Stomatal Optimization Models Ranked by Predictive and Functional Accuracy at Ecosystem Scales

Abstract:

Stomatal optimization models can improve estimates of water and carbon fluxes with relatively low complexity. Optimization principles have been developed at the leaf and xylem level but there is no consensus on which carbon cost or water penalty functions are most appropriate for ecosystem-scale applications. We implemented three existing analytical equations for stomatal conductance, formulated with different water penalty functions, in a big-leaf framework, and evaluated model variants against FLUXNET observations in a range of biomes. We used information theory to dissect controls of water supply and demand on evapotranspiration in wet to dry conditions and to quantify whether modeled interactions between soil moisture, vapor pressure deficit, and evapotranspiration are functionally accurate. We then ranked stomatal optimization models based on their predictive and functional performance, parameter uncertainty, and parsimony. Goodness-of-fit was high for all model variants. Water penalty functions based on xylem vulnerability did not substantially improve predictive or functional accuracy compared to marginal water use efficiency, despite their explicit representation of plant hydraulics. Our results have implications for identifying stomatal conductance models that are most robust to generalize in Earth system models.