Genetic Hotspots for Grain Size Uncovered in Durum Wheat

Durum wheat is the raw material of pasta, couscous, and some of the world’s most cherished foods, yet the genetic controls over the size and shape of its kernels have remained stubbornly opaque. A new genome-wide association study published in Theoretical and Applied Genetics has now mapped dozens of genomic regions linked to kernel traits in a diverse panel of durum wheat lines, offering breeders a molecular toolkit for one of the crop’s most economically important characteristics. The research, led by Ana Laura Achilli of the University of Saskatchewan and CERZOS-CONICET in Argentina, together with colleagues including Pablo F. Roncallo, Juan M. Rodrigo, Curtis J. Pozniak, and Viviana C. Echenique, focused on a collection of 170 genotypes drawn largely from Argentinian cultivars, a germplasm pool that reflects nearly a century of regional breeding history.

The importance of kernel morphology in durum wheat goes far beyond aesthetics. Kernel size and shape directly influence grain yield, because heavier individual grains contribute to overall biomass harvested per hectare, and they also determine semolina milling potential, the critical first step in producing high-quality pasta. Larger and more uniformly shaped kernels tend to yield more semolina with fewer broken particles, improving both the quantity and the quality of the final product. Despite these well-established relationships, the genetic architecture underlying kernel traits in durum wheat, a tetraploid species with genomes inherited from wild emmer ancestors, has been far less characterized than in its hexaploid bread wheat cousin. The new study set out to close that gap with a combination of careful field phenotyping and dense molecular marker coverage.

To capture the true genetic signal, the team grew the 170-genotype panel across four field experiments in Argentina, allowing them to evaluate how kernel traits behaved in different environments. Each genotype was measured for a suite of kernel characteristics, including thousand kernel weight, kernel area, kernel width, kernel length, kernel perimeter, and factor form density, a composite descriptor of grain shape. The researchers also computed best linear unbiased predictors, or BLUPs, a statistical technique that separates the stable genetic component of each trait from the noise introduced by environment and measurement error. High phenotypic variability was observed across the panel for all studied traits, and, crucially, heritability values were high as well, indicating that a substantial proportion of the observed variation is genetically controlled and therefore amenable to selection.

The correlation structure among traits provided early clues about which measurements matter most. Thousand kernel weight, the classic breeder’s proxy for grain size, showed strong correlations with kernel area and factor form density, followed by kernel width, length, and perimeter. This pattern suggests that the two-dimensional projected area of the grain and its overall shape density capture much of the genetic variation that drives grain weight, meaning breeders could potentially select on easily measured morphological traits as indirect indicators of the heavier-kernel phenotype they ultimately seek. Such indirect selection is particularly valuable in large breeding programs where measuring every trait on every line is impractical.

On the genotyping side, the panel was profiled with the 35K Axiom Wheat Breeder SNP Array, a single-nucleotide polymorphism platform designed for high-throughput screening of wheat breeding material. With thousands of genetic markers distributed across the durum wheat genome, the team could apply genome-wide association mapping, a statistical approach that scans for markers whose allelic states correlate with phenotypic differences across the population. The analysis, controlled for population structure and relatedness to avoid false positives, identified a total of 80 marker-trait associations exceeding a stringent false discovery rate threshold of P less than 0.00001, a cutoff that sharply limits the chance that any single association is a statistical fluke.

Not all of those 80 associations carry equal weight, and the researchers applied progressively stricter filters to isolate the most robust signals. Twenty-three marker-trait associations were detected either for the BLUP values averaged across all four environments or in at least two individual environments, indicating that these signals are stable rather than environment-specific. Even more compelling, ten SNPs located on chromosomes 1B, 2A, 2B, 3B, 6A, and 7A were identified by multiple mapping methods. Convergent evidence from different statistical models is widely regarded as one of the strongest indicators that a marker-trait association reflects a genuine underlying gene or regulatory element rather than an artifact of a particular analytical pipeline.

Three of these ten SNPs, all situated on chromosome 2A, emerged as the standout candidates because they were detected by all three mapping methods tested. When the team examined the genomic neighborhoods of these chromosome 2A markers, they found that the SNPs fell within two haplotype blocks, inherited stretches of DNA measuring 24.2 and 16.6 megabases respectively, and both blocks were strongly associated with most of the kernel traits measured. Haplotype blocks are particularly useful for breeders because the linked variants within them tend to be inherited together, meaning a single diagnostic marker can effectively track an entire chromosomal segment carrying the favorable alleles.

One of the most strategically important findings concerns what these haplotypes are not linked to. The chromosome 2A regions associated with kernel traits showed no linkage to Ppd-A1, the well-known photoperiod sensitivity gene on that chromosome. This independence matters enormously in applied breeding. Photoperiod genes control flowering time and adaptation, and markers linked to them can drag along unwanted changes in maturity when breeders select for other traits. Because the kernel-associated haplotypes are genetically independent of Ppd-A1, breeders can select for improved kernel size and shape without inadvertently altering the flowering behavior that determines whether a variety is suited to a particular growing region. The authors highlight these haplotypes as promising targets for further validation and for implementation in molecular-assisted breeding aimed at improving kernel-related traits in durum wheat.

The study also fits into a broader scientific conversation about the trade-off between grain weight and grain number, one of the central constraints on wheat yield improvement. Decades of research have shown that wheat tends to compensate: plants that set more grains often produce lighter ones, and vice versa, a physiological see-saw that has frustrated breeders seeking simultaneous gains in both components. Recent work, including studies of genes such as GNI-A1 in tetraploid wheat and efforts to manipulate cell wall properties in developing seeds, has explored ways to break this trade-off. Precise knowledge of the loci controlling kernel size, as delivered by this GWAS, gives researchers molecular entry points into that puzzle, allowing them to manipulate the weight side of the equation with greater specificity while monitoring effects on grain number.

For Argentinian durum wheat breeding specifically, the findings arrive at a moment of accumulated momentum. Earlier work by the same research group documented the genetic gains in grain yield and agronomic traits of Argentinian durum wheat from 1934 to 2015, as well as the population structure and allelic variation at major genes such as Rht-B1 and Ppd-A1 in the national germplasm. The new marker-trait associations extend that foundation by identifying the specific genomic regions that breeders can now track with DNA-based markers. With validation in additional populations and environments, the chromosome 2A haplotype blocks and the other robust associations on chromosomes 1B, 2B, 3B, 6A, and 7A could be converted into routine diagnostic tests, accelerating the development of durum varieties that pack heavier, more uniform kernels into every hectare, and ultimately delivering more semolina per harvest for the pasta and couscous markets that depend on this ancient and increasingly high-tech crop.

Subject of Research: Genetic mapping of kernel size and shape traits in durum wheat using genome-wide association studies

Article Title: Genome-wide association studies reveal genomic regions for kernel-related traits in durum wheat

Article References: Achilli, A. L., Roncallo, P. F., Rodrigo, J. M., Pozniak, C. J., & Echenique, V. C. (2026). Genome-wide association studies reveal genomic regions for kernel-related traits in durum wheat. Theoretical and Applied Genetics, 139(10), Article 289. https://doi.org/10.1007/s00122-026-05379-3

Image Credits: AI Generated

DOI: 10.1007/s00122-026-05379-3

Keywords: durum wheat, genome-wide association study, kernel size, kernel shape, thousand kernel weight, SNP markers, haplotype blocks, chromosome 2A, molecular breeding, semolina quality, Ppd-A1, grain yield