Analysis of variance components of testcross progenies in an autotetraploid species and consequences for recurrent selection with a tester.

For autotetraploid species the development of the concept of test value (value in testcross) leads to a simple description of the variance among testcross progenies. When defining directly genetic effects at the level of the value of the progenies, there is no contribution of triand tetragenic interactions. To estimate additive and dominance variances it is only necessary to have the population of progenies structured in half-sib or full-sib families; it is then possible to determine the presence of epistasis using a two-way mating design. When the theory of recurrent selection is applied dominance variance can be neglected for the prediction of genetic advance in one cycle as well for the development of combined selection when progenies are structured in families. The results are similar to those for diploids with two-locus epistasis. The more efficient scheme consists of the development of pair-crossing in off-season generations (for intercrossing) and simultaneous crossing of each plant to the tester. In comparison to the classical scheme, the relative efficiency of such a scheme is 41%. The use of combined selection will further increase this superiority.