Previous   |    Next

doi:10.14379/iodp.pr.352.2015

Preliminary scientific assessment

Expedition 352 successfully cored ~1.22 km of igneous basement created immediately following subduction initiation in the Bonin fore arc, together with 461 m of overlying sediment (Table T2). The original plan was to drill 2 sites to 750 m each. Drilling conditions limited depths of penetration at both sites and, if the original planned casing strategy had been used, little time would have been left for additional drilling at alternative sites. The decision to drill in the casing at both original sites saved ~10 days, which provided the time for drilling 2 additional sites (U1441 and U1442). This marks only the second time a casing string was drilled in since ODP Leg 196, and the first time a complete reentry system was deployed this way.

The basement core provides diverse, stratigraphically controlled suites of lavas and shallow intrusive equivalents of our target rock types: FAB and boninite related to seafloor spreading and earliest arc development. FAB were recovered at the two deeper sites (U1440 and U1441) and boninites at the two shallower sites (U1439 and U1442). Although recovery averaged ~21% for the igneous sections drilled (see “Principal results” for details), this recovery was sufficient to provide an excellent suite of samples for documenting the petrology, geochemistry, volcanology, and structure of the four basement sites. Onboard observations and data collection have already produced a significantly clearer understanding of both the development of the crustal architecture of the Bonin fore arc and the variations in sources and melting mechanisms through time for this region. The sediment core provides a record of the depth-time evolution of the fore-arc basement following subduction initiation and, through study of the interspersed ash layers, contributes to our understanding of the overall volcanic evolution of the region.

The overarching goal of Expedition 352 was to characterize the volcanic products of subduction initiation and early arc development and to use the results to understand better the subduction initiation process and its relevance to on-land geology. The objectives related to this goal and stated in the Scientific Prospectus for this expedition are listed below, accompanied by our assessment about how these objectives were addressed and the resulting discoveries. We conclude with a statement listing discoveries that went beyond those directly addressing the original objectives.

1. Obtain a high-fidelity record of magmatic evolution during subduction initiation by coring volcanic rocks down to underlying intrusive rocks, including radiometric and biostratigraphic ages.

One major achievement of Expedition 352 was the recovery of 4 lava sequences that provide evidence for the temporal evolution of volcanic activity during the nascent development of the IBM volcanic arc. These sequences cluster into 2 groups (Figure F19). The first group (Sites U1440 and U1441) addressed the chemostratigraphy and petrological evolution of FAB, and the second group (Sites U1439 and U1442) did the same for boninite-series igneous rocks. Defining the petrological/geochemical units and formulating ideas about sources and melting processes needed to generate these lavas onboard the ship was made possible by the analysis of cut rock surfaces using a Thermal Niton pXRF instrument. Expedition 352 is the first instance in which pXRF measurements have been successfully used to distinguish rock units chemically while core is being described. These real-time data also proved useful in targeting intervals for shipboard ICP-AES analyses on cores as they were recovered. The collaboration between the shipboard geochemists and petrologists emphasizes the importance of close coordination between different areas of expertise during core description and analysis, here resulting in the establishment of a new tool for use during basement coring. The targeted ICP-AES data collected onboard allowed us to generally classify the rocks based on major element data and provided important additional trace elements.

The formal naming of the volcanic rocks was carried out using modified IUGS protocols based on the three major element oxides, TiO₂, SiO₂, and MgO. This allowed us to identify parental boninites based on the IUGS criteria (Le Bas et al., 2000) of <0.5 wt% TiO₂, >52 wt% SiO₂, and >8 wt% MgO (Figure F20). Criteria based on those in Pearce and Robinson (2010) were used to extend the classification of parental magmas to boninite series. As Figure F20 shows and as is documented in the individual site chapters, these 2 diagrams provide the basis for the conclusion that the deeper sites (U1440 and U1441) predominantly recovered tholeiitic (FAB) lavas and dikes whereas the shallower sites (U1439 and U1442) predominantly recovered boninitic lavas and dikes.

