A Preliminary assessment of a Predictive Model for Stomatal Conductance Using 'Fylloclip' Sensors

The current trial investigated a low-cost leaf proximal sensor, known as ‘Fylloclip’, which detects the condensing water vapor resulting from the leaf transpiration process. This simple and inexpensive device enables an approximate evaluation of a plant's water status by comparing the daily patterns of leaf transpiration and solar radiation. In theory, when the plant is well hydrated, there is likely to be a strong correlation between condensation and sunlight. However, deviations from these patterns are indicative of a potential drop in the plant's hydration level. It is important to note that the sensor still requires validation with established parameters for assessing plants' water status. This study was conducted on one-year-old potted lemon trees grown in a greenhouse. Measurements included leaf transpiration (capacitance), stomatal conductance (g_s) assessed with a gas exchange analyzer, and climate data. Pearson correlation analysis was used to select relevant features, followed by a machine learning approach to develop a predictive model for g_s. The model incorporated capacitance data and other variables, including the leaf temperature, stem water potential, and air relative humidity. Among the tested models, Random Forest emerged as the most suitable, yielding an R² = 0.72 and a mean absolute error (MAE) = 13.53 between the observed and predicted g_s values. These results indicate that capacitance alone is insufficient for accurately predicting g_s. Integrating additional sensors, such as leaf temperature and relative humidity sensors, could improve the prediction accuracy. These preliminary results may serve as a starting point for developing models and algorithms to estimate plants' water status by correlating ‘Fylloclip’ output data with common indicators of plants' hydration status, such as the stem water potential and g_s. Moreover, this sensor may represent an affordable component of a Decision Support System (DSS) for efficient and sustainable irrigation management. The next step could be to test this sensor in an open field to obtain additional data supporting its validity.