Fig. 2: Deep-sea temperatures and δ¹⁸O seawater composition across ETM2–H2.

a δ¹⁸O_b values derived from N. truempyi and O. umbonatus that are corrected towards Cibicidoides (assumed seawater equilibrium). b Deep-sea temperatures based on Δ₄₇ and δ¹⁸O_b including the number of Δ₄₇ (and δ¹⁸O_b) measurements that are contained in the background, slope, and hyperthermal peak temperature bins. The yellow (background), red (slope), and purple (hyperthermal peak) bars indicate the range in δ¹⁸O_b values corresponding to the Δ₄₇ measurements that were used for compiling the three bins. The dark grey error bars on the Δ₄₇-based temperatures represent fully propagated analytical and calibration uncertainties as 68% (solid) and 95% (dashed) CI (Methods). The dashed red envelope on the δ¹⁸O_b-based deep-sea temperatures displays the uncertainty (95% CI) of the δ¹⁸O_b-temperature calibration of Marchitto et al.⁷⁷. Analytical uncertainty on the δ¹⁸O_b-based temperatures of the three bins is very small (±0.1 °C 95% CI) and is therefore contained within the data symbols (Methods). c δ¹⁸O_sw (in ‰ VSMOW) of the background, slope, and hyperthermal peak bins including uncertainties as 68% (solid) and 95% (dashed) CI. Values of δ¹⁸O_sw were calculated using the mean Δ₄₇-based temperatures in combination with the average δ¹⁸O_b values of the three bins in the δ¹⁸O_b-temperature relationship of Marchitto et al.⁷⁷ (Methods). The horizontal red and blue dashed lines show the assumed δ¹⁸O_sw value based on ice-free conditions and the present-day δ¹⁸O_sw value, respectively.