Leguminous Cover Crop Astragalus sinicus Enhances Grain Yields and Nitrogen Use Efficiency through Increased Tillering in an Intensive Double-Cropping Rice System in Southern China
Round 1
Reviewer 1 Report
Materials and Methods
What about water management?
Were any herbicides, pesticides, or other inputs applied?
cv. Luxiangyou 996 - is this a hybrid or inbreed? How does it compare to its competitors in terms of yield, disease resistance, etc.
Results
Discussion
What about previous work that indicates there is a cultivar effect on results from unconventional agriculture production methods? That is, some high yielding wheat cultivars don't perform as well as low yield cultivars under organic management methods. This could be a limitation to this study - only one cultivar was studied.
"To better evaluate the sustainability of vetch-double rice system, it is important to consider 247 options that have the potential both to improve rice production and to decrease chemical N demand." What does the word sustainability mean to the authors?
The authors should mention that not just yield is of importance to a rice farmer. Thus, future work should evaluate this cover crop effects on rice milling and cooking quality.
The authors used LSDs in their work. I suggest they review the following work and consider how to improve their paper accordingly.
“Is, or is not, the two great ends of Fate”: Errors in Agronomic Research"
Kimberly Garland Campbell,* Yvonne M. Thompson, Stephen O. Guy,
Marla McIntosh, and Barry Glaz https://www.researchgate.net/publication/273507773_or_the_two_great_ends_of_Fate_Errors_in_Agronomic_Research
"An often-used approach in agricultural research is to plant a large experiment, evaluate a single H0, and report LSDs for all pairwise tests. This approach is probably a significant waste of resources. Overuse of comparison-wise tests, such as the LSD, result in unacceptably high a error rates, while the overuse of methods that correct for experiment-wise error, such as the Bonferroni test, are often too conservative and inflated by irrelevant comparisons of no practical interest to the researcher, resulting in high b error rates. Agronomic experiments are frequently designed with complicated split-plot and factorial structures with multiple independent and often correlated effects. In these complicated structures, several hypotheses can be tested within a single experiment. These hypotheses need to be specified in advance so that the experiment can be designed to adequately test those of the greatest practical interest. We recommend using power analysis to provide a reality check so that experiments can be designed to adequately test hypotheses within resource constraints. During the design phase, the experimental parameters that can be manipulated include the size of the trial, the number of treatments, the number of replications, the desired effect size that is useful and practical, and the use of experimental designs that control for spatial variation or correlated errors. Contrast statements should be developed and a range of estimates of the power to detect small, medium, and large effect sizes should be calculated for all treatment effects and their interactions before conducting the experiment The planning of experiments should include attention to the relative costs of false positive and false negative errors. False positives are not the only potential risk. False negatives can carry just as much or even more negative impact when farmers make decisions based on conclusions from our experiments. For the future, additional work on optimal error rates for analysis of non-normally distributed data and for analysis of massively large data sets with correlated effects, for example, data from spectral reflectance indices, is still needed. Finally, if nothing else can convince researchers to reduce their reliance on null hypothesis testing, we need to consider that the H0 is always false with a large enough sample size. The actual interpretation of a failure to reject the H0 is that the H0 cannot be concluded to be true, so we cannot conclude it to be false. Thus, “the prevailing yes–no decision at the magic 0.05 level is a far cry from the use of informed judgment” (Cohen, 1990). Science would be better served if measures of variation, effect sizes, and relative costs of error were included in published reports of results. One of the most useful and intuitive measures of precision is the confidence interval and, if nothing else, calculating and reporting confidence intervals for observed treatment effects would add significantly to the ability of others to interpret and draw conclusions from our research. The results of our individual experiments are not the final point to be made about the systems that we study. Our research exists in the middle of a stream of scientific discovery. Optimally, our research validates the research of those who came before, and if we provide sufficient information to those who come after, then they will challenge and expand our research conclusions."
Author Response
Dear reviewer,
Thank you very much for your useful comments and suggestions on our manuscript. We have made corrections followed their suggestions and we also revised the language and the content of the manuscript. All the corrections were marked in red. Here is the details of point to piont responses to all the reviewer’s suggestions.
Materials and Methods：
What about water management?
Response: The water management was added with ‘The field was flooded after transplanting, and a floodwater depth of 3–5 cm was maintained until a week before maturity except that the water was drained at maximum tillering stage to reduce unproductive tillers’ in the text.
Were any herbicides, pesticides, or other inputs applied?
Response: Other managements such as herbicides and pesticides were added in the text. Weeds, pests, and diseases were intensively controlled to avoid yield loss according to the actual demands.
cv. Luxiangyou 996 - is this a hybrid or inbreed? How does it compare to its competitors in terms of yield, disease resistance, etc.
Response: The ‘cv. Luxiangyou 996’ is a wrong spelling word, cv. Luliangyou 996 is two-line hybrid cultivar. cv. Yueyou 712 is a three-line hybrid cultivar. The two varieties both meet the national (China) rice variety certification standards, and two widely grown in early-season and late-season rice cultivars in Hunan Province.
