EFFECT OF UNIQUE CYTOPLASMS IN RECIPROCAL CROSSES OF SOYBEAN

CytoptasmlcCytoptasmic inheritance studies concerning agronomic charac ters and seed quality traits have often produced contradictory results . The use of parental lines with similar cytoplasms may explain the se eming discrepancies, since differences among the cytoplasms were not I dentified in previous studies. Differences among cytoplasms have been identified in soybean [Glycine max (L.) Merr.] by means of chloroplast DNA (cpDNA) restriction fragment length polymorphism (RFLP) patterns. This study was conducted to determine if cytoplasmic effects exist in soybean. Ten soybean lines representing five cpDNA cytoplasmic groups were crossed to 'Harosoy 63' and 'Clark 63' to produce 40 F-1 and rec iprocal F-1 populations. Ten Harosoy 63 F-1 and reciprocal F-1 populat ions were advanced to the F-2 generation for evaluation in the followi ng year. The parents, crosses, and reciprocal crosses were evaluated f or the following agronomic characters: developmental stages, height, l eaflet characteristics, and seed size. The population means and varian ces of each reciprocal cross were examined. Reciprocal cross means wer e significantly different for agronomic characters in only 8% of the c rosses. Reciprocal crosses rarely differed with respect to the varianc es. Protein, oil, fatty acid, and amino acid composition were examined in composite samples of seed for each line. Significant differences w ere observed for only a few reciprocal crosses. Protein and oil concen tration each differed in only one reciprocal cross. Oleate and linolea te concentration differed the most often, accounting for eight of the 10 significant reciprocal cross differences observed for fatty acid ty pes in the F-1 and F-2 generations. Reciprocal cross differences invol ving the aliphatic amino acids were observed in four of the five cytop lasmic groups, but were not consistent with the cpDNA RFLP groupings. Consistent cytoplasmic effects, based on cpDNA RFLP groupings, were no t observed for any of the traits examined. The significant differences that were observed may have been due to nuclear x cytoplasmic interac tions, short-lived maternal effects, or Type I error.