Genetic control of seed weight in flax (Linum usitatissimum) and possible implications.

Mean seed weight data were obtained from the F1 and F2 of a six-by-six diallel cross with flax (Linum usita-tissimum L.). Pronounced reciprocal differences appeared in the F1, but had largely disappeared by the F2. The genetic control of mean seed weight was examined using two types of analysis of variance. The models underlying both analyses were fitted to the data by matrix methods supplying weighted least-squares estimates of the parameters in the models. Weights, the use of which dealt with the problem of variation in the reliability of means, were the reciprocals of the variances of individual cell (cross/self) means in the diallel data table. Elimination of redundant parameters supplied the minimum adequate models for each analysis type. Dominance was apparently masked by the large transient maternal effects in the F1, but surfaced in the F2, where dominance was towards larger mean seed weight. This may be coupled with findings elsewhere of possible advantages for larger seed weights to speculate on a role in preserving infrequent hybrid progeny among inbreeding (flax) species. Maternal effects producing larger seed size, plus dominance with the same result might be valuable, in conjunction with growth and competitive advantages conferred by larger seed, in preventing early elimination of rare hybrids.