Establishing of early discrimination methods for drought stress of tomato by using environmental parameters and NIR spectroscopy in greenhouse

Abstract

Early detection of drought stress in tomato (Solanum lycopersicum) is an important and critical issue. Water deficit occurs during seedling and flowering stages and it has great influence on the quantity and quality of tomato. In this study, two tomato lines, ‘Tainan-Yasu No. 19’ and ‘Yu Nu’ grew with and without irrigation in a greenhouse. Environmental parameters in greenhouse and NIR (near-infrared) spectrum were used as explanatory variables to establish logistic regression and partial least squares regression (PLSR) models for early detection of drought stress. The predictive performance of the logistic regression model which utilized the difference of temperature between leaf and environment as explanatory variable had the 0.90-0.93 accuracy and 0.91-0.97 area under the receiver operating characteristic curve (AUC) to predict the early drought stress. As for the PLSR models, the accuracy and AUC ranged from 0.84-0.91 and 0.63-0.68 for the models to differentiate drought stress from normal irrigation. The results of present study indicated that combination of a non-destructive method and the logistic model can be a potential and promising option for an early detection of drought stress in tomato in greenhouse.