Genotypic variation of yield-related traits in an irrigated rice breeding program for tropical Asia

Developing high-yielding rice varieties is critical to ensure global food security. To date, selection of promising genotypes is based on empirical evaluation for grain yield, but the relationship of agronomic traits to yield in tropical rice breeding is largely unknown. This study aimed to determine the extent of variation for 19 agronomic traits and interrelationships among traits in an irrigated rice breeding program at the International Rice Research Institute, Philippines. A large set of elite germplasm comprising advanced genotypes and varieties derived from the breeding pipeline was evaluated during dry and wet seasons. The broad-sense heritability ranged from 0.35 to 0.99 for all traits in both seasons. Grain yield for the whole plot (plot yield) was positively correlated with yield per plant, 1000 grain weight, and grain width in dry season, and yield per plant, 1000 grain weight, grain width, number of panicles per plant, and panicle exertion rate in wet season. Path analysis showed that the highest direct positive effect of traits on plot yield ranged from 0.25 to 0.45 in dry season and from 0.22 to 0.88 in wet season. Heat map bi-cluster analysis assigned genotypes into three main clusters in both seasons, while traits were grouped into three and five clusters in dry and wet seasons, respectively. The cluster analysis showed that spikelets per panicle, filled grains per plant, and yield per plant were key yield contributing traits. Identification of traits that were highly correlated with rice yield could be useful for developing new varieties adapted to tropical environments.