Flood Pulse Irrigation of Meadows Shapes Soil Chemical and Microbial Parameters More Than Mineral Fertilization

Round 1

Reviewer 1 Report

Land use and management practices can contribute to changes in SOM quality as well as diversity in the microbial community which should be investigated and monitored to reduce the negative environmental impacts. This will raise awareness of the impact of agricultural practices on the potential accumulation and stability of the SOM, that is susceptibility to mineralization and humification processes favouring C and N loss or accumulation. Therefore, I think that manuscript titled “Flood pulse irrigation of meadows shapes soil chemical and microbial parameters more than mineral fertilization” is very interesting  and covers current trends in soil protection activities.

The subject of this paper fulfils the scope of Soil Systems journal and was properly prepared according to the guidelines for the authors. The individual chapters was correctly organized but some issue need to be corrected or completed.

INTRODUCTION

The introduction section is generally well organized but requires a slight expansion of the subjected topic. Therefore, please describe the impact of the water content on carbon and nitrogen concentration in soil and other related parameters. Moreover, please indicate the optimal water content necessary for the proper and sustainable development of the soil ecosystem  (according to literature data). Are the recommendations for each type of meadow the same?

MATERIALS AND METHODS

Please indicate the total number of soil samples taken from the research area. The depth of sampling and the method of preparation of samples for microbiological and physicochemical analyzes were the same (0-7 cm)? Please explain it.

What types of soil were analyzed and what was the criterion for selecting the research objects?

Figure 1. Please mark individual objects to recognize the differences between them.

RESULTS

Please complete reference to table (l. 174, 194, 196, 199, 207, 210, 216, 218,223,227 and other)

L. 174-177: Why were such significant differences observed between the studied objects?

L.187-189: The flooding cannot change the texture of the soil while this sentence suggests it. Please reworded/rephrased this sentence. Please also discuss the flooding efficiency depending on the type of soil (in discussion section)

Table 1: Please describe the under the table the values in brackets – if it mean standard deviations, should be written with ‘±’ sign.

L.  204-207: What exactly could the adverse effects of hydration and fertilization on bacteria result from? Please explain this in the discussion section.

L. 242-244: Is it related with the competition between microorganisms, or rather with a favorable environment for the selected group of bacteria? Please explain this in the discussion section.

DISCUSSION

L. 266-273: Please discuss the date of fertilization, composition of the fertilizer (forms of components present) and soil types.

In the discussion, please refer to your results more precisely, explaining the potential reasons for your observations? The observations of other authors constitute the background for own results.

Author Response

Dear Reviewer,

we highly appreciate your efforts to improve the quality of our submitted article. Thank you for your revision and find below a point by point answer to your comments.

With kind regards

Kilian Kenngott

Comments and Suggestions for Authors

Land use and management practices can contribute to changes in SOM quality as well as diversity in the microbial community which should be investigated and monitored to reduce the negative environmental impacts. This will raise awareness of the impact of agricultural practices on the potential accumulation and stability of the SOM, that is susceptibility to mineralization and humification processes favouring C and N loss or accumulation. Therefore, I think that manuscript titled “Flood pulse irrigation of meadows shapes soil chemical and microbial parameters more than mineral fertilization” is very interesting  and covers current trends in soil protection activities.

The subject of this paper fulfils the scope of Soil Systems journal and was properly prepared according to the guidelines for the authors. The individual chapters was correctly organized but some issue need to be corrected or completed.

Thank you for this positive evaluation.

INTRODUCTION

The introduction section is generally well organized but requires a slight expansion of the subjected topic. Therefore, please describe the impact of the water content on carbon and nitrogen concentration in soil and other related parameters. Moreover, please indicate the optimal water content necessary for the proper and sustainable development of the soil ecosystem (according to literature data). Are the recommendations for each type of meadow the same?

Thank you for this comment. We agree, that some background could be described in more detail. Therefore we added following information with literature in line 44-53 :

Flooding is generally beneficial for carbon input and may increase soil organic carbon and nitrogen content [3,4]. This can be caused either by an improved plant growth and respectively increased root carbon deposition and litter input [3], or by organic matter accumulation due to reduced degradation efficiency of microbial communities if large soil zones become anaerobic [5]. However, the latter would significantly reduce the hay yield due to oxygen deficiencies for plant roots. Therefore, an optimal water management would keep the water content in a range that plant available water capacity can be maximal exhausted while oxygen supply keeps soil conditions in an oxic state. These optimal conditions also support microbial carbon mineralization which may level out an increased carbon input [5].

MATERIALS AND METHODS

Please indicate the total number of soil samples taken from the research area. The depth of sampling and the method of preparation of samples for microbiological and physicochemical analyzes were the same (0-7 cm)? Please explain it.

We added following information in line 119 to clarify the total number of samples:

Within each plot, seven samples were taken in the centroid and the vertexes of two concentrically oriented equilateral triangles (n = 7, see Figure 1) leading to a total number of 84 analyzed samples.

