April 2005 LIP of the Month

Corresponds to events #19-20 in LIP record database

The Ontong Java Plateau

J. Godfrey Fitton
School of GeoSciences, University of Edinburgh, Grant Institute, West Mains Road, Edinburgh EH9 3JW, UK.
Godfrey.Fitton@ed.ac.uk

John J. Mahoney
School of Ocean and Earth Science and Technology, University of Hawaii, Honolulu, HI 96822, USA.
jmahoney@soest.hawaii.edu

Paul J. Wallace
Department of Geological Sciences, 1272 University of Oregon, Eugene, OR 97403-1272, USA.
pwallace@darkwing.uoregon.edu

Andrew D. Saunders
Department of Geology, University of Leicester, Leicester LE1 7RH, UK.
ads@leicester.ac.uk

Introduction

The submarine Ontong Java Plateau (OJP) is the largest of the world’s large igneous provinces (LIPs). It covers an area of about 2.0x106 km2 (larger than Alaska, and comparable in size with Western Europe), and OJP-related volcanism extends over a considerably larger area into the adjacent Nauru, East Mariana, and possibly Lyra and Pigafetta basins (Fig. 1). With a maximum thickness of crust beneath the plateau of 30–35 km (e.g. Gladczenko et al. 1997; Richardson et al. 2000; Miura et al. 2004), the volume of igneous rock forming the plateau and filling the adjacent basins could be as high as 6x107 km3 (e.g. Coffin & Eldholm 1994). It represents the largest igneous event on Earth in the past 200 Myr.

Seismic tomography experiments show a rheologically strong but seismically slow upper mantle root extending to about 300 km depth beneath the OJP (e.g., Richardson et al. 2000; Klosko et al. 2001). Gomer & Okal (2003) have measured the shear-wave attenuation in this root and found it to be low, implying that the slow seismic velocities must be due to a compositional, rather than thermal, anomaly in the mantle. The nature and origin of this compositional anomaly have not yet been established.

The OJP seems to have been formed rapidly at around 120 Ma (e.g. Mahoney et al. 1993; Tejada et al. 1996, 2002; Chambers et al. 2002; Parkinson et al. 2002), and the peak magma production rate may have exceeded that of the entire global mid-ocean ridge system at the time (e.g. Tarduno et al. 1991; Mahoney et al. 1993; Coffin & Eldholm 1994). Degassing from massive eruptions during the formation of the OJP could have increased the CO2 concentration in the atmosphere and oceans (Larson & Erba 1999) and led or at least contributed significantly to a world-wide oceanic anoxic event accompanied by a 90% reduction in nannofossil palaeoflux (Erba & Tremolada 2004).

Collision of the OJP with the old Solomon arc has resulted in uplift of the OJP’s southern margin to create on-land exposures of basaltic basement in the Solomon islands (Fig. 1), notably in Malaita, Santa Isabel, and San Cristobal (e.g. Petterson et al. 1999). In addition to these exposures, the basaltic basement on the OJP and surrounding Nauru and East Mariana basins has been sampled at ten Deep Sea Drilling Project (DSDP) and Ocean Drilling Program (ODP) drill sites (Figs. 1 and 2). The most recent of these, ODP Leg 192, was the first designed specifically to address the origin and evolution of the OJP (Mahoney et al. 2001). Pre-Leg 192 research on the OJP has been reviewed by Neal et al. (1997). The principal results of Leg 192 have been presented in a Special Publication of the Geological Society, London (Fitton et al. 2004). This publication complements the recent thematic set of papers on the origin and evolution of the Kerguelen Plateau, the world’s second largest oceanic LIP, published in Journal of Petrology (Wallace et al. 2002).