In addition, lava compositions varied enough in the boninites to compel division into three series. We defined a basaltic boninite series to include lower SiO₂ concentrations for parental magmas than those allowed by the IUGS classification. The boninite field on the MgO versus SiO₂ diagram itself was divided into low-Si and high-Si boninite series following the concept of Kanayama et al. (2013). This subclassification reveals some important variations. Most importantly, relatively primitive high-Si boninites form the upper units of both boninite holes. Another significant finding is the abundance of evolved members of the boninite series (high-Mg andesites) in igneous Unit 2 of Hole U1442A and at the base of the lava sequence in Hole U1439C. Only one series was made for the FAB rocks based on their uniformly basaltic parent magma compositions. Nevertheless, FAB lavas have ranges in TiO₂ concentrations that require significant variations in parental magma compositions. For example, several rocks plot as basalt in the MgO versus SiO₂ diagram but on the boninite/basalt boundary of the TiO₂ versus SiO₂ diagram. We termed these lavas D-FAB (D for depleted), to our knowledge the first lavas of this composition to be recovered from in situ oceanic crust.

Note that only ICP-AES data can be classified in this way because SiO₂ and MgO could not be analyzed by pXRF. Moreover, SiO₂ is significantly dependent on alteration. As a result we developed proxy diagrams (Ti versus Cr and Ti/Zr versus Cr) that can informally classify the whole core even for relatively highly altered samples or for those not chosen for ICP-AES analysis. The classifications confirm those made by Figure F20 and have enabled each unit from each core to be classified in terms of magma type, as shown in the stratigraphic summary of Figure F19.

FAB Sites U1440 and U1441

The chemical compositions and mineral abundances of the lavas and dikes cored at Sites U1440 and U1441 are similar to each other as well as to those for FAB documented during nearby dives (see Ishizuka et al., 2011), as well as for FAB from the Mariana fore arc (Reagan et al., 2010). Lavas from Hole U1440B are underlain by a transition zone and then 40 m of what appears to be a sheeted dike complex, which we take to indicate that we drilled the entire extant FAB lava sequence. In contrast Hole U1441A had significantly less penetration, was capped by a talus deposit, and did not intersect intrusive rocks. Therefore, Hole U1440B has the best reference section for illustrating the chemostratigraphy of the FAB sites.

FAB lavas and dikes from Sites U1440 and U1441 are seen in Figure F20 to be tholeiitic rather than boninitic, and the total alkali-silica (TAS) plot of LaBas et al. (1986) shows them to be predominantly basalts. Exceptions are a few basaltic andesites with 52–53 wt% silica, and the highly differentiated andesite that makes up igneous Unit 6 at Site U1440 (see “Site U1440 summary”). With the exception of the Unit 6 andesites, FAB from our sites and other IBM fore-arc locations are characterized by low Ti/Zr ratios and low abundances of both elements relative to MORB (Figure F21), demonstrating that they are generated by higher degrees of melting or from a more depleted mantle than for lavas from mid-ocean ridges.

The concentrations of trace elements and CaO in FAB lavas are relatively diverse because of variations both in parental melt compositions and in degree of differentiation. Below the heterolithic breccia making up Unit 1 at Site U1441 and especially beneath Unit 4 in Hole U1440B, Cr and Mg concentrations generally increase with depth (Figures F22, F23), indicating an increase in the degree of differentiation upward. Nevertheless, these trends have significant overall fine-scale diversity and are punctuated by narrow sections of core with significantly more, or less, differentiated lavas. The andesites from Site U1440 Unit 6 mentioned above represent one such significant compositional excursion. These lavas have the lowest Ti/Zr and Sr/Zr values (Figure F21) of any of the FAB-related lavas. We attribute this to extensive plagioclase and titanomagnetite fractionation, perhaps in a small, isolated, and shallow magma reservoir. Other compositional excursions are represented by the high-Cr, and hence less fractionated, lavas in the two cores that make up Unit 3 in Hole U1440B and the D-FAB from Hole U1441A.

Superimposed on these magma chamber effects are variations in incompatible trace element ratios that represent variations in source compositions. Two units from Hole U1440B (Units 4 and 8) and some basal dolerite dikes have Sr/Zr ratios similar to those of average N-MORB (~1.25), indicative of mantle sources with little or no subducted material. Other units have relatively elevated Sr/Zr, but most are still within the range of MORB glasses taken from the laser-ablation inductively coupled plasma–mass spectrometry data set of Jenner and O’Neill (2012) (Figure F21), suggesting that any subduction influence was small, at least in terms of Sr input, throughout the genesis of most lavas from this site. A small set of samples do, however, plot clearly above the field of MORB glasses. This includes the aforementioned D-FAB samples from Hole U1441A, where the ultradepleted source may have made the subduction component more evident. Sr is, of course, mobile during alteration. Nevertheless, the consistency of Sr/Zr within most units likely reflects interbedding of lavas derived from mantle sources with different subduction inputs. The D-FAB unit, as noted above, has exceedingly low incompatible trace element abundances, the highest Ti/Zr (Figure F21), and the highest CaO concentrations of any uncalcified lava collected during the expedition. This lava was generated by the highest degree of melting, or from the most depleted source, of all FAB, and we tentatively interpret its high Ti/Zr and high CaO to reflect melting of its mantle source to near the point of clinopyroxene exhaustion.

