Groundcover Mulching in Mediterranean Vineyards Improves Soil Chemical, Physical and Biological Health Already in the Short Term

Round 1

Reviewer 1 Report

The paper addresses the research area related to "Farming and cropping systems" of the MDPI Agronomy journal. The subject is of interest for viticulture, especially under Mediterranean climate.

The paper is generally well organized. However, in my opinion, the shortage of this paper is related to the experimental design. The soil variability issue is presented in the ‘Introduction’ section (Lines 98-103). Still, it is not clear how the treatments (Figure 1) were implemented according to Figure 2 and 3.

A larger number of replicates would strengthen the results given the soil variability. A randomised completed block design has been defined in previous studies (e.g., Giese et al. 2014, Sci Hortic; Muscas et al. 2017, Agric Ecosyst Environ; Meissner et al. 2019, OENO One).

More detailed information is needed regarding “initial soil sampling campaign”. Are there any significant statistical differences between treatments/replicates (organic content, texture…)? And statistical differences between soil depths?

Lines 171-172 state “Each experimental plot was divided in three replicates according to the slope of the vineyard.” But no results are presented in the manuscript regarding the slope effect.

Author Response

All the line numbers refer to the track-changes version of the manuscript attached.

The paper addresses the research area related to “Farming and cropping systems” of the MDPI Agronomy journal. The subject is of interest for viticulture, especially under Mediterranean climate.

1. R1: The paper is generally well organized. However, in my opinion, the shortage of this paper is related to the experimental design. The soil variability issue is presented in the ‘Introduction’ section (Lines 98-103). Still, it is not clear how the treatments (Figure 1) were implemented according to Figures 2 and 3.

A.: We modified Figure 2 and 3 and added the treatments positioning at lines 388 and 404

2. A larger number of replicates would strengthen the results given the soil variability. A randomized completed block design has been defined in previous studies (e.g., Giese et al. 2014, Sci Hortic; Muscas et al. 2017, Agric Ecosyst Environ; Meissner et al. 2019, OENO One).

A: We agree with the reviewer on the potential benefits provided by a larger number of replicates. We also are aware that some of the studies cited in this work were based on randomized complete block design. Nevertheless:

• On a practical side: this is an on-farm trial and thus complex designs would have been difficult to implement and maintain.
• On a methodological side:

1. It should be noted that randomization in vineyards often refers to randomization between vine-rows and not within (e.g. chess table design). We argue that in CRBD experiments the soil homogeneity within the rows remained un-tested and the possible soil variability which very often characterizes vineyards is not taken into account in the evaluation of specific farming practices.
2. Considering the within-row soil variability, the randomization across vine rows (typically used in viticulture) would have not improved the statistical power of our analysis.
3. Complex designs may improve (e.g. chess table design) statistical power as compared with “row-randomized” design but also results in a high edge effect due to the limited size of the plots.
4. It has been clearly shown that knowledge of the spatial characteristics of the soil and the use of continuous soil covariates, strongly improves the statistical power of the experiments. As an example:

Alesso et al. 2019, tested different experimental designs both randomized and non-randomized with and without continuous covariates. 

"Our results showed that the addition of the spatial information, the analysis yielded more reliable estimates of treatment effects, keeping Type I error rates under the nominal rate of 5%, regardless of the spatial layout, whether it was randomized or not, and spatial structure. "

Also Rudolph et al., 2016 reported on the improved statistical power when continuous soil covariates (the author used ECa) are included in the statistical models. Moreover, the authors suggested the integration of different proximal sensors as we did in our study.

"In conjunction with our results showing that there are no gains from adjusted blocking when ECa is used as a covariate in ANCOVA, this suggests that adjusted blocking is not, in general, the best way to incorporate such covariates into experimental design and analysis".

(…)

"Other proximal sensed data such as propagation velocity of electromagnetic waves, natural emission of gamma radiation, soil reflectance spectra, soil pH and soil nutrients, obtained by ground-penetrating radar (GPR), γ-radiometry, reflectance spectroscopy, or ion-selective electrodes, could be helpful."

