Résumé
Phenology and time of flowering are crucial determinants of rice adaptation to climate variation. A previous study characterized flowering responses of 203 diverse indica rices (the ORYTAGE panel) to ten environments in Senegal (six sowing dates) and Madagascar (two years and two altitudes) under irrigation in the field. This study used the physiological phenology model RIDEV V2 to heuristically estimate component traits of flowering such as cardinal temperatures (base temperature (Tbase) and optimum temperature), basic vegetative phase, photoperiod sensitivity and cold acclimation, and to conduct a genome-wide association study for these traits using 16 232 anonymous single-nucleotide polymorphism (SNP) markers. The RIDEV model after genotypic parameter optimization explained 96% of variation in time to flowering for Senegal alone and 91% for Senegal and Madagascar combined. The latter was improved to 94% by including an acclimation parameter reducing Tbase when the crop experienced low temperatures during early vegetative development. Eighteen significant (P< 1.0 x 10(-5)) quantitative trait loci (QTLs) were identified, namely ten for RIDEV parameters and eight for climatic index variables (difference in time to flowering between key environments). Co-localization of QTLs for different traits were rare. RIDEV parameters gave QTLs that were mostly more significant and distinct from QTLs for index variables. Candidate genes were investigated within the estimated 50% linkage disequilibrium regions of 39 kB. In addition to several known flowering network genes, they included genes related to thermal stress adaptation and epigenetic control mechanisms. The peak SNP for a QTL for the crop parameter Tbase (P= 2.0 x 10(-7)) was located within HD3a, a florigen that was recently identified as implicated in flowering under cool conditions.