Discussion
What about previous work that indicates there is a cultivar effect on results from unconventional agriculture production methods? That is, some high yielding wheat cultivars don't perform as well as low yield cultivars under organic management methods. This could be a limitation to this study - only one cultivar was studied.
Response: Good point. We have ignored this limitation in our study, in our future study we will pay more attention to this.
"To better evaluate the sustainability of vetch-double rice system, it is important to consider options that have the potential both to improve rice production and to decrease chemical N demand." What does the word sustainability mean to the authors?
Response: In our study, the sustainability means with less chemical N addition and more grain yeild in rice production.
The authors should mention that not just yield is of importance to a rice farmer. Thus, future work should evaluate this cover crop effects on rice milling and cooking quality.
Response: Good point. To a rice farmer, the important goal is not just the yield, however, in this research we did not evaluate the effects on rice milling and cooking quality, we will pay more attention to the rice milling and cooking quality in our future work.
The authors used LSDs in their work. I suggest they review the following work and consider how to improve their paper accordingly.
“Is, or is not, the two great ends of Fate”: Errors in Agronomic Research"
Kimberly Garland Campbell,* Yvonne M. Thompson, Stephen O. Guy,
Marla McIntosh, and Barry Glaz https://www.researchgate.net/publication/273507773_or_the_two_great_ends_of_Fate_Errors_in_Agronomic_Research
"An often-used approach in agricultural research is to plant a large experiment, evaluate a single H0, and report LSDs for all pairwise tests. This approach is probably a significant waste of resources. Overuse of comparison-wise tests, such as the LSD, result in unacceptably high a error rates, while the overuse of methods that correct for experiment-wise error, such as the Bonferroni test, are often too conservative and inflated by irrelevant comparisons of no practical interest to the researcher, resulting in high b error rates. Agronomic experiments are frequently designed with complicated split-plot and factorial structures with multiple independent and often correlated effects. In these complicated structures, several hypotheses can be tested within a single experiment. These hypotheses need to be specified in advance so that the experiment can be designed to adequately test those of the greatest practical interest. We recommend using power analysis to provide a reality check so that experiments can be designed to adequately test hypotheses within resource constraints. During the design phase, the experimental parameters that can be manipulated include the size of the trial, the number of treatments, the number of replications, the desired effect size that is useful and practical, and the use of experimental designs that control for spatial variation or correlated errors. Contrast statements should be developed and a range of estimates of the power to detect small, medium, and large effect sizes should be calculated for all treatment effects and their interactions before conducting the experiment The planning of experiments should include attention to the relative costs of false positive and false negative errors. False positives are not the only potential risk. False negatives can carry just as much or even more negative impact when farmers make decisions based on conclusions from our experiments. For the future, additional work on optimal error rates for analysis of non-normally distributed data and for analysis of massively large data sets with correlated effects, for example, data from spectral reflectance indices, is still needed. Finally, if nothing else can convince researchers to reduce their reliance on null hypothesis testing, we need to consider that the H0 is always false with a large enough sample size. The actual interpretation of a failure to reject the H0 is that the H0 cannot be concluded to be true, so we cannot conclude it to be false. Thus, “the prevailing yes–no decision at the magic 0.05 level is a far cry from the use of informed judgment” (Cohen, 1990). Science would be better served if measures of variation, effect sizes, and relative costs of error were included in published reports of results. One of the most useful and intuitive measures of precision is the confidence interval and, if nothing else, calculating and reporting confidence intervals for observed treatment effects would add significantly to the ability of others to interpret and draw conclusions from our research. The results of our individual experiments are not the final point to be made about the systems that we study. Our research exists in the middle of a stream of scientific discovery. Optimally, our research validates the research of those who came before, and if we provide sufficient information to those who come after, then they will challenge and expand our research conclusions."
Response: Thanks for introducing the new method in experimental design and data statistics. In this paper, we use the traditional two-factor randomized block design and method analysis, and we will consider the recommended method in future studies.
Author Response File: 
Author Response.doc
Reviewer 2 Report
The current manuscript presents a study concerning that winter cover cropping of rice with leguminous vetch. According to the authors the proposed practice increases grain yields, N uptake and use efficiency compared to winter fallow. The experiment is rather well designed and meets the standards to draw clear conclusions. However, some parts and especially the statistics and their presentation of the results (tables and figures) needs more attention, so that the manuscript to become suitable for publication.
Corrections
“Astragalus sinicus” in the titles needs to be in italics
Line 20: the phrase “subsequent two rice crops” needs revision in order to explain the two cropping systems
Lines 41 to 43: I believe that they Need better wording using “Nitrogen leaching”
Line 48: These “improved management practices”
Lines 80 to 84: Is there any treatment that can be characterizes as farmer’s common practice? If not why you have considered not to include it in the experimentation design?
Line 112: “and grain weight per 1000 seeds”, “1000 grains weight”
Line 117: I would feel more confident if you add how N content was determined in vetch, similar to rice
Line 134: This is true only for FN200 and MN200 in year 2015 in late rice season. In all the other cases (FN100 and MN100, FN200 MN200) there were significant differences in year 2015 according to Table 2.
