The most likely candidate for a paleoceanographic signal in the bulk isotopic data is the increase in 
18O values in the middle portion of the Maastrichtian. The shift occurs within the same relatively narrow stratigraphic interval (equivalent to ~69–70 Ma with the age model used) at all six sites despite the fact that burial depths range from <600 m to >1300 m among the sites analyzed.
The consistent stratigraphic position argues that the shift at least partially reflects conditions at the time of deposition. That is, it is not simply a diagenetic effect related to burial and separate from paleoceanography. However, in Hole 1183A the shift corresponds to the level at which the quality of preservation changes, and in Hole 1186A the shift corresponds to the level at which the foraminiferal lysocline was crossed (P. Sikora and J. Ogg, unpubl. data). Poorer preservation in older samples might introduce a bias at all six sites studied, but the preservational changes are not the same at all sites. Further, the shift is in the opposite direction from what would be expected if dissolution preferentially removed material precipitated in warm surface water and was, thus, most out of equilibrium with bottom water. That said, the correlation between preservational evidence for Maastrichtian changes in the depth of the lysocline and carbonate compensation depth on OJP (J. Ogg et al., unpubl. data; P. Sikora and J. Ogg, unpubl. data) and isotopic shifts is intriguing. Both may be separate but related effects caused by paleoceanographic changes during the Maastrichtian (e.g., Frank and Arthur, 1999; Barrera and Savin, 1999; MacLeod and Huber, 2001).
Potential primary causes for the shift in 18O values include temperature, salinity, and seawater isotopic composition. Large regional changes in salinity seem unlikely for these sites because they are within a very large ocean basin, and possible global changes in seawater 18O seem to be contradicted by decreasing 18O values observed in correlative samples from the North Atlantic (MacLeod et al., 2000; MacLeod and Huber, 2002). If the shift seen on OJP is the result of temperature change, the ~0.5
increase suggests cooling of 
2°C. In support of cooling, increases in 18O values across the Maastrichtian have been documented in a variety of samples from widely distributed localities and independent evidence for widespread cooling is provided by floral, faunal, and biogeographic evidence (see data and references in Frank and Arthur, 1999; Barrera and Savin, 1999; MacLeod and Huber, 2001). If accurate, the temperature interpretation expands the documented area that experienced Maastrichtian cooling to the southwestern tropical and subtropical Cretaceous Pacific. The magnitude of the shift is similar to those reported for good to well-preserved planktonic foraminifers from Shatsky Rise and the Mid-Pacific Mountains (Barrera et al., 1997; Barrera and Savin, 1999) located north and east of OJP and Manihiki Plateau.
That a Late Cretaceous temperature signal may have survived in all these sections is remarkable. The samples are obviously lithified, and average 18O values vary by >2 among the sites. In addition, the most deeply buried and most highly lithified samples (Hole 807C) have 18O values that yield paleotemperature estimates that are most reasonable for planktonic carbonate in a tropical greenhouse ocean. On the other hand, the least lithified samples (Hole 288A) have 18O ratios that yield improbably cool paleotemperature estimates for this setting. The compositional uniformity within and among samples could make it difficult for burial diagenesis to completely obscure depositional trends, but it does not explain why the qualitatively least altered samples result in the apparently worst paleotemperature estimates, and vice versa. Planktonic foraminifers are relatively abundant in the samples from Hole 288A and, if those foraminifers largely represent deeper dwelling taxa, it could explain part of the difference, but only if the evidence for alteration is largely ignored.
The explanation we favor is that the OJP data set provides an example among sites (rather than downcore in a single site) of changes in the direction of diagenetic effects on 18O values with increasing burial depth (e.g., Schrag et al., 1992, 1995). Assuming that (1) 18O values of the bulk carbonate were initially low at all sites (because the carbonate precipitated dominantly in warm surface waters), (2) samples from Hole 288A are the least geochemically altered, and (3) samples from Hole 807C are the most altered, then 18O values must have first been shifted to higher values (Hole 288A) and then shifted back to lower values (Holes 317A to 1183A to 1186A to 289 to 807C, progressively more lithified). Under this model, the initial shift toward higher 18O values would represent early alteration/recrystallization in bottom waters/near-bottom pore waters that were cooler than contemporary surface waters—a process that may explain the cool tropics paradox (D'Hondt and Arthur, 1996; Pearson et al., 2001). The shift back toward low 18O values would represent late alteration and reflect increasing temperatures with increasing burial. Because samples in Hole 288A were never as deeply buried as samples at other sites, they only experienced the former, whereas alteration in Hole 807C coincidentally has resulted in 18O values that approach reasonable Late Cretaceous levels. Finally, alteration related to hydrothermal circulation and proximity to basalt crust seems to affect the oldest samples from Holes 1183A, 1186A, and 289A (Fig. F3; Table T2).
Surprisingly, the 13C signal provides less paleoceanographic insight than the 18O record. Values of 3VPDB are reasonable for Late Cretaceous surface water, and the large number of samples in five (of six) holes with 13C 
3VPDB suggests it approximates a background value. As with 18O values, Hole 288A represents one end-member of the range of isotopic values observed. A relatively high abundance of deeper-dwelling thermocline taxa among the planktonic foraminifers in this hole coupled with changing taxonomic composition through the section could explain the difference, but this possibility has not been tested.
The large 13C excursion at or near the K/T boundary in Holes 289, 807C, 1183A, and 1186A (Fig. F3, F4) seems to be a diagenetic artifact. Studies on samples from Hole 807C suggested this feature might be primary and unusually well expressed at that site because of the expanded nature of that section (Corfield and Cartlidge, 1993; Corfield et al., 1993). However, the duration and magnitude of the excursion is quite unusual for the increasingly well documented K/T record, and the excursion is not consistently expressed among these six sites. Further, the limestones in the vicinity of the K/T boundary are partially silicified at several sites. Preferential cementation of the K/T boundary interval may explain why the boundary interval has been recovered more often in Pacific drilling than expected, given average recovery rates for the Upper Cretaceous and Paleogene. The 13C excursion in Holes 289, 807C, 1183A, and 1186A could be a manifestation of this selective alteration. This possibility seems more likely than unusual paleoenvironmental conditions during the K/T boundary interval in the tropical Pacific, especially since faunal patterns across the boundary at many Pacific sites seem consistent with the impact hypothesis.
