The manuscript by Saini et al. is a continuation of a series of works where the effect of glacial-interglacial changes on planktic functional types and the carbon cycle is explored. Here they focus on differences between a portion of the onset of the last glaciacion (70 ky) and the Holocene. In particular the effect of different surface iron fluxes to the ocean on atmospheric CO2 and global ocean DIC is tested. The manuscript is clearly written. I have some comments regarding the methodology and presentation of results that should be addressed for the paper to be published by Climate of the Past. 

In the methodology, the authors give an analysis of different estimates of iron from dust solubilities, and they define a "most likely range" between 3 and 5%. Is this estimate also valid for preindustrial dust? For your preindustrial simulation you pragmatically chose 1% solubility. The estimate of CO2 effect of Southern Ocean fertilization is based on the comparison of 70 ky simulations that use 3-5 % Fe solubility in the Southern Ocean vs a 70 ky simulation that uses preindustrial iron fluxes with 1% solubility. Due to the high variations in estimates of Fe solubility in the ocean, which the authors correctly point out, I think that the case of same solubility between glacial and interglacials should be included in the "most likely scenarios". This will affect the minimum of your range of CO2 change estimates due to Fe fertilization of the Southern Ocean. 

The difference in results between using either of the two glacial dust flux estimates that you apply in your experiments appears to give a weaker response than changing the solubility. This is an important result, which means that the global carbon cycle is more sensitive to factor changes in soluble iron flux that to differences in the horizontal flux pattern. It would be a good idea to discuss the result in the paper.  

One of the novelties of the methodology is that it includes four functional types of phytoplankton: Normal phytoplankton, diazotrophs, coccolithofores, and diatoms. But how much does the inclusion of the new functional types affect the CO2 drop in the Southern Ocean fertilization experiments? In other words, how do the results of Saini et al. compare with a model with only regular phytoplankton and diazotrophs? It would be a benefit to the scientific community to have that comparison documented, since it would give us a hint on how necessary it is to include more planktic functional types in glacia-interglacial experiments. 

What's the CO2 difference between your 70 ky and preindustrial experiments? I imagine it is higher in the 70 ky experiment due to lower global EP. Is that effect overcompensated by the Southern Ocean fertilization effect? It would be important to note this in the paper, since in the real climatr system the onset of the glaciation included physical changes as well as the assumed iron fertilization. 

Another novelty of the model is the inclusion of a sediment model, which allows changes in global alkalinity. Does global alkalinity vary significantly among your experiments? Does the variation in CO2 that you find depend on global alkalinity changes at all? This is a topic that could be potentially addressed in this paper.

Minor comments:

Line 19: No need for the ~ symbol if you include uncertainty. 

Line 121: What leads? The sentence is not clear. 

Fig. 2: Color scale poorly chosen, it saturates in most places. 

Fig. 3: Why does the North Altantic show such a complex pattern of EP anomalies? Is it changes in convection, ventilation, the AMOC?

Fig. 4d shading colors not easily distinguishable. Why does terrestrial carbon go down in your simulations?

The global iron flux in experiments lambfe10% and lambfe3% is not shown anywhere. The fluxes should be plotted somewhere in the paper, since the results of these simulations are discussed in detail. 

Comment for “Impact of iron fertilisation on atmospheric CO₂ during the last glaciation”

Saini and co-author present nice modeling study of glacial carbon changes at 70 ka BP. The authors investigate, using a range of factorial analyses, the impacts of iron input and increased efficiency of biological pump in the glacial ocean. They simulate an upper limit for the CO2 decrease due to iron fertilization of 21 ppm. The manuscript is concise and well written. Figures and tables are illustrative and support the conclusions. I recommend publication of the manuscript after minor revision.

I have one concern that require revision. The author showed that an exponential decay response of CO₂ decrease to iron input to the Southern Ocean (SO) is due to the saturation of the biological pump. I agree with this conclusion, but some descriptions of the reason why the efficiency of the biological pump saturates in response to iron inputs are missing. As a result of the saturation of the increase in EP/NPP in the SO in response to the iron input, has the efficiency of the biological pump also saturated? For example, please add figures for changes in EP/NPP in SO in response to iron inputs in SO to Figure 4 and describe the nonlinear response of the ecosystem to changes in iron inputs.

It would also be important to add a similar figure for changes in nutrient concentrations to show changes in nutrient cycling with iron fertilization. In the case of large iron flux (e.g. lambre50%-47S), does the ecosystem response saturate due to nitrogen depletion instead of iron? I think these descriptions would be useful to the reader.

Minor comments

L174-175: It is interesting result that even though the two iron deposition patterns are very different, the effect on CO₂lowering is almost the same for the two patterns. Please add an explanation as to why the Weddell Sea convection would result in smaller changes in CO₂ for the different iron deposition patterns.

