Carnegie Mellon University

The worldwide leaf economic spectrum: How causal discovery algorithms forced me to re-imagine its generating causes: Bill Shipley

Abstract: This story begins with four leaf traits related to the main "economic" function of a leaf (i.e. to fix carbon that contributes to growth, reproduction and survival): specific leaf mass (SLM, the ratio of leaf dry mass: leaf projected surface area), leaf nitrogen content per mass (Nm, the mass of nitrogen per unit leaf dry mass), maximum leaf net photosynthetic rate (Am, the net amount of carbon fixed per unit leaf mass per unit time) and leaf lifespan (L, the average number of days before programmed leaf death). A large body of evidence shows that species of vascular plants all over the world, typical of every type of habit, display the same pattern of correlations between these variables. What causes this common pattern of covariation? It is impossible to answer question this using experimental manipulations. I (and many others) thought that we understood this from basic physiology but structural equations modelling shows us to be wrong. I explain how the Fast Causal Inference algorithm and the Tetrad representation theorem led me to a new hypothesis.