Three experimental plot forms (rectangular, one line - linear and single tree plot - STP) were evaluated for Eucalyptus spp clonal tests, using three experiments with 18 clones each. Three analysis models (ordinary least square - traditional ANOVA, mixed models with fixed clone factor or with random clone factor- REML/BLUP) were used. The first and second models showed similar results. With REML/BLUP, there was a reduction in the range of prediction compared to the amplitudes obtained employing the means. This reduction was proportionally higher with rectangular and STP plots. The ranking of clones was also similar to these two types of plots. Probably, there was a compensatory balance of alocompetition with STP plots, because it is possible to work with a higher number of repetitions and lower costs. Therefore, the use of STP plots results in resource saving, preventing losses in forest breeding programs.
Eucalyptus; experimental design; mixed models
