General comments

In their manuscript bg-2022-155, Wen et al. present the results of a series of nutrient additions experiments conducted in the South China Sea and western North Pacific, where in addition to bulk N2 fixation rates, species composition based on nifH gene abundance was analyzed in response to Fe and P amendments.

Overall, I enjoyed reading the manuscript, and I find the dataset is a useful addition to our understanding of the regulation of N2 fixation rates and how it is linked to species composition. The manuscript is well written, and the discussion insightful. Yet, there are a few points that I believe should be discussed/improved, specifically I have some concerns/queries with regard to replication and interpretation of qPCR results, as outlined below.

Specific comments

• Apparently, in several of the experiments there were only 2 replicates – I wonder whether the statistical analysis procedures are valid for two replicates? Could the authors at least indicate in each of the figures which data are averages of 2?
• The authors acknowledge that 15N label% was not measured, which is indeed a shortcoming, but as they state that the experimental procedure and the results were comparable to their previous study I believe it is acceptable. An average and stdev of label% in the previous study are given, but could the authors add any further details to help us understand how reproducible the approach was (number of replicates etc)?
• I wonder how the reports on Trichodesmium polyploidy (ca 100 genome copies per cell in field samples, e.g. Sargent et al. 2016 https://doi.org/10.1093/femsle/fnw244) affect the estimates of species composition based on nifH gene copies, as well as the trends observed in bioassays. Was polyploidy taken into account when the ‘dominant species (e.g., l. 351)’ were determined? And, taking into account the high level of polyploidy in Trichodesmium compared to the other species, couldn’t shifts in the species composition explain the mismatch in responses of N2 fixation vs nifH abundance (e.g., l. 343)?
• Related to this, I would suggest being more cautious about the use of the terms ‘abundance’ (e.g. abstract l. 26 ‘abundances of specific diazotrophs’, l. 119 ‘abundances of specific diazotroph phyla’) and ‘growth’ where actually gene copy number was measured. Specifically, in supplementary figure 1, I would suggest replacing ‘growth rate’ by ‘gene abundance’, since growth rate might imply that measurements of cell density or C concentration were made.

Minor comments

• 34-35 ‘the largest responses were always dominated by either Trichodesmium or UCYN-B’: this is not clear (responses in what?) – can it be clarified?
• 110-112 it is not completely clear from this why high spatial resolution is necessary - can this be justified better?
• 164 can more details on the gas-tight plastic bags (supplier) be added?
• 221 why were those Fe and P concentrations chosen – are there any references to add on how these relate to in situ concentrations in this area?
• 223 Can the authors supply some more details on the incubation system? Do I understand correctly that 10L carboys were placed in the on-deck incubator, and there was some kind of water jacket flushed with seawater for temperature control?
• Please supply more detail or a reference on the method for chl measurement
• 441 I believe the biochemical substitution of Fe and P deserves some more explanation (either here or at a later stage) - how could this work, are there specific mechanism/enzymes that can substitute Fe for P?
• 450-451 also the ‘serial limitation’ of N2 fixation by another resource deserves a few more words for explanation I believe – it is not clear how this would work
• 476 I think there might be more specific refences for the Fe demand of Trichodesmium (e.g., Kustka et al., https://doi.org/10.1016/S0923-2508(02)01325-6, 10.4319/lo.2003.48.5.1869, https://doi.org/10.1046/j.1529-8817.2003.01156.x)
• Could it be specified how the data in Fig. S1 was calculated? Is the growth rate in units d-1 the total increase in gene abundance over the 3 day bioassay experiment divided by 3?
• In the supplementary table showing parameters involved in nitrogen fixation rate calculations, the units for several of the parameters (e.g., depth, chl a, PON, NFR) are missing in the headers, please add these.

Technical corrections

• 62-63 check the order of references: in sequence of publishing year?
• 69 *did* not quantitatively match
• 72 *is* potentially crucial
• 76 I assume this should mean ‘contribute differently to the sinking flux carbon that small unicellular species (i.e., delete ‘that of’)’?
• 110 remain*s*
• 242 metal bound (remove hyphen)
• 266 below
• 313 and others: remove brackets around station names in the text

In this paper, Wen et al. reported the patterns and limiting factors of nitrogen fixation activity and diazotroph community in the South China Sea and western North Pacific. Through the iron and phosphate amendment experiments, they delineated the nutrient limitations of nitrogen fixation and diazotrophs for different water bodies in the study region. These findings may facilitate our understanding of the bottom-up controls of the diazotrophs in the study region, while I have some concerns regarding the methods, interpretations, and discussions in the current manuscript.

