Most of the phenotypic characters of interest to aquaculture are quantitative rather than qualitative. It is therefore important to understand the genetic architecture of such characters. The polygenic theory of quantitative characters (developed by Mather, 1943) envisaged a fairly large number of loci each with relatively small and equal effects acting in a largely additive way on a quantitative character. The theory made no assumptions about the heritability of the character so that two discrete characters might have similar numbers of loci concerned in their determination but very different values for genetic and environmental variance when these were partitioned from the total phenotypic variance. Over the years it has indeed been observed that relatively large numbers of loci (of the order of 50 according to the work of Shrimpton and Robertson 1988a and b) may be involved but also that effects of dominance and epistasis are frequently involved and that the magnitude of the effect produced by each locus can vary considerably (Mather, 1979).
Current informed thinking on genetic architecture is admirably described by Falconer and Mackay (1996). It can be summarised as follows:
A typical quantitative character is likely to involve:
1. a number of loci which may reach several tens in number;
2. genes acting in ways which may be additive, dominant, epistatic and interactive with environmental factors; and
3. considerable variation in individual locus effect (including many small effects from genes whose primary effect is elsewhere on the phenotype through pleiotropic effects).
The last point is quite important as it suggests that typically a small number of loci account for a very large fraction of the variation in the character (see Figure 3).
Figure 3. Graphical theoretical representation of the relationship between the number of loci determining a typical character and the cumulative proportion of the additive genetic variance account for by such loci.
The work of Devlin et al. (2001), which suggests that the benefits of transgenic technology in strains already subject to selection for the desired phenotype may be negligible, is relevant here. The selected strain of rainbow trout which they used might reasonably be assumed to be homozygous for favourable alleles at most of the loci of larger effect.
These observations lead inescapably to the conclusion that success in achieving the desired phenotype in transgenic animals will depend on the nature of the genetic architecture of the character concerned. Where gene action is largely additive more progress may be expected than where more complex aspects of gene action are seen. However, where gene action is additive success may still be disappointing as with the transgenic AFP referred to in Section 5.1. Hayes, Davies and Fletcher (1991) have shown that the AFP gene exists in multiple copies in the genome of the winter flounder Pseudopleuronectes amenis and that the number may be as high as 40. Thus, to produce successful AFP phenotypes through transgenesis is a highly complex and demanding objective as the effects of individual loci in this case are probably roughly equal.