Guiet et al. present recent developments in the BOATS global ecosystem model that improve ecological and economic processes to improve the fit of model results to biomass, catch and fishing effort data. BOATS is a size-spectrum global marine ecosystem model in which they mainly incorporated three new features: 1) a demersal guild in addition to the already existing pelagic guild, 2) a better representation of iron limitation and 3) a modified assumption of fish accessibility for fisheries including spatially variable fishing costs and catchability. The results show multiple improvements in the evaluation of various indicators, validating the necessity of these recent developments and the ability of the newly included mechanisms to reproduce the observed patterns.

I would like to emphasize the quality of the article, which is well written and overall clear on the description of the model's equations and assumptions. Well done to the authors, who have done a tremendous job in this respect. In addition, I very much appreciated the approach of developing and testing new mechanisms in the model by successive implementations. Thus, I think the article is timely as we need in marine modeling science to:

(i) Continue the development of existing models by including new robust mechanisms to understand and anticipate current and future human pressures on marine ecosystems

(ii) Have a clear description and documentation of model assumptions and equations to improve model usability, transparency and reproducibility. For these reasons, I fully support the authors' request to publish their paper in Geoscientific model development.

 I also very much appreciated the ingenuity of the model parameterization and the selection of the best set of parameters: their approach could be more widely used for other marine ecosystem models in a context where model validation and calibration methods are still under discussion and development.

The structure of the document is unusual in that it includes a section on the sensitivity analysis of the model to certain parameters between the “Materials and methods” section and the results section and presents some of the results of the sensitivity analysis in this section. The need for this information to better appreciate the results makes it acceptable. Furthermore, the results and discussion are brought together in one, which I find very appropriate in this type of technical paper.

My main criticism concerns certain parts of the discussion. I found that some results for which the model does not perform well are never discussed, whereas a discussion accompanied by a hypothesis about potential missing processes in the model or experimental data could help validate the model and understand the gap that remains between the data and the model. One of the aims of the paper was to improve the representation of iron-limiting zones. Are the zones better represented in the iron-limited zone of Boatsv2? How do you explain the areas where the model performs less well? Similarly, when the model underperforms in v1 and v2, can you speculate on the reasons for this underperformance? Some global hypotheses are mentioned at the end of the discussion, but specific hypotheses for the highlighted area would be useful (North Atlantic, Eastern Pacific, etc.). Maybe other global ecosystem models (Apecosm, DBEM…) with different assumptions perform well in different region: it could be helpful to formalize. For more examples, see the detailed line-by-line commentary.

My second point concerns certain parameter settings. I noticed that some values/hypotheses were given without explanation. Even if they are assumed or derived empirically, I think it would be useful to specify them for greater clarity. See also the line-by-line commentary for more information.

Lastly, I found some minor typo error that needs to be corrected (see lines by lines)

To close, I wish to emphasize that I consider this research is already of great quality.  My criticisms are simply intended to be helpful to developing/precising it.

L37: Add a source exploring multiple aspects of global fisheries dynamics

L39: the example of an ecosystem defined as HNLC could help non-specialists of this ecosystem to identify the type of ecosystem you are referring to.

L70: how is the vertical position of communities/captures estimated if the grid is 2D?

L80: First equation: The first term of the equation is a growth term, but it is negative. Is it the biomass that exceeds the group size threshold due to growth? If so, a sentence explaining this would be useful.

L86: If the minimum requirement is not met, this has no impact on mortality, why?

L95: "Primary production is equally distributed between groups". Why is this?

L103 and 104: The typography of the letter "phi" is different in this line and the next than in the rest of the text.

L107: Equation 3: same A0 as in the anabolism equation? If so, mention it either afterwards, or in the table of parameters.

L121-122: Do you think that using only the first 75 to estimate the temperature faced by the pelagic community is sufficient? Or could this be responsible for a bias in the representation of the community? If it is the second option, please discuss it

L136-138: I did not find the parameter em_o,k in the equation preceding (5). Explain why this information is given here or delete it.

L146: Since EK (t = 0) = 0, how do the dynamics of Ek begin?

L147: Table 1: Predator to prey mass ratio: this ratio is very high. Do you have a source that confirms this? How can the trophic scale parameter be interpreted biologically?

L157: "qk increases annually at a rate of 5%" : Where does this value come from? Is it realistic?

L171: Why is the reduction in primary production not enough to explain the change in fish growth? Do you have any arguments in favor of a change in trophic efficiency? How do they explain the mechanism in Galbraith et al?

L174: (NO-3 , in μM) is considered an indicator of iron limitation. Why is?

L199 and L278: What thickness is used to estimate Tbot? Does the thickness vary with bottom depth?

L208: How did you assume -0.8?

L221: Specify the activation energy of "growth and mortality" to help identify that these are 2 parameters.

L231: When fisheries target demersal species, do we agree that cost increases with distance from shore and depth? If so, I don't find this clear in the equation for the demersal community. If not, why not?

