Water Table Dynamics Control Carbon Losses from the Destabilization of Soil Organic Matter in a Small, Lowland Agricultural Catchment

Round 1

Reviewer 1 Report

General comments:
The manuscript “Water table dynamics controls soil organic matter destabilisation pathways in a small, lowland agricultural catchment” represents an important look at how soil organic matter dynamics are influenced by seasonally high water tables which are present in significant areas of
global soils and are an under-researched area in the literature on soil organic matter stabilization.
Overall the manuscript is straight forward and generally understandable. Before acceptance
several areas should be clarified:
1) Some additional details are needed in the stastical modeling section. The work relies on fitting empirical equations such as the Q10 approach and variation with adding additional variables such as WFPS and DOC contents. The model fittings and coefficient correlations are used to make assumptions on the controlling factors of FCO2. With time series data, there is the issue of serial autocorrelation (points closer in time being more highly correlated) and the fact that the dataset is then broken down by slope position and other factors to explore some of these issues. This results in model fits going from 71 data points in equation 1 for the whole catchment to 19 points for slope position and transect and in equation 2 from 45 points down to in some cases 9 points. Especially in equation 2 where there are 3 model parameters to be fit in addition to the Q10 value, caution should be exercised in interpretations. An alternative approach, might be to keep the whole catchment together, but then use a mixed-modeling approach, where Q10 could vary by transect or position, in addition to a random error term which could address the serial autocorrelation. This would allow all data points to be used in one model, while
still exploring changes due to transection/slope position.
2) There are points in the manuscript that can use editing of the grammar/sentence structure.
I have made some specific edits/suggestions below:
Specific comments:
Title: Change “controls” to “control”
Abstract:
General comment: It is not clear from the abstract if the measured variables, especially water table depth were naturally occurring or the result of manipulation.
L16: and microbial processes
L17: change “dynamic” to “dynamics”
L19: change “transect” to “transects”
L20: change “sites” to “site”
L22: It is not clear what the “main parameters of this dynamic” refer to. Were these predictors in a multi-regression model or treatment variables. Some more description is needed.
L24: What was the investigated depth? What is the catchment scale? This hasn’t been mentioned before.
Keywords: CZO should be written out
Introduction:
General comments:
Macro- and microporosity is talked about numerous times, but is never defined. The first mention in the introduction should include a definition.
L37: “…stabilization and destabilization are in evolution.” Not clear what evolution means in this sentence. Is it meant in development or in the works?
L78: change “soil porosity” to “soil pores”
L79-80: change “macroporosity” to “macropores” and “microporosity” “micropores”
L104: delete “than aerobic respiration take place.” Uncessary in sentence.
L104-105: change “…reduction is an important one” to “reduction is an important process.”
L107: change “macroporosity” to “macropores”
L109: Would this actually increase soil CO2 efflux temperature sensitivity or the apparent temperature sensitivity? Also as compared to aerobic conditions? On a molecular level it might
be argued that it is changing the substrate amount or the carbon use efficiency as opposed to the temperature sensitivity. Some more explanation is needed here.
L113: sorption capacity of what?
L115-116: Is this quotation from the article really necessary? The quote is rather general and the
preceding phrase describing the study is also lacking detail.
L119: “th” to “the”
Materials and methods:
L142-143: Change to “lowest elevations in the riparian zone.”
Figure 1. Description: “kervidy” needs capitalized.
L153: What was the average slope in the watershed or the slope at the different sampling points?
Why were these two particular transects selected?
L154: change to “lower part”
Table 1: How was pH measured? On a 1:1 soil:water mixture or another method?
L180-183: What time of day were the measurement taken? Was the order of measurements
randomly done or always in the same order?
L192: What kind of filters were used?
L194: delete “from”
L194-198: The description of the analyzer and procedure to estimate DOC and DIC is a bit
confusing. The analyzer is given first at the TOC analyzer and then it is mentioned how the samples were acidified and then measured via a combustion-oxidation method. Was this on the
TOC analyzer? Is there a reference for this method? How was the sample acidified?
L196:
L201: What wavelength was used for the determination of Fe2+?
L212-213: Does it mean the difference in the function or the different in model fit or model error? Please clarify.
L215: change “Fco2 are controlled” to “is controlled”
L231: Not clear what is “DOC is the concentration in the macroporosity” Was this the measured
DOC concentration from the piezometers?
General: It is not mentioned in the data anlaysis part how the serial collinearity of the samples (e.g. time series) was treated. Were all the time period modeled as one? Was there a seasonal aspect? Some more information is needed.
Results:
L262: Not clear what repartition refers to.
L263: Does it mean here atleast 0.5 mm?
L270: Replace “precipitations” with “precipitation”
L273: “began” is written twice in the sentence, delete second occurrence.
L335-350: This descriptive section of the DIC and DOC concentrations can probably be shortened, as there were no statistical tests done comparing different time points (restricted by the experimental design), so many phrases such as “slowly and regularly decreased” and “decreased steadily” are probably more description then necessary as there is no way to tell if
this is actually a statistical increase/decrease or just natural noise.
L369: not clear what situations “when the calculation of a p-value was possible” refers to.
Clarification is needed here.
L376: Suggestion to explicitly write “negatively correlated” in this phrase.
Figures: Sometimes a “,” is given instead of a “.” decimal separator.
L423: Not clear what is meant in this sentence that “residuals are important” and how this relates to the model results. Please elaborate.
L481: Part of sentence is missing.
L502: Does “this latter paper” refer to the current paper, if so change to “in the current study.”
L515: “transfer”
L553: sentence ending in “induced” seems to be an incomplete thought.
L572: a “humidity transect” could also be associated with air humidity. Suggested change to soil wetness transect or soil drainage transect.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors monitored two hillslope transects over 7 months and recorded surface CO2 efflux, soil temperature, soil water content as well as dissolved organic and inorganic carbon contents in soil pore water. Their aim was to understand how the changes in the water table level would influence the degradation of soil organic matter.

