Zoometric Characterization of Creole Cows from the Southern Amazon Region of Peru

Round 1

Reviewer 1 Report

General coment

dear authors,

the text has improved but the work is still confused and difficult to understand, especially for readers who are less familiar with the tools and strategies for measuring livestock biodiversity.

Some changes made make the manuscript inaccurate and misleading.

In particular, the statistical section should still be improved.

Specific comments

L 129-130: the dendrogram allows to evaluate the distance between samples and their eventual clusterization in homogeneous groups, but it does not describe the nature of the separation (ie on which variables the observed separation is based).

Rewrite the sentence.

L 130: if I understand correctly, you have computed the phenotype averages for each cluster, and compared these averages with an ANOVA for each trait.

If so, I would like to see a table with estimated averages and statistical differences.

L 134-136: Finally, you performed multivariate factors analysis (MFA), I suppose using the original zoometric indices.

Performing this analysis as validation of the three clusters found is an interesting and correct approach, but from what I see, it was not carried out in the best way.

The two tests mentioned have nothing to do with the distance identified by the clusters:

1) the Kaiser Measure of Sampling Adequacy (MSA) or even Kaiser-Meyer-Olkin (KMO), measures the difference between partial correlations and Pearson one.

A high value of this test (KMO> 0.6 or 0.8 depending on the authors) indicates the existence of a substructure (latent factor) that characterizes the data.

The biological meaning of this substructure can be interpreted by the weights (loadings) that each original variable takes in each extracted factor.

If you use this analysis, I expect to see tabulated the loadings of the extracted factors and a comment on them.

For example, in your case, Factor 2 separates cluster 2, but clusters 1 and 3 are no different.

While Factor 1 mostly separates clusters 1 and 3 with an intermediate position of cluster 2.

The analysis of the loadings of the two extracted factors would allow clarifying the nature of these separations (or equalities) in terms of original phenotypes in a multivariate context.

2) Regarding the Bartlett test, it only guarantees the homoscedasticity of the data, while to determine the distances between groups, the distances between centroids (multivariate means for each group) should be calculated, usually by Mahalanobis distance.

L 248: you write, here and elsewhere, about principal component analysis, attention because it is a very different analysis from MFA, the first studies the variance structure, the second the (CO)variance structure.

Tables 2, 3, and 4: I think that, with proper references in the text, these tables can be moved into the additional material.

Figure 3: What does "REGR factor score 1 for analysis 1" mean?

I see a plot of the animal scores in the two-dimensional space of the first two extracted factors.

It would be better to add (here or in the text) what amount of variance is explained by each factor.

Table 11: this is the most important and significant table of the whole work, as it indicates the real differences between the three identified groups.

My recommendation is to use the results of the analysis of variance (univariate analysis) and MFA (multivariate analysis) to underline (with scientific accuracy) what differentiates and what joins the three studied subpopulations.

If necessary by archiving assistance from an expert in statistics.

Additional material: The material presented in this section should also follow editorial and scientific rules.

Tables and figures should not be as they come out of the software,

ie table S1 in Spanish, figures S1 with the header at the top deformed and in Spanish.

Table S2: I think the percentage values are missing in Biotype 3 for the color pattern.

Author Response

Reviewer 1

First of all, thank you very much for the suggestions and comments to the manuscript, as they make it more succinct and allow us to reflect what we want to convey with our findings.

• Final del formulario

Principio del formulario

• L 129-130: the dendrogram allows to evaluate the distance between samples and their eventual clusterization in homogeneous groups, but it does not describe the nature of the separation (ie on which variables the observed separation is based).

Qualitative and quantitative traits were used to classify the animals, and they were graphed in a dendrogram after the cluster analysis. (L109-110). 

• L 130: if I understand correctly, you have computed the phenotype averages for each cluster, and compared these averages with an ANOVA for each trait.
• If so, I would like to see a table with estimated averages and statistical differences.

The averages were compared by ANOVA and Duncan's test for quantitative traits (Table 1, L156-157), and associations with qualitative traits using Chi-square test (Table 2 and 3, L174-176, L182-184).

