Guest Post from Connor Whitaker
I’m a student in a graduate-level discussion-based seminar that Michael is teaching: “Frontiers in Biodiversity Measurement.” As part of the course, we are expected to “take point” on one paper during the semester, lead the discussion, and craft a tutorial in R. We were given a lot of creative liberty with how to structure the tutorial, with the guidance that our classmates would get the most from it if we related the tutorial to the paper we were discussing that week.
Michael sent out an email about a week before the class began asking everyone to fill out a survey to assess knowledge, and also asked one brave (or in my case, impulsive and a little foolhardy) student to lead the discussion for the second week of class. I decided to take him up on the challenge to go first, and ended up with a finished product I’m quite proud of.
The paper for my discussion was the 2013 paper “Scale-dependent effect sizes of ecological drivers on biodiversity: why standardised sampling is not enough” by Jonathan Chase and Tiffany Knight. In a nutshell, the paper goes as follows:
- The effects of ecological drivers (any phenomenon that causes ecological change) are highly dependent upon scale, but this is not accounted for in many experimental studies.
- The authors used a model to investigate and compare several biodiversity metrics. Each of these metrics responds differently to variation in biodiversity across scale.
- They proposed a specific metric, ENSPIE, (the Effective Number of Species calculated from Hurlbert’s (1971) Probability of Interspecific Encounter) to assess to what extent the effects of various ecological drivers are scale-dependent or independent, and demonstrated this through simulating communities and comparing treatments across different metrics.
- The authors concluded with how researchers could use multiple biodiversity metrics in combination to draw more accurate conclusions and predictions about the direction and magnitude of ecological drivers.
I’ve been interested in modeling community behavior and dynamics for some time. Building off my own curiosity and research interests, I decided to base my tutorial off the paper’s simulation. Along the way, I discovered that one of the authors, Jonathan Chase, contributed to an R package, “mobsim,” which ported much of the functionality of their original simulation (written in Matlab) into R. This made writing the code in line with my prior level of programming experience.
The tutorial was designed to demonstrate that ENSPIE (Hill-Simpson diversity) is a powerful metric for deducing whether a driver is effecting a scale-dependent change, or if instead, the driver produces consistent effects across scale. This is critical for conservationists on a changing planet. Having a way to detect consistent effects across scale could enable better predictions based on field research, which is often done, by necessity, at a smaller scale than the effects it may be attempting to document. This will enable better understanding broad patterns of various drivers, and when we can and cannot extrapolate results from a smaller scale.
The tutorial code is available on my github. I hope you enjoy playing around with it as much as I enjoyed writing it.