Silicon-Mediated Physiological and Agronomic Responses of Maize to Drought Stress Imposed at the Vegetative and Reproductive Stages

Round 1

Reviewer 1 Report

The authors followed the previous recommendations.

Reviewer 2 Report

I have no further comments to this revised version, which address satisfactorily all my concerns in the previous revision.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Dear authors,

The manuscript Ning et al., present a relevant topic in agronomy. Many studies have been carray out in recent years related to protective effect of Si in crop production, but this manuscript showed a life cycle prepective that is few studied. 

I've some notes about the data presentation:

1) Why do you replicated 15 times for control and 9 times for treatments? These replications are the n of plants or temporal replications??

2) Why some graphics present the standard deviation bar completed and other present the upper side?

3) I think that you may have a mistake in significat character in the figure 4. The condition CK+Si in R2 for caratenoid contetnt has a\ "b", when the SD bar is so high. Please check it.

Author Response

Reviewer: 1

Dear authors,

The manuscript Ning et al., present a relevant topic in agronomy. Many studies have been carry out in recent years related to protective effect of Si in crop production, but this manuscript showed a life cycle perspective that is few studied.

I've some notes about the data presentation:

1. Why do you replicated 15 times for control and 9 times for treatments? These replications are the n of plants or temporal replications??

• Response: This is because we destructively collected plant samples at each drought stress duration from control treatments, and 15 is number of control pots, but not the sample sizes of plants.
•  

1. Why some graphics present the standard deviation bar completed and other present the upper side?

• Response: Thank you for your suggestion. We make the SE bars consistent for all figures.
• 3) I think that you may have a mistake in significant character in the figure 4. The condition CK+Si in R2 for caratenoid contetnt has a\ "b", when the SD bar is so high. Please check it.

• Response: Sorry for the mistake. The letters have been corrected accordingly.

Reviewer 2 Report

This is a well-written manuscript reporting the results of a study on the effects of Si application to alleviate drought stress in maize plants. The physiological changes underpinning these effects are also analysed. The authors found that Si might alleviate oxidative stress by enhancing the production of antioxidant enzymes, which seems to maintain higher stomatal conductance and photosynthesis. Unfortunately, the manuscript has some important flaws and inconsistent results that need to be revised by the authors.

1. The most important of these flaws is poor reporting of the details of the ANOVA (factors, levels, procedure to assess the meaning of interactions) and their results. The description of the ANOVA needs to be detailed in the statistical analyses section of Material and Methods. From the only report of these analyses in the manuscript (table3 and 4, analysis of yield and Si concentration, it seems that the authors performed a two-way ANOVA. I have many difficulties to understand how the authors applied such analyses given the experimental design. If I am not mistaken (following the description in material and methods), the authors initially distributed the plants into two Si application levels. Then, at three different crop stages, plants were further divided into two irrigation treatments. This can be described as a 2 x 3 x 2 factorial design that needs to be analysed by a three-way analysis of variance. In table 3 and 4, the authors list two of these main factors (Si and drought), and then the Si x Stage interaction. What about the stage effect and the Si x Drought and the Drought x Stage interactions? I can think of thre possibilities: either the authors ignored reporting the other factors and interactions or they customise the general lineal model excluding them. Alternatively, the authors could wrongly identified the factors and the analyses should be redone in an appropriate manner. In any case, it is required that the authors explain the analyses (factors, levels) in detail and they should report the result of the ANOVA for the rest of the studied parameters, not just yield and Si concentration, either in the text or preferably in tables. In the present state, results are referenced to the figures, which can only offer a visual representation aided by the homogeneous group separation of means. However, Tukey test cannot be used as the exclusive reference for difference between levels of factors, the significance of factors and interactions in the ANOVA must be shown and interpreted accordingly. Moreover, the group separation procedure should be further explained in the statistical analysis section. Looking at the figures, it seems that the authors performed the Tukey test to compare the Drought x Si combinations within the same stage. If that is the case (it should be explained), this needs to be preceded by the assessment of the Drought x Si interaction and the Drought x Si x Stage interaction. Alternatively, given the complexity of three-way designs, and for the sake of clarity, I believe it would be admissible to perform independent two-way analyses (Drought x Si) to each stage (statisticians would disagree with me).

