Previous   |    Next

doi:10.14379/iodp.pr.352.2015

Background

Izu-Bonin-Mariana system

The Izu-Bonin-Mariana (IBM) system is the type locality for studying oceanic crustal accretion immediately following subduction initiation. It is sufficiently old that it carries a full record of the evolution of crustal accretion from the start of subduction to the start of normal arc volcanism and sufficiently young that the key features have not been excessively disturbed by subsequent erosion or deformation. Intraoceanic arcs are built on oceanic crust and are sites of formation of juvenile continental crust (Rudnick, 1995; Tatsumi and Stern, 2006). Most active intraoceanic arcs are located in the western Pacific. Among these, the IBM system stands out as a natural scientific target. This predominantly submarine convergent plate boundary is the result of ~52 My of subduction (Ishizuka et al., 2011; Reagan et al., 2013) of the Pacific plate beneath the eastern margin of the Philippine Sea plate. Stretching for 2800 km from the Izu Peninsula, Japan, to Guam, USA (Figure F1), the IBM system (summarized in Stern et al., 2003) has been extensively surveyed and has become an important natural laboratory for International Ocean Discovery Program (IODP) expeditions aimed at understanding subduction initiation, arc evolution, and continental crust formation. A scientific advantage of studying the IBM system is its broad background of scientific investigation resulting from its designation as a focus site by the U.S. National Science Foundation MARGINS-Subduction Factory experiment and similar efforts in Japan. We know when subduction and arc construction began, even if the precise paleogeography is controversial, and there is a good time-space record of crustal development.

Petrologic evolution

The petrologic evolution of early stage magmatism in the IBM arc has been reconstructed mainly based on volcanic sections that are exposed on the fore-arc islands (Bonin Islands and Mariana Islands) and that have been recovered from Deep Sea Drilling Project (DSDP) and Ocean Drilling Program (ODP) fore-arc drill sites. Recent dredging and submersible studies have provided additional information. Consequently, we were able to predict the sequence of magmas likely to characterize the drill site and its surrounding region, which developed after subduction initiation and prior to establishment of a stable magmatic arc to the west by the late Eocene. The physical evolution of the fore arc and its associated lavas reflects the reorganization of mantle convection and slab-derived fluid flow in response to the changing behavior of the sinking Pacific plate. This evolution, from initial seafloor spreading and eruption of mid-ocean-ridge basalt (MORB)-like tholeiites to eruption of boninites to fixing of the magmatic arc ~150 km west of the trench (separated by a broad, dead fore arc), took 7–8 My (Ishizuka et al., 2011). The process is reflected in the succession of igneous rocks of the Bonin Ridge, which is described in greater detail below and depicted in the time-space diagram (Figure F2).

Early subduction-related volcanism

Diving and dredging on the fore-arc slope east of the Bonin Ridge and south of Guam recovered basaltic rocks from stratigraphic levels below boninite. These basalts have chemical compositions that are similar to those of normal MORB (N-MORB). However, they are not identical, and hence the term “fore-arc basalt (FAB)” was coined by Reagan et al. (2010) to highlight their distinctive setting and to emphasize that, in detail, FAB is a different magma type from MORB. Most of the reliable ⁴⁰Ar/³⁹Ar ages and U-Pb zircon ages of FAB are identical within error in both locations and indicate that FAB magmatism occurred from ~51 to 52 Ma, preceding boninite eruption by 2–4 My (Ishizuka et al., 2011; Reagan et al., 2013). Lavas with compositions transitional between FAB and boninites from DSDP Site 458 were dated at 49 Ma (Cosca et al., 1998). FAB and related gabbros are thought to relate to the first magmas produced as the IBM subduction zone began to form (Reagan et al., 2010).