While the technical side (planning, sampling, measurements, equipment used etc.) is mostly fine, the manuscript itself was badly prepared and is far from ready for journal submission. It actually reads like a first draft and I couldn’t shake the feeling that the manuscript was send out just because the holiday season was coming up and the authors wanted it out of their way, so the reviewers could work on it while they were away. I simply can’t believe that a manuscript with no less than 8 (!) authors would be in this shape if all authors would have done their job properly. Personally, I find it disrespectful and outrageous to send out half-baked manuscripts and hope the reviewers do the corrections and deliver ideas for discussion.

As I already passed more than a full day on reading this long and tedious manuscript I will refrain from listing all mistakes and errors I found. Instead I will only address the more urgent points and leave the polishing of the manuscript to those whose job it is a.k.a. the authors.

The first thing I noticed was that not even a basic English language revision was done as already seen in the title (subject „water table dynamics“ is plural while the verb „controls“ is singular). Apart from a long list of grammatical errors and the notoriously wrong use of the word „macroporosity“ (instead of „macropores“), the authors also construct unnecessary complicated sentences, which often end up with a wrong syntax and make it hard to follow the story. In other parts the language is sloppy and imprecise.

Abstract

L17-18: „these three destabilization pathways“ – according to the preceeding sentence these are: the leaching of DOC, the leaching of DIC and fluxes of CO2. Strictly speaking a pathway contains several stages (desorption of SOM -> solubilized DOC -> uptake and degradation by MOs -> CO2 = degradation pathway), while the named „pathways“ are actually processes. While the first one can be considered as a destabilizing process (as part of the SOM is removed), the last two are not. DIC and CO2 flux are the results of SOM degradation. The authors themselves state in line 466 that DIC is a by-product of DOC mineralization. SOM degradation or destabilization happens through physical, chemical or biological processes like partial dissolution and leaching, oxidation or microbial degradation. This imprecision in what is actually a degradation or destabilization process and what is the result of it (and can be used as a proxy) unfortunately runs through the whole text. The authors hence need to overhaul the whole manuscript.

Introduction

The introduction is way too long and should be shortened by at least a third.

L37: The concept of geosynthesis of geomacromolecules recalcitrant to biodegradation – doesn’t make sense (without the „geosynthesis“ it works)

L37-40: This is an oversimplification. The concept of recalcitrance due to chemical structure has not been completely thrown out, but was rather extended by the concept of accessibility. Kleber (2010) in Environmental Chemistry 7 (+ response to comment) nicely explains how different chemical structures require different amounts of oxygen for microbial degradation.

L120: Is DOC a primary control of Soil CO2 efflux? Another imprecision. What precisely is meant? Its structure? Its concentration?

