Reviewer 1:

Scientific significance:

The research concept is very clearly and simply stated in the Introduction. Hypotheses such as that framed and tested by the authors are challenging to undertake due to the difficulty in obtaining enough samples from a broadly distributed set of sites, using the same methods, in a timely fashion. Often these types of studies accrue data from multiple projects that were never intended to be considered collectively and so they may suffer from bias introduced because disparate research teams never coordinated and may have used different methods.  This research is an exception to that norm. This team has accomplished their research by using the WHONDRS program to contemporaneously collect a set of samples, using a common plan, to address their question.  The WHONDRS program sets a standard for how to carefully collect samples and corresponding reference data from a team of broadly distributed, motivated, self-selected collaborators and then to follow it up with detailed, systematic sample analysis. 

Thank you for sharing this encouraging perspective.

Scientific Quality:

The use of the WHONDRS program’s new and extensive database is notable and an exciting consideration of the data collected by the broad community of scientists who are in support of WHONDRS.  I appreciate the use of standardized samples (Field vs. Incubation samples) as a way to account for possible heterogeneity in the samples and control for skewed results that might occur due to such heterogeneity. Does this approach mean that their conclusions are contingent on only acquiring samples from settings that are homogeneous?  In other words, do they only have confidence that their findings hold whenever samples are strictly homogeneous?  Is it possible that using this technique they’ve removed samples that are naturally heterogeneous and worth including despite this challenging feature?  This might be a consideration in the interpretation of the results.

We plan to include a short discussion of this interesting issue/question in the Discussion section of the paper. 

I also wonder about the variation in extraction efficiency that the authors note in the Methods (pg 4).  What is the basis for the assumption that extraction efficiency is not systematically linked to respiration rate?  Is it possible that some compounds that are not easily extracted might also not be easily respired, i.e., that the “extractiveness” of a sample corresponds to its biological accessibility (in terms of respiration)?  The authors state that this assumption seems to be acceptable because it is extremely unlikely that extraction would be linked to respiration, but is there some evidence for this assumption?  Might it be possible that compounds exist that are both especially challenging to extract and challenging for microbes to respire?  Perhaps more background to support their assumption would come from organic biogeochemistry studies that have considered the nature of recalcitrant compounds.

We believe the reviewer is right in that there is technically a non-zero probability that there is some systematic link between extraction efficiency and respiration. Our specific analyses are focused not so much on the details of the organic chemistry or on the concentrations of individual molecules (which would be influenced by extraction efficiency), but rather on how many unique molecules were observed in each sample. A key point is that we do not use any information on peak intensities, which others have often used as a proxy for relative concentrations of individual molecular ‘species’ (technically, unique peaks in the mass spectra). Extraction efficiency should, in theory, have less of an influence on whether or not a given peak/species is observed, relative to its influences over relative concentrations of different peaks/species. In addition, if there was a systematic relationship between extraction efficiency and respiration rate, and if that bias was strong enough to systematically alter the number of observed peaks/species, this should lead to a significant relationship between peak richness (i.e., number of unique peaks/species in a given sample) and respiration rate. That is, the bias would cause there to be a relationship even though none exists in reality. In our data we did not, however, observe a clear statistical relationship between peak richness and respiration rates. Instead, we found that respiration rate is primarily related to dissolved organic carbon concentration (in the form of a constraint space). Organic matter peak/species richness explained a small, but not trivial, amount of variation in respiration beyond what carbon concentration explained on its own. As the reviewer notes, we cannot be completely sure that the variation explained by organic matter peak/species richness was not due to a bias related to extraction efficiency. However, even if it is due to a bias, that bias would ultimately be tied to the organic matter chemistry. As such, with or without a bias, we feel that we can confidently infer that some aspects of organic matter chemistry do explain a small, but not trivial, amount of variation in respiration rate beyond what is explained by organic carbon concentration on its own. It is nonetheless fair to add a caveat to this inference in that we cannot be 100% sure the relationship with organic matter chemistry is via molecular richness or something deeper in the detailed chemistry. In turn, we will add some text to the Discussion highlighting this line of reasoning and associated caveat.  

