Joint analysis of triple test crosses of a single population with associated and dispersed inbred testers

Abstract

If two pairs of pure breeding lines and their respective F1's are used as two sets of testers in triple test cross investigations of the same population, the data can be subjected to a combined orthogonal analysis of variance. Where one pair of pure breeding testers is relatively dispersed (usually the parents whose cross produced the population) and the other pair is relatively associated (usually extreme selections from the population), the orthogonal comparisons yield estimates of the additive genetic, dominance and epistatic components of variation and information about the genetical constitution of the testers. The theoretical expectations for the six principal orthogonal comparisons of the combined analysis are presented for the simple case which assumes no non-allelic interaction and a linkage equilibrium, and for more complex situations where one or both of these assumptions is relaxed. The combined analysis of two triple test crosses on 60 F∞, families derived from the cross of V2 and V12 of Nicotiana rustica, using V2, V12 and their F1 as one set of testers and two extreme F∞ families, D10 and D17 and their F1 as the other set, confirm earlier analyses of this cross. The estimates of the genetical components are intermediate in value between those of the separate, conventional triple test cross analyses but the standard errors of the dominance components are lower. In addition the combined analyses show that there have been major changes in the distribution of dominant alleles between the two sets of inbred testers as well as changes in the phases of alleles in the testers at loci that contribute to the epistatic variation. These changes are in the direction expected from the origins of the testers and their estimated coefficients of association/dispersion.