DNA metabarcoding is not quantitative, but how should we evaluate if sequence counts are a useful proxy for biomass?

Bruce Deagle¹

¹CSIRO (Australian National Fish Collection), Battery Point, Australia

Abstract:

It is generally accepted that sequence read counts obtained from high-throughput sequencing can provide precise estimates of proportional DNA composition of a sample. However, these proportions are not an accurate reflection of species biomass due to technical and biological biases. How inaccurate are these count proportions? Are they useful proxies for biomass? These questions are still unanswered in most study systems where DNA metabarcoding is applied.

We examine studies from the literature where researchers have sequenced “mock communities” of known composition and try to make sense of the results. We outline several challenges that make interpretation of these studies difficult. One issue is that many experiments use mixtures of DNA, these data are not directly comparable to those obtained from mixtures of organisms. Another issue is the use of statistical correlation to evaluate the agreement between what goes in and what comes out. This analysis can provide very misleading results, and the problem can be clearly illustrated with published data sets.

While there is no easy approach to evaluate if sequence counts are a useful, we outline some pragmatic ideas. We also outline our attempts to answer this question in a study system using DNA metabarcoding to study seal diet. Our analysis includes tissue mixtures, feeding trials, simulations, and sequencing of over 4500 field collected samples.

Biography:

Bruce is a principal research scientist at the CSIRO Australian National Fish Collection (ANFC) and he has had a diverse 20 year research career in the field of molecular ecology. Currently he is particularly interested in development and application of environmental DNA tools to study fish biodiversity, and applied genetic approaches for fish conservation and management.