the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 Jan 2026
Precipitation reduction overrides edaphic controls on nitrous oxide emissions along a soil carbon, texture and pH gradient in a cereal field
Abstract. Nitrous oxide (N2O) is a potent soil-borne greenhouse gas (GHG) which increases in the atmosphere due to the widespread use of synthetic nitrogen (N) fertilisers. Soil N2O emissions are intrinsically controlled by soil moisture and edaphic properties such as soil organic carbon (SOC) content, texture and pH. With a future climate projected to increase frequency and severity of droughts in northern Europe, understanding how these factors interact to affect N2O emissions is critical for predicting climate feedbacks. In this study, we investigated N2O emissions along a hillslope gradient in an agricultural field in southeast Norway, characterised by increasing SOC and clay content and decreasing pH from top to bottom. Eight rainout shelters were installed along the hillslope, nominally reducing precipitation by 49 %. N2O emissions were measured weekly using static chambers over two years during the snow-free period. In the first year, N2O emission measurements started two months after fertilisation and covered a four-month period, which included episodes of heavy rainfall; during this time, we found no effect of precipitation reduction or edaphic factors on emission rates. In the second year, reduced precipitation significantly decreased N2O emissions (~25 %). Under ambient precipitation, cumulative N2O emissions were positively correlated with SOC content and showed weak negative and positive trends with soil pH and clay content, respectively. No significant correlations were found in plots with reduced precipitation. Altogether, our findings illustrate that soil physicochemical controls on N2O emissions depend on the interaction between soil properties and climate. This has consequences for parameterising process-based N2O models driven by soil properties and weather and calls for more in-depth studies on interdependencies of edaphic and climatic drivers of N2O emissions.
- Preprint
(1065 KB) - Metadata XML
-
Supplement
(440 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-210', Anonymous Referee #1, 21 Oct 2026
-
RC2: 'Comment on egusphere-2026-210', Anonymous Referee #2, 22 Oct 2026
General evaluation
Overall, this manuscript addresses a relevant topic and provides interesting field-based information on the effects of reduced precipitation and the relationships between edaphic parameters and soil N₂O emissions. The experimental approach has potential to improve our understanding of how soil properties and moisture conditions interact in controlling N₂O emissions. The two-year experiment is particularly interesting because it captures contrasting environmental conditions and explores how precipitation may influence the relationships between spatial variation in soil properties and N₂O emissions. However, several aspects of the experimental design, statistical analysis and interpretation require further clarification or revision before the conclusions can be fully supported.
My main concerns relate to aspects of the experimental design and statistical analysis, the limited monitoring period in the first experimental year, and the relatively low measurement frequency, which may have resulted in important emission peaks being missed. The statistical approach also requires clarification, especially regarding the treatment of the paired rainout-control plots, the use of pseudoreplicates, the mixed-model structure, and the use of unpaired statistical tests despite the apparent paired experimental design. In addition, the relationships between cumulative N₂O emissions and individual soil properties should be interpreted more cautiously given the small sample size, multiple comparisons, and covariation among SOC, clay content and pH along the hillslope.
The Discussion and Conclusions should therefore more clearly acknowledge these limitations and ensure that mechanistic interpretations are proportional to the statistical evidence provided. With appropriate revisions to the statistical analysis and a more cautious interpretation of the results, the study could provide a useful contribution to understanding the combined influence of precipitation and soil properties on agricultural N₂O emissions.
Specific comments
- L9: The use of synthetic fertilizers represents only one source of N₂O emissions from agricultural soils. Organic fertilizers and crop residues can also be important sources. I therefore suggest using the broader formulation “due to the widespread use of nitrogen fertilizers”.
- L13: “decreasing pH from top to bottom” → Please see Comment 19c.