This is a relatively high-level view of the processes associated with the respiration of organic carbon in the hyporheic zone.  Regarding a more detailed inspection of the data, a couple things come to mind and these might be helpful to point out in the discussion.  Recognizing that the study was focused on trends that might be evident at the continental scale and accordingly required a collection of samples from a geographically vast area, it seems that there are some sample types that were not considered in the broad sampling effort.  Presumably, this is because collaborators could not be recruited from these areas to collect samples.  This might mean that certain watershed types as defined by regional climatic conditions, vegetation type, edaphic quality, regolith, underlying geology, stream gradient, etc. would have been under-represented in the dataset.  From the map, examples of missing areas seem to be rivers located in the upper Plains, the Basin and Range Province, and on the Pacific coast.  I cannot say that any such missing watersheds or river systems are critical to their story, but omission of these regions in this study suggest that they should be included in future studies.

We agree that there are some missing spatial and environmental domains. We are currently running a crowdsourced sampling campaign to help fill these gaps. For the current manuscript we will add some caveats and point to future needs as suggested by the reviewer. 

A tangential question: Are there systems other than marine and river corridors (as referenced at the bottom of pg 6) for which OM diversity and microbial respiration may have been considered?  Could soils and marine sediments be added to their list and considered in this regard and if so, then could they also be interesting reference points, or distinctive contrasts for this study performed on samples from hyporheic zones?  Conceivably, because of the way the WHONDRS work is conducted, this paper may be something of a landmark in having studied such a broad sweep of sample locations and might be used for future comparisons of microbially dominated ecosystems.

At the moment we are unaware of other ecosystem types that have been sampled and analyzed using the same methods used for this study. Consistency in methodology will be important for quantitative comparisons. However, we very much appreciate the idea of using the current dataset as one that can be added to for quantitative comparisons across diverse ecosystems. The dataset is open access and well-structured to enable reuse (i.e., it is as FAIR as we could make it) and expansion. We will include this nice idea near the end of the manuscript as another future direction for the community. Thank you for suggesting it! 

The Introduction includes reference to the importance of studying respiration in hyporheic zones. I think the Discussion could be improved by returning to this point and considering the how the findings may impact critical processes occurring in these riverine settings (i.e., where development or survival of larval/juvenile stages of fish species or aquatic invertebrates is fostered, where contaminant degradation occurs, where cold water refugia become important as rivers warm).  The authors consider this at the end (lines 240-247) ; however, I think something more about the implications of the paper findings would be helpful to include here and would underline the importance of the work to readers.

Thank you for another nice suggestion. We will include additional material in the discussion focused more on the implications for a broad audience.

Presentation Quality:

This paper is a concise, straightforward, and articulate test of a well-stated hypothesis using a large and unique dataset.  As a high-level view of the observed relationships between microbial respiration and OM diversity or OC concentration the paper succeeds in presenting the information.  The figures are all appropriate for explaining their observations.

Thank you for the encouraging thoughts. 

I suggest that the authors acknowledge the time and care taken by numerous scientists who sampled rivers and then contributed the hundreds of samples that were subsequently analyzed by the WHONDRS program.  I’m certain that the original WHONDRS paper does so; however, it seems appropriate to have such a statement in all of the papers using data from this program.

Another great suggestion, thank you. We will add more acknowledgement as suggested.

Reviewer 2:

Does the paper address relevant scientific questions within the scope of BG?

Yes – it reaffirms the importance of organic carbon (OC) concentration as a major control on hyporheic zone respiration and offers suggestions for further relevant research and suggests a possible role of Organic matter molecular richness on hyporheic zone respiration.

Thank you for the encouraging response.

Does the paper present novel concepts, ideas, tools, or data?

Mostly – more effort could have been made to fully analyse the data available in order to achieve the stated goals.

Please see our responses below associated with related reviewer feedback.

Are substantial conclusions reached?

Yes

Thank you for the encouraging response.

Are the scientific methods and assumptions valid and clearly outlined?

Yes, but effort can be made to further validate assumptions/methods

Please see our responses below associated with related reviewer feedback.

Are the results sufficient to support the interpretations and conclusions?

Yes, the authors clearly state the importance of organic carbon (OC) concentration (as found by other authors) as a major control on hyporheic zone respiration, and that OM richness may have an influence on hyporheic zone respiration

Thank you for the encouraging response. 

Is the description of experiments and calculations sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results)?

Yes

Thank you for the encouraging response.

Do the authors give proper credit to related work and clearly indicate their own new/original contribution?

