The 18O/16O ratio of leaf water is a useful signal in several contexts. It affects the oxygen isotopic composition of atmospheric carbon dioxide, thereby giving rise to the possibility of assessing terrestrial gross primary productivity (Farquhar et al. 1993; Ciais et al. 1997). It determines the isotopic ratio of oxygen released by plants from the splitting of water by photosystem II (Guy et al. 1987), thereby contributing to the Dole effect, an indicator of the balance between terrestrial and marine net productivity on millennial time scales (Bender, Sowers & Labeyrie 1994). It reflects genetic differences in stomatal conductance and these are recorded in the oxygen isotope composition of plant organic matter, and these in turn have correlated with progress in wheat yields (Barbour et al. 2000a).
Such applications of the leaf water signal vary subtly in what part of the leaf water they record. The imprint on CO2 depends on the compartment where carbonic anhydrase is active, most probably the chloroplast stroma for C3 plants, but the mesophyll cells for C4 plants (Reed & Graham 1981). That on O2 depends on water in the chloroplast thylakoid. The imprint on organic matter depends on the isotopic composition of water in the chloroplast stroma, and in the cytosol and at the sites where sugars are converted to cellulose and other materials.
It follows that a clear understanding of the spatial variation in 18O enrichment of leaf water is needed to inform the above applications (Yakir 1998). To that end we distinguish variation within the plane of the leaf from the gradients that must exist over the small distances between veins and the sites of evaporation. In the model finally developed here we simplify the leaf to a longitudinal structure, grass-like but rather simplified. It allows the isotopic composition of the water to vary on a large scale longitudinally, with radial variation on a small spatial scale caused by flow from the ‘longitudinal’ xylem to the evaporation sites. This radial flow is through radial ‘veinlets’ and then through the lamina mesophyll tissue. The model considers an isolated leaf, in which the vapour pressure and the isotopic composition of the ambient air are uniform, rather like the air in a well-mixed gas exchange cuvette.
Gan et al. (2002) considered contemporary models of leaf water in the light of their measurements in cotton leaves. They concluded that a composite model was needed that took account of: (a) the two-pools represented by veins and mesophyll (Allison, Gat & Leaney 1985), modified to include ground tissue in the vein ribs; (b) the competing effects of advection from the veins and diffusion from the sites of evaporation (Farquhar & Lloyd 1993); and (c) the progressive enrichment along a line of evaporating cells. With regard to (c), Yakir (1992) suggested the use of the string-of-lakes model of Gat & Bowser (1991) and this has been applied by Helliker & Ehleringer (2000).