A common irrigation strategy is to replenish the soil water reservoir according to evapotranspiration (ET). However, the ET from plants under deficit irrigation is not well explored and is normally assumed to be a constant fraction of their respective well-watered condition. In the current experiment, we hypothesized that the ratio between the ET of well-watered (WW) and water deficit (WD) grapevines is not constant but depends on the reference ET (ETO). The hypothesis was tested using lysimeters over two consecutive seasons to measure the ET of WD grapevines (Vitis vinifera L.) that were irrigated at 35 % of a second set of lysimeters with WW vines. The WD treatment started at veraison and resulted in a quick depletion of the soil water reservoir; thereafter a relatively stable soil water content (θ) and crop coefficient (KC) were measured. The ET of the WD vines was 20–75 % lower than that of the WW vines, depending on the reference evapotranspiration (ETO). Under high ETO, the difference between the treatments was much larger than under low ETO. The dynamic ratio between the ET of the treatments demonstrates the difficulty in predicting the ET of WD plants and suggests that irrigation according to a constant fraction from a WW plant might result in either excessive or insufficient irrigation amounts. The high correlation between instantaneous stomatal conductance (gs) measurements and KC emphasizes the advantage of utilizing gs to improve current irrigation models.
