Distinct associations of ScGDD1/BC12 transcript abundance and sequence variation with grain size and shape in rye

Grain size and shape are important determinants of cereal yield and quality, yet their genetic regulation remains insufficiently characterized in rye (Secale cereale L.). Here, we examined associations of ScGDD1/BC12 transcript abundance and sequence variation with grain morphology in a panel of 91 winter rye genotypes comprising 78 gene bank accessions and 13 breeding materials. Relative transcript abundance was quantified at 14 days after pollination, and grain morphometric traits were evaluated in mature grains. Genome-wide DArTseq data available for 90 genotypes were used to assess population structure and as covariates in association models. Principal component analysis performed exclusively on grain phenotypic traits showed that Grain_PC1 explained 62.88% of the variance and represented overall grain size, whereas Grain_PC2 explained 35.49% and contrasted grain elongation with circularity. ScGDD1/BC12 transcript abundance was higher in breeding materials than in gene bank accessions after adjustment for DArTseq-derived PC1-PC3. Higher transcript abundance was consistently associated with lower thousand-grain weight, grain length, width, area, and perimeter in population-structure-adjusted models. The effect directions and broadly similar magnitudes were retained after additional adjustment for material category, although none of these associations remained significant after false discovery rate correction in the sensitivity models. Sequence analysis identified 182 polymorphisms forming 55 haplotype clusters, while the predicted ScGDD1/BC12 protein sequence was fully conserved. No sequence variant was significantly associated with Grain_PC1 after correction for multiple testing. In contrast, seven variants were associated with Grain_PC2 and circularity after adjustment for DArTseq-derived PC1-PC3. These variants showed moderate-to-strong pairwise correlation r² = 0.334-0.942, indicating a linked association pattern rather than seven independent effects. Their effect directions and magnitudes were largely maintained in models including DArTseq-derived PC1-PC5, but the associations no longer met the false discovery rate threshold. These results identify distinct association patterns between ScGDD1/BC12 transcript abundance, sequence variation, and grain morphology. The findings support ScGDD1/BC12 as a candidate locus for further investigation but do not distinguish direct regulatory effects from linkage with causal variation or establish functional causality.