Boninite Sites U1439 and U1442

The compositions of boninite group lavas drilled at Sites U1439 and U1442 are chemically distinct from FAB. In contrast to FAB Sites U1440 and U1441, lavas in both of the boninite sites have compositions that become more primitive upward. Irrespective of degree of differentiation, TiO₂ concentrations generally decrease and SiO₂ concentrations generally increase upward in lavas from both boninite sites. Based on these compositional variations, the lowermost lavas classified as basaltic boninites and low-Si boninites, and the capping lavas as high-Si boninites resembling those that form the base of the lava sequence at Chichijima. Nevertheless, differences in the thicknesses of chemostratigraphic units are surprisingly great considering that the sites were only ~1.3 km apart (Figure F24).

Our best correlation between the stratigraphic records at the two sites places differentiated basaltic boninite to low-Si boninite series lavas at the base of both sites. The uniformly low Cr and MgO concentrations in these lavas suggest they represent magmas that fractionated in a magma chamber that persisted during the eruption of this unit. The thickness of this basal sequence appears to change from ~20 to ~200 m from Site U1439 east to Site U1442. Overlying the basal sequence at both sites are lavas of the low-Si boninite series that change in thickness from ~230 m at Site U1439 to 80 m at Site U1442. These lavas have higher and more variable Ti/Zr ratios than the underlying lavas (Figure F21), indicating that they were tapped from variably depleted mantle by variably high degrees of melting.

Mingling between magmas with high and low Cr concentrations is common in this unit. This demonstrates that some magmas ponded for long enough in the crust to undergo significant crystal fractionation and erupt, whereas others rose to the surface essentially unfractionated, and that these 2 magmas commonly mingled and erupted before significant mixing could occur. Unit 5 in Hole U1439C represents a compositional excursion to basaltic boninite-series lavas. These lavas differ from equivalent series lavas from deeper in this hole in that they have relatively high Cr concentrations in keeping with the more primitive compositions of lavas from adjacent depths in this hole. The high-Si boninites atop both holes have similar compositions. Although a few significantly differentiated lavas are present in these upper boninites, most are relatively primitive with Cr concentrations of 200–1600 ppm and MgO of 9–17 wt%. The extreme depletion of the mantle sources and degrees of melting for these lavas is reflected in the low TiO₂ concentrations, which are typically <0.3 wt%, and in low Ti/Zr ratios, which are <60.

Although coring in Hole U1442A ended in boninite-group lavas, Hole U1439C drilled ~50 m into a sequence of dolerites, with compositions spanning the range of Ti/Zr ratios of overlying lavas but with overall lower Cr concentrations. Our inference is that genetically related dikes underlie the boninitic-group lavas. We found no evidence for the presence of FAB resembling those drilled at Sites U1440 and U1441 beneath the boninite-group lavas.

DSDP Sites 458 and 459

Selected archive cores from Site 458 were onboard the JOIDES Resolution for training purposes during Expedition 352. We used this opportunity to analyze these cores using the pXRF instrument so that their compositions could be directly compared with those from our drill sites. The results show that the rocks from DSDP Site 458 are generally more differentiated than those drilled at Sites U1439 and U1442, with Cr concentrations ranging from 275 ppm to below detection limits. Site 458 Cores 28, 39, and 43 have TiO₂ and Zr concentrations resembling those of the low-Si boninites from the lower to middle sections of Holes U1439C and U1442A, whereas Core 47 has compositions similar to the andesite from Hole U1440A (Figure F21). The original geochemical data for DSDP sites (Wood et al., 1982; Meijer et al., 1981) demonstrate that Cores 43–45 from Site 458 and Cores 60–64 from Site 459 have compositions that are intermediate between those of our FAB and boninite sites. Our Cr and MgO concentrations and those published in Wood et al. (1982) indicate that the Site 458 and 459 lavas are typically highly fractionated. This further supports the existence of a magma chamber during the eruption of these transitional lavas.

