SCIENTIFIC OBJECTIVES

Hypotheses that involve mantle plumes in the formation of large igneous provinces include rapid-eruption, age-progressive, and episodic growth models.

1. Rapid-emplacement models are of two main types. The plume-head or plume-impact model (e.g., Richards et al., 1991; Saunders et al., 1992; Campbell, 1998) predicts that large oceanic plateaus are formed by widespread basaltic flood eruptions as the inflated head of a rising new mantle plume approaches the base of the lithosphere. An alternative model devised specifically for continental and continental-margin flood basalt provinces (White and McKenzie, 1989), sometimes called the plume-incubation model (e.g., Saunders et al., 1992), considers flood volcanism to result from cataclysmic pressure-release melting when a rift propagates above the enlarged top of a more or less steady-state mantle plume that has accumulated gradually (perhaps over several tens of millions of years) beneath thick, slow-moving continental lithosphere. Both types of model predict that the great bulk of magmatism occurs in only a few (probably <5) million years.
2. In contrast, an age-progressive, Icelandic style of construction in which oceanic plateaus are formed over much longer intervals (tens of millions of years) remains a distinct possibility for many plateaus (e.g., Mahoney and Spencer, 1991; Coffin and Gahagan, 1995; Ito and Clift, 1998).
3. Alternatively, plateau growth may occur in two or more discrete pulses of activity, dependent on mantle plume dynamics or the interplay between episodes of lithospheric extension and mantle melting (Bercovici and Mahoney, 1994; Larson and Kincaid, 1996; Neal et al., 2000; Ito and Taira, 2000).

As the world's largest oceanic plateau, the Ontong Java Plateau provides an important test case. Its great crustal volume implies partial melting of at least 1.5-4.0 × 10⁸ km³ of mantle, which virtually necessitates involvement of the lower mantle if the bulk of the plateau was formed in the 122-Ma event (e.g., Coffin and Eldholm, 1994). Melting on such a scale is not happening in the Earth's mantle today, and this consideration has helped fuel suggestions of fundamental differences between Cretaceous and Cenozoic mantle convection. However, if the plateau accreted more slowly over several tens of millions of years (like the much smaller Iceland Plateau) or in two or more discrete pulses, then this partial melting requirement is eased considerably, but simple plume-head models either would need to be modified significantly or would not apply. The few basement locations sampled before Leg 192 demonstrated that both 122- and 90-Ma lava flows are present in widely separated areas, but the importance of the 90-Ma episode remained unclear. In many places, 90-Ma lava flows may simply form a relatively thin carapace over a thick 122-Ma volcanic pile.

Laboratory and numerical modeling suggests that starting-plume heads should be strongly zoned because of entrainment of large amounts of ambient, nonplume mantle during their rise to the base of the lithosphere (e.g., Campbell, 1998). Thus, even if a plume's source region (usually assumed to be at the base of, or deep within, the mantle) is compositionally homogeneous, significant isotopic and trace element variability is nevertheless predicted in magmas erupted from different parts of the plume head or at different times. Major element compositions are predicted to vary as well, with magmas erupted above the hottest (axial) parts of the plume head having picritic (e.g., Campbell, 1998) or possibly even komatiitic (Storey et al., 1991) affinities, and more ordinary basaltic magma predicted above cooler, more distal regions. In this regard, the two most remarkable features of the Ontong Java Plateau basement samples available before Leg 192 were (1) the limited overall range of chemical and isotopic variation in the 122-Ma lava flows and (2) that the 90- and 122-Ma flows are so chemically and isotopically similar to each other. The isotopic and incompatible element results could indicate that the world's largest plateau had a much more homogeneous source (both relative to the scale of melting and in time) than predicted by plume-head models. Furthermore, the combined major and trace element data imply that storage and homogenization in large reservoirs was a dominant process in the evolution of the magmas.

Plume-head models predict as much as 1-3 km of dynamic uplift associated with the arrival of a large starting-plume head at the base of the lithosphere (e.g., Hill, 1991; Neal et al., 1997). The associated constructional volcanism also creates a much thicker crust than normal oceanic crust. The combination of these effects is predicted to elevate parts of a plateau's surface to shallow water depths or even cause portions to emerge above sea level. Indeed, significant portions of several plateaus are known to have been initially shallow or subaerial (e.g., Richards et al., 1991; Coffin, Frey, Wallace, et al., 2000). However, although most of the Ontong Java Plateau stands 2-3 km above the surrounding seafloor today, basement lava flows from all locations studied before Leg 192 were emplaced beneath fairly deep water, probably below the calcite compensation depth (CCD) in some cases (Neal et al., 1997; Ito and Clift, 1998; Michael, 1999). The reasons for this behavior and whether it is typical of the plateau as a whole were unknown, but critical for understanding how plateaus are constructed and for testing the plume-head model. Moreover, whether or not parts of the Ontong Java Plateau were shallow at the time of volcanism has important implications for how its emplacement affected large-scale climatic, oceanographic, and biospheric conditions. If significant amounts of magma were erupted in shallow water or subaerially, the flux of climate-modifying volatile species (particularly SO₂, Cl, and F) to the atmosphere would have been much greater than if the bulk of plateau volcanism occurred at greater water depths.

The physical volcanology of large-scale submarine lava flows is poorly known, as are the nature and scale of hydrothermal fluid fluxes associated with plateau magmatism. Knowledge of the physical volcanology of lava flows is important for understanding eruption mechanisms and how the volcanic pile accumulated, whereas data on hydrothermal activity are critical for understanding the environmental effects of plateau formation. Flows making up continental flood basalt provinces are typically 10-30 m thick and, in some cases, have been traced for distances of several hundred kilometers (e.g., Hooper, 1997). Areas distant from eruptive sources tend to be made up of simple flows, whereas compound flows are more indicative of relative proximity to eruptive vents. Previous sampling of the Ontong Java Plateau basement revealed that flow thickness varies from <1 m to 60 m, although most of the flows are in the 4- to 12-m range (e.g., Neal et al., 1997). The flows are dominantly simple, consistent with the locations of most pre-Leg 192 basement sites at the margins of the plateau and presumably far from their eruptive vents. Very little interlava ash has been found. With regard to hydrothermal activity, pre-Leg 192 Ontong Java basement sites show almost no evidence of anything but low-temperature seawater-mediated alteration in either the lava flows or overlying sediments (e.g., Babbs, 1997). This lack of higher-temperature hydrothermal alteration again is consistent with the inferred distance from eruptive vent systems. Major hydrothermal systems would be expected to be centered around major eruptive loci, postulated by Tejada et al. (1996) to be the comparatively shallow regions of the high plateau and eastern lobe.