
07fizh5332.417.015.12.321400.520.05141.433.630.613.271.090.471.550.291.900.431.300.211.350.210.950.040.120.12
07fizh5536.935.217.61.472090.590.01111.884.820.774.911.460.592.190.423.030.671.930.301.960.291.100.040.040.04
08fizh323.823.510.12.66890.430.05211.133.350.543.061.090.421.590.281.880.441.290.191.240.190.810.030.100.06
08fizh539.235.023.71.481140.630.02342.085.271.005.041.890.722.600.483.190.732.080.332.050.311.230.040.150.09
08fizh632.933.720.73.127260.580.082052.124.880.934.861.610.622.350.453.040.692.100.322.050.321.080.040.140.18
08fizh737.733.015.422.043640.660.54951.875.240.894.841.710.632.360.462.970.672.080.311.980.311.200.040.160.10
08fizh833.324.312.320.283570.610.172751.765.100.884.061.570.572.200.412.740.611.760.281.710.271.090.040.140.15
09vfz733.931.410.35.512630.630.04201.815.100.834.551.600.662.360.442.830.642.020.311.830.281.190.040.250.10
09vfz928.525.412.16.02910.540.13781.594.410.734.191.410.612.080.402.540.621.770.281.590.251.130.030.130.08
 
V2 Lavas (ppm)
08fizh1133.933.914.90.52450.690.0141.434.520.784.281.670.622.470.483.220.752.150.352.140.331.260.050.190.11
08fizh1334.032.014.40.48520.790.0151.614.700.844.651.860.682.610.533.550.812.500.372.450.381.300.050.130.10
08fizh1416.314.77.80.54400.510.0241.113.220.573.031.140.421.590.322.170.511.520.241.500.230.880.040.090.07
 
Dikes (ppm)
07fizh4241.937.112.010.111540.340.32850.692.130.402.320.920.401.490.292.140.481.420.231.480.220.710.030.030.03
08fizh143.832.416.27.393200.390.06611.092.610.472.611.050.451.690.342.370.551.650.261.680.260.710.020.060.08
08fizh1741.339.516.20.941130.430.02390.802.440.452.510.990.381.630.332.290.531.630.251.580.250.790.020.080.28
 
Boninite Dike (ppm)
08fizh450.951.710.45.23440.610.08310.691.550.221.050.410.160.680.161.070.250.840.140.990.160.300.050.200.12

[23] As mentioned above, the analyzed samples have suffered pervasive low-temperature alteration which may have affected the primary geochemical characteristics. We will first assess the effects of low-temperature alteration before discussing the primary geochemical signatures. Loss on ignition (L.O.I.) of analyzed samples ranges from 2.0 to 6.5 wt% and shows a positive correlation with CaO contents, suggesting an enrichment of CaO during alteration (Figure 5). This is consistent with abundant calcite as veins, and vesicle fillings and replacing olivine crystals [Alt and Honnorez, 1984]. On the other hand, Na2O contents show a weak negative correlation with L.O.I. However, the high Na2O in low L.O.I. samples does not always represent the primary nature of the sample, because plagioclase is totally altered to albite. Therefore, the high Na2O contents seem to have increased during the secondary processes. MgO contents do not correlate with L.O.I. but they might have been modified due to the formation of chlorite replacing olivine and filling vesicles [Humphris and Thompson, 1978a; Honnorez, 1981]. Therefore, we used immobile incompatible elements such as TiO2 and P2O5 for the following petrological consideration. Vesicles are commonly filled with quartz so that SiO2 may have increased in such samples.

Figure 5.

Bulk MgO, CaO and Na2O versus L.O.I (wt%). Major element compositions are recalculated on an anhydrous basis.

[24] Large-ion lithophile (LIL) elements such as Rb, Ba and Li, and Sr are generally highly mobile during ocean floor weathering and low-temperature hydrothermal alteration [Thompson, 1973; Humphris and Thompson, 1978b; Staudigel and Hart, 1983; Seyfried et al., 1984, 1998; Alt and Honnorez, 1984]. In fact these elements are highly scattered, suggesting an intensive modification of the initial concentration during the secondary processes. On the other hand, V, Y, Zr and Cr are generally immobile against such alteration processes [Thompson, 1973; Hart et al., 1974; Humphris and Thompson, 1978b; Pearce and Norry, 1979]. Tight positive correlations such as TiO2-Zr and Y-Zr support this view. In the following sections, we discuss the petrological and geochemical characteristics of the extrusive rocks on the basis of these immobile elements.