Geochemical data show that FAB magmas have light rare earth element–depleted rare earth element (REE) patterns, indicating derivation from a moderately depleted lherzolitic upper mantle, similar to that responsible for generating MORB (Figure F3). FAB in the IBM, however, has low Ti/V (14–16), a diagnostic ratio (Shervais, 1982) that distinguishes FAB from subducting Pacific MORB (26–32) and from Philippine Sea back-arc lavas (17–25) (Figure F4). Chemically and petrographically, Bonin Ridge FAB are indistinguishable from Mariana FAB. This strongly implies that FAB tholeiitic magmatism was associated with fore-arc spreading along the length of the IBM arc. Low concentrations of incompatible elements and low trace element ratios such as Ti/V imply that FAB magmas were derived from more depleted mantle and/or were larger degree mantle melts than typical Philippine Sea MORB (Reagan et al., 2010).

Pb isotopic compositions of FAB from the Bonin fore arc show that they are, like other IBM arc and back-arc magmas, derived from a mantle with Indian Ocean characteristics, as demonstrated by high Δ8/4 Pb compared to Pacific MORB (Ishizuka et al., 2011). Isotopic characteristics indicate some differences between the mantle sources of Philippine Sea MORB and FAB, including distinctly higher ⁸⁷Sr/⁸⁶Sr and ²⁰⁶Pb/²⁰⁴Pb (Figure F5), which may imply the presence of lithospheric mantle with inherited enrichment (Parkinson et al., 1998). Most significantly, incompatible trace elements derived from subducted crust do not unambiguously affect the source region of most FAB, although some FAB lavas from the Mariana fore arc have Pb isotopic compositions consistent with a weak influence of subducted Pacific crust (Reagan et al., 2010). Differences in isotopic and trace element characteristics between IBM FAB and MORB including that from the Philippine plate strongly imply that FAB does not represent preexisting ocean basin or back-arc basin crust trapped prior to subduction initiation, as originally concluded by Johnson and Fryer (1990) and DeBari et al. (1999) for MORB-like tholeiites recovered from the Mariana and Izu inner trench walls.

Lavas with compositions that transition upward between FAB and boninite were recovered at DSDP Leg 60 Sites 458 and 459 (the alternate site) and illustrate that FAB and boninite are genetically linked (Reagan et al., 2010). The oldest of these lavas have REE patterns similar to those of MORB but are more enriched in silica and have higher concentrations of “fluid-soluble” elements such as K, Rb, U, and Pb than FAB. These lavas also have Pb isotopic compositions that are more similar to lavas from the Pacific than to those from the Indian plate, supporting the contention that subducted fluids were involved in their genesis. The youngest lavas at Site 458 are strongly depleted in REE, somewhat resembling boninites but less magnesian and more calcic.

Boninite volcanism follows FAB volcanism as an integral part of the evolution of the nascent subduction zone. The type locality of boninite is in the Bonin (Ogasawara) Islands, an uplifted segment of the IBM fore arc. Boninites are better exposed on the Bonin Islands than anywhere else in the world, particularly on the island of Chichijima. ⁴⁰Ar/³⁹Ar dating indicates that boninitic volcanism on Chichijima took place briefly during the Eocene, between 46 and 48 Ma (Ishizuka et al., 2006). A slightly younger volcanic succession is exposed along the Bonin Ridge, including 44.74 ± 0.23 Ma high-Mg andesites (HMA) from the Mikazukiyama Formation, the youngest volcanic sequence on Chichijima, and 44.0 ± 0.3 Ma tholeiitic to calc-alkaline andesite from Hahajima. Four submersible Shinkai 6500 dives on the Bonin Ridge Escarpment mapped an elongate constructional volcanic ridge atop the escarpment and recovered fresh andesitic clasts from debris flows along the northern segment of the ridge; they also recovered HMA lava blocks from the escarpment northwest of Chichijima. Three samples of andesite collected from the Bonin Ridge Escarpment range in age from 41.84 ± 0.14 to 43.88 ± 0.21 Ma (Ishizuka et al., 2006).

