Comparison of experimental designs for estimating quantitative genetic parameters in fish

Four popular experimental designs for estimating variance components of additive genetic effects and (secondarily) dominance and maternal effects in fish (full-sib families, half-sib families, nested design and factorial design) were compared. The error variances of the estimators were formulated as functions of the true (hypothetical) values of genetic and environmental variances, depending on the experimental design considered, and whether the families were reared separately or mixed. Half-sib families and rectangular (many sires×few dams) factorial design were most useful for the estimation of heritability. The nested design appeared much less adequate. The minimum experimental capacity for obtaining accurate estimates was 100 lots in separate rearing or 1000 fish in mixed rearing. A larger experimental capacity is required when dominance and maternal effects as well as heritability are to be estimated because these different estimations, requiring contradictory conditions, cannot be simultaneously optimized.