Dear Pascale, dear co-authors,

Thank you very much for your thorough reply to the comments made by the reviewers. I invite you to submit a revised version of your manuscript.

The comments by the reviewers are well taken. There are only a few items that I would like to add.

One of your topics addresses multiple vegetation states. This is, indeed, a fascinating subject. You mentioned papers by Victor Brovkin and me. Hence it would be interesting to see whether you also find the differences in tropical large-scale circulation (e.g., a shift in the velocity potential) between the different states.

In your reply to reviewer 1 you mentioned that you would use the biomisation method to evaluate the different vegetation states. Perhaps I just mis-understood this point. It would be more appropriate to compare simulated vegetation patterns directly instead of diagnosing these patterns by using the indirect biomisation.

Reviewer 2 questions the use of constant aerosol forcing. The effect of changes in Holocene aerosol concentration is a topical question. (You mentioned the papers by Francesco Pausata, for example.) It seems that it is not at all clear what effect a change in mineral dust during the Holocene has had on Holocene climate and climate variability. The effect of mineral dust on precipitation, for example, strongly depends on the chosen values of optical properties of the aerosols.

Reviewer 2 also mentions the calendar problem. Looking at the recent paper by Bartlein and Shafer (see GMD Discuss.), the effect can be pretty large, if monthly means are concerned. Averaging over 4 months should have a smaller effect. Perhaps you can cite your 1997 paper with Sylvie Joussaume?

Line 35 of your manuscript: the term ‘intertropical convergence zone’ might be misleading, if you want to refer to the tropical rainbelt – see Sharon Nicholson’s critique which appeared in BAMS, Feb. 2018, pp. 337.

Lines 69 ff: model biases: You are right that an atmosphere-ocean GCM is a different model than an atmosphere-ocean-vegetation model. Looking at the Holocene West African monsoon, I would argue that the biases of an atmosphere-ocean GCM are larger than that of an atmosphere-ocean-vegetation model. The former generally produces too small monsoon rain than the later. But can you really compare the biases of different model types?

Line 78: A word on why you start at 6ka would be sensible. It is the old problem: the Holocene climate around 6ka reveals pretty strong changes.

Best regards,

Martin Claussen

The authors have done a great job in revising their paper. Both reviewers appreciate this effort and suggest minor revisions. I tend to agree with the reviewers and, hence, ask the authors to submit a revised version which will be seen only by the editor.

The reviewers, specifically reviewer 1, have raised a number of points which should be carefully considered by the authors. I am not sure whether it makes sense to put the tests of different model versions into the appendix. Reviewer 2 seems to like this careful testing, albeit it is somewhat tedious to read. Points b) and c) of reviewer 1 should definitely be considered. Finally, I am a bit worried about Figures 14 a, b, c and Figure A 2 b. These are too small.

Dear Pascale and coauthors,

You did a fine job responding to the questions and comments of the reviewers. Therefore I am pleased to accept your manuscript for publication as is.

Best regards,
Martin