Relationship between genetic distance among parents and genetic variance in populations of soybean

A major goal of soybean [Glycine max (L.) Merr.] breeding is the development and identification of high yielding transgressive segregants. Populations that have greater genetic variation should, on average, have more transgressive segregants than populations with lesser variance. The goal of this research was to study whether the genetic distances between parents of crosses were predictive of which crosses have the greatest genetic variance for yield and other agronomic traits. Genetic variance for seed yield, plant height and maturity date was estimated for three sets of populations in field tests. The first set included eight populations evaluated in 1989 and 1990, the second set included 21 populations evaluated in 1993 and a subset of 13 populations evaluated in 1994, and the third set included 24 populations evaluated in 1994 and a subset of 10 populations evaluated in 1995. The parents of the crosses were evaluated with RFLP markers to estimate RFLP distances (RFD) and for coefficient of parentage (CP) which was subtracted from one to measure genealogical distances (GD). Both GD and RFD were significantly (P < 0.05) correlated with genetic variance for plant height for the first set. There were no significant correlations between either genetic distance estimator and genetic variance for any trait in the second set. In the third set, GD was significantly correlated with maturity and plant height and RFD was significantly correlated with maturity. Genetic variance for yield was adjusted for maturity with covariance analysis. These adjustments reduced the estimates of yield variance and mostly reduced the correlations between yield variance and genetic distance estimates. To further evaluate the predictive ability of GD and RFD, the populations in each set were subdivided into two groups, one with parents that had the greatest genetic distance and the other with the least distance based on GD or RFD. The average genetic variance for yield of populations in the more distant group was greater than for the less distant group for the first and third set. These differences were significant for the third set in both years for RPD. These results indicate that although genetic distance can not accurately predict genetic variance of individual crosses, it can on average be useful to identify groups of crosses that will produce populations with greater genetic variance.