Furthermore we explain the depth of sampling in lines 130-133:

Top soil was chosen since it is known to be the zone with highest microbial activity, where the most pronounced effects can be expected. All physicochemical and microbiological analyses were obtained from the same topsoil samples to allow a direct comparability of the observed effects.

The preparation of the samples was different between measured parameters and is described in the respective sections. See for example lines 145 and lines 155:

Water extractable nutrients were extracted from 20 g of air dried soil shaken with 100 mL distilled water for 2 h.

Prior to the PLFA analysis, soil samples were freeze-dried to constant weight at -40 °C.

What types of soil were analyzed and what was the criterion for selecting the research objects?

Unfortunately we don’t have information about the soil type but we added available information and reported the soil texture in lines 218-221:

The soils of the investigated meadows are classified as Stagnosols. But due to human activities to adjust the slope for the irrigation system some of them could also be classified as Anthrosols. There was no systematic difference in soil texture between differently treated soils. Most of the soil types were loam and sandy loam, one of the flooded only soils was silty clay.

Accordingly to your comment we included the criteria for sampling site selection in lines 107-114:

After interviews with the farmers, sampling sites were selected with respect to the following criteria: Flooded sites needs to known to be flooded at least since 50 years and receive water three to five times per year. Fertilization should be known to be only inorganic during the last 50 years and in a similar intensity (40-75 kg N). The non-flooded sites should not have been flooded artificially for the last 50 years [24]. For this study on all sites Holcus lanatus L. (Poaceae) should be present in a sufficient amount to samples only under this species. By these criteria 12 sampling sites were selected according to flooding (yes/no), fertilization (yes/no, 40 to 52 kg ha^-1a^-1).

Figure 1. Please mark individual objects to recognize the differences between them.

Thank you for this recommendation. We have changed figure 1 and it’s caption accordingly:

...Symbols show the positions of meadows that were flooded (blue), non-flooded (red), fertilized (circles) and non-fertilized (triangles). ...

RESULTS

Please complete reference to table (l. 174, 194, 196, 199, 207, 210, 216, 218,223,227 and other)

Please excuse the missing references, this must have happened during conversion from .odt to .docx. We have updated all references accordingly.

L. 174-177: Why were such significant differences observed between the studied objects?

The greatest differences are observed within the flooded x fertilized treatment, with one plot showing clearly higher values. We assume these higher values are caused by the silt and clay content which are higher compared to the other flooded x fertilized soils (but not compared to the overall textures). However, we found no clear reason for this variability.

L.187-189: The flooding cannot change the texture of the soil while this sentence suggests it. Please reworded/rephrased this sentence. Please also discuss the flooding efficiency depending on the type of soil (in discussion section)

This is a misunderstanding. We do not want to imply that flooding changes the soil texture, but emphasis, that the texture is probably not responsible for the higher water content as discussed later in line 205:

However, the significantly higher water content together with a trend of increased C_org content suggests that, in the long-term, amount and quality of organic matter improved the water holding capacity in the flood-irrigated meadows.

We reworded the sentence as follows in lines 221-224:

There was no systematic difference in soil texture between differently treated soils. Most of the soil types were loam and sandy loam, one of the flooded only soils was silty clay. Since there was no systematic difference in soil texture between differently treated soils, the higher water content on the flooded soils is not caused by a higher WHC of the mineral part of the soil.

Table 1: Please describe the under the table the values in brackets – if it mean standard deviations, should be written with ‘±’ sign.

We agree that SD can be written as ± and we changed the table accordingly.

L.  204-207: What exactly could the adverse effects of hydration and fertilization on bacteria result from? Please explain this in the discussion section.

We agree that this can be pointed out clearer, we therefore changed lines 373 and following accordingly:

Additionally, copiotrophic bacteria such as Proteobacteria have been shown to be sensitive to drought stress, which is more likely to occur on non-flooded sites. [47]. Furthermore, Gram-negative bacteria are often associated with aerobic growth [44]. Bossio and Scow [41] as well as Drenovsky et al. [4] reported decreasing Gram-negative PLFA content with increasing soil water content which seems contradictory to our results. But in contrast to these rice field studies, the flood pulse irrigation method of the present study avoids stagnating water and anoxic conditions [1]. Anoxic conditions due to saturated soil and stagnating water can negatively affect Gram-negtive bacteria [8,44]. Our results, with slightly higher Gram-negative proportions indicate, that the flood pulse irrigation method avoids anoxic conditions as reported in [1] or promotes a resilient Gram-negative bacterial community that recovers within 8 to 10 weeks [6].

We expected a more pronounced reduction of pH in fertilized soils and a trend towards lower bacteria proportions on the expanse of fungi, as reported in literature. However, our results showed the highest and lowest pH in fertilized only soils and other chemical parameters seemed to be unaffected. Therefore we added the following explanation in lines 313 and 381-385:

The long-term input of ammonium with NPK fertilization is known to reduce soil pH, with potentially adverse effects on bacterial soil community [14]. Contrary, the application...

