The manuscript has now received two positive reviews, which include several suggestions for improvements and for some minor corrections to be made. I thank the authors for their careful and considered responses to the reviewers comments. The changes that the authors propose to make will allow for an increased level of detail for the discussion and interpretation of the results.

Whilst incorporating the proposed changes into a revised version of the manuscript, I have some additional minor comments to consider:
- reply to reviewer 1 page 2: discussion of uncertainties of the forcings. Thank you for adding this additional text. Is it possible for you to give some values for the differences in amplitude of the reconstructions so that it is more clear whether the reduced amplitude of Willeit et al. is still centred on the broad range of values from those you used? (or whether the data is offset in some way) You refer to the Berends et al. (2020, CPD) manuscript but it would be good if the reader doesn’t have to go digging into another manuscript to understand how different (or not) the CO2 data and the associated uncertainties on the forcing may be.
- reply to reviewer 2 page 3: comparison of the Ti/Al data with that of Grant et al. 2017. Thank you for clarifying the difference in the two records. Have you made a direct comparison of the two data sets where they overlap to investigate whether there would be any difference in interpretation (e.g. are the patterns and directions of change in alignment? Are there events which don’t match because of the incorporation of additional data in Grant et al. 2017?)
- reply to reviewer 2 page 4: if you are referring to million years ago it has been recommended to use “Ma” (e.g. https://www.geosociety.org/gsatoday/archive/22/2/article/i1052-5173-22-2-28.htm)
- reply to reviewer 2 page 4: Figure 3. I interpret the reviewer concern here reflecting the fact that the same colour of blue is used for Runoff 11 mm/day in panel (a) whereas it is used for Runoff 12 Power @23 kyr in panel b and then Runoff 11 Power @23 kyr in panel c. I agree that it would be good to use a different colour than blue in panels b and c. The colour scheme at present lends itself to a quick assumption that panels b and c are referring to Runoff 11 (since that is in panel a). Could you reconsider the colour here for either panel a or b&c?

I thank the authors for submitting their revised manuscript and responses to comments. Overall the changes have been positive and have addressed the reviewer comments.

Some minor corrections (line numbers refer to the tracked changes document):
- line 36-37: can you check the ordering of the phrasing here? Would it be better to refer to the inception of NHG occurring close in time to the Pliocene-Pleistocene boundary? (or by “Pliocene-Pleistocene transition” are you referring to a longer term shift in climate which starts in the Pliocene but ends in the early Pleistocene rather than the exact point of the stratigraphic boundary?). Would it be worth putting a time bracket on the “Pliocene-Pleistocene transition” if you are referring to a wider time interval?
- line 50: “including the Plio-Pleistocene transition”?
- line 61: “including” rather than “during”?
- line 125: “whereby” rather than “whereas”?
- line 135: can you please clarify whether the methods were the same as the cited paper, and if not, how were they different? You may need to include a few lines here to ensure that a reader can follow the method used in the lab to generate this new data.
- line 147: “shows a similar pattern”?
- line 273: “and included ice-sheet changes”
- lines 280+281: “constrained”
- lines 282-284: can you also note the ranges for the most recent boron isotope proxy reconstructions? (e.g. de la Vega et al. 2020 in Scientific Reports?)
- lines 297: what is “coaling”? (typo for something else?)
- line 312: I don’t think you need “following” at the start of the sentence