• L 134-136: Finally, you performed multivariate factors analysis (MFA), I suppose using the original zoometric indices.
• Performing this analysis as validation of the three clusters found is an interesting and correct approach, but from what I see, it was not carried out in the best way. The two tests mentioned have nothing to do with the distance identified by the clusters: 1) the Kaiser Measure of Sampling Adequacy (MSA) or even Kaiser-Meyer-Olkin (KMO), measures the difference between partial correlations and Pearson one. A high value of this test (KMO> 0.6 or 0.8 depending on the authors) indicates the existence of a substructure (latent factor) that characterizes the data. The biological meaning of this substructure can be interpreted by the weights (loadings) that each original variable takes in each extracted factor. If you use this analysis, I expect to see tabulated the loadings of the extracted factors and a comment on them. For example, in your case, Factor 2 separates cluster 2, but clusters 1 and 3 are no different. While Factor 1 mostly separates clusters 1 and 3 with an intermediate position of cluster 2. The analysis of the loadings of the two extracted factors would allow clarifying the nature of these separations (or equalities) in terms of original phenotypes in a multivariate context. 2) Regarding the Bartlett test, it only guarantees the homoscedasticity of the data, while to determine the distances between groups, the distances between centroids (multivariate means for each group) should be calculated, usually by Mahalanobis distance.

The distances between individuals to establish the three clusters were carried out by Ward's algorithm and Mahalanobis distance. Then, multivariate factor analysis contributed to reducing the traits number in six components, offering a KMO (0.4) moderate value to identify population substructure. Accumulated variance table of these six components is presented (Table 4, L192-193) and explain 92.1%. Bartlett test (p <0.001) indicates that application of multivariate factorial model in this analysis, is suitable. This procedure is explained in subtitle 2.3. Statistical analysis (L108-122).

• L 248: you write, here and elsewhere, about principal component analysis, attention because it is a very different analysis from MFA, the first studies the variance structure, the second the (CO)variance structure.

To explain accumulative variance and reducing the traits number, a multivariate factor analysis was used. The SPSS program performs the extraction by principal components method and provides a correlation value of each trait for each component, and we classify the traits in the component that has the highest correlation value (L 204-205). 

• Tables 2, 3, and 4: I think that, with proper references in the text, these tables can be moved into the additional material.

 Tables 2, 3 and 4 were removed to supplementary material.

• Figure 3: What does "REGR factor score 1 for analysis 1" mean? I see a plot of the animal scores in the two-dimensional space of the first two extracted factors. It would be better to add (here or in the text) what amount of variance is explained by each factor.

 The accumulated explained variance table for each component is added (Table 4, L192-193).

• Table 11: this is the most important and significant table of the whole work, as it indicates the real differences between the three identified groups. My recommendation is to use the results of the analysis of variance (univariate analysis) and MFA (multivariate analysis) to underline (with scientific accuracy) what differentiates and what joins the three studied subpopulations.

The findings regarding the classification into biotypes, comparison of their averages and multivariate factor analysis of qualitative and quantitative traits were prioritized.

Reviewer 2 Report

First of all, I wish to thank the authors for their response to the remarks made in the previous review. The presentation of the paper have been reduced and better organized, in particular in the presentation of the methods and the results. 

However, the main comments remain valid. In particular:

• The choice of the Fleckvieh grid is surprising, and the FAO guidelines for the characterisation of the genetic resources could have been useful;
• As the impact of the ecological zone is not studied, the presentation of the geographical distribution (table 1 and figure 1(table 2, 3, 4) ) is not really useful. It could have been presented in the supplementary material
• Presentation of the variables appears labourious and the tables 2, 3, 4 are difficult to read
• The presentation of the results, in particular through the table 5 to 8, is a bit heavy. For table 5 and 7, it could be more interesting to present only one value (%, with only one decimal, rather the number of animals). For table 6 and 8, where binary traits (Y/N) are presented, only the presence of the defect could be presented (Y)
• The description of the effects of age and lactation is rather long and present little interest. If the authors want to present these results, maybe put the only the variables where significative effects in the text, and present the whole tables as supplementary material
• More interesting is the multivariate analysis and the description of the 3 biotypes. This could be the main body of the results and discussion.