Although it has been shown that bacteria can be negatively affected by fertilization followed by reduced soil pH [14], we did not find such correlation. However, the effect of mineral fertilization on soil pH may also depend on composition and amount of applied fertilizer [51]. Therefore, the extensive management and the previously mentioned variation in soil pH observed in this study can explain why the relationship of soil pH and bacteria were not observed here.

L. 242-244: Is it related with the competition between microorganisms, or rather with a favorable environment for the selected group of bacteria? Please explain this in the discussion section.

We attribute this shift to favorable conditions for the respective microbial groups although a competition can not be excluded. We discuss the favorable conditions and the effect on microbial shifts in lines 393 and following:

The shift towards lower fungi/bacteria ratios in flood irrigated soils was caused by both, a smaller fungal and a larger bacterial community. We observed the trend that fungal abundance decreases with increasing soil water content. However, the effect of flooding on Gram-negative bacteria is significant and in absolute values higher than the reducing trend on fungi, which indicates that this community shift is stronger driven by changes in the bacterial than by the fungal community. Probably, the higher carbon content in flood pulse irrigated soil had a positive effect on both, bacteria and fungi, which mitigated the negative effects of higher water content on fungi. Nonetheless, our findings support hypothesis II, that changes in soil chemical parameters induced by agricultural practice, i.e. flood pulse irrigation, are reflected in SMC shifts, i.e. reduced fungi/bacteria ratio.

DISCUSSION

L. 266-273: Please discuss the date of fertilization, composition of the fertilizer (forms of components present) and soil types.

The exact date of fertilization is not known. Therefore we mention the season for fertilization and possible impacts on our results in lines 307-310:

Furthermore, the fertilizer is usually applied once per year in spring or early summer. Therefore, the remaining effects on soil nutrient contents (N, P, K) in autumn may be small, especially compared to the more recent flooding events.

Since we found no clear differences in soil nutrient content (N, P, K), we explain the reason for absence of differences in lines 301 and following:

The lack of an overall fertilization effect on soil chemical parameters can be best explained with the extensive agricultural practice of the meadows. With 40 to 52 kg ha^-1 a^-1, the application rate of mineral fertilizer on the investigated meadows is low compared to over 120 kg ha^-1 a^-1 of fertilizer applied on conventionally managed grassland cut twice a year [38].

However, we found indication that soil pH may be changed by the composition of fertilizer and discuss this in lines 312 and following:

The long-term input of ammonium with NPK fertilization is known to reduce soil pH, with potentially adverse effects on bacterial soil community [14]. Contrary, the application of the carbonate containing CAN fertilizers [43] buffers soil acidification. However, due to small replication size, this contrasting effect can only be considered as a trend within this study and a mitigation by flood pulse irrigation of altered soil pH in fertilized soil remains unclear.

In the discussion, please refer to your results more precisely, explaining the potential reasons for your observations? The observations of other authors constitute the background for own results.

We thoroughly revised the Discussion section. We referred to our results and compared with results of other authors and explained potential reasons for our observations e.g., in:

line 297:

In contrast to our expectations and to findings of some studies [10,34,35], we found no indication that mineral fertilizer application is associated with increased C_org, N_tot and water extractable NO₃^- content. However, other studies also found the opposite effects, e. g. accelerated humus mineralization after long-term application of mineral NPK fertilizers [36] or only a minor influence of mineral fertilization on C_org [37]. The lack of an overall fertilization effect on soil chemical parameters can be best explained with the extensive agricultural practice of the meadows. With 40 to 52 kg ha^-1 a^-1, the application rate of mineral fertilizer on the investigated meadows is low compared to over 120 kg ha^-1 a^-1 of fertilizer applied on conventionally managed grassland cut twice a year [38].

line 310:

However, we found indication that the types of fertilizers, i.e. NPK and CAN, have contrasting effects on soil pH if soils are not flooded, which can explain the high variation. The long-term input of ammonium with NPK fertilization is known to reduce soil pH [10], while the application of the carbonate containing CAN fertilizers [39] buffers soil acidification. However, due to small replication size, this contrasting effect can only be considered as a trend within this study and a mitigation by flood pulse irrigation of altered soil pH in fertilized soil remains unclear.

Line 333:

Total microbial biomass seemed to respond positively to the increased C_org contents in flood pulse irrigated soils. Zak et al. [40] linked higher plant productivity to increased C_org and consequently higher microbial biomass. This is in line with our findings and supports the idea, that flood pulse irrigation has a positive effect on soil microbial biomass due to increased plant productivity followed by increased C_org. However, causal relations are still not fully understood. An explanation might be a tendentiously elevated plant species richness on the flooded meadows [21].

line 345:

Beside plant diversity, also water availability for plants is improved by flood pulse irrigation [1] which is another important factor supporting microbial biomass [8,42]. The increased C_org on the flood-irrigated meadows additionally may have improved soil water retention presenting a positive feedback on water supply for both, plants and microbial biomass in a long term. However, a higher carbon loss due to prolonged favorable conditions for microbial activities did probably not compensate for the increased carbon input by enhanced plant growth.

