(Table S1) Stable oxygen isotope record of barite from DSDP Leg 85 sites

Oxygen isotopes in marine sulfate (d18O SO4) measured in marine barite show variability over the past 10 million years, including a 5per mil decrease during the Plio-Pleistocene, with near-constant values during the Miocene that are slightly enriched over the modern ocean. A numerical model suggests that sea level fluctuations during Plio-Pleistocene glacial cycles affected the sulfur cycle by reducing the area of continental shelves and increasing the oxidative weathering of pyrite. The data also require that sulfate concentrations were 10 to 20% lower in the late Miocene than today.

Biogenic sedimentation of DSDP Site 85-574

The equatorial Pacific is an important part of the global carbon cycle and has been affected by climate change through the Cenozoic (65 Ma to present). We present a Miocene (12-24 Ma) biogenic sediment record from Deep Sea Drilling Project (DSDP) Site 574 and show that a CaCO3 minimum at 17 Ma was caused by elevated CaCO3 dissolution. When Pacific Plate motion carried Site 574 under the equator at about 16.2 Ma, there is a minor increase in biogenic deposition associated with passing under the equatorial upwelling zone. The burial rates of the primary productivity proxies biogenic silica (bio-SiO2) and biogenic barium (bio-Ba) increase, but biogenic CaCO3 decreases. The carbonate minimum is at ~17 Ma coincident with the beginning of the Miocene climate optimum; the transient lasts from 18 to 15 Ma. Bio-SiO2 and bio-Ba are positively correlated and increase as the equator was approached. Corg is poorly preserved, and is strongly affected by changing carbonate burial. Terrestrial 232Th deposition, a proxy for aeolian dust, increases only after the Site 574 equator crossing. Since surface production of bio-SiO2, bio-Ba, and CaCO3 correlate in the modern equatorial Pacific, the decreased CaCO3 burial rate during the Site 574 equator crossing is driven by elevated CaCO3 dissolution, representing elevated ocean carbon storage and elevated atmospheric CO2. The length of the 17 Ma CaCO3 dissolution transient requires interaction with a 'slow' part of the carbon cycle, perhaps elevated mantle degassing associated with the early stages of Columbia River Basalt emplacement.