General Comments

Lambert et al.’s “Role of nitrogen and iron biogeochemical cycles on the production and export of dissolved organic matter in agricultural headwater catchments” provides an interesting contribution around possible mechanisms influencing dissolved organic carbon release into streams from agricultural soils. The manuscript provides a useful framework to explain why passive (water flow / discharge) transport of DOC from soils to streams does not work fully as a way to predict DOC concentrations in steam. The study proposes and investigates the role of Fe and NO3 in soils as active regulators of DOC release throughout the hydrological cycle. The study provides a useful contribution to our understanding of environmental regulators of stream DOC concentrations. I also think the results of the study highlight very clearly the heterogeneity of the system, which raises important and novel questions about the cumulative effects of soil-water interactions, material transport, and stream conditions.

I enjoyed reading the manuscript and after reflection have a few clarify questions and comments that I hope if addressed would strengthen the manuscript.

1. With respect to the results and statistical analyses, I did not understand why data were averaged and then used in the PCA at level of each lysimeter. I think the cluster and PCA approach works well here to reduced noise and find patterns, but based on the temporal patterns and dynamic connections between variables, it would appear to me that using each event by lysimeter would better match the papers intent. it Should flow path and spatial network be accounted for? Perhaps proximity of lysimeters to each other is important to explaining pattern?
2. I agree with the use of PARAFAC components as the sole DOM identifier and not using indices or peaks. The peak shape for each component looks normal, but I was surprised that the PARAFAC model did not include a protein-like peak. This seems odd and I can’t think of a study using PARAFAC that lacks some version of a protein-like peak even if contained within a multi-peak component. I am not certain what to make of this. Usually humic-like peaks dominant soils but protein-like peaks are still present and on an absolute basis can contain more protein-like materials than what would be found in a stream. Perhaps, the original uncorrected EEMs, blanks, and corrections could be revisited and verify that the model correctly represents their features. Perhaps including a few corrected-observed, modeled, and residual eem plots would be useful to highlight that protein-like peaks were not present. This way the reader can be reassured the model fit well the data. Assuming the absence of a protein-like component, I think this result needs to be discussed and clarified. Protein-like and less complex DOM forms are expected to be present in agricultural catchments and their absence would be interesting to explore more deeply in the discussion.
3. I think the connection between DOM quality and Fe-NO3 interactions could be explored more fully. The PCA results are mostly explored from a Fe, DOC, and NO3 point of view, but DOM seems to also be divided along the two clusters. I think more could be done to amplify the DOM spilt along PC1 within the two clusters. The discussion pulls in DOM quality as a possible mechanism and I think some of these links could be brought out more in the results. Perhaps variable influence scores for the PCA showed Fe, DOC, and NO3 were most important and this is why DOM was only partly included as a clustering agent. If so, I think it would be useful to the reader to acknowledge these influences.
4. The discussion did an excellent job connecting patterns and telling a story around the variables important to the release of DOC from streams into soils. I struggled a little seeing some of these patterns in the figures and I wasn’t always certain how to interpret the pattern within PCA clusters. I think it would be helpful to included a little more detail in the results that explains specific patterns visualized in the figures. This then could be revisited in the more detail already present in the discussion. In addition, perhaps there might be an X vs Y type approach that could be used to amplify the patterns expressed in the timeseries plots.

Specific Comments

Introduction

The introduction framed the study and the key background ideas needed to understand the results really well.

Methods

Figure 1: It seems like it would be useful to have lysimeter locations also included on the map. One of my questions was if there are spatial-correlations hidden within the clusters or across clusters that might be able to be explained by looking at location as a variable. Adding the lysimeter locations to the map, would at least help the reader see if spatial patterns seem relevant or not. The paper said some lysimeters of opposite clusters were within 1 m of each other. So their might not be a spatial pattern, but it might be useful to note this or acknowledge more fully the location of each lysimeter.

PARAFAC modeling data: It is true that PARAFAC is not influenced by Fe but the optical conditions of EEMs are impacted by iron levels. It might be useful to provide the range of estimated carbon and Fe(II) levels in the diluted sample. This way they could be used to assure the reader that Fe levels are below interference levels. With respect to absorbance scans, would it be possible to clarify if scans were made on filtered or filtered and diluted soil-water samples? High levels of Fe and NO3 have been suggested to impact absorbance scans towards the UV end of the spectrum.

Line180-181: Data average then normalized by lysimeter. Does this mean temporal resolution was lost? Please explain the reasoning for this. This collapses the data to n = 17 with 9 variables in multivariate analysis, which could impact how clusters and relationships form. Why not run a PCA on all data and then determine the centroid and error for each lysimeter and build clusters from the full dataset? By averaging, the link between variables within the analysis seems to be broken on a case-by-case basis. It seems in order to say at each event that Fe, DOC, and NO3 changed together, then the PCA should be able to form components with these connections available on a per event basis rather than across sampling location average for the year.

Figure 4 suggests correlations used but these analyses and purpose are not reported in methods. Could this approach be added to the methods and correlation coefficients be added to results?

