The authors present a supremely well executed study of N cycling rates in an oxygen deficient zones from well-controlled tracer incubations, from which they derive the relative contribution of respective processes to N2O production, and from which they document the sensitivity of said production pathways to dissolved oxygen concentrations. Their tracer incubations rely in part on site-preference measurements of isotopocules in order to determine pathways of production. Their data corroborate a dominance of denitrification in N2O production within the anaerobic regions of the water column, whereas multiple pathways operate concurrently in oxyclines. N2O production from ammonium, presumed to be catalyzed by nitrifiers, occurred dominantly through a hybrid pathway reliant on both ammonium and nitrite as substrates, whereas the hydroxylamine pathway (both N’s in N₂O from ammonium) was relegated to the well-oxygenated upper water column. The results and interpretation are highly informative, providing important constraints on pathways of N2O production and their respective sensitivity to oxygen.

I found the manuscript generally well written but, perhaps necessarily, a challenging read. I read it multiple times. The “cognitive challenge” arises from the inherent complexity of the topic and study design. It is also exacerbated by some structural elements of the manuscript that would benefit from revision: (a) The motivations for the study are not made clear in the introduction; (b) the general “order of operation” keeps jumping around in the results and discussion (I explain what I mean below), (c) there is a heavy reliance on supplementary materials, requiring a lot of back and forth.

I suggest a number of modifications that I think could improve ease of understanding by readers peripheral to the field of N2O isotopes who want to understand the findings and who also want to have a sense of the limitations of the findings.

Thank you for taking the time to thoroughly read and understand our paper, and for your constructive feedback. We have restructured the paper according to your suggestions and hope that it is easier to follow as a result.

The introduction does not effectively motivate the study. This study appears to be a companion to a published study where net rates of N cycling were determined from bulk tracer additions. I suppose that is why the bulk rate estimates figures were relegated to the supplements even though they are highly informative in the current context. Regardless, questions evidently emerged from the previous study that are presumably addressed herein, but these questions are not articulated in the introduction. I suggest the following paragraph sequence, which would make the intro more seamless:

The first paragraph alerts us that the study deals with nitrous oxide in oxygen deficient zones, with a justification of why N2O matters. In the second paragraph, the reader expects to learn where N2O is believed to come from in ODZ’s. Instead, the paragraph otherwise begins with what seems a separate (but related) topic, N2O production by archaea, ocean-wide, not necessarily in ODZ’s. In lieu, I suggest moving up the third paragraph to the second, to explain the current understanding that most N2O in ODZ's appears produced by denitrification. This would lead into a third paragraph that explains that nonetheless, a significant fraction appears to be produced by archaeal nitrification. I would present the current evidence that supports this hypothesis, in order to motivate “looking” for hybrid production, which is where this paper ultimately brings us.

Thank you for this helpful suggestion. We moved up paragraph three of the introduction (N₂O production via denitrification) and revised (formerly) paragraph two to focus more on motivating our discussion of hybrid production.

The fourth paragraph should be explicit in whether it is referring to naturally occurring isotopes or tracer isotopes, since the subsequent paragraph jumps into tracers. To better motivate the study, perhaps this section can explain what naturally occurring isotopocules have divulged about N2O production in ODZ’s specifically, and which questions remain unanswered – in order to link to the last paragraph of the intro.

We made it more explicit that paragraph four is about natural abundance isotopes. We also revised it to focus on the fact that hybrid N₂O production complicates the interpretation of natural abundance δ(¹⁵N^sp) because it draws from two different substrate pools.

IN the last paragraph, the motivation for measuring site preference on tracer experiments needs clearer articulation. What additional insights can it provide that natural abundance or bulk tracer experiments did not? And your results, as I see them, inform on more than a dependence of oxygen on hybrid production, correct? They (a) corroborate previous findings on relative pathways of N2O production (b) uncover that the hybrid pathway dominates production by nitrification and (c) production from hydroxylamine is not a thing except at the surface. Importantly, do the results confirm inferences from natural abundance tracers in the same system? These can be posed as questions to which the authors can return in the discussion. 

We added a sentence to the last paragraph saying that ⁴⁵N₂O^α and ⁴⁵N₂O^β measurements create an additional constraint on N₂O production rates and thus allow us to quantify different source process more precisely and accurately. As per your suggestion, we also detailed more thoroughly the different findings from this study.

Methods:

Line 200: I would rephrase to “…. contribution of 15N15NO to masses 46 and 31, which, while negligible at natural abundance, becomes important in tracer experiments.”

Corrected.

Equations 1-4: I think it would be wise to define ALL the terms in equations 1-4, for readers peripheral to this field who may still strive to understand the equations.

Corrected.

Line 245: Nitrate IS produced from nitrite when sulfamic acid (or any acid) is added to nitrite, due to the acid decomposition of nitrous acid. See Granger and Sigman 2009, Equations 6 and

Figure 2. And 15N nitrate is a probable contaminant of the 15N nitrite solutions.

We revised this section to say that our high t0’s are likely because NO₃^- is produced when sulfamic acid is added to NO₂^-(Granger and Sigman, 2009), so the sulfamic treatment probably chemically converted some ¹⁵N-NO₂^– tracer to ¹⁵N-NO₃^–; additionally, ¹⁵N-NO₃^–  is a probable contaminant of the ¹⁵N-NO₂^– tracer solutions.

Line 274: what is N exchange between substrates?

Sorry, “exchange” is probably the wrong word here. We have changed it to N transfer between substrates.

Line 280: These “pathways” were not discovered by Wan et al. 2023. The citations are unclear to me.

We changed these citations to “labeled as Pathway 1 in Wan et al., 2023…”.

Results:

I realize some of the data are published elsewhere but they are fundamental to navigating the paper. I suggest moving some of these back to the main text. In particular, the N2O production plots (mass 45 for each 15N substrate).

To clarify: none of the data included in this study have been published elsewhere. A companion paper (Frey et al., 2023) published rates of ammonia oxidation and N₂O production from ammonium measured in concurrent, but separate, experiments. Nevertheless, we have moved the ⁴⁵N₂O and ⁴⁶N₂O production plots into the main text. They are now figures 4 and 5.

I suggest presenting the results in order of dominance of rates, and sticking to this pattern in all subsequent text and figures. Denitrification is fastest; detailing it first helps contextualize nitrite oxidation rates, which are also very high, and ammonium oxidation rates, which are puny.

We changed the order of section 3.2, “Nitrification and nitrate reduction rates,” to talk about denitrification first, then nitrite oxidation, then ammonia oxidation.

Stick with one, NH3 or NH4 oxidation. It varies in the text.

We changed all of these to NH₃ oxidation.

Section 3.3 is very difficult to navigate. I read it multiple times. The term “high rates” is meaningless without context. Rates peak or not, but it can’t be argued that rates of 45N2O-alpha are high even in this context, at picomolar per day. In this regard, I suggest using picomolar in lieu of multiple decimals in the text and figures, which are tiresome. And the Figure S8 is nearly impossible to navigate as every panel has a different x axis range. Perhaps homogenize ranges for given isotopocule production? And I’m not sure why these figures are relegated to the supplements. I spent a long time looking at them. A long time…

• You’re right, in this section “high rates” is relative. We revised “high rates” to “relatively higher rates.”
• We changed all of the N₂O production rates from nM/day to pM/day.
• We homogenized the x-axis ranges for Fig. S8 as much as possible while still allowing the variation in each panel to be visualized.
• We moved Fig. S7 and Fig. S8 to the main text. They are now Figs. 4 and 5.

