Yield and Reproductive Growth of Simulated and Field‐Grown Soybean I. Seed‐Filling Duration 1

The existence of close associations between yield and seed‐filling duration in soybean [ Glycine max (L.) Merr.] populations may make indirect selection for yield feasible. However, reported positive associations have not been consistently strong. A dual approach was taken to examine the association between yield and seed‐filling duration estimates in soybean. First, 45 simulated soybean genotypes were generated with a simulation model of soybean reproductive growth. The hypothesis that a positive association exists between yield and seed‐filling duration was formulated using the simulated genotypes. Plant traits associated with the dependence of yield on seed‐filling duration were examined. Second, this hypothesis was tested in four field experiments using 88 random and 16 to 18 selected F7 determinate (MG VI1 and VIII) genotypes from two populations and their parents. Data on 11 additional field experiments were also used to compare visual seed‐filling duration estimates, as predictors of yield across environments, with predictions based on yield itself. Simulations predicted linear yield increases (r 2 = 0.46**, significant at the 0.01 probability level) with longer effective filling periods (EFP) and reproductive period duration (RPD), (r 2 = 0.48**). In field experiments among selected genotypes, usually small portions of the variation in yield were explained by EFP (r 2 =0.01 to 0.34) and RPD (r 2 =0.12). Associations of yield with R5 to R7 increased to r 2 =0.39** and 0.42**. Among random genotypes, associations of yield were either nonsignificant or positive with RPD (r 2 = 0.05 to 0.16**) and R5 to R7 (r 2 =0.08 to 0.23**). Simulations indicate that weak positive relationships of yield with seed‐filling duration can be expected due to differential associations of yield with other plant traits. Generally, nonsignificant and inconsistent correlations were observed between yield and seed‐filling duration estimates (EFP, RPD, R5 to R7) across field environments. Yield in any single year was a better predictor of yield than seed‐filling duration. Simulation analyses and field studies showed that while higher yielding soybean usually had longer seed‐filling duration, selection for this trait would not necessarily result in selection for higher yields.