General comments:

The paper of Pasquier et al. investigates the future changes of the ocean's biological and preformed carbon pumps by 2090s under two emission scenarios (RCP4.5 and RCP8.5) using a simple model biogeochemistry model and steady-state ocean circulation transport matrix models. They present a new partitioning of preformed DIC to separate the contributions of different pathways of preformed DIC to the ocean carbon storage and outgassing. Using this data-constrained model approach, they found that biological production declines only modestly in the future, while organic matter export declines more significantly due to the reductions in both biological production and export ratio.

The paper is well-written with clear figures and presents interesting results on the biogeochemical cycling and biological carbon pump. I only have specific minor comments and questions detailed below:

- The authors highlight the fact that the model results are to some extent imprinted by unrealistic circulation features of the ACCESS1.3 model, especially in the Southern Ocean and the deep ocean. The sensitivity of the results to the choice of this peculiar model should be more emphasize.

- This study does not account for the potential changes in the oceanic circulation and stratification due to melting of ice sheets, which could alter the ventilation and storage of carbon in the deep ocean. If this were the case, would it result in different behavior of the biological production or organic matter export?

- Why focusing on the RCP4.5 and RCP8.5 scenarios? This has to be justified.

- What are the implications of prescribed pCO₂ concentrations for the simulation? It would have been interesting to account for the feedbacks between the ocean carbon cycle and the atmospheric pCO₂ concentration, which may affect the future evolution of the preformed DIC.

- I found the description of the preformed carbon pump and how it differs from the solubility pump a bit confusing. I think it could be clarified.

- As explained by the authors, the effect of iron on phytoplankton growth is not included in this study to avoid complexity. How would it change the presented results if it was taken into account? (i.e. does it have an important role?)

Other comments:

L6: “experiences only modest declines”. Precise how much.

L8: the latter being driven

L19: In recent decades,

L30: …pump, preindustrial atmospheric pCO₂ concentrations

L55: remove the brackets of the sentence. Same at L77-78, L90-91, L128-129, L132, L258-259, L320-321, L375-376, L452-453 and in the caption of Figure 4.

L66-71: this paragraph has to be removed from the introduction since it is about the results of the study.

L80: atmospheric pCO₂. Same at L93.

L105-106: precise the reference used to choose the parameters of the PCO2 model

L112: …which in steady state obeys:

L133: the ocean through pCO₂ air-sea exchange

L138: Note that in our model, carbon

L156: “the exact same concentrations as traditionally defined preformed DIC”. This has to be precised. What are these concentrations? How are they chosen?

L178: what are the main mechanisms

L183: remove merely

L186: “the more sluggish circulation”. Precise the sentence with value for the circulation.

L186: decline by 12 and 19 % for the two scenarios compared to the preindustrial

L187-188: “due to the slower future circulation and decreased ventilation”. Please give values.

L191: for RCP4.5 and RCP8.5 respectively (compared to the preindustrial)

L191: “although the North Atlantic contains a patch of prominent cooling”. Does it have a local effect on nutrient and carbon uptake rates?

L204: RCP4.5 and RCP8.5 respectively

L220-221: in the RCP4.5 and RCP8.5 scenarios respectively, compared to preindustrial. Same in L253.

L230: “do not suffice”. Replace by are not enough.

Table 1: Add to the caption “…for preindustrial, RCP4.5 and RCP8.5 scenarios. Variations relative to the preindustrial are also shown”.

L271: remove a priori

L275: remove such

L302: and how they change in the future

L303: remove the [t]

Figure 6: make the dashed rectangle in the left corner more visible for the reader.

L326: To better understand the transport pathways of preformed DIC, we quantify the amount able to enter

L328: (not shown in Fig. 6) revealed that regardless of source-sink pair, more

L357: Italic font may be used for emphasis and used sparingly. Remove the italic style of the word less. Same in L361, 364, 365.

L437-439: remove the sentence to explain the SSP scenarios since it is not useful.

L461: remove below

L462: Liu et al. (2023)

L484: remove brought to light here

L491: this allows for the first time

L496: with declines of ~10 %, even for RCP8.5. Remove the italic style as well.

L500: RCP8.5-based steady-state scenarios respectively, in comparison to preindustrial

References: sort the references of DeVries in chronological order.

Review of “The ocean’s biological and preformed carbon pumps in future steady-state climate scenarios” by Benoît Pasquier et al.

Summary

The manuscript provides an analysis of the changes in the ocean carbon cycle between the preindustrial years and the 2090s, under the RCP4.5 and RCP8.5 future scenarios. Using a biogeochemical model forced by decadal-mean ocean circulations derived from a climate model, in steady-state, the authors find that (i) the counteracting effects of the decline in nutrient supply into the euphotic zone and the warming-enhanced phytoplankton growth rate induce a slight decrease in biological production, and (ii) the reduction in biological production and export ratio leads to a decrease in organic matter export. In addition, they assess the changes in preformed and regenerated dissolved inorganic carbon (DIC) inventories and cycles and find that due to the weakening of the circulation the inventory of regenerated and preformed DIC increases while the cycle of preformed DIC cycle becomes faster.

The manuscript makes a novel contribution by developing a partitioning of preformed and regenerated DIC in accounting for the source and sink processes. The manuscript is well written and organized. However I would like to raise several points, mostly regarding the discussion of the results, that should be addressed before its publication.

