Dear Dr Voarintsoa and co-authors,

You paper has now received comments from two reviewers and one additional short comment. Thank you for also providing responses to these reviews/comments.

The reviewers highlight major deficiencies in the original manuscript that will require substantial revisions. Your responses go someway to outlining how the reviewer comments will be address, but I do not think that they currently go far enough in revising the work to a level that would warrant publication in Climate of the Past. At the same time I recognise that palaeoclimate records from this region are incredibly sparse and that your new records have the potential to make an important contribution to that sparse data network.

I therefore ask you to work further with the author team to develop and fully revise this manuscript. If you feel that you can more completely address the reviewer comments then I would welcome seeing a revised version of the manuscript, and would anticipate sending this version for further review to check the thoroughness and suitability of the revisions.

In addition to responding to the reviewer comments I would also ask that you pay particular attention to:

*detailing the mineralogical assignment of the speleothems and how this leads to the isotopic correction along the length of the samples. In particular it is unclear how each isotopic sample is corrected for it’s aragonite:calcite composition when only a small number of discrete XRD analyses have been completed.

*detailing the U-series and age model details. In particular more detail needs to be given on initial 230/232 values and the effect that the chosen values have on bringing U-series ages into stratigraphic alignment. Age information also needs to be included on plots showing the isotopic time-series so that readers can evaluate how much “movement” of the records is possible based on age uncertainty.

*be careful with “wiggle matching”. The examples in the proposed revised figures are not convincing to me. Please be careful in how you approach this and in how robust your climatic interpretations are.

*more thoroughly compare your data with other records from the region (not just a map of the locations of other sites). The Scroxton et al 2017 QSR paper that also presents speleothem data from Madagascar should also be included in the revised interpretation of your records.

As major revisions are required you may need additional time to complete these. If this is the case, please let me know and we can set a suitable time frame for revisions.

Sincerely,
Nerilie Abram

Thank you for submitting your revised manuscript to Climate of the Past. The manuscript required major revisions and has been substantially improved following the advice of the two reviewers and the short comment made during the discussion phase. Rather than resend your manuscript to the original reviewers I have carried out a detailed review of the manuscript myself. There are still further revisions that I feel are needed before the manuscript could be accepted for publication in Climate of the Past. These are:

*Abstract: the abstract is much too long. Please revise so that the abstract provides a concise account of the motivation and findings of your study.

*Please make sure that there is consistency in your use of age abbreviations throughout the text. Make sure all abbreviations are defined on first use (e.g. ka) Avoid jumping between different age scales (e.g. ka BP, yr BP, cal BP and AD are all used in section 5.2.3).

*Please try to reduce the length of the text wherever possible, and in particular keep the text more focused on the processes/concepts that are important for this study. E.g. There are long discussions about IOD and ENSO in both sections 2.3 and section 5.6, but no interpretations are made regarding the role that these climate processes may have played in the climate of Madagascar during the Holocene, so these sections should be much shorter than they currently are, or incorporated better into the interpretations made with your records.

*There are some sections that are still far too speculative and should be removed from the text. Things that in my opinion can’t be justified include:
- the discussion in section 5.2.3 linking events in the last millennium with the speleothem isotope records. The lack of replication of the isotopic records between the two samples makes this conclusion very weak. I would suggest sticking just with the broader-scale “growth-non growth” information, rather than trying to interpret features of the poorly reproduced isotope records.
- Figure 8: Too speculative, suggest removing this figure and be more careful with associated discussion.
- Figure 9: Too speculative, suggest removing this figure and be more careful with associated discussion.
- Figure S18: Too speculative, suggest removing this figure and be more careful with associated discussion.

*The discussion of the 8.2ka event is still a bit problematic, particularly as it isn’t clear that this excursion in the Madagascar record is reproducible, or that it is significantly different from other isotopic excursions in the records. I would suggest further reducing the emphasis on this event, although it would be OK to still mention it as a possible link that requires further verification with additional high resolution records from this time interval. Also, in the section at line 297, please don’t describe this as the 8.2ka event (this is an interpretation, not a result): instead you can say something like “A prominent isotopic excursion is evident at xx ka BP +/- xx y.”

Minor comments:
*The sentence beginning line 59 needs to be rephrased to make sense.

*Line 303/304: I think that you are refering to “local karst conditions”, not invoking “local climate conditions” as a difference between the caves. Please clarify.

*Section 4.3: I was expecting to see details in here describing why the mineralogy was primary and not a secondary alteration product. Consider moving this information to this section.

*Line 417: start a new paragraph here to break up the long section of text.

*Line 523: I don’t think “6ka event” is right in this context. I believe that this is a 6ka time slice simulation, rather than a climate event?

*Line 560: There is no need to bring in “Bond Cycles” here when you don’t look at any other of the proposed cycles in your work.

*Line 629: I think you mean “migration” not “expansion” here, as this is followed by “and/or expansion”.

*Figure 1b: please fix formatting so that all components of the figure are legible and not covered by labels.

*Figure 1d-e: Show location within caves where samples were collected.

*Figure 2: it is very difficult to compare the age information of the two speleothems in the current format of this figure. Consider combining into a single panel with two depth axes so that it is more obvious to the reader how the two age scales compare.

*Figures 5 and 6: Information about aragonite and calcite sections of the speleothems has been lost in the revisions. This is important, so please consider adding this information into these figures.

Figure 7: it would be helpful to use a symbol to show Madagascar on the maps. Is it necessary to include the Walker and the Head and Gibbard subdivisions on this figure?

Figure 11: Add information beside these records to show information on their interpretation (e.g. double arrows showing direction of wet/dry or warm/cool, etc.)

Dear Dr Voarintsoa and co-authors,

Thank you for submitting your revised manuscript to Climate of the Past. I'm pleased to now be able to accept your paper for publication.

Sincerely,
Nerilie