Etc.

Reviewer 2 Report

The submitted manuscript entitled “Flood pulse irrigation of meadows shapes soil chemical and microbial parameters more than mineral fertilization” presents interesting results of the comparative study of flood pulse irrigation and mineral fertilization and it is within the scope of the “Soil Systems”. Introduction very well defined background of the study. Materials and methods are well described. Results and discussion are also well presented.

However, some minor corrections are required. 

#1 Line 47 the first time used abbreviation SMC should be explained

#2 Line 99 the mentioned first time Holcus lanatus should be follow by (L.)

#3 Lines 190 - 191 The description of Table 1 is slightly confusing because is not directly related to the columns' headings.

#4 Line 249 Figure 5 - is "Correlation of ... " but it is a linear trend line.

#5 Figure 6. The authors showed a trend line as a linear relationship. The value of R² and formula describing dependencies should be added to the figures. However, based on the distribution of points I have a feeling that maybe the logarithmic dependence with  0 < ∝ < 1 will better express the relationships between the Fungi/Bacteria ratio (Fig. 5b), Fungi mol% (Fig. 5c), and water content as well as the logarithmic dependence with ∝ > 1  will better express the relationships between the Bacteria mol% Fig. 5d) and water content. suggest for authors to try the analysis of the logarithmic dependence and compare the R2 values of linear and logarithmic dependencies to choose the better one. Maybe it will more fruitful for the discussion of results. 

Author Response

Dear Reviewer,

we highly appreciate your efforts to improve the quality of our submitted article. Thank you for your revision and find below a point by point answer to your comments.

With kind regards

Kilian Kenngott

Comments and Suggestions for Authors

The submitted manuscript entitled “Flood pulse irrigation of meadows shapes soil chemical and microbial parameters more than mineral fertilization” presents interesting results of the comparative study of flood pulse irrigation and mineral fertilization and it is within the scope of the “Soil Systems”. Introduction very well defined background of the study. Materials and methods are well described. Results and discussion are also well presented.

However, some minor corrections are required. 

#1 Line 47 the first time used abbreviation SMC should be explained

Thank you for this comment. We agree and now introduce, consistently wit PLFA, the SMC in the abstract. We changed following lines 19 and 51:

In this study we assessed shifts in soil microbial communities (SMC) as a response to changes in soil chemical parameters after flood pulse irrigation and/or fertilization of meadows.

Thus, also negative effects of flood pulse irrigation on certain groups within the soil microbial community (SMC) are possible.

#2 Line 99 the mentioned first time Holcus lanatus should be follow by (L.)

We agree and changed the line 112 accordingly:

For this study on all sites Holcus lanatus L. (Poaceae) should be present

#3 Lines 190 - 191 The description of Table 1 is slightly confusing because is not directly related to the columns' headings.

Thank you very much for this comment. This is indeed a residue of an older terminology. We have changed the table description accordingly:

Soil chemical parameters and soil texture of meadows that were neither flooded nor fertilized (control), flooded only, fertilized only and flooded × fertilized shown as means and standard deviation.

#4 Line 249 Figure 5 - is "Correlation of ... " but it is a linear trend line.

We agree and changed the figure caption as the reviewer suggests:

Figure 5. CorrelationLinear trend line of arbuscular mycorrhizal fungi (AMF)...

#5 Figure 6. The authors showed a trend line as a linear relationship. The value of R² and formula describing dependencies should be added to the figures. However, based on the distribution of points I have a feeling that maybe the logarithmic dependence with  0 < ∝ < 1 will better express the relationships between the Fungi/Bacteria ratio (Fig. 5b), Fungi mol% (Fig. 5c), and water content as well as the logarithmic dependence with ∝ > 1  will better express the relationships between the Bacteria mol% Fig. 5d) and water content. suggest for authors to try the analysis of the logarithmic dependence and compare the R2 values of linear and logarithmic dependencies to choose the better one. Maybe it will more fruitful for the discussion of results. 

Thank you for this very interesting suggestion. We have calculated the logarithmic dependencies and compared the old R² values of linear models (0.459 Fungi/Bacteria, 0.325 Fungi alone, 0.23 Bacteria alone) with R² values of log models (0.498 Fungi/Bacteria, 0.35 Fungi alone, 0.23 Bacteria alone). As you can see, R² values were not significantly improved and still very low. Therefore we prefer the more “parsimonious” simple linear model.

We added trend lines to visually show the trend that we see in the data but without clear statistical significance. Since the models residuals do not meet the assumptions, we do not want to suggest statistical validity to the reader by adding formula and R².

