UAV-based phenotyping identifies net assimilation rate as a diagnostic trait for synergistic enhancement of rice yield and grain quality

Abstract

Achieving rice yield-quality synergy, which is critical for breeding and agronomic practice, is hindered by dynamic regulatory gaps due to methodological constraints, while high-throughput unmanned aerial vehicle (UAV) phenotyping can enable breakthroughs by decoding dynamic traits at scale. This study conducted five experiments (EXP, 2022–2024; including nitrogen fertilization, multi-cultivar, and breeding material experiments) with UAV-based phenotyping to establish trait estimation models (EXP1-EXP3), enabling dissection of trait-specific contributions to yield-quality synergies via regression, multi-objective optimization, and path analysis (EXP4-EXP5), and identifying diagnostic traits in practice. Using UAV data, effective regression models were developed to monitor five rice traits: plant height (R² ​= ​0.89), aboveground biomass (R² ​= ​0.84), leaf area index (R² ​= ​0.61), canopy nitrogen content (R² ​= ​0.68), and leaf nitrogen content (R² ​= ​0.83), thereby systematically establishing 37 critical plant traits across the growth stages. Furthermore, feature importance analysis using extreme gradient boosting (R² ​= ​0.99) assessed the importance of these traits for yield and grain quality, and four common traits that were crucial for both yield and grain quality were identified. Notably, the synergistic yield-quality group exhibited 26.38–51.76% higher net assimilation rate (NAR) than the low-performance group (validated by multi-objective optimization), positioning NAR as a diagnostic marker for yield-quality synergistic enhancement. Path analysis revealed that NAR exerted positive effects on yield and grain quality, while yield indirectly influenced grain quality through eating quality. Overall, this study integrated UAV-based phenotyping and trait analysis, providing a novel insight into the synergistic enhancement of yield and grain quality.