Nonparametric richness estimators Chao1 and ACE must not be used with amplicon sequence variant data
Yongcui Deng, Alexander K Umbach, Josh D. Neufeld
Abstract
Microbial ecologists use alpha diversity metrics for estimating species richness and evenness from data obtained by high-throughput sequencing of small subunit ribosomal RNA genes. This perspective argues that the nonparametric richness estimators Chao1 and ACE should never be used with ASV data because the default process of generating amplicon sequence variants (ASVs) removes singletons, which are specifically required for these estimate calculations. In addition, retaining singletons, if/when possible, will contribute a large proportion of artifacts to this abundance category, leading to extreme richness overestimation. We recommend the use of alternative sequence clustering strategies and/or diversity metrics to avoid generating meaningless richness estimates from ASV data.