11 — How many dinosaur species were there?

The authors of this paper propose TRiPS, or “True Richness estimated using a Poisson Sampling model,” a technique they use to place the number of dinosaur species at around 2000 (they hypothesize 1936 specifically — in the range [1543-2468]).

The method consists of pulling observations from PaleoDB and classifying them by location, age, and other features: The authors run Birth-Death (BD) simulations, adding to the outcomes of Survival or Extinction a third option, Fossilization, with a probability modeled by math that takes more than one line of LaTeX to explain. (Actually, it looks based on the font that the authors wrote the formulas in MS Word, something I also noticed in the MantisBot paper).

The paper introduced me to the Common Cause hypothesis of paleontology — that the richness of samples found in a rock and the ease of interrogating that rock (as well as other factors) co-vary closely because they are caused by the same (potentially unknown) conditions. This is, of course, a huge confound for paleo statistics, I imagine; but the authors leverage it as a known-bias to inform their TRiPS model.

There are some very cool figures in this paper (if you’re reading the bioarXiv version like I am, scroll all the way down) that chart species richness across time. This is really neat because the minima in the chart match known extinction events — except for the Jurassic-Cretaceous extinction event, on which, claim the authors, this model casts doubt.