What Drives the Destiny of Tropical Forests? Lessons Learned from a Full Variability Partitioning of a Mechanistic Vegetation Model

Abstract:

There is no doubt that the fate of the terrestrial carbon sink in the coming century is linked to the productivity of the tropics. Tree mortality rates have been increasing in moist tropical forests due to key drivers including elevated atmospheric CO₂, climate warming, and drought. While CO₂ and climate temperature are expected to increase on average in the future, extreme climate events such as heat waves, droughts and floods are predicted to increase in both frequency and intensity. To survive, tropical plants must be able to react and adapt quickly to short term events while living in a climate of slowly shifting means.

The Ecosystem Demography model 2.2 with mechanistic hydraulics (ED2-hydro) has been shown to credibly predict the carbon dynamics of tropical forests under the current climate and rainfall manipulation droughts. Furthermore, due to its physics-based mechanics, it is appropriate to use ED2-hydro to extrapolate and explore hypothetical scenarios. In this study we use ED2-hydro to make 100 year projections at Barro Colorado Island, Panama, a lowland tropical forest incorporating within climate scenario variability, parameter uncertainty, and initial condition variability, resulting in a full variance partitioning of model predictive variability. Doing so allows us to quantify the main drivers behind the differences in model projections.

Previous work partitioning model uncertainty in long term hindcasts at Harvard forest found initial condition uncertainty and process uncertainty as dominant drivers of model uncertainty. Climate uncertainty initially contributed little but grew substantially over time, a pattern we expect to see in the tropics. With constraint, parameter uncertainty should contribute the least to model uncertainty, barring covariance with climate and initial conditions. Extending the variance partitioning approach to future projections in the tropics holds value, not only for the modeling community as we seek to improve model accuracy and precision, but also for the development of climate protection policy.