Elucidating the genetic architecture of yield components in pyramidale wheat (Triticum turgidum L.): a comparative approach using QTL analysis and genomic selection

Wheat grain yield is determined by three primary components, one of which is grain number per spike (GPS; or spike fertility). GPS is further determined by spikelet number per spike (StPS) and grain number per spikelet (GPSt). TN28, a pyramidale wheat accession, produces up to five grains per spikelet, substantially exceeding the number in typical cultivars. To elucidate the genetic architecture underlying TN28’s superior traits and determine the optimal breeding strategy for trait introgression, we evaluated yield-related traits in a biparental F 2 population derived from TN28 × Langdon across three environments over two years using quantitative trait locus (QTL) mapping and genomic selection (GS). QTL analysis using two mapping approaches detected no stable QTLs other than known flowering genes, indicating that TN28’s yield-related traits cannot be reliably introgressed via single- or few-marker selection. Conversely, GS using five prediction models demonstrated potential for accumulating favorable minor-effect alleles. The ridge regression model exhibited optimal performance for GPSt-related traits, with the top 20% of individuals selected on the basis of genomic estimated breeding values (GEBVs) achieving approximately 9% higher grain number per spikelet at the basal section (GPStB) and 7% higher grain number per spike (GPS) relative to the population mean. These results support that QTL mapping is effective for traits controlled by a single major-effect gene, whereas GS offers a more powerful strategy for traits controlled by multiple minor-effect genes. This study suggests that TN28’s superior yield-related traits may arise from the cumulative contribution of numerous minor-effect genes and establishes GS as the preferred tool for their improvement in breeding programs.