A comparison of three approaches to modeling leaf gas exchange in annually drought-stressed ponderosa pine forests

We tested, compared and modified three models of stomatal conductance at the leaf level in a forest ecosystem where drought stress is a major factor controlling forest productivity. The models were tested against 2 years (1999 and 2000) of leaf-level measurements on ponderosa pine (Pinus ponderosa Dougl. ex Laws.) growing in the Mediterranean climate of California, USA. The Ball, Woodrow and Berry (1987) (BWB) model was modified to account for soil water stress. Among the models, results of the modified BWB model were in the closest agreement with observations (r(2) = 0.71). The Jarvis (1976) model showed systematic simulation errors related to vapor pressure deficit (r(2) = 0.65). Results of the Williams, Rastetter, Fernandes et at. (1996) (SPA) model showed the poorest correlation with empirical data, but this model has only one calibration parameter (r(2) = 0.60). Sensitivity analyses showed that, in all three models, predictions of stomatal conductance were most responsive to photosynthetically active radiation and soil water content. Stomatal conductance showed little sensitivity to vapor pressure deficit in the Jarvis model, whereas in both the BWB and SPA models, vapor pressure deficit (or relative humidity) was the third most important variable. Parameterization of the SPA model was in accordance with the parameterization of the modified BWB model, although the two models differ greatly. Measured and modeled results indicate that stomatal behavior is not water conservative during spring; however, during summer, when soil water content is low and vapor pressure deficit is high, stomatal conductance decreases and, according to the models, intrinsic water-use efficiency increases.