The researchers analyzed 16S rRNA gene amplicon sequences, shotgun metagenomes and metatranscriptomes generated from the same lakes. Sequencing depth was 3.31 Gbps for amplicons, 170.91 Gbps for metagenomes and 34.11 Gbps for metatranscriptomes. They tested three ways of making the datasets comparable: mapping shotgun reads to amplicon-derived operational taxonomic units, or OTUs; profiling each dataset with a method-specific taxonomic marker; and attempting to recover 16S rRNA genes from shotgun assemblies for a cross-method OTU library.

The assembly-based strategy was discontinued because recovered sequences had highly variable lengths and included chimeras longer than 1,600 base pairs. These sequences could not be compared meaningfully with the V9 region represented by the amplicon OTUs. The analysis therefore focused on the two remaining approaches, which converged on the finding that sequencing technology introduces systematic differences in the resulting community profiles.

In the OTU-based comparison, 2,535 representative amplicon OTUs passed a relative abundance threshold of 0.001% per sample. They recruited 9,356,192 amplicon reads, 154,996 metagenomic reads and 19,979,498 metatranscriptomic reads, with every OTU recruiting reads across all samples. Although the technologies represented the same 26 prokaryotic phyla, alpha-diversity estimates differed significantly. Amplicon data produced lower richness and Shannon diversity than both shotgun datasets, while metagenomes and metatranscriptomes did not differ significantly in those measures. Friedman tests gave p-values below 0.001, with Kendall’s W values of 0.76 for richness and 0.84 for Shannon diversity.

Marker-specific profiling produced a more pronounced ordering. The researchers used the single-copy rpS3 gene for metagenomes, rRNA reads for metatranscriptomes and amplicon OTUs as an independent reference. This analysis represented 23 prokaryotic phyla in metagenomes and 51 in metatranscriptomes. For every lake, estimated diversity ranked metatranscriptomes above metagenomes and metagenomes above amplicons. The consistent ranking across lakes indicates that these differences were systematic methodological effects rather than lake-specific patterns.

Community-level comparisons also depended on the analytical strategy. PERMANOVA attributed approximately 16% of variation in relative-abundance data and 13% of variation in presence-absence data to sequencing approach, with method effects reaching about 41% in the marker-based analysis. Metatranscriptomic profiles were the most homogeneous, while amplicon profiles were the most heterogeneous. Procrustes and Mantel tests showed greater concordance between metagenomes and metatranscriptomes than between either shotgun dataset and amplicons. Among the OTU-based comparisons, metagenomes and metatranscriptomes had the strongest congruence, with R² = 0.68 and P = 0.001.

Differential-abundance analysis identified method-specific taxonomic patterns. Relative to metagenomes, amplicon data contained more OTUs assigned to Alphaproteobacteria and Cyanobacteria and fewer assigned to Gammaproteobacteria, mainly Burkholderiales. Polynucleobacter and Limnohabitans were strongly underrepresented in the amplicon data, matching a pattern previously attributed to primer bias in the same lakes. Metagenomes had higher relative abundances of Cytophagales and Chitinophagales within Bacteroidetes, whereas amplicons overrepresented Flavobacteriales. Corrections based on 16S rRNA gene copy number and a weighted primer score did not materially alter community composition, with Spearman correlations of 0.9925 and 0.9968, respectively.

Metatranscriptomes and metagenomes differed in only six OTUs by differential abundance. However, the DNA dataset used as a reference affected conclusions about transcription. Relative to metagenomes, 22 taxonomic classes had similar proportions in the metatranscriptomic and metagenomic data, and Clostridia, including Clostridium species, appeared transcriptionally active. When amplicon data served as the baseline, a smaller proportion of that population appeared to be transcribing 16S rRNA genes. The comparison therefore depended on the reference method, not only on the metatranscriptomic measurements themselves.

Four analytical approaches—Mantel tests, Procrustes analyses, BioEnv and distance-based redundancy analysis with stepwise selection—were used to assess how well each dataset reflected environmental variation. Metatranscriptomes showed the strongest global community–environment associations in the Mantel and Procrustes analyses, while metagenomes produced the most stable explanatory OTUs in the BioEnv and redundancy analyses. The full metagenomic redundancy model explained about 20% of environmental variation, but forward selection retained one Cyanobacteria-affiliated OTU that explained 9.4% of the variation, with an adjusted R² of 0.094, F = 3.08 and p = 0.021. Neither amplicon nor metatranscriptomic datasets produced significant models after stepwise selection.

The findings identify trade-offs rather than a universally preferred method. Amplicon sequencing is comparatively rapid and inexpensive but is affected by primer and amplification biases. Metatranscriptomics captures transcriptional activity but is sensitive to RNA degradation and transcriptional dynamics. Metagenomics offers broad genomic context and stable explanatory features, but requires greater sequencing and computational resources and can have reduced sensitivity to rare organisms. The observational comparison was limited to 21 European lakes and does not establish which approach is best under all conditions; the authors noted that controlled experiments such as batch cultures would be required for that determination.

Subject of Research: Comparison of amplicon, metagenomic and metatranscriptomic methods for studying lake microbiomes and their relationships with environmental conditions

Article Title: Constraining lake ecology via metagenomics, metatranscriptomics, and amplicon sequencing

Article References: Nuy, J. K., Bornemann, T. L. V., Beisser, D., Probst, A. J., & Boenigk, J. (2025). Constraining lake ecology via metagenomics, metatranscriptomics, and amplicon sequencing. Metabarcoding and Metagenomics, 9, Article e156215. https://doi.org/10.3897/mbmg.9.156215

Image Credits: AI Generated

DOI: 10.3897/mbmg.9.156215

Keywords: lake microbiomes, amplicon sequencing, metagenomics, metatranscriptomics, 16S rRNA, microbial diversity, environmental variation, European lakes, Constraining, lake, ecology, amplicon