L123-124: ...sampling soil solutions moving freely in soil macropores... (repeat over and over in the text)

M&M

Water table measurement is only mentioned in caption of Figure 1. As the data is displayed and used a description of the equipment should be added.

Table 1 – How was pH determined (H2O? CaCl2? Reference?)?

L189: How much soil pore water was sampled per replica?

L193: Filters are usually washed with part of the sample which is discarded. I assume the filtration unit did not have a problem with dead volume? Otherwise washing with ultrapure water leads to sample dilution.

L200: the AFNOR reference is missing

What about pH? Was pH not determined? If not, why?

Eq. 1 and 2: Is CO2 the same than F_CO2used so far in the text?  If yes, stick to the same naming!

Eq.4 and 5: I don’t really understand what the use of these equations actually is. Why are temperature and water content suddenly ignored. Furthermore, no results have been displayed.

Results

Results should be consistently presented in past tense.

Figures 2 and 3: y-axis legends use commas as decimal separator instead of  a period.

Marking the named periods A-C in the figures would help

The data evaluation and presentation needs to be thoroughly reworked.  I strongly recommend to describe by sites (K and G) and always in the same order. I don’t see how grouping and discussing two sampling points from separate transects should make sense.

L314-316: Discussing average values over the whole period is nonsense and doesn’t help. Especially as the significant difference is definitely not seen at the beginning of the sampling period.

Figure 3: I recommend grouping data by transects. DIC and DOC can be displayed on on scale as they are in similar orders of magnitude. CO2 can go on a secondary axis.

Use Figure type „X/Y with lines“. What about the error bars of DIC and DOC?

L335-337: same here. Average values over the whole period are useless.

L352-354: and again. Not useful.

L370: I seriously doubt that four sampling points are enough to represent an area of 5 km2! Especially as the sampling points specifically focus on slopes close to the streams.

I know that presenting lots of correlations is a difficult task, but the current way with tabeled values in the SI (which in itself is fine) and verbal descriptions of the correlations found is an imposition for the reader. Especially as the authors are not stringent in the order they describe the sampling points. Soemtimes they start with the up position, sometimes the down position, sometimes positions across transects are grouped. Result description needs to follow a very stringent line in order not to loose the reader. If the authors are convinced that two positions of different transects should be compared than they should do that in the discussion and give a good reason why they think they can and should be compared.

I recommend to loose the catchment scale as 4 sampling points are anyway too few for the whole catchment and focus on evaluating and discussing the two transects. The detected and discussed strong correlations should be displayed in figures (can also go to the SI). There are also highlighted correlations in the table which are not discussed. Also the linear regressions (at least the discussed ones) should be displayed in the SI.

Discussion

This section looks structured, but is not. Chapter 4.1 claims to deal with DOC concentration as a driver for CO2 efflux, but the first sentence is about temperature and soil water content. Most of the first paragraph then discusses Q10. It’s only the second half of this section which gets to the point and in an unnecessary cumbersome way.  In principle the authors wanted to state that while modeling the efflux using only Temp and WFPS as variables already does a good job, but the fit is much improved when DOC conc. is also considered as variable. (L440: „...explicitely representing DOC in Eq.1 may improve ..“ – the result section already showed that it does, no need for „suspense“ here). Switch the two paragraphs, get to the point right away and then substantiate your viewpoint.

L457: significant differences between the two topographic transects – of what exactly? The recorded data? The calculated correlations?

L465: Why are micropores suddenly important?

L465-468: Discussion of DIC needs to consider pH! This is actually acknowledged by the authors in lines 530 and 533-535, but nevertheless no pH-values of the soil solutions are reported. This is a weak point in the discussion.

L481: Then it is submited to thermodynamic rules and has a kinetic. – This is an example of why I believe no one really cared about this manuscript. With 8 authors someone should have spotted this useless and hollow sentence.

By the way, in chapter 4.2 „catchment“ is suddenly replaced by „landscape“.

Chapter 4.4

SOM respiration – that doesn’t make sense. SOM is the substrate, it’s the MOs which respire not the SOM.

Chapter 4.5

This section is completely superfluous. This is common knowledge and the current study is not needed to state this.