To sum up, in our context it was not possible to randomize the trial both for management reasons and due to the in-row soil variability. We, therefore, replicated the experiment in two farms and we conducted a thorough study of the soil variability and explicitly included in the statistical analysis through the variables clay, sand, silt, gravel, total limestone, soil organic matter, Mg and K so leading to an innovative approach never used before in viticulture studies, to the best of our knowledge.

Reference:

Alesso, C. A., Cipriotti, P. A., Bollero, G. A., & Martin, N. F. (2019). Experimental Designs and Estimation Methods for On‐Farm Research: A Simulation Study of Corn Yields at Field Scale. Agronomy Journal, 111(6), 2724-2735.

Rudolph, S., Wongleecharoen, C., Lark, R. M., Marchant, B. P., Garré, S., Herbst, M., ... & Weihermüller, L. (2016). Soil apparent conductivity measurements for planning and analysis of agricultural experiments: A case study from Western-Thailand. Geoderma, 267, 220-229.

3. More detailed information is needed regarding “initial soil sampling campaign”. Are there any significant statistical differences between treatments/replicates (organic content, texture…)? And statistical differences between soil depths?

A: Thank you for your suggestion, the point is well taken as statistical elaboration on the initial soil sampling campaign was not mentioned in the paper. We run Feasible Solution Algorithm (FSA) on 2017 data, prior to the implementation of the trial, to investigate differences across the inter-rows where the treatments would have been implemented which were coded as “Treatment-T0”. Treatment-T0 was not selected by FSA as a critical factor for SOM, N, P₂O₅ and K, thereby suggesting the suitability of the sites for the experiment. We added this consideration in Material and Method under statistical analysis from line 323 to line 369.

4. Lines 171-172 state “Each experimental plot was divided into three replicates according to the slope of the vineyard.” But no results are presented in the manuscript regarding the slope effect.

A: We carried out variables selection by exploring all possible variable subsets providing models with no interactions, selecting those models which minimized BIC. Such variables along with “Treatment” were used to feed the Feasible Solution Algorithm, allowing the algorithm to include interactions. FSA solutions are optimal in the sense that no single swap to any of the variables will increase the criterion function (BIC). So, variables were selected through the BIC of the model including only such variables, and the latter model was used as a starting point to select the optimal model (low BIC) including interactions. In all the models the slope was never selected. Moreover, the slope was also tested as random effect, bringing no model fit improvements.

Author Response File: Author Response.docx

Reviewer 2 Report

The submitted manuscript reports an extensive study carried out in central Italy about different training soil systems in vineyard soil. In this study, different physical, chemical and biological parameters are studied in 2 consecutive years. The treatments compared are traditional tilled soil and different types of plant cover (spontaneous vegetation and 3, alone or blended, cultivated plants seeded).

As a whole, it is very interesting to highlight its careful presentation and exposition, especially in the introduction section and discussion of results section. The approach is interesting regarding the issue and the increasingly widespread concern for the sustainability of agriculture and the conservation of the environment.

Unfortunately the length of the study is excessive from my point of view and the results presented are interesting, although not very novel since they go in the direction of most actual studies as the cited literature reports. However, the manuscript may be of interest for publication after a Major Revision:

Firstly, I would suggest deleting the entire study of soil variability by studying electrical conductivity and gamma ray. I am not a specialist on these subjects, but regarding the results presented, they do not seem to be relevant for the conclusions of the study. If the soils are homogeneous, it is enough to provide dispersion measures that can be reported in Table 1 with the classic parameters (standard deviation or coefficient of variation). If the soils are very heterogeneous, the study could not be carried out, it would be necessary to look for other farms or it would be necessary to carry out another statistical design with more blocks and more repetitions. Lines 96-118, 175-215, 287-317 and 328-339 could be deleted in addition to Figures 2 and 3. Consequently, the lines referring to this point in the abstract and in the conclusions must be deleted.

