Differentiation power of SNPs for assessing the phylogenetic structure of populations

The main advantages of single nucleotide polymorphisms (SNPs) compared to microsatellites are the ability for standardization and the low mutation rate (Fries and Durstewitz, 2001). Many studies compared microsatellites and SNPs for whole genome scans in humans, but the situation in livestock data remains to be studied. Because of the differences in population structure, mainly due to much smaller effective population sizes and directional selection in livestock populations, it is difficult to translate the results one to one from human to livestock. Differences in differentiation power of microsatellites and SNPs are thus expected. In humans the number of SNPs with an equivalent information content of one microsatellite locus varies between 2.7 (Chakraborty et al., 1999), 3.75 (Thalamuthu et al., 2004), 4.29 (Krawczak, 1999), or 5.56 (Glaubitz et al., 2003). In poultry, Schopen et al. (2008) found, that the number of needed SNPs to compensate every microsatellite was not constant. With 12 microsatellites they needed 2.3 SNPs for each microsatellite. They ascertained that more SNPs would be required for an increasing number of microsatellites. Herráez et al. (2005) found that 2.65 SNPs matched one microsatellite in Galloway cattle. The aim of this study was to assess the number of SNPs needed to reach the same differentiation power than 29 microsatellites. For that purpose, we first realized a Principal Component Analysis (PCA) and plotted each individual in a frame defined by the two first components. We then used Euclidean distances obtained from the first two coordinates of the individuals in the PCA-analysis to compare the differentiation power for both marker types.