- L16: N₂O measurements only started approximately two months after fertilization. However, substantial N₂O emission peaks may occur shortly after fertilizer application and contribute considerably to cumulative emissions. Although the implications of the limited monitoring period are discussed in Lines 262–269, please specifically address how the delayed start of monitoring may have affected the conclusion that N₂O emissions did not differ clearly between the control and rainout treatments. Could an earlier start of N₂O monitoring have affected this conclusion?
- L16: Why was the measurement period limited to only four months? Depending on weather and soil conditions, relevant N₂O emission events may also have occurred outside this monitoring period. Please justify the selected measurement period in the materials and methods section and discuss its implications for the cumulative emission estimates in the Discussion section.
- L31: Consider reporting the atmospheric N₂O mixing ratios for both 1750 and 2018, in addition to the relative increase.
- L38: “at low soil moistures” → “at low soil moisture”
- L38–39: “high soil moistures” → “high soil moisture.”
- L92: A total of 16 plots were established. Please clarify the level of replication in the experimental design. As pseudoreplication is mentioned in the statistical analysis, the distinction between experimental replicates, paired plots, and pseudoreplicates should be clearly explained here.
- L97: Please briefly describe the recent agricultural history and management of the experimental field, including previous crop(s), tillage practices, and management/incorporation of residues from the preceding crop.
- L99: Was the fertilizer applied as a single dose?
- L100: Please specify the area sampled for yield determination. The manuscript states that barley rows were cut over a length of 1 m, but it is unclear how many rows were included in each yield sample.
- Table 1:
- Please indicate in the table caption how dates are formatted (I assume DD.MM). Why are the sowing and fertilization dates for 2023 not reported?
- The rainout shelters appear to have been installed for a substantially longer period in 2024 than in 2023. Please explain the reason for this difference. Precipitation or drought during spring 2023 may have influenced subsequent N₂O emissions during the monitoring period, whereas in 2024 the rainout treatment apparently also reduced precipitation during spring. This difference in treatment duration should be considered when comparing the two experimental years. Also see comment 13.
- L109: Please explain the rationale for starting N₂O monitoring approximately two months after fertilization in 2023, particularly given that monitoring started considerably earlier in relation to crop management in 2024.
- L109: N₂O emission peaks frequently occur during the days and weeks following fertilizer application. Please explain why the sampling frequency was not increased during this period. With only one measurement per week, short-lived emission peaks may have been missed, potentially affecting the cumulative emission estimates.
- L109-110: What are the dimensions of the static chamber? What soil surface area is covered and what is the volume of the headspace used regarding the barley that is growing during spring and summer.
- L117: How were concentration-time relationships that did not adequately follow a linear pattern handled? Please specify the quality criteria used to accept or reject the linear fits used for flux calculation, for example the goodness-of-fit criteria and any other quality-control thresholds applied.
- L121: Please explain the rationale for selecting a sampling depth of 0–20 cm and for collecting two soil cores per plot for mineral N determination, which were subsequently pooled. How were the sampling locations within each plot selected?
- L125: It is mentioned that stable isotope ratios (C and N) were analysed but these data are not shown nor discussed.
- Table 2:
- The rainout shelters are 2.5 × 2.5 m, which I assume corresponds to the dimensions of the experimental plots. Please clarify what “metres from hilltop” refers to, for example the plot centre, upper edge, or another reference point. It is not clear how these distances relate to the plot dimensions.
- Please indicate the sampling replication for the different soil properties in the table caption (n=…), where relevant, to make the sampling basis of the reported values immediately clear
- With the exception of plots G and H, pH appears relatively similar across most positions along the hillslope. A clear gradual decrease in pH from the top to the bottom of the slope is therefore difficult to identify from Table 2. Please substantiate this spatial trend, for example graphically and/or statistically, or describe the pattern more cautiously in the Results.
- L144–153:
- The description of the linear mixed-effects analysis requires further clarification. Please provide the complete model structure, including the fixed and random effects, how the paired design and sampling time were incorporated, and how plots along the hillslope gradient were represented in the model. Please also explain explicitly how repeated measurements on the same plots were accounted for and whether temporal correlation was assessed. Finally, clarify how pseudoreplication was addressed.
