Abstract
Rice panicle architecture is a key target of selection when breeding for yield and grain quality. However, panicle phenotypes are difficult to measure and susceptible to confounding during genetic mapping due to correlation with flowering and subpopulation structure. Here we quantify 49 panicle phenotypes in 242 tropical rice accessions with the imaging platform PANorama. Using flowering as a covariate, we conduct a genome-wide association study (GWAS), detect numerous subpopulation-specific associations, and dissect multi-trait peaks using panicle phenotype covariates. Ten candidate genes in pathways known to regulate plant architecture fall under GWAS peaks, half of which overlap with quantitative trait loci identified in an experimental population. This is the first study to assess inflorescence phenotypes of field-grown material using a high-resolution phenotyping platform. Herein, we establish a panicle morphocline for domesticated rice, propose a genetic model underlying complex panicle traits, and demonstrate subtle links between panicle size and yield performance.