[25] Zr contents of the V1 samples range from 24 to 153 ppm, indicating that these samples span the range from the most primitive to most evolved MORBs [Basaltic Volcanism Study Project, 1981]. Samples from the LV1 and MV1 range in Zr from 42 to 153 ppm, and from 36 to 132 ppm, respectively, whereas the UV1 samples concentrate between 27 and 50 ppm. Thus, the LV1 is generally evolved and the UV1 is relatively primitive among the V1 samples. The V1 samples form two clusters on the Zr-TiO2and Zr-P2O5 variation diagrams (Figure 6), which subdivide the LV1 samples into high- and low-TiO2group. The high-TiO2group is characterized by higher incompatible element contents than the low-TiO2group. Although both groups are observed in the LV1 and MV1, the high-TiO2group dominates in the LV1 while the low-TiO2group dominates in the MV1. All the UV1 samples belong to the low-TiO2 group.

Figure 6.

TiO2, P2O5, Ni, Cr, Y, V, Zr/Y and (Nd/Yb)n ratios plotted against Zr. TiO2 and P2O5 contents are recalculated on an anhydrous basis. High Ni and Cr samples are omitted. Open and shaded fields are lava compositions by the previous studies; V1, Geotimes and sheeted dikes, and V2 field: Ernewein et al. [1988], Beurrier et al. [1989], Einaudi et al. [2003], Godard et al. [2003, 2006] Miyashita et al. [2003], and A'Shaikh et al. [2005]; Lasail and Alley field: Alabaster et al. [1982] and Lippard et al. [1986]. The arrow in the Zr/Y plot represents the fractionation trend of olivine, clinopyroxene and plagioclase up to 10, 30 and 60, respectively, and partition coefficients are after Fujimaki et al. [1984].

[26] These high- and low-TiO2groups show different trends in the Zr-Zr/Y variation diagram (Figure 6). The high-TiO2group has a limited Zr/Y ratio of 2.6–3.2, which slightly increases with increasing Zr contents. This suggests that the high-TiO2 group of the LV1 and MV1 was formed by the process of a simple fractional crystallization from the same parent magma [Fujimaki et al., 1984]. The low-TiO2group exhibits a broad range in Zr/Y from 1.6 to 2.6 that increases with increasing Zr contents, suggesting more diverse sources than the high-TiO2 group.

[27] REE concentrations of the V1 sequence exhibit similar characters to high field strength (HFS) elements. The high-TiO2group of the LV1 has Yb contents from 3.5 to 5.5 ppm, (La/Yb)n ratios from 0.8 to 1.0, and shows flat REE patterns with a slight light-REE (LREE) depletion (Figure 7a). On the contrary, the LV1 low-TiO2group ranges in Yb contents from 1.5 to 1.8 ppm and in (La/Yb)n ratios from 0.7 to 0.8 with slightly LREE-depleted REE patterns. Yb contents of the MV1 also show two distinct groups, which the high-TiO2group ranges from 3.0 to 4.5 ppm and the low-TiO2 group ranges from 1.5 to 1.9 ppm (Figure 7b). The MV1 high-TiO2group shows flat REE patterns ((La/Yb)n = 0.9–1.1) similar to those of the LV1. On the other hand, the MV1 low-TiO2group shows LREE-depleted patterns ((La/Yb)n = 0.5–0.7) (Figure 7b). The UV1 samples overlap the MV1 low-TiO2 group and have 1.4–2.1 ppm Yb and 0.5–0.8 (La/Yb)n ratios (Figure 7c).

Figure 7.

Chondrite-normalized REE patterns for (a) LV1, (b) MV1, (c) UV1, and (d) V2 sequence. The chondrite value is afterSun and McDonough [1989]. Shaded areas in Figures 7b–7d are the ranges of the LV1 lavas in Figure 7a.

