Genetic variation in a Norwegian germplasm collection of white clover (Trifolium repens L.) 2. Genotypic variation, heritability and phenotypic stability

Genotypic variation for important agronomic characters, was assessed between and within 11 local populations of white clover (Trifolium repens L.) using replicated clonal experiments at two locations in Norway during the years 1990 and 1991. Genotypic variances were highly significant (p<0.001) for winterhardiness, spring growth, foliage height, general performance, dry matter yield and seed yield, and for leaflet and internode lengths. Highly significant genotype x location interaction variances were also detected for all characters evaluated at both locations. Differences between years as regards the contribution of variance components to the phenotypic variance were substantial. In 1990, the genotypic variances exceeded or were in the same order as the environmental variances for most traits, and the broad sense heritability (h2(bs)) estimates varied from 0.43 to 0.66, with leaflet length and internode length having the highest heritabilities. In 1991, plant lethality reduced the genotypic variances for most traits and consequently reduced the heritability estimates. Seed yield exhibited the highest relative amount of genotypic variation, expressed as genotypic coefficient of variation (CV(G)). Genotypic variances were highly heterogeneous among populations for all traits measured in the first year. There was a tendency towards higher genetic variation within the local populations from the most marginal areas for several characters. Phenotypic stability of foliage height, which is one of the most important agronomic characters, was estimated on individual genotypes by means of Shukla's stability variance. Large differences in stability were revealed, and 47% of the genotypes contributed significantly to the genotype x locations interaction. Phenotypic stability of individual genotypes showed a good consistency over years. Rank correlation between average foliage height and stability was not significant (r = -0.08). This indicates that average performance and stability could be manipulated separately in breeding programmes.