Many current tree genetic trials are only suited for a general purpose ranking of genetic entries on a single site. Variance components and genetic estimates from trials with damaged trees, outliers, competition, and microsite effects will often be seriously inflated and have large standard errors. Most trials will encounter one or more of these problems. New robust techniques, spatial analysis, and inclusion of covariates may ameliorate the analysis situation somewhat, but a more rigorous definition of objectives and hypotheses are needed in the design phase to produce results that can prove the value of tree improvement efforts. A better understanding of the deployment "environment" is a prerequisite for more biologically oriented genotype × environment analyses.
No takes yet. Share an insight, caveat, or question.
Steen Magnussen (1993) studied this question.