|
Abstract:
|
The taxonomic diversity of a biological population is commonly used to characterise the ecosystem and serve as a marker for ecosystem health. Diversity can be assessed within samples (alpha diversity), or compared across samples (beta diversity). Plug-in diversity estimators are extremely common in the ecology literature however fail to account for rare and unobserved taxa, which comprise a large component of many microbial communities. However, and perhaps more importantly, the literature generally ignores the randomness of the observed sample. For this reason, standard errors for plug-in diversity indices are treated as zero, which severely understates their precision. We present an overview of methods that appropriately account for both rare taxa and sampling variability. We propose an approach for modelling diversity indices that permits formal inference for the effect of covariates and the presence of excess variability. The latter is of especial interest when biological replicates are under study, because the current theory is insufficient to confirm homogeneity of replicates with respect to community richness or evenness. Joint work with John Bunge and Thea Whitman.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.