More support for Earth's massive microbiome

Abstract

Until recently, our planet was thought to be home to ~ 10⁷ species, largely belonging to plants and animals. Despite being the most abundant organisms on Earth, the contribution of microbial life to global biodiversity has been greatly underestimated and, in some cases, completely overlooked. Using a compilation of data known as the Global Prokaryotic Census (GPC), it was recently claimed that there are ~ 10⁶ extant bacterial and archaeal taxa [1], an estimate that is orders of magnitude lower than predictions for global microbial biodiversity based on the lognormal model of biodiversity and diversity-abundance scaling laws [2]. Here, we resolve this discrepancy by 1) identifying violations of sampling theory, 2) correcting for the misuse of biodiversity theory, and 3) conducting a reanalysis of the GPC. By doing so, we uncovered greater support for diversity-abundance scaling laws and the lognormal model of biodiversity, which together predict that Earth is home to 10¹² or more microbial taxa. REVIEWERS: This article was reviewed by Alvaro Sanchez and Sean M. Gibbons.

Keywords: Biodiversity; Biogeography; Census; Macroecology; Scaling.

Fig. 1

Results of distributing N = 10⁷ individuals belonging to S = 10⁵ species across a 2-D landscape (see S1 Fig). Subplots (a-d): Estimated and true richness under a) an unrealistic scenario where taxa are uniformly distributed in space with similar abundances, b) where species have uneven distributions of abundance but uniform distributions in space, c) where species have even distributions of abundance but aggregated distributions in space, and d) an ecologically realistic scenario where species have both uneven distributions of abundance and aggregated spatial distributions. For c and d, mean and standard deviations of normally distributed species spatial distributions where chosen at random. Information, source code, and simulated data can be found at https://www.github.com/LennonLab/census

Fig. 2

Diversity-abundance relationships (DARs) for microorganisms and macroorganisms (i.e., plants and animals) using data from [1, 2]. Abundance (N) refers to the number of individuals or sequence reads in a sample. Coefficients and exponents of scaling equations are mean values from 10,000 boot strapped multiple regressions, with each regression based on 500 assemblages chosen by stratified random sampling. Each scatter plot represents a single random sample; hulls are 95% prediction intervals. All code and data for recreating these analyses can be found at https://www.github.com/LennonLab/census