Poster Presentation Society for Molecular Biology and Evolution Conference 2016

Counting with DNA: can species proportional biomass be estimated using high-throughput amplicon sequencing? (#341)

Bruce E Deagle 1 2 , Laurence J Clarke 1 2 , Austen C Thomas 3 4 , Andrew W Trites 4 , Karen J Westwood 1
  1. Australian Antarctic Division, Kingston, Tasmania, Australia
  2. Antarctic Climate & Ecosystems Cooperative Research Centre, University of Tasmania, Hobart, Tasmania, Australia
  3. Smith-Root Inc., Vancouver, Washington, USA
  4. Marine Mammal Research Unit, University of British Columbia, Vancouver, British Columbia, Canada

DNA metabarcoding using high-throughput amplicon sequencing is a powerful tool for characterizing species assemblages. However, using DNA metabarcoding to quantify relative species abundances is currently limited by both biological and technical biases that influence sequence read counts. To assess the problem and explore whether correction factors can improve quantitative estimates, we examined sets of known species mixtures using Illumina amplicon sequencing. In the first study, we characterised mixtures of single-celled marine protists containing either just diatom species or including a broader range of protists from several different phyla. Species-specific biases in recovery of 18S rDNA barcodes compared to cell counts were apparent, and were particularly strong for some diatoms. In mixes covering broad taxonomic groups, corrections based on cell volumes improved quantitative estimates of cell numbers, but this did not apply to diatoms. In the second study, we examined mixtures of tissue from prey species consumed by harbour seals. The goal was to improve diet estimates obtained through DNA-based analysis of faeces by accounting for differential recoverability of prey DNA. Our approach was to sequence mtDNA amplicons recovered from 50/50 mixtures of target fish species and a control species to generate relative correction factors (RCFs). These RCFs measured how under- or over-represented a particular species was relative to the control. In our experiment RCFs remained consistent regardless of the co-occurring species, but RCFs did vary with input ratio (i.e. biases changed when mixes were not 50/50). Still, 50/50 RCFs applied to DNA sequence counts from various mixtures of the target species significantly improved relative abundance estimates of proportions. Overall, our results indicate that correction factors applied to metabarcoding datasets improve estimates of relative species abundance in simple communities.