[28] The V2 sequence has similar bulk compositions to the V1 low-TiO2group. The V2 samples show low Zr contents ranging from 39 to 54 ppm, which overlap the range of the low-TiO2 group. However, TiO2, Y and V contents of the V2 samples tend to be higher than the V1 low-TiO2 group at a given Zr content (Figure 6). Also, Zr/Y ratios of the V2 samples tend to be lower than the latter at a given Zr content (Figure 6). Furthermore, (Nd/Yb)n ratios of the V2 samples from 0.69 to 0.74 are lower than 0.75–1.25 for the V1 low-TiO2 group, leading to a crossover of the REE patterns of the V1 and V2 samples (Figures 7d). Dikes intruding into the V1 sequence at 410 and 730 mab have similar compositions to each other and lower (Nd/Yb)n ratios (= 0.56–0.58) than any flows. A boninite dike in the UV1 at 770 mab is highly depleted in incompatible elements but has a slightly LREE-enriched pattern, which resembles to a typical U-shaped pattern of boninites from the Oman and other ophiolites [Ishikawa et al., 2002; Godard et al., 2003].

5. Mineral Chemistry

[29] Clinopyroxene phenocrysts and microphenocrysts in 37 samples were analyzed by EPMA (JEOL JXA-8600SX) at Niigata University using an accelerating voltage of 15 kV and a beam current of 1.3 × 10−8A. Counting time was 30 s on the peak and 20 s on the background. Correction procedures followed the ZAF-Oxide method. Representative clinopyroxene analyses are given inTable 4. Because preferential partitioning of TiO2, Al2O3, Na2O and REEs in [100] sector compared to other sectors is enhanced for clinopyroxene crystals in submarine lava due to rapid cooling, cares must be taken when comparing the variation of these elements in clinopyroxene [Dowry, 1976; Coish and Taylor, 1979; Gamble and Taylor, 1980; Lofgren et al., 2006]. We used core compositions of [001] sector only in order to exclude the compositional differences arose from this kinetic effect on the trace element partitioning (Figure 8).

Table 4. Representative Clinopyroxene Analyses by EPMA
Sample NameSiO2 (wt%)TiO2Al2O3Cr2O3FeOMnOMgOCaONa2OTotalCaMgFeMg#
Lower V1 Sequence
LV1a              
    07fizh549.711.103.130.0511.100.2514.2419.760.3799.714141.117.90.70
    07fizh1751.570.392.040.006.580.1717.3520.730.2299.0541.448.310.30.82
    07fizh2252.090.182.190.134.530.1918.3922.070.1899.9543.1506.90.88
    07fizh2452.110.152.150.035.140.1718.3221.540.1899.7942.249.97.90.86
LV1b              
    07fizh2651.490.622.160.169.260.2015.8219.850.2999.8540.444.814.70.75
    07fizh3350.460.912.900.089.250.3115.2620.530.32100.0241.943.314.70.75
 
Middle V1 Sequence
07fizh4053.070.092.760.742.800.0617.9421.680.1399.2744.451.14.50.92
07fizh4350.680.682.590.009.040.2316.0719.880.2899.4540.345.414.30.76
07fizh4552.900.151.950.114.840.1218.6921.020.1199.8941.451.27.40.87
07fizh4751.530.512.840.106.530.1617.1920.590.2199.6641.548.210.30.82
07fizh4852.860.152.170.433.630.0817.9822.370.1599.8244.649.85.70.90
 
Upper V1 Sequence
07fizh5254.680.141.900.254.430.0618.2220.170.1199.9641.251.77.10.88
07fizh5553.900.172.030.144.770.1518.1819.790.1499.2740.551.87.60.87
08fizh252.840.252.240.056.100.2518.5019.040.1799.4438.4529.60.84
08fizh353.060.372.250.005.820.1417.8919.220.2098.9539.551.19.30.85
08fizh652.870.202.400.165.510.1817.8620.610.1699.9541.449.98.60.85
08fizh753.030.152.640.215.560.1617.8920.010.1699.8140.650.58.80.85
08fizh853.470.231.570.065.870.0918.5220.000.1799.9839.751.29.10.85
 
V2 Lava
08fizh1150.220.312.020.078.740.1816.4319.800.1597.924046.213.80.77
 
Boninite Dike
08fizh452.560.081.940.216.210.2517.8620.220.0499.3740.549.89.70.84
Figure 8.

Clinopyroxene Mg# versus (a) TiO2, (b) Al2O3, (c) Cr2O3, (d) Wo mol%, and (e) Na2O. Open and shaded fields are clinopyroxene compositions of the V1 (Geotimes) and V2 (Alley) Units from Alabaster et al. [1982], Einaudi et al. [2003], and A'Shaikh et al. [2005].

