Processes driving nocturnal transpiration and implications for estimating land evapotranspiration

Evapotranspiration is a major component of the water cycle, yet only daytime transpiration is currently considered in Earth system and agricultural sciences. This contrasts with physiological studies where 25% or more of water losses have been reported to occur occurring overnight at leaf and plant scales. This gap probably arose from limitations in techniques to measure nocturnal water fluxes at ecosystem scales, a gap we bridge here by using lysimeters under controlled environmental conditions. The magnitude of the nocturnal water losses (12-23% of daytime water losses) in rowcrop monocultures of bean (annual herb) and cotton (woody shrub) would be globally an order of magnitude higher than documented responses of global evapotranspiration to climate change (51-98 vs. 7-8 mm yr(-1)). Contrary to daytime responses and to conventional wisdom, nocturnal transpiration was not affected by previous radiation loads or carbon uptake, and showed a temporal pattern independent of vapour pressure deficit or temperature, because of endogenous controls on stomatal conductance via circadian regulation. Our results have important implications from largescale ecosystem modelling to crop production: homeostatic water losses justify simple empirical predictive functions, and circadian controls show a fine-tune control that minimizes water loss while potentially increasing posterior carbon uptake.