- Given the paired experimental design, it is unclear why a Wilcoxon rank-sum test was used rather than a Wilcoxon signed-rank test, which accounts for paired observations. Similarly, please clarify why an unpaired rather than a paired t-test was used for DMY. Please justify treating the observations as independent or consider re-analysing these variables using tests that preserve the paired structure.
- Because control plots were consistently positioned uphill from their corresponding rainout shelter plots along a gradient in soil properties, please discuss whether this systematic spatial arrangement could confound the estimated precipitation-treatment effect.
- L185: The manuscript states that differences in mineral N content across both years were tested using a linear mixed-effects model. Please describe this analysis more explicitly in the Statistical Analysis section and provide the complete model structure.
- L186–188: Surface runoff along the hillslope could potentially introduce an artefact into the experimental design by redistributing water and nutrients between plots. Please discuss this potential limitation. If measures were taken to prevent or minimize runoff between plots, these should be described in the Materials and Methods section. Also subsurface water movement may have an impact on soil water dynamics. Shortly discuss this.
- L214–218: See Comment 20c regarding the use of an unpaired Wilcoxon rank-sum test for these data.
- L235–246:
- Please provide additional information on the regression analyses in the Statistics section, including the number of observations included in each regression and whether separate univariate models were fitted for each soil property, treatment and year. Given the small number of plots and the number of relationships examined across soil properties, please also consider the implications of multiple testing when interpreting individual p-values. Importantly, a significant relationship between a soil property and cumulative N₂O emission in the control treatment and a non-significant relationship in the rainout treatment does not in itself demonstrate that the relationships differ between treatments. If the authors wish to conclude that precipitation modified or “cancelled out” the influence of a soil property, this difference should be tested directly, for example through a soil property × precipitation treatment interaction. Otherwise, the interpretation should be formulated more cautiously.
- I also recommend avoiding terms such as “near-significant” or “near-significant trends” for relationships with p > 0.05. Instead, report the estimated relationships, R² values, and p-values without assigning additional categories of statistical significance.
- L270–272: Given the strongly varying soil moisture conditions during the experimental period, including the initially dry conditions followed by heavy precipitation events, together with temporal variation in mineral N availability, cumulative N₂O emissions integrate periods during which N₂O production may have been controlled by different limiting factors. This may partly explain the weak relationships between cumulative N₂O emissions and individual soil properties, particularly in the first year. Please discuss this limitation and consider more explicitly how temporal variation in soil moisture and mineral N availability may have affected the observed relationships between cumulative N₂O emissions and edaphic properties. This consideration is also relevant to the interpretation of the soil-property relationships discussed in Section 4.2.
- L321-322: “We found no relationship between POM and N₂O emissions.” and “suggesting that the positive correlation between SOC and N₂O was not driven by C availability.” As far as I can determine from the manuscript, POM was not included among the soil properties described in the Materials and Methods or Results. Although the manuscript states that these data are not shown, I suggest providing the POM data and describing the corresponding analysis, as POM may provide relevant information regarding C availability and the potential for denitrification. Please also indicate when and how the samples used for POM determination were collected.
- L334: “In our study, SOC and clay content were closely correlated” Please quantify the reported relationship between SOC and clay content by providing the correlation coefficient and p-value, as their covariation is important for the subsequent interpretation of their relationships with N₂O emissions.
- L341:
- “Both SOC and clay content were negatively correlated with pH.” Please provide the statistical evidence supporting these relationships, including correlation coefficients and p-values.
- Please avoid the term “near-significant.” The relationships should instead be described based on their direction and magnitude together with the reported p-values.