[30] Mg# (=Mg/[Mg + Fe*] where Fe* is total Fe as Fe2+) of the V1 clinopyroxene ranges from 0.65 to 0.93, which is comparable to or more varied than the previous reports for volcanic rocks from the Oman ophiolite [Alabaster et al., 1982; Ernewein et al., 1988; Umino et al., 1990; Einaudi et al., 2003; A'Shaikh et al., 2005]. This is consistent with the wide Zr range of bulk rock compositions in this area as described above. Mg# of clinopyroxene ranges widely from 0.65 to 0.90 for the LV1and from 0.74 to 0.93 for the MV1, while the UV1 clinopyroxenes range from 0.81 to 0.88, implying that the UV1 is comparatively less evolved and has a limited compositional range among the V1 lavas as the bulk rock compositions. As shown in Figure 8, analyses of the V1 sequence have a tight negative correlation between Mg# and TiO2, Mg# and Na2O, and a positive correlation between Mg# and Wo (=100 × Ca/[Ca + Fe* + Mg]) (Figures 8a, 8d, and 8e). Cr2O3 attaining 1.0 wt% decreases rapidly with decreasing Mg# (Figure 8c). Thus, the LV1, MV1 and UV1 clinopyroxenes show similar chemical variations.

[31] Mg# of clinopyroxene of the V2 samples spans also a wide range from 0.67 to 0.85 similar to that of the V1 sequence. TiO2 of clinopyroxene of the V2 samples, ranging from 0.21 to 0.56 wt%, shows a negative correlation with Mg# like the V1 samples, but is lower than the latter (Figure 8a). Such compositional trends are also recognized in Na2O contents (Figure 8e). Moreover, it is noted that clinopyroxene of the boninite dike in the V1 sequence is extremely depleted in TiO2 and Na2O (Figures 8a and 8e). Thus, the clinopyroxenes of the V2 samples and the boninite dike are apparently distinct from those of the V1 samples.

6. Chemical Stratigraphy

[32] Bulk-rock and clinopyroxene compositions shown above are examined in respect to the lithological stratigraphy. The lithological column and geochemical variations along the Wadi Fizh are illustrated inFigure 9.

Figure 9.

Stratigraphic variations of selected bulk-rock TiO2, Ni, Cr, Zr, Zr/Y, (La/Yb)n and clinopyroxene Mg#, TiO2 and Na2O. Grey crosses are sector zoned clinopyroxenes. See text for explanation.

[33] The LV1 lava shows overall smooth variations with evolved compositions (TiO2: 1.5–2.3 wt%; Zr: 87–153 ppm) except for a few spikes at 183, 250 and 330 mab. The incompatible elements such as Zr and Yb of the high-TiO2group gradually increase upward. Four samples belonging to the low-TiO2 group are characterized by less evolved but slightly enriched compatible elements. Three of them are lobate sheet flows below the jasper layer at 270 mab. These less evolved samples have similar (La/Yb)n and Zr/Y ratios to the other evolved rocks above and below. Clinopyroxene compositions are scattered but show broadly consistent variations with the bulk compositions.

[34] It is noticed that there is a gap in bulk and clinopyroxene compositions between above and below the jasper layer at 270 mab (Figure 9). The LV1b is characterized by more evolved bulk compositions (TiO2: 1.9–2.2 wt%; Zr: 118–153 ppm) and narrower variations than the LV1a (TiO2: 1.5–2.1 wt %; Zr: 86–141 ppm). Grayish pillow lava at 330 mab is the only exception that has a less evolved composition (Figure 2f). However, the entire LV1 shows similar (La/Yb)n ratios and a slight increase in Zr/Y ratios upward.

[35] The 0.4 m thick metalliferous sedimentary layer marks the sharp compositional gap between the LV1 below and the MV1 above. The lower part of the MV1 sequence consists of less evolved lavas with very low incompatible elements compared to the LV1b samples. The stratigraphic variations of the MV1 exhibit zigzag patterns with a wide range in TiO2 (0.6–2.1 wt%) and Cr (5–300 ppm). It is noticed that (La/Yb)n and Zr/Y ratios of the MV1 also exhibit zigzag profiles unlike those of the LV1. The overall zigzag patterns shown by the MV1 flows indicate that the MV1 basalts derived from a heterogeneous source or variable degrees of partial melting.