Yes, but more credit should be given to the contributors of the WHONDRS dataset as the work presented here is an analysis of data collected and analysed by a wide range of contributors

Thank you for this very important suggestion. We will add more acknowledgement as suggested.

Does the title clearly reflect the contents of the paper?

The title overstates the findings with respect to the authors findings in terms of molecular richness- The authors themselves state in the Abstract  “we found that organic carbon (OC) concentration imposes a primary constraint over hyporheic zone respiration, with additional potential influences of OM richness.” May I suggest to avoid any possible misunderstanding, that the title be adjusted to reflect the quoted text.  Also further discussed below with respect to the use of respiration maxima.

We propose the following title: “Maximum Respiration Rates in Hyporheic Zone Sediments are Primarily Constrained by Organic Carbon Concentration and Secondarily by Organic Matter Chemistry”

This is meant to address Reviewer 2’s comment here and their comment below related to focusing on the maximum respiration rates. It also addresses Reviewer 1’s comment about the potential bias introduced by variation in extraction efficiency, whereby the most robust inference is that there is something related to organic matter chemistry that has a secondary relationship with respiration rates (i.e., there is some chance it’s not molecular richness, but rather something deeper about the molecular properties of the organic molecules; see our responses to Reviewer 1). The revised title uses the more general language of ‘organic matter chemistry’ to allow for this possibility.

Does the abstract provide a concise and complete summary?

Yes

Thank you for the encouraging response.

Is the overall presentation well structured and clear?

Yes

Thank you for the encouraging response.

Is the language fluent and precise?

Yes

Thank you for the encouraging response.

Are mathematical formulae, symbols, abbreviations, and units correctly defined and used?

Yes

Thank you for the encouraging response.

Should any parts of the paper (text, formulae, figures, tables) be clarified, reduced, combined, or eliminated?

Yes, see detailed comments

Please see our responses below associated with related reviewer feedback.

Are the number and quality of references appropriate?

Yes

Thank you for the encouraging response.

Is the amount and quality of supplementary material appropriate?

Yes, but more effort could be made to potentially analyse the data in more detail, possibly leading to further significant scientific findings and conclusions.

Please see our responses below associated with related reviewer feedback.

Detailed comments

The authors aim to test and advance a proposed hypothesis from Lehmann et al. (2020) and seek to test this hypothesis of the presence of a negative relationship between respiration rates and OM molecular richness in the hyporheic zone on a continental scale using data collected from the WHONDRS consortium. The hyporheic zone is chosen due to its higher levels of hydrologic connectivity which may diminish influences of spatial isolation such as an OM stabilization mechanism. The authors research rejects the hypothesis of any direct relationship between respiration rate and OM richness, both using the full dataset of sample respiration rates and maximum respiration rates across the OM richness. The authors confirm previous findings that OC concentration could impose a primary constraint over maximum respiration rates, with OM richness acting as a potential additional (but less important) constraint. The authors use maximum respiration rates to show that the combined influences of OM richness and OC concentration are realized as a non-linear constraint space, with the vast majority of measured respiration rates falling well below the constraint boundary. They further suggest research into additional factors which act as controls over respiration, which drive respiration below its potential maximum. The significant relationship between OM richness / NPOC and respiration rate is only valid for the respiration maxima and not for all the data collected, this seriously limits this continental scale study to a very small dataset. I would be interested to know the model results for the entire dataset of Respiration rate vs OM richness / NPOC (similar to the other models done) shown in Figure 4. I believe the title again does not reflect this important detail of the study findings and could lead to misunderstandings. Maybe a title along the lines of “Maximum respiration rates in the subsurface of rivers is predominantly constrained by organic carbon concentration, modulated by molecular richness” may be more representative.

There are two primary points here, and each is addressed in turn.

For the model with the entire dataset of respiration rate vs. richness/NPOC, we will include those regression statistics in the supplemental material.

The title has been revised as discussed above.

L9-10: I would be cautious with the phrasing here to avoid a misinterpretation – What is the definition of the hyporheic zone referred to ? To my knowledge most definitions, including those of authors cited in the current manuscript (eg. Krause et al. 2011) define the hyporheic zone as a zone of mixing of shallow groundwater and surface water. Not all sections of the river bed subsurface exhibit surface and groundwater mixing.

We will more clearly define our meaning of ‘hyporheic zone’ as definitions vary across researchers.

