Bayesian calibration of a physiological model infers multiple interactive causes of drought-induced mortality

Background/Question/Methods

We conducted a Bayesian calibration of a steady-state model aimed at describing the physiological processes leading to plant death following extreme drought events. The model incorporates important feedbacks between water and carbon fluxes in plants and especially a representation of the interactions between leaf-level gas exchange, xylem and phloem water and sucrose transport and substrate utilization by the plant. We employ field measurements collected over three years describing water and carbon relations of declining Scots pine in Northern Spain to calibrate the model parameters using a Bayesian approach.

---

Results/Conclusions

 The model was constrained well by the field measurements and allowed estimating properties, such as phloem turgor, that could not be determined directly in the field. Simulation results supported the conclusion that multiple independent processes (run-away cavitation, negative carbon balance and cessation of phloem transport) were jointly to blame for the observed mortality. The inability of this model to isolate these proposed mechanisms of death may be an inherent feature of the drought-induced mortality process. Future efforts should concentrate on validating the model using independent data.