Finally, I found that the english language merit a strong revision.

Author Response

Reviewer 2

Dear reviewer, thank you very much for your helpful advice that we believe has helped us a lot in the implementation of the manuscript. We have tried to satisfy all your requests, suggestions and have responded to your comment in the text and in this document.

• The choice of the Fleckvieh grid is surprising, and the FAO guidelines for the characterisation of the genetic resources could have been useful;

The suggestion is very valuable, we were guided by the FAO guidelines to characterize coat colors and coat coloring pattern; however, for animal traits we rely on the Fleckscore system because there are some traits that we measure in this research, that FAO guidelines does not detail.

• As the impact of the ecological zone is not studied, the presentation of the geographical distribution (table 1 and figure 1(table 2, 3, 4)) is not really useful. It could have been presented in the supplementary material.

Table 1, 2, 3 and 4, and Figure 1 have been removed to supplementary material.

• Presentation of the variables appears labourious and the tables 2, 3, 4 are difficult to read

The content of Table 2 was described in the corresponding section in materials and methods. Tables 2, 3 and 4 were removed to the supplementary material section.

• The presentation of the results, in particular through the table 5 to 8, is a bit heavy. For table 5 and 7, it could be more interesting to present only one value (%, with only one decimal, rather the number of animals). For table 6 and 8, where binary traits (Y/N) are presented, only the presence of the defect could be presented (Y)

Tables 5 to 8 were modified. Table 5 and 7 are presented with a single value and a single decimal. For Table 6 and 8 only the presence of defect is presented.

• The description of the effects of age and lactation is rather long and present little interest. If the authors want to present these results, maybe put the only the variables where significative effects in the text, and present the whole tables as supplementary material.

The description of age and lactation effects was removed at the end of the manuscript and supplementary material.

• More interesting is the multivariate analysis and the description of the 3 biotypes. This could be the main body of the results and discussion.

The text was modified and the description of biotypes is presented at the beginning of results section.

Reviewer 3 Report

Dear authors

Please see notes on manuscript and comments on first submission. A number of these were not addressed.

Comments for author File: Comments.pdf

Author Response

Reviewer 3

Dear reviewer, thank you very much for your helpful advice that we believe has helped us a lot in the implementation of the manuscript. We have tried to satisfy all your requests, suggestions and have responded to your comment in the text and in this document.

• How many cows/animals measured?

We measured 29 Creole cows in Chachapoyas, 37 in Luya and 29 in Bongará; all are provinces of the Amazon region, Peru (L 72 and 73).

• Please clarify "straight muscularity"

Straight muscularity is a qualitative trait of bovines that have a straight line through the vertebral column, as parallel as possible to the ground.

• What is the significance of these results - there will be age differences

Yes, according to ANOVA, rear udder length is greater in cows with six teeth and less in cows with eight teeth and a full mouth. They have good productive aptitude because they could have a good udder capacity for milk production, despite being younger than the other groups.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

General comments:

The manuscript presents the characterization of Creole cattle in three districts of Peru characterized by a tropical climate and altitudes above sea level varying between 1300 and 3000 meters, approximately.

The authors point out the importance of characterizing the local population in order to develop appropriate conservation strategies.

The presented research increases the scientific and technical knowledge on a population that has not yet been studied. In this sense, it is an important contribution to preserving the biodiversity of endangered local breeds.

In this regard, the manuscript acquires a significance that goes beyond the apparent local interest of the work.

In general, the paper is reasonably well structured, but I think it is unclear in some parts and could be improved.

The introduction defines the problem adequately, the goal of the work and the experimental design are consistent.

The materials and methods section is well structured, but the statistical section should be improved, for a better understanding of the results.

Particularly, Ward's method should be explained conceptually as it represents the main statistic for the subdivision of the population into the three identified types.

Results section is very hard to read, I think that tabulation of some reported data could make the text more clear.

