Extended Data Fig. 7 | Predictions of RSL at three far-f ield sites (the
Seychelles, Western Australia and Mallorca) and one intermediate-field site
(the Bahamas). a–d, RSL predictions for the Bahamas from the full suite of
simulations that bound from above all of the coral data with the exception of
the earliest datum (at ~131 ka) for the COL ice history (a), the LAM ice history (b),
the HYB ice history (c) and the WAE ice history (d). e–h, RSL predictions for the
Seychelles from the full suite of simulations that lie above the three coral
records with an elevation of ~4 m for the COL ice history (e), the LAM ice history
(f), the HYB ice history (g) and the WAE ice history (h). i–l, RSL predictions for
western Australia from the full suite of simulations that bound from above all of
the coral data for the COL ice history (i), the LAM ice history (j), the HYB ice
history (k) and the WAE ice history (l). The age uncertainty is 2σ, and the depth
uncertainty ref lects the uncertainty in habitat depth. m–p, RSL predictions for
Mallorca from the full suite of simulations that fit the data within 50% of the
minimum misfit achieved for all simulations for the COL ice history (m), the
LAM ice history (n), the HYB ice history (o) and the WAE ice history (p). The age
uncertainty is 2σ, and the depth uncertainty ref lects the uncertainty in
speleothem water depth.