Please see the detailed explanation of these major points and all my detailed comments below. 

Main comments

1. The circulation model:  The authors forced the biogeochemical model using the fields of the ACCESS-1.3 model. They referred to previous studies (Bi et al., 2020; Pasquier et al., 2023) stating a surestimation in the mixed layer depth in this model (in particular unrealistically deep mixed layer in the Weddell and Ross seas), that could impact the carbon pump. Several questions arise from the existence of these biases in the circulation model:

• Could the author justify the choice of the ocean circulation ACCESS-1.3 simulations for this study? 
• The authors used optimized biogeochemical parameters, which partly corrects biases from circulation in the preindustrial conditions, if I correctly understand. Are these optimized parameters also used in the future scenario runs? If it is the case, how does this optimization influence carbon pump in the future states where these biases could have disappeared? 
• Could the authors expand the third point in the discussion section L. 406-420 by specifying how these MLD biases impact their results on carbon flow rates and inventories, and their main conclusions, given the “key importance of circulation changes”? Could they estimate the uncertainties of the results related to these biases? If these uncertainties are not negligible, at least, the authors should qualify, in the abstract and conclusion sections, the quantified changes found between the preindustrial and 2090s in both future scenarios. 

2. The preindustrial conditions: For the preindustrial run, the authors forced the biogeochemical model using averages of circulation and thermodynamics over the 1990s instead of preindustrial years. They justified this choice based on the minor changes in hydrodynamics between these two periods. Why did they not directly use averages of circulation over preindustrial years? I suggest adding some explanations on this point in the Methods section. 

3. Time integration: The authors considered averages on 10-year time slices. I suggest adding a small discussion on this relatively short time integration for which decadal variability is not handled.

4. Seasonality: In the discussion section (L. 408-410), the authors also mentioned uncertainties in the results associated with the absence of seasonality in their circulation model forcing. Could the authors be more specific and give an estimate of these uncertainties in the carbon pump and its plumbing, associated with this simplification, or refer to previous studies that could have estimated them?

5. Export of particulate organic matter: Pasquier et al. (2023) indicated that “particles are only submitted to gravitational sinking” in the coupled model. Is it also the case in the present study? I suggest clarifying how the export of organic matter is calculated in the Methods section. If the transport of POC is similar as done in Pasquier et al. (2023), could the authors specify or give an estimate of how including the advective-diffusive transport could quantitatively change preformed and regenerated DIC inventories and carbon flow rates, in particular in deep convection areas? 

Detailed comments

L. 66-71: I suggest removing this text summarizing the main results of the study from the introduction section.

L. 113: Did the authors consider the deposition of particulate organic matter (POM) on the floor, remineralisation of POM in the sediment, and the flux of dissolved inorganic matter from the sediment to the water column?

L. 181: I suggest rephrasing “ The future circulation of our states”.

L. 185-187: I suggest showing the nutrient supply and its changes instead of (or in addition to) the euphotic nutrient inventory and its changes, which result from several processes. 

L 192: Maybe better “and despite” instead of “despite”. 

L. 192-200: Please see my first major comment. The authors presented that the main changes in mean euphotic nutrient concentration are localized in areas where ML is unrealistically deep in the preindustrial simulation (Fig. B1d-e, Fig.  C2). Could the authors, in the discussion section, discuss or indicate an estimate of the uncertainties of the different contributions to DIC biological uptake changes and of the resulting uptake change, associated with the MLD biases, perhaps based on previous studies? Could the MLD biases impact the sign of the total production change and the first main conclusion?  

L. 201: Maybe better: “Changes in export ratio”

L. 202-203: I suggest defining export production at the beginning of “Changes in export ratios” section, instead of in L. 217-218.   

L. 208-215: In this paragraph, the authors listed mechanisms inducing changes in OC export ratio.  Did the authors quantify the changes of DOC and POC export ratios, DOC and POC exports, and euphotic DOC and POC remineralizations? I suggest adding these changes to the appendices. Related to my first main comment, could the authors give an estimate of the uncertainties of the magnitude of export ratio changes due the MLD biases? 

L. 217: “changes in carbon export production Jex itself”.

L. 223: Do you mean organic-matter production or export production? Please clarify.

L. 230: “suggest”

L. 259: “unrealistic deep ML”

L. 280-281: I suggest rephrasing this sentence.

L. 302: “these pathways change”

L. 303: delete [t]

L. 338: Do you mean “regenerated nutrients at intermediate depths”?

L. 345-346: I suggest deleting the repetitions with L. 253-254, if any. 

L. 367-368: the flow rate decreases/increases or the flow rate slows/speeds up? 

L. 391-393: I suggest replacing “a significant advance over” by “an enrichment of” 

L. 449-453: “the changes in the controls on carbon export and biological utilization identified by Boyd (2015)”: Could the authors list these changes and clarify the agreements with the study of Boyd et al. (2015)?

Appendices: I suggest (i) numbering the appendices and their figures in the order of reference in the main text, (ii) locating the figures after the title of the corresponding appendix (Fig. C1), (iii) changing the caption of Figure D2 (and Fig. D1), where the authors referred to Fig. 2 that is commented later in the main text. 

L. 533-535 and L. 613-618 : These lines should be located after the appendices.

Figure C2: I suggest adding the MLD changes for the two future scenarios as in Figure 2.