Another point of concern, related to the reporting of statistics, stems from the lack of reporting in figures caption. Bar figures should state the sample size (n) for bars and the type of error bar used (standard error, standard deviation). I am assuming that bars are standard error as usual (please change to s.e. otherwise). If this is the case, I found quite hard to believe some of the separation in homogeneous groups shown in the figures. There are many examples, but perhaps the most striking is the differences in carotenoids at R2 stage. Given the size of the error bar for D-Si, it is impossible that this treatment is significantly lower than the rest (even if the bar would be standard deviation). Numerous differences between bars are so small in comparison to the errors that it is unlikely they are statistically significant if the errors are reported as S.e. and they are correctly depicted. Please clarify and amend if necessary.

2. Experiments studying the effects of soil drying on plant physiology, must always report any sort of measure of either soil or plant water status, preferably both. The same way authors did measure Si concentration in leaves to assess the effectiveness of the application treatments and the interaction of other factors on Si uptake, the same should be done with water. I assume that the authors did not measure leaf water potential or water content (in my opinion, it is a shame that such an impressive array of measurements missed this very essential parameter that defines water status of the plant). However, it is very likely that they recorded pot weight and can retrospectively calculate, even approximately (estimating the weight of the maize plants) soil water content. This is an essential variable to report, as the interpretation of the authors in the discussion implicitly rely on the uniformity of water content within each irrigation treatment (across stages and Si treatments). However, if these two factors change transpiration rates or leaf area, the intensity of the drought could have been greater in those bigger more transpiring plant. This is unlikely for Si treatments, but it is most probable for stages. Thus, the authors attributed the smaller effect of drought in V6 compared to R2 to a lower sensitivity to drought in that stage, but it is most likely that the smaller size of the plant explain lower soil water depletion rates and lower intensity of stress over the 7 days period of stress application. The irrigation procedure to maintain soil water content at a constant level does not guarantee homogeneity of stress levels across treatments, which depends also on plant water uptake (see Puértolas et al., 2017, J. Exp. Bot.)

3. The relevance of determining plant (or at least soil) water status is illustrated by the discussion on the main finding of the paper: Si alleviate drought stress by increasing the levels of SOD and POD, which reduces oxidative stress. Drought, as expected, increased O₂^- and MDA and reduced SOD and POD in Si- plants, but not so on Si+. It is known that mild or initial water stress increase the expression of these enzymes, but in some experiments, higher intensity of stress induces a decline in the levels. The results presented here could then be interpreted in two ways, depending on the evolution of plant water status: If plant water status was similar between D-Si and D+Si, Si would be directly involved in decreasing oxidative stress by maintaining SOD and POD levels acting at the cellular level. Alternativelty, if plant water status was higher in Si+, it would mean that Si+ decreases oxidative stress by improving water uptake. Therefore the interpretation of the SOD and POD levels depend very much on the evaluation of plant water status. Without this, the authors can only speculate on the mechanisms explaining the observed results, which diminishes greatly the value of the study.
4. Related to the point above, in my opinion the structure of the discussion, divided into section dedicated to a group of physiological variables, is not correct. The discussion should be integrative and establish relationship between the different variables because all these variable are interrelated in the plant. For example, it is not credible that Si improved photosynthesis, without linking this to the enhanced stomatal conductance, which is possibly due to enhanced water uptake (according to some results in the literature). As mentioned before, the rest of the parameters studied could result from that enhanced water uptake, which would allow a better plant water status (again, it is a pity that the authors did not measure it). Alternatively, higher stomatal conductance in water stressed Si compared to no Si plants could be due to reduced ROS, which increase photosynthesis. All these interrelations must be discussed together and not separately. In this sense, the correlation analyses (by the way, please explain in the statistical analyses section in M&M) can be misleading in this type of studies, as correlation does not imply causality. The authors explain yield patterns from SOD and POD shifts between treatments following those analyses, but this cannot be taken as a proof of the main mechanism of Si action.

