Soyabean-seed Yield Response in Row- and Hill-plots under a Recurrent Selection Programme

The objective of this study was to evaluate F 4 lines in both row- and hill-plots for agronomic traits in order to test the effectiveness of selecting high-yielding S 1 soyabean (Glycine max (L.) Merr.) lines selected in hill-plots after two cycles of recurrent selection. There was a cumulative seed yield loss of 197 kg/ha from cycle C 0 to C 2 when the selections were evaluated in row-plots but a cumulative gain in seed yield of 533 kg/ha from C 0 to C 2 was observed when the selections were evaluated in hill-plots. Ranking of seed yield between row- and hill-plots showed a non-significant association, but correlation coefficients between the two methods of testing were positive and highly significant for height, lodging and maturity. Thus, evaluation of height, lodging and maturity can be carried out effectively with either method of planting. There were increases in plant height, in lodging scores and in days to maturity with each cycle of selection. These changes were evident in both row and hill-plots. The lack of seed yield response to selection as measured in row-plots and a non-significant association between the two plot arrangements indicated that seed yield evaluation by S 1 testing in hill-plots was not a reliable procedure in the identification of productive genotypes