Effectiveness of Seed Traps for Assessing Seed Rain in Periurban Grasslands
Katalin Szitár
Round 1
Reviewer 1 Report
This study explores an interesting subject, and its results bring useful clues that other investigators can apply in future research. Nevertheless, I have some suggestions for the authors:
Line 112: how many traps were placed on each station?
Table 1: the IDs for the funnel traps located at 0 cm and 70 cm are wrong.
Line 174: why did you choose to compare funnel traps with 45° tilted sticky traps?
Line 179: quantitative tests are more appropriate to test normality and homoscedasticity.
Figure 2 (Line 194) and others: please consider using not-transformed data to describe the density values.
Line 254: please reconsider some generalizations. Sticky traps at 45 cm did not show significant differences from the others, so it is incorrect to say that all sticky traps had higher richness. Taking this into consideration, this sentence: “The total species richness of anemochorous seeds and the species richness of anemochorous species from locally absent species in the standing vegetation was higher in sticky traps (whatever height) than in funnel traps (although there was no significant difference in total species richness in anemochorous seeds from funnel traps set at 0 cm).” could be changed to something like “The total species richness of anemochorous seeds and the species richness of anemochorous species from locally absent species in the standing vegetation WAS higher in sticky traps at 45 cm. Also, the first was lower in elevated funnel traps, while the second was lower in all funnel traps (no matter their elevation).”
Line 251: Density or richness?
Line 260: Please be consistent with the letters that are used for higher and lower treatments, because in some boxplots “A” is the highest mean value, and the same letter is used for the lowest mean in others.
Line 271: Supplementary Material Table S3 was not sent to me.
Line 286: “Contrary to the literature, we found that the most urbanized stations (S1 and S2) had the highest species richness in seeds either locally absent in the standing vegetation or anemochorous.” According to your results, this is only correct for anemochorous seeds from species locally absent in the standing vegetation. You can generalize for S1, but not for S1 and S2 together.
Line 289: “For example, S1 (like S7 and S5) had a relatively high proportion of woody surfaces in its surroundings, which could lead to a different and more diverse seed rain composition.” I think that this seems a bit of a stretch.
Line 344: diversity and richness are related, but not the same. Diversity includes richness and specific abundance, and it was not addressed in this study. Sticky traps were more effective in capturing species richness, but not capturing seed density, and this might affect diversity results.
Line 353: Or maybe the other shapes of sticky traps capture fewer seeds because they were not tilted.
Line 70: add an “i” in “ncluding”
Line 194: change “variability” for “variation”
Line 222: This should be Figure 3.
Line 234: This should be Figure 4.
Line 245: The using of the parenthesis is confusing, please rephrase this sentence for more clarity. It says “45 cm” but it should not be “25 cm”?
L 253: Consider changing the sentence from this: “Richness of seeds from locally absent species in the standing vegetation was higher in sticky traps at a height of 70 cm (and at 45 cm, although not significantly) than in all funnel traps (all heights combined).” to this: “Richness of seeds from locally absent species in the standing vegetation was higher in sticky traps at a height of 70 cm than in almost all funnel traps (all heights combined), except at 45 cm.”
Line 260: This should be Figure 5.
Line 274: This should be Figure 6.
Line 288: The sentence that starts whit “However, …” is confusing.
Supplementary Material S2: line 46 says “DENISTIES”. On the other hand, the first column of this table is wrong, as it should be referring to seed density, not seed richness. And finally, the second table of this file repeats the information of the first.
Appendix A: Lapsana and Polygonum have different fonts that the rest of the genus.
Author Response
We thank the reviewers for their thoughtful review of our manuscript, your comments have enabled us to greatly improve this article. You will find our answers to your comments here. We have also made some minor changes to the abstract. You will find our revised version of the manuscript in the attached file.
Line 112: how many traps were placed on each station?
Five traps were used on each station, we specified it in the sentence L114 to “. In each station, five traps were positioned in a line, and spaced 50 cm apart. No mowing was performed at the stations during the study.”.
