Using stomatal conductance as a water status indicator in grapevine: implementation into a decision support system for irrigation management

Prediction of grapevine water status can be performed nowadays with decision support systems (DSS). In some cases, the DSS predicts predawn leaf water potential (Ψ_pd), modelling the soil-plant-atmosphere continuum, and provides irrigation recommendations to maintain a desired Ψ_pd. Although Ψ_pd is a good indicator of water status, leaf stomatal conductance (g_s) can provide additional feedback on how the vine is responding to its water status from a physiological point of view. Moreover, leaf g_s measured with portable porometers can be a good ground-truthing alternative to pressure chamber measurements. The objective of this research was to propose a model of leaf g_s as a complementary indicator of water status. In 2017, Ψ_pd and early-morning, mid-morning and midday leaf g_s were measured eight days between budbreak and harvest in 10 plant replicates in four vineyards (‘Grenache’, ‘Cinsault’ and two plots of ‘Syrah’) under irrigated and non-irrigated conditions (about 640 measurements for each trait). Ψ_pd was able to explain 59% of the variability observed in leaf g_s and the best model consisted in an exponential regression including Ψ_pd and mean air temperature as independent variables (R²=0.70). In 2018, additional data was collected for model evaluation. The variation in leaf g_s that was explained by the previous model varied significantly between different data sets used for model validation (75% in a new plot of ‘Macabeu’, 55% for the same plots used in 2017, and 25% in a new plot of ‘Merlot’). These results may indicate that the development of a robust empirical model of leaf g_s may require a larger data set. The construction of models including specific coefficients for each cultivar and hour of measurement is another alternative to increase the robustness of g_s predictions but from a practical point of view they may be difficult to implement in a DSS. The coefficients of the model equation for all the existing cultivars are not known and the hourly specific models limit the time when the grower can perform g_s measurements to ground-truth the DSS. Having in mind that a generic model of leaf g_s using Ψ_pd and mean air temperature may not capture the variability of all existing vineyards, it could be strategic to implement a generic model into a DSS as a first step and improve the model as information on leaf g_s from multiple vineyards is collected from users.