Vegetation acts as a critical link between the geosphere, biosphere, and atmosphere, regulating the flux of water to the atmosphere via transpiration (E) and the input of carbon from the atmosphere to plants and soil via photosynthetic carbon assimilation (A). The rate of A is known to be seasonally dynamic, however, few studies have investigated how the ratio between E and A, known as the water use efficiency (WUE), changes with phenology. WUE directly impacts regional to global carbon and water cycles and lack of knowledge regarding the dynamics of WUE remains among the largest uncertainties in current earth system model (ESM) projections of carbon and water exchange in temperate forests. Here we attempt to reduce this knowledge gap by studying these dynamics across a range of eight deciduous tree species common to temperate forests of North America. Using gas exchange and spectroscopic measurements, we investigated seasonal patterns in leaf level physiological, biochemical, and anatomical properties, including the seasonal progress of WUE and foliar capacity for carbon assimilation, which corollate with seasonal leaf phenology. We incorporate these findings into a modeling framework that contains the same representation of A, E, and canopy scaling found in ESMs to explore the impact of parameterization, which tracks phenological status, on model forecasts. Our results indicate that both photosynthetic capacity and WUE are seasonally dynamic processes which are not synchronized. WUE increased from a minimum at leaf out toward a more conservative behavior at the mid-summer growth peak. This pattern was explained by a decreased stomatal aperture and a decrease in cuticular leakage with leaf aging. We also observed a seasonal increase in maximum carboxylation capacity, with maximum rates of A and modeled tree net primary productivity (NPP) occurring later toward the end of the summer. This change was primarily driven by an increase in foliar nitrogen content, and a shift in the ratio of Vcmax to Jmax between expanding and mature leaves. By applying our revised parameterization, which captures seasonal dynamics of gas exchange, into our model framework we aim to improve the process representation of leaf function in a temperate forest, and more faithfully represent dynamics of NPP and E in the early and late growth season.

A primary source of uncertainty in terrestrial biosphere model (TBM) projection of carbon uptake and water cycling from ecosystems is the relationship between CO2 assimilation (A) and water loss via stomatal conductance (gs). A common mathematical framework for modeling this relationship is the “Unified Stomatal model”, which relates A to gs over environmental conditions and is governed by two terms, the stomatal slope (g1) and intercept (g0). Given their importance in determining the relationship between forest productivity and climate, an accurate and mechanistic understanding of the g1 and g0 parameters is crucial, particularly in wet tropical broadleaf forests where changes in water cycling could impact global weather patterns. These stomatal parameters are estimated using leaf-level gas exchange by two alternative methods: (1) a response curve where the environmental conditions are modified for a single leaf, or (2) a survey approach, where repeated measurements are made on multiple leaves over a diurnal range of environmental conditions. We compare the curve and survey approaches by conducting a comprehensive measurement campaign in which we paired diurnal gas exchange surveys with leaf level response curves for the estimation of g1 and g0 on six tropical species across a full range of leaf phenological stages. We examine how these different estimates impact model projection of gs, and how the consideration of a diurnal effect on g1 and g0 can improve predictions relative to a model using parameter estimates which are fixed over the photoperiod. Our results showed that age is an important factor to consider in estimates of g0, however there was no effect of leaf age on estimates of g1. The survey approach identified a diurnal trend associated with g1 and g0, which when accounted for improved model projections of diurnal trends in gs. We found that while both approaches yield equally statistically valid estimates of g1 and g0 at a fixed point in time, they are not directly comparable across diurnal timescales, where shifting water supply and carbon demand lead to dynamic canopy scale water use efficiency (WUE). These results suggest that to improve the accuracy of modelled gs in tropical forests, TBMs should recognize and implement diurnal variation in stomatal parameters which are associated with diurnal shifts in WUE.