INTRODUCTION

Oxygen isotope data on benthic foraminifers from deep-sea cores have provided a useful history of Cenozoic climate change. Existing records exhibit minimum values during the early Eocene, consistent with an ice-free world with very warm deep-sea temperatures reaching ~12°C (Miller et al., 1987; Zachos et al., 2001) (Fig. F1). The early Eocene is inferred to be the warmest >1-m.y. interval of the past 70 m.y. and is generally considered a "greenhouse world." Based on a recent global compilation of deep-sea benthic foraminifer oxygen isotope records (Zachos et al., 2001), the long-term increase in Cenozoic ¹⁸O of ~3.2 must reflect some combination of deep-sea cooling and polar ice sheet growth. Because present-day ice sheets account for ~1 equivalent ¹⁸O, the remaining 2.2 is consistent with cooling of 9°–11°C over the course of the Cenozoic (Miller et al., 1987; Flower, 1999; Zachos et al., 2001). Three rapid increases in benthic ¹⁸O are generally inferred to include significant ice sheet growth, including increases near the Eocene/Oligocene boundary at ~33.5 Ma, during the middle Miocene at ~14 Ma, and during the late Pliocene at ~2.75 Ma.

The long-term increase in benthic ¹⁸O is interrupted by two major decreases during the late Oligocene at ~26 Ma and the early Miocene at ~17 Ma (Zachos et al., 2001). Compilation data from the late Oligocene suggest a >1 decrease in ¹⁸O following oxygen isotope event Oi2b (Miller et al., 1991) at ~26 Ma (Fig. F1). This decrease is attributed to a combination of deep-sea warming and Antarctic ice volume decrease (Miller et al., 1988, 1991; Zachos et al., 2001), leading to the peak of late Oligocene warmth. Significantly, data prior to the excursion come primarily from southern Indian Ocean Site 744 on Kerguelen Plateau (61°34.66´S, 80°35.46´E; water depth = 2307 m), whereas those afterward come from equatorial Atlantic Ocean Drilling Program (ODP) Site 929 on Ceara Rise (5°58.57´N, 43°44.4´W; water depth = 4356 m). Consequently, the magnitude and rapidity of the benthic ¹⁸O decrease is uncertain because few continuous high-resolution records with paleomagnetic age control exist for the late Oligocene.

Deconvolving the ice volume and temperature components of the benthic ¹⁸O signal is a continuing challenge in Cenozoic paleoceanography. Mg/Ca data on benthic foraminifers provide an independent means to track Cenozoic deep-sea temperature history (Lear et al., 2000, 2004; Martin et al., 2002; Billups and Schrag, 2002). Furthermore, the combination of Mg/Ca and ¹⁸O analyses in benthic foraminifers allows separation of temperature effects and isolation of the ¹⁸O composition of seawater (¹⁸O_sw). Existing Mg/Ca records exhibit a complex relationship to the benthic ¹⁸O compilation (Lear et al., 2000, 2004). Indeed, substantial mismatches are apparent in the late Oligocene and early Miocene. Deep-sea temperatures exhibit no mean increase during the late Oligocene that might parallel the benthic ¹⁸O decrease (Fig. F1), although they exhibit considerable variability of ~2°C (Lear et al., 2004). Accordingly, the late Oligocene benthic ¹⁸O decrease must mainly reflect ¹⁸O_sw decrease. Nevertheless, temperature variability of ~2°C could mask ~0.44 ¹⁸O_sw variability based on a recent calibration (Lynch-Stieglitz et al., 1999), reflecting some combination of high-frequency salinity and ice volume effects.

In this paper, we present high-resolution late Oligocene Cibicidoides mundulus stable isotope records from deep Pacific Hole 1237B, tied to the new geomagnetic timescale (GTS 2004) (Gradstein et al., 2004). These data define the magnitude and timing of the late Oligocene ¹⁸O decrease and associated ¹³C changes in the deep Pacific Ocean. Comparison of ¹³C records with those from the other major ocean basins addresses the role of Atlantic-Pacific deep circulation in this climate transition.