The line at 215 belongs with the previous paragraph. And it’s not clear whether this will be an example of rates varying in concert or not. Wordsmith accordingly.

Did you mean a different line? 215 is just after eqn. (6), “where ¹⁵N¹⁵N¹⁶O_excess represents the amount of ¹⁵N¹⁵N¹⁶O produced in the sample over the course of the experiment.”

Equation 13: In the case of nitrite where a higher concentration was added then intended, I would think that the flux derived therefrom, J, is no longer proportional to nitrite (zero order) at these concentrations. Does this matter?

If we compare the ¹⁵N-labeled ammonium treatment to the ¹⁵N-labeled nitrite treatment at the same experimental depth, the ⁴⁵N₂O and ⁴⁶N₂O production rates in the ¹⁵N-labeled nitrite treatment were far higher than those in the ¹⁵N-labeled ammonium treatment, even when normalized by atom fraction. This is visualized below. In fact, the rates of production of ⁴⁵N₂O and ⁴⁶N₂O in the ¹⁵N-labeled ammonium treatments were so small, comparatively, that they are visually indistinguishable from zero when plotted on the same scale as the rates of production of ⁴⁵N₂O and ⁴⁶N₂O in the ¹⁵N-labeled nitrite treatments.

Production of ⁴⁵N₂O, divided by atom fraction, in the ¹⁵N-NO₂^- treatment vs. ¹⁵N-NH₄⁺ treatment at the same experimental depths. Red diamonds indicate p⁴⁵N₂O^a/¹⁵Fand black diamonds indicate p⁴⁵N₂O^b/¹⁵F. b) Production of ⁴⁶N₂O, divided by atom fraction squared, in the ¹⁵N-NO₂^- treatment vs. ¹⁵N-NH₄⁺ treatment at the same experimental depths. In both plots, the dashed line is the 1:1 line.

Since the tracer concentration was much higher in the ¹⁵N-labeled nitrite treatment (5.00 µM) than in the ¹⁵N-labeled ammonium treatment (0.501 µM), this imbalance of ⁴⁵N₂O production supports the idea that there is some dependence of N₂O production rate on substrate concentration.

Line 337: Wording of sentence is awkward.

Revised to “The model solves for N₂O production rates, given a set of NH₄⁺ oxidation, NO₂^– oxidation, and NO₃^– reduction rates calculated in Sect. 2.5, eqn. (7) (Table S2).”

Line 395: How can nitrite oxidation rates possibly be negative?

The “negative” nitrite oxidation rates at two depths are likely an artifact of the elevated t₀d(¹⁵N) values in some of our ¹⁵N-NO₂^- treatments (discussed above). We have added this to the text.

Line 420: Remind the reader what “f” designates.

Done.

Equation 19: “AP” was designated as “15F” in equations above…

Changed to ¹⁵F.

Could p45excess result from misestimation of the actual atom percent of substrates the incubations? The rates are very small such having a small error on AP could potentially account for this? Or wrong proportion of carrier? I think Figure S9 may allude to this but the associated uncertainty needs to be better explained in the main text, whether or not the data evince unequal values of “f” beyond a reasonable “doubt”

Figure S9 (now Fig. S S7) alludes to this. The dashed lines in Figure S9 indicate the range of atom fractions in each type of experiment, which far exceeds the uncertainty in the atom fraction of any one individual experiment. So points above the dashed line indicate excess ⁴⁵N₂O production, beyond a reasonable doubt.

Figure 4: Present in order brought up in text, which is N2O production from nitrate first.

Is production from NH4+ only necessarily hydroxylamine oxidation? It is called that in some figure captions. If so, it would be much easier for readers if it were called hydroxylamine oxidation throughout.

The order of this figure (now Figure 6) has been changed. Sorry, “hydroxylamine oxidation” was a mistake — N₂O from NH₄⁺ could also include hybrid production using an internal NO₂^- pool. We have revised the figure captions to “N₂O production from solely NH₄⁺”.

Section 3.5: I would start with describing N2O production “as a whole”, followed by nitrate reduction (highest flux), etc… Same order of operation as suggested above.

We changed the order of section 3.5 to discuss N₂O production from nitrate first. We also changed the corresponding section of the discussion (Section 4.4).

Figure 4 d: the trace for ammonium oxidation differs from the corresponding trace in Figure 3 a.

Thank you for catching this. Figure 3a is correct. Not sure what happened with Figure 4d (now Figure 6d) but we corrected it.

Discussion:

Because the study is very complex, it would be beneficial for the discussion to begin with a paragraph that summarizes the dominant findings, rather than jumping into the deep end form the get go. In this regard, I would also get N2O production from denitrification out of the way first because it was the dominant flux, then discuss hybrid production. I find it interesting as well that production from hydroxylamine was virtually absent except at the surface – I think this merits more emphasis.

Thank you for this suggestion. We added a summary paragraph at the beginning of the discussion, and we changed the order of the discussion to 1) N₂O production from denitrification, 2) hybrid production, 3) production from solely NH₄⁺.

Section 4.3: I get that MOST N2O is produced by denitrification and 1/5 from hybrid production. Is that what is also inferred from natural abundance measurements, in these proportions? Curious minds want to know

Yes, this is indeed what we inferred from natural abundance measurements. Based on natural abundance site preference, we found that the near-surface [N₂O] maximum in was likely to be comprised of ∼20% N₂O produced via nitrification or archaeal N₂O production and ∼80% N₂O produced via denitrification (Kelly et al., 2021). We added this to the beginning of section 4.3.

Line 642: What do you mean by “allowed?” Need better wording.

Here we’re alluding to natural abundance measurements indicating that N₂O production from NO₃^- could be an important source of N₂O in the anoxic core of ODZs, as long as it has a positive δ(¹⁵N^sp). As you know, denitrification is usually assigned δ(¹⁵N^sp) ≈ 0‰ (Sutka et al., 2006), but some strains of denitrifying bacteria can produce N₂O with δ(¹⁵N^sp) > 0‰ (Toyoda et al., 2005; Wang et al., 2023). And so can denitrifying fungi (Sutka et al., 2008; Rohe et al., 2014; Yang et al., 2014; Lazo-Murphy et al., 2022). So, given that there are several potential sources of N₂O production from NO₃^- with a positive δ(¹⁵N^sp), the importance of N₂O production from NO₃^- in this study agrees with natural abundance work.

Line 650: qualify “this” , you mean the notion that internal pool are processed, not external…?

Yes, exactly. We changed “this” to “N₂O production from NO₃^- that utilizes an internal NO₂^- pool”.

Line 600: Reader is left hanging: What are the implications for mechanisms of production? Need a concluding sentence for the paragraph to bridge it to the next, or simply amalgamate with the following paragraph.

We re-wrote this paragraph and the following text to reflect the fact that most of our experiments actually have equal formation of ⁴⁵N₂O^α and ⁴⁵N₂O^β, and thus f = 0.5, which would imply that hybrid δ(¹⁵N^sp) would not vary in most of the tested conditions.

Paragraph at line 605: Reads like something that should be in results section.

