Contrasting Effects of Tillage and Landscape Structure on Spiders and Springtails in Vineyards

Round 1

Reviewer 1 Report

Please find below my opinion about the paper “Contrasting effects of tillage and landscape structure 2 on spiders and springtails in vineyards” by Pfingstmann et al.

This manuscript reports interesting results about the effect of soil management on spiders communities in a vineyard context, in relation with surrounding landscape and presence of prey (Collembola communities). Even if some additional informations are required, this paper represents an original progress for sustainable pest management in vineyards, due to its original approach, considering prey-predator relationship at a landscape scale.

This manuscript also presents a few concerns that may have to be corrected before it becomes suitable for publication. Some references should be added in the introduction to give a better overview of results already obtained in a agricultural context, and in the discussion on the effect of surrounding landscape on spiders. On the other hand, additional explanations should be given about statistical analyses : some analysis are not mentionned, and standard error should be prefered to standard deviation.

I therefore recommend a revision of the paper.

I hope these comments will be useful to improve the manuscript. Details are provided in the attached file.

Comments for author File: Comments.pdf

Author Response

Pfingstmann et al.´s reply to comments of Reviewer 1

We are grateful for the thorough reading of our manuscript and appreciate the constructive comments on improving our manuscript. In the revision we addressed all points raised by Reviewer 1. Please find below our point-to-point reply. All changes on the original text have been highlighted through tracked changes. We would also like to thank for these constructive comments in the revised acknowledgements.

Reviewer 1: Please find below my opinion about the paper “Contrasting effects of tillage and landscape structure 2 on spiders and springtails in vineyards” by Pfingstmann et al. This manuscript reports interesting results about the effect of soil management on spiders communities in a vineyard context, in relation with surrounding landscape and presence of prey (Collembola communities). Even if some additional informations are required, this paper represents an original progress for sustainable pest management in vineyards, due to its original approach, considering prey-predator relationship at a landscape scale. This manuscript also presents a few concerns that may have to be corrected before it becomes suitable for publication. Some references should be added in the introduction to give a better overview of results already obtained in a agricultural context, and in the discussion on the effect of surrounding landscape on spiders. On the other hand, additional explanations should be given about statistical analyses : some analysis are not mentionned, and standard error should be prefered to standard deviation. I therefore recommend a revision of the paper.

Authors: Thanks. We carefully revised the manuscript, added new references and provided more details on the statistical analyses.

L143. A part concerning univariate analyzes is missing

In the revised version of the manuscript we decided to remove original Figure 2 because we felt it showed redundant information already presented in the model analysis. To account for statistical differences between management option for Collembola activity density and diversity, we performed a Wilcoxon signed-rank test. This non-parametric statistical hypothesis test was chosen because we had two dependent samples from populations having the same distribution. This procedure is now described in the revised M&Ms.

L147. activity density ?

Thanks. We changed this accordingly.

L148. Keep the same vocabulary "activity density" or "abundance", and "taxon richness" or "diversity".

We now use the same terms throughout the manuscript.

L155. Define "AICc" and "AICmin".

Done. 

L157. activity density ?

Done.

L163. R²

Done.

L168. why species ? Families instead ?

Agreed, it should be called families, thanks for pointing this out.

L185. considering all families ?

In the revision we removed this information from the manuscript (see first comment).

L185. You should calculate standard error instead, to standardize by the number of replicates.

We removed this analysis in revised version. Please, see first comment.

L186. Which univariate analysis ?

Please, see first comment.

L189. To be explained in "2.3 statistical analyses" part. (In reference to Shannon index)

We added relevant information about the Shannon index.

L193. "The activity density"

Done.

L193. Indicate the mean values +- SE

Changed. 

L206-209. Which statistical analysis ?

Please, see first comment.

L213. Caption and asterisks should be written bigger.

Done.

L237. Captions are not readable.

Done.

L243. I don't understand how you can conclude that. To develop.

We rephrased this section and hope it is clearer now.

L237. Captions are not readable.

Captions have been enlarged.

Reviewer 2 Report

Overall review: This is an interesting manuscript, and addresses a relevant topic that merits study. However, I am largely concerned about the inconsistencies in results interpretation and presentation – the main Table 1 that overviews the models tested is not consistent with the results as written – they reference a certain number of “best” models in the text, but don’t bold all of these models in the Table, and particularly for spider diversity, the results as written and referenced in the abstract and discussion do not align with this Table as I viewed and interpreted it. For spider diversity, one of the best models is the null! I also provide some additional minor comments on literature gaps, ways to improve figures and better utilize the collected data, questions to seek clarity on the way methods were accomplished.

Abstract

I don't think the authors have the support to indicate that the reason they see some of these results is because of emigration from the field? This is speculation.