L286-287 Where is this information described in the method? I couldn't figure it out, so please let me know.

Saini et al. present an interesting modeling study in which the authors investigate the response of the global marine ecosystem to an increase in the supply of dust - and by inference Fe – to the ocean surface at the last glacial inception, 70 ka. The scarcity of Fe is indeed considered to limit biological/export production in large swaths of the world ocean, in particular in polar regions today and in the past.

Their sensitivity experiments indicate that increased availability of Fe conduces to a strengthening of the marine biological carbon pump, which contributes to sequester C in the ocean interior, decreasing atmospheric CO2 concentrations (4-16 ppm). What’s more, their model outputs suggest that most of the action is centered in the Southern Ocean (SO). Furthermore, the experiments suggest that the strengthening of the biological carbon pump consequent to Fe fertilization rapidly saturates as a result of complex interactions involving downstream effects, to me, the most interesting conclusion of the work.

The study is certainly interesting and worth publishing, yet I feel is it not sufficiently anchored in available palaeoceanographic records. High-resolution EP reconstructions across the MIS4/MIS5 interval are arguably limited, yet available records are often at odds with the model outputs - notably in the subsarctic North Pacific and the SO - which warrants a much more detailed/nuanced discussion. What’s more, given the sensitivity of the SO to increased supply of dust/Fe, I would urge the authors to distinguish the SAZ and AZ, as these regions have responded very differently in terms of EP.

Please find my comments below, which I hope the authors will find constructive.

l. 39 - Interestingly, d13CO2 data suggest that the drop in atmospheric CO2 centered around 70 kyrs was likely related to a combination of factors including enhanced C storage in the ocean subsurface and decreased air-sea gas exchange in the SO (Eggleston et al., 2016; Menking et al., 2022).

l. 49-51 & 227-233 - I find this argument somewhat misleading. I would encourage the authors to discuss the SAZ and AZ separately, as mentioned above. Palaeoceanographic data indeed suggest an increase in EP in the SAZ at the MIS4/5 transition (e.g. Martinez-Garcia et al., 2011/2014; Lamy et al., 2014; Thöle et al., 2019; Amsler et al., 2022). However, palaeoceanographic records from the AZ suggest a simultaneous decrease in EP (e.g. Anderson et al., 2009; Jaccard et al., 2013; Studer et al., 2015; Thöle et al., 2019; Amsler et al., 2022).

Fig. 1 – maybe the authors could include ice core d13CO2 data (Eggleston et al., 2016; Menking et al., 2022) in the figure?

l. 92/115 & Table 1 – could the authors provide some more context related to the dust flux estimates used in the simulations? The reason I’m asking is that the reconstructed dust/Fe fluxes to Antarctica based on ice cores measurements are much larger when compared to Fe fluxes quantified based on marine sediments. These differences may relate to different transport pathways and/or atmospheric dynamics over the SO and Antarctica during the last ice age.

l. 150 – is the AMOC both weaker and significantly shallower in the 70 kyr simulation?

l. 154-156 & 234-236 - Interesting… but these observations seem at odds with paleoceanographic reconstructions, which generally show a decrease in EP at the MIS4/MIS5 transition (e.g. Jaccard et al., 2005; Gebhardt, et al., 2008). These observations are consistent with paleoceanographic reconstructions spanning the last deglaciation, which consistently indicate lower EP during the LGM, even though Fe supply might have been higher then (e.g. Kohfeld & Chase, 2011 for a review)

l. 153 – I’m not sure to understand  what is meant by “strength of the deep water masses”?

l. 156 – what do you mean by “general phytoplankton”?

l. 157-163 – again, I would recommend discussing the SAZ and AZ separately.

l. 168-169 – why is that? Are ecosystems rapidly becoming N-limited as Fe availability increases?

l. 178-182 & 283 – This is not really surprising as EP was likely limited by the scarcity of N (or P) outside of the SO during the last ice age.

l. 179-188 – this conclusion is at odds with existing paleoceanographic reconstructions. Data suggest that the SAZ was particularly responsive to aeolian Fe supply, while ecosystems in the AZ mainly responded to (micro)nutrient supply from below via upwelling and vertical mixing (e.g. Jaccard et al., 2013). Moreover, N-isotope data suggest that the relative uptake of nitrate increased both in the SAZ (Martinez-Garcia et al., 2014) and the AZ (Studer et al., 2015; Ai et al., 2020) at the MIS4/MIS5 transition.

l. 211 - does the reduction in terrestrial vegetation and soil moisture imply more dust production (positive feedback)?

l. 247 – these observations would be consistent with a general decrease in deep ocean oxygenation at the onset of MIS4 (e.g. Jaccard et al., 2016; Amsler et al., 2022).