Please see supplement

Review of manuscript “Multi-model assessment of the deglacial climatic evolution at southern high latitudes” by Obase et al.

In the manuscript, the authors used six transient simulations for the last deglaciation to assess similarities and differences of their Southern Ocean/Antarctic changes, e.g. temperature, responses to AMOC and CO2, etc. The main issue is that different models have very different forcing histories (especially the freshwater schemes), which undermines the foundation of a MIP work. Under this context, the two simple models seems to a key to improving comparability of multi-model outputs, which I do acknowledge. Nevertheless, I did not see enough substantial improvements for our understanding of dynamics of Southern Ocean and roles of AMOC and CO2 in Southern Ocean. It seems that the experimental designs of HadCM3, MPIESM and iLOVECLIM in general follow the protocol, while the others are on their own. What about summarizing common features and mechanisms from the former and discussing potential uncertainties based on the latter? In addition, it would be important to first introduce detailed performances of each model on LGM and PI before discussing their results. There have been some for sea ice, which I think is really helpful.  One potential way is to also provide absolute values with the same axis-scale for different models. Attentions might also be given to statements that are based on results from models with different forcing schemes. I do like the Discussion which discussed potential contributions of different factors and existing uncertainties, directing next steps. Honestly, I’m not very familiar with the expected outcomes of a MIP paper especially in this case. Dynamic-wise, I would encourage the authors provide additional sensitivity runs to substantiate key inferences, for instance roles of the early FWF forcing/sea ice on the early SO warming, etc., even if based on one model.

Minor comments:

Line 25-26: In the T1 protocol paper by Ivanovic et al 2016, it is suggested that the FWF scheme in the core simulation should at least follow the associated ice sheet history, which obviously is not the case for all the models here. Since these transient runs were initially conducted with different research focuses and none of them are really focusing on Southern Ocean (e.g. identical SO forcing, or referring to SO proxies, etc.), this avoids me from considering it as a typical MIP study for SO.

Line 38-42: I’m not convinced since the models are forced differently and hence spread of results is expected.

Line 184: what’s the physical meaning of γ?

Line 200-202: α and β here are different from those in eq. (1). It would be clearer with subscripts.

Line 210-212: why is the initial SST(0) equal to 0? Why should it be the LGM values? Please provide more details about the way to generate the best fit parameters.