Reviewer 3 Report

GENERAL COMMENTS

Kenngott and co-workers report a study on changes in soil chemical parameters and microbial communities after flood pulse irrigation and/or fertilization of meadows, whose aim is providing knowledge that will lead to an optimized and more sustainable agricultural land use management with reduced fertilizer application. While the topic is relevant and within the scope of Soil Systems, this study has, in my opinion, two major drawbacks: firstly, it suffers from a weak experimental design; and secondly, it lacks important, complementary data.

There were actually two different types of fertilization (NPK or CAN), being however regarded in statistical analyses as a single type. Different fertilization leads to different effects on soil chemistry, some of them can be detected (e.g. pH) but others perhaps not (e.g. differential levels of some nutrients, but below the limit of quantification). Subtle changes in soil nutrient levels may have produced changes in soil microbial communities, which can keep masked (hidden) when averaging results from both types of fertilization. Additionally, there is no indication on how these two kinds of fertilization are distributed across the two flooding regimes. Did the three plots with the same fertilization have the same regime of flooding? If yes, each fertilization type was thus associated with a different regime of flooding; if no, fertilization types are not evenly distributed across the two flooding regimes. Moreover, the authors seem to have performed ANOVA analyses using a sample size of 21 replicates per treatment (7 samples * 3 plots) instead of 3 replicates (3 independent plots) per treatment, which introduces in my opinion an issue of pseudoreplication.

Secondly, neither data on plant productivity (biomass production) nor on plant diversity were obtained in this study, which are both relevant to contextualize the results on soil microbiology more solidly. Due to this lacking information, most of the discussion is speculative and only based on literature. So, this study has some taste of incomplete work as implicitly acknowledged by the authors in the last sentence of the manuscript (lines 383-386).

SPECIFIC COMMENTS

Lines 27-28. The authors refer to water-holding capacity (AWC), but actually they measured soil water contents. Although AWC has relationships with properties like texture and organic carbon content, it seems an overstatement to assert that flood pulse irrigation promotes higher water holding capacity.

The authors should state the number of replicates (n) per treatment (agricultural practices) in Table 1 and Figures 2, 3 and 4. There would arise an issue of pseudoreplication if 21 replicates per treatment (7 samples * 3 plots) were used.

Lines 169-189. The only parameter showing differences statistically significant was soil water content. Therefore, no solid conclusions can be drawn with regard the effects produced by agricultural practices on soil chemical properties. Moreover, the situation could turn worse due to pseudoreplication if the ANOVA analyses were performed using 21 instead of 3 replicates (plots) per treatment.

Line 205 and onwards. "Gram-negative bacteria were significantly increased by flooding to 61 ± 3 mol% compared to…” There is a risk that some readers may confuse relative abundances of microbial groups (%mol) with total abundances. So, it should be always stated what kind of abundance it is (e.g. "Relative abundance of Gram-negative bacteria were significantly…”). 

Line 193 and onwards. Because of different fatty acids have different molecular weight, total microbial biomass should be expressed in molar units per gram of soil (e.g. nmol/g).

Author Response

Dear Reviewer,

we highly appreciate your efforts to improve the quality of our submitted article. Thank you for your revision and find below a point by point answer to your comments.

With kind regards

Kilian Kenngott

Comments and Suggestions for Authors

GENERAL COMMENTS

Kenngott and co-workers report a study on changes in soil chemical parameters and microbial communities after flood pulse irrigation and/or fertilization of meadows, whose aim is providing knowledge that will lead to an optimized and more sustainable agricultural land use management with reduced fertilizer application. While the topic is relevant and within the scope of Soil Systems, this study has, in my opinion, two major drawbacks: firstly, it suffers from a weak experimental design; and secondly, it lacks important, complementary data.

There were actually two different types of fertilization (NPK or CAN), being however regarded in statistical analyses as a single type. Different fertilization leads to different effects on soil chemistry, some of them can be detected (e.g. pH) but others perhaps not (e.g. differential levels of some nutrients, but below the limit of quantification). Subtle changes in soil nutrient levels may have produced changes in soil microbial communities, which can keep masked (hidden) when averaging results from both types of fertilization. Additionally, there is no indication on how these two kinds of fertilization are distributed across the two flooding regimes. Did the three plots with the same fertilization have the same regime of flooding? If yes, each fertilization type was thus associated with a different regime of flooding; if no, fertilization types are not evenly distributed across the two flooding regimes. Moreover, the authors seem to have performed ANOVA analyses using a sample size of 21 replicates per treatment (7 samples * 3 plots) instead of 3 replicates (3 independent plots) per treatment, which introduces in my opinion an issue of pseudoreplication.