Relationship to the Louisville Hotspot

Several authors (e.g. Richards et al. 1991; Tarduno et al. 1991; Mahoney & Spencer, 1991) have favoured the starting-plume head of the Louisville hotspot (now at ~52†S) as the source of the OJP. Kroenke et al. (2004) have used a new model of Pacific absolute plate motion, based on the fixed hotspot frame of reference, to track the palaeogeographic positions of the OJP from its present location on the Equator back to 43†S at the time of its formation (~120 Ma). This inferred original position is 9† north of the present location of the Louisville hotspot, and suggests that this hotspot was not responsible for the formation of the OJP or, alternatively, that the hotspot has drifted significantly relative to the Earth’s spin axis (as the Hawaiian hotspot appears to have done; e.g. Tarduno et al. 2003). Antretter et al. (2004) have pointed out that the palaeomagnetic palaeolatitude of the OJP (~25†S) determined by Riisager et al. (2003a, 2004) further increases the discrepancy with the location of the Louisville hotspot.

Outcrops on the Solomon Islands

The thickest exposures of the OJP basement rocks in the Solomon Islands are found on the remote island of Malaita (Fig. 1). Petterson (2004) has presented the results of geological surveys that reveal a monotonous succession of Early Cretaceous tholeiitic pillow basalt, sheet flows, and sills (the Malaita Volcanic Group) 3-4 km thick. Rare and very thin interbeds composed of laminated pelagic chert or limestone suggest high eruption frequency and emplacement into deep water. The Malaita Volcanic Group is conformably overlain by a 1-2 km thick Cretaceous-Pliocene pelagic sedimentary cover sequence, punctuated by alkaline basalt volcanism during the Eocene and by intrusion of alnoite during the Oligocene. The alnoite intrusions contain an abundant and diverse suite of lower crustal and mantle xenoliths (e.g. Nixon & Neal, 1987). Thermobarometric analysis (Ishikawa et al., 2004) has shown that the xenolith suite represents a section through virtually the whole of the sub-OJP lithosphere. 

Age and biostratigraphy

The age and duration of OJP magmatism has not yet been established with any certainty. OJP basalts are difficult to date by the widely used 40Ar/39Ar method because of their very low potassium contents. Published 40Ar/39Ar data (Mahoney et al. 1993; Tejada et al. 1996, 2002) suggest a major episode of OJP volcanism at ~122 Ma and a minor episode at ~90 Ma. 40Ar/39Ar analysis (Chambers et al. 2002; L.M. Chambers, unpublished data) of samples from ODP Leg 192 Sites 1185, 1186 and 1187 (Fig. 1) gives ages ranging from 105 to 122 Ma. Chambers et al. (2002) suggest that their younger apparent ages (and, by implication, the data on which the 90 Ma episode is based) are the result of argon recoil and therefore represent minimum ages. Biostratigraphic dating based on foraminifera and nannofossils (Sikora & Bergen, 2004; Bergen, 2004) contained in sediment intercalated with lava flows at ODP Sites 1183, 1185, 1186 and 1187 suggests that magmatism on the high plateau extended from latest early Aptian on the plateau crest to late Aptian on the eastern edge. This corresponds to age ranges of 122–112 Ma (Harland et al. 1989) or 118–112 Ma (Gradstein et al. 1995). However, Re-Os isotopic data on basalt samples from these same four drill sites define a single isochron with an age of 121.5±1.7 Ma (Parkinson et al. 2002).

Petrology and geochemistry

The Malaita Volcanic Group (Petterson, 2004) has been divided by Tejada et al. (2002) into two chemically and isotopically distinct stratigraphic units: the Kwaimbaita Formation (>2.7 km thick) and the overlying Singgalo Formation (~750 m maximum exposed thickness). Basalt of the Kwaimbaita Formation was found to be compositionally similar to the basalt forming units C-G at ODP Site 807, on the northern flanks of the OJP (Fig. 1), whereas the Singgalo Formation is similar to the overlying unit A at Site 807. Thus, Kwaimbaita-type and Singgalo-type basalt flows with the same stratigraphic relationship are found at two sites 1500 km apart on the plateau (Tejada et al. 2002). A third basalt type, with higher MgO and lower concentrations of incompatible elements than any previously reported from the OJP, was recognised during ODP Leg 192 at Sites 1185 and 1187 on the eastern edge of the plateau (Mahoney et al. 2001). We use the term Kroenke-type for basalt flows with these characteristics because they were discovered on the flanks of the submarine Kroenke Canyon at Site 1185 (Fig. 1).