We moved this text down to our paragraph where we address the unequal production of ⁴⁵N₂O^α and ⁴⁵N₂O^β at certain depths, which anchored significant relationships between f and ambient [O₂] and potential density anomaly. The oxygen and potential density gradients may be proxies for changing archaeal community compositions at different depths in the water column, which may exhibit different patterns of incorporation of NO₂^—-derived N and NH₄⁺-derived N into N^α and N^β. It is also possible that we sampled a different “hybrid” N₂O-producing process at these depths, such as fungal co-denitrification (Shoun et al., 2012), which may proceed via a different pathway from archaeal hybrid N₂O production.

Line 610: Articulate fully for readers to catch up again “findings of unequal alpha vs. beta production during hybrid pathway have implications for interpretation of the natural abundance isotopes of N2O produced by hybrid process.”

We now write that “the equal formation of ⁴⁵N₂O^α and ⁴⁵N₂O^β led to values of f within error of 0.5 in most of our experiments (Table S4), and the mean value of cf across all stations and depths was 0.5±0.2. This means that during hybrid N₂O production, half of the N^α atoms were derived from NO₂^–, and half were derived from NH₄⁺ (likewise for N^β). These findings of equal ⁴⁵N₂O production have important implications for the natural abundance δ(¹⁵N^sp) of N₂O produced by the hybrid N₂O process...”

Paragraph at line 670: I don’t understand why the results here should be different than cited study.

Ji et al., (2018) did not include hybrid N₂O production in their estimates of N₂O yield. We added this to the text.

I remain perplexed by the following: In Figure S8, there is NO production of ⁴⁵N₂O from addition of ¹⁵NH₄⁺at 100 m at station 1, yet there is reportedly 50 nM/day N2O production from the hybrid pathway at this depth… Am I fundamentally misunderstanding something about the experimental design? The hybrid pathway requires some input from ¹⁵NH₄⁺which should be detected as ⁴⁵N₂O?

We can understand why this would be confusing. The model solves for the same rates of hybrid N₂O production in the ¹⁵NH₄⁺ and ¹⁵NO₂^- experiments. In this case, there is high ⁴⁵N₂O production in the ¹⁵NO₂^- experiment but very little ⁴⁵N₂O production in the ¹⁵NH₄⁺, so the model finds an intermediate value. Given that the ¹⁵N-NO₂^- spike was added at a higher concentration (5 µM) than the ¹⁵N-NH₄⁺ spike (0.5 µM), it is feasible that the ¹⁵N-NO₂^- generated a greater ⁴⁵N₂O signal than the ¹⁵N-NH₄⁺ experiment.

References

Granger, J. and Sigman, D. M.: Removal of nitrite with sulfamic acid for nitrate N and O isotope analysis with the denitrifier method, Rapid Commun. Mass Spectrom., 23, 3753–3762, https://doi.org/10.1002/rcm.4307, 2009.

Ji, Q., Buitenhuis, E., Suntharalingam, P., Sarmiento, J. L., and Ward, B. B.: Global Nitrous Oxide Production Determined by Oxygen Sensitivity of Nitrification and Denitrification, Glob. Biogeochem. Cycles, 32, 1790–1802, https://doi.org/10.1029/2018GB005887, 2018.

Kelly, C. L., Travis, N. M., Baya, P. A., and Casciotti, K. L.: Quantifying Nitrous Oxide Cycling Regimes in the Eastern Tropical North Pacific Ocean With Isotopomer Analysis, Glob. Biogeochem. Cycles, 35, e2020GB006637, https://doi.org/10.1029/2020GB006637, 2021.

Lazo-Murphy, B. M., Larson, S., Staines, S., Bruck, H., McHenry, J., Bourbonnais, A., and Peng, X.: Nitrous oxide production and isotopomer composition by fungi isolated from salt marsh sediments, Front. Mar. Sci., 9, 2022.

Rohe, L., Anderson, T.-H., Braker, G., Flessa, H., Giesemann, A., Lewicka-Szczebak, D., Wrage-Mönnig, N., and Well, R.: Dual isotope and isotopomer signatures of nitrous oxide from fungal denitrification – a pure culture study, Rapid Commun. Mass Spectrom., 28, 1893–1903, https://doi.org/10.1002/rcm.6975, 2014.

Shoun, H., Fushinobu, S., Jiang, L., Kim, S.-W., and Wakagi, T.: Fungal denitrification and nitric oxide reductase cytochrome P450nor, Philos. Trans. Biol. Sci., 367, 1186–1194, 2012.

Sutka, R. L., Ostrom, N. E., Ostrom, P. H., Breznak, J. A., Gandhi, H., Pitt, A. J., and Li, F.: Distinguishing Nitrous Oxide Production from Nitrification and Denitrification on the Basis of Isotopomer Abundances, Appl. Environ. Microbiol., 72, 638–644, https://doi.org/10.1128/AEM.72.1.638-644.2006, 2006.

Sutka, R. L., Adams, G. C., Ostrom, N. E., and Ostrom, P. H.: Isotopologue fractionation during N2O production by fungal denitrification, Rapid Commun. Mass Spectrom., 22, 3989–3996, https://doi.org/10.1002/rcm.3820, 2008.

Toyoda, S., Mutobe, H., Yamagishi, H., Yoshida, N., and Tanji, Y.: Fractionation of N2O isotopomers during production by denitrifier, Soil Biol. Biochem., 37, 1535–1545, https://doi.org/10.1016/j.soilbio.2005.01.009, 2005.

Wang, R. Z., Lonergan, Z. R., Wilbert, S. A., Eiler, J. M., and Newman, D. K.: Widespread detoxifying NO reductases impart a distinct isotopic fingerprint on N ₂ O under anoxia, Microbiology, https://doi.org/10.1101/2023.10.13.562248, 2023.

Yang, H., Gandhi, H., Ostrom, N. E., and Hegg, E. L.: Isotopic Fractionation by a Fungal P450 Nitric Oxide Reductase during the Production of N2O, Environ. Sci. Technol., 48, 10707–10715, https://doi.org/10.1021/es501912d, 2014.

General comments

This paper reports potential rates of N₂O production by several important pathways in the oxygen deficient zone in the Eastern Tropical North Pacific. The studied oceanic region is one of the major N₂O sources to the atmosphere, and therefore this work is crucial in understanding the origin of excess N₂O and predicting the future emission of this global-warming and ozone-depleting gas under changing oceanic environment.

The strong point of this paper is that the N₂O production rate and its dependence on dissolved oxygen concentration are determined for each of the possible N₂O formation processes using ¹⁵N-labeled substrates and isotopocule measurements. The authors found the significant contribution from hybrid N₂O production during ammonia oxidation in the near-surface and deep N₂O concentration maxima. They also found that the hybrid N₂O formation is enhanced in low-oxygen water and that N₂O can be produced by denitrification from nitrate even with oxygen concentrations higher than those considered to inhibit denitrification. I believe these findings will help us develop a clear picture of N₂O cycling in and around the ODZ.

Thank you for this positive and thorough assessment of our work.

However, I think this work has a couple of drawbacks. First, the authors use conventional second-order kinetics to analyze N₂O production processes in order to calculate the rate of each pathway from the N₂O isotopocule ratios obtained by the ¹⁵N-incubation experiments. Considering that the amount of tracers added is sufficiently higher than those in initial seawater, I don’t think it is always appropriate to assume that the rate is proportional to the concentration of substrates. I would like to see some justification or evidence on this assumption.

