Performance of upland rice families selected from segregant populations

The objective of this research was to analyze the agronomic performance of rice (Oryza sativa L.) families selected from divergent populations evaluated under different sowing times and locations in 1996/97. The source populations for the families were CNA x 5496, CNA x 6001, CNA x 6063 and CNA x 6102. A hundred families from each cross (population) were evaluated and submitted to five common treatments, which originated a three-replicate 20x20 lattice, grown at two locations, Lavras and Patos de Minas in 1997/98. In the next year (1998/99), the ten best and the ten worst-performing families from each population were again evaluated to confirm the superiority of families selected from the CNA x 5496 and CNA x 6001 populations. A 9x9 lattice with three replications experimental design was used. The two populations selected as the most promising in 1996/97 yielded the best families upon selection. Therefore, in a similar breeding program, it would be advantageous to carry out a high number of crosses, originating a great number of segregant populations, and to select the best ones, concentrating efforts on the evaluation of families extracted from superior populations. Such results confirm the efficiency of selection of source populations for family extraction.