Dear editor and authors,

Please refer to the attached comments.

Vogt at al. propose an original study reconciling past and recent literature focusing on the drivers of ocean heat uptake efficiency (OHUE), a major metric linked to the role of the ocean in mitigating climate change. Previously proposed drivers were either upper-ocean properties, meridional overturning metrics, or both, with limited understanding of the linkage between the two, particularly under Earth system modelling frameworks such as the Coupled Model Intercomparison Project (CMIP). Using an extensive ensemble of the last CMIP models, the authors identify the drivers of OHUE using global and regional approaches and explore the links between these drivers. Thus, the manuscript addresses particularly relevant scientific questions fitting the scope of the Ocean Science (OS) journal. The manuscript does not present any novel data or tools but definitely bring a very welcome and truly missing level of fundamental understanding by investigating the links between major ocean processes that are keys to improve our knowledge of the climate system. Substantial conclusions are reached, for example confirming previous findings on the role of upper-ocean stratification in the Southern Ocean on OHUE and clarifying the previously proposed role of the Atlantic meridional overturning circulation (AMOC).

Going through the review criteria of the OS journal:

The scientific methods and assumptions are valid and clearly outlined; the results are sufficient to support the interpretations and conclusions (see further down for suggestions); the description of experiments and calculations is almost perfectly complete, traceable, and precise to allow for reproducibility (see further down for suggestions); the authors give almost systematically proper credit to related work and clearly indicate their own original contribution (see further down for suggestions); the title clearly reflects the contents of the paper; the abstract provides an adequate, concise and complete summary; the overall presentation is well structured and clear, and so is the language that is fluent, particularly well-written, and precise, making it very pleasant to read, thank you very much for this. The mathematical formulae, symbols, abbreviations, and units are all correctly defined and used. The number and quality of references are also appropriate. No supplementary material is provided due to the use of appendixes. However, I still found some parts of the paper that should be clarified or improved, leading to relatively minor revisions.

Main comments

Use of different members from 3 models:

The Table A1 shows that different members were used for piControl and 1pctCO2 experiments in the 2 HadGEM3 models and the GISS model. Please correct if these are typos. If this is not typos, for these 3 models (10% of the model ensemble), the OHUE anomalies between 1pctCO2 and piControl experiments aiming at removing any model drift might generate artificial variability in OHUE in the analysed signal which could be of (negligible?) concern for statistic and signal processing methods. This is not the case for the application of EOFs limited to data from the piControl experiment, but these differences could alter the OHUE vs. driver correlations/relationships investigated. I presume that using different members between experiments for these models have been done to circumvent limitations in data availability of these models/experiments. If so, I suggest mentioning this aspect in the Methods section, ideally including an assessment of the implications.

CO₂-doubling OHUE vs. preindustrial ocean conditions:

The OHUE computed under doubling atmospheric CO₂ conditions relative to preindustrial from the 1pctCO2 experiment is systematically compared to ocean properties and meridional overturning strengths under preindustrial atmospheric CO₂ conditions. It would be interesting to duplicate this approach and determine if the relationships are similar using the same CO₂-doubling period for ocean properties and meridional overturning strengths. My guess is that the relationships would be similar. If it happens that they are not similar, why?

Model ensemble vs. observation for bias assessment:

It seems that CMIP6 “historical” experiments have been used for comparison with the ECCO state estimate in Figure 4, but the use of such experiment does not seem to be mentioned in the Methods section. Please mention it if so, including the list of members. Or did the authors sampled the 1pctCO2 experiments on a period corresponding to the mean atmospheric CO₂ levels found during ECCO’s spanning period (1992-2017)? The latter seems inadequate considering the availability of the CMIP6 historical experiments.

Detailed comments:

Line 86: “four questions”

Line 91: “idealized atmospheric CO₂ increase”

Line 93: “model outputs”

Line 94: “are regridded”

Line 117: Are all CMIP6 models providing sea water density as an output? If not, I presume that you computed it from temperature and salinity, maybe with a toolbox such as TEOS-10. I suggest mentioning this aspect in section 2.2.

Line 178: Bourgeois et al. (2022) did not link OHUE to *global* mean upper ocean stratification, but only to *Southern Ocean mid-latitude* upper ocean stratification.

Line 182: “(Fig. 1a)”

Line 184-186: I suggest mentioning the agreement with previous findings of the particularly strong link between OHUE and upper-ocean stratification in the Southern Ocean mid-latitude (Bourgeois et al., 2022; Liu et al. 2023).

Line 338-339: Suggest removing the parenthesis delimiting the last sentence.

Line 439: Please include acknowledgements to WCRP/CMIP/ESGF, and any mandatory disclaimer from the EU grant agreement, if so.

Figure 1: Panels a-b shows 26 dots in accordance with the ensemble size. However, the figure’s legend seems incomplete with only 21 models (of the last panel I presume). Please show the full legend.

Figures with maps: Please add longitude/latitude coordinates and consider using a common colorbar for several panels where appropriate: e.g. a single colorbar could be used in Figure 2, 3, 5, A1, A2, A4-6, A8, and A9. Please also consider increasing the font size of many of the very tiny panel’s titles from many of the appendixes figures.

Figure 6: I suggest using the same acronyms as earlier in the manuscript: M_SO and M_WMT instead of MOC_SO and MOC_WMT.

Thank you very much for addressing my comments.