Line 140: Yield components analysis must be summarized and analyzed over year like the other traits in Table 1. Only if the factor Year (Y) is non-significant you are able compare the results for each trait over year. Otherwise (significant differences among years), the results must be analyzed for each year separately. In this case, the different letters in the same column and within the same year in Table 3 mean that there is a significant difference.
Lines 154-155: The sentence in lowercase letters needs to be removed from the text
Line 157: See previous comment concerning the over year statistics.
Line 174: Define if the lower letters concern within the same year only in all years’ significant differences
Lines 189-190: There are cases where the SE’s area huge, indeed. Can you recheck them, or can you provide the Coefficient of Variances for each analysis?
Lines 223-224: Some comparisons with the present yield status will help to understand the magnitude of the findings and to the real world.
Line 237: C: N should be changed to C/N
References: I do recommend to include the publication DOI:10.1038/s41598-017-03919-y , compare results and discuss it.
Author Response
Dear reviewer,
Thank you very much for your useful comments and suggestions on our manuscript. We have made corrections followed their suggestions and we also revised the language and the content of the manuscript. All the corrections were marked in red. Here is the details of point to piont responses to all the reviewer’s suggestions.
Comments and Suggestions for Authors：
The current manuscript presents a study concerning that winter cover cropping of rice with leguminous vetch. According to the authors the proposed practice increases grain yields, N uptake and use efficiency compared to winter fallow. The experiment is rather well designed and meets the standards to draw clear conclusions. However, some parts and especially the statistics and their presentation of the results (tables and figures) needs more attention, so that the manuscript to become suitable for publication.
Response：Thank you for your useful comments on the paper. We have revised the manuscript as suggested.
Corrections
“Astragalus sinicus” in the titles needs to be in italics
Response：Thank you for reminding us the careless mistakes in the manuscript. It was revised.
Line 20: the phrase “subsequent two rice crops” needs revision in order to explain the two cropping systems
Response：The sentence has been rewritten.
Lines 41 to 43: I believe that they Need better wording using “Nitrogen leaching”
Response：Good point. This sentence has been rewritten.
Line 48: These “improved management practices”
Response: The sentence has been rewritten.
Lines 80 to 84: Is there any treatment that can be characterizes as farmer’s common practice? If not why you have considered not to include it in the experimentation design?
Response: Thank you for pointing out the farmer’s common practice. The farmer’s common practice is similar with the treatment FN200 (winter fallow and 200 kg N ha-1 input for both the early and late rice).
Line 112: “and grain weight per 1000 seeds”, “1000 grains weight”
Response: The sentence has been rewritten.
Line 117: I would feel more confident if you add how N content was determined in vetch, similar to rice
Response: Thank you for reminding us these N concentration analysis method. We have added this in the text.
Line 134: This is true only for FN200 and MN200 in year 2015 in late rice season. In all the other cases (FN100 and MN100, FN200 MN200) there were significant differences in year 2015 according to Table 2.
Response: Thank you for the reminding. Here, we just want to express the point that compared with 100 kg ha-1 chemical N fertilizer addition, vetch cover could increase grain yields.
Line 140: Yield components analysis must be summarized and analyzed over year like the other traits in Table 1. Only if the factor Year (Y) is non-significant you are able compare the results for each trait over year. Otherwise (significant differences among years), the results must be analyzed for each year separately. In this case, the different letters in the same column and within the same year in Table 3 mean that there is a significant difference.
Response: Good point. The factor Year (Y) is non-significant, so we can compare the results for each trait over year. In our manuscript, the different letters in the same column and within the same year in Table 3 mean that there is a significant difference.
Lines 154-155: The sentence in lowercase letters needs to be removed from the text
Response: The sentence in lowercase letters was explanation of the treatments in this study. Hence, we keep it in the text.
Line 157: See previous comment concerning the over year statistics.
Response: Thank you for the reminding.
Line 174: Define if the lower letters concern within the same year only in all years’ significant differences
Response: Thank you for the reminding. The indicators in the figure, such as nitrogen uptake and nitrogen utilization efficiency, were compared separately every year, and inter-annual comparisons were not made. Different lowercase letters were used to indicate that the results of multiple comparisons were significantly different, while the same letters indicated that the differences were not significant.
Lines 189-190: There are cases where the SE’s area huge, indeed. Can you recheck them, or can you provide the Coefficient of Variances for each analysis?
Response: Thank you for reminding us the SE. We have checked it and corrected it, Vertical bars represent the standard deviation in the old figures, and the Vertical bars represent the standard errors in the new figures.
Lines 223-224: Some comparisons with the present yield status will help to understand the magnitude of the findings and to the real world.
Response: Thanks for the reminding. There are many factors influencing the yield of rice. Even for the same variety, the yield will vary greatly in different soil climate and field management, so it can not be compared with the data outside the experiment.
Line 237: C: N should be changed to C/N
Response: The word has been changed to C/N as suggested.
References: I do recommend to include the publication DOI:10.1038/s41598-017-03919-y , compare results and discuss it.
Response: Thank you for your reminding. We have read this paper, and compare our results with their findings and discuss it in the text.
Author Response File: Author Response.doc