Table 1: the IDs for the funnel traps located at 0 cm and 70 cm are wrong.
You are right, thank you for noticing it. It is now corrected in table 1
Line 174: why did you choose to compare funnel traps with 45° tilted sticky traps?
In order to analyze the effect of trap type and height, we choose to compare funnel traps with the most common sticky trap tilt which is 45°. However, due to your question and the other reviewer recommendation, the first analysis of trap characteristics is now focused on height, trap type and tilt. Thus the analysis is now on funnel traps and sticky square traps with 45° and 90° tilt. The method and results part are updated to fit these changes.
Line 179: quantitative tests are more appropriate to test normality and homoscedasticity.
You are right. We did have a significant quantitative test for normality for traps characteristics analysis although residuals seem to fit visually. So we change ANOVA for nonparametric Kurskal Wallis test on these analyses. We updated methods and results (with figure 5, 6 and 7).
Figure 2 (Line 194) and others: please consider using not-transformed data to describe the density values.
We prefer to represent the data with the same transformation that we used for statistical analysis, which makes it easier to compare low values.
Line 254: please reconsider some generalizations. Sticky traps at 45 cm did not show significant differences from the others, so it is incorrect to say that all sticky traps had higher richness. Taking this into consideration, this sentence: “The total species richness of anemochorous seeds and the species richness of anemochorous species from locally absent species in the standing vegetation was higher in sticky traps (whatever height) than in funnel traps (although there was no significant difference in total species richness in anemochorous seeds from funnel traps set at 0 cm).” could be changed to something like “The total species richness of anemochorous seeds and the species richness of anemochorous species from locally absent species in the standing vegetation WAS higher in sticky traps at 45 cm. Also, the first was lower in elevated funnel traps, while the second was lower in all funnel traps (no matter their elevation).”
You are right, we were not precise enough with the results description. Since we changed the analysis, some p values have been modified. We rewrote the results with more precision.
Line 251: Density or richness?
Thank you for noticing this mistake. It is indeed Richness. The paragraph has been rewritten without this mistake.
Line 260: Please be consistent with the letters that are used for higher and lower treatments, because in some boxplots “A” is the highest mean value, and the same letter is used for the lowest mean in others.
Thank you. Figure 5 is corrected A is now always the highest value
Line 271: Supplementary Material Table S3 was not sent to me.
Sorry, this should be solved.
Line 286: “Contrary to the literature, we found that the most urbanized stations (S1 and S2) had the highest species richness in seeds either locally absent in the standing vegetation or anemochorous.” According to your results, this is only correct for anemochorous seeds from species locally absent in the standing vegetation. You can generalize for S1, but not for S1 and S2 together.
You are right. Sentence modified L 315 to: “Contrary to the literature, we found that the most urbanized stations (S1 and S2) had the highest species richness in seeds either locally absent in the standing vegetation or anemochorous for S1 only”
Line 289: “For example, S1 (like S7 and S5) had a relatively high proportion of woody surfaces in its surroundings, which could lead to a different and more diverse seed rain composition.” I think that this seems a bit of a stretch.
It is indeed. As we have added a canonical correspondence analysis in this new version of the manuscrit, the effects of the surrounding landscape have been more precisely defined. So we replace this sentence by “In addition, the CCA shows a strong effect of built surfaces on seed rain composition. This finding demonstrates the need to properly study urban landscape influence on seed rain at larger scale.” L317
Line 344: diversity and richness are related, but not the same. Diversity includes richness and specific abundance, and it was not addressed in this study. Sticky traps were more effective in capturing species richness, but not capturing seed density, and this might affect diversity results.
You are right, we changed diversity for “richness” L376.
Line 353: Or maybe the other shapes of sticky traps capture fewer seeds because they were not tilted.
It is a relevant comment. However, both cones and hemisphere are fixed at 90° to the plot but, due to their shape, the capturing surfaces are not perpendicular to the ground (expected for the top of hemispheres). Furthermore, these traps can capture seeds at 360°.
