Review for “Increased future ocean heat uptake constrained by Antarctic sea ice extent”, by Linus Vogt, Casimir de Lavergne, Jean-Baptiste Sallée, Lester Kwiatkowski, Thomas L. Frölicher, and Jens Terhaar

General comments

This manuscript presents a new emergent constraint relating present-day Antarctic sea ice extent with future ocean heat uptake in the Southern Ocean. Using this new relationship, an updated estimate of ocean heat uptake is found to be higher than the ensemble mean estimate, and that the uncertainty is greatly reduced. While the increased heat uptake estimate is at odds with previous literature using other emergent constraints, the present-day sea ice extent used here is argued to be a more reliable predictor.

While the novelty of the study might not be obvious at first, since it relies on emergent constraints to reduce uncertainty on an already investigated metric (namely future Southern Ocean heat uptake), this manuscript actually provides a very valuable contribution to the overall understanding of the future of the Southern Ocean. The use of a new predictor, Antarctic sea ice extent, provides a more robust relationship and the new estimate of Southern Ocean heat uptake, of opposite direction compared to previous studies, feeds the discussion for future global and regional climates.

The manuscript is well written but more importantly, the study is well conducted, with several complementary data sources and methodological approaches used to provide a robust conclusion, and all results properly discussed. Many loose ends are convincingly tied in this study, except maybe the use of other sea ice extent-related metrics (especially the seasonal amplitude; see specific comment #3). This makes it a valuable manuscript, and I recommend publication once some minor concerns and modifications have been addressed.

Specific comments

• While the cloud feedback is an important process well included and discussed in the manuscript, a short description or definition is lacking for readers less familiar with such a process. This could fit well either in the second paragraph of the introduction or in the short 2.3 section, along with a brief summary of the radiative kernel method.
• Alternative predictors: this paragraph highlights that the authors thought through about other predictors to strengthen their results. Yet, there is no clear mention of other ice-related predictors, such as austral winter sea ice extent or amplitude of the seasonal cycle of sea ice extent. CMIP models are known to show discrepancy for both in the Antarctic. Contrarily to the summer extent, winter extent is less constrained by geographical boundaries (opposite of the Arctic, where the winter extent is the one most constrained by land) and might therefore provide a better insight on the potential evolution of OHU. The seasonal amplitude is also likely to influence ocean heat uptake through similar processes as mentioned in the manuscript. It might be worth exploring, but I also strongly suspect that while using those metrics might change some of the quantitative aspects, it won’t modify the qualitative story of the study. I therefore leave to the authors to evaluate whether such a new analysis is worth the effort.
• l.336: As currently worded, this sentence is arguable: the CMIP ensemble mean does not reproduce the observed increasing sea ice trend until 2015. Instead, as shown by Liu (2025), some subset of the ensemble does pick up this increase, but the related members remain a minority (24 versus 53 simulating a decrease). The Liu (2025) study rather highlights the role of internal variability in the present-day sea ice extent trend and the capability of those models to capture this internal variability. Whether this lends confidence to the use of the CMIP5/6 ensemble for emergent constraints based on the pre-industrial or historical mean state is not obvious to me. I suggest removing this sentence, which does not bring much to the discussion, or else clarifying your argument.
• Figure 8 is poorly inserted in the discussion. This schematic is pleasing to look at and conveys an important summary of the results, but it is not described nor discussed in the text. Please link it better with the main text.
• l. 352-359: this paragraph is not clear to me. Do your results invalidate previous results or not? You explain that it does not, but the following sentence still mentions that we should expect larger future warming, while previous studies rather mentioned smaller future warming. Please reformulate and clarify.

Technical comments and small suggestions:

• l.91: please clarify units of OHU (J)
• In section 2.4, when describing the maths behind the emergent constraint, a short sentence linking the mathematical terms with the physical terms would have been helpful, to better link the theory with how it is applied. (e.g. here, N=28 CMIP models; one of the predictors x_i is the sea ice extent while one of the response variables y_i is the ocean heat uptake, etc.)
• l.141: missing reference for the uncertainty in satellite product being only half of inter-product spread.
• l.368: when talking about thermosteric sea level rise, a reminder of the results buried in the section 3.2 would be welcome (e.g. “more thermosteric level rise, estimated to an extra 0.7cm under SSP1-2.6 and 2.1cm under SSP5-8.5 with reduced associated uncertainties, more damage to marine ecosystems [...]”)

This study provides improved global ocean heat uptake (OHU) projections by identifying a relationship between present-day Antarctic sea ice extent and future OHU across an ensemble of climate models. Combining this relationship with satellite observations of Antarctic sea ice reduces the uncertainty of OHU projections under future emission scenarios considerably. This emergent constraint is based on a strong coupling between Antarctic sea ice, deep ocean temperatures, and Southern Hemisphere sea surface temperatures and cloud cover in climate models. The robustness of this emergent constraint is thoroughly discussed and comparison with previous constraints based on past warming trends has been done. Overall, this paper is carefully written with detailed analysis and discussion. I only have some minor comments.

1) Line 200: Sea ice loss is mentioned here which is different from the sea ice extent used in Figure 5. For consistency, it would be good to show similar figures using sea ice loss, although a significant correlation between sea ice loss and sea ice extent is identified in Figure 3a.

2) Line 259: As described in the Methods section, OHU is defined as the anomalous net air-sea heat flux integrated in space and cumulatively integrated in time, which is different from temperature. Does the region where there is significant correlation between zonal mean ocean warming and preindustrial sea ice extent in Figure 4 necessarily show “addition OHU”?

3) I wonder why some figures (e.g. Figure 2, Figure 5) show thinner stippling while some other figures (e.g. Figure 4) show much thicker stippling. It may not be due to different grid resolution, I guess. It would be good to use the same stippling across different figures in the whole paper.

4) Section 2.3: Could you provide more details on how the climate feedback parameters are computed? For example, showing some major equations would be helpful.