• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. Apr 2012; 190(4): 1547–1562.
PMCID: PMC3316663

Flowering Time in Maize: Linkage and Epistasis at a Major Effect Locus


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.

QUANTITATIVE variation results from the combined action of multiple genes and environments. The most widely used approaches to dissect the genetic architecture of complex traits and to relate genotypes to their phenotypic expressions are linkage and association mapping. As reviewed in Bergelson and Roux (2010), linkage mapping is essentially limited by (i) the allelic and phenotypic variation present in the parents and (ii) a restricted number of recombination events that do not allow distinguishing pleiotropic effects of a single genetic determinant from linkage between multiple determinants (Doerge 2002; Chen and Lubberstedt 2010). The success of genome-wide association studies (GWAS), on the other hand, depends on our ability to correct for the underlying relatedness (Pritchard et al. 2000; Zhao et al. 2007), but also on the relative contribution of rare variants to the phenotypic variation, often not captured in discovery association panels. Such rare variants may commonly control complex trait variation in human populations, perhaps explaining why association mapping has been applied with only moderate success (Manolio et al. 2009). To circumvent these limitations some authors have explored the possibility of combining linkage and association mapping in several model species including Arabidopsis (Zhao et al. 2007; Brachi et al. 2010) and maize (Yu et al. 2008; Kump et al. 2011; Tian et al. 2011). The so-called joint linkage–association mapping designs artificially equilibrate allele frequencies by using controlled crosses (Myles et al. 2009), thereby reducing the rate of false positives (Manenti et al. 2009) and false negatives (Zhao et al. 2007). Joint linkage–association mapping has been used in plants to dissect the genetic architecture of agronomic traits such as leaf architecture and disease resistance (Kump et al. 2011; Poland et al. 2011; Tian et al. 2011).

Long-term divergent selection experiments, set up to move a phenotype from its initial value, provide an alternative way to investigate the phenotype/genotype relationship. In particular, divergent selection experiments (DSEs) that start with a narrow genetic basis material and in which a single trait is moved toward a “low” and a “high” value, offer a unique resource to investigate the response to selection and its underlying genetic determinants. When the power of association and linkage mapping is limited by a considerable number of contributing loci (complex traits), DSEs typically overcome the effects of population history and selection at multiple traits simultaneously, thereby facilitating investigations on the genetic determinism and the dynamic changes in allele frequencies in response to selection for a single trait. DSEs have been undertaken to explore the genetic determinants of several phenotypic traits such as body weight in chicken (Johansson et al. 2010), mice (Keightley 1998), and Caenorhabditis elegans (Azevedo et al. 2002) and bristle number in fruitflies (Mackay et al. 2005). In plants, a DSE started in 1896 has consisted of selecting maize populations for kernel oil and protein concentration. A continuous response to selection was observed (Dudley and Lambert 2004), leading to an extreme differentiation between the low and the high populations. Crosses between those populations allowed the detection of 50 QTLs with small and additive effects, explaining 50% of the genetic variance for oil concentration (Laurie et al. 2004).

More recently, we undertook a DSE for flowering time starting from the maize inbred line F252 to investigate the relative contribution of de novo mutations and standing genetic variation to the variability of this trait (Durand et al. 2010). Our results showed that the response to selection was fast (seven generations) and significant in both the Early-flowering (Early) and the Late-flowering (Late) population. In addition, within the Late population, we reported the emergence of two subpopulations (a late and a very late subpopulation) characterized by an average flowering time difference of 4 days. Our results pointed to a major locus associated with the variation for flowering time in the Late population and explaining up to 35% of the total variance with mainly an additive effect (Durand et al. 2010). The underlying polymorphism, consisting of a restriction fragment length polymorphism at a candidate locus for flowering time, was present as residual heterozygosity in the initial F252 inbred line.

In plants, the switch from vegetative growth to reproductive growth is referred to as the floral transition and primarily determines flowering time. Flowering time is therefore a key factor in plant adaptation and is linked to developmental characteristics such as the plant height, the total number of leaves, and the grain fill. Flowering time is a complex trait that displays a large range of variability: from 92 to 196 days and from 35 to 120 days after planting, respectively, in Arabidopsis thaliana (Brachi et al. 2010) and maize (Colasanti and Muszynski 2009). More than 60 QTLs for flowering time with five to six major chromosomal clusters/regions have been detected in A. thaliana/maize (Chardon et al. 2004; Buckler et al. 2009; Brachi et al. 2010). In contrast to A. thaliana, where some QTLs have major effects on flowering time variation (Atwell et al. 2010), each QTL in maize contributes to a small part of the phenotypic variation (Buckler et al. 2009). In both model species, however, QTL effects are mainly additive, epistatic interactions contributing on average to <3% to the phenotypic variation (Buckler et al. 2009; Brachi et al. 2010).

While linkage studies have produced a large set of candidate genes for flowering time in maize (Chardon et al. 2004; Buckler et al. 2009), very few genetic determinants have been characterized at the molecular level. A few exceptions merit attention: the maize Indeterminate gene (Id1) (Colasanti et al. 1998), the delayed flowering1 gene (dlf1) (Muszynski et al. 2006), the vegetative to generative transition 1 (vgt1) (Salvi et al. 2007), the gene ZmCCT (Ducrocq et al. 2009; Coles et al. 2010), the Dwarf 8 gene region (Thornsberry et al. 2001; Andersen et al. 2005; Camus-Kulandaivelu et al. 2006), the CONSTANS-like gene (conz1) (Miller et al. 2008), the zfl1 gene (Bomblies and Doebley 2006), and the FT-Like ZCN8 gene (Meng et al. 2011).

The goal of the present study was to combine QTL mapping, association mapping, and developmental characterization of a very late mutant controlling a large part of the phenotypic variation for flowering time in maize in our DSE (Durand et al. 2010) to dissect the genetic mechanisms of its contribution to phenotypic variation. We describe the sequence polymorphism at the locus and show a pleiotropic effect on flowering time, plant height, and leaf number. Hence, while some interactions with the genetic background were consistent with epistasis, linkage between multiple causal loci can also explain changes in the sign and significance of effects. Although the causal mutation(s) remain undetermined, the extensive structural variation between the two alleles in the 3′ region of the Eukaryotic Initiation Factor 4A gene suggests that transposable element movements and associated epigenetic regulation may be responsible for the phenotypic variation.

Materials and Methods

Plant material

Starting from the commercial inbred line F252, we have conducted a divergent selection experiment for flowering time during 11 generations. From an initial seed lot of the commercial inbred line F252, an Early F252 and a Late F252 population were derived by successively selecting and selfing the earliest and the latest individuals at each generation. The selection procedure is described in detail in Durand et al. (2010) for the first 7 generations. By extension, generations 8 (G8) to 11 (G11) were obtained following the same procedure.

In addition, we previously identified a restriction fragment length polymorphism (RFLP) in the Late F252. This polymorphism, present as residual heterozygosity in the initial seed lot F252, segregates at a candidate locus (Qck5e06) for flowering time. The two alleles, referred to as fast and slow on the basis of their electrophoresis profile, were genotyped in the first 7 generations (Durand et al. 2010) and were associated with the emergence of two subpopulations within the Late F252, a Very Late subpopulation (Late-VL) and a Late population (Late-Not Very Late, Late-NVL). For clarity, we designate the two alleles as the f252vl and the f252nvl alleles, which correspond to the fast RFLP allele observed in the Late-VL population and the slow RFLP allele observed in the Late-NVL population, respectively. We have previously shown that the genotypes homozygous for the f252vl allele flowered on average 4 days later than the genotypes homozygous for the f252nvl allele. Note that all genotypes in the Early F252 population were homozygous for the f252nvl allele. A genealogy of the individuals selected from generation 0 to generation 11 is shown in Figure 1.

