Models of ecosystem water loss have been steadily improving through the incorporation of plant hydraulics into canopy stomatal conductance. Recently, a simple plant hydraulic model has emerged that can predict changes in canopy stomatal conductance to environmental drivers such as light, vapor pressure deficit and soil moisture. The model incorporates the now well supported hypothesis that most plants adjust stomatal conductance to regulate minimum leaf water potential to prevent excessive and potentially catastrophic cavitation. We have incorporated this plant hydraulic submodel into the Terrestrial Regional Ecosystem Simulator (TREES) model which simulates ecosystem fluxes of mass and energy. TREES can be used in both a forward and inverse mode to investigate how parameter uncertainty in the hydraulic submodel impacts predictive understanding of ecosystem water fluxes. We tested the ability of TREES to capture the dynamics of sap flux scaled transpiration and canopy stomatal conductance in both time and space across 14 species in disparate ecosystems including heterogeneous mixed forests of Wisconsin, boreal forests of Manitoba, subalpine forests of Wyoming and shrub steppe of Wyoming.

Results/Conclusions

Across all of these species, the hydraulic submodel was able to explain the tradeoff between high canopy stomatal conductance/rapid decline with increasing atmospheric drought and a low canopy stomatal conductance/slow decline with increasing atmospheric drought. The only species that violated the simple plant hydraulic model was boreal black spruce (Picea mariana). We were able to explain this discrepancy by changing the hydraulic model to allow the minimum leaf water potential of black spruce to decline as the stand ages went from 11 to 151 years. We further tested TREES by comparing simulations of spatially explicit tree transpiration to spatial transpiration data from the Wyoming, Wisconsin and Manitoba field sites. At each site, transpiration from over 100 individual trees was estimated from sap flux across a soil moisture gradient within a stand. Across all sites, we found that spatial autocorrelation in transpiration declined as atmospheric drought increased indicating a direct connection between temporal and spatial dynamics. TREES was able to capture this time and space connection qualitatively, but was not able to predict quantitatively the spatial distribution of transpiration. Our results show that we now have predictive understanding of transpiration dynamics in time across many disparate plant species. Our future work will focus on determining which environmental and tree variables may explain the distinct spatial patterning of transpiration and its response to environmental drivers in time