Statistical analysis
The transgenic lines were statistically compared to WT using a Student’st test at 5% of probability (P < 0.05) by using Microsoft Excel (Microsoft, Redmond, WA, USA). Regression analysis were carried out using SIGMAPLOT 14 (Systat Software Inc., San Jose, CA, USA). The equations from stomatal kinetics regressions were derived and the rate of change were fit to linear plateau using easyreg package in R 3.6.3 (Arnhold, 2018; R Core Team, 2020). Correlation analysis were carried out by Pearson correlation analysis using the Java-based CorrelationCalculator software (Basu et al., 2017). The metabolomics data were analyzed using the MetaboAnalyst platform (Chong et al., 2018). Multivariate analysis such as partial least square-discriminant analysis (PLS-DA) and orthogonal PLS-DA (orthoPLS-DA) was performed in Cube root-transformed and mean-centered data by using both Cube root and Pareto-scaling mode of the MetaboAnalyst platform, which is recommended to reduce the scale variability of metabolomics datasets (Xia & Wishart, 2011). Correlation-based metabolic networks were designed by using MetScape on CYTOSCAPE v.3.7.2 software (Karnovsky et al., 2012; Shannon et al., 2003).