Nondestructive Estimation of Circadian Time in Harvested Green Perilla Leaves Using Hyperspectral Data

Abstract

The circadian clock, an internal oscillator with a period of approximately 24 hours, plays an important role in the regulation of biological processes, and an understanding of circadian rhythms can be employed to improve the quality of plant production. Many studies have measured the circadian rhythms of plants and estimate their circadian times. However, the circadian time estimation methods used in previous studies are difficult to apply to commercial crops because they require extraction of plant contents such as RNA, which involves destroying plant tissues. In this study, we sought to develop a nondestructive method for estimating circadian time in harvested leaves of green perilla (Perilla frutescens var. crispa f. viridis). The results of RNA sequencing (RNA-Seq) show that the gene expression of perillyl alcohol depend on the circadian time. A hyperspectral camera captured the light reflectance of 141 wavebands from 350 to 1,050 nm on leaves, and machine learning using the reflectance data successfully estimated the circadian time corresponding to the harvest time. The study results demonstrate the potential for the nondestructive use of hyperspectral reflectance data in circadian time estimation and its applicability to improving the quality of plant production.