Abstract: Flowering time is a crucial rice trait that influences its adaptation to various environments, cropping schedules, and agronomic characteristics. Rice breeders have exploited spontaneous mutations in heading date genes to regulate the flowering time. In the present study, we investigated how breeders in Fukui Prefecture regulated days to heading while developing promising rice varieties. Genome-wide association studies (GWAS) identified Hd1, Hd16, and Hd18 as the major genes controlling days to heading in the population. However, we suspected that this highly bred population might exhibit genomic stratification, which could lead to spurious or false correlations in the GWAS. Thus, we also conducted correlation and partial correlation analyses, which uncovered another key heading date gene, Hd17, that GWAS failed to detect because of its linkage disequilibrium with the major effect gene Hd16. Examination of haplotype frequencies across different breeding periods revealed that the early-heading Hd16 (Hd16(E)) and late-heading Hd17 (Hd17(L)) were increasingly co-selected in the Hd1 functional population. Varieties carrying this Hd16(E)/Hd17(L) combination exhibited days to heading in the range of 70–80, which corresponds to the peak temperature and sunshine period and is also optimal for grain quality and yield components in the Fukui environment. The present study highlights that it is imperative to remain vigilant for Type I (false positives) and Type II (false negatives) errors when performing GWAS on highly bred populations and to implement appropriate countermeasures by accounting for gene-by-gene interactions established through the breeding process. We also discuss the effectiveness of Hd16(E), which is not used outside Japan for subtle days to heading control but is widely used in Japan at certain latitudes.

Key words: breeding, flowering, GWAS, heading date, partial correlation analysis, rice