Design of breeding programs

Abstract

The prospects of substantial genetic improvement have been well documented in several fish species (Gjedrem, 1997b). This potential should be exploited through effective and sustainable genetic improvement programs to develop strains for aquaculture with better performance, resource efficiency and product quality. Well-designed programs are also needed for the sustainable use of genetic resources as a safeguard against future changes in both production and market conditions (Hammond, 1994). The term design may be used to characterise the size and structure of the breeding nucleus with respect to the number of full-sib families and breeding candidates tested per round of selection as well as the mating and selection strategies applied. An optimum designed program may be defined as one that maximise the genetic gain for a given trait (or the breeding goal) over a given period of time for a set of predefined constraints. Important constraints are the available testing capacity in terms of number of families (tanks) and breeding candidates that can be tested and the tolerable rate of inbreeding. Of importance could also be the establishing the base population, methods to trace pedigree (e.g., physical tags versus DNA tagging) and the degree of connectedness required across levels of fixed effects (e.g., cohorts, test environments, generations) to obtain unbiased estimates of breeding values and genetic gain. The design of the multiplier units is also important, but so far no study on this has been published. The first large scale selective breeding program for farmed fish was set up for Atlantic salmon in the nineteen seventies (Gjedrem, 1992; Gjen and Bentsen, 1997). The design of this program was based on basic knowledge in quantitative genetics, experiences from livestock programs and available technologies. Full-sib families were reared separately until tagging size and family identification was obtained by cold-branding and fin-clipping (Gunnes and Refstie, 1980). With the exception of some improvements (i.e. more traits in the breeding objective, more families tested and earlier and individual tagging using passive integrated transponder (PIT) tags) rather few changes have been taken place since then. The establishment of other family based programs for salmonids and other fish species, developed mainly in the late nineteen eighties and nineties, have to a large extent followed the same design. The possibilities and constraints for the design of fish breeding programs have been discussed in general terms (Gall, 1990; Bentsen and Gjerde, 1994; Gjerde and Rye, 1997; Gjerde et al., 2002), but studies on optimum designs are few and limited to programs where truncation selection has been applied for a single normally distributed trait under individual (mass) (Gjerde et al., 1996; Villanueva et al., 1996; Bentsen and Olesen, 2002) and index selection (Villanueva and Woolliams, 1997; Trong, 2004).