Thank you very much, that you have red the manuscript so carefully. We could not decide the type of fertilizer applied on the fields, thus we focused on what is comparable between the fertilization regimes, namely the introduction of mineral nitrogen. Since this study is part of a bigger project, there were even more fields sampled and CAN and NPK was equally distributed. Unfortunately there was no Holcus lanatus on some of them which consequently had to be excluded. This is why we ended up with the distribution 2 * NPK and 1 * CAN on flooded × fertilized plots, 2 * CAN and 1 * NPK on flooded only plots. We agree that an investigation of fertilizing type effects would require a balanced design. We describe our design more precise now in lines 100-102:

This creates a mosaic of agricultural practice combinations comprising both agricultural practices, i.e. flood pulse irrigation and fertilization (see Figure 1). CAN was applied on one flooded and two non-flooded meadows, NPK was applied on one non-flooded and two flooded meadows. The practice combinations...

We agree, that the fertilization type is not evenly distributed across the flooding regimes. However, our hypotheses were developed beforehands and focused on the introduction of mineral fertilizer, in particular the introduction of mineral nitrogen. Thus we excluded the type of fertilizer from the statistics. However, it appeared to have an effect that can not be simply ignored and we included the type of fertilizer to explain variations that were observed. We are very careful with this inclusion as for example interpretation of differences in the soil pH and only speak of a possible explanation, see lines 301 - 307:

However, we found indication that the types of fertilizers, i.e. NPK and CAN, have contrasting effects on soil pH if soils are not flooded, which can explain the high variation. The longterm input of ammonium with NPK fertilization is known to reduce soil pH [6], while the application of the carbonate containing CAN fertilizers [36] buffers soil acidification. However, due to small replication size, this contrasting effect can only be considered as a trend within this study and a mitigation by flood pulse irrigation of altered soil pH in fertilized soil remains unclear.

We are also aware of the semblance of pseudoreplication and we agree, that parameters like pH or soil organic carbon may have this issue. However, we think that most chemical parameters and particularly the soil microbial community has a small scale heterogeneity, which gives the samples taken from the same plot some statistical independence. Furthermore, we did not ignore this in statistics but performed the chi-square ANOVA using linear mixed effect models including “Plot” as a random effect, and the F-ANOVA including “Plot” with an Error-term. This has similarly been performed in other studies (Vries et al. 2012, doi: 10.1111/j.1461-0248.2012.01844.x; Schirmel et al. 2018, https://doi.org/10.1016/j.apsoil.2017.11.025 ). Furthermore we modified lines 191 - 195 as follows:

Two-way analysis of variance (F-ANOVA) was used to test for significant effects and interaction of flooding and fertilizing including sampling plots as random factor. All tested models included the factor “Plot” as random effect to account for multiple samples per plot. Normality and homoscedasticity of residuals were evaluated with Q-Q and residuals vs. fitted plots. A χ²-ANOVA on ranks was applied to variables that did not meet parametric assumptions using linear mixed-effect models in the package lme4 [34].

Secondly, neither data on plant productivity (biomass production) nor on plant diversity were obtained in this study, which are both relevant to contextualize the results on soil microbiology more solidly. Due to this lacking information, most of the discussion is speculative and only based on literature. So, this study has some taste of incomplete work as implicitly acknowledged by the authors in the last sentence of the manuscript (lines 383-386).

We agree that a discussion together with results of plant productivity and diversity will complete the whole research question. The data were obtained and will be published soon in a separate article. We prefer to separately publish the above and below ground data because we believe that the complexity of these data could not be adequately reflected in just one article. Unfortunately, conclusions from the plant part of the project are still not available, however, the conclusions of this study will then appear in the planned plant paper.

SPECIFIC COMMENTS

Lines 27-28. The authors refer to water-holding capacity (AWC), but actually they measured soil water contents. Although AWC has relationships with properties like texture and organic carbon content, it seems an overstatement to assert that flood pulse irrigation promotes higher water holding capacity.

Thank you for this comment. We agree that this is an overstatement and changed it accordingly, lines 27 - 28:

Flood pulse irrigation seemed to promotes the build-up of a larger soil carbon and nitrogen pool as well as higher water holding capacity content and microbial biomass.

We also changed line 348 accordingly:

The increased C_org on the flood-irrigated meadows additionally may have improved soil water retention presenting a positive feedback on water supply for both, plants and microbial biomass in a long term.

The authors should state the number of replicates (n) per treatment (agricultural practices) in Table 1 and Figures 2, 3 and 4. There would arise an issue of pseudoreplication if 21 replicates per treatment (7 samples * 3 plots) were used.

We agree that the way of statistical analysis can be pointed out clearer. We therefore added following sentence to Table 1 and Figures 2, 3 and 4:

Analysis was done using n = 7 samples * 12 plots, including “Plot” as a random effect.

Lines 169-189. The only parameter showing differences statistically significant was soil water content. Therefore, no solid conclusions can be drawn with regard the effects produced by agricultural practices on soil chemical properties. Moreover, the situation could turn worse due to pseudoreplication if the ANOVA analyses were performed using 21 instead of 3 replicates (plots) per treatment.