Comments on the Quality of English Language
Line 70: add an “i” in “ncluding”
Thank you for noticing. Corrected
Line 194: change “variability” for “variation”
Indeed. Modified
Line 222: This should be Figure 3.
You are right, figures have been misnumbered. Corrected
Line 234: This should be Figure 4.
Corrected
Line 245: The using of the parenthesis is confusing, please rephrase this sentence for more clarity. It says “45 cm” but it should not be “25 cm”?
We understand this recommendation. We rephrased the sentence by: “After excluding Agrostis seeds, total seed density in funnel traps set at a height of 0 cm was found to be higher than for other trap types and heights, excepted 25 cm funnel traps.” L258.
L 253: Consider changing the sentence from this: “Richness of seeds from locally absent species in the standing vegetation was higher in sticky traps at a height of 70 cm (and at 45 cm, although not significantly) than in all funnel traps (all heights combined).” to this: “Richness of seeds from locally absent species in the standing vegetation was higher in sticky traps at a height of 70 cm than in almost all funnel traps (all heights combined), except at 45 cm.”
Thank you for this recommendation. Since we changed ANOVA to Kruskal Wallis. Some p value may have been modified. We have been careful not to repeat the same inaccuracies in the description of results.
Line 260: This should be Figure 5. Corrected
Line 274: This should be Figure 6. Corrected
Line 288: The sentence that starts whit “However, …” is confusing.
It is indeed. As we have added a canonical correspondence analysis in this new version of the manuscrit, the effects of the surrounding landscape have been more precisely defined. So we replace this sentence by “In addition, the CCA shows a strong effect of built surfaces on seed rain composition. This finding demonstrates the need to properly study urban landscape influence on seed rain at larger scale.” L317
Supplementary Material S2: line 46 says “DENISTIES”. On the other hand, the first column of this table is wrong, as it should be referring to seed density, not seed richness. And finally, the second table of this file repeats the information of the first.
Thank you for noticing. Corrected
Appendix A: Lapsana and Polygonum have different fonts that the rest of the genus.
Thank you for noticing. It is corrected.
Thank you again,
Best regards,
Authors
Author Response File: 
Author Response.docx
Reviewer 2 Report
Review of the manuscript ’Effectiveness of seed traps for assessing seed rain in periurban grasslands’
The manuscript deals with seed trap effectiveness in urban environments. The topic is timely with the rapidly growing attention to urban environments and restoration/regeneration potential based on plant species dispersal. I found it interesting, as only a few studies assess the dispersal potential of many plant species in urban areas. The manuscript is written in a good and enjoyable style.
The manuscript has two aims: looking for spatial variation in seed rain using funnel traps and assessing the effectiveness of the two types of traps, their height, and other characteristics.
The approach of assessing locally absent and wind-dispersed species is appropriate, as it may show longer distance dispersal, which is very relevant for estimating vegetation regeneration potential. Here I suggest showing more explicitly the effect of the stations' landscape composition in the ordination using redundancy analysis.
I think the second part comparing trap characteristics needs a bit more editing. I would make separate analyses for sticky trap plate size and shape and be more accurate when presenting the results. This means that we can state that there is a difference between two trap categories only when we find statistical differences. Now the text is more permissive. Overall, the manuscript can be published in Diversity after adequately addressing the following points.
Introduction
L48 I think that one of the merit and originality of the manuscript is that it was performed in an urban environment. The urban context, its specialities and its importance could be introduced in the Introduction part, too.
L84 Please reformulate Question 1 to comply better with the design, as it is only about funnel traps (and not sticky traps).
Methods
Fig.1C Please change the order of the columns to start with S1.
Table 1. Please use the ’no’ or ’-’ sign to note no tilt instead of ’/’. I also suggest removing the last column with the replicate numbers.
L109 Were the funnel traps placed at the ground level?