Figure 1
Origins of plant material derived from the F252 inbred line. (A) Genealogy of selected individuals of the DSE from generation G0 to generation G11. (B) Construction of S-LateF252-C and S-LateF252-T populations with alleles C and T, respectively, at the ...

In the present study, we chose one heterozygous individual for the candidate locus at G2 to generate by selfing two S1 homozygous individuals, respectively, for the f252vl and the f252nvl allele (Figure 1) to test the effect of the f252vl allele in an F252 genetic background. S1 individuals were further selfed to constitute S2 populations from which we derived S3 families (Figure 1). The population S-LateF252-T designates S3 families derived from the S1 individual homozygous for the f252vl allele and the population S-LateF252-C designates S3 families derived from the S1 individual homozygous for the f252nvl allele. Genotyping of the S1, S2, and S3 individuals was obtained by PCR assays as described below.


One of the goals of our study was to characterize the sequence variation leading to the length polymorphism generated by the RFLP probe QcK5e06 (GenBank accession no. CF045980.1) in combination with the EcoRI restriction enzyme (Durand et al. 2010). Previous RFLP mapping results have shown that the probe QcK5e06 mapped on bin 6:01–6:02 (Falque et al. 2005). The probe QcK5e06 colocalized with two markers: umc85a (66.4 cM) and umc1006 (125 cM) (Chardon et al. 2004). The region spanning those two markers encompasses the centromere (Schnable et al. 2009). When blasted against the B73 reference genome (http://www.maizesequence.org/blast) using default parameters, QcK5e06 revealed a strong homology with the Eukaryotic Initiation Factor 4A (eIF-4A; mRNA GenBank accession no. NM_001111404.1, GeneID 541649) duplicated at two positions, one on chromosome 5 (identity = 99.39%, length = 328 pb) and one on chromosome 6 (identity = 87.74%, length = 310 pb). As expected, the position on chromosome 6 corresponded to bin 6:02. We searched for the closest EcoRI sites on each side of the probe at these two locations and found that only the locus located on chromosome 6 could generate the fragment length we observed in our RFLP assay. We further used the B73 genomic sequence of chromosome 6 to design primers and PCR amplify (Supporting Information, File S1) the region spanning the two EcoRI sites (4803 bp) located 3′ and 5′ of the eIF-4A gene in a sample of five Late-VL individuals (homozygous for the f252vl allele) and five Late-NVL individuals (homozygous for f252nvl). Direct sequencing of the PCR product was performed by GENOSCREEN (Lille, France). While we were able to PCR amplify all f252nvl alleles, we failed to PCR amplify the 3′ region of the gene in the f252vl alleles.

To circumvent this problem in f252vl alleles, we performed an inverse PCR reaction. We digested the genomic DNA with EcoRI, ligated the digested fragments to obtain circularized molecules, and designed primers in divergent orientation on QcK5e06 to PCR amplify the region spanning the two EcoRI sites. We applied the following protocol: 300 ng DNA of each f252vl sequence was digested using 1 unit EcoRI (Biolabs) and 1× Buffer NEBuffer2 in a total volume of 25 μl during 3 hr at 37°, followed by inactivation for 20 min at 65° . A 50-μl ligation mix [1 unit T4 DNA ligase (Biolabs), 0.7× Buffer NEBuffer2, and 1 mM ATP] was added to 15 μl of digested DNA. Ligation was performed at 16° for 16 hr and followed by an inactivation at 70° for 10 min. DNA was subsequently purified using the QIAGEN (Valencia, CA) mini elute kit according to manufacturer’s protocol.

Ligated DNA was PCR amplified using two sets of nested primers. The first PCR was performed in a final volume of 25 μl, using 1 μl of purified ligation mix, 2 mM MgSO4, 1× High Fidelity PCR Buffer (Invitrogen, Carlsbad, CA), 0.2 mM dNTP, 0.4 μM each primer, and 1 unit Platinium Taq DNA Polymerase High Fidelity (Invitrogen). The thermocycling program consisted of 2 min of denaturation at 95°; 33 cycles of 45 sec at 94°, 45 sec at 58°, and 4 min 30 sec at 68°; followed by 5 min at 68°. The second PCR was performed with 1 μl of the first PCR product as DNA matrix following the protocol described above. PCR products from the nested PCR were run on 1% agarose TAE1 X gel at 120 V. Direct sequencing of the PCR product was performed by GENOSCREEN. Primer sequences are indicated in File S1.

Characterization of the candidate region

We annotated the sequence of the f252vl and the f252nvl alleles using information from eIF-4A mRNA sequence, repeatmasker (http://www.repeatmasker.org), and the GIRI Repbase database (http://www.girinst.org/repbase/index.html) (Jurka et al. 2005). We determined the positions of the introns, the exons, and truncated transposable elements insertions. We determined insertions–deletions (indels), and single-nucleotide polymorphisms (SNPs) by comparing the f252vl and the f252nvl alleles to the B73 reference genome (Schnable et al. 2009). We called b73 the allele corresponding to the B73 reference genome. The per-site nucleotide diversity between f252vl, f252nvl, and b73 was estimated by π (Nei 1978) at all sites, using DnaSP v5.0 (Librado and Rozas 2009).

Development of a codominant marker

We used one SNP at position 5752 bp (Figure 2) to develop an allele-specific marker, named thereafter gsyeIF-4Ao3. The specific base at gsyeIF-4Ao3 is C for the f252nvl allele (hereafter called the C allele) and the b73 reference sequence and T for the f252vl allele (hereafter called the T allele). Two specific forward primers were employed: the 5′-TGTAAAACGACGGCCAGTAAACTAGCAGACCACAAATG-3′ primer hybridized only C alleles and contained an M13 tail and 5′-GAACTAGCAGGCCATAAATA-3′ hybridized only T alleles. The reverse primer 5′-CTGGTGATGCAAGGTGCTTA-3′ was designed on a monomorphic region between f252nvl and f252vl alleles and in consequence amplified both alleles. Because of the M13 tail, amplification of the C alleles results in a longer PCR fragment than amplification of T alleles. PCR reactions were performed in a final volume of 20 μl with 15–30 ng of DNA, 1.6 mM MgCl2, 1× PCR buffer, 0.2 mM dNTPs, 0.4 μM for primers (0.2 μM was used instead for the primer containing the M13 tail), and 1 unit of Taq polymerase synthesized according to the Desai and Pfaffle (1995) protocol. We used a touchdown thermocycling program consisting of 2 min of denaturation at 94° followed by eight cycles of 30 sec at 94°, 30 sec at 65° decreasing to 60° by a step of 0.7° per cycle, and 30 sec at 72°; followed by 25 cycles of 30 sec at 94°, 30 sec at 58°, and 30 sec at 72°; and a final extension of 5 min at 72°. Twenty microliters of PCR product were run on 4% metaphore agarose gels in 1× TBE for 5 hr at 180 V.

Figure 2
Annotation and polymorphisms of the f252vl, f252nvl, and b73 alleles at the eIF-4A locus. (A) f252vl, f252nvl, and b73 (from the B73 reference sequence) are represented by a thick line. The RFLP probe QcK5e06 is shown with a pink horizontal line and ...

Using the allele-specific PCR protocol, we were able to genotype the SNP in the Late F252 population of the DSE and the S2 parents of both S3 families (S-LateF252-T and S-LateF252-C), as well as an association panel encompassing 375 inbred lines (Camus-Kulandaivelu et al. 2006). We were able to confirm all results from the previous RFLP genotyping (Durand et al. 2010) and are therefore confident of the reliability of our allele-specific PCR assay.