Boninites from the Bonin Islands are characterized by high MgO at given SiO₂ concentrations, low high-field-strength elements, low Sm/Zr, low REE, and a U-shaped REE pattern (Figure F3). These are “low-Ca boninites” (Crawford et al., 1989) and can be explained by low-pressure melting of depleted harzburgite that was strongly affected by slab flux. These boninites are isotopically characterized by high Δ7/4 Pb, high ⁸⁷Sr/⁸⁶Sr, and low ¹⁴³Nd/¹⁴⁴Nd relative to local MORB and FAB sources (Figure F5). In contrast to the FAB mantle source, which was not much affected by subduction-related fluids or melts, the boninite magma source manifests a major contribution from subducted pelagic sediment and oceanic crust. The boninites are also distinct from ~44 Ma lavas exposed on Hahajima Island and recovered by Shinkai 6500 diving on the Bonin Ridge Escarpment (Ishizuka et al., 2006). HMA from Chichijima and the Bonin Ridge Escarpment are more similar to relatively enriched boninitic lavas from ODP Site 786 (Pearce et al., 1992) and Guam, including having higher Sm/Zr at a given Zr content and higher REE and Ti concentrations compared to Chichijima boninites (cf. Taylor and Nesbitt, 1994). The HMA are isotopically distinct from the boninites (Figure F5) and were derived from a source mantle that was less affected by fluids or melts derived from the subducted plate.

Post-45 Ma, tholeiitic to calc-alkaline andesites from the Bonin Ridge and ~45 Ma rhyolites from Saipan (Reagan et al., 2008) exhibit strong characteristics of arc magmas: they are depleted in Nb and enriched in fluid-mobile elements such as Sr, Ba, U, and Pb. These characteristics indicate that, by 45 Ma, near-normal configurations of mantle flow and melting, as well as subduction-related fluid formation and metasomatism, were established for this part of the IBM arc. The Bonin Ridge Escarpment, Mikazukiyama Formation, and Hahajima andesites thus represent a transitional stage from the waning stages of fore-arc spreading (represented by FAB and perhaps boninites) and the stable, mature arc that developed in the late Eocene. These orthopyroxene-bearing, high-Mg, tholeiitic to calc-alkaline andesites erupted along the Bonin Ridge Escarpment as the arc magmatic axis localized and retreated from the trench. Post-45 Ma andesites (and basalts) do not show the influence of pelagic sediment melt from the slab (Figure F5). Instead, the mantle source seems to have only been affected by hydrous fluid derived mainly from subducted altered oceanic crust.

Overall, the geochemical and isotopic characteristics of the IBM arc along its entire length appear to have evolved in tandem with the formation of a new subduction zone and a new mantle flow regime by

1. Initial decompression melting with little to no slab flux, producing MORB-like basalt and fore-arc spreading (51–52 Ma),
2. Mixing of fluids or melts from subducted sediments and oceanic crust into an extremely depleted (harzburgitic) mantle to generate boninites (49–45 Ma), and
3. Continued influx of hydrous fluid input into increasingly fertile lherzolitic mantle to generate tholeiitic and calc-alkaline magma (post-45 Ma), marking the time when a mature, stable arc magmatic system was finally established (Ishizuka et al., 2006, 2011).

Note however that although we have established a general volcanic stratigraphy, it is a composite stratigraphy based on dredging, submersible grab sampling, and coring at widely spaced localities. There is no reference stratigraphic section to check this subduction initiation stratigraphy and, in particular, to identify the nature of the boundaries between the units and demonstrate that units have not been missed. Defining this stratigraphic section is the aim of this expedition.

