Genome-wide association study of grain shape variation among Oryza sativa L. germplasms based on elliptic Fourier analysis

Although grain shape is an important cereal breeding target, it has only been evaluated using simple measurements, e.g., the length:width ratio. We used elliptic Fourier analysis to evaluate grain shape variation and conducted whole genome association mapping of grain shape using a germplasm collection of Asian cultivated rice. The first principal component of elliptic Fourier descriptors, accounting for over 90% of the total variation, was associated with the length:width ratio of brown rice. That component was the most significant among the first ten components: the length:width ratio is the major genetic variation of rice grain shape. Bayesian multilocus association mapping detected five significant markers for this component; three might be linked to previously reported quantitative trait loci. Allelic effects of significant markers were visualized using inverse Fourier transformation, showing that the allele of a Japonica variety Nipponbare produced plumper grains in four of five significant markers. For the second to tenth principal components, varietal effects were significant (P < 0.001), although most accounted for less than 1% of the total variation in elliptic Fourier descriptors. Association mapping detected at least one quantitative trait locus in six of these nine components: breeding programs can improve the heritable shape characteristics associated with these components. For association mapping, elliptic Fourier analysis' accuracy and high throughput are suitable; it is readily applicable to cereal crops because it is not based on rice-specific morphological characteristics but rather on universal shape descriptors that can delineate any closed contour.