Akademik Veri Yönetim Sistemi
CROP SCIENCE, cilt.33, sa.3, ss.620-626, 1993 (SCI-Expanded)
Genetic distance measures are typically derived from multivariate statistical analysis of morphological, isozymic, or RFLP data but not agronomic data. Rational application of genetic distance measures to plant breeding requires some evidence that these measures are associated with agronomic traits such as seed yield. No such proof is available in soybean. The objectives of this study were to (i) evaluate agronomic diversity among five of the founding stock of Southern U.S. soybean employing F2 heterosis for seed yield, and (ii) investigate the association of agronomic genetic diversity with three genetic similarity (GS) estimates. Two estimates, GS1 and GS2, were constructed from multivariate analysis of biochemical and morphological traits in the founding stock grown in a phytrotron facility. The third, variance for 100-seed weight (GS3) in the F2 progeny of the founding stock, was chosen because 100-seed weight was the trait most heavily weighted in the statistical construction of GS1 and GS2. The five ancestral strains and their ten F2 progeny bulks were evaluated for seed yield and 100-seed weight at two locations in North Carolina in 1989 and 1990. Half of the 10 F2 progeny populations exhibited significant (P < 0.05) mid-parent heterosis for seed yield. Mean heterosis was 9.3% with a range of -1 to 19%. One F2 progeny bulk, that of Tokyo x Roanoke, produced only slightly less seed yield than two modern cultivars derived from 40 yr of soybean breeding. The F2 heterosis for yield was predicted by GS2 and GS3 but not GS1. GS2 and GS3 more clearly distinguished the differences among the strains than did GS1 because these estimates were based soley upon the genotypes under study while GS1 was developed to maximize discrimination among a much broader base of ancestral strains. We conclude that the application of genetic distance measures as predictors of agronomic genetic diversity should be made with caution. Their utility may be population dependent. The F2 heterosis we observed was not associated with previously published similarity indices developed from isozymic and RFLP data.