Finally, I think it is impossible not to mention in the introduction and in the conclusion sections that morphological characterization should always be combined with molecular characterization and that only the two methods used in combination can give concrete indications for planning of conservation.

Specific comments:

L 48: Eding end Laval 1999 is not the most suitable bibliographic reference in this case,

The whole chapter is about genetic diversity and its measurement, as opposed to phenotypic diversity: “there is an enormous amount of diversity. This diversity we can observe and measure directly is phenotypic. On the other hand the largest part of diversity is hidden, because it is genetic diversity,”

L 59: these bibliographical references are general, not specific to the Peruvian creole cattle,

L 74- 75: these are results and should not be presented in this section, just indicate the aim of the work and the tools used to achieve it,

L 109: sarebbe meglio prima introdurre le misure di base effettuate (quelle riportate nella tabella supplementare),

L 119-133: perhaps it is better to tabulate the values for greater readability of the text,

L 250 – 266: distinctive traits of the three types identified is the most important part of the research, these data should be tabulated, while the text should explain better how they were obtained,

L 296: replace II with III.

Reviewer 2 Report

This communication deals with the morphometric description of a local Creole cattle "ecotype" from the Amazona region in Peru. The Peruvian Creole cattle has been little studied so far, and the study of the diversity of this local genetic resource deserve interest.

However it appears several limitations in the design of data collection, and their statistical analisis, and also in the discussion of the results:

• First of all, the number of animals (only female) is rather low, and appears insufficient to really study the population stratification of the Peruvian Creole cattle.
• Some differences in ecological zone are described in table 1; but the influence of this parameter, which could have been interesting to analyse, have not been evaluated
• Nevertheless, the geographical distribution of the population studied appears rather small. It could have been interesting to study more cattle from a widder range of ecological zone in Peru, to infer adaptive pattern of various phenotypes to various ecological zones.
• In material and methods, the description of the classification traits is rather long (about 3 pages). The coat colour description in 16 classes does not rely on any genetic basis of cattle coat, which is based on the type of melanine, its dilution and presence / absence of white spot or other modifiers (brindles,...).
• It should have been interesting to explain why the Fleckwieh linear scoring presents an interest (as a mixed - purpose breed, I suppose).
• From a statistical point of view, the muscularity score (in a range of 68 - 93) could have been analysed as a continuous variable. For categorical traits, the number of classes appears much too high to describe the distribution of the low number of animals sampled.
• In the results, a table presenting the main statistical results of the ANOVA analysis or the chi-square test should have been included firstly, to describe what are the more variable traits and their variation  factor.
• 5 pages are dedicated to an extensive description of the variability of the morphometric traits explained by age or lactation, even when those factors don't have any effect, and thus are not relevant to discuss.
• It could have been interesting to study the ecological zone or the biotype effect on the traits, but there were not included.
• The more interesting part, the multivariate analysis, is poorly described in material and methods; and their presentation in the results is a bit laborious. A table presenting the main difference between the "biotypes" described would have been better to explain these differences.
• Finally, the implication of the work for the conservation or the use of this valuable genetic resources could have been interesting to include in the discussion and conclusion. As well as the interest of an more extensive study at the country scale.
• In the introduction, the authors link the origin of Creole cattle in Peru with the arrival of Christopher Columbus, in 1493, which is not true. From various sources, if this historical episode is linked to the introduction of domestic animals in the Caribbean and the Central America, Colombia or Venezuela, it seems that later events have driven the introduction of cattle in the southern cone of Latin America, through Rio de la Plata and Argentina (Rodero, A. & Delgado, J. vicente & Rodero, E. Primitive Andalusian livestock an their implications in the discovery of America. Arch. Zootec. 41, 383–400 (1992)). A comparison of morphometric indices or colour type to other Creole cattle from the southern american cone could have been interesting to discuss

Reviewer 3 Report

Dear Authors

This could be an interesting article, but language editing is first required to make it more clear.

Material and methods are incomplete and maybe you can add more animals to the study. Consult the FAO guidelines for morpho metric traits for conservation and characterization.

regards 

Comments for author File: Comments.pdf