Similarities and effectiveness of test environments in selecting and deploying desirable genotypes

Abstract

Rooted cuttings of 40 different clones (genotypes) of Picea abies (L.) Karst were planted in seven test environments (P, H, L, S, B, M and K) in northern Germany. Type-B genetic correlations between the pairs of test sites were estimated, and a cluster diagram based on inter-site genetic correlations for height was used to illustrate the general pattern of similarities among the test sites. Principal component analysis (PCA) showed that the first three principal components explained 97% of the variance in height, thereby providing a good summary of the relationships among the test sites. Test sites B and K accounted for a relatively high—and test sites H and M for a relatively low—proportion of genotype-environment interaction sums of squares and, consequently, for “ecovalence”. “Moderator” and “instigator” test sites for interactions in different site combinations were detected. On the basis of several similarity measures, we observed two distinct groups: sites M-S-P formed one group, while sites K-L formed the other; sites H and B were closer to the first group. The empirical data indicated that a good test site should have the following features: (1) low interaction behaviour, (2) low coefficient of variation, (3) high genotypic selectivity, (4) high coefficient of determination, (5) high efficiency of expected gain. Based on these criteria, site M was determined to be the best test site for screening genotypes to be planted in other environments. We concluded that one broadly adapted vegetative propagation population of P. abies can be used for the potential planting environments in northern Germany.