*2.2. Screening of Seed Aspect Ratio*

The seed aspect ratio data (Supplementary Table S2) were obtained by scanning seed images. The scanning images were processed by ImageJ 1.52a software (https://imagej. nih.gov/ij/notes.html) to generate phenotype data for the genome-wide association study. The seed aspect ratio was calculated as the seed major axis divided by the seed minor axis. Ten seeds per accession were scanned at the same time, and the seed aspect ratios of the ten seeds were averaged (Supplementary Figure S1). The phenotype data were analyzed using the R program to conduct a t-test and normal distribution in the accessions.
