Statistical analysis of individual assignment tests among four cattle breeds using fifteen STR loci1

Abstract

Assignment tests based on multilocus genotypes are becoming increasingly important to certify quality and origin of livestock products and assure food safety and authenticity. The purpose of this study was to determine the potential of microsatellites (STR) for determining the breed origin of beef products among cattle breeds present in the market. We typed 19 STR in 269 animals from 4 cattle breeds. Based on Wright's F-statistics, 4 loci were discarded, and the remaining 15 loci (FIT = 0.101, FST = 0.089, and FIS = 0.013) were used to compute the likelihood that each multilocus genotype of the total sample was drawn from its true breed instead of another breed. To avoid occurrence of zero likelihood when one or more alleles were missing from a tested breed, sample allele frequencies were estimated assuming uniform prior distributions. Log-likelihood ratio [log(LR)] distributions of the individual assignments were determined for all possible breed contrasts, and their means and SD were used to infer the true-positive and false-positive rates at several values of the log(LR). The posterior probability that the animals of a presumed breed were actually drawn from that breed instead of any another breed was then calculated. Given an observed value of log(LR) > 0 and assuming equal priors, these probabilities were > 99.5% in 10 of 12 possible breed contrasts. For the 2 most closely related breeds (FST = 0.041), this probability was 96.3%, and the probability of excluding the origin of an animal from an alleged breed when it was actually derived from another breed was similar.