If we understand the reviewer’s comment correctly, the concern is that the rates of N₂O production could have plateaued and not continued to increase (or decrease) with increasing (decreasing) substrate concentrations. In other words, the question is, what happens when you scale the N₂O production rate to the substrate concentration instead of assuming that the rate has hit its maximum value?

There are two important things to clarify:

• The model solves for the 2^nd-order rate constant that best fits the data, given a certain concentration of substrate. These 2^nd order rate constants are not necessarily applicable to ambient substrate concentrations; thus, we report the rates, not the rate constants.
• The substrate concentrations in eqns. (13) and (14) are the total concentration of substrate including the tracer and carrier additions, not just the ambient concentrations of each substrate. Because these substrate concentrations do not vary much in the incubations, eqns. (13) and (14) effectively amount to the same thing as assuming N₂O production has plateaued and hit a maximum rate. Nonetheless there are some cases where the substrate concentrations change over the course of an incubation, and we assess below how this would influence our results.

The experiment with the highest rates of ammonia oxidation was at station PS3, feature “interface2” (63m, Table S1). Here, the rates of ammonia oxidation were 4.68 nM/day (Table S2). In the ¹⁵N-NH₄⁺ experiment, the starting ammonium concentration was 0.52 µM, and the starting nitrite concentration was 1.61 µM. This includes the ¹⁵N-NH₄⁺ tracer addition and ¹⁴N-NO₂^- carrier addition. Then, the modeled hybrid N₂O production rate declines by 1% over the course of the experiment: 

[(0.52-0.00468)(1.61+0.00468)]/[(0.52)(1.61)]∙100=99%

Likewise, the modeled N₂O production from solely ammonium declines by 2%:

[(0.52-0.00468)(0.52-0.00468)]/0.52² ∙100=98%

And the modeled rate of N₂O production from NO₂^- increases by 0.5%:

[(1.61+0.00468)(1.61+0.00468)]/1.61² ∙100=99%

Even in the experiment with the highest nitrite oxidation rate, from the secondary nitrite maximum (182 m) at station PS3, the modeled rate of N₂O production from NO₂^- only declines by 12% over the course of the experiment, and the modeled rate of N₂O production from NO₃^- only increases by 4% over the course of the experiment.

What if we compare the ¹⁵N-labeled ammonium treatment to the ¹⁵N-labeled nitrite treatment at the same experimental depth, since the tracer additions were unequal (5.00 µM ¹⁵N-NO₂^- vs. 0.501 µM ¹⁵N-NH₄⁺)? The ⁴⁵N₂O and ⁴⁶N₂O production rates in the ¹⁵N-labeled nitrite treatment were far higher than those in the ¹⁵N-labeled ammonium treatment, even when normalized by atom fraction. This is visualized below. In fact, the rates of production of ⁴⁵N₂O and ⁴⁶N₂O in the ¹⁵N-labeled ammonium treatments were so small, comparatively, that they are visually indistinguishable from zero when plotted on the same scale as the rates of production of ⁴⁵N₂O and ⁴⁶N₂O in the ¹⁵N-labeled nitrite treatments.

Production of ⁴⁵N₂O, divided by atom fraction, in the ¹⁵N-NO₂^- treatment vs. ¹⁵N-NH₄⁺ treatment at the same experimental depths. Red diamonds indicate p⁴⁵N₂O^a/¹⁵Fand black diamonds indicate p⁴⁵N₂O^b/¹⁵F. b) Production of ⁴⁶N₂O, divided by atom fraction squared, in the ¹⁵N-NO₂^- treatment vs. ¹⁵N-NH₄⁺ treatment at the same experimental depths. In both plots, the dashed line is the 1:1 line.

Since the tracer concentration was much higher in the ¹⁵N-labeled nitrite treatment (5.00 µM) than in the ¹⁵N-labeled ammonium treatment (0.501 µM), this imbalance of ⁴⁵N₂O production supports the idea that there is some dependence of N₂O production rate on substrate concentration. The 2^nd order kinetics in our model allow us to capture that dependence.

Second, the contribution of suspended particulate matter to N₂O formation is not adequately taken into account in the interpretation of the results. Although the authors discuss the algal N₂O production as an alternate source of N₂O, it seems that they do not pay more attention to other particulate matter. Why don’t they consider potential N₂O production/consumption at anoxic microsites inside the particles? Although I don’t know any reports on experimental evidence of such N₂O production, at least one paper suggested that active microbial CH₄ oxidation occurs within the oxic/anoxic boundary of sinking particles (Sasakawa, M.et al., 2008. JGR: Oceans, 113(C3). https://doi.org/10.1029/2007jc004217).

We agree with the reviewer that particle-associated denitrification is a potential alternative N₂O source, especially at the highly productive coastal station. We have added particle associated N₂O production and consumption to the discussion of potential alternative sources of N₂O.

In summary, I recommend the publication of this paper after addressing the issue above and specific points below.

Specific comments

L64–66. Do the authors also mean NO does not undergo exchange with outside NO? In addition, are all the references listed here appropriate to cite? I cannot find the “evidence of nitrate reduction to N2O without exchange with an extracellular nitrite pool” in Monreal et al. (2022) and Toyoda et al. (2023).

Yes, the process that we refer to here is N₂O production from externally sourced nitrate without exchange of intermediates outside the cell, including NO. This is implicated in both of the cited papers as a major source of N₂O in the eastern tropical North Pacific and Bay of Bengal, respectively (Monreal et al., 2022; Toyoda et al., 2023). We have clarified this in the text.

L108–110. Is the STOX sensor identical with “Optode” in Table S1? It is confusing because “chemiluminescent optode” appears later in section 2.3.

Apologies for the confusion here. The measurements from STOX sensor mounted on the rosette are different from the optode measurements reported in Table S1. We have removed the mention of the STOX sensor since we do not report any of its measurements.

L131–133. I appreciate the authors’ effort to avoid oxygen contamination, but isn’t there any possibility that this procedure might reduce the oxygen concentration to the level lower than in situ seawater?

This is indeed a concern, which is why only anoxic depths (where the ambient dissolved oxygen was below detection) were purged with He gas. Depths with low but non-zero ambient oxygen were not purged. The creation of a He headspace should also result in a small reduction in the dissolved oxygen in the sample after equilibration. In this case, however, the He headspace was so small (2 mL) that it did not outweigh or even compensate for the oxygen contamination introduced during sampling. This is shown in Figure S1.

L161–162. How were the fiber optic cables pulled out of the bottle without air contamination?

We apologize for the confusion. The FireSting fiber optic cables never enter the bottles, themselves. Instead, the fiber optic cables measure the signal from the oxygen sensor spot placed inside the bottles through the glass wall of the bottle. This has been clarified in the text.

L165. Could the fiber optic cables, not the sensors, be really calibrated?

The fiber optic cables were indeed calibrated with a two-point calibration, using an oxygen sensor spot mounted inside a bottle containing 30 g/L sodium sulfite solution (0% saturation) and a sensor spot mounted inside a bottle containing air-equilibrated seawater (100% saturation). The same two calibration bottles were used for all four of the fiber optic cables, effectively correcting them to the same scale. Differences in detection limit between sensor spots were accounted for by first performing this two-point calibration procedure to correct for differences between fiber optic cables, then measuring the minimum oxygen concentration measured by each sensor spot in helium-purged seawater (purged at 100 mL/min for 90 minutes, equal to 56 volume exchanges). We have added this explanatory text.