L15-17 / 25-26:  Since the hyporheic zone is specifically mentioned, is the data used from WHONDRS exclusively from the hyporheic zone (HZ)?

We use the definition of the hyporheic zone as those sediments through which surface water enters and at some point returns to the surface water channel. Collections of sediments were restricted to shallow (~3-10cm depth) fine-grained sediments. As such, we make the assumption that surface water moves through those sediments and returns at some point to the water channel. In turn, we assume that all samples are reasonably conceptualized as hyporheic zone sediments. We will include a more detailed description of our definition, assumptions, and sampling methods in the revised manuscript.

L24-25: What are the potential “other variables” that the results indicate are secondary influences on Hyporheic zone respiration (other than OM concentration) ? Could the authors  hypothesise based on literature which exists on the topic? Maybe lability, presence/ density of double/triple bonds, ring structures ?

This is a very interesting and important direction to be heading. We feel there are a broad range of possible mechanisms and will very briefly point to a couple possibilities here in the Abstract. Specifically, we will point to microbial biomass and sediment physical properties such as grain size and surface area.

L31-33: I would stress here not only contaminant removal, but more relevant to the paper, increased CO2 evasion (respiration) and DOM alteration within the HZ. Several papers exist on the topic eg. Nature Comms. and Scientific Reports

We will edit the text here to more directly link the HZ to CO2 evasion.

L41-46: I would argue that the classification of the molecular diversity in terms of structural complexity (eg. presence and number of ring structures, C:H, C:O ratios, N containing molecular formulae potentially indicating proteins, etc) and not simply number of unique organic molecules (after all the authors present FTICR-MS data) is also important for this. I would be interested which effect the different fractions of DOM molecules have on respiration. Have the authors explored DOM diversity in the level? I think it would be very interesting to identify groups of molecules that lead to higher respiration rates versus other groups.

This is a very interesting direction, though going down this path opens up a huge variety of analyses (e.g., >10 mean properties, Rao’s functional diversity for each of >10 properties, and up to three dendrogram-based methods integrating across properties). Each of those 25-30 analyses will need to be modeled against respiration rates in terms of whole-dataset and maximum values in both univariate and multivariate regressions. In total that will lead to ~100 additional analyses, with associated figures and statistical models. This will greatly expand the number of required figures and length of the Results and Discussion, leading to a very different paper. One of the strengths, in our opinion, of the current paper is that it is very tightly focused with a clear message. Our preference is to point to the need/opportunity for these additional analyses in the Discussion of the manuscript. We would appreciate editor guidance on this. Which direction is pursued will have major impacts on the paper and level of effort required to revise the manuscript.

L50-56: I am not convinced that all the data used from WHONDRS is actually from the hyporheic zone, can you confirm that it is ?

Please also see our responses above. In short, sediments were collected from ~3-5cm depth, relative to the riverbed surface, and we assume that surface water enters and flows through these shallow sediments and at some point returns back to the surface channel. Per the definition we will include in the manuscript, we consider this to be hyporheic exchange such that we consider the sediments to be part of the hyporheic zone. We will include these assumptions and considerations in the Methods and Discussion sections.

L100 – 101: This seems counter intuitive to me. You inverted ratios that were less than 1 ? Please explain further

We will provide clarification of this approach in the associated section. For our purposes the important consideration is the proportional difference between the Field and the Incubation NPOC concentrations. The same proportional difference could lead to ratios below or above 1 depending on whether Field or Incubation NPOC is higher. For our analysis we simply needed to know the proportional difference, not whether Field NPOC was higher or lower than Incubation NPOC. In turn, we simply inverted the Field-to-Incubation NPOC ratio if it was below 1 so that all proportional differences were more quantitatively comparable.  

L 106-118: Is the use of a Michaelis-Menten function and the half saturation truly more justifiable than the use of a least squares approach with a pre-determined limit on the tolerated difference between the “replicate” Field and Incubation NPOC samples (maybe 20%) that would justify removal.  Please explain.

We will provide more details of our rationale in the manuscript, along the lines of the following. This is a data quality control challenge and there are a variety of ways in which one could approach quality control of the data. In all quality control approaches there is a tradeoff between increasing confidence in data and removing so much data that statistical analyses become impossible. Our approach was to increase data confidence up to an inflection point beyond which there appeared to be diminishing returns. Based on the functional form of the data, it appeared that a Michaelis-Menten function fit the data very well and has the nice feature of estimating the half saturation constant, which we considered to be a practically useful inflection point.    

