Genetic control of soybean seed isoflavone content: Importance of statistical model and epistasis in complex traits

A major objective for geneticists is to decipher genetic architecture of traits associated with agronomic importance. However, a majority of such traits are complex, and their genetic dissection has been traditionally hampered not only by the number of minor-effect quantitative trait loci (QTL) but also by genome-wide interacting loci with little or no individual effect. Soybean (Glycine max [L.] Merr.) seed isoflavonoids display a broad range of variation, even in genetically stabilized lines that grow in a fixed environment, because their synthesis and accumulation are affected by many biotic and abiotic factors. Due to this complexity, isoflavone QTL mapping has often produced conflicting results especially with variable growing conditions. Herein, we comparatively mapped soybean seed isoflavones genistein, daidzein, and glycitein by using several of the most commonly used mapping approaches: interval mapping, composite interval mapping, multiple interval mapping and a mixed-model based composite interval mapping. In total, 26 QTLs, including many novel regions, were found bearing additive main effects in a population of RILs derived from the cross between Essex and PI 437654. Our comparative approach demonstrates that statistical mapping methodologies are crucial for QTL discovery in complex traits. Despite a previous understanding of the influence of additive QTL on isoflavone production, the role of epistasis is not well established. Results indicate that epistasis, although largely dependent on the environment, is a very important genetic component underlying seed isoflavone content, and suggest epistasis as a key factor causing the observed phenotypic variability of these traits in diverse environments.