Extended Data Fig. 3: Comparison of δ18O generated as part of this study with equivalent published data and Δ47-based temperature estimates derived using different Δ47-temperature calibrations.
From: Relatively warm deep-water formation persisted in the Last Glacial Maximum
a. Kernel density estimate plots showing the distribution of δ18O obtained from two independent analyses of benthic foraminifera from the same sediment samples for the mid-to-late Holocene and LGM (Δ47, this study, blue; stable isotope analysis as per ref. 14, teal). p-values from Welch’s t-tests indicate no significant difference between the two datasets (in both cases p > 0.05), confirming that the analyses yield indistinguishable δ18O. Close agreement between the published data and published δ18O data from around the Northwest Atlantic suggests that the δ18O measured in this study were unaffected by contamination, e.g., from clays. b and c. Comparison of mid-to-late temperature estimates derived using different published Δ47-temperature calibrations: M21 (equation (1))68, DG23 (equation (5))103, A21 (equation (1))104, and DV24 (equation (28))105. Dashed lines in b indicate the approximate modern in-situ temperature at each core site, based on GLODAP 2022 data47. The M21 calibration was selected for its strong statistical agreement with modern deep ocean temperatures and methodological consistency, despite being derived, in part, from planktic foraminiferal Δ47 data. This calibration was therefore applied to glacial samples as well.