[36] The basal part of the UV1 sequence shows low incompatible element contents with a limited compositional range (TiO2: 0.6–1.0 wt%; Zr: 35–54 ppm) than the lower sequence. The lowest UV1 sample is also characterized by the highest compatible element content (Ni: 59 ppm; Cr: 199 ppm). Because the top of the MV1 lavas have high Zr contents up to 118 ppm, there is a sharp compositional gap through the transition from the MV1 to the UV1 at 670 mab. The zigzag profile of Zr/Y ratios of the UV1 implies that the UV1 lavas were derived from a slightly heterogeneous source.

7. Discussion

7.1. Effusive Section in the Wadi Fizh Area: Change From on- to off-Ridge Setting

[37] The result of seafloor observation indicates that the lava flow morphology correlates well with the gradient of slopes irrespective to a subtle difference in lava composition. Lobate sheet flows associated with submarine pahoehoe flows develop on flat, subhorizontal seafloor. In contrast, elongate pillows preferentially form on slopes with a gradient larger than approximately 5° and dominate on slopes >15° on submarine volcanoes off Hawaii Islands and on the rise slopes of the southern EPR at 14°S [Auzende et al., 1996; Umino et al., 2002; Gregg and Smith, 2003; Tominaga and Umino, 2010]. Gentle slopes around fast spreading ridges develop on (1) the axial summits and troughs and (2) the ridge flanks more than several kilometers off axis. High-resolution topographic exploration of the axial regions on fast spreading ridges has depicted narrow (1–200 m) and shallow (3–50 m) fault grabens and collapse pits on lobate sheets [Ballard et al., 1979; Gregg and Chadwick, 1996; Fornari et al., 1998; Engels et al., 2003; Fundis et al., 2010]. The occurrence of elongate pillows is only local in small scale on slopes inside the axial troughs and as pillow mounds formed around vents at low extrusion rates during the waning stage of eruptions [Fornari et al., 1998]. The superfast-spread upper oceanic crust drilled in ODP Hole 1256D has a 864 m thick extrusive layer consisting exclusively of massive and sheet flows, although most flows are highly fractured and brecciated. Only a limited interval of pillow lava occurs at 700–800 m below seafloor, 200 m above the top of the sheeted dike complex, which is correlated to pillow-dominant facies emplaced on the ridge slope [Tominaga and Umino, 2010]. The lowermost 200 m of massive flows and breccias at Site 1256 hence deposited on the ridge crest. Lobate sheet flows and bulbous pillows dominate in the lowermost V1 sequence, while elongate pillows occur at 230 mab. Applying the same idea on volcanostratigraphy in Hole 1256D to the V1 extrusive sequence, we correlate the lowermost 230 m of the LV1 to axial facies extruded and emplaced on the subhorizontal paleoridge crest and the thick pillow lava interval from 230 to 410 mab to the slope facies which was emplaced on the steeper slopes outside the ridge crest (Figure 10). The jasper layer at 270 mab, which divides the LV1 to LV1a and LV1b, manifests the hiatus of lava accumulation on the ridge flank. A jasper layer is considered to be precipitated from hydrothermal fluid as colloidal silica and iron-oxyhydroxide particles, which is easy to be dispersed and subject to erosion without capping by later volcanic products [Halbach et al., 2002; Grenne and Slack, 2003]. Its limited continuity and thin occurrence of the jasper layer at 270 mab suggest that it settled on local depressions during a short repose period before overlain by the subsequent flows.

Figure 10.

Schematic model of lava accretionary processes of the Wadi Fizh extrusive sequences on the basis of the geological and geochemical observations (modified after Tominaga and Umino [2010]). The LV1a consists of flows erupted and emplaced on the ridge crest, filling the axial trough. The LV1b consists of pillow lavas and lobate sheet flows emplaced on the ridge flank. After a short repose period, the MV1 lavas were emplaced off axis, with evolved flows transported from the ridge summit and less evolved flows erupted in situ. The UV1 erupted in situ from off-axial vents.

