Tree mortality due to drought has major impacts on forests, more so under climate change scenarios. Yet, our ability to predict the conditions that kill trees is hindered, in part, by poor understanding of the mechanisms by which trees avoid dehydration and the thresholds of dehydration that cause mortality. When soil water is unavailable, trees generally slow water loss by closing stomata and shedding leaves and fine roots, helping to maintain water and hydraulic conductivity within stems. However, these mechanisms are often ineffective, leading to drought mortality associated with failure of stem hydraulic conductivity. I subjected eight-month old seedlings of, Cavanillesia platanifolia, a stem-succulent tropical forest tree, to drought conditions and monitored their leaf and root surface area, stem water potential, wood water deficit, stem hydraulic conductivity, and mortality in order to test drought survival limits, to better understand what thresholds are associated with mortality, and to examine the process of stem dehydration during drought.
Results/Conclusions
The seedlings survived and maintained stem water potential >-2 MPa for up to eight months despite soil water potential reaching <-10 MPa and leaves and fine roots being shed within 2 months of drought. During the experiment, the water loss rate from the wood was nearly constant, reaching a wood water deficit of 800 kg m-3. Mortality was associated with >80% loss of stem hydraulic conductivity and with a phase change in the relationship between stem water potential and wood water deficit. Near death, small amounts of water loss were associated with large drops in stem water potential. These results confirm that tree mortality is associated with stem hydraulic failure. They also point to the depletion of wood water reserves as a threshold for hydraulic failure. Since water loss continued after leaf and fine root shedding, it was likely lost through the bark. Indeed, simply accounting for bark water vapor conductance and stem-to-air vapor pressure deficit accurately predicted stem water deficit throughout the experiment. These results suggest that predicting drought mortality, even in large trees, depends on understanding when trees lose access to soil water, their stem water reserves, and their rate of water vapor loss through bark.