Quantitative metagenomics for marine prokaryotes and photosynthetic eukaryotes

Abstract

High-throughput sequencing has provided unprecedented insights into microbial biodiversity in marine and other ecosystems. However, most sequencing-based studies report only relative (compositional) rather than absolute abundance, limiting their application in ecological modeling and biogeochemical analyses. Here, we present a metagenomic protocol incorporating genomic internal standards to quantify the absolute abundances of prokaryotes and eukaryotic phytoplankton, which together form the base of the marine food web, in unfractionated seawater. We applied this method to surface waters collected across 50°N to 40°S during the 29^th Atlantic Meridional Transect. Using the single-copy recA gene, we estimated an average bacterial abundance of 1.0 × 10⁹ haploid genome equivalents per liter. Leveraging a recent report that the psbO gene is typically single-copy in phytoplankton, we also quantified eukaryotic phytoplankton. Metagenomic estimates closely aligned with flow cytometry data for cyanobacteria (slope = 1.03, Pearson's r = 0.89) and eukaryotic phytoplankton (slope = 0.72, Pearson's r = 0.84). Compared to flow cytometry, taxonomic resolution for nano- and picoeukaryotes was greatly improved. Estimates for diatoms, dinoflagellates, and Trichodesmium were considerably higher than microscopy counts, likely reflecting microscopy undercounts and potential ploidy variation. These findings highlight the value of absolute quantification by metagenomics and offer a robust framework for quantitative assessments in microbial oceanography.

Keywords: absolute quantification; internal standards; metagenomics; phytoplankton.

Figure 1

Broad agreement between metagenomics-based estimates of cyanobacterial absolute abundances and those made by FCM, as well as pigment patterns. (A) The AMT29 cruise track across the Atlantic Ocean is shown (see sampling stations in Fig. S1). The background color map represents the mean surface chlorophyll a concentration for October–November 2019 [22]. (B) Cell counts of cyanobacteria estimated using spike-in metagenomics and flow cytometry (FCM) methods. (C) Concentrations of cyanobacterial pigments zeaxanthin and divinyl chlorophyll a (DVChl-a) measured in surface samples. (D) Scatterplot between cyanobacterial abundances estimated by recA-metagenomics and FCM. Pearson’s correlations are presented for all samples (slope = 1.10) and separately for the first 33 stations (slope = 1.03) (darker dots only, red correlation statistics). (E) Scatterplot of cyanobacterial haploid genome equivalents based on recA and psbO (slope = 0.72; Pearson’s r = 0.99) gene markers through spike-in metagenomics.

Figure 2

Relative and absolute abundance estimates in AMT29 samples. (A) Relative abundance of major taxa at the domain (left) and order (right) levels derived from universal 3-domain rRNA amplicon sequencing. (B) Absolute abundance of major prokaryotic taxa at the domain (left) and order (right) levels estimated using quantitative metagenomics via recA for bacteria and radA for archaea.

Figure 3

Comparison of abundances of eukaryotic phytoplankton show metagenomics broadly agrees with FCM of nano-and pico-eukaryotes, but metagenomics-based abundances of dinoflagellates and diatoms were much higher than microscopy. (A) Total eukaryotic haploid genome equivalents, estimate of cell counts via equation (4)—see text, calculated using psbO-based metagenomics, alongside nano- and pico-eukaryote counts determined by FCM. (B) Eukaryotic cell counts derived from FCM for coccolithophores and from microscopy for dinoflagellates and diatoms. (C) Cell counts of major eukaryotic groups identified using psbO genes, i.e. the same metagenome results as in panel A, broken down by major groups. Note the numerically dominant groups from metagenomes are expected to be largely in the pico and nano size range, with cell diameters <20 μm (see text). Samples marked with circular dot were resequenced. (D) Total eukaryotic cell count estimates in the 10 resequenced samples. The resequencing effort yielded 26 289 psbO sequences, comprising 25 079 from cyanobacteria and 1210 from eukaryotes. (E) Absolute abundance estimates from (D) divided into major of eukaryotic groups.