Other comments that might improve further versions of the manuscript.

Line 97. Is it possible that Na can have an effect? It would have been better to apply the same amount of Na+ in the Si- treatment. Please discuss and justify.

Line 98-99. Length of each stage? Were plants always well-watered outside of the drought periods (please specify)

Line 107. Frequency of irrigation to maintain soil water content? Oscillation of soil moisture content (higher and lower) between irrigations. As mentioned before, it is necessary to show soil moisture evolution in the four Si x Drought combinations at least during the stress period in each stage.

Figure 2. Letters in Leaf area V12 panel are wrong. It should be b b a a.

Line 156-157. Chlorophyll a is not is not significantly higher is D+Si at V12

Figures. Please improve the consistency of error bars design. They all should be either double or upper, not a mixture of both.

Author Response

Reviewer: 2

This is a well-written manuscript reporting the results of a study on the effects of Si application to alleviate drought stress in maize plants. The physiological changes underpinning these effects are also analysed. The authors found that Si might alleviate oxidative stress by enhancing the production of antioxidant enzymes, which seems to maintain higher stomatal conductance and photosynthesis. Unfortunately, the manuscript has some important flaws and inconsistent results that need to be revised by the authors.

1. The most important of these flaws is poor reporting of the details of the ANOVA (factors, levels, procedure to assess the meaning of interactions) and their results. The description of the ANOVA needs to be detailed in the statistical analyses section of Material and Methods. From the only report of these analyses in the manuscript (table3 and 4, analysis of yield and Si concentration, it seems that the authors performed a two-way ANOVA. I have many difficulties to understand how the authors applied such analyses given the experimental design. If I am not mistaken (following the description in material and methods), the authors initially distributed the plants into two Si application levels. Then, at three different crop stages, plants were further divided into two irrigation treatments. This can be described as a 2 x 3 x 2 factorial design that needs to be analysed by a three-way analysis of variance. In table 3 and 4, the authors list two of these main factors (Si and drought), and then the Si x Stage interaction. What about the stage effect and the Si x Drought and the Drought x Stage interactions? I can think of three possibilities: either the authors ignored reporting the other factors and interactions or they customise the general lineal model excluding them. Alternatively, the authors could wrongly identified the factors and the analyses should be redone in an appropriate manner. In any case, it is required that the authors explain the analyses (factors, levels) in detail and they should report the result of the ANOVA for the rest of the studied parameters, not just yield and Si concentration, either in the text or preferably in tables. In the present state, results are referenced to the figures, which can only offer a visual representation aided by the homogeneous group separation of means. However, Tukey test cannot be used as the exclusive reference for difference between levels of factors, the significance of factors and interactions in the ANOVA must be shown and interpreted accordingly. Moreover, the group separation procedure should be further explained in the statistical analysis section. Looking at the figures, it seems that the authors performed the Tukey test to compare the Drought x Si combinations within the same stage. If that is the case (it should be explained), this needs to be preceded by the assessment of the Drought x Si interaction and the Drought x Si x Stage interaction. Alternatively, given the complexity of three-way designs, and for the sake of clarity, I believe it would be admissible to perform independent two-way analyses (Drought x Si) to each stage (statisticians would disagree with me).

