Abstract
Genomic selection has increased genetic gain in many dairy cattle populations, but mostly in numerically large, transboundary breeds. Potential for genomic selection in small breeds tends to be limited by lower accuracy of genomic prediction, resulting from a smaller training population for the prediction model. However, accuracies can be improved by augmenting the training population with data from related populations, called multi-breed prediction. Genetic marker data used for genomic prediction also allow characterization of inbreeding and genetic diversity, and can be used for inbreeding management. Breeders of Icelandic Cattle, an example of a local dairy breed, aim to implement genomic selection. This thesis aimed to investigate the feasibility of genomic selection in Icelandic Cattle, characterize relationships to other populations, estimate inbreeding, and to study methods to manage inbreeding in a dairy cattle population with a genomic breeding program.
Paper I studied relationships of Icelandic Cattle with other breeds, admixture in the Icelandic population, and population structure within breed. Icelandic Cattle is a genetically distinct breed that has been almost completely isolated for a long time. The most related breeds are some of the Nordic traditional breeds. The population has only minor structure. The observed structure was due to differences in relationships of animals to recent bulls with high impact.
Paper II studied inbreeding, linkage disequilbrium and selection signatures in Icelandic Cattle. Levels and rates of inbreeding are not excessive. Linkage disequilibrium in Icelandic Cattle reflects its long isolation and suggests that high genomic prediction accuracy can be achieved. We identified three regions that had recently been the target of selection.
Paper III compared accuracy of tradtional prediction, based on pedigree, with prediction based on both pedigree and genomic infomation, for milk yield, fat yield, protein yield and somatic cell score. Accuracies of young, genotyped animals were 13, 23, 19, and 20 percentage points higher with genomic prediction than with pedigree-based prediction. There were only slight improvements for non-genotyped animals. Assuming the genomic prediction accuracies obtained, current selection intensities, and shortened generation intervals, genetic gain could be increased by 50% relative to the current breeding scheme.
Paper IV simulated a small dairy cattle population with genomic selection. The study compared the effect of using different kinship matrices for management of inbreeding using optimum contribution selection. Using genomic relationships was more efficient than using pedigree relationships. The genomic relationship matrix should be constructed using reference allele frequencies that are estimated from base animals.
The lack of close relationships to other breeds suggests a high conservation value for Icelandic Cattle, but also indicates small benefits from multi-breed prediction. However, lack of within-population structure and relatively high linkage disequilibrium benefit genomic prediction. A more thorough study is needed to find the optimal structure of a genomic breeding program. The future breeding program should aim to increase genomic prediction accuracy by: (i) genotyping cows, (ii) adding sequence variants to the genotyping chip, (iii) improved models, iv) possibly using multi-breed prediction. Optimum contribution selection should be implemented using a genomic relationship matrix.
Paper I studied relationships of Icelandic Cattle with other breeds, admixture in the Icelandic population, and population structure within breed. Icelandic Cattle is a genetically distinct breed that has been almost completely isolated for a long time. The most related breeds are some of the Nordic traditional breeds. The population has only minor structure. The observed structure was due to differences in relationships of animals to recent bulls with high impact.
Paper II studied inbreeding, linkage disequilbrium and selection signatures in Icelandic Cattle. Levels and rates of inbreeding are not excessive. Linkage disequilibrium in Icelandic Cattle reflects its long isolation and suggests that high genomic prediction accuracy can be achieved. We identified three regions that had recently been the target of selection.
Paper III compared accuracy of tradtional prediction, based on pedigree, with prediction based on both pedigree and genomic infomation, for milk yield, fat yield, protein yield and somatic cell score. Accuracies of young, genotyped animals were 13, 23, 19, and 20 percentage points higher with genomic prediction than with pedigree-based prediction. There were only slight improvements for non-genotyped animals. Assuming the genomic prediction accuracies obtained, current selection intensities, and shortened generation intervals, genetic gain could be increased by 50% relative to the current breeding scheme.
Paper IV simulated a small dairy cattle population with genomic selection. The study compared the effect of using different kinship matrices for management of inbreeding using optimum contribution selection. Using genomic relationships was more efficient than using pedigree relationships. The genomic relationship matrix should be constructed using reference allele frequencies that are estimated from base animals.
The lack of close relationships to other breeds suggests a high conservation value for Icelandic Cattle, but also indicates small benefits from multi-breed prediction. However, lack of within-population structure and relatively high linkage disequilibrium benefit genomic prediction. A more thorough study is needed to find the optimal structure of a genomic breeding program. The future breeding program should aim to increase genomic prediction accuracy by: (i) genotyping cows, (ii) adding sequence variants to the genotyping chip, (iii) improved models, iv) possibly using multi-breed prediction. Optimum contribution selection should be implemented using a genomic relationship matrix.
| Originalsprog | Engelsk |
|---|
| Forlag | Århus Universitet |
|---|---|
| Antal sider | 144 |
| Status | Udgivet - nov. 2022 |