DISCUSSION AND SUMMARY

Scientific Results
Our principal objectives for this cruise were to establish the composition, age, and eruptive environment of the basement volcanic rocks at each site. We also hoped to determine the early subsidence history from the overlying sedimentary succession at each site and to ascertain the ages of sequence boundaries observed in the seismic record. Many of these objectives were partially met through shipboard studies but will require more comprehensive shore-based work to be achieved fully.

Basement Ages
Shipboard biotstratigraphic analysis brackets the age of basement at Sites 1183 and 1186 between the early and middle Aptian; thus, the upper levels of basement crust at these sites belong to the 122 (±3)-Ma phase of Ontong Java Plateau volcanism. At Site 1185, biostratigraphic age controls on basement are poor but suggest that basalt flows of two ages are present. The upper 15 m (and possibly ~125 m; i.e., the upper group of basalt units in Fig. 26) of lava flows appear to be between latest Cenomanian and Albian, possibly late Albian, in age; that is, micropaleontological data suggest they were emplaced between ~93 and 112 Ma (in the time scale of Gradstein et al., 1995). The lower 92 m of Kwaimbaita-type flows (the lower group in Fig. 26) appear to be late Aptian or older. The simplest interpretation, in the absence of radiometric age data, is that the lower flows belong to the widespread ~122-Ma event and the upper ones to the 90 (±4)-Ma event documented in lava flows at Site 803, on Santa Isabel and San Cristobal, and in ash layers at DSDP Site 288 (see Fig. 3). The nearly identical, and unusual, low-Ti, high-Mg compositions of the upper group of flows at Site 1185 and the entire lava sequence drilled at Site 1187 suggest that all the low-Ti, high-Mg flows are related and were formed at the same time. However, biostratigraphic data suggest that the Site 1187 flows are late Aptian or older (>115 Ma). Therefore, we have a discrepancy, the resolution of which must await ⁴⁰Ar-³⁹Ar dating of basalt samples and/or refinement of the biostratigraphic age estimates.

Nevertheless, the evidence now available from Leg 192, combined with age data for Legs 30 and 130 sites, indicates that an immense part of the high plateau was formed in the ~122-Ma event: the central region (Sites 289, 1183, and 1186), northern flank (Site 807), and probably much of the eastern flank (lower group of flows at Site 1185, possibly Site 1187). Basement crust in Malaita and much of Santa Isabel also was formed in this event, and possibly the lowest part of the section exposed in San Cristobal (Neal et al., 2000). With the resolution of existing sampling and ⁴⁰Ar-³⁹Ar and biostratigraphic ages, the duration of this event could have been as great as ~7 m.y., or much shorter. Importantly, the central region of the main plateau appears to have been largely bypassed by post-122-Ma eruptive episodes. Evidence for these episodes is found exclusively in locations around the eastern margins of the main plateau (Site 803, possibly Site 1187, and the upper group of basalt at Site 1185) and on the eastern salient (Santa Isabel, Malaita, San Cristobal, Site 1184, and in ash layers at Site 288, at the boundary between the salient and main plateau). Thus, the ~90-Ma episode now appears to have been volumetrically minor in relation to the ~122 Ma event and later episodes to have been still less important. This conclusion is one of the major results of Leg 192. In sharp contrast, in the southern and central Kerguelen Plateau and Broken Ridge, substantial volumes of magma appear to have been emplaced over a period of ~30 m.y. (Pringle and Duncan, 2000).

The thick sequence of volcaniclastic rocks at Site 1184 demonstrates that at least locally significant volcanism occurred on the northern part of the plateau's eastern lobe in the middle Eocene (~41-43 Ma in the time scale of Berggren et al., 1995). On the southern part of the eastern lobe, the 44-Ma alkalic lavas of the Maramasike Formation in Malaita (Tejada et al., 1996) reach a maximum thickness of 900 m (Petterson et al., 1997). The age of the Site 1184 volcaniclastic rocks and the Maramasike lavas is close to that of the ~43-Ma major change in Pacific plate motion (e.g., Duncan and Clague, 1985), and it is tempting to suggest that volcanism in one or both cases may have occurred in response to a change in the stress field of the plateau.

Petrology and Geochemistry of Igneous Rocks
Fundamental results of Leg 192 are that Kwaimbaita-type basalt was the only type we encountered at Sites 1183 and 1186 and that it lies below the 125 m of low-Ti, high-Mg basalt at Site 1185. Units C-G at Leg 130 Site 807 on the far northern flank of the high plateau are also of the Kwaimbaita type. Thus, Kwaimbaita-type lava flows cover an immense region of the high plateau. Furthermore, despite the considerable distances separating these sites, the total range of elemental variation is surprisingly small (Fig. 10, Fig. 11, Fig. 12, Fig. 13). Other magma types, whether ~122 Ma or younger, such as the low-Ti, high-Mg basalt discovered on Leg 192 and the Singgalo type at Site 807, in Santa Isabel, and particularly abundant in Malaita, appear to be present mainly on the margins of the plateau. No Singgalo-type basalt was recovered at any of the Leg 192 sites. We conclude that the Kwaimbaita magma type was by far the most abundant type produced, at least during construction of the upper levels of basement crust. Although shore-based isotopic work on Leg 192 basement samples is needed, the implication from the shipboard elemental data is that the magmas were derived from a very homogeneous and voluminous mantle source.

