The migration of maize from tropical to temperate climates was accompanied by a dramatic evolution in flowering time. To gain insight into the genetic architecture of this adaptive trait, we conducted a 50K SNP-based genome-wide association and diversity investigation on a panel of tropical and temperate American and European representatives. Eighteen genomic regions were associated with flowering time. The number of early alleles cumulated along these regions was highly correlated with flowering time. Polymorphism in the vicinity of the ZCN8 gene, which is the closest maize homologue to Arabidopsis major flowering time (FT) gene, had the strongest effect. This polymorphism is in the vicinity of the causal factor of Vgt2 QTL. Diversity was lower, whereas differentiation and LD were higher for associated loci compared to the rest of the genome, which is consistent with selection acting on flowering time during maize migration. Selection tests also revealed supplementary loci that were highly differentiated among groups and not associated with flowering time in our panel, whereas they were in other linkage-based studies. This suggests that allele fixation led to a lack of statistical power when structure and relatedness were taken into account in a linear mixed model. Complementary designs and analysis methods are necessary to unravel the architecture of complex traits. Based on linkage disequilibrium (LD) estimates corrected for population structure, we concluded that the number of SNPs genotyped should be at least doubled to capture all QTLs contributing to the genetic architecture of polygenic traits in this panel. These results show that maize flowering time is controlled by numerous QTLs of small additive effect and that strong polygenic selection occurred under cool climatic conditions. They should contribute to more efficient genomic predictions of flowering time and facilitate the dissemination of diverse maize genetic resources under a wide range of environments.

In a previous study, we identified a candidate fragment length polymorphism associated with flowering time variation after seven generations of selection for flowering time, starting from the maize inbred line F252. Here, we characterized the candidate region and identified underlying polymorphisms. Then, we combined QTL mapping, association mapping, and developmental characterization to dissect the genetic mechanisms responsible for the phenotypic variation. The candidate region contained the Eukaryotic Initiation Factor (eIF-4A) and revealed a high level of sequence and structural variation beyond the 3'-UTR of eIF-4A, including several insertions of truncated transposable elements. Using a biallelic single-nucleotide polymorphism (SNP) (C/T) in the candidate region, we confirmed its association with flowering time variation in a panel of 317 maize inbred lines. However, while the T allele was correlated with late flowering time within the F252 genetic background, it was correlated with early flowering time in the association panel with pervasive interactions between allelic variation and the genetic background, pointing to underlying epistasis. We also detected pleiotropic effects of the candidate polymorphism on various traits including flowering time, plant height, and leaf number. Finally, we were able to break down the correlation between flowering time and leaf number in the progeny of a heterozygote (C/T) within the F252 background consistent with causal loci in linkage disequilibrium. We therefore propose that both a cluster of tightly linked genes and epistasis contribute to the phenotypic variation for flowering time.

Using advanced intermated populations has been proposed as a way to increase the accuracy of mapping experiments. An F3 population of 300 lines and an advanced intermated F3 population of 322 lines, both derived from the same parental maize inbred lines, were jointly evaluated for dry grain yield (DGY), grain moisture (GM), and silking date (SD). Genetic variance for dry grain yield was significantly lower in the intermated population compared to the F3 population. The confidence interval around a QTL was on average 2.31 times smaller in the intermated population compared to the F3 population. One controversy surrounding QTL mapping is whether QTL identified in fact represent single loci. This study identifies two distinct loci for dry grain yield in the intermated population in coupling phase, while the F3 identifies only a single locus. Surprisingly, fewer QTL were detected in the intermated population than the F3 (21 vs. 30) and <50% of the detected QTL were shared among the two populations. Cross-validation showed that selection bias was more important in the intermated population than in the F3 and that each detected QTL explained a lower percentage of the variance. This finding supports the hypothesis that QTL detected in conventional populations correspond mainly to clusters of linked QTL. The actual number of QTL involved in the genetic architecture of complex traits may be substantially larger, with effect sizes substantially smaller than in conventional populations.

The adaptation of maize varieties to lower N-fertilization could become a selection goal for sustainable agriculture. The question is then to know whether it is better to select at low N-input (direct selection) or at high N-input (indirect selection). To answer this question with specific material, a recurrent selection was developed from the same population during two cycles, with two independent selections: one for adaptation to low N-input and the other for adaptation to high N-input. Populations of each cycle were then evaluated with the initial population both at low and at high N-input. The results show a greater efficiency of indirect selection than of direct selection. This was interpreted as being due to a lower heritability at low N-input, with higher environmental variance and genotype x environment interaction than at high N-input. After one cycle of selection, genetic variance at low N-input was even lower than genetic variance at high N-input. During the two selection cycles there was no specific adaptation of the material to the breeding N-conditions. A low N-input environment appeared as an unfavourable environment for selection. The response to grain yield was associated with a decrease in kernel protein content. However there was an increase in grain protein yield.