The manuscript by Justino et al. presents new model simulations designed to test the impact of astronomical forcing and changes to the West Antarctic ice sheet, sea ice, and ocean circulation to explain the climate conditions recorded during Marine Isotope Stage 31. This unusual interglacial within the Pleistocene is relatively warm in certain regions, and has been a source of debate and discussion for some time. It is valuable to try to test the relative importance of the external (orbital) and internal climate processes which might explain this interglacial stage.

Both reviewers have serious concerns about the study. Reviewer 1 is particularly concerned with the potential issue of model bias leading to erroneous conclusions on the main climate patterns (and their controls). I agree with the reviewer that this needs to be much more clearly explained in the main text. The reply by the authors notes that they will address these concerns in a revised manuscript: this definitely needs to be done, and very carefully, to ensure that there is detail in the statements to back up their claims. The authors likely need to do some serious redrafting of some early sections of the manuscript to ensure that there is a good structure and depth to this discussion – because accepting any other claims in the manuscript is heavily dependent upon this.

There are also concerns about how certain the authors can be about some of the patterns and trends – the statistical significance is rarely outlined, and visual comparison tends to be used. However, the formatting of the graphics does not always help to see these patterns clearly either (Reviewer 1 has many constructive comments on how to improve this). Without some statistical justification of the patterns being significant, it is difficult to accept some of the arguments made here about the key processes operating during MIS 31. There has been some improvement in the figures shown in the 'reply to reviews' - these versions should be incorporated into a revised manuscript.

I found the treatment of the data-model comparison rather vague as well. A small selection of data available for this MIS is shown, but just ‘blue=cold, red=warm’, and there are notes to ‘reasonable agreement’ or not. This is not a robust comparison, and misses other available data for this time window. Reviewer 1 also highlights some errors in the reporting of data interpretations, which need to be corrected, and Reviewer 2 is also concerned about a lack of quantitative comparison. The reply to reviewers 1 and 2 by the authors is not very convincing: it should still be possible to show some or all of the data as anomalies, even with error estimates given proxy/age uncertainties. There will also be uncertainty in the models, but showing how similar/different they can be is an informative exercise. (n.b. in the revised figure 2 with the data shown, some of these sites look to have moved – e.g. sites 806 and 849 in the Pacific should both sit on/close to the equator? Check all plots for accuracy). I would like to see an improved treatment of the display of the proxy data and its citation and explanation in the main text.

The graphics are not always supportive of the text: the scale bars are not always clear to read, different panels of the same figure can have different scales (e.g. why is figure 4b just an excerpt of the global scale, when 4d is global? Would it not be useful to directly compare both). Reviewer 1 has made many constructive annotations on the manuscript to address these. Some of them have been taken up. The seasonal sea ice differences in Figure 1 are not shown, but I agree with the reviewer that that would be a very useful plot.

Both reviewers comment on a need to improve the citations of the key literature, and also the style of writing. The annotated comments by Reviewer 1 will be particularly helpful to the authors in this endeavour. A lot of care needs to be taken to avoid descriptive terms for the model comparisons or data-model comparisons: statistics will help this, but terms like ‘agree reasonably well’ are very subjective, and could be seen as a weak correlation by others. The authors also need to be careful not to keep directing the reader to other sources to find the details required for a reader to understand or evaluate what is going on: some of the details of model bias may be published elsewhere, for example, but you must state what that bias may/may not be here too.

A revised version should make sure that all pledges to make changes in the 'replies to reviews' are carried out.

The authors have made a number of major revisions in response to the concerns of the two reviewers and the editor. Overall, the manuscript is much improved. The increased detail in and improved presentation of the graphics is particularly valuable. However, there are still some areas where the reviewers concerns have not fully been addressed. Some of these are significant (e.g. detailing the model bias and what impact that has on the conclusions about circulation drivers) whereas others are minor but were requested by reviewers to seek clarity in the statements. There also remain quite a number of typos and grammatical mistakes. I have attached an annotated version of the manuscript here. The “informality” in the writing noted by a reviewer continues through the revised manuscript (e.g. “it has to be noted that…”). The Editor also sought a reduction in the informality of discussing relationships without statistical support or detailing the relationships. These informalities need to be removed or edited in a revised submission.

A critical comment of the reviewers was the need to be clear on the potential biases of the model in areas of sea ice formation, and what that might mean for the OHT in particular. This still requires a few sentences of explanation (I have made a suggestion for page 4). Noting that “the main sites of NADW formation are properly located” (page 9) is vague, and it isn’t clear from figure 3 that this is the case, since the model outputs are only compared to the CTRL simulation, not to known oceanography.

I have also made suggestions on the manuscript about some re-structuring of text, to help the flow. This is especially important in section 2 where the methods and tests are outlined.

A revised submission also needs to cross-check the corrections indicated by reviewer 1 on their annotated script. There are many typos or suggestions for change which the reviewer 1 made, which have not been considered but without any clear rationale. I have highlighted these in the annotated manuscript. A revised submission of this manuscript should address these concerns carefully.

I should also note that providing annotations on the pdf was quite problematic (I was often unable to be precise in where my comment was placed), so the comments are often shaded and then noted so that you see the broader area in which the comment applies.

Thank you for taking the time to carefully address the previous comments and concerns of the reviewers and the Editor.

To proceed, please upload your final version of the manuscript, addressing the following technical corrections:
- remove the italicised sections where you have highlighted your responses to previous comments
- page 4 line 13: can you insert a reference or two here to support your statement about some models being warmer than others?
- title of section 3 "Design of the sensitivity experiments"
- delete page 5 lines 13-14 "Our simulation... MIS 31 interval" (you essentially said it on line 6, and it interrupts your discussion of freshwater impacts)
- page 7 line 16-17 notes impact of WAIS topography on sea ice, but this seems to interrupt the overall discussion on SSTs, whereas lines 24-25 then note the impact on sea ice. Should the first of these be moved to coincide or to replace the second? I'm a bit confused about why you cite Supp Fig 1c in the second example when the first cites Figs 2e and 2f?
- page 7 final line (into page 8): is this finding from your results, or do you need to cite one of the data papers which shows this?
- page 8 line 9 "these locations" - clarify if this is Lake E and ODP 982, or remove the paragraph break so that the two sentences are together.

Thank you for submitting your work to Climate of the Past.