• Response: We totally agree with the reviewer’s suggestions. Your points are well taken and are fully considered by the authors. The major concern is that we didn’t carry out a three-way analysis of ANOVA for each variable, since actually there three factors for the experiment, that is drought level, drought stage, and Si treatment. To address the reviewer’s main concerns, we conduct a three-way analysis for all variables. We found that there exist no drought × stage × Si interactions for most variables, except several variables like leaf area, and O₂^-. The outcome of the three-way analysis suggests that the way we compared the effects of treatments on variables was suitable in our previous manuscript. To make it clear that the three-way analysis was well performed, we add the outcomes of our three-way analysis for each variable at the beginning of the paragraphs. We hope the description can meet the reviewer’s demand.

1. Another point of concern, related to the reporting of statistics, stems from the lack of reporting in figures caption. Bar figures should state the sample size (n) for bars and the type of error bar used (standard error, standard deviation). I am assuming that bars are standard error as usual (please change to s.e. otherwise). If this is the case, I found quite hard to believe some of the separation in homogeneous groups shown in the figures. There are many examples, but perhaps the most striking is the differences in carotenoids at R2 stage. Given the size of the error bar for D-Si, it is impossible that this treatment is significantly lower than the rest (even if the bar would be standard deviation). Numerous differences between bars are so small in comparison to the errors that it is unlikely they are statistically significant if the errors are reported as S.e. and they are correctly depicted. Please clarify and amend if necessary.

• Response: We totally agree with the reviewer. We are sorry for missing the key information of sample size and lacking the declaration of standard error for the means of data. To clarify it, we add additional description to the captions of each figure. We hope the modification can meet the reviewer’s request.

1. Experiments studying the effects of soil drying on plant physiology, must always report any sort of measure of either soil or plant water status, preferably both. The same way authors did measure Si concentration in leaves to assess the effectiveness of the application treatments and the interaction of other factors on Si uptake, the same should be done with water. I assume that the authors did not measure leaf water potential or water content (in my opinion, it is a shame that such an impressive array of measurements missed this very essential parameter that defines water status of the plant). However, it is very likely that they recorded pot weight and can retrospectively calculate, even approximately (estimating the weight of the maize plants) soil water content. This is an essential variable to report, as the interpretation of the authors in the discussion implicitly rely on the uniformity of water content within each irrigation treatment (across stages and Si treatments). However, if these two factors change transpiration rates or leaf area, the intensity of the drought could have been greater in those bigger more transpiring plant. This is unlikely for Si treatments, but it is most probable for stages. Thus, the authors attributed the smaller effect of drought in V6 compared to R2 to a lower sensitivity to drought in that stage, but it is most likely that the smaller size of the plant explain lower soil water depletion rates and lower intensity of stress over the 7 days period of stress application. The irrigation procedure to , which depends also on plant water uptake (see Puértolas et al., 2017, J. Exp. Bot.)

• Response: We agree with what the reviewer suggested for the manuscript. Our major objective is to investigate the effects of Si addition on the responses of maize plants to drought stress, which was kept at the same level (50% field capacity) for all deficit treatments, in comparison with none Si addition treatment. Also, the genotype of maize plants was kept constant, as well as the other field management throughout the whole experiment. I agree that maintaining soil water content at a constant level does not guarantee homogeneity of stress levels across treatments, and the physiological responses depended on moisture content which in turn was influenced by the plant water uptake rate. Irrigation frequency in our experiment was also kept constant. Drought levels were imposed as the same through different irrigation amounts. What’s more, we try our best to minimize the heterogeneity of stress levels between treatments by increasing sample sizes and replicates. We hope that the potential individual differences in stress levels existing in various plants were decreased to the fullest by increasing replicates and sample sizes.

1. The relevance of determining plant (or at least soil) water status is illustrated by the discussion on the main finding of the paper: Si alleviate drought stress by increasing the levels of SOD and POD, which reduces oxidative stress. Drought, as expected, increased O₂^-. and MDA and reduced SOD and POD in Si- plants, but not so on Si+. It is known that mild or initial water stress increase the expression of these enzymes, but in some experiments, higher intensity of stress induces a decline in the levels. The results presented here could then be interpreted in two ways, depending on the evolution of plant water status: If plant water status was similar between D-Si and D+Si, Si would be directly involved in decreasing oxidative stress by maintaining SOD and POD levels acting at the cellular level. Alternatively, if plant water status was higher in Si+, it would mean that Si+ decreases oxidative stress by improving water uptake. Therefore the interpretation of the SOD and POD levels depend very much on the evaluation of plant water status. Without this, the authors can only speculate on the mechanisms explaining the observed results, which diminishes greatly the value of the study.

