Yield improvement in soybeans using recurrent selection

Abstract

The objective of this study was to evaluate the potential of recurrent selection for improving seed yield in a soybean population. Seventeen parents, chosen for their high yield potential in the sub-tropical environments of Southern and Central Queensland were intermated using a diallel cross mating system. F1 plants were intercrossed in pairs for two more generations and then allowed to self for two generations to produce the base population. The selection criterion was the seed yield of S1 lines grown at Hermitage Research Station. The selection intensity was approximately 10%. After five cycles of recurrent selection the mean yield of the population and the mean yield of the lines selected for recombination had improved by 17% and 54% respectively, in comparison with one of the original parents, Davis. Progress from recurrent selection was evaluated using fifty randomly selected lines from the base population and from each of the first three cycles of selection. The average gain per cycle in seed yield, averaged over five sites was 128 kg ha-1 or 5.4% of the yield of the base population. Greater progress (9.8% per cycle) was measured at the testing site for selection (Hermitage Research Station) which was heavily infested with phytophthora root and stem rot than at three other sites (2.7% per cycle) which has nil or low levels of disease. Broad sense heritabilities for the base population and for cycles 1, 2 and 3 in the evaluation trial were 0.34, 0.25, 0.13 and 0.14 respectively compared with a mean heritability of 0-64 for the first five cycles of the selection experiment. Expected gain from further cycles of selection was estimated at 102 kg ha-1 per cycle for three replicates at a single site. In the evaluation trial correlated responses to selection for yield were found in a number of other traits. Days to flowering was reduced by 0.8 days/cycle while days to maturity increased by 1 day/cycle. Seed shattering score was significantly reduced from 0.98 to 0-69 while there were only minor changes in plant height, lodging and seed coat quality score. Resistance to root and stem rot and seed coat mottling, due to primarily to soybean mosaic virus, were also significantly increased. The improvement in phytophthora resistance was expected as the field used for S1 yield testing had become infested with the disease. A covariate analysis indicated that about half the improvement in seed yield at the Hermitage site was due to the correlated response of increased resistance to phytophthora root and stem rot.