Tejada et al. (2004) used radiogenic-isotope (Sr, Nd, Pb, Hf) ratios to show that Kwaimbaita-type basalt is found at all but one of the OJP drill sites and therefore represents the dominant OJP magma type (Fig. 3). Singgalo-type basalt, on the other hand, appears to be volumetrically minor. Significantly, Kroenke-type basalt is isotopically identical to Kwaimbaita-type basalt (Tejada et al. 2004) and may therefore represent the parental magma for the bulk of the OJP. Age-corrected radiogenic-isotope ratios in Kroenke- and Kwaimbaita-type basalts show a remarkably small range (Fig. 3). Tejada et al. (2004) model the initial Sr-, Nd-, Pb- and Hf-isotope ratios in these two basalt types as representing originally primitive mantle that experienced a minor fractionation event (e.g. the extraction of a small amount of partial melt) at ~3 Ga or earlier. 

The remarkable homogeneity of OJP basalts is also seen in their major- and trace-element composition (Fitton & Godard 2004; Fig. 4). Fitton & Godard (2004) use geochemical data to model the mantle source composition and hence to estimate the degree of partial melting involved in the formation of the OJP. Incompatible-element abundances in the primary OJP magma can be modelled by around 30% melting of a peridotitic primitive mantle source from which about 1% by mass of average continental crust had previously been extracted (Fig. 5). The postulated depletion is consistent with the isotopic modelling of Tejada et al. (2004).

An independent estimate of the degree of melting is provided by Herzberg (2004), who uses a forward- and inverse-modelling approach based on peridotite phase equilibria. He obtains values of 27% and 30% for fractional and equilibrium melting, respectively. Further support for large-degree melting is provided by the platinum-group element (PGE) concentrations determined by Chazey & Neal (2004). The PGEs are highly compatible in mantle phases and sulphides, so their abundance is sensitive to degree of melting and sulphur saturation. Concentrations of PGEs in the OJP basalts are rather high, and consistent with around 30% melting of a peridotite source from which sulphide phases had been exhausted during the melting process. Some basalt samples have PGE abundances that are too high to be accounted for by a standard model peridotite source, and an additional source of PGEs may be needed. Chazey & Neal (2004) speculate that a small amount of material from the Earth’s core may have been involved in the generation of OJP magmas.

To produce a 30% melt of peridotite requires decompression of very hot (potential temperature >1500 °C) mantle beneath thin lithosphere. Thin lithosphere is consistent with the suggestion by Kroenke et al. (2004) that the OJP may have formed close to a recently abandoned spreading centre. Alternatively, lithospheric thinning could have resulted from thermal erosion caused by the upwelling of hot plume material. Ishikawa et al. (2005) have obtained Sm-Nd ages of ~160 Ma on spinel lherzolite and gabbro xenoliths collected from alnöite intrusions on Malaita, and these authors conclude that this represents the age of the upper oceanic lithosphere beneath this part of the OJP. If this interpretation is correct then at least the southern part of the OJP was erupted onto thick 40-Myr-old lithosphere, well away from a spreading centre. Ishikawa et al. (2005) also note that the closest known ~160 Ma crust is some 1800 km away, off the north-eastern margin of the plateau. This implies that either a fossil triple junction or large-offset transform lies buried beneath the OJP, and thus permits the possibility of much younger lithosphere somewhere to the north of Malaita.