2. Use the results of Objective 1 to test the hypothesis that fore-arc basalt lies beneath boninites and to understand chemical gradients within these units and across their transitions.

We expected FAB to be present at the base of the Bonin fore-arc volcanic succession and a sequence of boninite-series lavas to be present atop these FAB. Our pre-expedition drilling strategy, in fact, was to drill a single section from boninite through to FAB dikes. We did not, however, encounter this stratigraphy at any of the drill sites. Instead, the presence of dikes at the base of the sections at Sites U1439 and U1440 provides evidence that these lavas are underlain by their own conduit systems and that FAB and boninite group lavas are likely offset more horizontally than vertically.

The separation of compositions for the FAB and boninite sites is best seen on the plot of Zr versus TiO₂ concentrations (Figure F21). On this plot, the trenchward FAB sites (U1440 and U1441) have high Ti/Zr ratios that completely separate them from the low Ti/Zr ratios, and low absolute values of Ti and Zr, of the boninite sites (U1439 and U1442). This gap appears to be filled by the compositions of lavas from Sites 458 and 459, providing some evidence that we drilled the 2 end-members of a FAB–boninite spectrum, but there is a continuity of compositions within which transitional members do exist. The implication is that FAB erupted closer to the trench than boninite, and we speculate that lavas of transitional compositions may have erupted at an intermediate distance from the trench.

3. Use drilling results to understand how mantle melting processes evolve during and after subduction initiation.

The results summarized in the assessment of Objective 2 are best explained by temporal evolution, not just of the mantle melting process itself, but also of location of that melting and of the nature of the plumbing system taking the melts from the mantle to the surface. Although the compositions of FAB lavas erupted at Sites U1440 and U1441 vary through time in terms of the amount of slab fluid involved in their genesis and their extent of differentiation, all are relatively evolved, with most Mg concentrations lying within the range 5–8 wt%. These lavas could have been fed from magma chambers that persisted throughout the eruptive history of FAB.

In the boninitic section, the lowermost lavas at Sites U1439 and U1442 also are relatively differentiated, with a thickness that increases to the east (i.e., from Site U1439 to U1442). Stratigraphically higher lavas at both sites are both differentiated and primitive in composition, with mingling common between the two. Primitive high-Si lavas derived from the most depleted mantle cap the stratigraphy at both sites. The changes in composition support a model in which a persistent magma chamber system was present early during genesis of boninite group lavas, especially at eastern Site U1442. This persistent chamber system disappeared by the time high-Si boninites erupted at both sites. If the lower boninites formed at a ridge axis, these high-Si boninites could represent off-axis magmatism, perhaps the start of a proto-arc.

Overall, the evidence supports the presence of at least a small subduction contribution to the genesis of FAB lavas and dikes. As noted above, the Sr/Zr ratio is equivocal for all but a small set of FAB lavas, including the most depleted where the subduction component is most visible. However, the drilled FAB (Figure F21), as well as those previously sampled from the IBM region (Figure F4), plot in the island arc field on a V-Ti plot, likely reflecting water-enhanced melting resulting in higher oxygen fugacity and a greater degree of melting (or melting of more depleted sources) than is normally prevalent at mid-ocean ridges (Shervais, 1982).

Boninites all were generated after the addition of a more significant flux of water-rich fluid, and perhaps a melt, from the subducting slab. The degree of depletion of TiO₂ (Figure F21), as well as CaO, increases upsection, implying that the overall depletion of the mantle continues from the FAB group through into the boninite group lavas.

These results indicate that seafloor spreading related to subduction initiation and eruption of FAB, having begun at ~52 Ma (Ishizuka et al., 2011; Reagan et al., 2013), also migrated from east to west. We believe, on the basis of the consistently evolved lower lavas and dikes, that spreading was rapid and, like fast-spreading centers (e.g., the East Pacific Rise; Langmuir et al., 1986), an axial magma chamber was present. Melting was largely decompressional during this period, but subducted fluids significantly affected at least some of the melting. Relatively fast spreading continued migrating to the west of the subduction zone through the eruption of the differentiated basaltic boninites. One plausible hypothesis is that the spreading rate had by that time declined such that a persistent magma chamber could no longer be maintained, allowing progressively more primitive lavas to erupt (cf. slow-spreading ocean ridges, Dick et al., 1989; Langmuir et al., 1992). The high-Si boninite at the top of the lava sequence might then represent the final magmatic phase of such a ridge when most erupting lavas were primitive. Alternatively, the high-Si boninites could represent off-axis eruptions resulting from continued melting of depleted mantle trenchward of the westward migrating ridge axis.

