High throughput phenotyping using automated imaging system reveals the relationship between seed yield and agronomic...

This study utilized plant phenomics image analysis technology to explore the agronomic characteristics of rice cultivars, aiming to enhance growth stability, yield potential, and digital data for rice breeding. RGB images were captured at three lateral angles during the growth period of the plants using ScanLyzer, LemnaTec. A total of 42 agronomic traits were analyzed across 102 rice cultivars, categorized into three maturing groups. In addition, to evaluate the measurement accuracy, 9 phenotypic traits, the panicle length (Pl), panicle count (Pc), and number of seeds were also measured destructively after harvest. Parameter estimated revealed that the Pl trait exerted the strongest positive effect on seed production across all groups analyzed, with coefficients (β) of 0.459 for the entire population, 0.456 in the early-maturing group, 0.537 in the medium-maturing group, and 0.574 in the medium-late maturing group (p < 0.05). Other traits, such as maximum area (Am), and maximum height (Hm), also positively influenced seed production but to a lesser extent. Notably, duration of maximum value of rice plant width had a significant negative effect in the early-maturing group (β = -0.369, p < 0.05). Correlation analyses revealed strong positive relationships between seed production and various traits across maturity classes, notably with days to maximum height, Pl, Pc, and seed count. Additionally, panicle length and count emerged as pivotal factors influencing seed numbers. These findings underscore the varying impacts of agronomic traits on seed yield depending on cultivars and maturity groups, offering valuable insights for the selection of rice cultivars aimed at optimizing seed production.