Plant & Animal Genomes XVI Conference Plant & Animal Genomes XVI ConferenceJanuary 12-16, 2008
Town & Country Convention Center
San Diego, CA
Anna K Sonesson1 , Theo H.E. Meuwissen2 , Hossein Yazdi2
The aim of this paper was to investigate, by computer simulation, accuracy of breeding values (ACC) in a general genomic selection scheme for aquaculture using different phenotypic recording strategies. The design was an extension to current family-based fish breeding schemes. The association between genetic markers and phenotypes were estimated in sibs of the candidates. Breeding values of candidates were calculated based on their genotypes and the estimates of the genotype effects. 1000 generations of random selection of a population with an effective population size of 200, a genome with 10 chromosomes of 100 markers and 100 QTL per chromosome preceeded the actual selection scheme to create mutation-drift balance. In the selection scheme, 100 sires and 100 dams were selected for a trait with a heritability of 0.4 from 3000 candidates and 3000 sibs were phenotypically recorded. When recording phenotypes every generation, ACC was 0.67 in generation one, decreased therafter probably due to the Bulmer effect to 0.6, where it stayed until generation six when it was further decreasing. When costs were reduced by recording only every second generation, ACC was around 0.4 in generations without recording and around 0.6 in generations with recording. When recording phenotypes only in generation one, ACC quickly decreased from generation two onwards. When recording phenotypes only the first three generations, ACC was kept at 0.6 until generation three when it again quickly decreased. Thus, estimates of the marker effects and accuracy of selection deteriorate rather quickly when phenotypic recording stops in an ongoing selection scheme.