Fig. 1. Map of the southeastern Atlantic Ocean, showing the location of Site 1090 (6°2′S, 8°E). The locations of Site 744 (61°S, 82°E) on Kerguelen Plateau and Site 689 (64°S, 3°E) on Maud Rise are also shown. The approximate positions of the present-day Subantarctic Front and the Antarctic Polar Front are shown.

Fig. 2. Carbonate content of 1090 sediments based on Ca concentrations, the solid circles represent age control points based on oxygen isotope and magnetostratigraphy (Billups et al., submitted), the solid triangles represent age control points based on nannofossil datums ( Marino and Flores, in press). The Eocene/Oligocene transition is shaded in gray. Several intervals of extremely low carbonate content are observed during the Oligocene; however, the dissolution event from 250 to 254 mcd occurs in the earliest Oligocene and can be correlated to dissolution events observed at other sites in the Southern Ocean. Galeotti et al. (in press) place the Eocene/Oligocene boundary at 270 mcd based on planktic foraminifera. We adopt the nannofossil-based biostratigraphy of Marino and Flores (in press) for this study.

Fig. 3. Seven point smoothed percent terrigenous determined by Al normalization plotted against age. Gray shaded area represents the Eocene/Oligocene transition. The record is more variable during the Eocene and becomes more cyclic in the Oligocene and Miocene. Values exceeding 100% are artifact of the normalization process, meaning the end-member chosen for normalization purposes does not necessarily represent the actual end-member source.

Fig. 4. Seven point smoothed Al/Ti plotted versus age. The Al/Ti ratio is highly variable during the middle Eocene with maximum values higher than lithogenic sources. Granites have the highest Al/Ti ratio of greater than 40, which is exceeded during the middle Eocene. Late Eocene Al/Ti ratios are similar to those found in average continental crust (15.8) or average upper crust (27) suggesting a continental source (Taylor and McLennan, 1985). Values from about the Eocene/Oligocene boundary to 16 Ma are generally lower (Al/Ti=10) and more similar to oceanic sources, such as oceanic crust or basalt ( Taylor and McLennan, 1985).

Fig. 5. Scatter plots of P and Ba concentrations versus Ti concentrations. A positive correlation with Ti suggests P and Ba are at least partially associated with terrigenous material.

Fig. 6. Seven point smoothed P concentrations (top), P/Al ratios (middle) and P/Ti ratios (bottom) plotted versus age. Because P is positively correlated with Ti, a terrigenously derived metal, we use P/Al and P/Ti ratios as proxies for export production rather than P concentrations. Most concentrations range from 20 to 65 μmol P g⁻¹, with elevated concentrations in the late early Miocene. Elevated P/Al and P/Ti ratios are interpreted as increases in export production. The most notable increases occur during the middle Eocene (44–42 Ma) and across the Eocene/Oligocene transition (shaded area).

Fig. 7. Seven point smoothed Ba concentrations (top) Ba/Al ratios (middle) and Ba/Ti ratios (bottom) plotted versus age. Ba concentrations undergo a dramatic and permanent change at 32.8 Ma. Prior to 32.8 Ma, Ba concentrations are low, averaging 5 μmol g⁻¹, and after 32.8 Ma Ba concentrations are higher and more variable, averaging 21 μmol g⁻¹. Ba/Al and Ba/Ti ratios exhibit similar peaks to those seen in P/Al and P/Ti records (Fig. 6), with maxima from 44 to 42 Ma and across the Eocene/Oligocene transition, which we interpret as increases in export production.

Fig. 8. Fe (solid triangles) and P (open circles) concentrations plotted against age. Salamy and Zachos (1999) suggested eolian Fe-fertilization increased productivity in the earliest Oligocene. From 34 to 32.8 Ma, Fe and P concentrations appear positively correlated, with peaks in both concentrations coincident; however, the exact relationship between Fe and productivity at this time is unknown.