Concepts and procedures are presented for the analysis of progeny trials that incorporate clonal replication as a means to resolve variance arising from nonadditive gene effects. Components of variance from the linear model may be expressed in terms of expected covariances among relatives, and these, in turn, may be used to derive approximations of additive, dominance, and epistatic components of genetic variance. In addition to the usual assumptions applied to conventional progeny trials, the use of this expanded genetic model in the analysis of tests with clonal replicates assumes that the greatest portion of the total epistasis is due to interactions involving groups of more than two or three loci. If this assumption is not satisfied, estimates of additive and dominance variance, including those from trials without clonal replicates, will be contaminated by a large fraction of epistasis, and total epistasis will be underestimated by a corresponding amount. Heritability and gain formulae for alternative selection and deployment schemes are developed and illustrate the use of genetic parameters in the comparison of seedling and clonal reforestation strategies.

Height growth at 10 years from striking was assessed for clonally replicated full-sib black spruce (Piceamariana (Mill.) B.S.P) families tested at three locations in central Nova Scotia. Variance components were interpreted according to an additive–dominance–epistasis genetic model and used to derive comparative estimates of gain from various selection and deployment strategies. Field performance at 5 and 10 years was compared with that of the original ortets and families growing in a 25-week greenhouse study, by means of phenotypic and genetic correlation, and rank-change analyses. Between age 5 and 10, the additive portion of the total genetic variance for height decreased from 66 to 38%, while the dominance portion increased from less than 3 to 13%, and the epistatic portion from 31 to almost 49%. As a consequence, narrow-sense heritability estimates were lower at age 10 and gain estimates also decreased, particularly for those strategies that capture gain primarily from additive effects. Although correlations between field performance and early growth measurements were generally poor, the strongest were found at the half-sib level; full-sib correlations were somewhat weaker and those between clone means and early ortet performance were small and not statistically significant. The strongest age–age correlations were those that involved family mean seedling weight in the greenhouse. Family rankings based on early oven-dry biomass production also showed the most consistent agreement with ranking after 5 and 10 years of field testing, although the analyses suggest that effective early selection is probably limited to culling the worst 25% of the families based on biomass. Genotype–environment interactions were statistically significant, although these were limited to only 2% of the phenotypic variance in height growth; furthermore, strong genetic correlations between environments suggested that these interactions have little impact on selection efficiency.