variance. Some authors have speculated that including dominance in the model will increase the power of detecting QTL, however, it is not clear under this framework what is the most powerful test of linkage. Also to what extent variance components can be biased if dominance is not accounted in the model of analysis. I present a method to account for dominance when mapping QTL and investigate how estimates of parameters can be biased when dominance is not incorporated in the model of analysis. The results show that when the true model involves dominance, the random additive model produced upward bias estimates of the additive variance due to the QTL; the actual bias is clearly a function of the magnitude of the dominance variance, however, the loss in power is negligible, for the different population structures simulated. Further- more, the position estimates are unbiased, independent of the degree of dominance at the QTL. It can be concluded that variance component method using additive effects for QTL detection seems to be robust to the underlying genetic model of the QTL, although model checking is required to obtain unbiased estimates of the variance components and to test different modes of inheritance. doi:10.1016/j.aquaculture.2007.07.131 Genetic diversity analysis and management of turbot (Scophthalmus maximus) broodstocks assisted by microsatellite markers C. Bouza a , A. Pino a , J. Castro a , M. Hermida a , M. Vilariño a , A. Riaza b , I. Ferreiro b , P. Martínez a a Departamento de Genética, Facultad de Veterinaria, Universidad de Santiago de Compostela, 27002 Lugo, Spain b Departamento I+D Stolt Sea Farm S.A, Lira, 15292 Carnota, A Coruña, Spain A commercial turbot broodstock was characterized using a set of 12 microsatellites suitable for population analysis and parentage assignment in turbot. A brood- stock of 175 individuals, which comprised six different wild and cultured origins (parental groups) was analyzed. The amount and distribution of genetic diversity was estimated. The whole broodstock showed high genetic variation at the 12 microsatellite loci similar to that previously described in natural popula- tions and other farm stocks of this species. Allelic richness appeared more suitable than heterozygosity to detect loss of genetic diversity for microsatellites. No significant differences in genetic variability were observed between the groups considered and the wild population. However two parental groups showed more than 20% decrease in allelic richness compared to the wild sample and the remaining groups. The broodstock showed moderate–high genetic differentiation (GST 3%), in accordance with the different geographic origin of parental groups. Relatedness coefficients in the absence of pedigree information were also evaluated in the broodstock to avoid endogamic crosses. Several groups of putative full-sibs were detected using appropriate statistical packages. The analysis of genetic variation across three generations of selection revealed no significant differences between parental groups and filial generations (F 1 –F 3 ). These results demonstrate the usefulness of molecular tools for management of broodstocks along genetic selection programs. doi:10.1016/j.aquaculture.2007.07.132 Performances of relatedness coefficients using actual microsatellite family data from a turbot selection program A. Pino a , J. Castro a , M. Hermida a , M. Vilariño a , C. Bouza a , A. Riaza b , I. Ferreiro b , P. Martínez a a Departamento de Genética, Facultad de Veterinaria, Universidad de Santiago de Compostela, 27002 Lugo, Spain b Departamento I+D Stolt Sea Farm S.A, Lira, 15292 Carnota, A Coruña, Spain A set of 154 full-sib turbot (Scophthalmus maximus) families of 6 individuals pertaining to a family selection program were genotyped for 12 microsatellite loci to ascertain the statistical properties (error, bias, variance partitioning) of three relatedness estimators commonly used in simulation studies. Pedigree relationships of the 154 families could be traced back until the P generation using the information available to identify parentage relationships among these families. A set of 141 full-sib (FS), 15 half-sib (HS) and 96 parent-offspring (PO) families and 68 unrelated (UR) groups were used to obtain relatedness coefficients with these estimators. Statistical packages were employed to obtain related- ness coefficients for all pairwise individuals within each parentage tested with the three estimators using family data (SPAGeDi 1.1) or a wild population (RELATED- NESS 5.2) as references. A slight downwards syste- matic bias for all parentages tested was observed for all estimators when using family data as reference, S288 Abstracts / Aquaculture 272S1 (2007) S238–S321