Search

Durum wheat (Triticum turgidum subsp. durum) is a key crop worldwide, yet its improvement and adaptation to emerging environmental threats is made difficult by the limited amount of allelic variation included in its elite pool. New allelic diversity may provide novel loci to international crop breeding through quantitative trait loci (QTL) mapping in unexplored material. Here we report the extensive molecular and phenotypic characterization of hundreds of Ethiopian durum wheat landraces and several Ethiopian improved lines. We test 81,587 markers scoring 30,155 single nucleotide polymorphisms and use them to survey the diversity, structure, and genome-specific variation in the panel. We show the uniqueness of Ethiopian germplasm using a siding collection of Mediterranean durum wheat accessions. We phenotype the Ethiopian panel for ten agronomic traits in two highly diversified Ethiopian environments for two consecutive years, and use this information to conduct a genome wide association study. We identify several loci underpinning agronomic traits of interest, both confirming loci already reported and describing new promising genomic regions. These loci may be efficiently targeted with molecular markers already available to conduct marker assisted selection in Ethiopian and international wheat. We show that Ethiopian durum wheat represents an important and mostly unexplored source of durum wheat diversity. The panel analyzed in this study allows the accumulation of QTL mapping experiments, providing the initial step for a quantitative, methodical exploitation of untapped diversity in producing a better wheat.

The first step to QTL cloning is the discovery of genetic loci linked to phenotypes of interest. Increasing the power, definition and speed by which such genetic loci are pinpointed on the genome is crucial for the success of this challenging task. New breeding designs emerged to address this need by increasing diversity and recombination events in QTL mapping panels. We discuss the MAGIC maize, a 1,636 recombinant inbred lines mapping panel we developed by intercrossing eight diverse Zea mays inbred lines in a balanced scheme. Each MAGIC recombinant inbred line is a unique mosaic of the diversity of the original founders, the population altogether collecting more than a hundred thousand recombination events. The increased diversity in elevated minor allele frequency permits the MAGIC maize to bridge the flexibility of association mapping approaches to the power of linkage mapping, speeding up the discovery of candidates for cloning. The MAGIC maize leverages transcriptomics data and full sequencing of the founder lines to reach single-gene definition. We genotyped and phenotyped a subset of the MAGIC maize and used it to show the general features of the population, identifying candidate genes for complex phenotypes. Mapping power simulations showed that the MAGIC maize may allow efficient QTL mapping even when few hundred lines are phenotyped, favoring its use in controlled conditions experiments. We discuss the perspectives that the MAGIC maize discloses in maize QTL mapping and cloning, and the further steps to exploit this resource, including the production of recombinant intercrosses for the study of heterosis.

Unrooted phylogeny of the MM population, magnified from Fig. 3a. Founders’ placement is highlighted with the corresponding colors. Note that genetic distances between MM lines are evenly distributed throughout. (PDF 20148 kb)

Genomic composition of the MM population. Panel a shows the distribution of the number of recombination events in MM RIL. The observed mean value (80.9) is close to the expectancy (79.8; see text). In panel b colored bars refer to each founder contribution to the MM population, as calculated from MM genome reconstruction. On top of the bars, the observed contribution in percentage. The red line refers to one-eighth, or 12.5%. IBS regions also influence parental contribution estimation. Panel c shows mean genomic similarity of founder lines per chromosome (increasing similarity from red to white as reported by the scale on the right). (PDF 2927 kb)

QTL scan for ear height. In panel a, the full model scan. On the x axis, the physical position from Chr 1 to 10. On the y axis, the LOD score. Red and green thresholds represent strong (P

Chromosome-wise distribution of polymorphic markers (as fraction of markers polymorphic per 1 Mb bin, gray line), marker density (as number of markers standardized per chromosome, yellow shading), and heterozygosity (as average heterozygosity in 1 Mb bins, red line). Note the two different scales on the y axes. Centromere positions are marked with black triangles. The observed heterozygosity is not related to polymorphism distribution. Marked enrichments for heterozygous loci are present on Chr 3, Chr 8, and Chr 10, as apparent from Additional file 3: Figure S1. (PDF 709 kb)

Detail of the local LD decay distribution along each chromosome. The 25th percentile (blue line) and 75th percentile (red line) of the LD distribution within ±1 Mb each marker is shown. Centromere positions are marked with black triangles. The r2 quantiles for each marker-based window are averaged in sliding windows of 100 markers in size. Some deviations between the general shape of the two distributions are evident, notably in Chr 2 and Chr 6. (PDF 12383 kb)

Linkage disequilibrium within the flowering time QTL confidence interval. Panel a shows the heat map considering all imputed SNPs in the region. From white to red, increasing LD. Along the diagonal, SNP marker index, not proportional to physical position (noted on edges). ZCN8 position is marked with an asterisk. The haplotype of highest significance is framed in blue. Panel b shows mean LD in a sliding window. On the x axis, physical position on the genome. The red box marks the haplotype of highest significance according to the association approach. LD is generally low in the QTL region, but a few distinctive peaks are visible: the most central to the QTL interval is short upstream ZCN8 (position marked with *), which lacks imputed SNPs. (PDF 2509 kb)