Introduction

Clearly written, but I found the literature cited for tillage effects (vs ground cover) on arthropods to be rather thin. The authors make it clear that work like this in vineyards is rare, but I don’t think that means not acknowledging other work studying tillage effects on spiders for example. Thorbek and Bilde (2004 Journal of Applied Ecology) is a rather classic study in this area, and Mary Barbercheck has also done quite a bit of work examining the impacts of tillage on soil arthropods including spiders and collembolans (Barbercheck et al 2009 Envt Monitoring & Assessment, Rivers et al. 2018, Jabbour et al. 2016 etc). Ann Rypstra has done a lot of work examining habitat complexity effects on spiders – also not included anywhere in this MS.

Methods

General question – the authors place 6 traps per site, I assume the trap numbers per site are summed together for analysis? Or are these subsamples are handled in analysis in another fashion? Please clarify in the methods.

121: I’m confused why springtails are being assessed with regards to pest control potential of grape berry moth? Is it because they are considered alternate prey for spiders and other generalist predators? Maybe spell that out here for readers.

148-149: “spider abundance and diversity covering all the possible combinations mentioned about” I don’t know what this is referring to – all the possible combinations mentioned above? This needs to be edited for clarity and comprehension for the reader.

152: I think AICs should be AICc here?

157-159: what is the “total springtail population” being referred to here?

In addition to richness and Shannon, authors may consider characterizing these communities according to evenness as this has been found to be important driver in some other agricultural studies (ie Crowder et al 2010). Also, I am generally curious to know more about the springtail communities – perhaps telling us not just about impacts on diversity and density but also on springtail community composition through multivariate analyses would be quite interesting, since the authors have already gone to all the trouble to identify them to species which is quite a lot of work!

Results:

Figures: For Figure 2 & 3, it would be more information if the authors can overlay the sample points on the graph – this is an option in R. Also, the colors in these graphs as is cannot be discerned by those with red-green color blindness, so the authors should consider changing this color scheme.

Figure 2: In A, Is this activity density per site? Or per trap?

Table 1. This table is not well-explained. It should be specified that in the columns “Total spider density” and “Total spider diversity” that these are the AICc values for each of these response variables. In the caption, Coll = Collembola density correct? Maybe specific that.

Also in Table 1 – should there be a 5^th model bolded for total spider density? Maybe 531.09? Because 5 “best” models are referenced in the text, but only 4 are bolded. Should a 4^th model be bolded for total spider diversity? Because 4 “best” models are referenced in the text but only 3 are bolded – maybe the 111.58?

Ln 230-233: I find this paragraph quite confusing because when I look at Table 1, the 4 lowest AICc values for total spider diversity models are for the null model, collembola only, mgmt. only, and mgmt + collembola….nothing involving SNE? So, not sure where all of the text in this paragraph comes from? Nor the justification for Figure 4C+D.

Figure 4 is a nice way to visualize the interaction for Figure 4A+B. Please be consistent with capitalization of axis labels “Total Spider activity density” should be “Total spider…”

Discussion

272-274: I don’t understand this conclusion based on the results. How can the authors compare how closely linked spiders are in this system to SNE than in other systems? They found that SNEs are important in their models for spider abundance.

278: I don’t think your models showed spider diversity increasing with SNE, as mentioned above.

Author Response

Pfingstmann et al.´s reply to comments of Reviewer 2

We are glad that Reviewer 2 found our study interesting and are grateful for the constructive suggestions on improving our manuscript. In the revision we addressed all points raised by Reviewer 2. Please find below a point-to-point reply. All changes on the original text have been highlighted with tracked changes. We would also like to thank for these constructive comments in the revised acknowledgements.

Reviewer 2

Overall review: This is an interesting manuscript, and addresses a relevant topic that merits study. However, I am largely concerned about the inconsistencies in results interpretation and presentation – the main Table 1 that overviews the models tested is not consistent with the results as written – they reference a certain number of “best” models in the text, but don’t bold all of these models in the Table, and particularly for spider diversity, the results as written and referenced in the abstract and discussion do not align with this Table as I viewed and interpreted it. For spider diversity, one of the best models is the null! I also provide some additional minor comments on literature gaps, ways to improve figures and better utilize the collected data, questions to seek clarity on the way methods were accomplished.

Authors: Thank you for the thorough reading of our manuscript. We carefully considered the concerns raised and have to admit that there were indeed some inconsistencies in the original manuscript. We are grateful that these aspects were pointed out by Reviewer 2 and apologize for the confusion this may have caused. 

Methods

General question – the authors place 6 traps per site, I assume the trap numbers per site are summed together for analysis? Or are these subsamples are handled in analysis in another fashion? Please clarify in the methods.

