Effect of linkage disequilibrium on inferences of population structure and introgression of iberian and black honey bees

Identification of population structure, a primary goal in population genetics, is easily performed because there is a number of methods available, implemented by user-friendly software packages. However, the user must be cautious when inferring population structure because spurious results may be obtained when there is strong linkage disequilibrium. With recent development of high-density SNPs we have now more power to interrogate the honey bee genome. However, the greater the number of loci genotyped the greater the chance of scoring loci that are linked. In addition, events such as population bottleneck, small effective population size, genetic drift, and admixture may also generate strong linkage disequilibrium. According to Kaeuffer et al. (2007), correlation rLD is the best way to deal with linkage disequilibrium. These authors recommend removing loci with rLD higher than 0.5 when inferring structure. In this study we used the GoldenGate Assay of Illumina to genotype over 1221 loci in individuals sampled from populations of A.m. iberiensis and A.m. mellifera. In this dataset we used the genetic distance between SNPs and rLD to test the effect of linkage in the number of clusters and the introgression level inferred by the clustering method implemented in the software STRUCTURE.Fundação para a Ciência e Tecnologi