L177. Which does this optode mean, STOX or chemiluminescent? (see above)

Again, apologies for the confusion. We refer here to the chemiluminescent optode measurements and have removed any mention of the STOX sensor from the text.

L233–238. Because the sample for N₂O measurements were poisoned with HgCl (L151), remaining sample could damage the denitrifying bacteria. How did the authors get around this problem?

Samples are diluted in the bacterial media, so that the effective concentration of HgCl₂ that the bacteria experience is lower than typical for poisoning. In addition, the denitrifier method uses a high concentration of bacteria and no adverse effects from addition of HgCl₂ have been observed.

In test runs, we found no statistically significant difference in the δ(¹⁵N) of standards (USGS32, USGS34, and USGS35) prepared with and without HgCl₂. This was true of standards prepared with 20 nmol NO₃^- and 10 nmol NO₃^-.

L269. Why were not individual uncertainties for δ(¹⁵N-NO₂^-) measurements estimated? Was there no need to apply the procedure for δ(¹⁵N-NO₂^-) because of larger peak area obtained?

Our method of estimating individual uncertainties was developed to deal with low NH₃ oxidation rates, which generated low peak areas in δ(¹⁵N-NO₃^-) samples. Since the rates of NO₃^– reduction were generally much higher than the rates of NH₃ oxidation (Table S2), a parallel method was not needed to estimate individual uncertainties in samples measured with the azide method, i.e. δ(¹⁵N-NO₂^-) measurements. This has been clarified in the text.

 L317. In the work by Frey et al. (2023), time course of N₂O production was analyzed with Michaelis-Menten kinetics and Km values of 0.017–0.018 mM were obtained for oxycline at stations PS2 and PS3. In the present study, NH4+ was added at 0.5 mM, two orders of magnitude higher than the Km values. This means the rate of N2O production should reach to the maximum value, irrespective of substrate concentration.

See response above regarding the representation of N₂O production kinetics in our model.

L336, eq (16). Following the convention used for eq (14), 1/2 of the right-hand side of this equation should correspond to the ammonia consumption rate.

Eq. (14) contains the factor ½ because that converts the rate of ammonia consumption in nM-N/day to N₂O production in nM-N₂O/day. We have clarified this in the text.

L566–568. Describe more details about the “different conditions”. It seems the location and cruise are identical between the two studies. Were date or time different? What were the differences in other hydrographic/chemical parameters?

It is important to note that where our samples overlapped with this previous work, we observed similar results (>90% hybrid production). The depths where we observed a smaller proportion of hybrid production had not been sampled in previous work; it is possible that we sampled different microbial communities there, acclimated to different levels of ammonium, nitrite, and dissolved oxygen. This has been clarified in the text.

L590. On the basis of which data can this claim be made? Fig. S9 shows a clear deviation from the relationship expected for N2O production from a single substrate pool, but it does not present how the relation would be if NH4+ and NO2- were used in the ratio 1:1.

That’s true. We don’t actually present evidence of the 1:1 ratio of NH₄⁺ to NO₂^-; instead, hybrid N₂O production is operationally defined in our model as a 1:1 combination of N derived from NH₄⁺ and NO₂^-, which is generally consistent with previous work (Stieglmeier et al., 2014). Any combination of N derived from NO₂^- with a second N derived from NO₂^- would be included in the N₂O production from NO₂^- pool; likewise, any combination of N derived from NH₄⁺ with a second N derived from NH₄⁺ would be included in the N₂O production from solely NH₄⁺ pool. The question, then, is what reaction would be specific enough to have one N derived from each substrate, but not specific enough to govern ¹⁵N placement in the resulting N₂O? One such reaction could be the combination of NH₄⁺ and NO₂^- to form an intermediate such as hyponitrite (HONNOH or ^–ONNO^– in its deprotonated form), which reacts to form N₂O via breakage of one of the N–O bonds, resulting in N₂O that contains a 1:1 ratio of NH₄⁺: NO₂^–. With a precursor such as hyponitrite, equal formation of ⁴⁵N₂O^a and ⁴⁵N₂O^b could be achieved with non-selective N–O bond breakage. We have revised the discussion accordingly.

L614–616. I cannot understand whether the authors consider the N-O bond breakage occur randomly or at specific site regardless of ¹⁵N distribution in the intermediate containing two N-O bonds. I see that the former case corresponds to f = 1/2, and δ¹⁵N^sp will become equal to ε (i.e.,

¹⁴N-O bond at one side of the intermediate molecule is more likely to be broken than

¹⁵N-O bond at the other side). In the latter case, however, what happens if the bond cleavage resulting in N^b of N₂O does not proceed due to the slower rate for ¹⁵N than ¹⁴N? We cannot rule out the possibility that the intermediate go back to substrate in such a case, but it accompanies N-N bond breakage, which should require more energy than N-O bond breakage. Rather, it appears that all intermediates are eventually converted to N₂O. Then we don’t need to consider ε for the N^b-O bond breakage.

Here we assume the former case: that either N-O bond could break, not at a specific site.

L623–625 and 674–677. I agree that denitrification is not likely to proceed in the aerobic water column, but how about the microsites within suspended particles which might provide anaerobic condition?

Good point — it is also possible that particle-associated denitrification is a potential driver of the δ(¹⁵N^sp) minimum observed in Popp et al. (2002) (L623-625). We have added this to the text. We have also added particle-associated denitrification as a potential contributor to our observed N₂O production from denitrification at higher-than-expected dissolved oxygen levels (L674-677).

L632 (caption), It would be helpful if x-axis includes the full range of f (0 to 1).

We modified the x-axis to include the full range of f.

L721–728. It seems that the authors assumes the first case I pointed out above. I cannot follow why the resulting site preference becomes variable.

Thank you for making this point. We rephrased the conclusions to focus on the fact that we see more or less equal production of ⁴⁵N₂O^a and ⁴⁵N₂O^b in most of our experiments, which would imply that hybrid δ(¹⁵N^sp) does not vary.

L768 (eq A10) and L769. “slope2” and “intercept2” do not appear in eq (A10). Is this equation correct?

Thank you for catching this error. Eqn. (A10) was indeed written incorrectly. We corrected eqn. (A10) to include slope₂ and intercept₂ (now called m₂ and b₂).

Table S3. If I understand correctly, f is applicable only to hybrid N2O production. Why values (including 0) are listed even when hybrid production rate is zero?

Thank you for catching this error. We have removed the f values in Table S3 (now table S4) and Fig. S12 (now Fig. S10) for experiments where the hybrid production rate is zero. There are some very small but significant rates that were hidden due to how the numbers were rounded. The rates in Table S4 have been converted to pM/day to fix this issue.

Technical corrections

L24. O in N2O should not be subscript.

Corrected.

L38. The error for the value “0.85” should be “0.03”?

Corrected.

L43. The “m” in “mmol/kg” must be mu.

Corrected.

L202, eq (3). It seems unnatural to write down ¹⁸R_VSMOW numerically, but not for¹⁷R_VSMOW.

Corrected.

L266. Use a single character for parameters such as rate and slope.

Corrected.

L293. nitrifier-denitrification using extracelluar NO2-.

Corrected.

L302 (eq. 8). Subscripts “i” and “k” in the summation terms should be “n”.

Corrected.

L486 (Caption of Fig. 5). …total N2O production at stations PS1 (a), …

Corrected.

