Evaluation of the Accuracy of Nondestructive Measurements of Stem Length and Leaf Area Using a 3-D Sensor and Its Potential Application in Predicting Tomato Phenotypes

Plant phenotype measurements for large numbers of samples are typically labor-intensive and challenging, especially for traits that require destructive measurements. To improve the efficiency of phenotypic measurement, we evaluated the accuracy of nondestructive measurement of individual leaf area (LA) and stem length (SL) of tomato segregating lines, derived from a cross between Japanese and Dutch F1 varieties, using a Kinect game console as a 3-D sensor in this study. Two types of software were used for nondestructive measurements using the sensor: one for measuring LA and another for measuring SL. The values obtained were compared with the actual values obtained using destructive measurement. The results showed that the correlation coefficients between the SL values obtained using destructive and nondestructive measurements (kPH) ranged from 0.95 to 0.99, indicating a very high accuracy throughout the 91-day cultivation period. The correlation coefficient between the LA values obtained using destructive and nondestructive measurements (kLA) was 0.762 at 60-days after planting; however, a high correlation coefficient of 0.897 was observed for the 91-day cultivation period. These results indicate that the two nondestructive measurement methods with two types of software are effective in SL and LA measurements. Regression analysis was performed on three traits, namely SL, LA, and total dry weight (TD), using the values of kPH and kLA obtained from nondestructive measurements and two composite variables obtained by multiplying or dividing the two values for a total of four explanatory variables. The regression analysis resulted in high coefficients of determination of 0.9883, 0.8068, and 0.9789, for SL, LA, and TD, respectively for the 91-day cultivation period. These results reveal that phenotypic traits can be predicted based on the values obtained from the nondestructive measurements presented in this study.