In-field high throughput phenotyping and phenotype data analysis for cotton plant growth using LiDAR

Abstract. In-field high throughput phenotyping technologies need to be developed in order to accelerate plant breeding approaches for securing global crop supply. In this study, a terrestrial LiDAR-based high throughput phenotyping system was developed, and an RTK-GPS was used to provide spatial coordinates during data collection in the field. Cotton plants were scanned over the period from 45 to 109 days after planting (DAP). The 3D model of cotton plants was reconstructed based on the LiDAR and RTK-GPS data, from which two morphologic parameters—canopy height and projected canopy area—were derived. Plant growth analysis were conducted over the monitoring period. Four different cultivars showed similar growth trend and growth rate pattern. They kept on growing up until 88 DAP, and then varied slightly. However, the absolute values were cultivar specific. A special cultivar (cultivar 2) was found in this study, which obtained the highest height, smallest area and the lowest yield. The correlation between canopy height and yield kept on increasing over the period from 88 to 109 DAP and then varied slightly for cultivar 1, 3 and 4, whereas cultivar 2 showed a special variation pattern of decreasing over the monitoring period. Overall, the correlation between projected canopy area and yield kept on increasing, although the absolute values were cultivar specific. It obtained higher R² value than the correlation between canopy height and yield. The developed LiDAR-based high throughput phenotyping system makes it possible to generate 3D model of cotton plants with sufficient spatial resolution, which is necessary for measuring morphologic parameters and conducting growth dynamics analysis.