- L344: “The decline in pH downslope” → A clear gradual decline is difficult to identify from Table 2, particularly across the upper and middle parts of the hillslope. Please substantiate the relationship between hillslope position and pH, for example using a figure and/or an appropriate statistical analysis, or use more cautious wording. Also see Comment 19c.
- L346–348: “This interconnectedness indicates that even if pH appears to be a dominant factor controlling N₂O emissions, its effects are modulated by SOC and clay content, making it impossible to fully separate their contributions.” This conclusion appears stronger than can be supported by the analyses presented. Correlations among SOC, pH, and clay content demonstrate covariation among these soil properties, but do not demonstrate that SOC or clay content modifies the effect of pH on N₂O emissions. Demonstrating such modulation would require testing interaction effects. Given the small sample size and the collinearity among these variables, however, such analyses may have limited statistical power. I therefore suggest rephrasing this conclusion to state that the covariation among SOC, pH, and clay content makes it difficult to disentangle their individual associations with N₂O emissions.
- L373–380: The conclusion provides a useful synthesis of the main findings. However, I suggest explicitly acknowledging that these findings require confirmation in additional experiments, particularly given the relatively small number of experimental units and the limited N₂O monitoring period in the first experimental year.
Citation: https://doi.org/10.5194/egusphere-2026-210-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 942 | 339 | 147 | 1,428 | 209 | 123 | 227 |
- HTML: 942
- PDF: 339
- XML: 147
- Total: 1,428
- Supplement: 209
- BibTeX: 123
- EndNote: 227
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
A solid and interesting contribution on how and under what circumstances edaphic factors determine soil N2O emission from agricultural land. The approach to study such effects in a real field setting is commendable, but also challenging. At some points the description of used methodology is lacking and presentation of data could be improved. The discussion and interpretation is in order with sufficient care not to jump to conclusions. There is, however, room for further improvement of the manuscript - see details below. Overall, minor revisions.
‘atmospheric mixing ratio of N2O’ sounds rather complicated, perhaps just present change in ppmv?
P2 L35 is this increased and more extreme precipitation expected in summer as well? Not so clear from the current sentence
L50 better not use forms like ‘soil’s’ in scientific writing
P2 L61 Is this proneness of physically protected SOC to release by drying-rewetting really established. Seems quite uncertain still at this stage. Nuance perhaps
P3L69 pH often covaries with clay% as well – mention worthy here as well
P3 A general hypothesis on how the here covered soil factors may mediate the effect of precipitation regime onto N2O emission is missing.
M&M
L93-94 what probes were used (EM50 is just the data-logger model). Seems doubtful that the measuring principle was in fact as stated TDR. Should this not be via capacitance probes instead?
Measurement at just 5cm depth is a weakness of the experimental design as obviously part of N2O is produced at deeper depth, where different moisture conditions likely occur than at 5cm.
L99 provide more detail on the composition of the mineral N fertilizer: NH4+ based?
Table 1 26.3 notation is confusing
L116 emission rates are not directly measured – buildup of N2O was, rephrase and explain in more detail how soil effluxes were derived. In this respect: in L119 it is not clear if fitting a linear model to N2O buildup in the chambers was justified. Why not a curvilinear model?
L121 “spectrometry” be more precise
P11 3.4 Why not use multivariate regression analysis to investigate the effect of
There are a few concerns or unclarities with the used approach:
With respect to measurement of N2O emissions: What were the dimensions of the chambers used for N2O emission detection? Were crops included in the chambers? Was their height progressively increased with the growing crop? Did you only take samples after 30 and 60 min or also at onset to know ambient background concentrations at time 0min?
With respect to the plots: What were the dimensions of the plots: same as the rainout shelters? Was there in any way some separation of soil under these shelters and surrounding soil. Surely with the loamy to clayey texture lateral moisture movement is from wetter surrounding soil into the rainout-shelter covered plots is likely. This seems to be also suggested by L 264-266: “Later in the first year, extreme precipitation events (> 40 mm day-1; Fig. 2E) cancelled out differences in soil moisture between the control and rainout shelter plots (Fig. 2D).” This limitation of the setup should be recognised.
