Selección genómica en ganado ovino de aptitud lechera

  1. Sánchez Mayor, Milagros
Supervised by:
  1. Juan José Arranz Santos Director
  2. Ricardo Pong Wong Director

Defence university: Universidad de León

Fecha de defensa: 06 November 2020

Committee:
  1. Beatriz Gutiérrez Gil Chair
  2. Noelia Ibáñez Escriche Secretary
  3. Pau Navarro Martínez Committee member
Department:
  1. DEP. DE PRODUCCIÓN ANIMAL

Type: Thesis

Abstract

The molecular revolution brought by the emergence of genomics in the last years of the 20th century has led to considerable developments in the field of animal breeding. A large amount of information on the variability of the genome of animal species has been an invaluable resource that has allowed the establishment of new selection strategies, including what is known as genomic selection/prediction. The efficiency of the genomic selection is mostly determined, among other factors, by the structure, characteristics and size of the population where the breeding program is applied, the genotyped animals, as well as the features of the so-called reference population. The main objective of this thesis is to study the possibilities of using molecular markers for the implementation of genomic prediction in the genetic improvement program of the Churra sheep breed managed by ANCHE, when the resources are limited. For this purpose, we have proposed three different studies. The first study aimed to evaluate the potential benefits of different genomic prediction methods (GBLUP, DGBLUP and ssGBLUP) when using a suboptimal reference population, in the selection program of the Churra breed managed by ANCHE. The benefit of genomic prediction methods were assessed in different targeted candidate groups: (i) genotyped ewe candidates with half-sib sisters in the training population, (ii) non-genotyped ewe candidates that are daughters of genotyped rams, (iii) the youngest non-genotyped ewe candidates at the bottom of the pedigree not closely related to genotyped individuals and (iv) genotyped rams. The accuracy of the prediction was assessed as the correlation between the corrected phenotypes and the different genetic values (GEBV) on six milk production traits for the different candidate groups analysed in this study. The use of genomic information in the evaluation did not yield significant benefit in the precision of the GEBV of female candidates, compared to using only pedigree information. However, for the scenario involving male candidates the accuracies of the genomic prediction methods yielded a higher precision in the estimation of their GEBV. To better understand these results, two simulation studies were performed. The first simulation aimed at assessing whether the SNP chip used in the Churra population has enough information to improve genetic prediction. Because this approach simulates data for genotyped individuals, only genotyped ewe candidates were assessed. The second simulation study aimed at evaluating the effect of data structure on the benefit of genomic selection on the Churra sheep. In this simulation, all candidate groups considered in real data were evaluated. The results in the two simulation studies suggested that the use of molecular information can benefit the predictions in the Spanish Churra population, yielding better predictions for genotyped male and female candidates than those obtained using the standard BLUP using pedigree information. The second study aimed to evaluate the effects of genotyping strategies and the proportion of genotyped candidates on genetic gain when using the ssGBLUP. The ssGBLUP method allows information from genotyped and non-genotyped animals to be included in a single analysis, which generally leads to better predictions in genotyped animals. This method allows the option of not genotyping the whole group of candidate animals and, thereby, reducing the cost of the selection program. However, this raises the need to determine the optimal set of candidates to be genotyped to ensure that significant benefit continues to be gained from ssGBLUP. The scenarios compared were: (i) three genotyping strategies on the candidates to be genotyped (RANDOM: candidates to be genotyped were chosen at random; TOP: the best candidates based on their phenotype were genotyped, assumed to be known at the time of the genotyping decision; and, EXTREME: genotyped candidates were those with the best and the worst phenotype) and (ii) six proportions of genotyped candidates. The comparison criterion was the cumulative gain and reliability of the GEBV. The results were based on a simulation study. The results showed that the highest genetic gain across several generations of selection is achieved with TOP, followed by EXTREME and RANDOM. The advantage of TOP over RANDOM ranges from 38-108% of the extra response to that achieved by the standard BLUP, across the whole range of tested proportions of the candidates to be genotyped. The performance of EXTREME was close to that observed with RANDOM when the proportion of genotyped candidates was low, but improved as the proportion increased. However, when comparing the strategies on the reliability of their GEBV, EXTREME had the highest values followed by RANDOM, with TOP being the strategy that produced the lowest reliability. The ranking of genotyping strategies showed the same trend when considering only the group of genotyped candidates. The third study aimed at evaluating novel traits that can benefit from the genomic selection process. For this purpose, cheese making traits (CMTs) were analysed in a commercial population of Assaf dairy sheep. The study estimated genetic parameters of these traits and their correlation with milk production and milk quality traits. In addition, a study was carried out on the indirect response to selection in coagulation and cheese yield traits when using as proxies milk production, milk composition and functional traits routinely measured in the Spanish Assaf selection programme. Of the milk samples included in the study, 13% did not coagulate, whereas approximately 37% showed slow coagulation. The estimated heritabilities for the milk coagulation and cheese yield traits were moderate. Based on the estimated phenotypic correlations, our study supports that a higher number of somatic cells is related to a delay in milk gelification and a lower curd firmness. Many of the estimated genetic correlations showed large standard errors Overall, this study provided the first report on phenotypic and genetic parameters of cheese traits in the Spanish Assaf sheep population. Finally, the results of the second part of the third study are based on the expected response of the cheese traits (not routinely measured) when the milk traits (routinely measured) were used as proxies and taken into account in the selection criteria. Differences in the information available when selecting male or female candidates were considered, and, also, the impact of the current selection index of Assaf Spanish sheep was evaluated, and a new index was suggested. Overall, the results showed that a selection index using milk traits could be optimized to improve cheese traits. The proposed new index will produce substantial genetic progress in milk coagulation properties and individual cheese yield traits and will continue to achieve the same rate of genetic gain in milk yield traits when the current Assaf index is used. In conclusion, the profitability of genomic selection requires that the economic gain associated with the genetic progress should exceed the added cost of the scheme. Hence, the inclusion of more economically valued traits such as CMTs and the reduction of genotyping cost would make genomic prediction more viable in the Churra breed and other economically less important populations.