• Response: Our results were in accordance with the second conclusion. That is if plant water status was higher in Si+, it would mean that Si+ decreases oxidative stress by improving water uptake. Previous studies have shown that Si addition can increase soil water uptake, as well as leaf water potential and relative leaf water content. Our result was mainly based on the previous study’s conclusion. Although the plant water status was not formally measured, we had visual expression of plant water status as follows (see the Fig. 1. below).
•  
• Fig. 1. Visual expression of plant water status for (A) Si+ and (B Si-) under the same drought level at the R2 stage of maize plants.

1. Related to the point above, in my opinion the structure of the discussion, divided into section dedicated to a group of physiological variables, is not correct. The discussion should be integrative and establish relationship between the different variables because all these variable are interrelated in the plant. For example, it is not credible that Si improved photosynthesis, without linking this to the enhanced stomatal conductance, which is possibly due to enhanced water uptake (according to some results in the literature). As mentioned before, the rest of the parameters studied could result from that enhanced water uptake, which would allow a better plant water status (again, it is a pity that the authors did not measure it). Alternatively, higher stomatal conductance in water stressed Si compared to no Si plants could be due to reduced ROS, which increase photosynthesis. All these interrelations must be discussed together and not separately. In this sense, the correlation analyses (by the way, please explain in the statistical analyses section in M&M) can be misleading in this type of studies, as correlation does not imply causality. The authors explain yield patterns from SOD and POD shifts between treatments following those analyses, but this cannot be taken as a proof of the main mechanism of Si action.

• Response: Agree with the reviewer’s comments on the structure of discussion section. Although related variables were discussed separately, we added the discussion on their correlations with other factors in discussion section. For example, we further discussed the relationships between water uptake and SOD, POD, and CAT activities, as well as the relationships between photosynthesis and physiological parameters. At the ending part of discussion section, all factors were integrated together to identify their relationships in terms of grain yields and biomass. We hope our revised discussion can satisfy the reviewer’s concerns.Line 97. Is it possible that Na can have an effect? It would have been better to apply the same amount of Na+ in the Si- treatment. Please discuss and justify.
• Other comments that might improve further versions of the manuscript.

• Response: Modified. Control plants were treated with an equivalent Na amount of Na₂SO₄. We have added this detailed information accordingly.
• Line 98-99. Length of each stage? Were plants always well-watered outside of the drought periods (please specify)

• Response: The length of each drought stage is 7 d. All plants were always well-watered outside of the drought periods.
• Line 107. Frequency of irrigation to maintain soil water content? Oscillation of soil moisture content (higher and lower) between irrigations. As mentioned before, it is necessary to show soil moisture evolution in the four Si x Drought combinations at least during the stress period in each stage.

• Response: We start irrigation whenever the soil water content is less than 65% field capacity using pot weighing method. The irrigation was stopped when soil water content reaches 80% field capacity. Therefore, there is no accurate frequency of irrigation. Irrigation was triggered according to real-time monitoring of soil water content estimated based on pot weight. It is regretful that we didn’t measure the oscillation of soil moisture content between irrigation events, nor the soil moisture evolution during the stress period in each growth stage. We try our best to explain the response of variables to drought stress using published mechanism in the discussion.
• Figure 2. Letters in Leaf area V12 panel are wrong. It should be b b a a.