What I miss in the discussion is a consideration of the land use. The authors chose two different transects with different vegetal cover. This is not discussed at all. While they removed all vegetation for efflux measurement, the rhizospheres and SOM chemical structure in both sites are likely different and do influence the sampling of soil solution. Accessibility surely is one important factor for degradation to happen, but chemical structure cannot be completely ignored even if modelers don’t like it as „quality“ is difficult to quantify.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The study tackles a very interesting aspect of the OC dynamics, which is the affect of topography and moisture content on DOC, DIC and CO2 fluxes. I think that the study is written well. There are some minor editorial mistakes in the language, and the introduction section may be shortened and made more concise. Also, the dependence of water table dynamics on the topography may be clearly outlined including in the abstract. The statistical methodology is sound, and explained quite well. I do not have any additional comments on this study and recommend that it may be accepted in its present form. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

General comments:

Overall the manuscript is much improved and reads better.

There are still a few concerns regarding the statistics and data presentation. The initial review comment regarding autocorrelation with the time series data was not addressed for Equations 1 and 2. Additionally, there are the results of statistical tests which are not explained in the materials and methods, such as comparisons of FCO2, DOC, DIC between and within transects.

Also one of the main findings is that including DOC along with WFPS and temperature improves prediction of FCO2. Due to the fact that there was no water present in the lysimeters at certain sampling dates, there was no DOC on some dates when FCO2 was measured. This also varied depending on transect and position. To be able to compare the two equations/approaches, the same number of data points should be use. This will reduce the overall number, but would facilitate a more straightforward comparison. I’ve made more specific comments and suggestions below.

Please see the specific comments below:

Abstract:

L23: “encounters with microorganisms…” This can be implied, but was not really measured, even indirectly in the study. I would suggest dropping this phrase as microbial processes is already mentioned right after that.

L25: Not clear what “the main descriptors of this dynamic…” is referring to. Is it water table dynamics? Also do descriptors refer to the measured properties of the water table dynamics that were used in the multi-regression models to explain DIC, DOC, and FCO2?

Missing in the abstract is the modeling that was used to explore some of these relationships, the only mention is L26-27 where there is a mention of FCO2 being correlated to DOC, which leaves out some of the other exciting findings from the study.

Introduction:

L63: Please give a specific size for defining macropores.

L81: Here does “water velocity” refer to downwards flow, lateral flow, or surface flow? Please specify. 

L84: Is this increase in DOC only due to a release of carbon from Fe, or can it also be seen as a build up due to the suppression of microbial action due to anaerobic conditions?

L95-104: What I think is missing here is a short explanation of why this site is important. The reader can surmise that it is important as it is part of the French CZO, but for the broader applications of this study, it would reinforce the findings if the reader knows for example that this site represents lowland agricultural sites of NW France or in general northern mainland Europe, or it is a certain physiographic region, or represents certain soil types.

Materials and Methods

L116: Are the soils formed from schist residuum or are they colluvial in nature, especially being at the footslope position?

L118: Please give some more detail here. How long has the watershed been under the current land use?

Table 1: Are the values for “days where water table was above 10 cm soil depth” correct? It seems surprising that for the well-drained soil, there were 80 days, while for the poorly and very poorly soils, it was less. 

Also “Fe total”, is this soil Fe or solution Fe?

L164: What type of screen is used? What is the mesh size?

L173: change to “filtered through a 0.2 um cellulose acetate membrane…”

L186: space need in between “uncertainties” and “were”

L188: The initial step of what? I assume modeling or data analysis, but please specify.

L194-195: This is a bit confusing here. The sentence refers to the latter study, ref. 38, which found that DOC and temperature, even though were modeled with an exponential equation, the equation ended up being very close to linear over the analyzed range.

L253-254: Suggestion to change “pumping of water by plants” to precipitation is relatively high and there is low evapotranspiration”

L255: “organic surface horizons” This implies an “O” horizon, which from the description in the materials and methods and also the general soil characteristics, is not present here. They are mineral soil horizons. It would be better to either term this the “organic matter-rich surface horizon” or just “surface horizon.”

Results

DIC, DOC, and FCO2 values are being compared between transect (K and G) and within a transect in several places in the text (e.g. L301-303, L325-327, L341-343), with means, standard deviation and p-values reported. How were these statistical comparisons made? This should be explained in the materials and methods. The question, along with the time series analysis, is are the assumptions for statistical comparison met, that is independent samples, homogenatity of variance, etc. I would imagine that within a transect, different values are correlated. If a statistical comparison is still absolutely necessary, in my opinion, it would be needed to account for this spatial autocorrelation within a transect and use some type of nested model where both transect and position within a transect is included. 