[38] The wide variations of the LV1 incompatible elements such as TiO2, Zr and Yb are ascribed to the coexistence of genetically different groups of lavas. As demonstrated by the Zr-Zr/Y variations (Figure 6), the low-TiO2LV1 magma cannot be produced by a simple fractional crystallization of the high-TiO2LV1 magma. On the contrary, the high-TiO2group shows a narrow range in Zr/Y ratios through the entire LV1 sequence. It is noted that Zr/Y ratios are weakly correlated with Zr, implying an overall differentiation of the LV1 flows upsection. The LV1 high-TiO2 group is thus derived from the same source and experienced progressive degrees of fractional crystallization with time. Although most elements show smooth stratigraphic variations, TiO2 concentration jumps over the short quiescence period at 270 mab from the less evolved LV1a to the more evolved LV1b. The LV1b consists of thick (∼130 m) flows with very limited, evolved compositions which was emplaced on the ridge flanks mainly as pillow lava. Their homogeneously evolved compositions, in addition to the large total thickness, suggest that they were derived from eruptions sufficiently large to extrude voluminous flows which flooded over the axial crest and flowed down on the ridge flanks.

[39] The 0.4 m thick umber dividing the LV1 and MV1 contains more heavy metals than pelagic sediments and is considered to have been correlated with hydrothermal activities [Robertson and Fleet, 1986; Karpoff et al., 1988]. Unlike the jasper layer at 270 mab, the umber layer is traced more than 3 km, suggesting a significant depositional period. Assuming a sedimentation rate of 0.01 mm/a [e.g., Berger, 1974], the 0.4 m thick umber represents a 40 thousand year time interval. If we assume a half-spreading rate of 5–10 cm/a, the MV1 should have deposited more than 2–4 km off axis from the top of the LV1. Although this is only a rough approximation, the distance of 2–4 km seems to be in accordance with the change in characteristic lava morphology from elongate pillows in the LV1 to common lobate sheet flows in the MV1. We suggest that the MV1 flows were emplaced off axis on a subhorizontal abyssal plain by traveling a long distance from the ridge crest or by in situ eruptions at off-axis vents.

[40] The lowermost 70 m thick flows in the MV1 above the umber are characterized by less evolved and depleted compositions with lower (La/Yb)n and Zr/Y ratios compared to the LV1, suggesting that these flows may be either derived from a more depleted source or formed by a higher degree of partial melting. The upper two third of the MV1 sequence comprises both evolved and less evolved flows, which alternate upsection. Recurrence of less evolved flows through the MV1 suggests a discrete source other than the axial vents that fed the more evolved LV1 flows. Evolved flows in the upper two third of the MV1 have similar compositions to the LV1 high-TiO2 group with the overall same Zr/Y ratios, indicative of a common source of these two. The evolved, more fertile lavas intervened with the MV1 may be flows that traveled a long distance through lava channels or tubes from the ridge axis [e.g., Soule et al., 2005]. The zigzag pattern of the geochemical variations in the MV1 sequence may represent alternating sources of the on- and off-axis magmatism. This geochemical pattern is similar to the northern EPR at 9°N where pillow ridges and mounds formed above off-axial source vents are sometimes covered with large-scale flows from the ridge axis [Perfit et al., 1994].

[41] The UV1 sequence is characterized by less evolved and depleted homogeneous compositions belonging to the low-TiO2group. The less evolved character suggests that the volcanism occurred significantly off axis beyond the limit of distance that evolved, voluminous flows can travel from the ridge axis. The presence of the fissure vent at 854 mab is the evidence of such off-ridge volcanism. We suggest that the UV1 lavas were generated from a homogeneous depleted source mantle. An isolated volcano is also reported from the Wadi Shaffan area [Einaudi et al., 2000, 2003]. These findings suggest the common presence of off-ridge volcanism on the Neotethys ocean floor.

[42] It is noted that the UV1 may be correlated to the Lasail Unit, because they have similar geological and petrological features. The Lasail Unit was defined by Alabaster et al. [1980]as a fractionation sequence from basalt through andesite to felsite which directly overlie the ‘axis’ lavas. The basaltic member is typified by small, distinctly gray-green, non-vesicular, aphyric and frequently bun-shaped pillow lavas. The Lasail Unit shows more primitive compositions (TiO2: 0.3–0.9 wt%; Zr: 20–50 ppm) than the Geotimes Unit (TiO2: 0.6–1.5 wt%; Zr: 50–200 ppm) [Alabaster et al., 1982; Lippard et al., 1986]. The UV1 and less evolved lavas of the MV1 which are roughly plotted in the field of the most primitive V1 and sheeted dikes, and the Lasail Unit by the previous studies (Figure 6). Alabaster et al. [1982]also claimed that the Lasail Unit is of seamount origin on the basis of their sporadical distributions. Both the UV1 and the less evolved flows in the MV1 were formed by off-ridge volcanism which is broadly regarded as a part of the mid-ocean ridge magmatism. We thus conclude that the Lasail Unit is a subunit of the V1, represented by the less evolved MV1 and UV1 in the Wadi Fizh area, and is essentially generated as a part of the mid-ocean ridge volcanism.

