Variability of hydraulic traits across ten aspen (Populus tremuloides Michx.) stands in Utah: Implications for stomatal response modeling and drought stress vulnerability assessment

Background/Question/Methods

Aridity and drought events are expected to increase in many regions due to global climate change. Predicting their impact on forest productivity and tree mortality remains a challenge. A recently developed stomatal control model, based on an optimization criterion between photosynthesis gain and hydraulic risk, enables predicting stomatal responses under different environmental conditions (light, CO₂, temperature, soil moisture, relative humidity). Its parameterization requires plant traits such as xylem vulnerability curves (VCs), leaf to basal area ratio (LA/BA), leaf specific conductivity (LSC), and maximum rate of carboxylation (V_max). In order to evaluate the impact of plant trait variability on model predictions and the applicability of this model to large distribution areas we: (i) analyzed trait variability in ten aspen (Populus tremuloides Michx.) stands located across Utah; (ii) compared observed midday water potentials, Ψ_midday, and native stem conductance, k_s, and conductivity, K_s, to predicted values obtained running the model with individual tree traits vs. mean stand traits vs. mean species traits; and (iii) used the model to infer the combinations of predawn xylem pressure (Ψ_predawn) and vapor pressure deficit (VPD) that would threaten these stands by reducing net carbon assimilation (A_net) or increasing whole plant PLC relative to unstressed maxima.

Results/Conclusions

Root VCs were similar among stands and their mean 50% loss in hydraulic conductivity (P₅₀) ranged from -0.5 to -1.4 MPa. Stem VCs differed among stands and the mean P₅₀ of stems for stands ranged between -1.2 and -3.9 MPa. LA/BA and LSC were significantly different among stands; stand means ranged 1,851–2,815 cm²/cm² and 1.2–8.4 mmol s^-1 m^-2 MPa^-1, respectively. V_max estimations varied from 38 to 240 µmol m^-2 s^-1. The mean deviation between the observed and predicted Ψ_midday was 0.44 MPa running the model with individual tree values (n=59), 0.29 MPa running it with stand means (n=10), and 0.18 MPa running it with species mean for each stand (n=10). This suggests that running the model with mean species values provides better Ψ_midday predictions. Observed native K_s correlated with modeled k_s values for individual trees (r=0.44, p<0.001) but not for stand and species means (p>0.05). The model showed that A_net is negative at Ψ_predawn ≤-0.7 MPa for high VPD (4.25 Pa) and ≤-1.7 MPa for low VPD (1 Pa). 60-70 whole plant PLC increases mortality risk, and these values were reached at -1.1 and -1.5 MPa Ψ_predawn, respectively. Only one stand reached this threshold.