The authors present a detailed analysis of ocean carbon cycle feedbacks, which is a useful extension of the

recently published global CMIP6 carbon cycle feedback paper (Arora et al. 2020). The manuscript goes beyond 

the Arora et al. study in that it focuses on the contribution of different ocean basins to the feedbacks and 

also explores circulation changes (using AMOC strength as a proxy) and their relation the feedbacks. The 

authors diagnose the basin wide contributions of preformed (saturated/disequilibrium) and regenerated carbon 

pools to the feedbacks. The description of the methodology is more detailed and contains additional diagnostics 

compared to the Arora et al. study. This manuscript is clearly within the scope of Biogeosciences, and I 

believe it will be of great interest to ocean carbon cycle community. The manuscript is generally well written

(with some exceptions pointed out below) and I recommend it for publication in Biogeosciences after a few points 

detailed below have been addressed by the authors.

Main points

1) Title: I don't think "controls of feedbacks from different ocean basins" is a good title. I am not a

native speaker, but this sounds a bit odd to me. What the authors present is the "contribution of different 

ocean basins to carbon cycle feedback", and feedbacks are also attributed to different processes (including 

AMOC). As also noted further down, I don't think that the wording "control of AMOC on feedbacks in CMIP6 

models" is appropriate. For CMIP6 models, the authors show a correlation between pre-industrial AMOC and AMOC 

weakening. "Control" implies a detailed mechanistic explanation, in my opinion. This is beyond the scope of 

this study, but therefore I would avoid using the word "control" here.

2) The authors base their definition of feedbacks on changes in DIC-inventories, which, as they note, makes only 

very little difference at the global scale. However, at the regional scale the difference can be large (see 

Fig. 3), and therefore I suggest to use a different symbol for the feedbacks based on DIC-inventories. Since 

the feedbacks derived from carbon fluxes are the standard definition, I would use something like beta*/gamma*

for the feedback estimate based on inventory changes. This would also simplify the discussion at the beginning 

of Section 3, where the two beta/gamma definitions are compared (the authors could then just write "beta*" instead 

of "beta, estimated from the regional ocean carbon storage").

Also, it seems the authors point out that the feedback definition based on inventory changes makes more sense

than the "traditional" one based on accumulated fluxes (line 290-292: "...to gain more mechanistic insight, so 

as (i) to account explicitly for the ocean transport of carbon..."). Here I would disagree: From the feedback

perspective, the flux at the air-sea interface (and changes to it) is the process we are interested in. Transport of 

carbon below the ocean surface leads to a disconnect between the actual feedback process at the surface, and where 

DIC-inventory changes are diagnosed (nicely illustrated by Fig. 3). Don't get me wrong here: I think the method the 

authors use is extremely useful to gain a global to large-scale regional understanding of ocean carbon cycle 

feedbacks, but there is a price to pay. In lines 290-292 it sounds like the authors are selling this "price to pay"

as an advantage. Maybe the authors can re-consider their wording here?

3) Figure 3 and related discussion at the beginning of Section 3: This is one of the core figures of the manuscript,

but it doesn't account for model uncertainty. I think it would strengthen the manuscript if the authors could 

expand this figure by 4 panels visualising the model spread (or standard deviation) for the 2 beta/gamma pairs, and 

add a brief discussion of where the main model uncertainty lies (and how and why this is different for the two 

definitions of regional feedbacks). Also for beta/gamma based on the DIC inventory, it would be great to split this 

further into the components (sat/diss/reg; this Figure could go into the Appendix). 

4) Figure 5 and related discussion: This Figure seems to be flawed:

  -why does the total ocean volume add up to only 99%?

  -why do the different contributions to beta/gamma add up to different percentages (between 91 and 99%)?

The discussion in lines 332-337 is unclear to me: "In the well ventilated Atlantic Ocean, the additional heat

penetrates into the ocean interior and is not confined to the ocean surface, which limits the effect of the 

reduction in solubility with warming". But the definition of gamma_sat doesn't care whether a water parcel is 

at the surface or not. I don't think that this is can be the explanation (same in the next paragraph for the 

Pacific). Please clarify this.

5) Nowhere in the manuscript it is stated how the ocean basins are defined. Where is the delineation between the 

Southern Ocean and the other basins? What about the Arctic Ocean? Is it included in the Atlantic or omitted? What 

about marginal seas? Also, the definition of AMOC strength is not given. Please add this information.

Minor points



lines 1-3: This sentence is complicated, and the wording isn't very precise (it is not a "competition between 

the increase in atmospheric CO2" but "a competition between the response to the increase in atmospheric CO2..."). 

Please consider rewording this sentence and maybe splitting it into two.

Equation 9: "Delta f" is not defined. Here, I would find it worthwhile writing the equation first in terms of 

DIC_sat, and state that DIC_sat=f(CO2,....). Then write down explicitly what Delta f means.

Equation 10 and 11: Here, I think, it would be easier and shorter to just express beta and gamma in terms of I_sat 

(without writing out the f-terms explicitly - since I_sat is already defined in terms of f just a few lines above).

Equation 18 and 19: Here, I also think it is much easier to understand if beta and gamma are just expressed in terms 

of I_reg, which is defined just a few lines above in Eq. 17.

line 84-85: This is an assumption, not a conclusion, so starting the sentence with "Hence..." is not appropriate. 

