Stomatal and non-stomatal limitations of photosynthesis for four tree species under drought: A comparison of model formulations

Abstract Drought strongly influences terrestrial C cycling via its effects on plant H 2 O and CO 2 exchange. However, the treatment of photosynthetic physiology under drought by many ecosystem and earth system models remains poorly constrained by data. We measured the drought response of four tree species and evaluated alternative model formulations for drought effects on photosynthesis ( A ). We implemented a series of soil drying and rewetting events (i.e. multiple droughts) with four contrasting tree species in large pots (75 L) placed in the field under rainout shelters. We measured leaf-level gas exchange, predawn and midday leaf water potential (Ψ pd and Ψ md ), and leaf isotopic composition (δ 13 C) and calculated discrimination relative to the atmosphere (Δ). We then evaluated eight modeling frameworks that simulate the effects of drought in different ways. With moderate reductions in volumetric soil water content ( θ ), all species reduced stomatal conductance ( g s ), leading to an equivalent increase in water use efficiency across species inferred from both leaf gas exchange and Δ, despite a small reduction in photosynthetic capacity. With severe reductions in θ , all species strongly reduced g s along with a coincident reduction in photosynthetic capacity, illustrating the joint importance of stomatal and non-stomatal limitations of photosynthesis under strong drought conditions. Simple empirical models as well as complex mechanistic model formulations were equally successful at capturing the measured variation in A and g s , as long as the predictor variables were available from direct measurements ( θ , Ψ pd , and Ψ md ). However, models based on leaf water potential face an additional challenge, as we found that Ψ pd was substantially different from Ψ soil predicted by standard approaches based on θ . Modeling frameworks that combine gas exchange and hydraulic traits have the advantage of mechanistic realism, but sacrificed parsimony without an improvement in predictive power in this comparison. Model choice depends on the desired balance between simple empiricism and mechanistic realism. We suggest that empirical models implementing stomatal and non-stomatal limitations based on θ are highly predictive simple models. Mechanistic models that incorporate hydraulic traits have excellent potential, but several challenges currently limit their widespread implementation.