Phenotypic evaluation of S3 families

One hundred seventy-two of 178 S3 families from the S-LateF252-T population and 90 of 94 S3 families from the S-LateF252-C population (Figure 1B) were evaluated in 2008 and 2009. A subsample of those S3 families, 109 and 71 from the S-LateF252-T and S-LateF252-C populations, respectively, was evaluated for both years. Each year, S3 families were grown in a randomized two-block design at Gif-sur-Yvette (France). This design consisted of replicating S3 families in two rows (one per block) of 40 individuals each at density 90,000 plants/ha in 2008 and in two rows (one per block) of 25 individuals each at density 75,000 in 2009. We measured the flowering time (corresponding to the time when both male and female had flowered, expressed as days after July 1) and the plant height (in centimeters from the ground to the base of the panicle). We counted the final number of leaves produced that was recorded taking into account the senescent leaves.

We decomposed phenotypic values of S3 families (Yijkl) using


where μ is the average value across S3 families, yi is the year of the experiment (i = 2008, 2009), popj is the population (j = S-LateF252-T, S-LateF252-C), genok is the S3 family (k = family number representing one S2 individual), blockil is the block (l = 1, 2) within each year of the experiment, and εijkl is the residual.

To perform subsequent analyses without having to incorporate a year effect at each step of the process, we decided to correct phenotypic data by the year effect as follows:


Phenotypic data, containing Zijkl values, are available following the procedure explained in File S2. Using phenotypic data corrected for the year effect, Zijkl, we determined the average phenotypic value of S3 families, the variance within population and between populations, and Pearson’s correlation coefficients between characters. We performed Student’s t-tests to compare significant differences between S-LateF252-T and S-LateF252-C populations. All analyses were conducted using R (R Development Core Team 2011).

Developmental characterization

We characterized the development of six genotypes from the DSE, three sampled from generation G7 and three sampled from generation G11 (Figure 1): two Early genotypes (homozygous for the f252nvl allele in the Early F252), two Late-NVL genotypes (homozygous for the f252nvl allele in the Late F252), and two Late-VL genotypes (homozygous for the f252vl allele in the Late F252). Two genotypes from the initial F252 seed lot were used as a control (both were homozygous for the f252nvl allele) for a total of eight genotypes. Each genotype was selfed to produce offspring. The offspring were divided into two sets.

One set of offspring from the eight genotypes was sown in five rows of 25 plants and used for the meristem developmental characterization. Every other day and during 42 days, 2 plants per genotype were dissected and the apical meristem was examined to (i) follow the dynamics of apparition of the number of initiated leaves (including leaf primordia) and (ii) determine the time of transition from the vegetative phase to the floral phase (floral transition) when the meristem starts elongating, referred to as stage C according to Irish and Nelson (1991). We used the number of initiated leaves through time to determine the time of floral transition by adjusting a two-phase segmented linear regression model (Burnham and Anderson 2002). The first phase and the corresponding slope corresponded to the rate of leaf emergence. The second phase described the cessation of leaf production when the meristem started its floral transition. The second phase therefore coincided with the final number of initiated leaves, when the leaf production reached a plateau. The time of the floral transition was estimated for each genotype from the regression analysis at the inflection point between the two phases. Time was expressed in degree days following Ritchie et al. (1991) (using the base temperature Tb = 6° and the optimum temperature To = 30° as parameter values).

A second set of offspring from the eight genotypes was sowed in two rows of 25 (50 plants per genotype) and used for the plant developmental characterization. We measured the final number of leaves and the plant height (in centimeters from the soil to the base of the panicle) as well as the number of individuals that had flowered (both male and female). The mean and standard deviation for flowering time were obtained on 50 individuals per genotype by adjusting the values to a cumulative Gaussian distribution. We performed pairwise Student’s t-tests to compare the rate of initiated leaves, the final number of initiated leaves, the time to floral transition, the final number of leaves, the plant height, and the flowering time between genotypes (Early, Late-NVL, Late-VL, and control) at generations G7 and G11. Phenotypic data used for the developmental characterization are available following the procedure explained in File S2. All analyses were conducted using R (R Development Core Team 2011).

Association mapping

We used the codominant PCR marker gsyeIF-4Ao3 (designed to reveal a T/C SNP) located in the region surrounding the eIF-4A gene to genotype an association panel of maize inbred lines. Originally the panel encompassed 375 inbred lines. It was phenotyped for 37 development (Table 1, see File S1) and kernel-related quality traits (Table 1) (Manicacci et al. 2009) and genotyped using 55 SSR markers (Camus-Kulandaivelu et al. 2006). On the basis of the SSR markers, Camus-Kulandaivelu and collaborators have established a structure matrix (Q) and a kinship matrix (K) and assigned each panel line to one of the five genetic groups: Tropicals (T), European Flints (EF), Northern Flints (NF), Corn Belt Dents (CBD), and Stiff Stalks (SS). In the present analysis, we considered 317 lines including 62 T, 57 EF, 50 NF, 133 CBD, and 15 SS lines for which amplification with the PCR codominant marker was successful. Note that the 317 lines encompassed 235 inbred lines with a group membership >75% and 82 lines with a mixed ancestry (group membership ≤75%).

Table 1
Abbreviations and descriptions of the traits measured on the panel

We tested the association between trait variation and the SNP, using a linear model to perform an analysis of covariance (ANCOVA) accounting for the SNP effect, the population structure Q, and the interaction SNP × Q. To assess the necessity of accounting for the kinship K in a mixed linear model, we tested the null hypothesis of the absence of the SNP effect under the linear model, using permutations of the genotypes at the SNP within each genetic group. A genetic group was assigned for each genotype by considering the highest probability of membership. We observed a uniform distribution of P-values (data not shown), suggesting that relatedness among lines within a genetic group (K) did not cause an inflation of the number of small P-values. It was therefore unnecessary to include K in the linear model, and we hence used the following equation for the rest of the analyses,


where Yij is the adjusted phenotypic value (see File S1) of the inbred line i and the allele j at the SNP, μ is the intercept, β(k) is the regression coefficient of the structure group k, αj measures the allelic effect of the SNP, and (αβ)j(k) is the effect of interaction between the group k and the allele j at the SNP. This model allowed us to test whether the effect of the SNP depended on the genetic group, thereby revealing possible epistatic interactions.

Model (3) has a hierarchical structure and we used Fisher’s test to compute the global significance of the SNP × Q interaction by comparing the residual mean square of the full model to the residual mean square of a model without interaction; i.e., (αβ)j(k)=0. Similarly, the significance of the SNP main effect was tested by comparing the residual mean square of the model without interaction, i.e., (αβ)j(k)=0, to the residual mean square of a model with only the genetic group effect, i.e., (αβ)j(k)=0 and αj = 0.

The same model was applied for each of the 37 phenotypic traits measured in the association panel. A false discovery rate (FDR) correction for multiple testing was applied using the R package fdrtool (Strimmer 2008). In our data sample, the 5% FDR corresponded to P-values <0.04. The coefficient of determination, RSNP2, which measured the proportion of the variation accounted for by both the SNP effect and the SNP × Q interaction, was determined by RSNP2=(SSSNP+SSSNP×Q)/SST, where SSSNP, SSSNP×Q, and SST are, respectively, the sum of squares of the SNP effect, the cumulated sum of squares for all interactions, and the total sum of squares from the ANCOVA (Equation 3). We deduced the corresponding percentage of RSNP2 explained by the interaction only.