• Response: Sorry for the mistake. The letters in the panel have been corrected accordingly.
• Line 156-157. Chlorophyll a is not significantly higher is D+Si at V12

• Response: Sorry for the mistake. The mistake has been corrected accordingly.
• Figures. Please improve the consistency of error bars design. They all should be either double or upper, not a mixture of both.

• Response: Done

Round 2

Reviewer 2 Report

The manuscript has been improved in this revision, with appropriate changes in the description of methods, and the discussion. I am still disappointed with the lack of reporting of soil water status; I assume that the authors did not record the pot weight before each irrigation during the experiment, which would have allowed monitoring soil water content. Despite this is a major drawback for the study, I think still offers valuable and novel data. Nevertheless, the most important change (it was not acceptable in the past state) is the appropriate analysis of the results using the three-way ANOVA that suits to the experimental design. However, these appropriate analyses revealed many inconsistencies in the interpretation of the results that I tried to compile comprehensively below, together with other comments. Before detailing them, I could suggest that perhaps the use of the three-way ANOVA approach is too stringent in this experiment, as the comparison of some values across stages could be difficult due to high inter-stage variability. Even though it is not statistically so accurate, I might suggest to perform a two-way ANOVA separately to each stage. That would sacrifice understanding the overall effect of the stage (maybe less interesting) in return for improving the capacity  to understand the Si x Drought interaction (which is probably more relevant to the study). The authors would need to make a decision of which alternative to choose, but always remembering that the interpretation of the results based on the post-hoc test must not contradict the ANOVA analyses (which has a higher hierarchy in the protection against Type I error).

Maize plant growth:

No need to say that the triple interaction is not significant all the time (only when it is). In contrast, it could be said which factors and interactions are significant in each subsection (but this is a style choice).

The effects of drought stress and Si application on plant height varied across stages (drought decreases H and LA except for R2).

This can be said of LA, but not of height as the Drought x Stage interaction is not significant for height.

Photosynthesis and chlorophyll content:

However, Si fertilization significantly improved the photosynthetic rate, transpiration rate, stomatal conductance, and chlorophyll a and chlorophyll b contents of the maize under drought stress conditions imposed at the V12 and R2 stages, whereas no significant differences were observed for these selected variables at the V6 stage regardless of whether Si was applied or not

To make that distinction between the effects of Si application in V12 and R2 from those in V6 under drought conditions, the Si x Stage x Drought interaction should be significant. Moreover, not even drought x Si interaction is significant in any case. Again, Tukey test must be used according to what is designed for and not to ‘mitigate’ lack of significance in ANOVAs  

Osmotic solute contents:

Soluble protein and sugar contents were substantially augmented by Si application under drought stress irrespective of the growth stage

Perhaps the authors should remark: but not under well-watered conditions (in this case the Drought x Si interaction is significant).

Si addition had no effect on proline content in the maize plants when drought stress was imposed at the V6 and V12  stages, and even decreased the proline content at the R2 stage

It is not possible to say this because neither Si effect, Si x Drought or the triple interaction are significant. In this case, it looks like at R2 there is a clear D x Si interaction, so, again, maybe is a good idea to analyse each stage separately.

Under the control conditions, no significant difference in osmotic solute contents was observed across the three stages with or  without Si treatment, with the exception of soluble sugars in the +Si treatment, which were  appreciably higher than the –Si treatment at the V12 and R2 stages

I cannot see that effect in V12, only in R2 (in this case it can be said indeed, because the D x Si x Stage interaction is significant).

Superoxide radicals and MDA contents:

The authors treat the results of O₂^- and MDA together, when according to the ANOVA they differ. It is true that in both cases the three main effects are significant. However, while for oxygen radicals all interactions are significant, for MDA none are. Thus, it cannot be said that MDA is reduced by Si addition only under stress, because there is no Si x Drought interaction (again, perhaps analysing each stage separately, would improve clarity).

Antioxidant enzyme activity:

Drought stress caused an obvious decrease in the activities of SOD, POD, and CAT, with the exception of SOD at the V6 stage (Fig. 7).

