Use of microsatellite loci to classify individuals by relatedness

TLDR

The study examines microsatellite loci for estimating relatedness in wild vertebrate populations and evaluates how locus number and heterozygosity affect misclassification rates. Researchers measured allele frequencies at 20 microsatellite loci in wild mice, used the data to simulate expected relatedness distributions for full, half, and unrelated pairs, validated the simulations with lab‑bred families, and tested UPGMA clustering to recover family groups. The method discriminates unrelated from full‑sib pairs with ≥97% accuracy and half‑sib pairs with >80% accuracy; a parent‑offspring criterion identifies only 15% of full‑sib, 2% of half‑sib, and none of unrelated pairs; simulations match real data; locus number and heterozygosity strongly influence misclassification, with higher heterozygosity halving the required loci; and UPGMA clustering perfectly recovers family relationships.

Abstract

This study investigates the use of microsatellite loci for estimating relatedness between individuals in wild, outbred, vertebrate populations. We measured allele frequencies at 20 unlinked, dinucleotide-repeat microsatellite loci in a population of wild mice (Mus musculus), and used these observed frequencies to generate the expected distributions of pairwise relatedness among full sib, half sib, and unrelated pairs of individuals, as would be estimated from the microsatellite data. In this population one should be able to discriminate between unrelated and full-sib dyads with at least 97% accuracy, and to discriminate half-sib pairs from unrelated pairs or from full-sib pairs with better than 80% accuracy. If one uses the criterion that parent-offspring pairs must share at least one allele per locus, then only 15% of full-sib pairs, 2% of half-sib pairs, and 0% of unrelated pairs in this population would qualify as potential parent-offspring pairs. We verified that the simulation results (which assume a random mating population in Hardy-Weinberg and linkage equilibrium) accurately predict results one would obtain from this population in real life by scoring laboratory-bred full- and half-sib families whose parents were wild-caught mice from the study population. We also investigated the effects of using different numbers of loci, or loci of different average heterozygosities (He), on misclassification frequencies. Both variables have strong effects on misclassification rate. For example, it requires almost twice as many loci of He = 0.62 to achieve the same accuracy as a given number of loci He = 0.75. Finally, we tested the ability of UPGMA clustering to identify family groups in our population. Clustering of allele matching scores among the offspring of four sets of independent maternal half sibships (four females, each mated to two different males) perfectly recovered the true family relationships.