Fig. 4: Meridional migrations of the ACC over the last glacial cycle and their connections to climate and CO2 change.

a Atmospheric CO272 (black) and EPICA Dome C (EDC) temperature change48 (gray). b Offset between CO2 and predicted CO2 based on a linear relationship between Antarctic ice core temperature and CO212 (dark red). c Compilation of SST at 50°–60°S52 (pink) and global SST change66 (purple). d Changes in ACC latitude calculated from ΔSST (3357−3353) (thick blue solid and dashed lines) and the uncertainty envelope (lighter shading). The dashed line represents where the measured ΔSST corresponds to an uncertainty range of ACC latitude change and the ACC latitude change is estimated as the average of the upper bound and lower bound (see “Methods”). e Changes in the difference between air temperatures at the moisture source (Tsource) and ice core site (Tsite) of Vostok (dark gray) relative to modern conditions reconstructed from ice deuterium excess73, and obliquity81 (yellow). f Offset between Indian AZ diatom-bound δ15N and predicted diatom-bound δ15N based on a linear relationship between Antarctic air temperature and diatom-bound δ15N12 (brown), g Combined record of Indian AZ diatom-bound δ15N as an indicator of upwelling12 (dark green) and predicted diatom-bound δ15N based on a linear relationship between Antarctic air temperature and diatom-bound δ15N12. The numbers at the top indicate Marine Isotope Stages. The vertical gray bars indicate different cooling periods of low obliquity (111.5–116.5 ka in MIS 5d) or lowering obliquity (1.5–6.5 ka in the late Holocene) during warmer conditions, low obliquity during glacial conditions (31.5–36.5 ka in MIS 3), and high obliquity during intermediate conditions (89–94 ka in MIS 5b).