Strategies to improve the efficiency of single-tree plot in eucalyptus breeding

Abstract

Single-tree plot (STP) is widely used in eucalyptus breeding programs. It has the advantage of allowing the use of a large number of repetitions, which contributes to improve the experimental precision. Its use, however, has been questioned due to a possible influence of missing plants in the plots, the effect of neighbors and its viability using the lowest number of repetitions. From the above, the present study was carried out with the objective of: verifying the influence of neighbors in experiments in STP; verifying the influence of missing or dominated plants; to evaluate a minimum number of repetitions in these experiments, without reducing the selective accuracy; to identify possible influences of the number of repetitions in the inferences to be obtained in the interaction clones x environments. To answer these questions, three research studies were carried out. The experiments were carried out in four Brazilian states by the company Suzano S.A. Data on the average annual increment (MAI, m3.ha-1.year-1) at three years of age were used. The design was always in randomized complete blocks. In the first study, STP data were used to verify the influence of neighbors, missing and dominated trees. The effect of missing and dominated trees was studied through the correction between the MAI of each clone/plot and the average of a clonal MAI of the eight neighbors. Analyses were performed considering non-adjusted data, data adjusted with missing and data adjusted with missing and dominated trees. In the second study, the effect of the number of repetitions was evaluated from six experiments with 30 repetitions of STP. The effect of the number of repetitions varying from 2 to 29 was simulated. For each number of repetitions, 1000 resampling, without replacement. In each simulation, the effect of the number of repetitions in the estimates of the general mean, the genetic variance and the accuracy was estimated. In the third study, to verify the influence of the number of repetitions on the inferences obtained from the CxA interaction in the STP experiments. The number of repetitions was simulated, in the estimation of genetic and phenotypic parameters, mainly involving the joint analysis of all environments and possible pairs of environments. It was verified that in STP, the performance of the clones is not influenced by the behavior of the neighbors. No increase in experimental accuracy was observed in any type of adjustment. The ranking of clones was not changed, not justifying the correction of missing or dominated trees. In the second study, it was found that the effect in the mean accuracy estimates were very similar in the different numbers of repetitions. However, the standard deviation of the estimates reduced with the increase in the number of repetitions. In the third study, it was found that the inferences about the interactions clones x environments were very similar, regardless of the number of repetitions. Again, the deviations associated with each estimated parameter reduced with the number of repetitions, without, however, causing significant changes in the results of the joint analyses.