For a subset of traits (10 of 37), the ANCOVA revealed a significant effect of the SNP among genetic groups, or of the SNP × genetic group interaction, or both. We discarded 2 of these traits, ear insertion height (EARHT) and leaf number above top ear (LFNBa), that were completely correlated with plant height (PTHT) and leaf number below top ear (LFNBb), respectively (data not shown). With the resulting 8 associated traits, we investigated the simultaneous effect of the SNP on the phenotypic variation by performing a set of multivariate analyses. We used model (3) to predict the least-squares mean of each inbred line Yij(k) for each trait and performed a principal component analysis (PCA). Inbred lines were projected on the first two PCA axes and colored according to their genetic group (T, EF, NF, CBD, or SS) and their genotype at the gsyeIF-4Ao3 SNP (T or C) (see File S2 for the resulting eigenvectors). We performed an ANCOVA (Equation 3), using the coordinates of the two first axes to test the SNP main effect and the SNP × Q interaction effect following the procedure described above on the 37 traits. We also computed the centroid of each origin (genetic group)-by-genotype (at gsyeIF-4Ao3) combination. The centroid represents the average phenotypic value of the genetic group, for either the T or the C allele at the gsyeIF-4Ao3 SNP. We obtained phenotypic landscapes of the T or the C allele by plotting the centroids of the 8 associated traits. The so-called radar plots were drawn for each genetic group independently, using the R package plotrix (Lemon et al. 2007). Finally we performed (i) a MANCOVA using the 8 associated traits (multivariate, Equation 3) and (ii) an ANCOVA separately for each of the two PCA axes (univariate, Equation 3). Both analyses were used to estimate the predicted phenotypic value Yj(k) of each of the 317 inbred lines by fixing their group k membership to 100% and considering allele j at the SNP. Within each group, Student’s t-tests were performed to compare the effects of the T vs. the C alleles at the T/C SNP. All analyses were performed using R (R Development Core Team 2011).


DNA sequence polymorphisms around the eIF-4A gene

We characterized the sequence variation revealed by the QcK5e06 RFLP probe between f252vl, f252nvl, and b73 alleles in the DSE. The probe QcK5e06 exhibited a strong homology to the Eukaryotic Initiation Factor 4A (eIF-4A; mRNA GenBank accession no. NM_001111404.1, GeneID 541649) duplicated on chromosome 5 and 6. We determined here that the polymorphism was located on chromosome 6 and used both the B73 reference genome and inverse PCR to isolate the complete sequence spanning the two EcoRI restriction sites and annotate the eIF-4A gene and the surrounding 3′ region (Figure 2). The latter contained two solo LTRs from the PREM2_ ZM_ LTR (score = 3910, accession no. AF090447) and the Copia6-ZM_I (score = 1079, accession no. CON1_ CON61) families, both belonging to the Copia superfamily. We also found an additional insertion in the most distal region of the f252vl allele encompassing an EcoRI site (Figure 2). This insertion is responsible for the generation of a fast RFLP allele. Because of the restricted sequence length of this insertion (60 bp), we were, however, unable to determine its origin.

Overall, the annotation of the region in both the f252vl and the f252nvl alleles revealed a surprisingly high level of sequence and structural variation in and beyond the 3′-UTR region of eIF-4A (Figure 2). While the amount of nucleotide diversity between f252nvl and b73 was low (π = 0.03%), and structural variation was consistent with previous observations in maize, i.e., three small indels (two of 1 bp and one of 14 bp) over the entire length of the region (2866 bp), the level of nucleotide and structural variation between f252vl and f252nvl was unusually high. Hence the average pairwise nucleotide difference was close to 5.7%, 60-fold higher than the pairwise nucleotide difference observed between f252nvl and b73. Note that both nucleotide and structural variation between f252vl and f252nvl were unequally distributed with no variants detected in exon 4 of the eIF-4A gene, relatively few in the 3′-UTR, and many downstream of the 3′-UTR, a region bearing intricate insertions of transposable elements.

We used sequence information to develop an SNP-based codominant specific PCR marker (gsyeIF-4Ao3) revealing a T/C polymorphism (Figure 2). This marker was used to genotype the SNP in the Late F252 population of the DSE. Note that while T and C alleles at the SNP were associated, respectively, with the f252vl and the f252nvl allele in the DSE (Figure 1), we did not check this association in other types of material such as the association panel; we therefore refer to the T and C alleles instead of f252vl and f252nvl to avoid confusion.

Phenotypic and developmental variation associated with gsyeIF-4Ao3 polymorphism in the F252 genetic background

We undertook two approaches to compare the phenotypic effect of the T and C alleles at SNP gsyeIF-4Ao3. First, we performed a field evaluation of S3 families fixed for the T (S-LateF252-T) and the C allele (S-LateF252-C), respectively (Figure 1). Second, we characterized the development of one Early (homozygous for the C allele), one Late-NVL (homozygous for the C allele), one Late-VL (homozygous for the T allele), and one control F252 genotype (homozygous for the C allele) from the DSE populations at generations G7 and G11 (Figure 1).

Results of the first approach are shown in Figure 3. As expected, both S-LateF252-C and S-LateF252-T flowered later than the control but, quite surprisingly, the two populations did not differ significantly (P-value = 0.472) for flowering time (Figure 3A). This result contrasts with our previous study that revealed an association between the candidate locus (RFLP marker Qck5e06) and flowering time in the Late F252 genealogy (Durand et al. 2010). We found, however, significant differences between S-LateF252-C and S-LateF252-T for plant height and final number of leaves (both P-values <10−15; Figure 3, B and C), the individuals of the S-LateF252-T population being taller and producing on average one more leaf than those of the S-LateF252-C population.

Figure 3
Distribution of flowering date (A), plant height (B), and final number of leaves (C) in the S-LateF252-C and S-LateF252-T populations. S-LateF252-C and S-LateF252-T populations are represented, respectively, with shaded and hatched bars. The average phenotypic ...

The tendency of the Late-VL genotypes issued from the Late F252 DSE population to produce significantly more leaves (Figure 4B) was also confirmed in the second developmental approach (both P-values <10−15). But in contrast to the S-LateF252-T/S-LateF252-C comparison, it translated into a later flowering (P-value <10−15, Figure 4A). For instance, at G7 for a sum of temperature of 844 (corresponding to 83 days after seedling), none of the Late-VL genotypes flowered compared to 80%, 97%, and 87% for Late-NVL, control, and Early genotypes, respectively. At G11, for a sum of temperature of 844 only 3% of Late-VL flowered compared to 91%, 95%, and 95% for Late-NVL, control, and Early, respectively. We measured the number of initiated leaves and the apical meristem stage every 2 days to determine the time to flowering transition and the rate of leaf initiation as shown in Figure 4C. The values reported in Figure 4D confirmed a significant delay in floral transition in Late-VL compared to the other genotypes (P-value <10−15). No significant difference in the rate of initiated leaves and in the final plant height (data not shown) was observed between genotypes having the C or the T allele (P-value = 0.29 and P-value = 0.49, respectively).

Figure 4
Developmental comparison of Early, Late-NVL, Late-VL, and control genotypes chosen from the divergent selection experiment at generations G7 and G11. Time is expressed in sum of temperature. (A) Cumulative distribution of flowered plants (i.e., both anthesis ...

Association mapping

To validate further the association between allelic variation at eIF-4A and variation of quantitative traits related to plant development and architecture, and to test for interaction with the genetic background at the species level, we genotyped the gsyeIF-4Ao3 SNP in 317 inbred maize lines (Camus-Kulandaivelu et al. 2006). This panel was structured in five genetic groups (Tropical, European Flint, Northern Flint, Corn Belt Dent, and Stiff Stalk) and was evaluated for 37 traits (Table 1).