L506 (Caption of Fig. 6). I cannot see “values of a and b in white boxes”, but a legend (without box) showing the fitting function in each panel.

Apologies, the Copernicus system seems to remove any transparent objects (including these white boxes) from figures if they are saved as vector files. Changing the figure format to .png or .jpeg seems to fix this issue.

L529. “0.12 nM N2O/day” seems to correspond to “0.11” in Table S3.

0.12 was the correct number. We have changed the units of Table S3 (now Table S4) to pM N₂O/day to make the numbers easier to read.

L614. Add equation number to the first equation, or continue the eq (24) from the first line by deleting “d(¹⁵N^sp)” in the left-hand side.

Corrected.

L754 (eq A2) and L755. It is confusing to use same character “m” and “b” in eq (A2) and the general equation for linear function.

Changed terms "m" and "b" to "A_measured" and "A_blank".

L757 and elsewhere. Parameters in equations A3–A7 and A10 should be written with a single character (and subscripts).

Corrected.

L972. Fix the author lists of Prokopiou et al. (2017).

This reference has been removed.

Title page of supplement says the file contains 14 figures, but I can see only 12.

Corrected.

 Figure S1. Add “a” or “b” to each panel.

Corrected.

Figure S2. It would be helpful if the region of ambient nitrate between 20 and 50 mM is enlarged because the delta values look significantly higher than natural values.

We added a panel (b) with values between 20 and 50 µM. They are indeed elevated.

Figure S4. Fix the explanation of panels a–d so that the figures and caption are consistent.

Corrected.

Figure S7, caption. Fix the typo “bluen”.

Corrected.

Figure S12, caption. Panel (b) is plotted against sigma theta, not nitrite.

Corrected.

This study presents very interesting findings on N2O hybrid production in marine environment. The complex approach applying 15N tracing methods in 3 different treatments with simultaneous measurements of d15N alfa and beta is very innovative and applied here for the first time in a real case study. Authors present the improved method of calculations of d15N alfa and beta in traced experiments, which has been integrated into the isotopomer-calculation software. These points are making this study important in further development of N2O-isotope based research, since the presented approach may broaden our interpretation potential of N2O isotopolocule studies.

However, the manuscript needs minor revision. Due to complexity of the experimental approach and results description, some aspects are difficult to follow by the reader and some information is missing. I suggest some technical corrections for this (below).

But more importantly, I disagree with the conclusion that hybrid N2O formation results in incorporation of N atoms from 2 substrates into different positions of N2O molecule (alfa and beta) - because this is not supported by your data. Most of your samples indicate the opposite - that N is located in both position independently of the substrate - which you describe very nicely in section 4.2. Below, in the specific comments, I also explain my points in more detail.

Very true. We have rephrased this part of the discussion, as well as the conclusions and abstract, to center around the fact that we do see equal formation of ⁴⁵N₂O^α and ⁴⁵N₂O^β in most of our experiments, which would indicate that hybrid site preference does not vary after all.

Specific comments:

L80: Actual definition of delta values is

(Rsample/Rstandard–1)

factor 1000 is just due to expression in permil notation, should be omitted in the definition

We removed the factor of 1000 from the definition.

L 151: 2%-92% - that wide range? is this correct?

This is correct. We added 1 µM ¹⁵N-NO₃^- to all of our experimental depths, regardless of the ambient NO₃^- concentration, resulting in a wide range of atom fractions due to the wide range of ambient NO₃^- concentrations. At depths where ambient NO₃^- is high, however, and thus the atom fraction is low, the rate of N₂O production from NO₃^- is high enough that we still get a detectable signal in ⁴⁵N₂O and ⁴⁶N₂O (see Figures S7 and S8).

L194: It should be described more precisely how much was added, depending on the concentration and enrichment level? I understand this was just a dilution procedure for mineral nitrogen isotope measurements? Or also for N2O measurements? It is a bit misleading because this chapter title is N2O isotopocule measurements... so I am not sure if my understanding is correct. Or you have diluted mineral nitrogen forms in your experiment to dilute your produced N2O in the headspace? Why not to dilute the N2O sample with any technical N2O gas to get respective dilutions?

The first paragraph of Section 2.4 describes the sample preparation procedure, immediately prior to running liquid samples for nitrous oxide isotopocules. Since we run liquid samples on the purge-and-trap system (see below), we need to protect the purge-and-trap system from highly ¹⁵N-enriched NH₄⁺, NO₂^-, and NO₃^- dissolved in the sample. To accomplish this, 100 µL of ¹⁴NH₄Cl, Na¹⁴NO₂, or K¹⁴NO₃ carrier was added to each sample a final concentration of 54 µM, 262 µM, or 27 µM, respectively, to bring ¹⁵N tracer levels below 5000 ‰. We have clarified the above in the text.

section 2.4: Actually you do not say how you finally collect your gaseous N2O samples - which volume, which containers, which procedure? Were the N2O samples colleted once only from each bottle or regularly in some time intervals?

We apologize for any confusion here. The purge-and-trap system completely extracts the dissolved N₂O from the sample (incubation) bottle and is described in greater detail in McIlvin and Casciotti (2010). So, one bottle = one sample. Time series are constructed by sacrificing triplicate bottles over a time course, rather than resampling the incubation bottles over time.

We describe how liquid samples were collected for incubation in section 2.2, “sample collection.”

Equation 3: what value was assumed for D17O?

Δ(¹⁷O) was assumed to be 0. We have added this to the text.

Figure 2: should the yellow arrow between NH4 and NO2- go in both directions? since this represents formation of hybrid N2O with cellular NO2-, right?

We added an arrow representing hybrid N₂O with cellular NO₂^-. The vertical arrow was between NH₄⁺and NO₂^- was a bit confusing since it did not represent an N₂O production processes, only NH₄⁺oxidation to NO₂^-. We made the vertical arrows colorless to indicate that they are not N₂O production processes.

Equation 18: I think this definition, with some explanation why this is possible should appear in methods section 2.6 This does not fit in results section. Same with Eq. 19

We respectfully disagree. Section 2.6 describes the modeling framework, and the model does not use eqns. (18) and (19). Actually, the modeling framework is a much more nuanced way of estimating the rates of hybrid N₂O formation than simply using eqns. (18) and (19). Eqns. (18) and (19) are just a way of showing that hybrid N₂O production is indeed occurring in our experiments, which we do in section 3.3.

L 550: Why there is such large difference in NH3 oxidation with different studies? - it should be discussed - is this due to different analytical approaches?

There are several factors that may have contributed to Travis et al. (2023) measuring higher rates of ammonia oxidation than our study or that of Frey et al. (2023). The incubations in Travis et al. (2023) were performed at different depths than ours, so they likely captured different microbial communities, different light levels, different chemical conditions (nitrate, dissolved oxygen, etc.). This is further exaggerated by the fact that the oxycline was moving up and down during the course of our occupation of PS3, so even experiments performed at the same depth on different days would likely sample different biogeochemical conditions. Finally, the incubations performed in Travis et al. (2023) were fully aerobic, whereas ours were generally low-oxygen and gas-tight. For example, the dissolved oxygen in our incubation with the highest rates of ammonia oxidation was 2 µM (see tables S1 and S2).

 We also needed to make a correction: the highest rate of ammonia oxidation measured by Travis et al. (2023) was actually 90±2 nM/day, not 48.7 nM/day.