Table 2: It isn’t mentioned anywhere if all of the model coefficients were significant or not. Also there are no error terms for any of the estimates. This is probably not needed or very interesting for a,b, and c, but would be useful for Q10, since this is discussed in the results and would give some indication for example the uncertainty for the Kdown value of 5.36 in the first equation.

In this table, maybe think about what is most important to show. As is presented now, it seems like the main comparison is how each position varies from the whole data set. My take-aways from reading the table as it is arranged now are 1) that error is reduced in most cases by modeling each slope position independently regardless of equation, 2) the estimated Q10 of the whole data set is usually lower then when each position is modeled regardless of equation, and 3) that equation 2 performs better than equation 1 in this order. For the discussion, I do not think #1 is that important in the overall results but rather #3 and possibility #2. As a suggestion, compare modeling results for equations #1 and #2 for the full data set and then by position. Additionally to make them more comparable and to reduce confounding factors, the dataset can be reduced for Equation 1 so they are the same sampling points used for equation 2. At the moment, the # of DOC is different by transect and position, so this makes any comparison between Equation #1 and #2 difficult. If comparing Q10 values between the different transects and positions is very important, then a mixed modeling approach would be appropriate where you could allow Q10 to vary by transect X position, or if it would not be significant just by position. That way a unified model could be used which would then estimate show the effect of position on Q10.  

L363 & L367: No clear what “analyzed with the four points” refers to? Does this mean that season was used as a model factor or that the values were averaged by season. Not clear.

L397-398: The AICc is actually lower for all 4 points and Kdown for Equation 2.

Discussion

L433: What is meant by “moisture function” in this sentence? Does it mean improve an equation using soil moisture content when modeling respiration?

L439: What was positively correlated?

L436 & 439: Change “under an exponential equation” to using an exponential equation.

L441-446: Comparing equation 1 to equation 2 (inclusion of DOC) here is problematic, as the datasets used for either equation are different. For example Equation 1 for all 4 points n = 71, while for Equation 2 n = 45. Without having any information on how the two populations are, it is difficult to draw any conclusions from the R2’s and parameter fits (i.e. Q10) and to say if the modeling improvement is due to the inclusion of DOC or rather it is just due to the subset of data points used.

L595: needs to be space between “CO2” and “gas” and “gas” needs to not be in superscript.

Figure 3: Please provide a symbol key in the first graph for the DIC, DOC, and FCO2 symbols. This will make the interpretation much easier as compared to going down to the figure description.

Author Response

Dear colleague,

You will find my answers in green in the following. As you will see, according to your comments, modeling FCO2 with Equation 1 (WFPS and soil temperature as variables) has been tested using the same dataset as for Equation 2 (WFPS, soil temperature and DOC as variables). With this approach, the improvement of the modeling was much less evident since the only improvement was in the p-value of the Q10 parameter for Kdown location. As a consequence, the modification of Equation 1 into Equation 2 has been removed from the manuscript and the paragraph 4.1 has been reworded.

Many thanks for your time and consideration in improving our manuscript.

Laurent Jeanneau

NB: Line numbering in this document corresponds to the document with apparent modifications (“track-change” version).

General comments:

Overall the manuscript is much improved and reads better.

There are still a few concerns regarding the statistics and data presentation. The initial review comment regarding autocorrelation with the time series data was not addressed for Equations 1 and 2. Additionally, there are the results of statistical tests which are not explained in the materials and methods, such as comparisons of FCO2, DOC, DIC between and within transects.

The part of the article dealing with the development of Equation 2 by adding DOC concentration to WFPS and soil temperature as variables to model F_CO2 was removed. Statistical tests were finally not used since the assumptions of normality and homogeneity of variance were not met.

Also one of the main findings is that including DOC along with WFPS and temperature improves prediction of FCO2. Due to the fact that there was no water present in the lysimeters at certain sampling dates, there was no DOC on some dates when FCO2 was measured. This also varied depending on transect and position. To be able to compare the two equations/approaches, the same number of data points should be use. This will reduce the overall number, but would facilitate a more straightforward comparison. I’ve made more specific comments and suggestions below.

