Comparison of Seed Images with Geometric Models, an Approach to the Morphology of Silene (Caryophyllaceae)

Round 1

Reviewer 1 Report

Dear Authors,

This is a very good and interesting subject in manuscript. I have added some minor changes as highlighter on the pdf file.

I am of the opinion that it is a valuable article for the Journal when it is published.

Best regards.

 

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

Thank you very much for reading the manuscript and for your positive comments, which together with the corrections made have helped to improve its quality.

We have corrected all the typos in the text indicated in the PDF, with the only exception of the word “reniformia”. We have checked again in the books of Boissier and Rohrbach and this is the way they use the adjective in reference to Silene seeds.

Please find attached the PDF with our answers to your questions.

With best regards,

Emilio Cervantes

Corresponding author

Author Response File: Author Response.pdf

Reviewer 2 Report

The methods developed by the authors in previous papers are applied to a further series of species. However, in comparison with the previous paper in Taxonomy, I do not find added value and I see three major methodological issues:

1.  First, the methods - the authors set out from cardioid originally - are getting more and more complex. A variety of "simple" models need to be added.  Fourier methods are included.

2. Using the "outline" from photographs in both views does not do justice to the seeds of Silene. In other taxonomy papers on Silene, SEM is used, which shows the real complexity of the seeds, and the variability.  

3. The authors started from the cardioid, which has a clear geometric meaning with all kinds of properties. This showed promise to understand teh geometry.  This is now completely lost.  The method is inbetween this geometrical approach and the morphometric approaches (e.g. with elliptic Fourier analysis).

4. It is is also unclear (and not discussed) how this method would contribute to the taxonomy and phylogeny of Silene, since they acknowledge more data are needed (which could lead to more complex models to be added). 

5. It is also unclear how this method would be comparable, better than, or opposite to findings in methods for  molecular phylogeny of Silene.

 

Author Response

Reviewer 2

Dear Reviewer,

Thank you very much for reading the manuscript and for your commentaries that have contributed to improve its quality. Your questions are answered in the following paragraphs.

The methods developed by the authors in previous papers are applied to a further series of species. However, in comparison with the previous paper in Taxonomy, I do not find added value and I see three major methodological issues:

1. First, the methods - the authors set out from cardioid originally - are getting more and more complex. A variety of "simple" models need to be added.  Fourier methods are included.

An original contribution of this article is that the method based on the comparison of seed silhouettes with geometric images has been applied to 51 new species from very diverse geographic origins, and results have been obtained for the J-index values corresponding to the lateral and dorsal views of the seeds for 27 species.

It is true that Silene seeds resemble the cardioid and the seeds of many species give high similarity values (J index) to the corresponding model. However, there is a drawback using the cardioid as the only model, namely that the figure is less discriminating than the combination of other figures derived from it. Thus, for the side views, we have two kinds of models derived from the cardioid: Those that are more convex (closed), and those being less convex (more open) than the cardioid. This is indicated in the introduction, as well as in the discussion. The reason for discarding the cardioid as a model in the final presentation of the results is stated at the end of the Materials and Methods section. It reads:

The results obtained with LM1 (the cardioid) were excluded because it is less discriminant than the other lateral models.

As for the dorsal views of the seeds, we have also closed (DM1 to DM4; DM10), open (DM7 to DM9; DM11-DM13) and intermediate (DM5 and DM6) models. The combination of these two sets of lateral and dorsal models gives a good description for the seeds of Silene species. Our aim is to establish the correlation between these models and to show their usefulness for Silene taxonomy. This aspect is now expanded in the last section of the results, entitled:

The Relationship between Geometric Models and Taxonomic Sections

It contains a dendrogram based on seed morphology.

In relation to the increased number of models and Fourier Transform methods, please see our response to question 3 below.

 

2. Using the "outline" from photographs in both views does not do justice to the seeds of Silene. In other taxonomy papers on Silene, SEM is used, which shows the real complexity of the seeds, and the variability.  

We agree that images obtained by scanning electron microscopy can be more informative, and indeed show in great detail the complexity of the seeds. However, classical photography, when applied with good equipment and quality objectives, also provides useful information about the surface and contours of the seeds. It is not our intention to propose classical photography as an alternative to SEM. We do not advocate it as a better or more complete technique, and consider it a complementary technique that can be particularly useful for the analysis of a larger number of seeds.

3. The authors started from the cardioid, which has a clear geometric meaning with all kinds of properties. This showed promise to understand the geometry.  This is now completely lost.  The method is inbetween this geometrical approach and the morphometric approaches (e.g. with elliptic Fourier analysis).