Secondly, it is not clear if capillary rise from groundwater played a significant role in the moisture balance. The variation in groundwater depth along the elevation gradient could be important for the used experimental design. Downhill, groundwater will get closer to the surface. Via capillary action this could result in moderation of the topsoil water content and cancel out part of the rainout shelter effect. On the hill, such capillary rising water input should be less and the effect of rainout shelters may be expected to be stronger. Could this have played a significant role here?
Results & discussion:
The discussion on deconvolution of effects of SOC, pH and clay reads well and makes sense. However, the choice to just look for correlations between single soil variables and N2O emission may appear quite surprising to readers. When dealing with soil predictors that are mutually correlated the reader will wonder why multivariate regression analysis was not employed? Please explain what motivated you to not use a regression model with account of multicollinearity between pH, SOC and clay% (and BD)? If the answer to this question is that simply too little observations were made to fit such models, than alternatively the authors could attempt to at least make the potential effect of combined occurrence of driving conditions for N2O more apparent visually. E.g. to better see if combinations of high soil NO3- and WFPS or temperature coincided or not with high N2O emission rates for either treatment, an alternative presentation, might be useful wherein N2O emission is plotted against two or three drivers. It is the combined effect of drivers that needs to be considered, wherein suboptimal conditions for N2O emission for one predictor may (or may not) be compensated for by the other: e.g. in Conen et al (2020; Global Change Biology, 6, 417-426) it emerges that at low temperature but high WFPS N2O emissions could peak; likewise at relatively low WFPS but relatively high temperature this appeared possible just as well. It could be interesting to use such a presentation approach as well for selected combinations of environmental drivers and/or soil variables with then a clear distinction between both treatments and perhaps using a colour scheme or something of the sort to scale emission rates.
L262-263 “The rainout shelters were installed end of June, about two months after fertilisation, which likely limited mineral N availability.” I fail to see the logic in this statement. Rewrite also considering the following sentence.
With respect to huge precipitation in September 2023, it is relevant to in the discussion also touch upon NO3- leaching. In all plots topsoil NO3- dropped to minimal levels. Could this have been the resultant of N-leaching (which obviously also depends on factors like soil texture and structure)? In principle we may then also expect a substantial difference in N-leaching between control and rain-shelter plots, but with the huge precipitation occurred such effect was not observable here.
Observations covered nearly two years, which fine, but as in any recent field study in NW-Europe, we were faced with strongly deviant weather conditions. This now impedes to generalize outcomes. Perhaps a point to recognize a bit stronger in the discussion.
The discussion could do without L299-305 as this deals with expected trends in N2O emission during a period when emissions were not measured.
L310-311 while true, would be good to briefly repeat how you came to this conclusion.
L318 there was no presentation of POM data and separation of POM was also not described. L318 thus comes out of the blue. Needs to be corrected: either add the POM data from your companion study or refer to it.
The is left wondering why in the M&M mention is made of δ13C measurements as these were apparently not further used in the data analysis.
L337-339 the effect of clay content on N2O emissions is of course not confounded to its effects onto aggregation, soil water content and SOC. Obviously, at increasing clay content, gas exchange itself slows down, regardless of clay effects onto soil water content. Could be made a bit clearer here.
L360 better “N2O/N2 product ratio”
L366 the connection with nitrification is somewhat confusing. Best to just keep this out of your explanation and instead explain that rain exclusion moved soils away from denitrifying conditions rather then towards nitrifying conditions.
L377 better to use another term than ‘when soil moisture is not limiting’: limiting would mostly be interpreted as limiting for vegetation uptake, while here you really want to refer to the situation where in fact there is an abundance of moisture setting the stage for denitrification, then co-controlled by some of the investigated edaphic factors. ‘limiting’ does not entirely well capture this.