Feedback significantly influences the simulated effect of CO2 on seasonal evapotranspiration from two agricultural species

Abstract

SummaryThe direct effect of elevated carbon dioxide on evapotranspiration over a growing season was investigated by scaling up single‐leaf gas exchange measurements on soybean and corn plants grown and measured at three carbon dioxide concentrations. Stomatal conductance decreased markedly with increasing carbon dioxide in these species under most conditions. Coupled soil–vegetation–atmosphere models were used to scale up these single‐leaf level measurements to simulate evapotranspiration at the regional scale from planting to harvest. The coupled modelling system introduced feedbacks over the season that are not present at the measurement level, which decreased the effect of carbon dioxide on evapotranspiration. Four sets of simulations were performed to evaluate specifically the magnitude of four feedbacks; two resulting from scale, surface layer and mixed layer feedback, one resulting from soil evaporation and one resulting from the interactions of stomatal conductance and the simulated canopy microclimate (physiological feedback). The feedbacks occurring from scale were consistent with previous analytical work indicating that transpiration becomes less dependent on stomatal conductance at larger scales. Evaporation from the soil has been generally neglected in past studies on carbon dioxide effects, but was especially important in decreasing the effects of carbon dioxide on evapotranspiration and showed a seasonal dynamic. The feedback resulting from physiological responses has also received less attention than the feedbacks from scale, but was only moderately important in these simulations. We also investigated the seasonal dynamics of how the observed increase in leaf area at elevated carbon dioxide affects evapotranspiration. Considering all the feedbacks and the observed increase in leaf area at elevated carbon dioxide, the simulated decrease in evapotranspiration was not negligible but was much less than the decrease in stomatal conductance. At the regional scale and maximum complexity in our model, the simulated decrease in seasonal evapotranspiration at doubled carbon dioxide (700 μmol mol–1) was 5.4% for soybeans and 8.6% for corn.