Abstract. To understand causal relationships in past climate variations, it is essential to have accurate chronologies of paleoclimate records. The last deglaciation, which occurred from 18 000 to 11 000 years ago, is especially interesting, since it is the most recent large climatic variation of global extent. Ice cores in Antarctica provide important paleoclimate proxies, such as regional temperature and global atmospheric CO₂. However, temperature is recorded in the ice while CO₂ is recorded in the enclosed air bubbles. The ages of the former and of the latter are different since air is trapped at 50–120 m below the surface. It is therefore necessary to correct for this air-ice shift to accurately infer the sequence of events.

Here we accurately determine the phasing between East Antarctic temperature and atmospheric CO₂ variations during the last deglacial warming based on Antarctic ice core records. We build a stack of East Antarctic temperature variations by averaging the records from 4 ice cores (EPICA Dome C, Dome Fuji, EPICA Dronning Maud Land and Talos Dome), all accurately synchronized by volcanic event matching. We place this stack onto the WAIS Divide WD2014 age scale by synchronizing EPICA Dome C and WAIS Divide using volcanic event matching, which allows comparison with the high resolution CO₂ record from WAIS Divide. Since WAIS Divide is a high accumulation site, its air age scale, which has previously been determined by firn modeling, is more robust. Finally, we assess the CO₂/Antarctic temperature phasing by determining four periods when their trends change abruptly.

We find that at the onset of the last deglaciation and at the onset of the Antarctic Cold Reversal (ACR) period CO₂ and Antarctic temperature are synchronous within a range of 210 years. Then CO₂ slightly leads by 165 ± 116 years at the end of the Antarctic Cold Reversal (ACR) period. Finally, Antarctic temperature significantly leads by 406 ± 200 years at the onset of the Holocene period. Our results further support the hypothesis of no convective zone at EPICA Dome C during the last deglaciation and the use of nitrogen-15 to infer the height of the diffusive zone. Future climate and carbon cycle modeling works should take into account this robust phasing constraint.