Does inclusion of plant hydro-dynamics improve drought-response predictions of dynamic vegetation models?

Background/Question/Methods

Climate change and land conversion is causing shifts in precipitation patterns across the Amazon basin; yet, there remains considerable uncertainty about the resilience of its rainforests under increasing water stress in the coming century.  Simulations by four leading dynamic vegetation models—CLM3.5, ED2, IBIS and JULES—revealed limited ability to replicate both the 20% decline in aboveground biomass and compositional shifts observed in two long-term large-scale ecosystem drought experiments in the eastern Brazilian Amazon. These four models performed poorly in part because they were not parameterized to mechanistically represent the competitive differences in hydraulic traits that exist between species, which prior to this study was unknown.  Therefore, we measured xylem vulnerability to cavitation and sensitivity of stomatal conductance to leaf water potential, as determined by turgor loss point (TLP), in order to parameterize a hydrodynamic pipe model for inclusion in the Ecosystem Demography (ED2) dynamic vegetation model. Measurements were made on upper canopy branches and leaves of four genera common to both drought experiment sites.  Each genus was placed a priorinto one of four functional categories: drought tolerant versus intolerant based on the drought experiment results and early versus late successional based on wood density.  We evaluated the hypothesis that these two important traits for regulating the plant hydraulic system would be significantly different between each of the plant functional groups and thus explain the observed differential mortality rates.

Results/Conclusions

P50 values for both xylem cavitation and stomatal closure (i.e. TLP) occurred at 1.0 MPa and 0.5 to 0.8 MPa, respectively, higher water potentials in the drought intolerant plant functional group compared to the tolerant group.  In comparison, the early versus late successional plant functional groups showed no significant differences in xylem P50 and TLP.  These hydraulic trait values were used to parameterized a hydrodynamic pipe model, which effectively tracked the flow of water through trees of each functional group and showed differential water-use between the drought tolerant and intolerant groups, thus suggesting differential mortality rates.  The ecological implications during droughts of these differences in hydraulic traits can be evaluated fully once implemented in ED2, which currently tracks competitive dynamics between trees, but lacks a realistic formulation for water movement through the soil-plant-atmosphere continuum.