Statistical Estimation of Uncultivated Microbial Diversity

Abstract

The full microbial richness of a community, or even of an environmental sample, usually cannot be observed completely, but only estimated statistically. This estimation is typically based on observed count data, that is, the counts of the representatives of each species (or other taxonomic units) appearing in the sample or samples. “Abundance” data consists of counts of the numbers of individuals from various species in a single sample, while “incidence” (or multiple recapture) data consists of lists of species appearing in several or many samples. In this chapter we consider statistical estimation of the total richness, i.e., the total number of species, observed + unobserved, based on abundance or on incidence data. We discuss parametric and nonparametric methods, their underlying assumptions, and their advantages and disadvantages; computational implementations and software; and larger scientific issues such as the scope of applicability of the results of a given analysis. Some real-world examples from microbial studies are presented. Our discussion is intended to serve as an overview and an introduction to the literature and available software.