Demystifying Chao1 with Good-Turing

Abstract

To estimate the ”biodiversity” in a particular area – or, in other words, the number of unique species living in a given environment – ecologists usually have no other option than to rely on incomplete samples. For all sorts of practical reasons, it is virtually impossible to spot all species that are actually present in an area, and hence certain species will be missing in the counts. Consequently, an important research question in ecology is how to reliably estimate the resulting ”bias” between the number of species that was observerd and the true number of unique species in an area. In this brief note, I try to demystify a famous unseen species model, Chao1, in terms of Turing’s work. I hope that this will help people unfamiliar with the method to understand exactly how it works, and, perhaps more importantly, how (not) to interpret the calculated values