Domesticated species such as cattle have evolved under natural and artificial selection, leading to cattle breeds (both Bos taurus and Bos indicus) that display a broad phenotypic spectrum. As a result of intensive artificial selection together with artificial insemination, highly productive global cattle breeds have been developed that are replacing local, native cattle breeds. This study focuses on analyses of how selection has modified the genomes of different cattle populations having diverse breeding histories, especially with regard to milk production and adaptation. For that, both gene and genome level studies were conducted. 

The molecular architecture of two quantitative trait loci (QTL) was investigated in different species and breeds using two candidate genes for milk yield, GHR and PRLR. The intracellular parts of the two genes were sequenced from over 10 cattle breeds and from different Artiodactyla species. The study revealed divergent selection pressures on GHR and PRLR genes among Artiodactyl species. The GHR gene was more divergent within genus Bos than between different species among the Bovinae linage. Nonsynonymous mutations have accumulated in the PRLR gene in pigs, possibly implying that PRLR has been either target of directional or artificial selection in pigs. 

SNP markers covering the whole genome at medium density were used to search for effects of artificial selection in different types of cattle breeds and to compare the genetic relatedness of differently selected breeds. This revealed evidences that GHR gene has been a target of selection in certain cattle breeds. In addition, several other genomic regions were found to be targets of selection. Most of them were not shared between the breeds but a region on chromosome 16 was found to be under selection in six breeds. Clear genetic separation between the turano-mongolicus type breed and other Bos taurus breeds was found by both whole genome SNP data and the GHR gene sequence. The within breed diversity was relatively similar for all breeds even if the histories of the studied breeds varied substantially. The estimates of effective population sizes calculated from whole genome SNP data varied from extremely low (24) to moderately high (150). 

In the last stage of the study, whole genome sequences were used for genome-wide association study (GWAS) to find genomic regions affecting milk, protein and fat yield in Nordic dairy cattle. The association study confirmed the existence of milk QTL on cattle chromosome 20, at the GHR gene, whereas no support for the QTL at PRLR gene was gained. Several thousand additional candidate SNPs with effect on milk production were located from eight cattle chromosomes. However, establishing the true causative variant remains challenging even when the densest possible marker map is used because of linkage disequilibrium. 

Taken together, this thesis provides genetic information from various Northern Eurasian cattle breeds that can be used for example for conservation decisions and gives a map of selection signatures for them. These selected genome regions may contain variation that would provide valuable traits for changing climate conditions. The knowledge of the genetic background of milk production and the effect of artificial selection is essential when breeding organizations are making decisions how to maintain and improve their genetic material.