Derivation of the dominant, evolved, Kwaimbaita magma type through fractional crystallisation of the primitive Kroenke-type magma is consistent with the isotopic (Tejada et al. 2004) and geochemical (Fitton & Godard 2004) evidence, and with melting experiments carried out by Sano & Yamashita (2004). Sano & Yamashita’s (2004) results show that the variations in phenocryst assemblage and whole-rock basalt major-element compositions can be modelled adequately by fractional crystallisation in shallow (<6 km) magma reservoirs.

The submarine emplacement of most of the OJP resulted in low-temperature alteration of the basalts through contact with seawater. The alteration ranges from slight to complete, and unaltered olivine and glass were found in some of the basaltic lava flows sampled in the drill cores. A detailed study of the alteration processes has been reported by Banerjee et al. (2004), who show that alteration started soon after emplacement and is indistinguishable from that affecting abyssal basalt. There is no evidence for high-temperature alteration in any of the basalt recovered from the OJP.

Volatile content and water depth of emplacement of OJP magmas

Glass from the rims of basaltic pillows was recovered from most drill sites on the OJP, and this glass preserves a record of the volatile content of the magmas at the time of eruption. Roberge et al. (2004) have shown that water contents in the glasses are uniformly low (Fig. 6) and imply water contents in the mantle source that are comparable with those in the source of mid-ocean ridge basalt. This is an important observation because it shows that the large degrees of melting estimated for the OJP magmas cannot have been caused by the presence of water but require high temperatures. The sulphur contents of OJP glasses confirm Chazey & Neal’s (2004) inference of sulphur-undersaturation in the magmas.

The water depth of lava emplacement controls the CO2 content of the glasses, and data obtained by Roberge et al. (2004) imply depths ranging from about 1000 m on the crest of the OJP to about 2500 m on its eastern edge (Fig. 6). From these data Roberge et al. (2005) estimated maximum initial uplift for the plateau of 2500-3600 m above the surrounding seafloor, values that are considerably smaller than the ~4000-6000 m expected from dymamic uplift due to a mantle plume coupled with the isostatic effects of increased crustal thickness. Roberge et al. (2005) also used their data to show that the OJP has subsided by ~1500 m since its emplacement. This value is significantly less than the 2700-4100 m expected from thermal subsidence. The discrepancies in the amount of uplift and subsidence have not yet been explained.

The total amount of CO2 released during formation of the OJP is difficult to determine without reliable information on primary magmatic CO2 contents and precise knowledge of the duration of volcanism, but Roberge et al. (2004) calculated a maximum value that is around ten times the flux from the global mid-ocean ridge system. Erba & Tremolada (2004) estimated that the 90% reduction in nannofosil palaeofluxes that they link to emplacement of the OJP requires a three- to six-fold increase in volcanogenic CO2.

Subaerial eruptions on the eastern salient of the Ontong Java Plateau

One of the most exciting discoveries of ODP Leg 192 was a thick succession of basaltic volcaniclastic rocks at Site 1184 on the eastern salient of the OJP (Fig. 1). Drilling at this site penetrated 337.7 m of tuff and lapilli tuff, before the site had to be abandoned through lack of time. A detailed volcanological study by Thordarson (2004) concluded that the volcaniclastic succession was the result of large phreatomagmatic eruptions in a subaerial setting. This setting contrasts strikingly with that of the lava flows sampled on the main plateau and in the Solomons, which were all erupted under deep water (Roberge et al. 2004; Petterson 2004). Thordarson (2004) divided the succession into six subunits or members, each representing a single massive eruptive event. Fossilised or carbonised wood fragments were found near the bottom of four of the eruptive members (Mahoney et al. 2001). The volcaniclastic succession at Site 1184 provides the only evidence so far for significant amounts of subaerial volcanism on the OJP.

