What is the functional composition of a landscape in Borneo, or of one in central England? Knowing the answer is critical for mapping ecosystem functioning (for example, carbon fluxes), for conservation work (for example, understanding where landscapes have more functions and may be more resilient), Traditionally, answering this question could be done based on one of two approaches. The first approach has high precision but low spatial coverage, and involves intensive field surveys at a few focal locations. The second approach has high spatial coverage but low precision, and involves remote sensing to classify landscapes into broad vegetation types (e.g. forest vs. tundra). Finding a compromise between these approaches has been challenging.

Ethan Butler and colleagues just published a nice paper mapping plant functional traits across the entire globe. I am a minor co-author. The study is one of the first efforts to do this comprehensively (although some other work by Irena Simova and myself will also be coming out soon!). The study draws upon global databases of plant occurrences and trait measurements, as well as global maps of environment and land cover. Via some very clever Bayesian spatial modeling methods they are able to interpolate trait distributions within grid cells over the entire planet, and show that the model predicts distributions consistent with observations in key test locations.

The paper, Mapping local and global variability in plant trait distributions, is just out in PNAS – you can read it online or get a PDF copy. Most of the heavy-duty math is in the appendix, so the main text is very accessible. Please give it a read!