Application of a Bayesian nonparametric model to derive toxicity estimates based on the response of Antarctic microbial communities to fuel-contaminated soil

Ecotoxicology is primarily concerned with predicting the effects of toxic substances on the biological components of the ecosystem. In remote, high latitude environments such as Antarctica, where field work is logistically difficult and expensive, and where access to adequate numbers of soil invertebrates is limited and response times of biota are slow, appropriate modeling tools using microbial community responses can be valuable as an alternative to traditional single-species toxicity tests. In this study, we apply a Bayesian nonparametric model to a soil microbial data set acquired across a hydrocarbon contamination gradient at the site of a fuel spill in Antarctica. We model community change in terms of OTUs (operational taxonomic units) in response to a range of total petroleum hydrocarbon (TPH) concentrations. The Shannon diversity of the microbial community, clustering of OTUs into groups with similar behavior with respect to TPH, and effective concentration values at level x, which represent the TPH concentration that causes x% change in the community, are presented. This model is broadly applicable to other complex data sets with similar data structure and inferential requirements on the response of communities to environmental parameters and stressors.