L 600: why, which process can be responsible for this? Very important observation! You could give more details to these points - which processes dominated there, what was the N2O flux (rather high or low) or how it is possible to interpret these data?

See comments in response to L 613.

L 613: But in the first and second paragraph in this section 4.2 you showed that the values originating from NO2 and NH4 are mixed and finally the formed N2O has randomly situated 15N atoms from NO2 and NH4

I see, below the Eq (24) you explain, in most cases it is equally distributed but in some it is not. But why? The reader is a bit lost here

In the second paragraph you described very precisely how the hybrid formation may function and why we get equal distribution, and this is very convincing. So, the few cases with f unequal 0.5 must be due to some other process, some different mechanism? I understand this is rather an exception than a rule for hybrid formation – but you define this as a rule in Eq.24 (and then repeat this as final conclusion).

This is very important to describe this correctly here because for NA studies we do not know f, hence your conclusions here will be crucial for d15N-SP interpretations in NA studies.

Thank you for these comments. We revised the discussion in section 4.2 to reflect the fact that the majority of our experiments have equal formation of ⁴⁵N₂O^α and ⁴⁵N₂O^β and f within error of 0.5. This is actually a very important finding for the interpretation of natural abundance N₂O isotopocules because it implies that hybrid N₂O would indeed have a constant δ(¹⁵N^sp), despite being derived from two different sources. We revised section 4.2, the conclusions, and the abstract to reflect the equal formation of ⁴⁵N₂O^a and ⁴⁵N₂O^b seen in most of our experiments and the implications of f being equal to 0.5.

L 630: ok, but maybe you can sum up what were the conditions for the samples with f unequal 0.5 in your studies

I believe that it is rather not the hybrid process that behaves sometimes like this and sometimes the other way but rather admixture of some other processes, or the issue with the usage of cellular and extracellular NO2-. What about possible fungal co-denitrification that may show different mechanism?

I think you have so much data that maybe some hypotheses can be made?

The experiments with unequal ⁴⁵N₂O^α and ⁴⁵N₂O^β formation spanned a range of oxygen concentrations, depths, and substrate concentrations, and no clear patterns emerged. We do note that significant relationships emerged between f and ambient [O₂] (R² = 0.84, p < 0.001; Fig. S12a) and potential density anomaly (R² = 0.72, p < 0.001; Fig. S12b), although both relationships exhibited a large amount of scatter. These oxygen and potential density gradients may be proxies for changing archaeal community compositions at different depths in the water column, which may exhibit different patterns of incorporation of NO₂^—-derived N and NH₄⁺-derived N into N^α and N^β. We now note this in the text.

Thanks for the suggestion that we may have sampled a different “hybrid” process at these depths, such as fungal co-denitrification (Shoun et al., 2012), which may proceed via a different pathway from archaeal hybrid N₂O production. We added this alternative to the text.

L 703: Have you observed any activity, any N2O production in HgCl2 poisoned treatments? Would be interesting to report what was the "background" N2O production, since in some studies it appears quite high.

Was this in the expected range of abiotic N2O production?

We agree with the reviewer that there is a concern about abiotic reactions between NO₂^- and HgCl₂. In our ¹⁵N-NO₂^- experiments, the t₀ samples did not have δ(¹⁵N-N₂O) or δ(¹⁸O-N₂O) outside of the natural abundance range, which would have indicated an abiotic reaction between the ¹⁵N-NO₂^- tracer and HgCl₂. In comparison, we do see some elevated δ(¹⁵N-NO_x) in these samples (Figure S2), indicating that the sulfamic acid treatment may have converted some ¹⁵N-NO₂^- to ¹⁵N-NO₃^-, and/or that there was ¹⁵N-NO₃^- contamination in our ¹⁵N-NO₂^- tracer. We still believe that it is important to measure t₀’s in case an abiotic reaction should shift the baseline and it is necessary to account for this shift.

L 724: But this conclusion is not supported by the previous sentence. From the mechanism you describe it is expected that the alfa and beta positions are independent of the substrate origin.

I do not agree with this conclusion since MOST of your samples do not support this, only in few cases you observed differences in alfa and beta position, so rather the opposite conclusion should be given here, with an indication that there are also some exceptions, with not fully understood mechanism (in my opinion resulting from admixture of processes which has not been taken into consideration - e.g. fungal co-denitrification - which you admit in section 4.6, that fungal N2O can be an important source and it is not included in your model). You have actually concluded this at the end of your section 4.2 properly. You cannot simplify this into different direction in the conclusions because people will mostly read only conclusions, and this is very important point impacting the interpretations of natural abundance N2O isotopocule studies very much.

We revised the conclusions to reflect the fact that we see equal formation of ⁴⁵N₂O^α and ⁴⁵N₂O^β in most of our experiments, and thus that hybrid N₂O is not likely to have a variable δ(¹⁵N^sp). This is an equally strong conclusion because it implies that it may be possible to define a δ(¹⁵N^sp)endmember for hybrid N₂O formation.

733: These observations can be also due to fungal activity since fungal species usually tolerate higher oxygen levels than bacteria.

Thank you for pointing this out. We added fungal denitrification as a potential explanation for some of the N₂O production from denitrification at higher oxygen levels than expected, both in the conclusions and in section 4.3, “Rates of N₂O production via denitrification”.

References

Shoun, H., Fushinobu, S., Jiang, L., Kim, S.-W., and Wakagi, T.: Fungal denitrification and nitric oxide reductase cytochrome P450nor, Philos. Trans. Biol. Sci., 367, 1186–1194, 2012.

The authors present an impressively thorough analysis of N₂O isotope systematics from a field study in the oxygen deficient zone of the eastern tropical north Pacific (ETNP) – a region well studied for its redox active nitrogen cycle. Through a suite of ¹⁵N labeling experiments and the leveraging of those results, the paper lays out a complex yet compelling argument for the ecological distribution of various pathways of N₂O production. Taking the isotopic scrutiny to the next level, the paper presents a powerful and novel analytical model that leverages both the relative formation of singly labeled (⁴⁵N₂O) and doubly labeled (⁴⁶N₂O) as well as the site-specific labeling of the inner (alpha) and outer (beta) N atoms across all experiments (e.g., ¹⁵N labeled NH₄⁺, NO₂^- or NO₃^-) to solve for relative contribution of N₂O formation pathways. To my knowledge, such a sophisticated analysis has not been braved – and the authors should be commended for it.

The authors also use their results to evaluate the O₂ sensitivities of each of the formation pathways under these field incubation conditions, tying the results to both in situ O₂ and incubation levels of O₂ (which sometimes differed from in situ). These results show that adopted thresholds for N₂O production by denitrification (for example) may not be as hard and fast as previously thought. The data provide quantitative relationships from which models can be built for estimating wider patterns in N₂O production.

Especially unique and thought-provoking was the model analysis interrogating the possible impact on natural abundance site-preference compositions in N₂O as arising from hybrid formation – especially the proposed involvement of a symmetric intermediate. I very much enjoyed Section 4.2 which carefully walks the reader through the logic of the analysis and argues for the hybrid pathway involving formation of a symmetric intermediate (such as hyponitrite). Equation 24 demonstrates how, with a symmetric intermediate (and a 50/50 contribution of NH₄⁺ and NO₂^- precursors) – the actual composition of the precursors does not impact site preference. However, if this 50/50 proportion varies (as they observe in some incubations) – then this assumption falls apart – and could in fact explain or demonstrate that the site preference values for hybrid N₂O formation may vary under differing ambient conditions. While exceptionally nuanced, I found the arguments laid out in this section to offer real strides forward in our collective understanding.