L171 Was any p adjustment used? It would be nice if not, as many post hoc tests were implemented. For details, see, for example, Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, 57(1), 289-300. Please also give an R package citation for ANOVA and Tukey post hoc tests.
L173 Why were sticky plates with the 90º tilt not tested in the first ANOVA together with funnel traps and sticky plates with 45º? It would have been logical as the two most common tilts are 45 and 90º.
L174 It would be nice to differentiate between the trap characteristics in the tests. I would test plate size (S25_45 vs Ssmall) and plate shape (S25_45 vs Scone and Ssphere) separately.
L177 The R version citation would be even more important than citing R studio.
Results
L181 Please refer to Appendix A here.
L193 Please renumber the figures as there are two of Fig. 2.
Fig. 2-4. Please consider using the original scale for density data (for Fig. 3 and 4 for sure) but maybe for Fig. 2, too instead of log scale. I do not understand why you stated to use square root scale for the representation. It is true (the text on the Y-axis says otherwise), and if yes, please explain.
Please be more accurate with the results.
L186 For example, it is not true that S6-7 had higher total seed densities than S1-2 because they were not statistically different.
L189 Total species richness in S1 was higher only (based on the post hoc tests) when compared to S3 and S5. Locally absent species richness was the highest in S1 – this piece of information could be included in the results.
L221 I suggest using Detrended Correspondance Analysis instead of simple CA to avoid the artificial arch effect, as it can be seen in Fig. 1 and the compression of the points at the two ends of the gradient. As it is artificial, the separation of S6 seems to be invalid (L216).
Please use the same colours for S1-7 in the figures (ANOVA and ordination). The symbols could be larger for the stations in the ordination graphs, as well, for better visibility.
Another idea could be using RDA analysis with the possibility to see the effects of impervious surface or natural habitat ratio (used as environmental variables) on the species composition caught in the traps.
L251 F25 is not different from F0 based on the statistics.
L251-259 How do you deal with the higher area of sticky traps when explaining differences with funnel traps? I find it relevant to compare sticky and funnel traps at the same heights. Based on this, species richness (for all four species groups) is higher in sticky traps compared to funnel traps at 70 cm and only at that height.
L259 It would be nice to check for zero inflation in the case of total and locally absent anemo_richness, as the graph shows a lot of zeros in the dataset.
L267-270 Small square sticky trap effectiveness should be compared only to a sticky trap with the same tilt and height. The same applies to the different trap shapes.
Discussion and Conclusion
L306-312 I would delete it as it is a repetition for Introduction L40-52, not giving direct connections to the results.
L328 In this subpart of the Discussion, please place your results into context based on other literature sources next to the review of Chabrerie and Allard (2005).
L330-331 and L366-367 seem to contradict each other.
L342 Write ’lower total species richness’ instead of ’lower species richness’ as it was only true for this variable.
L344-346 This is not true because although the mean richness of all species and locally absent species were higher, they were not statistically different from the means for sticky traps at the height of 25 cm.
Author Response
We thank the reviewers for their thoughtful review of our manuscript, your comments have enabled us to greatly improve this article. You will find our answers to your comments here. We have also made some minor changes to the abstract. You will find the modified version of the manuscript in the attached file.
Introduction
L48 I think that one of the merit and originality of the manuscript is that it was performed in an urban environment. The urban context, its specialities and its importance could be introduced in the Introduction part, too.
Thank you for this relevant comment, we add L50 : ”However, contrary to other human dominated landscapes, cities are characterized by a high heterogeneity at many scales [25] which can induce different ecological responses compared to other landscapes. In addition, understanding how landscape composition and configuration affect seed flows is therefore a prerequisite to addressing biodiversity issues in cities [19]. “
L84 Please reformulate Question 1 to comply better with the design, as it is only about funnel traps (and not sticky traps).
We have modified objective 1 : L85 “Our aim is therefore (1) to investigate the effectiveness of seed traps in capturing variations in seed rain (composition, density and richness) in herbaceous habitats located in different landscape contexts using funnel traps”.
