Statistical efficiency of six progeny test field designs on three loblolly pine (Pinustaeda L.) site types

The environmental variance, defined in this study to include all nonfamily variances, associated with family mean estimates for height were compared among six progeny test field designs on three loblolly pine (Pinustaeda L.) progeny test sites in southeastern Oklahoma and southwestern Arkansas. Twenty-five-tree (or 30-tree) square (or almost square) plots, 5- and 10-tree row plots, 5-tree noncontiguous plots, 2-tree row plots, and single-tree plots were compared. The 5-tree noncontiguous, 2-tree, and single-tree plot designs consistently resulted in lower environmental variance associated with family means than the larger plots. The noncontiguous plot design resulted in environmental variances almost as low as those for the single-tree plot design, except on a site with a strong gradient. The environmental variance associated with family means was as small with 30 trees per family arranged in single-tree or 5-tree noncontiguous plots as with 60 trees per family arranged in 5-tree row plots on the most uniform site.