Maybe use "In the carbon cycle feedback framework introduced by Friedlingstein et al. (2003,2006) it is assumed 

that..." or similar.

line 266: "...where gamma_n includes the non-linearity of ocean carbon cycle feedbacks". This is confusing, it sounds

like the definition of gamma_n would be different from that of gamma, which is not the case. I suggest to delete this.

Figure 3: To make this figure consistent with all other results, please add NorESM2-LM (fgco2 for the BGC run is 

available).

line 286: What do the authors mean by "asymmetries"? Please clarify.

line 323-324: "By definition, the contribution of each basin to beta_sat and beta_dis is approximately proportional 

to the ocean volume contained in each basin...". I see that this is the case for beta_sat, but for beta_dis this 

depends on ventilation which is not related to the volume. Maybe delete beta_dis here?

line 445-447: Please check and reword this sentence (consider splitting in two).

line 476-477: "...which is mainly due to the disequilibrium carbon pool and the reduction in the physical 

ventilation with climate change." The second part of this sentence is a conclusion, isn't it? Then it would 

be more appropriate to write: "...which is mainly due to the disequilibrium carbon pool, indicating that the 

Atlantic has the strongest reduction in the physical ventilation with climate change."

line 487: This is also seen in Schwinger et al. 2014

line 491-492: "The inter-model variability in gamma amongst CMIP6 models is relatively large compared with beta...". 

I think it is worth mentioning that this is not true in terms of the absolute feedback strength: In terms of PgC 

taken up by the ocean, it is still the uncertainty in beta that plays the dominant role.

line 513: "...controlled by the AMOC weakening..." As pointed out above, the authors find a correlation, so in my

opinion the term "control" should be avoided here. Please consider rewording.

Technical

lines 26-28: Please check the grammar and logic of this sentence.

line 35: modes -> models

line 43: "defined on" -> "defined based on"

line 83: "..such as for example leading to..." please check grammar

line 121: "at the surface" is confusing. Maybe better: "is the part of DIC that has been transferred from the surface 

into the ocean interior..."

line 129: "is a unit conversion" please spell out from which to which unit.

line 180: "The term inside the first {} brackets..." -> "The first term in curly brackets..."

line 199: "to the alkalinity" -> "to alkalinity"

line 285 South -> Southern

line 328: necessary -> necessarily

line 349: "by the ocean ventilation" -> "by ocean ventilation"

line 356: delete "now"

line 372-373: "is the preindustrial" -> "denotes the preindustrial state" (or similar)

line 470: "...of 26% to 30%..." -> "...between 26% and 30%..."

1. interpreting changes in the distributions of the different carbon pools is not always straightforward as you might hope (see spatial complexity in Arora et al., 2020). Nevertheless,  basin-scale changes in carbon storage provide a natural integration of  a spatially complex air-sea CO2 flux pattern.  
2. Even using this approach it is not possible to completely separate the various processes. For example,  the regenerated component combines both biological  and ventilation changes because the strength of the biological pump is also dependent on the ventilation.   It seems that describing the changes in terms of their impacts on the different carbon pumps could be helpful.   
3. Basin-scale variations in the change in carbon storage can’t be interpreted directly as the importance of the region to the global carbon cycle feedbacks, because the source (and mechanism) of change could have been from outside the region. 

1. Clearly present and give distinguish between the  feedback parameters and the carbon inventories calculated from the  i) air-sea CO2 fluxes and ii) interior carbon inventories. In regional analyses this distinction becomes very important (as the authors explain). For example, you could use ‘betaf’  and ‘If’ and ‘betas’  and ‘Is’ for beta and carbon inventories calculated from carbon  fluxes  and carbon storage respectively. 
2. Please include a discussion of the merits/limitations of the flux vs storage approaches and their complementarity. 
3. Move the regional calculation of the feedback parameters (263–274) to this section. 
4. The basins have not been defined. Please provide the precise boundaries used for the basin-scale analysis show boundaries on the maps in Figure 3. 

1. Start this section with something like your line 115. “To gain insight …”
2. Include detail here on the how the various carbon pools  are used to diagnose ocean processes (ventilation, biology…) and the limitations of this approach.
3. I assume  methodology used to calculate the carbon pools is identical to that used in Arora et al., 2020?  Please elaborate. If so, it would be useful to inform the reader. 
4. Include here only the methodology used to calculate the changes in the carbon pools and leave the theoretical explanations that are used to better understand their  evolution to section 3. 

1. Either include a more detailed examination of the impact of transport on regional betas and gammas in a dedicated results& discussion section including the figures in the Appendix.  This would be interesting if it extends our understanding beyond what is presented  in Frolicher et al., (2015). 
2. Or, the spatial distribution of the  ocean carbon-cycle feedbacks in figure  3 could be moved into and discussed in the methodology section where the air-sea CO2 flux and carbon storage feedback parameters are introduced and compared. Here the figures could be discussed in terms of the complementarity of the flux and storage approaches.

1. either grouping the  contributions from each carbon pool into a plot for each basin rather than for each component. 
2. or presenting the contribution of each component to the basin-scale feedback parameters as percentage contributions, so that it is easy to clearly see how the relative contributions of the different components to the total feedback vary between the regions.