L322: 12 is not fixed in Table 1. Is this a reason not to recalibrate the trophic scale?

L324: (h ζ1) becomes (h and ζ1),

L355: "5 parameters (6 including the trophic scale)". Why do you make a distinction here?

L386: Figure 3: Acceptable range in addition to mean harvest could be useful ([70,150]x10^6) + Why is there an overall overestimation of pelagics?

Figure 4c: why the change in variability over time?

Table 3: why does v2-Bio* & Πβ seem to be the best model? *v2-Bio = v1 + αcorr + (Πψ : delete parenthesis

L389: These Australian LMEs: are they deep or iron-limited zones?

L445: How can mortality be negative? Perhaps more information on this term in the additional parameters table might help to understand where it comes from.

Figure 5a: How do you explain the increasing trend in the model for the 2 versions, which is not observed in the data? Is it linked to the exponential response to temperature? If so, it could be interesting to discuss what other temperature responses could have been used, and how they might impact the model.

L453: “Although the temperature dependence of mortality (ωa,λ) is not significantly different from the initial values, the optimized values suggest a stronger sensitivity of growth compared to mortality for the pelagic community (ωa,A −ωa,λ = +0.047 eV), and a stronger sensitivity of mortality for the demersal community (−0.082 eV).”. Is it supported by experimental studies of fish thermal responses?

L464: "indirectly allows larger asymptotic sizes (m∞) that are exposed to greater natural mortality; however, since m∞ is fixed". There's a contraction in that sentence, isn't there? If not, it needs to be explained differently.

L488: For ecological or economic reasons?

L495: Can you add a hypothesis about the reasons?

L540: What are the expectations in terms of the impact on carbon sequestration?

Table A1: n_k = 3. Why is this so? Is the model sensitive to this parameter?

Table A2: Temperature units. K or °C. Why use the 2 units?

The paper “BOATSv2: New ecological and economic features improve simulations of High Seas catch and effort” describes a set of significant advances in the BiOeconomic mArine Trophic Size-spectrum (BOATS) model.  The main improvements are that BOATSv2 now resolves a distinct benthic pathway for delivering energy to demersal and benthic fisheries and includes spatially variable fishing costs and catchability.  They also integrate iron-dependent fish growth rates and fishing effort targets that were developed in previous work.  After re-calibrating the model, BOATSv2 performed similarly for coastal systems showed marked improvement in its representation of high seas fisheries catch, which was significantly over-estimated by BOATSv1.  The addition of the benthic pathway was the primary driver of this improvement, with spatially variable fishing costs and catchability providing secondary yet notable further improvement.

I found the BOATSv2 improvements documented in this paper and the resulting improvements they yielded in high seas catch to be significant.  The approaches for parameterizing the array of ecosystem, fishing and economic factors should be of general interest to the community, as should the approach to evaluation and optimization.  I did, however, find some aspects of the presentation challenging.  While I was generally convinced that the approaches were reasonable, there were cases where limited and/or gaps in the model description left me with questions.  I also felt that the organization could be improved.  My two main comments are thus:

1) The presentation of BOATSv1 in Section 2.1 needs improvement.  I understand that the authors don’t want to spend too much time reviewing previously published model dynamics but the reader still needs to understand the BOATSv1 foundation to follow the rest of the analysis herein.  My advice is to give yourself an extra ~50 lines or so and put yourselves in the shoes of someone who has not read the Carozza article.  Clearly define all of the parameters and quantities you mention and provide enough narrative to give the reader a quantitative and qualitative understanding of the model dynamics.  I have tried to provide specific suggestions below that I hope are useful.

2) While the conclusions were ultimately clear, the presentation of Results needs improvement.  For example, when I reached the Results section (section 5) I was surprised because the primary results of the study (e.g., the improvement in fidelity with high seas fish catch) had just been presented in Section 4.  I also thought the results may contain too many detours and details that risked blunting the main conclusion.  Please consider describing the optimization methodology in the methods, but moving the optimization results to the results section.  The intermingling of methods and results may have contributed to my sense that there were too many detours and details, but a final round of editing for brevity and focus would be beneficial.  Page 18-19, for example, had a lot of material that, while interesting, was secondary to the primary messages.  Again, I have tried to provide specific suggestions for your consideration below.

Overall, I think this is a substantive paper that documents meaningful model advances and skill improvements that will be of general interest to the modeling community and serve as a valuable reference for future BOATS applications.  I also think, however, it would benefit from a final round of edits for clarity and brevity to ensure that it has the impact it should.  Please view my comments in that constructive spirit.

Specific comments:

Abstract: Clarify that the novel features described starting on line 7 are novel to BOATS but not necessarily novel in the field.  You need to define large marine ecosystems for the uninitiated.  Also, there seems to be a wonderful opportunity to state the factors that were responsible for the model improvements after line 13.  Please take it!