The initial extreme depletion of the sources for all boninite group lavas was likely related to FAB generation. This depletion resulted in melting of harzburgite by the time the high-Si boninites were generated. Melting of such mantle to the point of clinopyroxene exhaustion resulted from a strong influx of a subduction component from the subducting plate.

4. Test the hypothesis that the fore-arc lithosphere created during subduction initiation is the birthplace of supra-subduction zone ophiolites.

By recovering the first sections of in situ subduction initiation oceanic crust, Expedition 352 successfully demonstrated that fore-arc lithosphere created during subduction initiation could be a potential birthplace of suprasubduction zone ophiolites. This was another major achievement of this expedition.

Specifically, Expedition 352 drilled the end-members of the suprasubduction zone ophiolitic spectrum: oceanic lithosphere created from FAB magma (Sites U1440 and U1441) and oceanic lithosphere created from boninite magma (Sites U1439 and U1442). Although we do not have the complete crustal sections, the fact that two sites rooted in dikes following penetration of oceanic volcanic rocks is consistent with such an interpretation. Moreover the recovery of gabbros and peridotites by dredging and submersible from the deeper parts of the fore arc supports the likelihood that at least the deeper section has an ophiolitic structure.

The relative ages of the 2 sets of sites is critical for any interpretation and will be an important part of the follow-up work. Although they have not yet been dated, comparable rocks recovered from the fore arc prior to the expedition leads us to expect that the age of the FAB section lies between 51 and 52 Ma (Ishizuka et al., 2011; Reagan et al., 2013), whereas the age of the boninite section is somewhat younger (low-Si boninite from Site 458 is ~49 Ma, the oldest high-Si boninite from Chichijima is 48 Ma; Cosca et al., 1998; Ishizuka et al., 2006).

There are ophiolites that have full lava sequences based on only one magma type and, for these, the Expedition 352 sites could provide good analog sections. Some well-preserved examples of ophiolitic upper crust derived entirely from FAB magma include the Western Mirdita complex (Dilek et al., 2008) and the Coto Block of the Zambales ophiolite (Yumul, 1996), which only have FAB lavas (using FAB in the broadest sense of any tholeiitic basalt formed by spreading and located in a fore-arc setting). Examples of ophiolitic crust derived entirely from boninitic magma include Betts Cove in Newfoundland (Bédard et al., 1998) and the Acoje Block of the Zambales ophiolite (Yumul, 1996). In their study of Betts Cove, Bédard et al. (1998) inferred that this entirely boninitic ophiolite formed in a fore-arc setting, a conclusion supported by the core from Sites U1439 and U1442.

However, we are aware that the sequence of events in which fore-arc basalts are overlain by boninites is also common in many ophiolites. Well-documented examples include the Troodos, Oman, Eastern Mirdita, and the Pindos/Vourinos ophiolites (see review by Dilek and Flower, 2003). Here, volcanic rocks of FAB compositions (again, in broad sense) are overlain by lavas, or intruded by dikes, of boninitic composition. Compositionally composite sequences such as these are not comparable with our drilled sections, although the magmatic sequence of FAB followed by boninite in these ophiolites is consistent with our inference that FAB magmas predated boninite magmas in the Bonin fore arc. It is possible, indeed likely, that such composite lithosphere is located between our 2 sets of sites, where there was insufficient sediment for us to drill. However, we are fortunate that deep-sea drilling in the Mariana fore arc (at Sites 458 and 459, our alternate site) did recover lavas of both boninite and FAB composition, as well as lavas transitional between the two, within the same drilled section.

Thus, between our Bonin fore-arc drill sections and those of DSDP in the Mariana fore arc, we usefully cover much of the compositional spectrum of the world’s SSZ ophiolites. We note that the IBM fore arc may not be an exact analog for all SSZ ophiolites. Nevertheless, the general concept of increasing mantle depletion and subduction flux in a highly extensional subduction initiation setting could be a viable general model for their evolution.

Top of page   |    Previous   |    Next