Additive genetic effects of single chromosomal segment introgressions and epistatic effects of paired introgressions on quantitative traits in maize (Zea mays L.)

Epistasis effects estimated in conjunction with quantitative trait locus (QTL) mapping are often confounded by segregating background genomes or QTL x background interactions. To improve the precision of estimating QTL epistatic effects, 16 near-isogenic lines (NILs), resulting from the introgression of chromosomal segments of Tx303 into the genetic background of inbred line B73, were crossed in all pair-wise combinations. Marker-assisted selection and self-fertilization were employed to create double-introgression NILs (dNILs) homozygous for two introgressed segments. The resulting 127 dNILs, their 16 parental, single-introgression NILs, and inbred lines B73 and Tx303 were evaluated as inbred lines, per se, as well as in testcross combinations with Mo17, in replicated field trials to measure the effects of introgressions singly (additive effects) and in pairs (epistatic effects). Across traits tested, significant additive effects involving single NILs were detected in 2% to 28% of the tests in the inbred trials and between 2% and 40% of the tests involving single NILs in testcross in the hybrid trials. Significant epistatic effects were identified in 2% to 19% of the tests including inbred dNILs and in 2% to 27% of the tests including testcross dNILs. Generally, the frequency and magnitude of epistatic interactions were less than those of additive effects for both the inbred and hybrid trials. Across traits, between 0% and 100% of the significant epistatic interactions involved chromosomal segments that did not display independent significant additive effects. There was minimal congruency between chromosomal regions displaying significant genetic effects in the inbred and hybrid trials. These results suggest that epistatic interactions can affect predictions of phenotypic performance based solely on additive effects detected either in inbred lines or hybrid combinations. Such predictions might be biased and warrant testing in both population types.