Modeling tree recovery and mortality following drought events

Abstract:

Elevated tree mortality in response to drought has been observed in most global forests in recent decades. Yet, forest die-off during such events is often patchy—with some individuals surviving, while others die. Our goal is to understand some of the basic mechanisms responsible for tree survival after drought events, and the period during which recovering trees are especially vulnerable to die from subsequent droughts. To do that, we develop a minimal model of water and carbon dynamics for the pinyon pine (Pinus edulis), a species that has suffered extensive die-off in the southwestern United States during recent decades. Our model describes the coupled dynamics of non-structural carbohydrates (NSC) and functional xylem area (X), as proxies for carbon and water status. Our main assumption is that the major mechanism for xylem recovery is the construction of new xylem (via radial growth) over long time scales of months and years. On the one hand xylem construction demands NSC expenditures, and on the other the functional xylem area prescribes the flow of water, thus influencing the tree carbon dynamics and the accumulation of NSC. In this deterministic model, the xylem area and carbon reserves following a drought event determine the fate of the trees, either death or full recovery over a span of many years. This division of the (X, NSC) space into two distinct basins depends on parameters related to the tree’s strategy in partitioning NSC between maintenance and growth respiration, the cost (in carbon units) of growing new xylem, and the Huber value. Critically, a tree that barely manages to survive can spend many years in the vicinity of the border between these two basins, being vulnerable to die even from mild drought events, many years after the first drought to reduce its functional xylem area. We show how our model provides a useful conceptual framework for the evolving understanding of tree death from both hydraulic failure and carbon starvation. We end our discussion by incorporating into the model uncertain parameter values and stochastic environmental conditions, yielding mortality probabilities on larger spatial scales. While tree mortality during extreme drought is visually obvious and alarming, our model indicates that surviving trees may be at an increased long-term risk of mortality.