<p>Soil salinity poses a global threat to crop production. Early understanding of plant physiological responses to salinity stress can be critical to implementing timely stress management strategies. One of the initial plant physiological responses to salinity is a reduction of transpiration. This study used papaya as a model crop to better understand the effect of salinity on whole-plant transpiration using a greenhouse experiment. Treatments consisted of four electrical conductivity (EC_IR) levels of irrigation water: ~ 0 (tap water), 2, 4, or 8 dS m^− 1 were considered for the experiment. An automated phenotyping platform measured whole plant transpiration from the time of papaya transplanting until they reached approximately 15 weeks of age. Five machine learning models: extreme gradient boosting (XGBt), categorical boosting (CATBt), light gradient boosting (LAGBt), random forest (RF), and decision tree (DT) were fitted to the transpiration data and machine learning algorithms were deployed on a new data set. The impact of salinity on transpiration started to become evident 16 days after initiation of salinity treatments, where only the 8 dS m^− 1 treatment induced a significant decline in transpiration. All machine learning models efficiently captured salinity-induced impacts on transpiration. The use of salinity as an input feature improved the performance of all machine learning models. Salinity contributed up to 32% to the predictive capability of the machine learning models by improving the R², root mean squared error (RMSE), and mean absolute error (MAE) of the machine learning models up to 19, 25, and 25%, respectively. Overall, the extreme gradient boosting model outperformed all the machine learning models. Furthermore, deployment of machine learning algorithms to a new data set effectively indicated a critical level of EC_IR of 6 dS m^− 1 on whole-plant transpiration above which transpiration significantly drops. Apart from transpiration the effect of salinity on biomass, the concentration of Na⁺ and Cl⁻ contents in leaves and roots were evident. Overall, machine learning models can be useful tools for capturing salinity-induced impact on plant water use and its integration with crop stress management practices could be valuable.</p>