Secondly, the excessive number of acronyms makes the reading of the manuscript tedious and ,sometimes, difficult the understanding. The authors should try to reduce that number of acronyms and/or put their full meaning in more occasions. An specific table of acronyms should help…

Specific corrections:

Line 46: Where it says “course” texture, I think you mean “coarse” texture.

Line 154: It would be interesting to report the depth of working of the discs.

Line 161: It would be interesting to explain for figure 1 what type of work is done in the streets indicated as buffer.

Line 217: The dates of sampling should be as close as possible: the results in October, November or January, which are the dates that the authors report, may have important differences, especially in the biological indices of soil activity. In any case, the sincerity of the authors is appreciated.

Line 241: SSI is not widely used, at least I am not familiar with it. Furthermore, looking at the results, the differences, although significant according to the analysis of variance, are very small and surprisingly do not vary with depth.

Line 261: It would have been interesting to do the biological quality index (QBS) of the soil at the beginning of the study (year 2017), in any case it would be very similar to the result obtained for the traditional tillage treatment.

Line 326: Add dispersion measures to this table if the entire section on the study of soil variability is eliminated. In this same table, the results of silt, clay and sand are often given in percentage.

Line 350: The analysis of variance (table3) for  the organic matter of the soil (SOM) and for the following parameters provides redundant information with respect to the figures presented (figure 4 and following). Considered if it is necessary.

Lines 598, 630 and others: When discussing the results, it is not necessary to write the letters e.g. before of the reference numbers.

Author Response

All the line numbers refer to the track-changes version of the manuscript attached.

1. Firstly, I would suggest deleting the entire study of soil variability by studying electrical conductivity and gamma-ray. I am not a specialist on these subjects, but regarding the results presented, they do not seem to be relevant for the conclusions of the study. If the soils are homogeneous, it is enough to provide dispersion measures that can be reported in Table 1 with the classic parameters (standard deviation or coefficient of variation). If the soils are very heterogeneous, the study could not be carried out, it would be necessary to look for other farms or it would be necessary to carry out another statistical design with more blocks and more repetitions. Lines 96-118, 175-215, 287-317 and 328-339 could be deleted in addition to Figures 2 and 3. Consequently, the lines referring to this point in the abstract and in the conclusions must be deleted.

A: The study of soil variability performed in this article is one of the paper's novelty. In Italian vineyards, like in many other parts of the world, the soil spatial variability is generally high, because of natural geological and pedological discontinuities and the human activity to prepare the land for vineyards (land leveling, deep plowing, etc.). Often, such soil spatial variability is not taken into account in agronomic tests, considering soil homogeneous in all the study plots with no statistical test demonstrating such homogeneity within blocks. In other cases, the study plots are relatively homogeneous, but they are very small and not completely representative of the natural soil variability. This paper tried to take into consideration the natural soil variability of the studied vineyards in the statistical models used, verifying the effect of this variability on the treatments results. This is already explained in the introduction “rows 99-105”(first manuscript submitted).

Further details are provided in the reply to comment 2 and 4 to reviewer 1 and in the modified section of the statistical analysis from line 323 to line 369.

2. Secondly, the excessive number of acronyms makes the reading of the manuscript tedious and , sometimes, difficult the understanding. The authors should try to reduce that number of acronyms and/or put their full meaning on more occasions. A specific table of acronyms should help…

A: As suggested we reduced the number of acronyms. We specifically removed SPR (soil penetration resistance) and SSI (soil structure stability index)

3. Line 46: Where it says “course” texture, I think you mean “coarse” texture.

A: Yes thank you for your suggestion

4. Line 154: It would be interesting to report the depth of working of the discs.

A: We added the depth of the disc as requested in lines 183, 189

5. Line 161: It would be interesting to explain for figure 1 what type of work is done in the streets indicated as buffer.

A: Details on the management of the buffer inter-rows were included as requested in line 203

6. Line 217: The dates of sampling should be as close as possible: the results in October, November, or January, which are the dates that the authors report, may have important differences, especially in the biological indices of soil activity. In any case, the sincerity of the authors is appreciated.

