Convergent approaches to determine an ecosystem's transpiration fraction

Abstract

The transpiration (T) fraction of total terrestrial evapotranspiration (ET), T/ET, can vary across ecosystems between 20–95% with a global average of ∼60%. The wide range may either reflect true heterogeneity between ecosystems and/or uncertainties in the techniques used to derive this property. Here we compared independent approaches to estimate T/ET at two needleleaf forested sites with a factor of 3 difference in leaf area index (LAI). The first method utilized water vapor isotope profiles and the second derived transpiration through its functional relationship with gross primary production. We found strong agreement between T/ET values from these two independent approaches although we noted a discrepancy at low vapor pressure deficits (VPD). We hypothesize that this divergence arises because stomatal conductance is independent of humidity at low VPD. Overall, we document significant synoptic-scale T/ET variability but minimal growing season-scale variability. This result indicates a high sensitivity of T/ET to passing weather but convergence toward a stable mean state, which is set by LAI. While changes in T/ET could emerge from a myriad of processes, including aboveground (LAI) or belowground (rooting depth) changes, there was only minimal interannual variability and no secular trend in our analysis of T/ET from the 15 year eddy covariance time series at Niwot Ridge. If the lack of trend observed here is apparent elsewhere, it suggests that the processes controlling the T and E fluxes are coupled in a way to maintain a stable ratio.