Optimising resources and management of genetic variation in fish-breeding schemes with multiple traits

• Investigate effect of increasing the number of families in a breeding program.
• To what degree will increasing the number of tanks suffer from diminishing returns?
• Two different methods for the implementation of optimum contribution are tested.
• Benefit from additional families highly dependent on selection method.
• Optimal number of families around 200 to 300 if economic considerations are included.

When setting up a breeding program for fish, an optimal breeding scheme is sought, and especially the number of families to use is a pivot parameter in this regard. This simulation study tests a range of probable number of families, with the use of two different methods for implementation of optimum contribution procedures in fish: one based on individual quotas and one with family quotas. Schemes are compared at the same prescribed rate of inbreeding. The breeding goal consisted of two correlated traits, one that could be measured on all selection candidates, the second only on full-sibs. The number of families ranged from 50 to 400, whereas the number of offspring per full-sib family was fixed at 50. Average genetic gain for generations 5 to 15 was used for comparing the schemes, and the rate of inbreeding per generation was restricted to 1%. The individual-based method gave the overall highest genetic gain, but the superiority for this method was most evident for the breeding schemes with a high number of families. The biggest difference between the two methods tested stems from the fact that the family-based method furnished a relatively larger proportion of the gain on the first trait; measurable only on the informants. For the individual-based method, this trait had negative or almost no gain when the genetic correlation was negative. The study also showed that although the total gain did not differ too much, the choice of method could highly influence the specific gain in each of the two traits. It is concluded that for the parameters and assumptions used in this study, the optimal number of families for both methods are likely to be around 200 to 300 if economic considerations are also included.