PNNL

Abstract

?Accurate trait-environment relationships and global maps of plant trait distributions represent a needed stepping stone in global biogeography and are critical constraints of key parameters for land models. Here, we use a global data set of plant traits to map trait distributions closely coupled to photosynthesis and foliar respiration: specific leaf area (SLA), and dry mass-based concentrations of leaf nitrogen (Nm) and phosphorus (Pm). ?We propose two models to extrapolate geographically sparse point data to continuous spatial surfaces. The first is a categorical model using species mean trait values, categorized into plant functional types (PFTs) and extrapolating to PFT occurrence ranges identified by remote sensing. The second is a Bayesian spatial model that incorporates information about PFT, location and environmental covariates to estimate trait distributions. Both models are further stratified by varying the number of PFTs. ?The performance of the models was evaluated based on their explanatory and predictive ability. The Bayesian spatial model leveraging the largest number of PFTs produced the best maps. ?The interpolation of full trait distributions enables a wider diversity of vegetation to be represented across the land surface. These maps may be used as input to Earth System Models and to evaluate other estimates of functional diversity.