Season-Specific Trait Priorities for Direct-Seeding Performance in Double-Cropping Rice Revealed by Multi-Trait Evaluation

Direct-seeded rice offers advantages in labor efficiency and mechanized production, but differences between early- and late-season growing conditions may influence genotype performance and trait selection priorities in double-cropping systems. In this study, 32 rice genotypes were evaluated under direct-seeded and manually transplanted conditions during the early and late growing seasons of 2021 in South China. Grain yield, relative yield retention (RYR; direct-seeded yield/manual-transplanted yield), and agronomic trait variation were analyzed to characterize seasonal performance and develop a multi-trait direct-seeding suitability index (DSI). Mean grain yield under direct seeding was lower than that under manual transplanting in both seasons, with RYR averaging 0.614 and 0.671 in the early and late seasons, respectively. Direct-seeded grain yield showed moderate cross-season consistency (Pearson’s r = 0.605; Spearman’s ρ = 0.572), indicating that genotypes with higher productivity in one season generally performed relatively well in the other. Individual agronomic traits showed varying seasonal responses, although their associations with grain yield and RYR were generally modest and did not remain significant after false-discovery-rate correction. Integrated trait prioritization identified contrasting seasonal patterns: lodging-related traits and chalkiness-related traits received relatively high priority in the early season, whereas seed-setting rate and several grain-quality traits were more prominent in the late season. The DSI combined standardized direct-seeded grain yield with a multi-trait index to assess overall genotype performance. DSI rankings were strongly associated with grain-yield rankings (ρ = 0.91), while the multi-trait index provided complementary information. The DSI showed high internal stability across sensitivity analyses, and a reduced model based on five priority traits produced rankings closely aligned with the full model (ρ = 0.984). These findings highlight the value of combining grain productivity with season-specific trait priorities for evaluating direct-seeded rice genotypes.