Multi-environment forest genetics trials allowed studying
the magnitude and importance of the genotype by environment
interactions. The importance of such interactions is
indicated by the genetic correlations among environments,
viewing a trait in two environments as two distinct traits
(Falconer and Mackay 1996). Rank-change interaction is
reflected in departures of rA between sites from +1. The
available estimates of additive genetic correlations from
pairs of traits across-sites for growth traits at the same age
are presented in Table 7. The average estimate of additive
genetic correlations between pairs of sites, regardless of age
for DBH, was relatively high (0.67), and seven out of the
12 estimates tested did not differ significantly from +1,
indicating no additive genetic by environmental interaction.
Estimated correlations were significantly lower between
PRIN and VALL for DBH3 (0.38) and between CAST and
VALL for DBH4 and DBH6 (i.e., 0.35 and 0.58, respectively).
These last estimates, although still positive, were
imprecise, but could reflect different soil characteristics
(i.e., clay soil at CAST and well-drained and sandy soil at
VALL). The estimated additive genetic correlation for HT
between the trials at GUAM and MAYO and between
CAST and PRIN, were weak and negative (−0.09) or
positive (0.14) with high standard errors and differed
significantly from +1. They could be partly due to
establishment problems at two sites (i.e., CAST and PRIN).
Therefore, the interaction (departure from rA between
sites=1) found for HT should be interpreted cautiously.
Furthermore, these estimates of additive genetic correlations
across-sites for DBH and HT are generally imprecise