Abstract
Phenotyping methods for seed morphology are mostly limited to two-dimensional imaging or manual measures. In this study, we present a novel seed phenotyping approach utilizing lab-based X-ray microscopy (XRM) to characterize 3D seed morphology, internal structures, and cellular analysis from a single scan. Seeds of pennycress (Thlaspi arvense L.) an oilseed cover crop, were scanned and segmented using a machine learning model. Seed morphological analysis and a coat thickness map was applied to compare seed volumes of four genotypes. Notably, the 3D seed volume measurement alone was not enough to reveal differences in seed coat between the genotypes. Applying a seed coat thickness map revealed that the Large-Golden genotype had a thinner seed coat compared to wildtype despite a greater seed coat volume. Cellular analysis showed that cotyledon cell size was a driving factor for larger seeds. XRM was compared to traditional seed morphology traits and positive correlations were observed. Between Large-Golden and wildtype, differences could only be resolved by XRM highlighting the limitations of 2D seed area measures. These results demonstrate that XRM can provide quantitative measures extracted from seeds for enhancing our understanding of seed structure, development, and facilitate breeding efforts for enhanced seed quality and crop performance.