Genomic selection of dairy cows
Sélection génomique des vaches laitières
Résumé
Genomic selection has revolutionized breeding in dairy cattle, at least on the male pathway. This thesis focuses on the female side. First, the genotyping tool most adapted to females was defined. The first study conducted within the Eurogenomics consortium assessed the value of using the commercially available Illumina® 3K SNP chip. The allelic imputation error rate was 4% with the national reference population, and the loss in reliability of GEBV when using imputed genotypes instead of real genotypes was 0.05 (2% and 0.02 respectively with the combined Eurogenomics reference population). In a second study, alternative in silico low density chips were described. Their imputation accuracy was 1 to 2.5% higher than the initial commercial 3K. The imputation accuracy not only depends on the number of markers, but also on MAF and spacing. A novel imputation strategy, fast and accurate, based on existing software, was described. Then, the construction of the new Bovine LD panel, adapted to many breeds and specifically dedicated to imputation, was detailed. This tool is well adapted for the genotyping of females in dairy cattle at a reasonable cost. A second main aspect of this thesis was to study how performances of genotyped cows fit within the current genomic prediction model. An experimental design was set up to assess the effect of potential biases such as preferential treatment on genomic predictions. Two genomic evaluations were performed, one including only daughters performances of proven bulls, and another one including phenotypes for both males and females. Two traits were studied: milk yield, which is prone to preferential treatment and somatic cell count. Two groups were considered: elite females genotyped by breeding companies and randomly selected cows genotyped in a side project. For several measures potentially related to bias, the elite group presented for milk yield a different pattern than for the other trait/group combinations. The study demonstrated that including own milk performances of elite females induced over-estimated genomic evaluations. Such a bias has two major consequences: it may affect genomic predictions equations, and it may induce overestimated breeding values for the cow and her close relatives. Different possible solutions to properly include such performances in genomic predictions were described and their potential impacts were compared. Finally, the benefits of genotyping heifers either by breeding companies or by farmers were discussed. A review of several simulation studies was conducted. Selecting bulls dams based on their genotypes appears to be crucial within a breeding scheme. Indeed, it is as important as using young bulls for artificial insemination. Using genotyping tools to select heifers to replace culled cows is more controversial. The return on investment for the famers depends on the cost of genotyping, the replacement rate as well as the economic value of the expected genetic improvement. Several herd management decisions could be facilitated when using genomic breeding values. A positive interaction exists between genomic selection within herd and several reproduction practices such as embryo transfer or use of sexed semen. Their combination may help in solving the issue that dairy cattle faces today related to the decrease of performances for health traits such as fertility.
Domaines
Génétique animale
Loading...