Phospholipid fatty acid (PLFA) analysis as a tool to estimate absolute abundances from compositional 16S rRNA bacterial metabarcoding data

Highlights

• Substrate type may lead to different biases in environmental DNA monitoring (eDNA).

• We performed eDNA and phospholipid fatty acid analysis (PLFA) of 5 substrates.

• Variation in eDNA and PLFA views of bacterial composition differed among substrates.

• PLFA and relative abundances of taxa were combined, enabling samples to be compared.

• Our approach increases intercomparison of eDNA data from different substrates.

Abstract

Microbial biodiversity monitoring through the analysis of DNA extracted from environmental samples is increasingly popular because it is perceived as being rapid, cost-effective, and flexible concerning the sample types studied. DNA can be extracted from diverse media before high-throughput sequencing of the prokaryotic 16S rRNA gene is used to characterize the taxonomic diversity and composition of the sample (known as metabarcoding). While sources of bias in metabarcoding methodologies are widely acknowledged, previous studies have focused mainly on the effects of these biases within a single substrate type, and relatively little is known of how these vary across substrates. We investigated the effect of substrate type (water, microbial mats, lake sediments, stream sediments, soil and a mock microbial community) on the relative performance of DNA metabarcoding in parallel with phospholipid fatty acid (PLFA) analysis. Quantitative estimates of the biomass of different taxonomic groups in samples were made through the analysis of PLFAs, and these were compared to the relative abundances of microbial taxa estimated from metabarcoding. Furthermore, we used the PLFA-based quantitative estimates of the biomass to adjust relative abundances of microbial groups determined by metabarcoding to provide insight into how the biomass of microbial taxa from PLFA analysis can improve understanding of microbial communities from environmental DNA samples. We used two sets of PLFA biomarkers that differed in their number of PLFAs to evaluate how PLFA biomarker selection influences biomass estimates. Metabarcoding and PLFA analysis provided significantly different views of bacterial composition, and these differences varied among substrates. We observed the most notable differences for the Gram-negative bacteria, which were overrepresented by metabarcoding in comparison to PLFA analysis. In contrast, the relative biomass and relative sequence abundances aligned reasonably well for Cyanobacteria across the tested freshwater substrates. Adjusting relative abundances of microbial taxa estimated by metabarcoding with PLFA-based quantification estimates of the microbial biomass led to significant changes in the microbial community compositions in all substrates. We recommend including independent estimates of the biomass of microbial groups to increase comparability among metabarcoding libraries from environmental samples, especially when comparing communities associated with different substrates.