A: We agree with the comments but unfortunately it was not possible to sample on the same dates every year due to climatic factors and difficult soil conditions which would have severely impacted the quality of the data.

7. Line 241: SSI is not widely used, at least I am not familiar with it. Furthermore, looking at the results, the differences, although significant according to the analysis of variance, are very small and surprisingly do not vary with depth.

A: Yes differences are quite small and only concerned tillage vs spontaneous grassing. We found a bug in the script and corrected table 7 and related models, from line 577 to 584. Results did not change significantly.

8. Line 261: It would have been interesting to do the biological quality index (QBS) of the soil at the beginning of the study (year 2017), in any case, it would be very similar to the result obtained for the traditional tillage treatment.

A: We agree but unfortunately we could not sample QBS in 2017.

9. Line 326: Add dispersion measures to this table if the entire section on the study of soil variability is eliminated. In this same table, the results of silt, clay, and sand are often given in percentage.

A: We kept the section on soil variability and reported the texture as concentration in table 1.

10. Line 350: The analysis of variance (table3) for the organic matter of the soil (SOM) and for the following parameters provides redundant information with respect to the figures presented (figure 4 and following). Considered if it is necessary.

A: Thank you for your suggestion. We agree with the reviewer and erased table 3 as the information was already presented in the text and in the figure.

11. Lines 598, 630, and others: When discussing the results, it is not necessary to write the letters e.g. before of the reference numbers.

A: we erased all the “e.g.” as requested

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The manuscript was revised and completed according to my notices.

Author Response

Thank you very much for your contribution in improving the clarity and quality of the paper.

Reviewer 2 Report

After checking the article as it remains after the first review, I have verified that the authors have considered most of the indications given. However, there are two issues that should be reconsidered both by the editor and the authors:

I do not know if the journal has a length limit for published articles, but it seems to me that the length is still excessive. I suggested to cut the article eliminating the study of soil uniformity since it does not report relevant information for the results obtained, from my point of view. However, if the editor does not find problems with the length of the article, I have no inconvenient to include the mentioned study.

On the other hand, it was indicated that Table 3 provides redundant information with the following figure (point 10 of 1^st the review). The authors agree with this point and have deleted it. The same thing happens with the rest of the analysis of variance tables that precede the successive figures and that also provide redundant information. I think that these tables should also be eliminated, making the article more compact and not losing information or quality of the representations.

Author Response

Point by point answer to Reviewer 2

After checking the article as it remains after the first review, I have verified that the authors have considered most of the indications given. However, there are two issues that should be reconsidered both by the editor and the authors:

I do not know if the journal has a length limit for published articles, but it seems to me that the length is still excessive. I suggested to cut the article eliminating the study of soil uniformity since it does not report relevant information for the results obtained, from my point of view. However, if the editor does not find problems with the length of the article, I have no inconvenient to include the mentioned study.

A: We acknowledge the inputs of the reviewer which were taken into account and improved the paper. Concerning this last point, the journal does not have a word limit and we believe that it is relevant to keep the information on soil variability. This is critical to understand the innovative method that we used to take such variability into account and improve our regression models.

On the other hand, it was indicated that Table 3 provides redundant information with the following figure (point 10 of 1st the review). The authors agree with this point and have deleted it. The same thing happens with the rest of the analysis of variance tables that precede the successive figures and that also provide redundant information. I think that these tables should also be eliminated, making the article more compact and not losing information or quality of the representations.

A: We thank the reviewer for the comment. Still, we would like to keep the ANOVA tables as they provide relevant information on the results of the statistical analysis. We erased Table 3 as we were showing in the graphs all the significant factors and interactions. However, this does not apply for all the dependent variables as this would result in an excessive number of graphs and additional text. Furthermore, the ANOVA tables also provide valuable information on: (i) the significance of factors and interactions, (ii) degrees of freedom, (iii) chi-square, and sum of square. Having said that if the editor agrees with the comment of the reviewer we will act accordingly.