We tried to model F_CO2 with Equation 1 with the same sampling dates as for Equation 2. With this approach, the improvement of the modeling with the addition of DOC as a variable was much less evident since the only improvement was in the p-value of the Q₁₀ parameter for Kdown location. As a consequence, the modification of Equation 1 into Equation 2 by adding DOC as a variable using an exponential equation has been removed from the manuscript. The paragraph 4.1 has been reworded.

Please see the specific comments below:

Abstract:

L23: “encounters with microorganisms…” This can be implied, but was not really measured, even indirectly in the study. I would suggest dropping this phrase as microbial processes is already mentioned right after that.

It has been removed.

L25: Not clear what “the main descriptors of this dynamic…” is referring to. Is it water table dynamics? Also do descriptors refer to the measured properties of the water table dynamics that were used in the multi-regression models to explain DIC, DOC, and FCO2?

Yes, the word “descriptors” refers to the dynamic of the water table. It has been modified.

Missing in the abstract is the modeling that was used to explore some of these relationships, the only mention is L26-27 where there is a mention of FCO2 being correlated to DOC, which leaves out some of the other exciting findings from the study.

A sentence has been added.

Introduction:

L63: Please give a specific size for defining macropores.

Macropores in soil are defined as >75µm (Brewer R., 1964). It has been added (line 93).

L81: Here does “water velocity” refer to downwards flow, lateral flow, or surface flow? Please specify. 

It refers to lateral flow. It has been added (line 116)

L84: Is this increase in DOC only due to a release of carbon from Fe, or can it also be seen as a build up due to the suppression of microbial action due to anaerobic conditions?

The assumption in the cited references is the release of organic carbon previously adsorbed to iron oxides.

L95-104: What I think is missing here is a short explanation of why this site is important. The reader can surmise that it is important as it is part of the French CZO, but for the broader applications of this study, it would reinforce the findings if the reader knows for example that this site represents lowland agricultural sites of NW France or in general northern mainland Europe, or it is a certain physiographic region, or represents certain soil types.

A precision has been added in the text (line 146-147).

Materials and Methods

L116: Are the soils formed from schist residuum or are they colluvial in nature, especially being at the footslope position?

They are mainly formed from schist residuum at backslope positions and are colluvial in nature at footslope positions. It has been added in the text (line 159-161).

L118: Please give some more detail here. How long has the watershed been under the current land use?

French Brittany has been one of the most productive agricultural regions in Europe since the Second World War. It has been added in the text (line 167-168).

Table 1: Are the values for “days where water table was above 10 cm soil depth” correct? It seems surprising that for the well-drained soil, there were 80 days, while for the poorly and very poorly soils, it was less. 

This apparent contradiction could come from the difference in land-use, or that they are not located along the same topographic transect.

Also “Fe total”, is this soil Fe or solution Fe?

This is soil Fe. It has been added in the table 1.

L164: What type of screen is used? What is the mesh size?

The well screen is a PVC tube pierced with 1 mm slots. It has been added in the text (line 218).

L173: change to “filtered through a 0.2 um cellulose acetate membrane…”

The modification has been done.

L186: space need in between “uncertainties” and “were”

The modification has been done.

L188: The initial step of what? I assume modeling or data analysis, but please specify.

“Initial step” was useless and has been removed (line 245)

L194-195: This is a bit confusing here. The sentence refers to the latter study, ref. 38, which found that DOC and temperature, even though were modeled with an exponential equation, the equation ended up being very close to linear over the analyzed range.

It seems that you get the point.

L253-254: Suggestion to change “pumping of water by plants” to precipitation is relatively high and there is low evapotranspiration”

This has been modified (line 312-313).

L255: “organic surface horizons” This implies an “O” horizon, which from the description in the materials and methods and also the general soil characteristics, is not present here. They are mineral soil horizons. It would be better to either term this the “organic matter-rich surface horizon” or just “surface horizon.”

The first suggestion has been chosen (line 314).

Results

DIC, DOC, and FCO2 values are being compared between transect (K and G) and within a transect in several places in the text (e.g. L301-303, L325-327, L341-343), with means, standard deviation and p-values reported. How were these statistical comparisons made? This should be explained in the materials and methods. The question, along with the time series analysis, is are the assumptions for statistical comparison met, that is independent samples, homogenatity of variance, etc. I would imagine that within a transect, different values are correlated. If a statistical comparison is still absolutely necessary, in my opinion, it would be needed to account for this spatial autocorrelation within a transect and use some type of nested model where both transect and position within a transect is included. 

