Search

The links between water and crop yield are well known. In agricultural systems, maximum yield and maximum water productivity (WP; yield divided by water use) are not always compatible goals. In water-limited situations, optimal solutions must be reached by finding a compromise between the levels of crop production and WP. The tradeoffs between production and WP are reviewed here and the dominant effects of the environment on WP are examined. Genetic improvement for WP generally has yield tradeoffs, whereas management measures devised to improve WP also enhance yield. It is shown that partial closure of the stomata in response to environmental stimuli has a variable impact on canopy transpiration, depending on the degree of coupling between the canopy and the atmosphere. In contrast to the behaviour of the major herbaceous crops, WP increases in some woody crops in response to water stress, suggesting that biomass and transpiration are not linearly related, and that deficit irrigation should be successful in these species. Avoiding high evaporative demand periods (e.g. through tolerance to low temperatures) is an important option that aims to increase production and WP. A case study is presented for improving sunflower (Helianthus annuus L.) yield and WP in temperate environments.

We tested the hypothesis that the transpiration (λE p ) of high-coupled canopies, such as olive groves, may be calculated on a daily basis with sufficient precision by the Penman-Monteith 'big leaf' equation, by a model of bulk daily canopy conductance (g c ) capable of scaling for canopy dimension. Given the limited data required, such a model could replace the standard approach (ET 0 × K c ) for calculating olive water requirements, enhancing the precision of estimates. We developed a specific model of daily g c for unstressed olive canopies that was calibrated by transpiration measurements obtained by water balance from a 2-year experiment in a mature orchard with λE p ranging from 0.6 (February 1993) to 11.5 (July 1994) MJ m -2  day -1 and where leaf area index (L) changed from 1.25 to 2.5. The model uses the intercepted fraction of daily PAR and a linear function of average daytime temperature. The model was validated with λE p data collected by eddy covariance in a 3-year experiment conducted in a growing orchard that differed in L and cultivar from the one used in the calibration. The g c model, when used in the Penman-Monteith equation, gave very good daily λE p predictions for all seasons during 3 years, ranging from 0.5 (November 1998) to 5.5 (June 2000) MJ m -2  day -1 , indicating that the goals of dealing with the dependence of olive g c on L and of simulating the seasonal variations in g c were achieved. A comparison with the Jarvis g c model, calibrated with 2 months of measured g c hourly data, showed that the g c model developed here performed better than the Jarvis model for the 3-year dataset. The exception to this was the period in which the Jarvis model was calibrated. This indicates that (1) the Jarvis model did not account for the seasonal variations in g c of the olive trees; and (2) the spatial and temporal scale assumptions required in the calibration of g c generate seasonal errors in the simulated bulk daily λE p for this crop. The applicability of this bulk g c model is restricted to well watered olive canopies and to the one-layer approach of calculating λE p but it could be adapted to rain-fed canopies in the future.