Results

When exploring the PCA clusters, I think it would be useful to include more emphasis on DOM patterns or provide the variable influence scores that demonstrate that DOM is not a key influence on the clusters

I am not certain I understand Figure 7 or logic in Figure 6. The discussion suggests that the pattern highlights the role of heterogeneity in soils. Perhaps this idea could be linked here so that it’s clear that the patterns don’t match. Adding discharge or rainfall data to this figure might also make it easier for the reader to see the argument in the discussion around passive and active transport.

Discussion:

L369-372: This is a really interesting idea that very local active processes within a watershed can keep the collective view in a stream relatively constant even though the soil-water patterns are more dynamic. I wonder if there might be a way to tease this idea further out into the results so that it is clear how this idea connects to the broad level patterns in Figure 7 and the very messy lysimeter by lysimeter patterns. I wonder further how much DOC releases from a set of cluster 1 vs cluster 2 like soils is needed to maintain the DOC stream pattern once discharge (passive transport) is accounted for? Would it be possible to roughly estimate load from each cluster as a mixing model with two end members?

Technical Corrections

Two periods line 105

(XXX) line 137 after probe name

Line 235: “followed” should be “follow”

Line 244 % with space before number

L316: biodissolution repeated twice

Figure 1 caption, please also explain what PK3 represents.

General comments

This manuscript presents results from a sampling campaign in the riparian zone shallow groundwater and draws conclusions on the redox conditions influencing DOM exports into the stream of a small agricultural catchment. This work builds on preceding works in the same catchment and extents the previous findings and hypotheses. The manuscript is a good match for HESS and should be of interest for scientists working on catchment water quality.

The manuscript is written in a concise way. Figures are mostly good and references up to date. My specific comments given below add up to a quite long list but are not substantial. My most critical point is the temporal averaging of lysimeter data that nees a better justification. However, from my point of view some work is needed to bring this manuscript into a final acceptable form.

Specific comments

Abstract

The abstract uses DOC while title and manuscript text uses DOM. Homogenize that?

L7: Check this first sentence. Not clear what seasonal variations are meant here. Seasonal variations of environmental conditions that control DOC exports or controls of the seasonal variations of the DOC export itself?

L8: Why “nitrates” not “nitrate”?

L13: “visit” is maybe not the best choice here. I hope you also sampled them.

L14ff: Increase of DOC concentrations and release into the soil water seems to be the same thing. Our do you mean release into surface water?

L15: I have mixed feelings about “notably due to”. Is that your interpretation of the data or a proof? Maybe another choice can make that level of certainty in the underlying processes more clear.

Introduction

L36-39: Consider to state the hydroclimatic conditions under that these statements were made. Also consider the work of Winterdahl in this context (10.1002/2013gb004770).

L41-42: Is this statement underpinned with papers later in the text? Maybe it make sense to state that time-scale question after the next section? Here it seems quite strong without underpinning.

L43-56: Again some mentioning of climatic settings may help. I noted that often there is the assumption that DOC export works basically similar from boreal to Mediterranean/ subtropic conditions. Partly justified but (not exhaustively) stating where some of the findings have been made would be great. See also Laudon (10.1029/2012gl053033).

L67-72: I have problems following the line of argumentation in this sentence. Why is the impact of Fe reduction (on DOC export? Not stated here) limited but then rather favoring conditions are mentioned.

L75-77: This is true but it would be fair to cite some studies that make this attempt of bringing together soil and stream water (Knorr 2013, Dupas et al. 2015 – though P-centered, or even Seibert et al. 2009 and Ledesma et al. 2015).

L82-91: Maybe mention if and how stream water quality was monitored as well.

Material and methods

L102: Some more details on the riparian soils would be helpful to later on make more clear, why a depths of 15 cm was chosen.

L107: Figure reference is misleading here since there is no land use in this figure.

L122: Some details on the zero tension lysimeters would be helpful.

L128: Not sure if “degradation” is the right word.

L129: “consecutive dates of data” sounds a bit strange.

L132: Figure 1 reference does not fit here.

L137 and L146: Remove the XXX.

L157: On which bases was the decision made to dilute the sample.

L177ff: How were missing data handled?

L181ff: Does normalization not include an averaging to a mean of zero and a standard deviation of 1? Is the averaging justified? Temporal variability may be higher than spatial variability and maybe each single sample would be better in the PCA? This needs more justification somewhere in the paper.

Results

L189ff: In a stricter sense this opener connects your catchment to a general behavior in the region that would actually better fit the discussion and, moreover, needs referencing. I suggest to just state three phases. You may mention the general regional behavior in that type of catchment in the introduction or method section also. Moreover, the sharp mid-months boundaries of the phases need a criterion to be mentioned here. Is this visually decided?

Fig. 2: Consider to indicate the phases in the three time series.

L197: You mention frequent intense rainfall with 2 mm/d but wrote about moderate precip of 5 mm/d above. Are the numbers correct?

L201ff: Why not giving precipitation values for this third period?

L209ff: The chapter on fluorescence properties does not fit here in my opinion. It feels more naturally for me to first state water quality in terms of concentrations and then jump to the DOM properties in detail. But you may have a good reasoning at hand.

