Yield Improvement in a Soybean Population by Utilizing Alternative Strategies of Recurrent Selection1

Abstract

Maximum genetic gain from recurrent selection requires an efficient and effective strategy for selection of superior individuals from a population and intermating these selections to form a new population. This research was conducted to compare genetic response to alternative strategies of recurrent selection for seed yield in soybean [Glycine max (L.) Merr.]. Four strategies initiated in the AP6 soybean population represented a factorial arrangement of one or two stages of replicated yield testing before selection of superior F4‐derived or S2‐derived lines and one or three generations of intermating between cycles. A fifth strategy, representing a conventional breeding program, was initiated in a population formed by sampling the first of three generations of intermating used to synthesize AP6. The conventional breeding strategy utilized a two‐stage yield test of F4‐derived lines obtained from a limited number of single crosses each cycle. The response to selection was determined by evaluating composites of the parents of each cycle at three locations during 2 yr. The number of cycles completed for the strategies varied from two to five. The average genetic gain per cycle was 9.6 ± 1.1 g m−2 for the conventional breeding strategy, 6.0 ± 1.1 g m−2 for the twostage test and one intermating, 5.5 ± 1.8 g m−2 for the two‐stage test and three intermatings, 1.8 ± 0.6 g m−2 for the one‐stage test and one intermating, and 0.3 ± 1.1 g m−2 for the one‐stage test and three intermatings. The ranking of the five strategies for genetic gain per year was the same as for genetic gain per cycle. The results indicated that multiple generations of intermating did not enhance genetic gain and that the two‐stage test, as conducted in this study, was superior to the one‐stage test for yield improvement.