We agree that our conclusions with respect to the soil chemical parameters have to be stated more carefully. We therefore always use formulations like “appeared to” (line 191), “general trend” (line 200) or “seemed to” (line 202) to describe our observations. The lack of statistical evidence is also clearly stated in lines 203 - 208:

Yet, no significant differences in soil chemical parameters between agricultural practices were observed due to the large variability between the plots within the same agricultural practice. However, the significantly higher water content together with a trend of increased C_org content suggests that, in the long-term, amount and quality of organic matter improved the water holding capacity in the flood-irrigated meadows.

We also accordingly changed the formulations in the conclusion section, line 423 and following:

Our results suggest that flood pulse irrigation in extensively used hay meadows has a positive effect on soil fertility. The tendency of an enhanced build-up of soil organic carbon and nitrogen pool and significantly higher water contents are accompanied by significant effects on the microbial community composition upon flood pulse irrigation. Beside an increased soil microbial biomass, also a greater fraction of rather copiotrophic Gram-negative bacteria within the soil microbial community characterizes the flooded compared to non flooded soils.

…

The suggested positive effects of flooding can be explained as a ...

Line 205 and onwards. "Gram-negative bacteria were significantly increased by flooding to 61 ± 3 mol% compared to…” There is a risk that some readers may confuse relative abundances of microbial groups (%mol) with total abundances. So, it should be always stated what kind of abundance it is (e.g. "Relative abundance of Gram-negative bacteria were significantly…”). 

Thank you for this comment. We agree that formulations like “fungal PLFA content” may be confusing. We therefore further on speak of “proportion” rather than “content” and changed the text accordingly. See for example lines 243 and following:

The proportions of Gram-negative bacteria were significantly increased by flooding to 61 ± 3 mol% compared to the non-flooded counterparts with 58 ± 4 mol% (p = 0.032, χ²-ANOVA; Figure 3A). The proportion of Gram-positive bacterial PLFA content seemed to be lower in flooded only soils

Line 193 and onwards. Because of different fatty acids have different molecular weight, total microbial biomass should be expressed in molar units per gram of soil (e.g. nmol/g).

Thank you for this comment. We changed statistics, Figure 1 and text accordingly. See for example lines 230 - 235:

Total PLFA contents were significantly higher in flooded soils (8 ± 3 nmol g^-1) compared to non-flooded soils (5 ± 1 nmol g^-1; Figure 2A). Contrary, there was a trend of reduced total PLFA content when soils were fertilized (6 ± 3 nmol g^-1) compared to non-fertilized soils with 7 ± 2 nmol g^-1 (Figure 2A). The positive effect of flooding on total PLFA content was not reflected in the relative PLFA content, i.e. when the PLFA content was related to C_org (PLFA/C_org, Figure 2B).

Round 2

Reviewer 1 Report

Dear Authors,

I propose to introduce additional minor corrections to the manuscript:

Introduction:

The abbreviations that appear for the first time should be explained e.g. SMCs in L. 53

Please carefully check the repetitions in the text (L. 71)

Materials and methods:

Please check the applied unit (L. 98)

In the figure 1 please mark the city / place (reference for the location of the experimental place) and the compass rose in in the upper right corner

Author Response

Dear Reviewer,

thank you again for your helpful remarks. Please find below our responses.

With kind regards

Kilian Kenngott

Introduction:

The abbreviations that appear for the first time should be explained e.g. SMCs in L. 53

Thank you for this comment. SMC is now defined in line 53 accordingly:

It is well known, that soil microbial communities (SMC)s are sensitive to soil water regime [7–11].

Please carefully check the repetitions in the text (L. 71)

We have changed the line as follows:

The response of the SMC structure to drying and re-wetting cycles depends on the duration and frequency of inundation and the inundation frequency [19–21][19–21].

Materials and methods:

Please check the applied unit (L. 98)

Thank you for this comment. The unit in-deed can lead to misunderstandings since it can be read as kilogram fertilizer per hectare and year. We changed the line accordingly:

fertilizer applications ranging from 40 to 52 kg N ha^-1a^-1 [25].

In the figure 1 please mark the city / place (reference for the location of the experimental place) and the compass rose in in the upper right corner

We agree and changed the map accordingly.

Reviewer 3 Report

I regret to say that, in my opinion, the revised manuscript does still not meet the level of scientific soundness that it would be expected for publication. Two different types of fertilization, and moreover, in an unbalanced experimental design may create confounding variation and may lead to misinterpretations. Being very careful when interpreting some noticed result (e.g., pH) does not rule out other potential (unnoticed) effects that could have drove the soil microbial communities. On the other hand, I doubt that in this case pseudoreplication disappears by using “Plot” as random effect. Nonetheless, I am not enough skilled in Statistics to definitely refute this authors' approach; it would be interesting hear from a statistician. Finally, not providing data on aboveground production and plant diversity leaves this manuscript incomplete and too speculative.