[43] The V2 lavas are also characterized by the depleted compositions like the LV1 low-TiO2 group. The V2 sequence is considered to be generated in an immature arc setting [Ishikawa et al., 2002], that makes the most outstanding difference of the V1 and V2 sequence. The presence of vesicular pyroclastic deposits interbedded with the V2 lavas in Ghayth 20 km south of the study area suggests more hydrous conditions for the V2 magma, in accordance with the arc-like setting [Umino et al., 1990]. The best indicator that discriminates the V2 from the V1 sequence is clinopyroxene chemistry [Alabaster et al., 1982; Umino et al., 1990; Adachi and Miyashita, 2003]. The V2 clinopyroxenes show lower incompatible elements than the V1 clinopyroxenes at an Mg# <0.85 (Figure 8). Distinction of the V2 from the V1 low-TiO2group is less obvious in bulk-rock chemistry, however, the V2 sequence is characterized by slightly lower Zr/Y and (Nd/Yb)n ratios than the V1 low-TiO2 group (Figure 6). The low Zr/Y ratios indicate a more depleted source or a higher degree of partial melting for the V2 than the V1 low-TiO2 group. Furthermore, lower (Nd/Yb)n ratios of the V2 lavas result in the REE patterns that cross those of the V1 lavas, indicative of discrete sources for the V1 and V2 magmas (Figure 7d).

7.2. Volcanism at a Segment Boundary

[44] The bulk-rock compositions of the V1 sequence in Wadi Fizh show a larger compositional variation ranging the previous Geotimes and V1 Unit from other areas [Ernewein et al., 1988; Beurrier et al., 1989; Einaudi et al., 2003; Godard et al., 2003, 2006; A'Shaikh et al., 2005]. The sheeted dike complex of the Wadi Fizh area coincidentally exhibits the broadest compositional range in the northern Oman ophiolite [Miyashita et al., 2003]. In contrast, the sheeted dikes have less evolved, limited compositions toward the segment center of the paleo-spreading axis at Wadi ath Thuqbah 25 km south of the study area [Umino et al., 2003]. Seismic experiments along the EPR and the Galapagos Spreading Center have shown that melt lenses are small or absent below the segment ends due to lower supply rates of magma to the upper crust [Hooft et al., 1997; Blacic et al., 2004]. This suggests that primitive melts supplied from the lower crust have more chances to extrude without mixing with the evolved magmas in a shallow magma chamber. Meanwhile, rifting of the upper crust under magma deficient conditions leads to the development of faults and tension fissures which provide pathways for deep hydrothermal circulation. Indeed, deep seawater infiltration into the lowermost crust has been demonstrated by the high water-rock ratios for the alteration minerals along the Wadi Fizh section [Kawahata et al., 2001]. This leads to a colder crustal condition that promotes fractional crystallization and differentiation of magma in a shallow magma chamber than at the segment centers. In addition to this, both evolved and less evolved flows derived from the adjacent ridge segment may add a further complexity to the segment boundary lava compositions. Eventually, the large compositional variation of the V1 extrusive sequence is consistent with the model of Adachi and Miyashita [2003]that the Wadi Fizh area corresponds to the leading edge of a propagating paleo-ridge segment. Infrequent axial eruptions at the segment boundary were followed by relatively long repose periods for hydrothermal deposits to precipitate.

[45] Off-axis volcanoes are more abundant in second- or third-order segment ends along the EPR [Scheirer and Macdonald, 1995; White et al., 2006a, 2006b]. Generally, lavas erupted from off-axial vents show a more diverse composition than those extruded from axial vents, since the off-axial lavas may retain the signatures of the source mantle heterogeneity due to a lack of persistent crustal magma chambers, which otherwise are obscured by mixing of differentiated and replenished primitive melts and assimilation of crustal materials [Perfit et al., 1994; Sims et al., 2002, 2003; Hall and Sinton, 1996; Geshi et al., 2007]. The zigzag compositional profiles of the MV1 are in good agreement with the characteristics of such an off-ridge environment as exemplified by the off-axis lavas at EPR 9°N [Perfit et al., 1994]. The UV1 is most likely generated by off-axis volcanism as indicated by the presence of the fissure vent accompanied by the pillowed ridge within the off-axial pahoehoe flows. The lower supply rates of magma at the off-axial segment end allowed extrusions of less evolved UV1 magmas directly from the lower crust.

