A Maximum Entropy Approach for Inferring Ecosystem Hydraulic Trait Composition

Abstract:

Ecosystems consist of countless individual plants that interact and compete for nutrients, water and energy within their environments. These interactions present a scaling challenge for many terrestrial biosphere models. Data scarcity further limits the reliable parameterization of a wide range of hydraulic and physiological traits that vary within and between species. Therefore, there is a need for data-efficient approaches for inferring the ecosystem composition of these traits, so that they can be used to understand the dominant drivers of ecosystem water, carbon and energy exchange and facilitate the scaling of individual processes to the ecosystem level.

In recent years, the principle of maximum entropy (MaxEnt) has achieved success in explaining macroecological patterns such as the species abundance distribution and the abundance-energy relationship. MaxEnt is a statistical selection principle that treats microscopic states of a system as stochastic processes and imposes a limited number of physical macroscopic constraints on the system to estimate the most probable distribution of the microscopic states, which is associated with the highest information entropy. In this work, we aim to use MaxEnt to reveal the most probable composition of key plant hydraulic traits within ecosystems. To do so, we will constrain the probability distribution of individual plant fluxes (microscopic) with ecosystem-scale fluxes (macroscopic) derived from remotely sensed datasets (such as evapotranspiration, sensible heat flux, and gross primary productivity). We assume that plant hydraulic regulation governs water transport within mass and energy balances at the individual-level, which provides a mechanistic connection between the observed ecosystem states and the individual hydraulic trait distributions. The resulting trait distributions can be used to better characterize inter- and intra-specific trait variability within ecosystems and predict its influence on ecosystem resilience to climate extremes.