The frequency of both alleles in each of the five genetic groups is indicated in Figure 5. Considering all genotypes, the minor allele frequency ranged from 14% to 42% among groups and therefore both alleles segregated at intermediate frequency (>10%). At a 5% FDR, we found seven traits associated with a main effect of the SNP (Table 2): the leaf number (total, LFNB; and below the ear, LFNBb), the female and male flowering time (FFLW8 and MFLW8), the embryo to kernel weight (EMB), the plant height (PTHT), and the saturated fatty acid (SATUR). For these seven traits, the model explained a large part of the phenotypic variance, respectively 52.81%, 50.91%, 52.45%, 48.66%, 19.20%, 30.50%, and 23.21% for LFNB, LFNBb, MFLW8, FFLW8, EMB, PTHT, and SATUR (Table 2). The effect associated with both the SNP main effect (among groups) and the SNP × group interaction, RSNP2, was generally low for these traits (between 1.94% and 3.52%, Table 2) and the SNP × group interaction accounted for >33.53% of the effect (Table 2). In addition, we found one trait with no significant SNP main effect but a significant SNP × group interaction: BRNB, the panicle branch number (P-value = 0.0149, Table 2). For that trait, 97.16% of the RSNP2 was explained by this interaction (Table 2). In sum, we found that the gsyeIF-4Ao3 SNP was associated with the variation of eight traits and revealed significant interactions with the genetic background.

Figure 5
Frequencies of the T (solid) and C (shaded) alleles at SNP gsyeIF-4Ao3 among the five genetic groups of the association mapping panel. The panel comprises 317 inbred lines with 62 Tropicals (T), 57 European Flints (EF), 50 Northern Flints (NF), 133 Corn ...
Table 2
Significant traits and PCA axes associated with SNP gsyeIF-4Ao3 and interactions with genetic background

For each allele (T or C), we determined the least-squares mean (lsmeans) within group for the eight associated traits using the parameters of the linear model (3). We next performed Student’s t-tests to compare T and C lsmeans and evaluated the sign and significance of the difference in the effect of the two alleles (Table 3). For each trait, there was at least one genetic group where T and C genotypes had a significantly different phenotypic mean. These results were consistent with the ANCOVA results (Table 2) and revealed genotype-by-group interactions. Note, for instance, the change of the T allelic effect sign for the panicle branch number (BRNB) between the Tropical group (−0.59) and the European-Flint group (+0.42). Likewise for PTHT we observed a change of the T allelic sign between the Corn Belt Dents and the European Flints, the main effect of the latter being borderline significant (P-value = 0.0669, Table 3). The F252 line, from which we started the divergent selection experiment, belongs to the Corn Belt Dent group in which the T was associated with earlier flowering, smaller plants, and the production of fewer leaves than the C allele (Table 3). In the divergent selection experiment instead the T allele was associated with late flowering and the production of additional leaves. Such an inversion of effect was again consistent with pervasive interactions between the alleles at gsyeIF-4Ao3 and the genetic background.

Table 3
Effect of the T allele at SNP gsyeIF-4Ao3 within genetic groups for eight associated traits and the first two PCA axes

We performed a principal component analysis on the predicted phenotypic values (Equation 3) of the 317 inbred lines for the eight associated traits (LFNB, MFLW8, LFNBb, FFLW8, PTHT, SATUR, BRNB, and EMB). The two main PCA axes explained 94% of the total phenotypic variation (Figure 6A). As shown in Figure 6B, the major contributors to the first PCA axis were traits related to flowering time, leaf number, and plant height. Hence, the first axis clustered groups according to their flowering time: early-flowering European and Northern Flints, intermediate-flowering Stiff Stalks and Corn Belt Dents, and late-flowering Tropicals. Along this axis, however, the T allele had a direction of effect that depended on genetic group (Figure 6A). For instance, the average phenotypic value for the T allele was higher than that for the the C allele in the European Flint group and the reverse was true in the Corn Belt Dent group. The ANCOVA (Table 3) actually revealed a positive direction of effect of the T allele in European Flints (+0.74) but negative (−0.97) in Corn Belt Dents. Overall the T allele was significant in the Corn Belt Dent group either when testing the first PCA axis coordinates or when testing the traits contributing to this axis (PCA1, LFNB, MFLW8, LFNBb, FFLW8, and PTHT, Table 3). The second axis explained 23% of the variation and was mostly determined by a combination of traits related to the kernel development and content. In the Tropicals, the T allele was significant for the second PCA axis (PCA2, Table 3) and also for the traits that mainly contribute to that axis when tested separately (SATUR, BRNB, and EMB, Table 3). Overall, these observations were consistent with a significant SNP main effect along the two first PCA axes (PCA1 and PCA2, Table 2) and a borderline significant SNP × Q effect (Table 2).

Figure 6
Multivariate analysis using the predicted phenotypic values of the 317 inbred lines for the eight associated traits. (A) Principal component analysis with the first two axes explaining 71% and 23% of the variation (94% total). Genotypes that have the ...

We used the eight associated traits (seven with a global effect among groups and one with an SNP × group effect, Equation 3) to draw the phenotypic landscapes of the T and the C alleles at gsyeIF-4Ao3 in each genetic group (Figure 6C). While the per-trait analysis revealed a complex pattern of effects (Table 2), this multitrait analysis clearly revealed differences between the phenotypic landscapes of T and C alleles in Corn Belt Dent, Stiff-Stalk, European Flint, and Tropical origins. In contrast, the Northern Flints displayed a similar landscape for alleles T and C (closer to the T allele of European Flints), except for plant height (PTHT, Figure 6C).

Overall, the association analysis revealed first the existence of strong interactions of the T and the C alleles with the genetic background. Second, while the effect of the alleles on each trait was relatively small and borderline significant, the multitrait analysis revealed much sharper patterns.


The motivation of the present study was to characterize a candidate locus for flowering time in maize and to investigate the genetic mechanisms underlying its contribution to phenotypic variation. We combined multiple approaches from developmental characterization to QTL and association mapping to dissect the relation between the genotype at this candidate locus and phenotypic variation.

Because the time of floral transition coincides with the end of the plant vegetative growth, flowering time correlates with a number of physiological and phenological variables. We therefore investigated the association between the candidate polymorphism and the variation of several quantitative traits linked to development, weight, and quality of kernels in an association panel of maize, genotyped for a SNP belonging to the candidate region. The two alleles were common (Figure 5) with a minor allele frequency >14% in each genetic group (Tropicals, European Flints, Northern Flints, Corn Belt Dents, and Stiff Stalks). Because the variant was of potentially large effect (35% of the phenotypic variation in the Late F252) and the genetic structure of the association panel has been well characterized (Camus-Kulandaivelu et al. 2006), conditions were favorable to test for the association between the candidate locus and phenotypic variations. We found a significant association among groups for seven traits including leaf number, leaf number below the ear, female and male flowering time, embryo to kernel weight, plant height, and saturated fatty acid (Table 2). Corroborating these results, the candidate locus, located in bin 6.02, not only colocalized with a QTL for flowering time in meta-analysis (Chardon et al. 2004) but also colocalized with one QTL for the number of leaves (Koester et al. 1993) and with three QTLs for plant height (Beavis et al. 1991; CIMMYT 1994; Openshaw et al. 1997). Altogether we therefore confirmed the association between the candidate locus and a number of traits including flowering time. The allele effect was generally low on all traits (<3.5%, Table 2). Since flowering time is intimately linked to population structure, correcting for it probably led to an underestimation of the measured effect (Yu et al. 2006). The same lack of power has been documented by Camus-Kulandaivelu et al. (2008) when estimating the effect of two SNPs (B173 and B316) associated with flowering time variation located in the tb1-dwarf8 region in maize. Their effect ranged from 1.55% to 1.60% when accounting for underlying genetic structure, but was estimated at 10.3% when structure was not accounted for (Camus-Kulandaivelu et al. 2008).

