Analysis of genotype-by-environment interaction in a multisite progeny test with Scots pine for supporting selection decisions

Abstract

AbstractIn multisite forest genetic experiments, the presence of genotype × environment interaction (GEI) is common. GEI may negatively affect the estimates of genetic variance and hamper selection decisions in tree breeding programs. Several measures exist to evaluate the stability of tested genotypes’ performance across environments with a choice of the method likely affecting breeders’ decisions. In this study, we evaluated variation in diameter and height growth performance in the progeny test established at 4 sites with 80 open-pollinated half-sib families of Scots pine. We found significant variation among examined progeny at age 10, reaching up to 31% for diameter and 20% for height depending on site, and significant GEI in both traits. We estimated contribution of each family to GEI using various methods and tools of GEI analysis—AMMI, GGE biplots, heterogeneity of regression coefficients (bi’s), the deviation mean squares from regression (s2di) and Kang’s yield-stability index (YSi). Despite the presence of the cross-over interaction, family ranks did not vary much among sites. The selections based on the phenotype, YSi and restricted bi corresponded well to each other leading to the expected response to selection up to 7.8% on diameter and 4.4% on height, whereas those based on the AMMI stability variance were different and lead to a slight loss in both traits. We discuss our results in the context of the usefulness of those measures of genotype stability for tree breeding programs and propose the procedure to follow for making selection decisions in forest experiments with GEI.