I also found particularly useful the demonstration of how go about combining probabilistic analysis of N₂O formation (e.g., stochastic distribution of ¹⁵R between alpha and beta positions) with the ¹⁵N labeling exercise (where an excess of doubly labeled N₂O (15-15-16) may arise depending on formation pathways). Introduction of this ‘excess’ term allows for the application of site-specific composition to determine N atom sources under ¹⁵N labeling circumstances. To my knowledge, this approach has not been leveraged previously – and thus the manuscript contains a wealth of valuable methodological information – which I found laid out very clearly. Thus, the paper should also stand as a useful model for work beyond N₂O dynamics in ODZs – and could provide a model for application to a range of other systems.

Overall, because of the complex nature of the work - this paper is a beast to get through. That being said – it is excellently written and offers a wealth of value for really pulling apart the complexity of environmental N₂O formation. I provide some minor editorial comments below which hopefully help to highlight some areas that could be clarified. I recommend publication.

We are sincerely grateful for this positive and thorough evaluation of our work. Thank you for taking the time to work through the many aspects of this paper.

Specific Comments:

There is a lot of complex discussion of N₂O isotope systematics – which are notoriously challenging to understand. I can see that the authors are very careful to be clear in explaining most things and using careful wording for helping the readers follow the logic.

What were isotope effects used for NH₄⁺ oxidation, etc.? Table?  Would variation of these values (for example) impact the error estimates as mentioned in L350-352?

We added a supplementary table (now Table S3) of the isotope effects used in the model for NH₃ oxidation, NO₂^- oxidation, NO₃^- reduction, and N₂O reduction. Since we’re dealing with tracer-level ¹⁵N, though, natural abundance-level isotope effects are unlikely to affect the model results. No isotope effects were applied to N₂O formation. 

Table S3. Fractionation factors used the time-dependent numerical model.

While I recognize here a nomenclature used for isotope ratios (e.g., “δ(¹⁵N)”) has been adopted to be in line with some recent protocols, I find the use of the extra set of parentheses extremely distracting, unnecessary, and confusing. While I’m sure that the adoption of such conventions was intended to help clarify, the addition of more symbols into these terms does not help the reader and frankly muddies the message. I may very well be a minority here, but simply don’t see the logic in these new conventions (especially in the context of N₂O which is already complex enough). I see zero value in adopting the new nomenclature, and though probably futile, would suggest the authors stick to the nomenclature that has been in use for decades (e.g., δ¹⁵N).

The justification for writing δ values with parentheses, e.g., δ(¹⁵N), is that δ is the quantity symbol and “¹⁵N” is the label. I started using this notation in Kelly et al. (2023) in order to reflect the recommendations in the latest SI Brochure (https://www.bipm.org/en/publications/si-brochure/ ) and I continue its use here for consistency and semantic precision. I understand that this is a change from the conventions in the field and is likely to be unpopular, but I ask the reader to bear with me for now, and perhaps the notation will become less confusing at it is more widely adopted.

The paragraphs starting on Line 610, together with Equation 24 and Figure 8 worked to convince me that when the proportion of NO₂^- and NH₄⁺ to hybrid N₂O formation is equal (and the intermediate is a symmetric molecule), then the actual ¹⁵N content (or δ¹⁵N value) of those substrates does not play a role in the emergent site preference value. Why then on line 725 in the conclusion – do the authors state that these values do matter (even if 1:1 contribution)? Is it not true that the hypothetically variable site preference values from hybrid N₂O formation actually emerge from variations in the 50/50 (or 1:1) contribution – and that only in those cases will the values of the substrates play into the site preference of the product N₂O (as in Figure 8)? Please clarify.

Thank you for this comment. When the contributions of NO₂^- and NH₄⁺ to each N position are equal, hybrid site preference doesn’t depend on the isotopic composition of either substrate. You could hypothetically have N₂O containing a 1:1 ratio of NO₂^- and NH₄⁺, but with N^α always derived from NO₂^- (f = 1), and in this case site preference would depend strongly on the isotopic composition of each substrate. But in most of our experiments, N^α is equally derived from NO₂^- and NH₄⁺, which would imply that hybrid site preference does not vary. This means that it may even be possible to identify an isotopic endmember for hybrid N₂O production, which would be very useful to the natural abundance N₂O isotopocule community. We have revised the discussion and throughout the paper to reflect this majority case. This is an important clarification of the results, so we are grateful to you (and the other reviewers) for pointing this out.

Technical Corrections:

Methods: Perhaps I missed this somewhere. What volume of sample was collected for the N₂O analyses? 160ml serum bottles? Foil bags?

In 2.2, “Sample collection,” we state that “Incubation samples were filled directly from Niskin bottles into 160 mL glass serum bottles (Wheaton) using Tygon tubing. Incubation bottles were overflowed three times before being capped and sealed with no headspace using gray butyl rubber septa (National Scientific) and aluminum crimp seals.” In response to this comment and a similar comment from Reviewer 2, we added a clarification that time series were constructed by sacrificing triplicate bottles over a time course, rather than resampling the incubation bottles over time.

L24:  N₂O formatting

Corrected.

L25: ‘forward running model’ – unclear what this means… numerical model? Analytical model? Is there some terminology you could use here to help clarify?

Changed to “time-dependent numerical model”.

L 86: instead of ‘unlinked to’ (which seems a little awkward) maybe consider ‘independent from’

Corrected.

L134: Was the introduction of this background N₂O done as a gas or in dissolved form?

Gas form. Added to the text.

L148: … to provide enough total NO₂^- …

Corrected.

L229:  Here referring to the precision being lower, but the standard deviations being higher is a little confusing.  Perhaps refer to the precision being ‘poorer’?

Corrected.

L245:  …another explanation would be that the ¹⁵NO₂^- tracer actually may have contained some amount of ¹⁵NO₃^- to begin with.

Added to the text.

L253:  seawater water?

Corrected.

L256:  …precision for the denitrifier and azide methods is typically better…

Corrected.

L336: here the word ‘exchange’ is used to refer to movement of ¹⁵N from one pool to another occurring through biologically mediated processes. I would suggest using the word ‘transfer’ and not ‘exchange’ – as exchange is often used to refer to abiotic (or enzyme mediated) equilibration between two distinct pools.

Corrected here and throughout the text.

L348: extra comma

Corrected.

L395: With respect to the apparent negative nitrite oxidation rate – can any explanation here be invoked? Is this a real phenomenon or just some random analytical artifact that can’t be easily explained?

The “negative” nitrite oxidation rates at two depths are likely an artifact of the elevated t₀δ(¹⁵N) values in some of our ¹⁵N-NO₂^- treatments (discussed above). We have added this to the text.

L456: sediment-water interface?

This measurement was made at 898 m, which was very close to the bottom depth at station PS3. Clarified in the text.

L490: N₂O production pathways

Corrected.

L725:  depends on the ¹⁵N content of each substrate

The conclusions have been modified to reflect the fact that we actually see approximately equal placement of NO₂^--derived N and NH₄⁺-derived N in N^α and N^β, and thus that hybrid site preference may actually be constant.