Correlation among various traits associated with the SNP (gsyeIF-4Ao3) was well described by the multivariate analysis (Figure 6). Hence, the first PCA axis that explained 71% of the phenotypic variation was determined essentially by correlated developmental traits such as PTHT, MFLW8 and FFLW8, and traits related to the leaf number (LFNB and LFNBb) (Figure 6B). Such correlations may result from pleiotropy as defined by Chen and Lubberstedt (2010): (i) a single gene affecting different traits, (ii) several polymorphisms within a gene affecting different traits, or (iii) several genes in tight linkage affecting different traits. Pleiotropy is frequent for genes involved in the control of flowering time. For example, in A. thaliana, nonfunctional alleles at the FRIGIDA (FRI) gene conferring early flowering (Johanson et al. 2000) also decrease the number of nodes and branches on the inflorescence (Scarcelli et al. 2007). In maize, the vgt1 mutation corresponding to an intergenic region 70 kb upstream of an APETALA2 (AP2)-like transcription factor reduces the flowering time and the node number (Salvi et al. 2007). Likewise, intragenic linkage between causal polymorphisms has been documented in two genes controlling variation in flowering time: dwarf8 (Koester et al. 1993; Schön et al. 1994; Peng et al. 1999; Thornsberry et al. 2001; Andersen et al. 2005) and vgt1 (Buckler et al. 2009).

One of the most interesting findings of our study is the variation in sign and magnitude of allelic effect on flowering time. Indeed, the effect of the T allele on flowering time was (i) positive, i.e., leading to a later flowering of the Late-VL (homozygous for T allele) compared to Late-NVL genotypes (homozygous for C) in the developmental characterization of the divergent selection experiment (Figure 4A); (ii) not significant on flowering time when comparing S-LateF252-T and S-LateF252-C populations (Figure 3A); and (iii) negative, i.e., conferring an earlier flowering in the T inbreds compared to C inbreds, in the Corn Belt Dent group of the association panel (Table 3, Figure 6A). The correlated response on plant height and the number of leaves was visible in the association panel (Figure 6B). However, depending on the genetic group, the T allele was either associated with early flowering and smaller plants with less leaves (CBD inbreds) or the reverse, late flowering, taller plants, and more leaves in European Flints (Figure 6A). In contrast Late-VL genotypes (T allele) in the DSE at G7 and G11 were late flowering (Figure 4A) and produced more leaves (Figure 4B) than Late-NVL genotypes (C allele) while having the same plant height (data not shown). Finally, S-LateF252-T families produced more leaves and were significantly taller than S-LateF252-C families while having the same flowering time (Figure 3).

The consistency of the statistical association of the T allele with variation of at least one trait in all materials coupled with significant interactions with the genetic background strongly suggests that multiple determinants, including our candidate region, probably in tight linkage, are responsible for the observed phenotypic variation (Figure 7). We found two lines of arguments consistent with this hypothesis. First, there was an extremely high sequence divergence between f252vl and f252nvl alleles (5.7% at the nucleotide level), suggesting an overall lack of recombination in that region, perhaps causing long-distance linkage disequilibrium (LD). The extent of these haplotypes and their association with the SNP in S3 families and in the association panel remain to be elucidated; i.e., because this region appears depauperate from SNP characterization in the current available data sets (HapmapV1 data set downloaded at http://www.panzea.org/), we were unable to evaluate LD. Second, the closest gene to the eIF-4A gene located 155 kb away is a ubiquitin carrier protein (GRMZM2G102421). In A. thaliana, double mutants for two enzymes involved in the histone monoubiquitination (ubc1 and ubc2) exhibit early flowering (Cao et al. 2008).

Figure 7
Variation of the sign of the effect of the f252vl allele with the genetic background. Plant materials are genotypes derived from the DSE experiment used for developmental characterization (left plot), S-LateF252-C and S-LateF252-T populations (center ...

While a cluster of tightly linked genes dominated by additive effects and occasional recombination could explain the inversion of effect of the alleles between different types of materials, epistasis could also account for the interaction effects that we have detected in the association panel. Although flowering time in maize is believed to be a trait dominated by additive interactions, i.e., <2% as estimated by GWAS (Buckler et al. 2009; Brachi et al. 2010), the power to detect those interactions in GWAS is extremely low after correcting for multiple tests (Mackay et al. 2009). Furthermore, epistatic interactions in the control of flowering time have been documented in A. thaliana (Caicedo et al. 2004), sunflower (Blackman et al. 2010), and maize (Vega et al. 2002). Interestingly, Buckler et al. (2009) have reported the presence of multiallelic series at the vgt1 locus and showed that several sites within a locus contribute to flowering time variation. In our case, it is possible that the T allele is combined with other site polymorphisms generating intragenic epistasis that causes, for instance, the observed inversion of effect in the CBD, a material known to have multiple origins (Doebley et al. 1988).

We therefore argue that pleiotropy in the form of a cluster of tightly linked genes and epistasis contribute to the phenotypic variation for flowering time and can produce the complex patterns we observed. The dissection of this genetic system is far from being complete, but our collection of approaches has brought new insights into the genetic bases of flowering time. Given the apparent low level of historical recombination in the region and the expected amount of LD, it will be difficult to identify with precision the underlying causal mutation(s) at this locus. Nevertheless eIF-4A appears to be a strong candidate. This gene belongs to the eukaryotic initiation translation factor (eIF) family and is involved in the recruitment of one of the ribosomal subunits (40s) with the mRNA. Feng et al. (2007) have demonstrated that a loss-of-function mutation in the eukaryotic translation initiation factor 5A (eIF-5A) gene strongly alters A. thaliana growth and development, reducing the plant size and the number of all adult organs. Recently, Lellis et al. (2010) have shown that double knockout A. thaliana mutants for the eukaryotic translation initiation factor 4G (eIF-4G) flowered on average 10 days later than the wild types, with a reduction in germination and growth rate. eIF-4A is known to directly interact with eIF-4G (reviewed in Kawaguchi and Bailey-Serres 2002). If indeed eIF-4A is involved in the phenotypic variation for plant height and number of leaves, it raises the interesting possibility that sequence variation in the 3′-UTR region may actually contain causal polymorphisms. Hence, this region exhibits a number of polymorphisms (six SNPs and four small indels) between f252vl and f252nvl alleles while exon 4 did not contain a single polymorphism. 3′-UTR regions can interact with miRNA and play a role in the silencing of gene expression by mRNA degradation or translational inhibition in the presence of target (Huntzinger and Izaurralde 2011). We performed a BLAST search with the 3′-UTR region against a miRNA database (http://sundarlab.ucdavis.edu/smrnas/) but did not find evidence for such target sites. Given the present limited knowledge and data on miRNA regulation sites and underlying mechanisms, this hypothesis cannot be discarded but would need further validation through allele-specific expression assays. While the nearly sequenced Mo17 maize inbred line displays the T allele in the association panel, sequence data are not yet available in the 3′-UTR of the eiF-4A gene (http://www.panzea.org/). We will therefore need to sequence BACs using next generation technology to better characterize the surrounding structural and sequence polymorphisms. The study of the molecular evolution of this region in an extended panel of maize inbred lines will give precious insights about the extent of LD in the region, the haplotype structure, and its association with phenotypic variation.


We thank Céline Ridel, Carine Remoué, and Matthieu Falque for their help in the gsyeIF -4Ao3 marker development. We are grateful to Aurélie Bourgais, Delphine Madur, Valérie Combes, and Fabrice Dumas for their contribution to the DNA extractions and/or genotyping and Cédric Dufour, Sophie Jouanne, and Denis Coubriche who participated in field experiments. We also thank two anonymous reviewers for their constructive comments on the manuscript and Steve Chenoweth, the associate editor, for helping us to improve statistical analyses. Field experiments were financed by Institut National de la Recherche Agronomique (INRA). E.D. was supported by a Ph.D. fellowship from INRA and Centre National de la Recherche Scientifique.