Tectonic evolution

It has been generally accepted (Bloomer et al., 1995; Pearce et al., 1999; Stern, 2004; Hall et al., 2003) that the IBM subduction zone began as part of a hemispheric-scale foundering of old, dense lithosphere in the western Pacific (Figure F6). The beginning of large-scale lithospheric subsidence, not true subduction but its precursor, is constrained to predate the 51–52 Ma age of igneous basement of the IBM fore arc (Bloomer et al., 1995; Cosca et al., 1998; Ishizuka et al., 2006). The sequence of initial magmatic products is similar everywhere the forearc has been sampled, implying a dramatic episode of asthenospheric upwelling and melting associated with magmatism and seafloor spreading over a zone that was perhaps hundreds of kilometers broad and thousands of kilometers long. It is clear from the extensive geochronology for IBM fore-arc rocks that this episode took place ~52 My ago, and was followed by a period of shallow hydrous melting through about 44–45 My ago (Figure F2). Expedition 352 drilling intends to sample these parts of the tectonic history of the IBM arc.

Interestingly, these time-space trends in IBM fore-arc composition can be found in many ophiolite terranes. The world’s largest ophiolite, the Semail ophiolite of Oman/United Arab Emirates has long been known to exhibit a stratigraphy of FAB-like tholeiites overlain by depleted arc tholeiites (e.g., Alabaster et al., 1982), and recent discoveries of boninites in the upper part of the sequence (Ishikawa et al., 2002) confirm the full trend from tholeiite to boninites. Other large, complete ophiolites with complex fore-arc-type stratigraphies involving tholeiites and boninites include the Troodos Massif of Cyprus, the Pindos Mountains in Greece, the Bay of Islands ophiolite in Newfoundland (Canada), and numerous others distributed through most of the world’s mountain belts (e.g., Pearce et al., 1984; Dilek and Flower, 2003). Many of these stratigraphies are economically significant, with associated volcanogenic massive deposits and/or podiform chromite mineralization.

The presence of boninites is in itself an important tectonic indicator, requiring a combination of shallow melting, high water content, and depleted mantle. Boninites are defined by the International Union of Geological Sciences (IUGS) to have >52 wt% silica, <0.5 wt% TiO₂, and >8 wt% MgO and can usefully be distinguished from basalts on a diagram of Ti8 versus Si8, where Ti8 and Si8 refer to the oxide concentrations at 8 wt% MgO (Pearce and Robinson, 2010). On this projection (Figure F7), the earliest lavas are basalts (FAB) that plot in the MORB field. Later lavas (~48–44 Ma) plot as boninites before compositions eventually become basaltic again with eruptions at, for example, Hahajima. This appears to be a characteristic of subduction initiation, but to properly interpret its tectonic significance we need the full lava stratigraphy to know whether the basalt–boninite transition is gradational or episodic or has both magma sources available simultaneously. Drill core would also enhance the opportunity to obtain glass samples that can be analyzed for volatile and fluid-mobile element concentrations.

After a brief period of spreading, lavas began to build atop the newly formed crust and retreat from the trench, at the same time changing composition, perhaps first from FAB to boninite and then from boninite to calc-alkaline and tholeiitic arc magmas. The timing and rate of the migration of the magmatic locus away from the trench remains uncertain, but it is clear that the locus of magmatism reached the location of the first magmatic arc on Guam and the edge of the Ogasawara escarpment within ~8 My. This left vast tracts of infant arc crust “stranded” to form the IBM fore arc, which cooled and remained “frozen” in its primitive state. Understanding the formation of fore-arc crust is clearly critical for understanding the formation of subduction zones (and the magmatic responses), growth of arcs, evolution of continental crust, and origins of ophiolite.

Structure and thickness of fore-arc crust

The most detailed trench-orthogonal published images of IBM fore-arc crustal structure in the region of interest come from a seismic refraction/reflection study by Kamimura et al. (2002) in a region some distance north of the section from Sites U1439–U1442. While recognizing that the actual crustal structure for the drill sites might be slightly different than that shown, we infer from the study of Kamimura et al. (2002) that the crust beneath the drill sites is 6–8 km thick—slightly thicker than normal oceanic crust. In detail, the crust beneath this part of the fore arc can be divided into 5 identifiable layers. The first layer (V_P = 1.8–2.0 km/s) is mostly composed of thin sediment; this layer is actually very variable and Sites U1439–U1442 were originally chosen to have at least 100 m of sediment in order to facilitate drilling and casing operations of the uppermost part of the holes. The second layer (V_P = 2.6–3.3 km/s) is 1–2 km thick and probably consists of fractured volcanic rocks and dikes; this information contributed to our precruise estimate of 1.25 ± 0.25 km as the likely lava thickness that we needed to drill in order to reach the sheeted dikes.