We agree with this comment; however p-values are often used to ascertain an observed difference. Here, the Student t-test was used to calculate the p-values, but the assumptions of normality and homogeneity of variance were not met. Then the non parametric Kruskal-Wallis test was used (r function: kruskal.test) however the p-values were higher than 0.4, then it was not possible to reject the null hypothesis. Then p-values were removed from the text.

Table 2: It isn’t mentioned anywhere if all of the model coefficients were significant or not. Also there are no error terms for any of the estimates. This is probably not needed or very interesting for a,b, and c, but would be useful for Q10, since this is discussed in the results and would give some indication for example the uncertainty for the Kdown value of 5.36 in the first equation. 

In this table, maybe think about what is most important to show. As is presented now, it seems like the main comparison is how each position varies from the whole data set. My take-aways from reading the table as it is arranged now are 1) that error is reduced in most cases by modeling each slope position independently regardless of equation, 2) the estimated Q10 of the whole data set is usually lower then when each position is modeled regardless of equation, and 3) that equation 2 performs better than equation 1 in this order. For the discussion, I do not think #1 is that important in the overall results but rather #3 and possibility #2. As a suggestion, compare modeling results for equations #1 and #2 for the full data set and then by position. Additionally to make them more comparable and to reduce confounding factors, the dataset can be reduced for Equation 1 so they are the same sampling points used for equation 2. At the moment, the # of DOC is different by transect and position, so this makes any comparison between Equation #1 and #2 difficult. If comparing Q10 values between the different transects and positions is very important, then a mixed modeling approach would be appropriate where you could allow Q10 to vary by transect X position, or if it would not be significant just by position. That way a unified model could be used which would then estimate show the effect of position on Q10.  

Modeling F_CO2 with Equation 1 using the same sampling points used for Equation 2 was tested, and the comparison shows that the modeling was not improved by adding DOC. This could be due to the low number of sampling points. This point was not the most important message of the article, as a consequence testing the Equation 2 was removed from the article. Adding DOC as a variable to model F_CO2 could be a good idea but the study should surely include more sampling points to have a robust comparison. Equation 1 was kept and p-values and standard errors for Q₁₀ were added in Table 2.

L363 & L367: No clear what “analyzed with the four points” refers to? Does this mean that season was used as a model factor or that the values were averaged by season. Not clear.

It means when the four locations were grouped. In the previous version, the wording was “at the catchment scale”, but the term “catchment” was under question since we analyzed only four points along two topographic gradients. The wording was changed into “when the four points were analyzed together” (line 450).

L397-398: The AICc is actually lower for all 4 points and Kdown for Equation 2.

Since the modeling of FCO2 with DOC as a variable was removed, AICc was useless and it has been removed.

Discussion

L433: What is meant by “moisture function” in this sentence? Does it mean improve an equation using soil moisture content when modeling respiration?

It means the modeling of soil heterotrophic respiration as a function of soil moisture. It has been modified in the paragraph 4.1. (line 573)

L439: What was positively correlated?

FCO2 and DOC were positively correlated concentrations as mentioned in the text.

L436 & 439: Change “under an exponential equation” to using an exponential equation.

It has been modified.

L441-446: Comparing equation 1 to equation 2 (inclusion of DOC) here is problematic, as the datasets used for either equation are different. For example Equation 1 for all 4 points n = 71, while for Equation 2 n = 45. Without having any information on how the two populations are, it is difficult to draw any conclusions from the R2’s and parameter fits (i.e. Q10) and to say if the modeling improvement is due to the inclusion of DOC or rather it is just due to the subset of data points used.

According to this comment, the discussion on the inclusion of DOC concentration as a variable in the modeling of FCO2 has been removed since the present dataset is not sufficient to test it.

L595: needs to be space between “CO2” and “gas” and “gas” needs to not be in superscript.

A space has been added and “gas” has been written in superscript.

Figure 3: Please provide a symbol key in the first graph for the DIC, DOC, and FCO2 symbols. This will make the interpretation much easier as compared to going down to the figure description.

It has been done.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Please see attachment

Comments for author File: Comments.pdf

Author Response

Dear colleague, additional information has been added in the text on the four points that you have indicated. You will find more precisions in the attached document. Thank you for your time and consideration that have improved the quality of the manuscript.

Have a good day,

Laurent Jeanneau

Author Response File: Author Response.pdf