Microbial community analysis in incompletely or destructively sampled systems.

Analyses of microbial community dynamics are often constrained by the destructive, indirect, and incomplete nature of most sampling techniques. These methodological constraints compel assumptions that are rarely verified about the relationships among separate communities. We evaluated the consequences for community analysis of the common assumption that separate microbial communities are described by the same species abundance distribution. Sample data were generated from simulated communities in which the species abundance distributions were the same or were different. Samples from communities that had the same number of species or were described by the same species abundance distribution sometimes had significantly different numbers of species. Samples from simulated communities that had different species number-species abundance distribution combinations sometimes contained indistinguishable numbers of species. When sampling from independent communities described by unknown distributions (e.g., microbial communities on plant surfaces), the simulations showed that standardization of sample size (number of individuals or colony-forming units) does not guarantee samples of equal proportions of the total species in a community. Sample sizes that are logistically feasible for many microbial systems will provide only limited information for differentiating species numbers or species abundance distributions among separate communities over time. For ecologists studying destructively or incompletely sampled communities this seriously influences both the sample designs that are reasonable and the questions that can be addressed in such systems.