To determine the appropriate replication numbers for selection and estimation of genetic parameters economically and with appropriate precision, 21 four-year-old half-sib progenies of taxi-branco (Sclerolobium paniculatum Vogel) were studied in a randomized block design plantai in savanna belonging to the Agroforestry Research Center of Amapá-Embrapa Amapá, Amapá, Brazil. Estimates of genetic and phenotypic parameters were made for the growth characteristics height, diameter at breast height (DAP) and biomass, with the number of replications varying from two to ten. The efficiency of selection with different numbers of replications was compared to the experiment with ten replications. There was no advantage in using more than three replications, because the experimental precision did not change with more replications; the same occurred with the estimates of the genetic parameters and their associated standard deviations. The selection efficiency was high, showing almost full agreement between progenies selected using smaller replication numbers and with ten replications. This suggests that a smaller replication number (three) can be used, permiting the evaluation of larger numbers of progenies. This is a promising strategy to increase the chances of success in the selection process of Sclerolobium paniculatum.
family evaluation; experimental precision; selection efficiency; progeny test