Fig. 3: Changes in atmospheric CO₂ and global mean annual surface air temperature (GMT) after AMOC collapse down under various baseline CO₂ levels.

(a, b) display the atmospheric CO₂ concentration and GMT response to freshwater hosing over time in the fully coupled model setup. (c) Illustrates the difference in atmospheric CO₂ concentrations between equilibrium states before and after AMOC collapse (\(\Delta {{{{\rm{CO}}}}}_{2}^{AMOC}\)) for the fully coupled and land-carbon-off model setups. \(\Delta {{{{\rm{CO}}}}}_{2}^{AMOC}\) is computed using the average over the final 500 years (6500–7000) minus the average over the first 500 years (0–500) of the simulation under each CO₂ baseline. e Presents the corresponding ratio of \(\Delta {{{{\rm{CO}}}}}_{2}^{AMOC}\) relative to the baseline CO₂ (pre-hosing atmospheric CO₂ levels) for both the fully coupled and land-carbon-off model setups. In the fixed CO₂ model setup, atmospheric CO₂ concentrations remain constant throughout the entire simulation. d Illustrates the additional GMT changes due to AMOC collapse (ΔGMT_AMOC) as a function of ΔGMT_baseline, where ΔGMT_baseline represents the difference between the initial GMT under each baseline CO₂ level and the pre-industrial GMT. Specifically, ΔGMT_baseline is calculated using the first 500-year average GMT under each baseline CO₂ level minus the first 500-year average pre-industrial GMT (Supplementary Fig. 4a). Panel (f) presents the ratio of ΔGMT_AMOC to ΔGMT_baseline for the fully coupled, land-carbon-off, and fixed-CO₂ model setups. All results in are based on experiments utilizing 0.2 Sv hosing to assess responses following AMOC collapse.