We decided not to pool samples since our intention was to perform generalized linear mixed models. This way we can set the vineyard as a random factor and consider all variability due to the different vineyards of the study within this factor. With this method we want to focus on the target variables of the study: local management, landscape and alternative prey availability. We changed M&Ms to clarify this.

121: I’m confused why springtails are being assessed with regards to pest control potential of grape berry moth? Is it because they are considered alternate prey for spiders and other generalist predators? Maybe spell that out here for readers.

Thanks for pointing this out. This aspect was indeed confusing. Actually, we wanted to study the effect that alternative preys can have on generalist predators such as spiders. We added a statement to clarify this. 

148-149: “spider abundance and diversity covering all the possible combinations mentioned about” I don’t know what this is referring to – all the possible combinations mentioned above? This needs to be edited for clarity and comprehension for the reader.

With this statement we wanted to say that all possible combinations of explanatory variables with their interactions were taken into account to build the model battery. We clarified this statement in the revised text.

152: I think AICs should be AICc here?

Agreed, thanks.

157-159: what is the “total springtail population” being referred to here?

Total springtail population refers to total springtail activity density. We clarified this statement. 

In addition to richness and Shannon, authors may consider characterizing these communities according to evenness as this has been found to be important driver in some other agricultural studies (ie Crowder et al 2010).

We are aware that evenness can be an important driver in biocontrol aspects. Indeed, we included it in our previous analysis. Nevertheless, the results obtained with this index were not significant and we decided to not present it in the manuscript. We made this decision because of a lack of significance and because the effect of evenness is on biocontrol but not on activity density of natural enemies. Since we did not assess any measurement of biocontrol success in our study we would prefer to keep the original parameters also in the revised manuscript.

Also, I am generally curious to know more about the springtail communities – perhaps telling us not just about impacts on diversity and density but also on springtail community composition through multivariate analyses would be quite interesting, since the authors have already gone to all the trouble to identify them to species which is quite a lot of work!

As the objective of this paper was to assess the effect that springtails can have on spider abundance we did not decide to include the analysis suggested by the reviewer on this work. We agree that this community analysis can be very interesting. Indeed, suggested analyses on springtails were reported in a previous work from this project (Buchholz et al. 2017 Soil biota in vineyards are more influenced by plants and soil quality than by tillage intensity or the surrounding landscape. Sci Rep 2017, 7, 17445, doi:10.1038/s41598-017-17601-w.).

Results:

Figures: For Figure 2 & 3, it would be more information if the authors can overlay the sample points on the graph – this is an option in R. Also, the colors in these graphs as is cannot be discerned by those with red-green color blindness, so the authors should consider changing this color scheme.

Thank you for these suggestions. After carefully considering the comments of Reviewer 1 and Reviewer 2 we decided to delete Figure 2 from the revised manuscript. Information presented in original Figure 2 was actually redundant with the information of the model analysis where is clearly stated that management has an effect on both spider activity density and spider diversity. We have rearranged the figure numbers in the text accordingly. We changed Figure 3 according to this comment.

We hope Reviewer 2 accepts our motivation for these changes. 

Figure 2: In A, Is this activity density per site? Or per trap?

Original Figure 2 has been removed from the revised manuscript. Please see comment above. 

Table 1. This table is not well-explained. It should be specified that in the columns “Total spider density” and “Total spider diversity” that these are the AICc values for each of these response variables. In the caption, Coll = Collembola density correct? Maybe specific that.

We have changed caption of table 1 according to the reviewer comment.

Also in Table 1 – should there be a 5th model bolded for total spider density? Maybe 531.09? Because 5 “best” models are referenced in the text, but only 4 are bolded. Should a 4th model be bolded for total spider diversity? Because 4 “best” models are referenced in the text but only 3 are bolded – maybe the 111.58?

Reviewer is correct, thanks for pointing this out and apologize for this error. The text is correct and there were two values of AICc that were not bolded. We changed this accordingly.

Ln 230-233: I find this paragraph quite confusing because when I look at Table 1, the 4 lowest AICc values for total spider diversity models are for the null model, collembola only, mgmt. only, and mgmt + collembola….nothing involving SNE? So, not sure where all of the text in this paragraph comes from? Nor the justification for Figure 4C+D.

Thanks again for your thorough reading and pointing out this error of ours. We made a mistake when transferring the AIC values to the table. The value of the model “Col” should be the value of the model “Mgmt + SNE + (Mgmt x SNE)“ and vice versa. We apologize for this error.

Figure 4 is a nice way to visualize the interaction for Figure 4A+B. Please be consistent with capitalization of axis labels “Total Spider activity density” should be “Total spider…”

Done.