Modeling time variant transpiration schemes with StorAge Selection (SAS) functions: a methodology to integrate dynamic plant water capacitance in ecohydrologic simulation

Background/Question/Methods

Plant water storage is a critical determinant of the role vegetation plays in the water cycle. Much of the literature exploring plant capacitance considers only time-invariant transit times. Thus, these studies have not established how plant attributes and environmental conditions dynamically influence storage of transpiration and most popular ecohydrologic models neglect capacitance altogether.The transport of water through porous media can be mathematically characterized with StorAge Selection (SAS) functions. The probability distributions of SAS functions dynamically characterize how stored water molecules of different ages probabilistically combine to comprise outflow. While SAS functions have previously availed the approximation of the transit time of water solutes across landscapes, they have not commonly been utilized at the plant scale. We studied the time-varying influence of a suite of plant physiological and environmental variables on the relationship between stored and transpired water within three conifer species (Tsuga canadensis, Pinus strobus, Abies balsamea) using SAS functions fit to data collected through an isotope tracer-based experiment.Plants were enclosed within a gas-exchange apparatus for the collection of transpired water vapor. We applied 2H and 18O tracers and generated almost 2,500 isotope-based transpiration samples under varying climate and soil conditions, generating high resolution breakthrough curves of transpiration.

Results/Conclusions

Transpiration sourced from all plants indicated significant tracer concentration within a day of application (mean: +300‰, sd: ±90‰), with P. strobus demonstrating the highest enrichment and T. canadensis the lowest (directly corresponding with leaf area). Tracer breakthrough curves followed a gamma distribution pattern and a steady decline tracer recovery measured every 3 hours to ambient tracer levels 2 weeks following the commencement of the experiment. All plants demonstrated diurnal patterns in tracer emission, following daily transpiration peaks and declines. The results show methodological success, enabling us to assemble the highest resolution stable isotope-based dataset of plant transpiration.We utilize a non-parameteric statistical learning-based model to fit daily SAS functions to breakthrough data an explore how changes in plant physiology, climate, and soil moisture drive dynamism in plant capacitance. Finally, we integrate empirically validated relationships between SAS functions and studied environmental and plant variables into a land surface model to allow for the first direct dynamic simulation of plant capacitance.This project provides a scalable framework to measure and model plant water storage, significantly improving the accuracy of ecohydrological models in estimating the role vegetation plays in regulating the water cycle.