Communicating editor: S. F. Chenoweth

Literature Cited

  • Andersen J. R., Schrag T., Melchinger A. E., Zein I., Lübberstedt T., 2005.  Validation of Dwarf8 polymorphisms associated with flowering time in elite european inbred lines of maize (zea mays l.). Theor. Appl. Genet. 111: 206–217. [PubMed]
  • Atwell S., Huang Y. S., Vilhjálmsson B. J., Willems G., Horton M., et al. , 2010.  Genome-wide association study of 107 phenotypes in Arabidopsis thaliana inbred lines. Nature 465: 627–631. [PMC free article] [PubMed]
  • Azevedo R. B. R., Keightley P. D., Laurén-Määttä C., Vassilieva L. L., Lynch M., et al. , 2002.  Spontaneous mutational variation for body size in Caenorhabditis elegans. Genetics 162: 755–765. [PMC free article] [PubMed]
  • Beavis W. D., Grant D., Albertsen M., Fincher R., 1991.  Quantitative trait loci for plant height in four maize populations and their associations with qualitative genetic loci. Theor. Appl. Genet. 83: 141–145. [PubMed]
  • Bergelson J., Roux F., 2010.  Towards identifying genes underlying ecologically relevant traits in Arabidopsis thaliana. Nat. Rev. Genet. 11: 867–879. [PubMed]
  • Blackman B. K., Strasburg J. L., Raduski A. R., Michaels S. D., Rieseberg L. H., 2010.  The role of recently derived FT paralogs in sunflower domestication. Curr. Biol. 20: 629–635. [PMC free article] [PubMed]
  • Bomblies K., Doebley J. F., 2006.  Pleiotropic effects of the duplicate maize FLORICAULA/LEAFY genes zfl1 and zfl2 on traits under selection during maize domestication. Genetics 172: 519. [PMC free article] [PubMed]
  • Brachi B., Faure N., Horton M., Flahauw E., Vazquez A., et al. , 2010.  Linkage and association mapping of Arabidopsis thaliana flowering time in nature. PLoS Genet. 6: e1000940. [PMC free article] [PubMed]
  • Buckler E. S., Holland J. B., Bradbury P. J., Acharya C. B., Brown P. J., et al. , 2009.  The genetic architecture of maize flowering time. Science 325: 714–718. [PubMed]
  • Burnham K. P., Anderson D. R., 2002.  Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer-Verlag, Berlin/Heidelberg, Germany/New York.
  • Caicedo A. L., Stinchcombe J. R., Olsen K. M., Schmitt J., Purugganan M. D., 2004.  Epistatic interaction between Arabidopsis FRI and FLC flowering time genes generates a latitudinal cline in a life history trait. Proc. Natl. Acad. Sci. USA 101: 15670–15675. [PMC free article] [PubMed]
  • Camus-Kulandaivelu L., Veyrieras J. B., Madur D., Combes V., Fourmann M., et al. , 2006.  Maize adaptation to temperate climate: relationship between population structure and polymorphism in the Dwarf8 gene. Genetics 172: 2449–2463. [PMC free article] [PubMed]
  • Camus-Kulandaivelu L., Chevin L. M., Tollon-Cordet C., Charcosset A., Manicacci D., et al. , 2008.  Patterns of molecular evolution associated with two selective sweeps in the Tb1–Dwarf8 region in maize. Genetics 180: 1107–1121. [PMC free article] [PubMed]
  • Cao Y., Dai Y., Cui S., Ma L., 2008.  Histone h2b monoubiquitination in the chromatin of FLOWERING LOCUS C regulates flowering time in Arabidopsis. Plant Cell Online 20: 2586–2602. [PMC free article] [PubMed]
  • Chardon F., Virlon B., Moreau L., Falque M., Joets J., et al. , 2004.  Genetic architecture of flowering time in maize as inferred from quantitative trait loci meta-analysis and synteny conservation with the rice genome. Genetics 168: 2169–2185. [PMC free article] [PubMed]
  • Chen Y., Lubberstedt T., 2010.  Molecular basis of trait correlations. Trends Plant Sci. 15: 454–461. [PubMed]
  • CIMMYT, 1994.  QTL data for populations Ki3 × CML139 and CML131 × CML67. Maize Genome Database. Available at: http://www.agron.missouri.edu.
  • Colasanti J., Muszynski M., 2009.  The maize floral transition, pp. 41–55 in Handbook of Maize: Its Biology, edited by J. Bennetzen and S. Hake. Springer, New York.
  • Colasanti J., Yuan Z., Sundaresan V., 1998.  The indeterminate gene encodes a zinc finger protein and regulates a leaf-generated signal required for the transition to flowering in maize. Cell 93: 593–603. [PubMed]
  • Coles N. D., McMullen M. D., Balint-Kurti P. J., Pratt R. C., Holland J. B., 2010.  Genetic control of photoperiod sensitivity in maize revealed by joint multiple population analysis. Genetics 184: 799–812. [PMC free article] [PubMed]
  • Desai U. J., Pfaffle P. K., 1995.  Single-step purification of a thermostable dna polymerase expressed in Escherichia coli. Biotechniques 19: 780–784. [PubMed]
  • Doebley J., Wendel J., Smith J., Stuber C., Goodman M., 1988.  The origin of cornbelt maize: the isozyme evidence. Econ. Bot. 42: 120–131.
  • Doerge R. W., 2002.  Mapping and analysis of quantitative trait loci in experimental populations. Nat. Rev. Genet. 3: 43–52. [PubMed]
  • Ducrocq S., Giauffret C., Madur D., Combes V., Dumas F., et al. , 2009.  Fine mapping and haplotype structure analysis of a major flowering time quantitative trait locus on maize chromosome 10. Genetics 183: 1555–1563. [PMC free article] [PubMed]
  • Dudley J. W., Lambert R., 2004.  100 generations of selection for oil and protein in corn. Plant Breed. Rev. 24: 79–110.
  • Durand E., Tenaillon M. I., Ridel C., Coubriche D., Jamin P., et al. , 2010.  Standing variation and new mutations both contribute to a fast response to selection for flowering time in maize inbreds. BMC Evol. Biol. 10: 2. [PMC free article] [PubMed]
  • Falque M., Décousset L., Dervins D., Jacob A. M., Joets J., et al. , 2005.  Linkage mapping of 1454 new maize candidate gene loci. Genetics 170: 1957–1966. [PMC free article] [PubMed]
  • Feng H., Chen Q., Feng J., Zhang J., Yang X., et al. , 2007.  Functional characterization of the Arabidopsis eukaryotic translation initiation factor 5a – 2 that plays a crucial role in plant growth and development by regulating cell division, cell growth, and cell death. Plant Physiol. 144: 1531–1545. [PMC free article] [PubMed]
  • Huntzinger E., Izaurralde E., 2011.  Gene silencing by micrornas: contributions of translational repression and mrna decay. Nat. Rev. Genet. 12: 99–110. [PubMed]
  • Irish E. E., Nelson T. M., 1991.  Identification of multiple stages in the conversion of vegetative to floral development. Development 112: 891–898.
  • Johanson U., West J., Lister C., Michaels S., Amasino R., et al. , 2000.  Molecular analysis of FRIGIDA, a major determinant of natural variation in Arabidopsis flowering time. Science 290: 344–347. [PubMed]
  • Johansson A. M., Pettersson M. E., Siegel P. B., Carlborg Ö., 2010.  Genome-wide effects of long-term divergent selection. PLoS Genet. 6: e1001188. [PMC free article] [PubMed]
  • Jurka J., Kapitonov V. V., Pavlicek A., Klonowski P., Kohany O., et al. , 2005.  Repbase Update, a database of eukaryotic repetitive elements. Cytogenet. Genome Res. 110: 462–467. [PubMed]
  • Kawaguchi R., Bailey-Serres J., 2002.  Regulation of translational initiation in plants. Curr. Opin. Plant Biol. 5: 460–465. [PubMed]
  • Keightley P. D., 1998.  Genetic basis of response to 50 generations of selection on body weight in inbred mice. Genetics 148: 1931–1939. [PMC free article] [PubMed]