This manuscript gives me a general impression that the authors kept mentioning another paper of their group (Wen et al., 2022), instead of sticking to the findings of the current study. For example, they discussed a lot about the iron to nitrogen supply ratio, which seems like a highlight in Wen et al., 2022, but this ratio was not analyzed in the current study. Also, the dissolved iron to dissolved inorganic nitrogen ratio was not available in this manuscript. I suggest the authors focus more on the direct and interactive effects of iron and phosphorus availability on the diazotrophs for the discussion.

Given that the iron and phosphorus limitations of diazotrophs are the major focus of this study, the authors should describe the iron and phosphorus availability in the study region based on their real data, like ambient concentrations of iron and phosphate, instead of speculating. Also, the iron and phosphate concentrations in the nutrient amendment experiments should also be reported. This information is particularly important when the authors discuss the reasons and biogeochemical implications for the experimental results.

Besides, the authors should be aware that they only analyzed some commonly observed diazotroph groups (i.e., part of the diazotroph community) with qPCR assays. In other words, some of the unanalyzed diazotrophs might be important N-fixers at some stations. Some unanalyzed diazotrophs might even pop up during the 3-day incubation. So, they need to be cautious when comparing the patterns of nitrogen fixation rates and diazotroph abundances. The authors may also consider conducting nifH amplicon sequencing for reconstructing the whole diazotroph community in the study region. Also, initial diazotroph abundances of the incubations should be reported as well.   

L28-30: It is better to avoid hypothesis/speculation in the abstract. The iron to nitrogen supply ratio was not directly measured in this study as the authors stated in L416.

L34-L35: “the largest” and “always” seem subjective and inaccurate based on Figure 5. Also, there was no significant response at some stations where Trichodesmium dominated.

L38-40: Why? I did not see any evidence from this study supporting this speculation directly.     

L136-137: This sentence is not informative, as the depths are not labeled in the figure. Nevertheless, the depths of the seafloor are not important here.

L152-156: Were the water samples collected from different depths (2-5m) exposed to different degrees of dissolved iron contamination from the research vessel? Did the authors measure dissolved iron concentration for these water samples?    

L160: The results of primary production were not described or discussed. Was primary production also measured in the nutrient amendment experiment? The Chl-a and primary production from the experiments may be helpful when the authors discuss/speculate about the competition between diazotroph and non-diazotrophic phytoplankton (L28; L403).

L226: Did you measure iron concentration in the 15N2 enriched water? The preparation of 15N2 enriched water may introduce iron contaminants (Klawonn et al., 2015).

Klawonn, I., Lavik, G., Böning, P., Marchant, H. K., Dekaezemacker, J., Mohr, W., & Ploug, H. (2015). Simple approach for the preparation of 15− 15N2-enriched water for nitrogen fixation assessments: evaluation, application and recommendations. Frontiers in microbiology, 6, 769.  

L307: The description of S3 is confusing. Does “ab” in Figure 4 mean no significant difference with a and b? If that’s the case, S3 seems not “independent co-limited”. Please clarify. 

L317-323: Please clarify the exact number of replicates for each treatment. Also, I doubt the statistical significance based on duplicates (n=2). The limitation of replication should be stated clearly in the manuscript. It is also the same for Figure 5. Were initial nitrogen fixation rates (i.e., the rates of the seawater from the pump) measured?  

L350: As you only analyzed part of the diazotroph community, you may consider revising “diazotroph community structure” to “abundances of analyzed diazotrophs”.

L367 and Figure 5: How about UCYN-A1? UCYN-A1 was also abundant at K1 and WP based on Figure 3, while they disappeared in the nutrient amendment experiment (Figure 5). Also, the initial diazotroph abundances should also be displayed in Figure 5.

L382: There is no doubt about Kuroshio being a hotspot of nitrogen fixation, while the low rate at K1 is not the “increasing evidence” as stated here.

L385: Abundances of diazotrophs do not necessarily mean their contribution to nitrogen fixation.

L412-431, 453-464, etc.: The contents (mostly iron to nitrogen supply ratio) of Wen et al. 2022 are worth mentioning, but, they should be reduced significantly in the discussion. As said, the iron to nitrogen supply ratio was not analyzed in this study.

L440: I think another reason would be that the analyzed groups did not represent the entire diazotroph community.  There could be other diazotroph groups, which were not analyzed in this study, influenced by treatments. 

L473-475: However, the nitrogen fixation at S3 and S4 was mostly iron-limited, while the Trichodesmium abundances there were not affected by iron addition treatment. All these pieces of finding should be considered when you discuss the relationship between iron and Trichodesmium in the NSCS.