Choice of drill sites

The Bonin fore arc was chosen because it had the advantage of being in the same region as Chichijima (Bonin Island), the type locality for the key boninite rock type. It is part of a complete ophiolite section that has been sampled by dredging and diving (Ishizuka et al., 2011) and has full site survey data (Kodaira et al., 2010, pers. comm., 2013).

Two important hypotheses to be tested by drilling have been (1) that subduction initiation produces a consistent volcanic stratigraphy (from oldest to youngest): FAB, transitional lavas, low-Ca boninites, enriched HMA and related rocks, and normal arc volcanic rocks (Reagan et al., 2010), and (2) that this sequence was originally stacked vertically before erosion and therefore represents an in situ analog for sections through many SSZ ophiolites. Sites U1439–U1442 (Figure F8) were chosen to maximize the likelihood of testing these hypothesis because the sheeted dike/FAB contact was approximately located during Shinkai 6500 diving in 2009 along the inner wall of the Bonin Trench, near a location where the drill could spud into a sediment pond and sample the lower part of the fore-arc volcanic succession.

Figure F9 summarizes the distribution of rocks sampled during three expeditions: YK04-05, the first manned submersible (Shinkai 6500) diving survey of the western escarpment of the Bonin Ridge (Ishizuka et al., 2006); R/V Hakuho-maru KH07-2, which dredged 19 stations along the length of Bonin Ridge; and YK09-06 in the proposed Site U1439–U1442 area (Ishizuka et al., 2011). They show that, in particular,

1. Overall, there is an ophiolite-like sequence in the inner trench wall of lavas, dikes, gabbros, and peridotites;
2. Of the lavas and dikes, MORB-like tholeiites occupy the deepest part of the trench-side slope of the ridge (i.e., the easternmost part of the ridge). These are chemically indistinguishable from FAB as defined by Reagan et al. (2010) from the Mariana fore arc;
3. Boninites crop out to the west and upslope of the FAB/MORB outcrops; and
4. Younger tholeiitic/calc-alkaline basalt to rhyolite crop out on the western Bonin Ridge and are especially well exposed on the western escarpment.

The 2009 diving survey using the Shinkai 6500 examined and better established the igneous fore-arc stratigraphy exposed on the trench-side slope of the Bonin Ridge (YK09-06 cruise: 24 May–10 June 2009; Ishizuka et al., 2011). The northernmost dive area near 28°25′N for this survey (Dives 1149, 1150, 1153, and 1154) was located near the drill sites (Figure F8).

The deepest dive (1149) sampled gabbro and basalt/dolerite and appears to have traversed the boundary between the two units. The lower slope traversed during Dive 1149 is composed of fractured gabbro, whereas pillow lavas were observed in the uppermost part of this dive at ~6000 m water depth. Dives 1153 and 1154 surveyed upslope of Dive 1149 between 6000 and 5200 m water depth. These two dives found outcrops of gabbro and dolerite, as well as fractured basalt lava cut by dikes. The contact between basalt and dolerite was thought to be ~5400 m based on these results, and Site U1440 was chosen to drill through this contact (Figure F10). The shallowest dive (1150; 4600 to 3700 m) recovered volcanic breccia and conglomerate with boninite. Thus, the boundary between boninite and basalt was estimated to lie at ~4000 m water depth. Site U1439 was chosen to drill through the transition zone from boninite to basalt. The resulting drilling has shown that this boundary might be geographically limited and lie to the east.

Top of page   |    Previous   |    Next