  • Koester R. P., Sisco P. H., Stuber C. W., 1993.  Identification of quantitative trait loci controlling days to flowering and plant height in two near isogenic lines of maize. Crop Sci. 33: 1209–1216.
  • Kump K. L., Bradbury P. J., Wisser R. J., Buckler E. S., Belcher A. R., et al. , 2011.  Genome-wide association study of quantitative resistance to southern leaf blight in the maize nested association mapping population. Nat. Genet. 43: 163–168. [PubMed]
  • Laurie C. C., Chasalow S. D., LeDeaux J. R., McCarroll R., Bush D., et al. , 2004.  The genetic architecture of response to long-term artificial selection for oil concentration in the maize kernel. Genetics 168: 2141–2155. [PMC free article] [PubMed]
  • Lellis A. D., Allen M. L., Aertker A. W., Tran J. K., Hillis D. M., et al. , 2010.  Deletion of the eIFiso4G subunit of the Arabidopsis eIFiso4F translation initiation complex impairs health and viability. Plant Mol. Biol. 74: 249–263. [PMC free article] [PubMed]
  • Lemon J., Bolker B., Oom S., Klein E., Rowlingson B., et al. , 2007.  plotrix: various plotting functions. Available at http://cran. r-project. org/src/contrib/Descriptions/plotrix. html.
  • Librado P., Rozas J., 2009.  DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25: 1451–1452. [PubMed]
  • Mackay T. F. C., Lyman R. F., Lawrence F., 2005.  Polygenic mutation in Drosophila melanogaster. Genetics 170: 1723–1735. [PMC free article] [PubMed]
  • Mackay T. F. C., Stone E. A., Ayroles J. F., 2009.  The genetics of quantitative traits: challenges and prospects. Nat. Rev. Genet. 10: 565–577. [PubMed]
  • Manenti G., Galvan A., Pettinicchio A., Trincucci G., Spada E., et al. , 2009.  Mouse genome-wide association mapping needs linkage analysis to avoid false-positive loci. PLoS Genet. 5: e1000331. [PMC free article] [PubMed]
  • Manicacci D., Camus-Kulandaivelu L., Fourmann M., Arar C., Barrault S., et al. , 2009.  Epistatic interactions between Opaque2 transcriptional activator and its target gene CyPPDK1 control kernel trait variation in maize. Plant Physiol. 150: 506. [PMC free article] [PubMed]
  • Manolio T. A., Collins F. S., Cox N. J., Goldstein D. B., Hindorff L. A., et al. , 2009.  Finding the missing heritability of complex diseases. Nature 461: 747–753. [PMC free article] [PubMed]
  • Meng X., Muszynski M. G., Danilevskaya O. N., 2011.  The FT-like ZCN8 gene functions as a floral activator and is involved in photoperiod sensitivity in maize. Plant Cell Online 23: 942–960. [PMC free article] [PubMed]
  • Miller T. A., Muslin E. H., Dorweiler J. E., 2008.  A maize CONSTANS-like gene, conz1, exhibits distinct diurnal expression patterns in varied photoperiods. Planta 227: 1377–1388. [PubMed]
  • Muszynski M. G., Dam T., Li B., Shirbroun D. M., Hou Z., et al. , 2006.  Delayed flowering1 encodes a basic leucine zipper protein that mediates floral inductive signals at the shoot apex in maize. Plant Physiol. 142: 1523–1536. [PMC free article] [PubMed]
  • Myles S., Peiffer J., Brown P. J., Ersoz E. S., Zhang Z., et al. , 2009.  Association mapping: critical considerations shift from genotyping to experimental design. Plant Cell Online 21: 2194–2202. [PMC free article] [PubMed]
  • Nei M., 1978.  Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: 583–590. [PMC free article] [PubMed]
  • Openshaw S., Frascaroli E., Stefani T., Hobart B., Arbuckle J. A., 1997.  How many genes control plant height in maize? Maize Genetics Conference Abstract. 39: 1.
  • Peng J., Richards D. E., Hartley N. M., Murphy G. P., Devos K. M., et al. , 1999.  ‘green revolution’ genes encode mutant gibberellin response modulators. Nature 400: 256–261. [PubMed]
  • Poland J. A., Bradbury P. J., Buckler E. S., Nelson R. J., 2011.  Genome-wide nested association mapping of quantitative resistance to northern leaf blight in maize. Proc. Natl. Acad. Sci. USA 108: 6893. [PMC free article] [PubMed]
  • Pritchard J. K., Stephens M., Rosenberg N. A., Donnelly P., 2000.  Association mapping in structured populations. Am. J. Hum. Genet. 67: 170–181. [PMC free article] [PubMed]
  • R Development Core Team, 2011.  R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna.
  • Ritchie J. T., Nesmith D. S., 1991.  Temperature and crop development, pp. 5-29 in Modeling Plant and Soil Systems, edited by J. Hanks and J. T. Ritchie. ASA, CSSA and SSSA, Madison, WI.
  • Salvi S., Sponza G., Morgante M., Tomes D., Niu X., et al. , 2007.  Conserved noncoding genomic sequences associated with a flowering-time quantitative trait locus in maize. Proc. Natl. Acad. Sci. USA 104: 11376–11381. [PMC free article] [PubMed]
  • Scarcelli N., Cheverud J. M., Schaal B. A., Kover P. X., 2007.  Antagonistic pleiotropic effects reduce the potential adaptive value of the FRIGIDA locus. Proc. Natl. Acad. Sci. USA 104: 16986–16991. [PMC free article] [PubMed]
  • Schnable P. S., Ware D., Fulton R. S., Stein J. C., Wei F., et al. , 2009.  The B73 maize genome: complexity, diversity, and dynamics. Science 326: 1112–1115.
  • Schön C. C., Melchinger A. E., Boppenmaier J., Brunklaus-Jung E., Herrmann R. G., et al. , 1994.  RFLP mapping in maize: quantitative trait loci affecting testcross performance of elite European flint lines. Crop Sci. 34: 378–389.
  • Strimmer K., 2008.  fdrtool: a versatile r package for estimating local and tail area-based false discovery rates. Bioinformatics 24: 1461. [PubMed]
  • Thornsberry J. M., Goodman M. M., Doebley J., Kresovich S., Nielsen D., et al. , 2001. Dwarf8 polymorphisms associate with variation in flowering time. Nat. Genet. 28: 286–289. [PubMed]
  • Tian F., Bradbury P., Brown P., Hung H., Sun Q., et al. , 2011.  Genome-wide association study of leaf architecture in the maize nested association mapping population. Nat. Genet. 43: 159–162. [PubMed]
  • Vega S. H., Sauer M., Orkwiszewski J. A. J., Poethig R. S., 2002.  The early phase change gene in maize. Plant Cell Online 14: 133–147. [PMC free article] [PubMed]
  • Yu J., Pressoir G., Briggs W. H., Vroh Bi I., Yamasaki M., et al. , 2006.  A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nat. Genet. 38: 203–208. [PubMed]
  • Yu J., Holland J. B., McMullen M. D., Buckler E. S., 2008.  Genetic design and statistical power of nested association mapping in maize. Genetics 178: 539–551. [PMC free article] [PubMed]
  • Zhao K., Aranzana M. J., Kim S., Lister C., Shindo C., et al. , 2007.  An Arabidopsis example of association mapping in structured samples. PLoS Genet. 3: e4. [PMC free article] [PubMed]

Articles from Genetics are provided here courtesy of Genetics Society of America


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • EST
    Published EST sequences
  • Gene (nucleotide)
    Gene (nucleotide)
    Records in Gene identified from shared sequence links
  • MedGen
    Related information in MedGen
  • Nucleotide
    Published Nucleotide sequences
  • PubMed
    PubMed citations for these articles
  • Taxonomy
    Related taxonomy entry
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...