No interaction between Stage and Drought for SOD

Conversely, SOD, POD, and CAT activities were remarkably enhanced by Si application,

Not true for SOD (no Si significant effect)

However, no significant  differences were observed in the activities of SOD, POD, and CAT regardless of Si treatment under  well-watered control conditions

Could be correct for POD (significant Drought x Si interaction), but not for SOD or CAT.

Grain yield and silicon concentration:

It is difficult to understand the selection of the averages shown in table 3 and 4. The subsection Drought is show the average of the drought treatment in the three stages (I assume pooled across the two silicon treatments), and then the well-watered treatment (I assume pooled across the two silicon treatments and the three stages). Why do the authors chose to do that. It does not make sense. Neither do showing the average of the two Si treatments and then the average of the  Si x Drought x stage combination but only for the drought plants. Why did not the authors use the same layout than for the rest of the variables (grouped bar figures)?

Drought stress did not affect Si concentrations in the leaves regardless of Si fertilization

 According to the ANOVA table, this is not true. Drought is significant for leaves, stem and soil Si concentration.

The highest Si concentrations in the stems and soil were observed in the D-V6 treatment, followed by the D-V12 treatment and then the D-R2 treatment

Irrelevant, as there are no Drought x stage interaction

Figures (what does it mean mean of three replicates where n=9? Probably I did not see that the authors specified the number of replicates in your first version (sorry, my mistake), but now I am confused about this n=9

Author Response

Reviewer1:

The manuscript has been improved in this revision, with appropriate changes in the description of methods, and the discussion. I am still disappointed with the lack of reporting of soil water status; I assume that the authors did not record the pot weight before each irrigation during the experiment, which would have allowed monitoring soil water content. Despite this is a major drawback for the study, I think still offers valuable and novel data. Nevertheless, the most important change (it was not acceptable in the past state) is the appropriate analysis of the results using the three-way ANOVA that suits to the experimental design. However, these appropriate analyses revealed many inconsistencies in the interpretation of the results that I tried to compile comprehensively below, together with other comments. Before detailing them, I could suggest that perhaps the use of the three-way ANOVA approach is too stringent in this experiment, as the comparison of some values across stages could be difficult due to high inter-stage variability. Even though it is not statistically so accurate, I might suggest to perform a two-way ANOVA separately to each stage. That would sacrifice understanding the overall effect of the stage (maybe less interesting) in return for improving the capacity  to understand the Si x Drought interaction (which is probably more relevant to the study). The authors would need to make a decision of which alternative to choose, but always remembering that the interpretation of the results based on the post-hoc test must not contradict the ANOVA analyses (which has a higher hierarchy in the protection against Type I error).

• Response: We totally agree with the reviewer’s comments. Since Si x Drought interactionx stage interaction is not significant for most variables after three-way ANOVA analysis. Therefore, we focused much more on the effects of Si x Drought interaction, which were also the main aspects of the research. Then, we modified all the figures and the corresponding description in each related paragraphs. We hope the revised version now has been in a good shape, and address most of your concerns.

Maize plant growth:

No need to say that the triple interaction is not significant all the time (only when it is). In contrast, it could be said which factors and interactions are significant in each subsection (but this is a style choice).

The effects of drought stress and Si application on plant height varied across stages (drought decreases H and LA except for R2).

This can be said of LA, but not of height as the Drought x Stage interaction is not significant for height.

• Response: We modified the sentences accordingly.

Photosynthesis and chlorophyll content:

However, Si fertilization significantly improved the photosynthetic rate, transpiration rate, stomatal conductance, and chlorophyll a and chlorophyll b contents of the maize under drought stress conditions imposed at the V12 and R2 stages, whereas no significant differences were observed for these selected variables at the V6 stage regardless of whether Si was applied or not

To make that distinction between the effects of Si application in V12 and R2 from those in V6 under drought conditions, the Si x Stage x Drought interaction should be significant. Moreover, not even drought x Si interaction is significant in any case. Again, Tukey test must be used according to what is designed for and not to ‘mitigate’ lack of significance in ANOVAs

• Response: We have separately analysed data for each stage and found that Silicom and Drought×Si interaction was detected significant for photosynthetic rate, transpiration rate, stomatal conductance of the maize under drought stress conditions imposed at the V12 and R2 stages.

