Causal inference in Evolutionary Ecology and Conservation

Disentangling cause-effect relationships is a primary, although often understated, goal in ecological and evolutionary studies. However inferring causality is a daunting task in the absence of controlled and randomized experiments, which, more often than not, are unfeasible in ecology and evolutionary biology. With few exceptions field ecologists and evolutionary biologists thus renounce to make any inference about causality from their observatory studies, resigning to the sobering precept - we all learnt during our undergraduate statistics courses - that correlation does not imply causation. In the last 20 years new powerful and elegant methods for causal inference from observational data, based on graph theory (e.g. Structural Equation models and Path analysis) have been developed and only recently introduced in ecology and in evolutionary biology. The further development of these methods is currently a very active field of research but the application of these methods in ecology, evolution and conservation is yet only at the surface of its potential. I have applied these methods extensively during my career, using them to disentangle the complex relationships between corticosteroid hormones, behaviour and life history traits (Decristophoris et al. 2007, Costantini et al. 2012, Corlatti et al. 2012) and recently contributed to introducing them to molecular ecology (Brambilla et al. 2015). In collaboration with Dr. Alejandro Gonzalez-Voyer (UNAM, Mexico), I have recently extended causal inference to phylogenetic comparative studies developing a method for Phylogenetic Path Analysis (von Hardenberg and Gonzalez-Voyer, 2013, Gonzalez-Voyer and von Hardenberg 2014), an approach that we are now developing further in a Bayesian Structural Equation Modeling framework. This work has been supported by a University of Chester International Research Excellence Award, funded under the Santander Universities scheme.