Line 41: Define LMEs and include a reference so that readers understand this definition.

Section 2: As described in general comment 1, I found this section challenging.  I would allow yourself more space to present these core dynamics clearly. I have provided a few specific suggestions to improve it that I hope are useful:

Table 1 currently includes only a limited subset of the parameters and quantities discussed in this section and a lot of detail on optimization procedures that aren’t described until much later.  The parameter/quantity definitions are what the reader needs now.  Please provide them for all the parameters/quantities discussed in this section and save the additional details of the optimization for when the reader needs them.

I didn’t understand why you chose to include the growth/recruitment function for the smallest size class with the MvF equation on line 80.  The general growth expression is given later and the most natural place to deal with the recruitment function seems to be around line 100 in the current text.  I suggest you start with MvF and then unpack the growth functions as they arise in the text in a consistent manner.

Equation (2): should the (1-Phi_k) be carried over to the right-most quantity?

Line 90-99: Improved representation of the energy flow between phytoplankton and fish ends up being one of the major required improvements between BOATSv1 and BOATSv2.  Fully understanding this important change requires a clearer description of the BOATSv1 parameterization.  The productivity symbols need clearer definitions, the shape and rationale for the energy spectrum and the trophic scaling need to be more clearly described.  I could not find any description of how the characteristic size of the phytoplankton was determined.  It is unclear which groups NPP is being partitioned across and why assuming that it is even is sensible (e.g., if one group has much higher biomass than another, wouldn’t it make sense for more NPP to go to the one with higher biomass?).  Please expand this section as needed so that the reader understands how growth and energy flow constraints were handled in COBALTv1 so they can fully understand how these change in COBALTv2.

Line 111-119: Would this paragraph be better saved for a discussion of the optimization procedure?

Line 112: Are these natural logs or base 10?

Line 135-138: It is difficult to understand what is being described here without seeing the relationship and where the parameters sit within it.

Please ensure that each parameter in eqs. (5) and (6) is clearly defined.

Figure 1: Please expand this caption so that the reader understands what is plotted.  It also seems like colorbars are needed in several places.

Line 164, Section 3.2: “Novel” is a tricky word to use.  Do you mean new features relative to prior versions of BOATS?  Be more precise with what you mean here.

Eq. (8): What does the superscript “corr” indicate?  This occurs later in the text as well in association with other quantities, but I was never completely sure what it was meant to indicate.

Eq. (10): Are the costs additive?

Line 215-217: How was the size of the phytoplankton set?  I don’t think this was ever mentioned in Section 2.

Line 235: Was there a rationale for choosing 370 km?

Figure 2: As in figure 1, please expand the caption to make the meaning of this figure clear.  It gives the impression, for example, that the selection of 11 parameters comes from BOATSv1.  I don’t believe this is the case.

Line 359-377: The description here related to the calibration results and the performance of the calibrated model against observed patterns seems like a Result (see general comment 3).  Line 376-377, for example, reveals perhaps the most prominent result – the improvement in high seas catch.   I would consider describing the calibration procedure in the methods and moving the Results to the Results section where they can be concisely presented after digesting the methodology. 

Section 4.3 also suffers a bit from this mix of Methods and Results.  Also, following general comment 2, there are many details and detours on pages 18-19.  While each may be interesting, there is a risk of pulling attention away from the main messages.  Part of this may be addressed with a clearer separation of Methods and Results, but I would also encourage the authors to think carefully about how to present the results they feel are most critical as concisely as possible.  There also seem to be a number of small discrepancies between values listed in the text and those in Table 3.  Please ensure that these are synchronized.

Line 393-394: I’m not sure what you mean by: “This suggests that analogous parameterizations of heterogeneous costs and catchability will generate comparable variability in LME catches.”

Line 395: Assuming this is relative to Watson, should BOATSv2 be 0.64?

Line 404-405: The effect of heterogenous costs seems quite small on the fraction of high seas catch.  I found this surprising, and I could not find this result in Table 3.  It looks like the high seas fraction with heterogenous costs is ~0.14, not 0.3?

Results and Discussion section: This seems out of place.  Haven’t most of the primary results have already been revealed in the prior section (see general comment 3).

Line 445: The difference in the mortality constant seems to merit some additional discussion.  How should the reader interpret this ecologically?  Perhaps you could close the loop on this issue on line 464 where compensating effects are discussed?  Finally, you may want to clarify that a negative value of this parameter, which I understand is an exponent, does not imply negative mortality?

Line 467-469: Should the covariance of the pelagic and demersal temperature dependence be interpreted as an indicator that two parameters may not be needed (i.e., demersal and pelagic species exhibit the same response?).  I was curious why you chose to allow a different temperature dependence for pelagic and demersal.

Line 488-89: A reference to Patrick Lehodey’s SEAPODYM work seems like it could be useful here?

Line 502-504: Do you think that unresolved effects of hypoxia may play a role in the eastern tropical Pacific