Modeling effects of rare taxa in community ecotoxicology

Background/Question/Methods   The majority of taxa exhibit rarity in their abundance or distribution, but the effects of rare taxa in community ecotoxicology remain controversial. On one hand, including rare taxa might be expected to increase the abundance of intolerant organisms in a sample, thus increasing community dose-response sensitivities. On the other hand, excluding rare taxa may reduce variability within reference communities, thus increasing the probability of detecting non-reference conditions. We used modeling and simulation techniques to evaluate these contrasting scenarios. We used quantile regression techniques to model benthic macroinvertebrate community responses to a land-use stressor gradient in the mid-Atlantic highlands, USA. We then applied quantile models to simulate ecological degradation within reference communities of the Shenandoah National Park, USA.

Results/Conclusions   The effects of rare taxa varied among community indices and stress levels. Community-similarity indices (i.e., Bray-Curtis distances) exhibited greater sensitivity to simulated stressors than measures of taxonomic completeness (observed/expected taxa, O/E) for detecting low levels of stress. Excluding rare taxa improved O/E precision (i.e., decreased standard deviation of reference distributions) and increased O/E sensitivity at intermediate stress levels. However, the sensitivity of Bray-Curtis distances diminished when rare taxa were excluded from index calculations. Our results demonstrate that community ecotoxicology indices are not equivalent as early-warning indicators and suggest that rare taxa can either strengthen or weaken early-warning signals of ecological change depending on the index.