Fig. 3: Be precise in the captions. What temperature is this? Mention that C-F are soil water concentrations? Homogenize captions and axis – e.g. nitrates – N-NO3? The latter also applies to the other figures.

Fig. 4: Does it make sense to give nitrate in B on a log axis?

L231ff: I suggest to also give a Pearson’s correlation coefficient to underline the claim of a linear relationship. However, nitrate-iron relationship seems to be far away from linear and rank correlation would be maybe a better fit.

L234: “Some” lysimeter not following that cannot be seen in the data. You probably did the correlation across all data. Maybe state lysimeter-individual correlation coefficient (mean, range) to point out that some do not follow the general behavior?

L243: This title should make clear that this is about soil water quality.

L244f: Would be good to write clearly what variance is explained. “discriminating lysimeters depending on the degree of Fe(II) biodissolution” does not sound well written and seems to be more a discussion than a result. I would leave that statement more to the lines 251-253 at the end of the section.

L254ff: In this chapter it may be elaborated on if averaging the lysimeters in time was justified? Does one lysimeter seem to switch from high-nitrate:low-Fe to low-nitrate:high-Fe over time? This is hard to be seen in Fig. 6 as here all lysimeters of one sampling data and one cluster are averaged.

Fig. 6: Same as fig. 2 – indicating the different wetness phases would be very helpful here.

L258ff and 270ff: I don`t understand the idea of a flushing dynamic. Does that mean the solutes are flushed out and replaced by a different water with a different quality? Moreover, I find this to be more an interpretation of that data and thus more a discussion part.

Discussion

L290: appearance, abundance maybe better than apparition.

L303ff: What about a temperature effect here? Or can rainwater just not infiltrate deep enough?

L306f: I don’t understand the role of the hydraulic gradient here. Translating to soil water/ shallow groundwater travel time? What I read is more about depth to groundwater than hydraulic gradient.

L307: If DOC and SRP are similarly affected by oxygen and iron presence, why are they weighted differently in the two clusters?

L316: Double word biodissolution here.

L318f: I am not sure if I understand what is meant by “net depletion pattern”.

L327 (and 313): Spatial patterns are not shown but may be helpful and interesting? So maybe map the clusters back to the catchment figure? Could be in the SI or an additional panel in Fig. 5.

L337: Water circulation is maybe not the most precise word here. Does soil water really circulate?

L353-357: This is an interesting part but would, at this process-scale, better fit the end of chapter 4.1? But I am not fully sure either. I noted that this is an initial statement on something explained in more detail below (L373ff). Maybe make more clear that details are given in the following text?

L366-367: I know what you mean here but generally a “supply-limited pool” is associated with a dilution behavior not a flushing behavior. So, check your choice of words here and maybe better describe what concentration dynamics you see at this point in time in the stream.

Conclusion

L401: The word “but” indicates some contradiction which I do not clearly see here. Fe-reduction can only establish when nitrates are not present, right?

L409: Any reference for the “wetter winter” statement?

L404ff: This last section in the conclusions seem to make a new story on the relation of the small-scale redox soil processes to the long-term trends. While I appreciate this part I think this may be already part of the introduction and motivation.

This study reports on measurements of DOC quantity and quality, ferrous Fe, nitrate, P and pH in riparian soils for an agricultural headwater catchment in western France. Zero-tension lysimeters were installed in riparian soils at 15 cm depth. 17 lysimeters were sampled weekly to biweekly between November 2022 and June 2023. From data analysis, reductive Fe dissolution and the associated DOC mobilisation were found to be driven by the availability of nitrate. Redox driven mobilisation of DOC happed during the relatively cool and wet winter months.

General comments:

This is a well written paper that would benefit from the analysis of a few more aspects insufficiently addressed in the manuscript: (i) The molar ratio at which Fe and DOC were mobilised should be reported and could be compared to ratios reported in the literature. This would indicate if the processes interpreted from the data are reasonable. (ii) Also, there may be an indirect mobilisation of DOC due to a pH increase with Fe reduction which could be evaluated from the data presented. Note that an indirect mobilisation of DOC with a pH increase would probably increase OC to Fe ratios compared to a pure mobilisation due to the dissolution of iron minerals.

Specific comments:

L39-42: Isn’t this a contradiction? DOC export cannot be source and transport limited at the same time.

L54-56: This indicates a source limited DOC export, which contradicts the transport limitation stated above. I am confused. Perhaps, the conceptual model of source and transport limitation for DOC export needs to be explained better.

L79: hypothesis instead hypothese

L179: Is DOC part of the mineral composition of soil waters?

L284: I am not sure if Skerlep et al. is an appropriate reference here: Does this paper report really on Fe reduction during winter periods?

L316: delete biodissolution

L317: delete as long

L364-372: Here you again discuss your conceptual model of source versus transport limitations of DOC mobilisation. Perhaps a sketch of the conceptual model would help the reader to better understand this.

L384: Do you equate DOC production with redox driven mobilisation of DOC here?

L386: delete main

L409: Trends were not previously discussed but are mentioned now suddenly in the conclusion section.

Figure 4: Why is relation between Fe and DOC is closer than between Fe and P?