Methods
Fig.1C Please change the order of the columns to start with S1.
You are right, this order is easier to read. Figure 1C has been change in order to strat with S1
Table 1. Please use the ’no’ or ’-’ sign to note no tilt instead of ’/’. I also suggest removing the last column with the replicate numbers.
Modified, “/” have been changed by “-”.
L109 Were the funnel traps placed at the ground level?
Yes indeed. It was not very clear so we added L109 “placed at ground level”.
L171 Was any p adjustment used? It would be nice if not, as many post hoc tests were implemented. For details, see, for example, Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, 57(1), 289-300. Please also give an R package citation for ANOVA and Tukey post hoc tests.
Indeed, we did use a p adjustment. In this new version we used the Benjamini and Hochberg procedure for both the Tukey test and the Dunn test. It is now precisedin the Data analysis method subpart.
L173 Why were sticky plates with the 90º tilt not tested in the first ANOVA together with funnel traps and sticky plates with 45º? It would have been logical as the two most common tilts are 45 and 90º.
This is a very relevent comment. Following this recommendation, we now analyze funnel traps in comparison with all sticky square traps (45 and 90° tilt). A second analysis compare S25_45 and Ssmall. And finally, all forms are compared to square sticky traps. Figure 5, 6 and 7 have been updated, as long as methods, results and supplementary tables.
L174 It would be nice to differentiate between the trap characteristics in the tests. I would test plate size (S25_45 vs Ssmall) and plate shape (S25_45 vs Scone and Ssphere) separately.
This is a very useful comment. Following this recommendation, we now analyze funnel traps with all sticky square traps (45 and 90° tilt). A second analysis is S25_45 and Ssmall. And finally, all forms compared to square sticky traps. Figure 5, 6 and 7 have been updated, as long as methods, results and supplementary tables.
L177 The R version citation would be even more important than citing R studio.
Absolutely, we modified it L193 “R (version 4.3.0) “
Results
L181 Please refer to Appendix A here.
You are right. We now refer to appendix A here. L197 : “A total of 3279 seeds were found, corresponding to 57 taxa (23 genus and 24 species). The genus Agrostis accounted for a third of the seeds collected (Appendix A).”
L193 Please renumber the figures as there are two of Fig. 2.
You are right, figures have been misnumbered. Corrected
Fig. 2-4. Please consider using the original scale for density data (for Fig. 3 and 4 for sure) but maybe for Fig. 2, too instead of log scale. I do not understand why you stated to use square root scale for the representation. It is true (the text on the Y-axis says otherwise), and if yes, please explai
We used this scale in order to have the same scale on the four density boxplot to allow visual comparison. We used root square scale to stretch up the densities (that are log transformed) for locally absent species which are between 0 and 0.1 for more clarity.
Please be more accurate with the results.
L186 For example, it is not true that S6-7 had higher total seed densities than S1-2 because they were not statistically different.
It is true, we modified L193 : “S6 and S7 presented higher total seed density taking all species together compared to S3, S4 and S5, while S2 presented higher total density of anemochorous seeds.”.
L189 Total species richness in S1 was higher only (based on the post hoc tests) when compared to S3 and S5. Locally absent species richness was the highest in S1 – this piece of information could be included in the results.
We changed L196 “Total species richness was higher in S1 compared to S3 and S5. Total anemochorous species rich-ness was also higher in S1 compared to all stations except S2.“.
L221 I suggest using Detrended Correspondance Analysis instead of simple CA to avoid the artificial arch effect, as it can be seen in Fig. 1 and the compression of the points at the two ends of the gradient. As it is artificial, the separation of S6 seems to be invalid (L216).
We agree that an arch effect is present in this analysis, reflecting the existence of a major gradient in the data set. However, the first objective of the analysis is to compare the position of the sampling units within and between the stations, and we prefer not to apply a segmental detrending method .In response to your comments, we have decided to analyze this gradient using a separate analysis (see comments below).
