Revisiting the “a posteriori” granddaughter design


  • George Wiggans Animal Genomics and Improvement Laboratory, ARS, USDA
  • Joel Ira Weller Institute of Animal Science, ARO, The Volcani Center, Bet-Dagan 50250, Israel


granddaughter design, quantitative trait locus, Holstein, confidence interval, genome


An updated search for quantitative trait loci (QTLs) in the Holstein genome was conducted using the a posteriori granddaughter design. The number of Holstein sires with ³100 genotyped and progeny-tested sons has increased from the previous 52 to 71 for a total of 14 246 sons. The bovine genome was divided into 621 segments of ~100 markers each. The sons of each bull were divided into two groups based on which paternal haplotype was transmitted to each son for each chromosomal segment. Significance was tested for each economic trait for each chromosomal segment by a linear model that included the effect of paternal haplotype nested within father. Thirty-three traits were analyzed: yield (milk, fat and protein and component percentages), milk somatic cell score, productive life, daughter pregnancy rate, heifer and cow conception rates, service-sire and daughter calving ease, service-sire and daughter stillbirth rates, 18 conformation traits and the net merit genetic-economic index. Fifty-five chromosomal regions met a significance criterion of P < 10-14 compared with 30 regions in the previous analysis based on 52 grandsire families with 9 178 sons. All traits had at least one significant effect, except for protein yield, daughter stillbirth rate and four conformation traits. Confidence intervals (CIs) of 90% were determined for all effects by application of a non-parametric bootstrap. Length of CIs ranged from 2 to 15 chromosomal segments. In all cases, the CI included only part of the chromosome. No significant relationship between log probability (P) of the effect and CI length was found, even though Ps ranged from 10-14 to 10-41 on chromosome 3 for protein percentage. At least six of the regions displayed a bimodal effect distribution in the bootstrap analysis, which indicates more than a single QTL segregating on the chromosome. Results for yield traits were compared with those recently reported for Australian Holsteins, which found effects with a nominal P of <10-20 on five chromosomes (excluding effects on chromosome 14, which clearly result from the DGAT1 gene) when each single-nucleotide polymorphism (SNP) effect was estimated as a fixed effect. For U.S. Holsteins, a nominal P of <10-6 was found in this study for the same traits in nearly the same chromosomal locations, except for effect of fat percentage on chromosome 27. The identified CIs provide promising locations for study of sequence data to identify causative polymorphisms.