Osmotic solute contents:

Soluble protein and sugar contents were substantially augmented by Si application under drought stress irrespective of the growth stage

Perhaps the authors should remark: but not under well-watered conditions (in this case the Drought x Si interaction is significant).

• Response: Done

Si addition had no effect on proline content in the maize plants when drought stress was imposed at the V6 and V12 stages, and even decreased the proline content at the R2 stage

It is not possible to say this because neither Si effect, Si x Drought or the triple interaction are significant. In this case, it looks like at R2 there is a clear D x Si interaction, so, again, maybe is a good idea to analyse each stage separately.

• Response: We very agree with you. We have separately analysed data for each stage and found that Drought×Si interaction was detected significant for proline at R2 stage.

Under the control conditions, no significant difference in osmotic solute contents was observed across the three stages with or without Si treatment, with the exception of soluble sugars in the +Si treatment, which were appreciably higher than the –Si treatment at the V12 and R2 stages

I cannot see that effect in V12, only in R2 (in this case it can be said indeed, because the D x Si x Stage interaction is significant).

• Response: Agree! We modified the sentences accordingly.

Superoxide radicals and MDA contents:

The authors treat the results of O₂- and MDA together, when according to the ANOVA they differ. It is true that in both cases the three main effects are significant. However, while for oxygen radicals all interactions are significant, for MDA none are. Thus, it cannot be said that MDA is reduced by Si addition only under stress, because there is no Si x Drought interaction (again, perhaps analysing each stage separately, would improve clarity).

• Response: We modified the sentences accordingly, and analysedeach stage separately.

Antioxidant enzyme activity:

Drought stress caused an obvious decrease in the activities of SOD, POD, and CAT, with the exception of SOD at the V6 stage (Fig. 7).

No interaction between Stage and Drought for SOD

Conversely, SOD, POD, and CAT activities were remarkably enhanced by Si application,

Not true for SOD (no Si significant effect)

However, no significant  differences were observed in the activities of SOD, POD, and CAT regardless of Si treatment under  well-watered control conditions

Could be correct for POD (significant Drought x Si interaction), but not for SOD or CAT.

• Response: We modified the sentences accordingly.

Grain yield and silicon concentration:

It is difficult to understand the selection of the averages shown in table 3 and 4. The subsection Drought is show the average of the drought treatment in the three stages (I assume pooled across the two silicon treatments), and then the well-watered treatment (I assume pooled across the two silicon treatments and the three stages). Why do the authors chose to do that. It does not make sense. Neither do showing the average of the two Si treatments and then the average of the  Si x Drought x stage combination but only for the drought plants. Why did not the authors use the same layout than for the rest of the variables (grouped bar figures)?

• Response: We have make the layout of Table 3 and 4 the same as the grouped bar figures through deleting the average data of both drought treatments and Si treatments.

Drought stress did not affect Si concentrations in the leaves regardless of Si fertilization

According to the ANOVA table, this is not true. Drought is significant for leaves, stem and soil Si concentration.

• Response: Sorry for the mistake. We have corrected it accordingly.

The highest Si concentrations in the stems and soil were observed in the D-V6 treatment, followed by the D-V12 treatment and then the D-R2 treatment

Irrelevant, as there are no Drought x stage interaction

• Response: We have deleted the incorrect description.

Figures (what does it mean mean of three replicates where n=9? Probably I did not see that the authors specified the number of replicates in your first version (sorry, my mistake), but now I am confused about this n=9.

• Response: The sample size is 3. We have correct n=9 to n=3.