Three of the six eruptive members at Site 1184 contain blocky glass clasts with unaltered cores, and these cores allow the reliable determination of the composition of the erupted magma. White et al. (2004) used microbeam techniques to determine the major- and trace-element compositions of samples of the glass. The glasses are very similar in composition to the Kwaimbaita-type and Kroenke-type basalts sampled on the high plateau. Each member has a distinct glass composition and there is no intermixing of glass compositions between them, confirming Thordarson’s (2004) conclusion that each is the result of one eruptive phase, and that the volcaniclastic sequence has not been reworked. White et al.’s (2004) major and trace element data for the glass clasts suggest that the voluminous subaerially erupted volcaniclastic rocks at Site 1184 belong to the same magmatic event as that responsible for the construction of the main plateau. Thus the OJP would have been responsible for volatile fluxes into the atmosphere in addition to chemical fluxes into the oceans. Both factors may have influenced the contemporaneous oceanic anoxic event (Sikora & Bergen 2004; Erba & Tremolada 2004).

The geochemical evidence (White et al. 2004; Fitton & Godard 2004) linking the phreatomagmatic eruptions recorded at Site 1184 to the formation of the main plateau is supported by the Early Cretaceous age implied by the steep (~54†) magnetic inclination preserved in the volcaniclastic rocks (Riisager et al. 2004). However, this evidence appears to be contradicted by the presence of rare Eocene nannofossils at several levels within the succession (Bergen 2004). In an attempt to resolve this paradox, Chambers et al. (2004) applied the 40Ar/39Ar dating method to feldspathic material separated from two basaltic clasts, and to individual plagioclase crystals separated from the matrix of the volcaniclastic rocks. The clasts gave minimum age estimates of ~74 Ma, and the plagioclase crystals a mean value of 123.5±1.8 (1s) Ma. Thordarson (2004) and Chambers et al. (2004) suggest that the Eocene nannofossils were introduced later, possibly along fractures.

A mantle plume origin for the Ontong Java Plateau?

Testing the plume-head hypothesis for the formation of giant ocean plateaus was one of the principal objectives of ODP Leg 192, and many of the results are consistent with such an origin. The discovery of high-MgO Kroenke-type basalt allows us to calculate the composition of the primary magma and hence deduce the nature of the mantle source and the degree of melting. Isotopic (Tejada et al. 2004) and chemical (Fitton & Godard 2004; Chazey & Neal 2004) data are consistent with a mildly depleted peridotite mantle source, and phase-equilibria (Herzberg 2004) and trace-element (Fitton & Godard 2004; Chazey & Neal 2004) modelling independently constrain the degree of melting of this peridotite source to around 30%. Melting to this extent can only be achieved by decompression of hot (potential temperature >1500†C) peridotite beneath thin lithosphere. To achieve an average of 30% melting requires that the mantle was actively and rapidly fed into the melt zone, and a start-up mantle plume provides the most obvious mechanism. A plume-head impinging on thin lithosphere theoretically should have caused uplift of a sizeable part of the plateau above sea level, as in Iceland, and, indeed, at least part of the eastern salient was emergent (Thordarson 2004). However, the abundance of essentially non-vesicular submarine lava and the absence of any basalt showing signs of subaerial weathering show that all the other sampled portions of the OJP were emplaced below sea level (e.g. Neal et al. 1997; Mahoney et al. 2001). Volatile concentrations in quenched pillow-rim glasses suggest eruption depths ranging from 1100 m at Site 1183 to 2570 m at Site 1187 (Roberge et al. 2004).

We have not yet been able to resolve the paradox of apparent high mantle potential temperature (Fitton & Godard, 2004; Herzberg, 2004) coupled with predominantly submarine emplacement, and this is difficult to reconcile with a mantle-plume origin for the OJP. Fitton & Godard (2004) discussed three alternatives to a plume origin for the OJP: (1) meteorite impact, (2) a mantle source more fertile than peridotite (e.g. eclogite), and (3) hydrous mantle.

