This paper analyses the Southern Ocean carbon cycle from simulations of the Australian Community Climate and Earth System Simulator Earth System Model, ACCESS-ESM1.5, which has been applied for the PMIP4 scenario lig127k on the Last Interglacial (LIG). The argument is made that since this was the warmest interglacial of the last 1 million year it might serve as an analog for changes which might be expected in the future due to anthropogenically caused glabal warming.

The methods and results are well written, and the figure are very informative. I suggest some improvement in the introduction during the framing of the research question (see details below).

However, my major point is that I have some difficulties with this suggested analogy of the Last Interglacial with future warming. The analysis presented here shows that during the simulated LIG the westerly winds have been shifted equatorwards resulting in weakenend winds south of 50°S. This process is then responsible for different upwelling pattern and is important for quite a bit of changes in the Southern Ocean carbon cycle. For the future warming, it is now anticipated that westerly winds will shift polewards and get stronger, thus the opposite of what has been found for the LIG. I therefore strongly suggest to reframe the article in a way that its interpretation is largely restricted to the LIG. This reframing probably includes a change in the title. I would even go so far in pointing out in the dicsussion, that due to these shifts in wind pattern found here the LIG is no good analog for what to expect from future warming for the Southern Ocean carbon cycle. The reason for these differences in winds patterns are suggested to be due the the changes in orbital parameters (which seems to make sense), and this shows that past analogies for the future are often problematic.

Details:

- lines 18-19: „of which 40% has been attributed to the SO“ It is not clear on what this 40% is related to since before it is said “Both land and ocean act as sink“. Does it refer only to the ocean part? And you probably mean that land and ocean EACH absorbs 25% of the anthropogenic emissions. You might furthermore consider citing the most recent version of the global carbon project here, thus „Friedlingstein et al 2020“ (instead of 2019) https://doi.org/10.5194/essd-12-3269-2020


- line 35: „The Last Interglacial (LIG, 129-115 thousand years ago, ka) was the warmest interglacial of the last million years“. This statement is problematic, since the paper PAGES2016  cited here analyse only the last 800 kyr. 

- line 36ff: „The warmer climate at the LIG is primarily attributed to a stronger northern hemispheric summer insolation (Laskar et al., 2004) owing to the orbital configuration of higher eccentricity and obliquity (Berger, 1978), rather than higher greenhouse gas concentrations as projected for the future“. The role of orbital forcing vs greenhouse gases on temperature have been analysed in detail in Yin and Berger 2012 DOI 10.1007/s00382-011-1013-5. 

- line 38: „The LIG is associated with sea levels 6-9 m higher than pre-industrial (PI) (Dutton et al., 2015)“. This knowledgee on LIG sea level has recently been revised downward, please reframe according to  Dyer etal (2021) https://doi.org/10.1073/pnas.2026839118. 

- lines 91-94: Two different DIC tracers. I cannot remember that one of the tracers (the one not being constant at 280 ppm) is ever mentioned again. If so, it can be deleted here. You should also mention here, that since atmospheric CO2 is prescribed this approach misses the feedbacks which are related to CO2 in/outgassing. Also, absolute CO2 fluxes are biased since the C cycle is simplified by this fixed CO2, which is acceptable for these interglacial conditions, but nevertheless might introduce a bias.

- line 110: γ_SST = 0.0423°C^-1 is called the Revelle factor. I am completely lost here. For me, the Revelle factor R is the relative change in atm CO2 over the relative change in DIC (unitless) R = Δ(CO2)/CO2 / Δ(DIC)/DIC, eg. Egleston et al (2010) doi:10.1029/2008GB003407, while you here describe some temperature-dependency. Please revise, or explain.

- line 19: weaker and stronger upwelling: by how much stronger or weaker?

- Fig 3g: xaxis title is missing 

A review of “Marine carbon cycle response to a warmer Southern Ocean: the case of the Last Interglacial” by Choudhury et al.

The authors present the ocean carbon cycle response to a warmer climate, as simulated by an Earth System Model with the last interglacial (LIG) radiative forcing. The authors compared the Southern Ocean air-sea CO2 exchanges between the LIG and the pre-industrial (PI), and attributed the difference to changes in SST, SSS, DIC, and alkalinity using a decomposition method. A major finding includes a greater CO2 outgassing during the LIG compared to the PI, caused by warmer SST overwhelming the effects of surface DIC decreases. As the authors noted, exploring the Southern Ocean carbon cycle sensitivity to a warmer climate is an important research topic. The analyses/interpretations of the model results are also convincing. However, I have a few comments that might help improve this study.  

