Antagonistic pleiotropy at the stem solidness 1 locus balances wheat stem architecture and yield components under terminal drought

Key message: Stem structural investment enhances grain size and yield under drought, challenging the traditional view of vegetative-reproductive trade-offs. Antagonistic pleiotropy at the SSt1 locus reveals a genetic constraint between stem diameter and solidness. Incorporating genotype-by-environment into genomic selection models improves prediction accuracy by 46.5%. Abstract: Terminal drought is the primary environmental constraint on wheat productivity in the Mediterranean basin, a challenge intensifying under climate change. While stem biomass investment is known to promote grain-filling under stress, its underlying genetic architecture remains poorly understood. We evaluated a bread wheat diversity panel (n = 295; WheatMore) across three Mediterranean environments: well-watered (635 mm), terminal drought (228 mm), and rain-fed (468 mm). Our objectives were to characterize the phenotypic and genetic trade-offs between stem traits and yield components, identify associated genomic regions, and implement multi-environment genomic selection models. Increased investment in stem structural biomass, specifically dry weight, diameter, and peduncle length, was positively correlated with grain size and yield per spike, particularly under terminal drought. These results suggest that robust stem architecture enhances source strength without penalizing yield, as increased grain weight compensates for the biomass investment. Genome-wide association identified 80 significant SNPs and 77 gene hits, including 14 stable, 61 environment-specific, and 21 pleiotropic loci. Notably, pleiotropic regulation at the stem solidness1 locus (SSt1) on chromosome 3B revealed a genetic trade-off between stem diameter and solidness mediated by antagonistic alleles, highlighting a constraint for simultaneous optimization. Multi-environment genomic prediction incorporating genotype-by-environment interaction outperformed single-environment models, increasing prediction accuracy by 46.5%. These findings provide breeders with markers and predictive tools to develop wheat stem ideotypes to optimize climate adaptability.