Studies on genotype × spacing interactions for herbage yield, using a modified diallel analysis

Forty-four full-sib families from a diallel cross between 10 Lolium ((multiflorum × perenne) × perenne) clones were planted in the autumn at 0·6 × 0·6 m and 0·l × 0·l m spacings. Fresh weight of individual plants was measured on three occasions during the following spring and summer. The analysis of the diallel cross was extended to account for combining ability × spacing interactions, and to estimate between-spacing combining ability correlations and their sampling errors.Family × spacing interactions became increasingly significant at successive harvests and always contributed substantially to the total phenotypic variance. Much of the family × spacing variance was due to the general combining ability (GCA) × spacing component. Despite these large and significant interactions, the correlation between GCA values at the two spacings was high and positive at each harvest, so that selection for yield at wide spacing should lead to a correlated increase in yield at close spacing. Although the experiment was of relatively short duration, the results suggest that recent criticism of the conventional technique of selection among widely spaced plants in pasture plant breeding programmes may not have been entirely justified.