Alternativas para o maior sucesso na seleção recorrente com espécies do gênero Corymbia

Abstract

In the last decades in Brazil, evaluations of experiments using species of the genus Corymbia have intensified, especially due to the better quality of the wood. In order for the progenies of these species to be evaluated efficiently, attention must be directed towards the quality of the experiments and selection. Therefore, two studies were conducted using data from evaluation experiments of half-sibling progenies. In the first study, the objectives were to verify if there are differences in progeny survival (PS) between locations and evaluation ages, if dominated plants affect accuracy estimates, and how potential damages from differences in PS could be mitigated. Additionally, early selection efficiency was examined. Experiments with C. maculata conducted in three locations and C. torelliana in four locations at three and six years were used. The experiments were conducted in alpha-lattice with 40 replications and single- plant plots. Besides PS, diameter at breast height (DBH) in centimeters was obtained. Analyses disregarding dominated plants were also performed. It was found that although there were differences in PS among progenies and experiments, none of the strategies used to mitigate damages due to differences in PS significantly altered the progeny classification. There was a high association between estimates at three and six years, indicating that early selection is feasible. The second study involved evaluating the DBH data from the CM and CT experiments mentioned earlier, with the following objectives: estimating genetic and phenotypic parameters, using this information to obtain alternatives for combined selection among and within progenies, verifying the feasibility of selecting the best individuals of each species across multiple locations using reference clones from the experiments, discussing different alternatives for recombination and the consequences on expected progress with selection. It was found that accuracy estimates were of medium to high magnitude in all experiments. CM experiments showed lower PS rates, which contributed to the increase in average DBH. This likely resulted in the superiority of heritability among progeny plants, which were also superior for CM. For both species, it was found that the gain from combined selection was significant at all locations. However, it was observed that in combined selection, the selection intensity among progenies was very low, even considering that heritability among progenies on average was much higher than among individuals. It was noted that in some progenies, the number of selected individuals was high. Evidently, this excessive imbalance in the number of selected individuals from each progeny is not advisable for a recurrent selection program, as the representativeness of the progenies in recombination would be very different. It is possible to select the best individuals across multiple locations and form a single improved population, adjusting for common witnesses. The average of the witnesses was much higher than the averages of the progenies, indicating that several more selective cycles will be necessary for CM and CT species to advantageously replace commercial clones.