Using Genomics to Measure Phenomics: Repeatability of Bull Prolificacy in Multiple-Bull Pastures

Round 1

Reviewer 1 Report

The manuscript by Gary L Bennett and co-authors showcases a study leveraging genomic data to accurately estimate bull prolificacy in multiple-sire mating situations ie. potentially non-pedigreed scenarios. Two objectives are clearly stated and addressed successfully: to estimate the repeatability of bull prolificacy and to explore the potential of genotyping pools of DNA from potential offspring to estimate the prolificacy of bulls. The topic falls within the scope of the journal Agriculture and I believe this work could be relevant to both the readers specialized on beef cattle breeding and genomics, and also to a wider audience with interest on advance methodology for the efficient and effective analysis of genotype datasets in species of agricultural relevance. The claims they make for how the method may be applied are not overstating its usefulness.  In addition, the manuscript is exceptionally well written using a professional yet vibrant writing style allowing for an enjoyable reading.

I have three concerns for the authors and editors to consider:

1. I appreciate that for most of their arguments about bull performance in multi-sire mating systems, the authors provide one reference from the US and one from Australia. However, when it comes to highlighting the power of DNA pooling in genomic selection only two references are given (14 and 15) both from a combination of the same authors. A the very least, the authors should be aware two more publications within the context of multiple-sire mating and DNA pooling of unpedigreed cows (https://pubmed.ncbi.nlm.nih.gov/27898866/) and progeny (https://pubmed.ncbi.nlm.nih.gov/31710679/).
2. Since they are talking about a “phenotype”, not a management tool, perhaps they could add a calculation (like the one provided in the second reference above), that shows how much genotyping would need to be done to generate a reference population for genomic prediction of the trait.
3. I found the literal narrative describing the analytical methodology to estimate sire proportions in each pasture and hence prolificacy (lines 157-206) quite dense and perhaps a step by step algorithm could assist. That could reflect my limitation preferring equations to words in these situations. However, my question is how does this method compare to simply creating a genomic relationship with both individual sires and pools of progeny and then estimating the prolificacy of bulls based on the genomic relationship between a bull and the pools (of course weighted by the size of the pool and adjusted to the total number of animals)?  

Author Response

The manuscript by Gary L Bennett and co-authors showcases a study leveraging genomic data to accurately estimate bull prolificacy in multiple-sire mating situations ie. potentially non-pedigreed scenarios. Two objectives are clearly stated and addressed successfully: to estimate the repeatability of bull prolificacy and to explore the potential of genotyping pools of DNA from potential offspring to estimate the prolificacy of bulls. The topic falls within the scope of the journal Agriculture and I believe this work could be relevant to both the readers specialized on beef cattle breeding and genomics, and also to a wider audience with interest on advance methodology for the efficient and effective analysis of genotype datasets in species of agricultural relevance. The claims they make for how the method may be applied are not overstating its usefulness.  In addition, the manuscript is exceptionally well written using a professional yet vibrant writing style allowing for an enjoyable reading.

I have three concerns for the authors and editors to consider:

1. I appreciate that for most of their arguments about bull performance in multi-sire mating systems, the authors provide one reference from the US and one from Australia. However, when it comes to highlighting the power of DNA pooling in genomic selection only two references are given (14 and 15) both from a combination of the same authors. A the very least, the authors should be aware two more publications within the context of multiple-sire mating and DNA pooling of unpedigreed cows (https://pubmed.ncbi.nlm.nih.gov/27898866/) and progeny (https://pubmed.ncbi.nlm.nih.gov/31710679/).  Response:  We agree.  We have added these 2 articles to literature cited and appreciate the suggestions.
2. Since they are talking about a “phenotype”, not a management tool, perhaps they could add a calculation (like the one provided in the second reference above), that shows how much genotyping would need to be done to generate a reference population for genomic prediction of the trait.  Response: This is an excellent suggestion which is worthy of future work both by us and others.  However, we think the analysis Reviewer 1 suggests would be complicated and it goes beyond the scope of this work which is to show that bull prolificacy in multi-bull breeding pastures is repeatable and that the cost of phenotyping bull prolificacy can be reduced by pooling without sacrificing much in accuracy.
3. I found the literal narrative describing the analytical methodology to estimate sire proportions in each pasture and hence prolificacy (lines 157-206) quite dense and perhaps a step by step algorithm could assist. That could reflect my limitation preferring equations to words in these situations. However, my question is how does this method compare to simply creating a genomic relationship with both individual sires and pools of progeny and then estimating the prolificacy of bulls based on the genomic relationship between a bull and the pools (of course weighted by the size of the pool and adjusted to the total number of animals)?   Response:  First question, we agree and have substituted the quadratic programming description with a R script added to supplementary files.  In addition, to eliminating a difficult to read and comprehend section it provides an R function that readers can use to do the computations themselves on their own data.  Second question, we agree that the analysis that the reviewer suggests would be useful and worthwhile but that it goes well beyond the scope of the manuscript and would not be simple.

Reviewer 2 Report

The manuscript used real genomic data to estimate bull's prolificacy and later conducted simulations to evalute the reliability of estimated parameters. The obtained results provide genomic evidence for practical use of multiple-bull breeding pastures.  Minor changes are needed.

1. Lines 28-29, "Beyond the genome to phenome paradigm, some phenotypes used for management or inputs to genetic evaluation may be best estimated by genomics." It would be great to be more specific about some phenotypes. The sentences can be changed to "Beyond the genome to phenome paradigm, some phenotypes used for management or inputs to genetic evaluation, particularly those are notoriously difficult to be measured, may be best estimated by genomics."
2. It would nice to reorganize "2. Materials and Methods" section with subtitles for long paragraphs. At least, one subtile use real data and another subtile use simulated data.
3. In "4. Discussion" section, I would like to suggest the authors add few sentences to mention the limitations of this study.

Author Response

The manuscript used real genomic data to estimate bull's prolificacy and later conducted simulations to evalute the reliability of estimated parameters. The obtained results provide genomic evidence for practical use of multiple-bull breeding pastures.  Minor changes are needed.

1. Lines 28-29, "Beyond the genome to phenome paradigm, some phenotypes used for management or inputs to genetic evaluation may be best estimated by genomics." It would be great to be more specific about some phenotypes. The sentences can be changed to "Beyond the genome to phenome paradigm, some phenotypes used for management or inputs to genetic evaluation, particularly those are notoriously difficult to be measured, may be best estimated by genomics."  Response:  We agree and have incorporated Reviewer 2 comment into the manuscript where suggested.
2. It would nice to reorganize "2. Materials and Methods" section with subtitles for long paragraphs. At least, one subtile use real data and another subtile use simulated data.  Response:  We have added some subtitles.  However, we only simulated errors, pool construction and technical error.  So our simulation is a combination of real allele frequencies with simulated error which we think is preferrable and more conservative to not simulating error.
3. In "4. Discussion" section, I would like to suggest the authors add few sentences to mention the limitations of this study.  Response:  We added some sentences on limitations of the study.  For example, the study did not include F1 cross cows which are used a lot in the commercial sector.  F1 cross cows may reduce precision because there might alleles that are not inherited in the calve that are reflected in the dam pooling allele frequency.