RESUMO
In fish breeding, full-sib families are often kept in separate tanks until individuals are large enough to be tagged and pooled. This practice induces substantial environmental variation common to full sibs (VFS) in BW. We used multigeneration data on rainbow trout to investigate how variation among families in early rearing density affects BW at different ages and environments (fresh water and sea), and whether variance parameters and ranking of breeding candidates change when density is either excluded or included as a regression term in a multitrait animal model. Increasing density displayed a consistent negative relationship with full-sib tank-mean BW at the end of fry-stage when family sizes were equalized (r2 of linear regressions 11 to 31%). In 4 out of 6 year classes, the significant negative association between density and BW also remained through the family-tank period until tagging at 6 mo of age (r2=3 to 19%). In some year classes, early density had a carry-over effect on later BW means, reaching up to the age of >2 yr (after the second and third growing season). Yet, the association was generally weaker at later ages and varied from nonexistent to both significantly negative and positive (r2=0 to 6%). For each BW, the inclusion of early density in genetic model primarily captured the variance that was otherwise attributable to VFS. The reduction of VFS was most pronounced in tagging BW (21% difference between the models), where common environmental effects were moderately high (c2=0.18 in the model without density). For later BW traits, the difference in VFS was 6 to 8% (c2=0.04 to 0.05). The changes in genetic, residual, and phenotypic variances were generally small in the model, including density. Similarly, only a slight change in the heritability estimate of any BW was found (differences of h2 0.2 to 1.3% between models). Correlations between EBV obtained by the 2 models were highly positive in each BW trait (r range 0.94 to 1.00), indicating that ranking of the breeding candidates remained consistent, regardless of whether density was accounted for or not. Our findings highlight the importance of equalizing family sizes to decrease VFS in rainbow trout growth. If selection is practiced for EBV of BW recorded at an age of >2 yr, and equalization is done early enough, the effects induced by different initial density can be sufficiently corrected for by the common full-sib effect in models used for genetic evaluation.