Hard and soft bread wheat (Triticum aestivum L.) show contrasting seed vigour and storage pest resistance: A test of market class versus grain-hardness effects

Grain hardness is the trait used to divide bread wheat ( Triticum aestivum L.) into hard and soft market classes, but whether hardness itself, as opposed to other traits that co-vary with market class, predicts seed vigour and resistance to the rice weevil, Sitophilus oryzae (L.), is unresolved. We compared ten hard and ten soft genotypes, which by construction differed widely and non-overlappingly in grain hardness index (GHI: hard 72.0-88.0; soft 12.7-24.3), for seed biochemical and physical traits, standard germination and seedling vigour, and susceptibility to 60-day S. oryzae infestation and oviposition. Hard genotypes had significantly higher GHI, bushel weight, zinc and protein content; soft genotypes had significantly higher iron content, germination (95.3% vs 87.6%) and post-infestation germination loss (28.2 vs 4.6 percentage points; Welch t = −9.92, P < 0.001). Across all 20 genotypes, GHI correlated strongly with seed damage (r = −0.914), live adult counts (r = −0.898) and oviposition (r = −0.506, all P < 0.05). However, because the two market classes were pre-selected and do not overlap in GHI, we additionally tested whether these associations held within each class. After centring GHI within class, the within-class slope was non-significant for every outcome (all P ≥ 0.55; within-class |r| ≤ 0.33), while the between-class (hard vs soft) effect remained highly significant in every model. Genotype-aggregated binomial and quasi-Poisson models, fitted to the reported proportions and counts using the protocol's known sample sizes, gave the same pattern. We conclude that, within the range and design tested here, market class - not a continuous, within-class hardness gradient - is the trait associated with resistance and seed-vigour outcomes; whether hardness is itself causal, or a correlated marker for other Ha-linked or class-confounded traits, cannot be resolved from this two-class design and awaits confirmation in populations spanning a continuous hardness range (e.g., segregating or near-isogenic lines) with replicate-level data. GHI nonetheless remains a practical, non-destructive marker for prioritising further screening, but we no longer describe it as a validated, quantitative breeding criterion on the strength of this dataset alone.