As indicated in the answer to the first question, starting from the cardioid as a single model, we applied seven additional lateral models and twelve dorsal models. There are species whose seeds fit relatively simple models well, while others need specific models. The Fourier Transform is of great help in obtaining models for the later. The application of Fourier analysis to obtain new models is a useful technique that has been developed in our group for this purpose. This particular aspect has been now expanded in the new section at the end of results and in the new added penultimate paragraph in the discussion section.

4. It is is also unclear (and not discussed) how this method would contribute to the taxonomy and phylogeny of Silene, since they acknowledge more data are needed (which could lead to more complex models to be added). 

A summary of the results obtained in this and other publications has been added as a dendrogram at the end of the results section. It shows that the method based on the comparison of seed silhouettes with geometric images classifies the seeds broadly in a similar way to methods based on DNA sequence analysis.

5. It is also unclear how this method would be comparable, better than, or opposite to findings in methods for molecular phylogeny of Silene.

The last two questions address important issues for this work which have been described in more detail in a new section at the end of results and are now illustrated with a dendrogram.

 

 

Reviewer 3 Report

This is an interesting and important paper on seed morphology of Silene, a taxonomically complex genus. The paper is well-prepared. I have only a few suggestions and comments.

It may be reasonable to mention why the seeds tend to be kidney-shaped in side view. Apparently, because of the campylotropous ovule.

When calculating the J index, an important point is minimizing differences between the seed silhouette and the model. Apparently, this should be related to model size and best fitting of relative positions of the silhouette and the model. Was any computer algorithm used here or the model was just manually adjusted to the silhouette? There may be a danger that the manual adjustment provided a kind of systematic error.

I suggest summarizing a total knowledge on seed geometry of Silene acquired using these quantitative methods onto molecular phylogenetic tree. Such an additional figure will be highly useful.

Figure 18: the best model of the lateral view selected here looks like suboptimal.

Minor suggestions:

Line 38: closed -> closely

Line 41:  current -> currently

Lines 142-145: Latin names should be in italics.

Line 149: apparently the whole subtitle should be in italics?

Line 180, 185, 186, 190, 199, 209, 212, 221 etc.: italics.

Line 501: Complex -> complex

Author Response

Dear Reviewer,

Thank you very much for reading the manuscript and for your commentaries which have helped to improve its quality. Your questions are answered in the following paragraphs.

 

It may be reasonable to mention why the seeds tend to be kidney-shaped in side view. Apparently, because of the campylotropous ovule.

This has been indicated in a new paragraph, the penultimate now in the discussion section. It reads:

The cardioid (reniform or kidney-shaped) shape of the seeds is associated with the campylotropous ovule, with most of the morphological axis curved, allowing a great morphological variety in ovules and seeds [24], which is associated with various types of asymmetries [14]. 

When calculating the J index, an important point is minimizing differences between the seed silhouette and the model. Apparently, this should be related to model size and best fitting of relative positions of the silhouette and the model. Was any computer algorithm used here or the model was just manually adjusted to the silhouette? There may be a danger that the manual adjustment provided a kind of systematic error.

 

The model is adjusted to the silhouette manually. Indeed, there is a systematic error that we have estimated to be around 1% in the measurements with seed samples that fit well to the models (J index superior to 90). This value increases in samples containing heterogeneous seeds (the adjustment is more irregular).

I suggest summarizing a total knowledge on seed geometry of Silene acquired using these quantitative methods onto molecular phylogenetic tree. Such an additional figure will be highly useful.

A dendrogram has been made based on data on the comparison of seed images with lateral and dorsal models for the seeds of 40 species and is shown in Figure 21 and commented at the end of the results.

Figure 18: the best model of the lateral view selected here looks like suboptimal.

We agree that there were only partial similarity between the model and the seed images shown in the previous image. The new version of this figure shows alternative seed images of this population that more closely resemble the models.

The minor suggestions indicated have been corrected.

Round 2

Reviewer 2 Report

Thanks for the improved paper.  The new section on taxonomy and dendrogram are very useful, but must be improved for various reasons.

In M&M reference is made to a matrix for certain models; Should be in Supplementary material

In Table 1, species used in this study, there are 53 entries.  In the dendrogram there are 42 species. However I find that about 18 or 19 of the species in the dendrogram cannot be found in Table 1.  Only 25 of the 52 species in Table 1 are found in the dendrogram.

This can obviously lead to distortions. In particular In the Silene group only about 8 species are found in Table 1. 

The paper has improved but this needs to be addressed in the proper way. And the results or discussion has to be extended to address this all.

 

Author Response

Dear reviewer,

Thank you very much for your commentaries concerning the dendrogram that have helped to improve the quality of the article.

A new version of the dendrogram has been added which contains all the species listed in Table 1, without any additional species.

The data used to develop the dendrogram are now included in Table A7 in the Appendix.

The results section has been expanded with comments on the dendrogram, and these aspects are also indicated in the Conclusions section.

Looking forward to your answer. Yours sincerely,