One of the most exciting discoveries of Leg 192 was the low-Ti, high-Mg basalt of Sites 1185 and 1187, along the eastern edge of the plateau. The Kwaimbaita magma type represents a high total percentage of partial melting, probably on the order of 18%-30% (Mahoney et al., 1993; Tejada et al., 1996; Neal et al., 1997). The low-Ti, high-Mg basalt requires significantly higher total amounts of partial melting. Small amounts of basalt with rather similar elemental characteristics are found along some oceanic spreading axes and in Iceland (Hemond et al., 1993; Hardarson and Fitton, 1997), but understanding how the apparently large volume of these flows on the eastern Ontong Java Plateau originated poses a formidable challenge. An explanation of their origin and of the relationship of their mantle source region to the Kwaimbaita and Singgalo sources awaits more precise knowledge of their age, and comprehensive elemental and isotopic data from shore-based studies.

By the Eocene, the plateau had drifted thousands of kilometers from its 90- and 122-Ma positions (e.g., Yan and Kroenke, 1993). Yet, surprisingly, the mantle source of the Eocene volcanism at Site 1184 appears to have been compositionally similar to the source of the Kwaimbaita-type ~122- and ~90-Ma basalts (e.g., Fig. 12). The same is true of the Paleocene and Eocene tholeiitic basalts in San Cristobal, which closely resemble the older basalt groups both elementally and isotopically (Birkhold-VanDyke et al., 1996; Neal et al., 2000). Although late stage tholeiitic magmatism could have been caused by upwelling of mantle completely unrelated to the source that formed most of the plateau, the compositional similarity of the later tholeiitic basalts and those formed many tens of millions of years earlier argues that the sources of the later basalts were rather closely related to the mantle that formed the bulk of the plateau. Our working hypothesis is that the sources of such late-stage tholeiitic volcanism resided in fertile portions of the plateau's lithospheric mantle root that did not melt, or possibly were veined (i.e., refertilized) by migrating melts, at 122 (and/or 90) Ma. Such regions would be capable of melting to form tholeiitic magma if heated sufficiently, but the specific causes of melting remain obscure at present.

Basement Paleolatitude
Estimates of basement paleolatitude at Sites 1183, 1185, 1186, and 1187 are in the 19°-25°S range, well to the north of the ~35°-42°S location suggested for the central high plateau between 125 and 90 Ma in the plate reconstruction of Neal et al. (1997) and even farther from the present position of the Louisville hot spot (~50°S) proposed by several workers to be the source of the plateau. Part of the difference may be explained by flexuring and faulting of the plateau's crust associated with postconstructional subsidence, although this possibility remains to be evaluated. Likewise, the amount of true polar wander and the distance the Louisville hot spot has drifted since the Early Cretaceous (if it existed then) are unknown. Presently, in agreement with Nd-Pb-Sr isotopic data for pre-Leg 192 basement samples and the lack of a postplateau seamount chain corresponding to a plume tail (e.g., Neal et al., 1997; Tejada et al., 2000), the paleolatitude data do not appear to support a Louisville hot-spot origin for the Ontong Java Plateau.

Site 1184 presents a paradox in that the -54° mean magnetic inclination of the volcaniclastic sequence implies a paleolatitude of ~35°S. Assuming the middle Eocene biostratigraphic age is accurate, this value is much farther south than the expected Eocene paleolatitude of ~15°-20°S for this part of the plateau. At this site, postdepositional tilting of the sequence appears unlikely to explain a difference of this magnitude, and we currently have no satisfactory explanation for the discrepancy.

Eruptive Environment and Paleoenvironmental Impact
The volcaniclastic sequence at Site 1184 formed in shallow water but represents magmatism that occurred long after the Aptian phase of plateau construction. Basement rocks at the four Leg 192 sites on the main plateau were emplaced well below sea level, as were those of Sites 289, 803, and 807, and the Ontong Java Plateau basement sections in the eastern Solomon Islands (see "Background" section above). Site 1183, on the crest of the main plateau, appears to have been by far the shallowest site originally (Fig. 37), but the virtually vesicle-free pillow lava flows there were probably erupted at a depth of at least several hundred meters. Indeed, the only evidence that any part of the main plateau was at least briefly shallow or emergent in the Aptian is provided by the two thin (<1 m recovered) layers of laminated vitric tuff deposited as turbidites at the base of the sedimentary sequence at Site 1183, a vitric tuff immediately above basement at Site 289 and, possibly, abundant glass shards in Aptian limestone at Site 288 (Andrews, Packham, et al., 1975). If a large proportion of Ontong Java Plateau magmas were erupted under shallow water or subaerially, then the flux of climate-modifying volatiles (particularly SO₂, Cl, and F) to the atmosphere would have been considerable (e.g., Michael, 1999). Because our results indicate that most of the plateau formed substantially below sea level, its large-scale environmental effects were probably quite limited. The magnitude of hydrothermal exchanges between seawater and the plateau's magmatic systems is unknown; all of the basement rocks recovered on Leg 192 have been affected only by low-temperature alteration processes. Overall, the types and amounts of alteration at all of the sites are similar to those found in normal (nonplateau) seafloor formed at spreading centers. The only evidence from Leg 192 sediments of relatively short-lived biological "deserts" that might be associated with the huge amount of Aptian volcanism on the plateau is in the thin, barren ferruginous claystone layers immediately above basement at Sites 1183 and 1187.