The use of cross prediction methods in a practical potato breeding programme.

Most previous studies on cross prediction methods have examined relatively few crosses, particularly in relation to the numbers involved in most breeding programmes. In this paper the feasibility of using cross prediction methods was examined in a practical potato (Solanum tuberosum) breeding scheme by the analyses of progeny from 52 crosses. The variate considered was breeder's preference, a visual assessment made of the harvested tubers to estimate their commercial potential. The results showed that it was possible to identify the superior crosses. Cross prediction based simply on the mean preference scores, averaged over scorers and clones within progenies, estimated on seedlings or first clonal year plants, provided the best estimate of a progeny's performance in the third clonal generation. Predictions based on the expected proportion of clones that would transgress a given target value also provided a good indication of a progeny's potential. The poorest prediction was obtained by using the observed frequency of desirable clones in a progeny sample. The implications for potato breeding are discussed.