Dear Dr Brown,

I am now in receipt of three reviews and your response to them. Reviewers 1 and 2 suggest minor revisions, primarily having to do with the latest paleoclimate literature and the interpretations that may be derived from it. Your responses suggests that you will satisfactorily address such concerns.
Reviewer 3 requests deeper revisions: looking at relative SST rather than raw SST, with which you concur, and an analysis of changes in the Walker circulation. I agree with the idea that such an analysis would help clarify the mechanisms behind ENSO changes in the various simulations, but I am sympathetic to the claim that it would substantially lengthen the paper. As you point out, the work you present, and that of other manuscripts in this special issue, is primarily descriptive, so it can serve as a basis for deeper diagnostics and analyses.
I therefore agree with your suggestions, and look forward to a revised manuscript along those lines.

As you carry out such revisions, please follow all technical guidelines outlined here : https://www.climate-of-the-past.net/for_authors/submit_your_manuscript.html

All the best,

Julien Emile-Geay

Dear Dr Brown,
I have read the revised manuscript and your response to the reviewers, which show that you have conscientiously implemented the reviewers' feedback. I do, however, have two minor suggestions before I can formally accept the article for publication.

Firstly, many of the figures use stippling as a means to indicate model consensus. While this convention is widely applied in the climate community, I would argue that it is extremely misguided: it obscures areas of consensus (which are presumably of greater interest) and focuses the eye on the rest of the field (which is presumably of lesser interest, i.e. carrying a less robust signal). I would therefore encourage you to flip this convention, stippling the areas of model disagreement. If you need a precedent, you may find one in Hu, J., J. Emile-Geay, and J. Partin (2017), Correlation-based interpretations of paleoclimate data – where statistics meet past climates, Earth and Planetary Science Letters, 459, 362–371, doi:10.1016/j.epsl.2016.11.048. Just because a convention is widely applied, doesn't make it legitimate, and I have yet to hear anyone articulate a clear rationale for it.

Secondly, the definition of ENSO amplitude is rather vague, and as such open to misinterpretation. As far as I can tell, it is the raw standard deviation of the NINO3.4 index, without any kind of bandpass-filtering applied. This is dangerous, as it lumps all timescales together, including trends (if they are present). It would be more rigorous to apply a 2-8 year bandpass filter, to make sure that the statistic in question is not capturing decadal, multidecadal, or longer, variability. This seems like an appropriate safeguard given that the manuscript focuses on ENSO, not lower-frequency tropical Pacific variability.

I look forward to a revised version of the manuscript, and want you to know that the CP editorial board is cognizant of the delays that COVID-19 imposes on the publication/revision process.

Best regards,
Julien Emile-Geay, Special Issue editor

Dear Dr Brown,
thank you for your evaluation of the effect of filtering; I am satisfied that the changes observed are driven by variations in the interannual band (not lower-frequency ones), and am happy to recommend the paper for publication. On the subject of covering up areas of interest with stippling, I understand that you prefer to stick with a flawed tradition rather than challenge it. I rest my case here, but hope that the issue can get one day get further attention at higher levels of the IPCC.

The technical staff will be in touch about next steps for publication. Congratulations again and thank you for submitting your best work to Climate of the Past.