1. Although this study focuses on the potential changes in the Southern Ocean carbon cycle during the LIG, there are no discussions of how the simulated changes are compared with other observation or model based studies. The authors discussed it for some physical variables such as SST and sea ice extent, but no discussions regarding biogeochemical properties (which are the focus of this work).
2. There are no validations of the model SST, SSS, DIC, and alkalinity against observations. Climatological mean DIC and alkalinity data are available through the GLODAP project where the authors can download observation-based estimates for the preindustrial period. The validation could be important because the simulated depth gradients of DIC and alkalinity can affect their redistributions during the LIG, which in turn control ocean surface pCO2 changes through their surface changes. Of course, the model would not be perfect, but potential biases due to the model deficiencies need to be discussed.
3. I am not sure how the inference of future carbon cycle change would be useful based on the simulated steady-state response during the LIG. What are the logics behind the linkage?
4. The authors focused on the Southern Ocean carbon cycle responses to a global scale climate change. In such a steady-state model setup, the globally integrated air-sea CO2 exchange should be close to zero unless there are imbalances caused by riverine input and sedimentary burial. The excess CO2 outgassed through the Southern Ocean needs to be taken up elsewhere. Where and how does this uptake occur? A discussion of this would be useful.

Specific comments

Line 10: “changes in sea ice exert a minor control” Is this also true on regional scales?

Line 11-12: what are the logics behind the linkage between the simulated LIG steady-state response and future transient responses in the coming century?

Line 18: 25% each or together?

Line 100: 275 ppm for radiative forcing only? The authors earlier stated that they used 280 ppm.

Line 110 and elsewhere: To my best understanding, the Revelle factor refers to a sensitivity of pCO2 with respect to DIC only, not to SST, SSS and ALK.

Line 169: could the increased uptake south of 62S be due to reduced sea ice? Reduced convection in the bottom water formation regions should decrease CO2 uptake.

Line 188-202: The net change in pCO2 is really small at 1.2 uatm due to compensating effects of SST and DIC+ALK on pCO2 change. Perhaps this might be smaller than measurement or PI pCO2 uncertainties? When averaged over the Southern Ocean, it is true that SST is a dominant factor. However, the dominating factors would be different regionally because the contributions from SST, DIC, and ALK seem to be highly variable in space. The authors discussed the zonally averaged contributions, but the contributions seem to be also variable in a longitudinal direction as well. Would it be useful to provide a spatial map to identify the dominant factor at each grid cell? For example, SST dominance in red, DIC dominance in yellow, ALK dominance in blue, and SSS dominance in green?

Line 206-207: negative? The red line seems to suggest positive values south of 75S in Fig. 3g.

Line 218: It is interesting to see the larger SST increases over the Southern Ocean compared to other ocean surfaces, which is responsible for the net increase in oceanic pCO2 over the Southern Ocean during the LIG compared to the PI. Perhaps, it would be useful to discuss why the SST increase is larger over the Southern Ocean than the global average during the LIG?

Line 228-230: why does the northward shift of AAIW lead to lower O2 and higher DIC and PO4? If it is the shift that causes the changes, we might expect some dipole patterns?

Line 283: The role of sea ice cover change can be tested offline. You can calculate air-sea CO2 fluxes based the simulated air-sea pCO2 difference with and without the sea ice cover changes.

Line 301: There is a study suggesting an increase in surface productivity over the Southern Ocean in a warming climate (e.g., Kwiatkowski et al. (2020)).     

Figures: please add latitude and longitude labels to maps

References

Kwiatkowski, L., Torres, O., Bopp, L., Aumont, O., Chamberlain, M., Christian, J. R., Dunne, J. P., Gehlen, M., Ilyina, T., John, J. G., Lenton, A., LI, H., Lovenduski, N. S., Orr, J. C., Palmieri, J., & Santana-Falcón, Y. (2020). Twenty-first century ocean warming, acidification, deoxygenation, and upper-ocean nutrient and primary production decline from CMIP6 model projections. Biogeosciences, 17, 3439-3470. https://doi.org/10.5194/bg-17-3439-2020