Increased grain weight conferred by GW2 mutations in wheat does not translate into yield gains in multi-year field trials of near-isogenic lines.

Multiple studies have identified genes affecting grain morphology, yet their capacity to deliver yield gains under field conditions remains unclear. We performed a multi-year, multi-location factorial evaluation of GRAIN WIDTH2 (TaGW2) mutants in hexaploid wheat using BC4 near-isogenic lines, sowing density treatments, and semi-dwarfing RHT1 backgrounds. Loss-of-function mutations in TaGW2 increased grain size and thousand grain weight (TGW) additively, with the aaBBDD single mutant showing the most stable single-locus effect, while the aabbdd triple mutant achieved ∼20% higher TGW across 12 field trials. However, overall grain yield remained unchanged or slightly reduced, reflecting a compensatory trade-off with grain number. Spike phenotyping of both main and secondary tillers showed comparable increases in TGW and spike yield despite fewer grains per spike, indicating that limited yield gain primarily reflects reduced spike number per unit area rather than decreased spike-level productivity. Effects were stable across sowing densities, whereas interactions with semi-dwarfing alleles were allele specific: RHT-B1b partially suppressed TGW gains and accentuated yield penalties, whereas RHT-D1b maintained the large-grain phenotype and productivity. Across experiments, the TaGW2-A1D1 double mutant increased TGW (∼14%) while maintaining yield stability, identifying it as a promising genotype for breeding. We conclude that TaGW2 is a reliable modifier of grain size but not yield in isolation.