Author Response

Dear Reviewer,

thank you again for your helpful remarks. Please find below our responses.

With kind regards

Kilian Kenngott

I regret to say that, in my opinion, the revised manuscript does still not meet the level of scientific soundness that it would be expected for publication. Two different types of fertilization, and moreover, in an unbalanced experimental design may create confounding variation and may lead to misinterpretations.

We agree that unnoticed effects may be hidden behind the fertilizer type effects. However with the current experimental design we cannot detect them because, as you mentioned, the design is not balanced for these effects. However, this was also not the aim of this study. The aim was to test whether soil microbial communities (SMC) responses to changes in soil chemical parameters after flood pulse irrigation and/or fertilization of meadows and not which effects different fertilizer types would have. But this is also one outcome of this study, that the fertilizer types might have an effect and should thus be replicated in the next experiment. However, we also agree that the attention of the reader should be drawn to this issue. Thus we added the following sentences to the manuscript:

Line 320-322: However, this effect cannot be judged by the current experimental design because it is not balanced for the type of fertilizer.

Line 427-431: However, the current experimental design was not aimed at and is thus not appropriate for a test of the effect of different fertilizer types. Therefore, hidden effects of the fertilizers with respect to their type cannot be excluded and a future study balanced for the fertilizer types might shed more light on this question.

Line 439-443: In contrast, extensive mineral fertilization seems not to improve the nutrient status of soils in a long-term experiment. But our results also suggest that additional investigations with respect to different fertilizer types are required before an effect of mineral fertilization can be finally judged.

Being very careful when interpreting some noticed result (e.g., pH) does not rule out other potential (unnoticed) effects that could have drove the soil microbial communities.”

We agree that unrecorded co-variates may have influenced our results. This is an inherent limitation of any field study. Recording and statistically evaluating every potential co-variate possible, however, will not solve this issue since a high number of co-variates will inflate the likelihood of mistakenly identifying effects as significant although they were not (Type I error). We tried to counteract this with a rigorous hypothesis- and process-driven approach. This involved better understanding I) how mineral fertilization and flood irrigation change soil physicochemical parameters and II) how this eventually propagates to changes in the soil microbial community. For this reason, we decided not to include a high number of other variables into one big model but let the statistics reflect our initial hypotheses as follows: “physicochemical parameters depends on treatment” and “SMC depends on physicochemical parameters” (matching with the chapters/story).

“On the other hand, I doubt that in this case pseudoreplication disappears by using “Plot” as random effect. Nonetheless, I am not enough skilled in Statistics to definitely refute this authors' approach; it would be interesting hear from a statistician.”

Under the following link it was shown that the pseudoreplication disappears using the random effect “plot”: (https://www.middleprofessor.com/files/applied-biostatistics_bookdown/_book/models-with-random-effects-blocking-and-pseudoreplication.html#pseudoreplication)

We totally agree to the statement on the site that “subsamples from batches are not replicates. Inference from a model fit to subsampled observations without modeling the batches is called pseudoreplication...”. However, they also provide an example calculation for the analysis of only mean values of batches (in our case plots) and linear mixed model analysis. They state:

“One interesting result to note is the equivalence of the standard error, test statistic, p-value, and confidence intervals of the linear model on the batch means and the linear mixed model with a random intercept (but no random slope) only. The two are equivalent.”

Including batches (or plots) as random effect in linear mixed effect models is a valid way to account for subsamples.

“Finally, not providing data on aboveground production and plant diversity leaves this manuscript incomplete and too speculative.”

We agree that these data would be very helpful to interpret the overall processes. Data about the aboveground plant productivity and diversity were collected in other sub-projects of the research framework initiative “Aufland” and the latter are already published elsewhere [24,25] and referred to in this script in

Line 344: An explanation might be a tendentiously elevated plant species richness on the flooded meadows [24].

Line 366-368: Beside a reduced C_org, a possible explanation for a reduced microbial biomass due to fertilization may be a reduced plant diversity on the fertilized plots [24,45].

Line 417-418: One explanation of the apparently robust AMF community could be the higher plant diversity, which was shown by Müller et al. [24] on the same plots.

24. Müller, I.B.; Buhk, C.; Alt, M.; Entling, M.H.; Schirmel, J. Plant Functional Shifts in Central European Grassland under Traditional Flood Irrigation. Appl. Veg. Sci. 2016, 19, 122–131, doi:10.1111/avsc.12203.

25. Müller, I.B.; Buhk, C.; Lange, D.; Entling, M.H.; Schirmel, J. Contrasting Effects of Irrigation and Fertilization on Plant Diversity in Hay Meadows. Basic Appl. Ecol. 2016, 17, 576–585, doi:10.1016/j.baae.2016.04.008.

Data about plant productivity belong to another working group and we have no access to these data.