Background: The availability of larger-scale SNP data sets in the chicken genome allows to achieve a higher resolution of the pattern of linkage disequilibrium (LD). In this study, 36 k and 57 k genotypes from two independent genotyping chips were used to systematically characterize genome-wide extent and structure of LD in the genome of four chicken populations. In total, we analyzed genotypes of 454 animals from two commercial and two experimental populations of white and brown layers which allows to some extent a generalization of the results.
Results: The number of usable SNPs in this study was 19 k to 37 k in brown layers and 8 k to 19 k in white layers. Our analyzes showed a large difference of LD between the lines of white and brown layers. A mean value of r2 = 0.73 ± 0.36 was observed in pair-wise distances of < 25 Kb for commercial white layers, and it dropped to 0.60 ± 0.38 with distances of 75 to 120 Kb, the interval which includes the average inter-marker space in this line. In contrast, an overall mean value of r2= 0.32 ± 0.33 was observed for SNPs less than 25 Kb apart from each other and dropped to 0.21 ± 0.26 at a distance of 100 kb in commercial brown layers. There was a remarkable similarity of the LD patterns among the two populations of white layers. The same was true for the two populations of brown layers, while the LD pattern between white and brown layers was clearly different. Inferring the population demographic history from LD data resulted in a larger effective population size in brown than white populations, reflecting less inbreeding among brown compared to white egg layers.
Conclusions: We report comprehensive LD map statistics for the genome of egg laying chickens with an up to 3 times higher resolution compared to the maps available so far. The results were found to be consistent between analyzes based on the parallel SNP chips and across different populations (commercial vs. experimental) within the brown and the white layers. It is concluded that the current density of usable markers in this study is sufficient for association mapping and the implementation of genomic selection in these populations to achieve a similar accuracy as in implementations of association mapping and genomic selection in mammalian farm animals.