French (Fr)English (United Kingdom)

Research Topics

Our research is characterized by quantitative genetics approaches closely related to breeding methodology and the management of genetic diversity. They aim at:


  • understanding the effect of historical and modern breeding on the evolution and adaptation of open pollinated and hybrid varieties, in terms of phenotypic variation, global organization of genetic diversity and polymorphism at specific loci.
  • investigating the genetic determinism of complex traits, in view of direct applications in breeding through marker assisted selection and to gain a better understanding of the type of genetic effects which are involved. A specific attention is paid to flowering time, productivity under abiotic environmental constraints and heterosis,
  • optimizing the breeding process, from genetic resources to variety development.


This research involves theoretical and experimental approaches, the development of statistical methods and decision support tools. Experimental approaches involve the development of original genetic materials in maize, their genotyping and their phenotyping. They are conducted by team members and also benefit to a large extent from the support of INRA experimental structures.


1. Effect of migrations and breeding on the organization of maize genetic diversity

1.1   Molecular polymorphism in maize germplasm

New technologies of genotyping and sequencing offer new opportunities for genetic studies. We contribute to the development of new methods and tools such as:

  • the first 50k SNP chip (collaborative project, Ganal et al., 2011, Plos one). We used this chip to genotype of around 1200 maize lines, 250 landraces and 230 Highly Recombinant Inbred Lines.
  • the discovery of new polymorphisms such as SNPs or indels or more complex ones as CNVs in collaboration with ABI and GEAR teams. To discover these new polymorphisms, we achieved deep reseqencing of 5 inbred lines including the French reference line Fv2 (projects Amaizing and CNV-maize) and of 120 lines at a lower depth (projects Amaizing,, and Cornfed).


1.2   Consequences of evolution and breeding on the organization of genetic diversity

We aim at deciphering how evolution and modern breeding shaped the genetic diversity.

  • Following our previous studies on traditional maize varieties in Europe (Tenaillon & Charcosset, 2011, C R Biol), we extended our research to Asia and Africa within an international cooperation (Generation Challenge Program). More than 800 landraces were genotyped to evaluate the contribution of American maize genetic groups to the world diversity, hybridization areas were pointed out (Mir et al., 2013, Theor Appl Genet).
  • We showed the emergence of new genetic groups resulting from recent breeding (Truntzler et al., 2011, Theor Appl Genet).


1.3   Modelling of linkage desequilibrium

We have been studying the range and variability of linkage disequilibrium (LD) along the genome in different panels and evaluated the impact of genetic structure on LD.

  • In a panel of 375 lines, the average LD has been found to be quite broad (197 kb for a R2 of 0.1) and to be amplified by the genetic structure of the population (Bouchet et al., 2013, Plos One). Population structure and kinship must therefore be taken into account in the different models of association mapping (Veyrieras et al., 2007b, Crop Sci, Mezmouk et al., 2011, Theor Appl Genet)
  • We evaluated several tools for haplotype reconstruction and imputation. Our results show that lines can share large IBD regions (coll. B. Servin et B. Mangin, INRA Toulouse). We use this information on local identity to develop models for GWAS and QTL detection in multiparental context (see below).

2. Genetic determinism of complex traits

2.1. Effect of linkage on the variation of quantitative traits. QTL fine mapping

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 agronomical traits. We observed that:

  • The genetic variance for grain yield is significantly lower in the intermated F3 population, which suggests that single QTL in the classical F3 population should generally correspond to clusters of QTL in coupling phase  (Huang et al., 2009, Genetics)
  • This hypothesis was validated by the fine mapping of one QTL that showed that its apparent pleiotropic effect for grain yield and quality was due to two different QTLs with small effects.

2.2. Multiparental QTL mapping designs and allelic series

We contributed to the development of the BioMercator software to carry out meta-analysis of QTLs (in collaboration with the ABI team of the lab, Sosnowski et al., 2011, Bioinformatics, available at We used this software to reduce the confidence intervals of QTLs for silage quality (Truntzler et al., 2011, Theor Appl Genet) and water drought tolerance traits (Welcker et al., 2011, Plant Physiol). To get further insight into the allelic diversity of QTLs, we performed QTL detection in connected-multiparental designs, considering IBD probabilities of chromosome segments of parental lines  (Bardol et al., 2013, Theor Appl Genet).


2.3. Heterosis

We extended the North Carolina design III (NCIII) by using three populations of recombinant inbred lines derived from three parental lines belonging to different heterotic pools, crossed with each parental line to obtain nine families of hybrids. Most of the QTL detected for grain yield are located in pericentromeric regions and display apparent overdominance effects and limited differences between heterozygous genotypes, whereas for grain moisture predominance of additive effects was observed (Larièpe et al., 2012, Genetics).


3. Fine mapping and diversity organisation at major QTL


The fine mapping of QTLs was carried out by analysing of a large number of recombinants or by association studies.

vgt1 region - The distribution of the early allele fits with the environmental conditions and thus, reinforces the role of this locus in adaptation to the temperate area (Ducrocq et al.,2008, Genetics).

ZmCCT - A major QTL was localised near ZmCCT (170 kb) in an hypo-recombinating region (Ducrocq et al.  2009, Genetics) and its effect on earliness was later confirmed by other colleagues. This locus displays a limited number of haplotypes (fewer than for vgt1 locus), the effects of which were confirmed using near isogenic lines.

Tga1 and Su1 - A QTL affecting different agronomic traits was mapped in the region of Tga1 (involved in the domestication process) and Su1 (carbohydrate metabolism) on chromosome 4. Both genes show a strong decrease of nucleotidic diversity compared to teosinte.

Zcn8 and genome wide studies - Association genetics conducted with the 50k SNP chip highlighted a major effect of Zcn8 locus, which most likely corresponds to QTL vgt2 (Bouchet et al., 2013, Plos One).  This contributes to deciphering the major contribution of the vgt1 - vgt2 region to flowering time variation in maize.



4. Optimization of marker assisted selection


Following our results on the advantages of leading recurrent marker assisted selection for the multiparental schemes (Blanc et al., 2006, Theor Appl Genet; Blanc et al., 2008, Euphytica), we developed the OptiMAS software (Valente et al., 2013, Journal of Heredity, available at to follow the transmission of favorable alleles in breeding schemes in order to choose the best parents to be crossed to increase the probability of obtaining the molecular ideotype at the targeted QTL.

We also strongly invest in the evaluation of genomic selection, especially for complex traits. This approach is based on the genotyping and phenotyping of a reference sample of individuals for the calibration of a model, which is then applied to predict the value of individuals based only on genotyping data. This avoids QTL detection step, which increases prediction efficiency for highly complex traits.

We showed that his approach was suitable in a multiparental context for a complex trait such as yield (Bardol et al., submitted) and that its efficiency can be significantly increased by optimizing the calibration of the reference population (Rincent et al., 2013, Genetics). A R-code was developed to help choosing the lines to be included in the reference population to increase the reliability of genomic predictions and is available on request.