In animals, large or small body size may carry health costs. Large deviations from physiologically benign growth
range can divert resources from vital body functions and cause developmental defects, thereby constraining the
welfare and productivity of animals. When analyzing the phenotypic costs associated with growth variation, it is
unclearwhether slow or fast growth should be determined relative to an individual's own family background or relative
to population's overall genetic level. That is, should the body size of an individual be related to its parental
(genotypic) average, instead of a general population mean? Using the data from 60,518 individual rainbow trout,
Oncorhynchus mykiss, we tested whether the negative and positive deviations of tagging body weight
(one-summer-old fish, average weight 51 g) relative to either a common populationmean or sire-family means differently
predict the probability ofmortality and vertebrae defects during the subsequent grow-out period. Based on
the logistic regression analysis, neither types of bodyweight deviation predicted the probability of vertebrae defect.
In contrast, the probability of mortality showed a significant negative relationship with deviations of weight. The
smallest individuals with large negative deviation from their sire-family means possessed the highest risk of
death. However, there was a high positive correlation between the deviation types calculated from sire-family
and population means (r=0.93), and thus their ability to predict mortality was very similar. This finding does
not support our physiological capacity hypothesis stating that the phenotypic vitality costs would be more strongly
associated with an individual's body size relative to its family's genetic mean level, rather than relative to the overall
population mean. Rapid fingerling growth was not phenotypically related to either of the vitality costs studied. The
largest fish, both with respect to population and sire-family means, had the lowest risk of death and were no more
or less liable to developmental vertebral defects, compared to their smaller counterparts. These results emphasize
the importance of narrowing growth variation, especially from the lower end of the growth profile, using mating
and selection designs to reduce fish mortality during the communal grow-out period.