Please use the same colours for S1-7 in the figures (ANOVA and ordination). The symbols could be larger for the stations in the ordination graphs, as well, for better visibility.
It is indeed very important. We change boxplot colors to fit CA colors
Another idea could be using RDA analysis with the possibility to see the effects of impervious surface or natural habitat ratio (used as environmental variables) on the species composition caught in the traps.
Thank you for this relevant recommendation. We have added in this version of the manuscript a CCA in order to check for the gradient influence on seed composition. We find the results interesting and plan to continue this analysis on a larger landscape scale in a follow-up paper.. We rather used a CCA than a RDA to analyze the relations with an unimodal metric.
L251 F25 is not different from F0 based on the statistics.
Thank you for this recommendation. Since we have changed the ANOVA to Kruskal Wallis tests, some p value may have been modified. We have been careful not to repeat the same inaccuracies in the description of results.
L251-259 How do you deal with the higher area of sticky traps when explaining differences with funnel traps? I find it relevant to compare sticky and funnel traps at the same heights. Based on this, species richness (for all four species groups) is higher in sticky traps compared to funnel traps at 70 cm and only at that height.
Thank you for this interesting comment. We choose to study densities to take into account area surfaces differences between traps and we choose not to take it into account for richness. However, Jackel, A.-K. and Poschlod, P. 1994 have shown that area surfaces do not affect the species richness of seeds captured.
L259 It would be nice to check for zero inflation in the case of total and locally absent anemo_richness, as the graph shows a lot of zeros in the dataset.
Indeed, we tried to use Zero inflation but it is hard to apply to our datasets. However, for trap characteristics comparison only, we change the ANOVA to Kruskal Wallis to take into account the non-normality distribution of our data.
L267-270 Small square sticky trap effectiveness should be compared only to a sticky trap with the same tilt and height. The same applies to the different trap shapes.
This is a very useful comment. Following this recommendation, we now analyze funnel traps with all sticky square traps (45 and 90° tilt). A second analysis is S25_45 and Ssmall. And finally, all forms compared to square sticky traps. Figure 5, 6 and 7 have been updated, as long as methods, results and supplementary tables.
Discussion and Conclusion
L306-312 I would delete it as it is a repetition for Introduction L40-52, not giving direct connections to the results.
You are right, it is deleted.
L328 In this subpart of the Discussion, please place your results into context based on other literature sources next to the review of Chabrerie and Allard (2005).
We understand your comments and have again looked for relevant references to discuss our results. We have added a reference to the study of Jackel, A.-K. and Poschlod, P. 1994. . We found that very few studies used different seed traps and only Chabrerie and Allard and Arruda et al. analyze the difference in effectiveness between seed traps. However, we precise : “ Very few information on how traps characteristics affect seed rain assessment are available and none of it studied urban landscape” L361.
L330-331 and L366-367 seem to contradict each other.
Thank you, it is indeed. We change the conclusion to : “Sticky traps with a 45° tilt at a height of 70 cm were the most effective here in capturing both density and species richness, when focusing on locally absent species” L368. In this way, although total seed density is higher in funnel traps, as studies need to capture information about locally absent species, sticky traps are more efficient.
L342 Write ’lower total species richness’ instead of ’lower species richness’ as it was only true for this variable.
This is true. We add “total L369.
L344-346 This is not true because although the mean richness of all species and locally absent species were higher, they were not statistically different from the means for sticky traps at the height of 25 cm.
This is true. Since we changed ANOVA to Kruskal Wallis, some p value may have been modified. We have been careful not to repeat the same inaccuracies “In addition, the 70 cm high sticky traps captured the highest in seeds from outside the sta-tion and locally absent anemochorous seeds excepted with S25_90 and S25_45 and F0. “ L371.
Thank you again,
Sincerly,
Authors.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Thank you for your detailed responses, and the changes made following my suggestions. I have no further comment, and suggest publishing the ms in Diversity.