L126-130: Would FTICR-MS not yield information on molecular formulae, C:H, C:N, C:O ratios and thus indicate apparent lability? This may give further useful information.

This is related to a comment above about adding additional evaluations of organic matter chemistry to the paper. Our preference is to keep the paper’s analyses as they are and not expand into a large suite of additional analyses. FTICR-MS data are incredibly rich in terms of offering nearly limitless ways of using the data to study organic matter chemistry. As noted above, we feel a strength of our paper is that we have avoided the temptation to include a huge variety of exploratory analyses, and instead have focused on specific analyses tied to specific hypotheses.

L161-165: Just for clarity, was the maximum respiration rate in each bin plotted against the corresponding 1/NPOC value for that respiration rate or against an average of the bin ?

We will provide this detail in the revised manuscript. In short, used the 1/NPOC that corresponded to the maximum respiration rate as the x-axis variable.

L177-179: A skewed distribution is a possible indicator of another key controlling factor that was not taken into account by the model / study, correct?

At this point in the paper we are describing the distribution of measured rates and are not developing statistical models to explain variation in the data. Biogeochemical hot spots are broadly acknowledged as being essential components of ecosystems and their presence will, by definition, lead to skewed distributions of biogeochemical rates. In turn, we interpret the observation of a skewed distribution as indicating that we sampled enough sites to capture biogeochemical hot spots across the contiguous U.S. We find this an encouraging outcome of the study. In addition, we were able to develop highly explanatory statistical models of the constraint space that includes both low rates and the hot spots. In turn, we feel that we have accounted for the necessary factors, given the goals of our study.

L185-188: While the hypothesis sounds reasonable, I am not completely convinced by the data presented in the current graph. There are only three (out of ten) points making up the negative slope on the right of the graph showing a decrease in respiration rate with OM richness above 4000 unique peaks. The point representing the highest OM richness corresponds to almost double the respiration rate of the point representing the bin before it. Maybe using the maxima from 15 or 20 bins would make the relationship clearer ?

In this section we argue that there is no evidence to support the hypothesis that higher DOM richness leads to lower respiration rates. We believe the reviewer is saying the same thing here (i.e., the data presented in Fig. 3 are not consistent with the hypothesis). In turn, we believe there are no modifications to be made to the associated text. If we misinterpreted the reviewer’s comments, we would be happy to reevaluate, and await editorial guidance along those lines.

L201 – 209 : Given the authors analysis of the results, is the title of the paper truly justified ? Is it possibly a bit of an overstatement of the role of OM richness ? Should the title reflect more the statements in L 211 – 212?

Please see above for discussion of our plans for revising the title.

Figure 1: It seems that the samples were biased toward rivers in lower altitudes and flatter terrain (possibly lower gradient rivers?), as well as away from the central section of the USA. Could this have excluded some important environments/factors that are important for a “continental scale” model ? Also, what does the map look like showing the spatial distribution of final data point locations that were analysed for the model ?

It is an important caveat for all observational studies that all outcomes can be made only with respect to the sampling locations that were used. As the reviewer notes, we our sampling did miss some parts of the contiguous U.S., and in particular the upper midwest region. We did, however, sample across a broad range of environmental conditions such as stream order (1st to 8th) and land cover compositions (e.g., forest cover ranging from 0-97 % and urban cover ranging from 0-28%). Given the breadth of sampled environments, we have confidence in our outcomes and inferences, but agree that it is appropriate to call out some caveats and limitations related to the distribution of sampling locations. Text summarizing these limitations will be added to the manuscript, likely in the Methods and in the Results and Discussion.

The reviewer also asked to see the spatial distribution of samples that defined the constraint space. We didn’t made that map previously, but agree it could be insightful. We plan to include such a map in the supplemental material of the revised manuscript.

Figure 4/5: The full dataset is shown here. Why wasn’t a model for the full dataset calculated and results shown as comparison as done previously in Fig. 3 ?

We will include statistics within a supplemental table to summarize models applied to the whole datasets across Figs 4 and 5. Given the non-linear nature of the relationship we will specifically fit and report on negative exponential models. This is the same functional form fit to the constraint boundary so should also provide a useful and direct quantitative comparison in terms of model fits (i.e. R2 values will be used to compare models).