Abstract
Estimating colony numbers of invading Bombus terrestris in Japan is the most urgent task for eradicating the alien invasive species. Nests of bumble bees are hardly found in nature. A prospective alternative is estimating the number of nests with polymorphic genetic data of individuals foraging in the field. Full-sib reconstruction from such genetic data is NP-hard in a strict sense, but several heuristic methods have been proposed. Among them, likelihood methods have been often applied for reconstructing full-sibs in nature, but they require relatively large sample sizes for obtaining accuracy in allele frequencies. Alternative methods calculate genetic distances between paired individuals and cluster them into family groups, but there are also pros and cons for applying these methods against real data. We proposed another heuristic pairing and clustering method, Shared Loci Correspondence Analysis (SLCA), for reconstructing full-sib families of haplodipoid populations. In SLCA, genetic similarities among individuals are numbers of loci shared by the paired individuals. The similarity scores are sorted by the correspondence analysis for clustering them into full-sib families. SLCA could be applicable to quite small samples and succeeded to correctly estimate the number of bumble bee nests in the field. We also improved estimation accuracy of SLCA by blunting the similarity index among individuals. The modified SLCA performed almost perfectly against challenging problems when numbers of loci and alleles were equal or greater than eight. Future direction of improving SLCA is discussed.
