Search

Improving cold tolerance in maize (Zea mays L.) is an important breeding objective, allowing early sowings which result in many agronomic advantages. Using as source the F(2) population of B73 × IABO78 single cross, we previously conducted four cycles of divergent recurrent selection for high (H) and low (L) cold tolerance level, evaluated as the difference (DG) between germination at 9.5 °C and at 25 °C in the germinator. Then, we pursued the divergent selection in inbreeding from S(1) to S(4). This research was conducted to study (1) the direct response to selection (by testing ten S(4) L and ten S(4) H lines), (2) the trait inheritance (in a complete diallel scheme involving four L and four H lines), (3) the associated responses for cold tolerance in the field (at early and delayed sowings) and (4) the responses for other traits, by testing the ten L and the ten H lines at usual sowing. Selection was effective, leading to appreciable and symmetric responses for DG. Variation among crosses was mainly due to additive effects and the ability to predict hybrid DG based on parental lines DG was appreciable. Associated responses for cold tolerance traits in the field were noticeable, though the relationship between DG and these traits was not outstanding. High tolerance was also associated with early flowering, short plants, less leaves, low kernel moisture, red and thin cob, and flint kernels. These divergently selected lines can represent valuable materials for undertaking basic studies and breeding works concerning cold tolerance.

The evaluation of recombinant inbred lines (RILs) per se can be biased by inbreeding depression in case of allogamous species. To overcome this drawback, RILs can be evaluated in combination with testers; however, testers can carry dominant alleles at the quantitative trait loci (QTL), thus hampering their detection. This study was conducted on the maize (Zea mays L.) population of 142 RILs derived from the single cross B73 × H99 to evaluate the role of different testers in affecting: (1) QTL detection, (2) the estimates of their effects, and (3) the consistency of such estimates across testers. Testcrosses (TCs) were produced by crossing RILs with inbred testers B73 [TC(B)], H99 [TC(H)], and Mo17 [TC(M)]. TCs were field tested in three environments. TC(B) mean was higher than TC(H) mean for all traits, while TC(M) mean was the highest for plant vigor traits and grain yield. As to the number of detected QTL, tester Mo17 was superior to H99 and B73 for traits with prevailing additive effects. Several overlaps among the QTL were detected in two or all the three TC populations with QTL effects being almost always consistent (same sign). For traits with prevailing dominance–overdominance effects, as grain yield, the poor performing tester H99 was clearly the most effective; fewer overlaps were found and some of them were inconsistent (different sign). Epistatic interactions were of minor importance. In conclusion, the three testers proved to affect QTL detection and estimation of their effects, especially for traits showing high dominance levels.