1. Widespread melting of the mantle following the impact of an asteroid or comet has been suggested to provide a means of avoiding uplift (e.g. Ingle & Coffin, 2004), but this has been disputed (Korenaga 2005). Ishikawa et al. (2005) note that the presence of ~160 Ma oceanic lithosphere beneath Malaita appears to rule out an impact site anywhere in the southern region of the plateau, although a site farther north is possible. In any case the magma resulting from impact would be generated entirely within the upper mantle and should normally be expected to have the chemical and isotopic characteristics of Pacific mid-ocean ridge basalt. OJP basalt is isotopically (Tejada et al. 2004) and chemically (Fitton & Godard 2004) distinct from Pacific mid-ocean ridge basalt. Furthermore, no mass extinction occurred at the time of OJP formation, even though the required impactor would have had a diameter (estimated at 20 km by Ingle & Coffin (2000)) significantly greater than that thought to have been responsible for the extinctions at the Cretaceous-Tertiary boundary. Further arguments against an impact origin for the OJP have been advanced by Tejada et al. (2004).
2. A composite mantle source composed of peridotite and eclogite could in principle offer a solution to the OJP paradox of high melt productivity without significant uplift because it allows large degrees of melting in mantle with modest potential temperature. However, this can only work efficiently if isolated pods of eclogite can melt independently of the more refractory peridotite host (Cordery et al. 1997). We can rule out a pure eclogite source for the OJP magmas because eclogite melts to a liquid with higher SiO2 and lower MgO than the parental Kroenke-type magma (Yasuda et al. 1997; Takahashi et al. 1998; Yaxley & Green 1998). Tejada et al. (2002) modelled the major- and incompatible trace-element composition of OJP basalt by mixing large-degree melts from eclogite with small-degree peridotite melts but found that melting of peridotite alone (Mahoney 1993; Tejada et al. 1996; Neal et al. 1997) fits the data better. Similarly, Figure 5 shows a very good fit between the estimated composition of the primary OJP magma and that of a 30% equilibrium melt of a peridotite source. The remarkable agreement in degrees of melting calculated independently through phase equilibria (Herzberg 2004) and trace element modelling provides compelling evidence in favour of a peridotite source. Both approaches lead to values of ~30% melting for the OJP primary magma. An enriched mantle source composed of peridotite + eclogite (e.g. Yaxley 2000) could produce voluminous magma at lower Tp, but such a source would inevitably have higher concentrations of incompatible elements. It would therefore require correspondingly higher degrees of melting to produce magma with the low concentrations of incompatible elements that are found in Kroenke-type basalt. Any temperature advantage gained through the more efficient melting of an enriched source would be lost by the need for a higher Tp to produce larger-degree melts. Melting needs latent heat, and this can only be supplied by a high-Tp mantle source.
3. Hydrous mantle provides an equally unsatisfactory solution to the paradox because studies on OJP basaltic glass (Michael 1999; Roberge et al. 2004) show that the water content of the magmas was very low. Small amounts of water in the mantle would, in any case, cause the formation of small melt fractions at greater depth than with anhydrous mantle and therefore decrease the average melt fraction rather than raise it (Asimow & Langmuir 2003).

Concluding remarks

As a result of drilling during ODP Leg 192, we now have a much clearer view of the range and distribution of basalt types on the plateau, and we have identified a potential parental magma composition represented by Kroenke-type basalt. The age and duration of magmatism are still uncertain because we have still only scratched the surface of the 30- to 35-km-thick OJP crust. However, it now seems plausible that almost the entire plateau formed in a single, widespread magmatic event at ~120 Ma. The identification of a thick succession of volcaniclastic rocks at Site 1184 shows that at least part of the plateau was erupted in a subaerial environment, though most of it was erupted under deep water. We conclude that the start-up plume hypothesis appears to fit more of the observations than do any of the alternative hypotheses, but the lack of uplift of the magnitude predicted by the plume hypothesis and the lack of an obvious hotspot track remain to be explained.

References