8. Conclusions

[46] Temporal and spatial variations of extrusive layers from the ridge crest to off-ridge region were investigated in geological and geochemical terms along Wadi Fizh in the northern Oman ophiolite. A successive exposure of the V1 extrusive sequence is divided by a 0.4 m and a 0.8 m thick metalliferous sedimentary layer into three subsequences: the lower V1 (LV1), the middle V1 (MV1) and the upper V1 (UV1).

[47] The lowermost 230 m of the LV1 consisting of lobate sheet flows and bulbous pillow lavas formed on the flat ridge crest. The presence of elongate pillows at 230 mab represents a lithofacies emplaced on relatively steep slopes flanking the ridge. The LV1 is subdivided into two subsequences by a jasper layer at 270 mab. The stratigraphic variations of the lower subsequence (LV1a) show an evolved chemical compositional profile with a few less evolved levels. Similar Zr/Y ratios through the most LV1 sequence indicate that the LV1 originated from the same source material. Lavas in the upper subsequence (LV1b) were emplaced near the foot of the ridge slope where the gentle skirts extend away from the ridge. The 130 m thick evolved compositions of the LV1b represent voluminous lava flows fed from the axial source.

[48] Assuming a half-spreading and sedimentation rate to be 5–10 cm/a and 0.01 mm/a, respectively, the 260 m thick MV1 sequence was emplaced 2–4 km off the ridge axis. Predominance of lobate sheet and pahoehoe flows indicates that the MV1 flowed onto a subhorizontal off-axial abyssal plain. The evolved and less evolved flows in the MV1 sequence with wide compositional spectra represent distinct magmas which differ in the degrees of partial melting and/or fractional crystallization. These alternating flows comprise those erupted in situ at off-axis vents and those transported from axial vents. On the other hand, the UV1 sequence is a product of in situ off-axis volcanism as is evidenced by the fissure vent at 854 mab. Restricted depleted compositions of the UV1 lavas indicate an independent source other than the axial magma.

[49] The UV1 and less evolved lavas of the MV1 can be correlated to the Lasail Unit on the basis of the geological and petrological similarity. The Lasail Unit is considered to be a subunit of the V1, which formed by off-ridge volcanism. The V2 lavas can be discriminated from the V1 lavas by their lower bulk Zr/Y and (Nd/Yb)n ratios, and clinopyroxene compositions more depleted in TiO2 and Na2O.

[50] Wide compositional ranges shown by the V1 extrusive sequence comprising both evolved and less evolved magmas are consistent with those of the present upper oceanic crust at a segment end. Seawater penetration into the deep crust enhanced cooling and promoted fractional crystallization of subsurface magmas, which led to the production of the evolved V1 lavas and the sheeted dikes. At the same time, magma-deficient condition at the segment end enabled less evolved magmas from the deeper crust to extrude without mixing in the shallow magma chamber, and thereby resulted in the large spectrum of lava compositions. These magmas inherited the heterogeneities of the source mantle and erupted in situ from off-axis fissure vents to form pillow ridges within the field of pahoehoe flows in the MV1 and UV1.

Acknowledgments

[51] We thank to Hilal Al Azri, Salim Hamaed Al-Busaidi, Salim Omar Abdullah Al-Ibrahim, Mohamed Alaraimi, Durair A'shaikh (Ministry of Commerce and Industry of Oman), and Japanese Embassy in Oman for their kind support during field survey. The authors thank N. Tsuchiya (Iwate Univ.), T. Hiraga (Univ. of Tokyo), S. Tanaka, T. Hashimoto (Toshiba Home Tech. Co.), Y. Nogawa (Mitsubishi Material Techno Co.) and M. Hayashi (Foster Elec. Co., Ltd.) for their assistance during the field work. We are also indebted to E. Takazawa, N. Fujibayashi and T. Kurihara (Niigata Univ.) for fruitful discussions. Thoughtful reviews by S.M. White and anonymous reviewers greatly improved the manuscript. This study was supported by Sasagawa Scientific Research Grant from The Japan Science Society (21–711M) and Financial Support for Graduate Students from Niigata University 2010, 2011.