Congruency of quantitative trait loci detected for agronomic traits in testcrosses of five populations of European maize

Congruency of estimated positions and effects of QTL in different samples of the same cross or different crosses is an indicator of the reliability of these estimates and their usefulness in marker-assisted selection (MAS). We investigated the influence of the sample and genetic background on QTL congruency among five populations of European maize (Zea mays L.). Three samples derived from the same cross comprised 344 (A X B-I) and 109 (A X B-II) F-2:3 as well as 71 F-4:5 (A X B-III) lines. Two other crosses comprised 109 (A X C) and 84 (C X D) F-3:4 lines. All lines were topcrossed to the same inbred tester and evaluated in four or five environments. A combined linkage map of RFLP marker data from all five populations was used in composite interval mapping (CIM). The total number of QTL identified for five agronomically important traits was 42 in A X B-I, 18 in A X B-II, 20 in A x B-III, 28 in A X C, and 23 in C X D. Averaged across traits, the proportion p of the genetic variance explained by these QTL varied between 50.4% in the largest population A X B-I and 30.7% in a population of considerably smaller size (A X B-II). Cross validation (CV) yielded substantially lower estimates of p. Between 10 and 24% of the 42 QTL from A X B-I were also detected within a 20-cM interval in the other four populations. Incongruent QTL among A X B samples were due to the low power of QTL detection and the large bias in QTL estimates. The genetic correlations between predicted (based on QTL positions from one population) and observed phenotypic values in another population were highest among A X B samples with a maximum of 0.68 for plant height. Congruency of QTL was found for kernel weight, protein concentration, and plant height and was mainly attributable to one or few QTL of moderate to large size. If more cost-effective than phenotypic selection, MAS will he promising for these traits.