A large open pollinated progeny test of Pinus caribaea var hondurensis planted at eight coastal sites in Queensland, Australia, was studied to determine the importance of genotype-environment interactions. Families were established as groups, or populations, according to their origin and were measured approximately six years after planting. Among the four populations studied, representing 90 families, statistically significant population x site interactions were found for stem diameter (P < 0.05) and straightness score (P < 0.001), but not for height. Partitioning of the population x site interaction revealed that 55% of the interaction variance in straightness could be accounted for by differences between populations in the slope of the regression on site means. For diameter, the figure was only 7%. It was considered likely that the large regression component for straightness was due to disparate scoring scales used over the sites. Least-squares site means for diameter ranged from 13.3 cm to 16.8 cm. At all sites, the Queensland families had the highest mean diameters (15.9 cm), followed by Fiji and New Caledonia families (15.5 cm), Culmi (14.9 cm) and Rus Rus (14.3 cm). Mean heights at each site ranged from 7.8 m to 10.6 m. At each site, populations ranked the same for height as for diameter. Straightness scores ranged from site means of 1.03 to 3.24, with population mean scores of 2.34 for Queensland families (straightest), 2,06 for Fiji and New Caledonia families, 1,61 for Culmi families and 1.79 for Rus Rus families. With one very minor exception, rankings of the populations were consistent across sites. For practical purposes, interactions were of little importance for the traits studied. The breeding strategy adopted for Pinus caribaea var hondurensis in Queensland is consistent with this result.
