Modeling bacterial species abundance from small community surveys.

Abstract

Taxon abundance patterns are a central focus in evolution and ecology, providing the basic architecture of natural assemblages and potential clues to their formative processes. To better interpret species abundance patterns in natural microbial communities, we examined the consequences of three fundamental types of abundance patterns--uniform, geometric, and lognormal distributions. Theoretical communities were constructed based on the three pattern types with 2000 to 20,000 species and 10(7) to 10(10) individuals. The "dominant" species (species 1) among models that differed only in pattern type generally varied in abundance by 1 to 3 orders of magnitude. At the extremes among all the models examined, the dominant species comprised as much as 16% and as little as 0.005% of the total community. Analysis of the models and comparison with seven published surveys suggests that entire soil bacterial communities do not routinely exhibit Preston's cannonical subset of lognormal distributions. Use of the models to evaluate survey limitations showed that common diversity indices are generally sensitive to sample size over the range (50 to 200 clones) commonly used for microbial communities, emphasizing the need to compare surveys of similar size. The results collectively demonstrate that uniform, geometric, and lognormal distributions have profoundly different experimental and ecological consequences. Further, defined abundance models provide a simple quantitative tool for evaluating abundance patterns in clone libraries (even small ones) from natural communities.