Empirical stomatal conductance models reveal that the isohydric behavior of an Acacia caven Mediterranean Savannah scales from leaf to ecosystem

Abstract

Canopy conductance (g(c)) is the main controller of plant-atmospheric interaction and a key element in understanding how plants cope with drought. Empirical g(c) models provide a good inference as to how environmental forcing affects surface water vapor and CO2 gas exchange. However, when facing water scarcity, soil moisture or plant water availability becomes the primary controller. We studied g(c) in an Acacia caven (Mol) savannah in Central Chile under Mediterranean-type climate conditions that present distinguishable wet and dry seasons. We calibrated an empirical g(c), in order to account for whole canopy gas exchange with g(c) measurements from three different data sets: (1) an inversion of the Penman-Monteith equation in combination with a Shuttleworth and Wallace model (PMSW) for evapotranspiration from sparse canopies; (2) an inversion of the Penman-Monteith (PM) based on the big leaf approach and (3) a set of leaf stomatal conductance (g(s)) ground based measurements taken throughout the season and scaled up to the canopy level. Then the semi-empirical Farquhar-Ball-Berry (FBB) g(c) model was added to the comparison to evaluate if the inclusion of a mechanistic component for photosynthesis would improve the prediction of g(c). Models performance was assessed with ground based leaf gas exchange measurements during both wet and dry seasons. Acacia's g(c) showed a high synchronicity with soil moisture, exhibiting the typical isohydric behavior of this kind of vegetation. The addition of the Shuttleworth and Wallace modifier to the Penman-Monteith equation did not yield a better calibration for the multiplicative model when compared to the one calibrated with the PM g(c) data set, however this does not directly certifies that PM itself is a better estimator of g(c) in sparse canopies. Furthermore, scaling issues such as ecosystem heterogeneity and patchiness must be considered when applying these estimations to a watershed level for both eco and hydrological reasons. These empirical models demonstrated to be a good tool for predicting stomatal behavior for this kind of vegetation. Nevertheless, the effect of deep soil moisture on plant water status must be integrated in g(c) estimations in order to improve model's performance.