Review of Li et al.

In this manuscript, Li et al have implemented benthic denitrification using the empirical function derived by Bohlen et al (2012) on Uni-Vic ESM, and parameterized using chl, O2, NO3, PO4, N* observations. Relatively speaking they found considering N* to constrain their model has given better model performance when compared to the observations from WOA13. They found that this implementation has resulted in better simulation of N2 fixation patterns in the world ocean, and more realistic simulation of ODZs, suggesting that inclusion of benthic denitrification in ESMs are a key piece to improve model performance.

I found the procedure and inferences are sound. This work also highlights the importance of consideration of benthic processes in ESMs which are currently underrepresented, and its potential in improving model predictions. I have no major concerns with the present study; however, I feel the following comments will increase the readability of the paper.

Minor comments:

Line 23: replace “however, even though” with in spite of.

Line 33: “..thus will be..”

Line 35-39:  maybe expand a bit on this, exactly what you are going to test?

Line 55: “..description of the..”

Line 75: Please define what "H" version is meant for, and briefly mention how it is different from OPEM.

Line 170-171: it would be good to expand a bit more on this that why this might be the case. Give some perspective.

Line 180-181: Expand a bit more on the export production. What percentage of NPP is exported to the deep ocean and how does it compare to the estimates elsewhere.

Line 197: “fixed-N nitrogen” only one nitrogen.

Line 279-280: Not clear to me. Please explain. In figure 8, Calib_N* - no_bdeni values are mostly positive, that means Calib_N* has more O2 in the ocean, I see that in figure 6 as well. But how does the prediction of ODZs in no_bden case. If is a bit confusing, please clarify in the text.

287-298: It will be nice to add a third row of figures on figure 9 plotting the EP ratio. Also, can you comment on the EP ratio how it compares to any other literature (observation or modeling)? It is good to put this information as the rain rate to the seafloor will ultimately drive the denitrification in the model.

305: I think you have not put any number of observational estimates of N fixation, if not please provide some number in relevant places.

Define what is biome when it first occurs. If Arbian sea has one, how does it compare with other modelled areas/seas?

325-326: Define what is w_d0 and w_dd, when it appears first.

339-340: first part of this sentence is not clear.

345: “demonstrates”

Figure 2: Any reference for NPP to compare with?

Figure 5: explain why you did not include Indian ocean and southern ocean.

Figure 6 caption: Typo.

Figure 8 caption: three solutions or N*?

Table 1 caption: units. Typo.

Table 1: did you get the reference range from somewhere? please clarify.

Table A1: Why do you provide bar on top of TN? It is not defined in the table.

Review of “Global impact of benthic denitrification on marine N2 fixation and primary production simulated by a variable-stoichiometry Earth system model”

This manuscript describes the improvement of an Earth system model by incorporating benthic denitrification, highlighting the importance of benthic denitrification in shaping the global distributions of NPP, N2 fixation, oxygen, etc. In addition, the authors conducted a large ensemble of simulations and applied the Latin-hypercube sampling method to choose the model parameters. The simulation identified that the N* is an essential parameter to calibrate the model, which has an important implication for future model calibration. Overall, the manuscript was well-written and organized, and the references cited are up-to-date and appropriate. I recommend publication after making the following clarifications and modifications.

I'd like to see more discussion on the pattern of N2 fixation. Currently, the distribution patterns in the S. Pacific and N. Indian Ocean (Fig. 4a) seem not correct to my eyes. What causes such distribution, is there anything to do with benthic denitrification?

In section 4 "Model and calibration limitations", the authors discussed multiple ways to improve the model, including applying different parameterization schemes for upper and deeper ocean sinking speed and resolving the dependency of remineralization rate on O2. The methods described are all reasonable. But why not try them, this is part of model development.

Some minor suggestions

Line 83 (Somes and Oschlies, 2015) --> Somes and Oschlies, (2015)

Line 92 (Pahlow et al., 2020; Chien et al., 2020) --> Pahlow et al., (2020) and Chien et al., (2020)

Line 149 Best models choices --> Best model choice.

Line 305 Arabian --> Arabian Sea