Pruning the Tree: Comparing OTUs and ASVs in High-Throughput Sequencing of 5S-IGS Nuclear Ribosomal DNA in Phylogenetic Studies

Abstract

Amplicon sequencing of the nuclear ribosomal 5S RNA gene arrays is highly promising for genotaxonomy, to delineate genetic resources of species and trace their evolution. However, the huge amount of data retrieved with this approach is difficult to manage and prone to redundancy, error, and computational difficulties. Reducing the amount of data per sample without losing molecular-phylogenetic signal is therefore a crucial step for downstream analyses. In this work, we compared operational taxonomic units (OTUs, 100% identity) and amplicon sequence variants (ASVs) from 5S intergenic spacer (5S-IGS) amplicons of seven beech species (Fagus spp.) obtained with two widely used and competing bioinformatics tools, MOTHUR and DADA2. We assessed quantitative (total number of sequences retrieved, proportion of sequences unique or shared across the different data sets) and qualitative (sample diversity, data ruggedness and congruence of the obtained phylogenies) differences among sample data profiles obtained with the two methods, and the capacity of the inferred phylogenies to capture diagnostic 5S-IGS variant types. Over 70% of processed reads were shared between OTUs and ASVs. Despite a strong reduction (> 80%) of the representative sequences, DADA2-ASVs identified all main 5S-IGS variants known for Fagus, reflecting the phylogenetic, taxonomic and diversity patterns expected for each sample. In contrast, MOTHUR generated large proportions of rare OTUs and ASVs that complicated the obtained phylogenies and were inference-wise redundant. We conclude that differences in the sequence variation detected by the two pipelines are minimal and do not result in different phylogenetic information. The more effective and computationally more efficient DADA2 ASVs may thus replace OTUs in future 5S-IGS studies dealing with complex bioecological phenomena such as hybridisation, polyploidisation, drift and inferring evolutionary pathways of species systems, especially when using increasingly large sample sets.

Keywords: 5S nrDNA; DADA2; Fagus; MOTHUR; high‐throughput sequencing; phylogenetic analysis.

FIGURE 1

Venn diagram depicting the number of representative sequences unique to or shared across the different pipelines of data analysis.

FIGURE 2

Full maximum likelihood trees for sample #04 (F. caspica). Main 5S‐IGS types are annotated, asterisks denote reference tips after Denk et al. (2024). Trees are ordered by increasing tip sample size reduction. ML trees of all other samples are provided as File S3.

FIGURE 3

Collapsed maximum likelihood trees inferred on the DADA2 ‘non‐pooled’ ASV data sets and rooted on the Lineage O variants for the six samples belonging to Fagus subgen. Fagus, with the main types annotated (following Denk et al. 2024). For comparison, pie charts showing the relative abundance of main 5S‐IGS sequence variants as defined and discussed by Denk et al. (2024) are provided. Fagus japonica tree available in File S3.

FIGURE 4

Neighbour‐nets based on intersample ‘phylogenetic Bray–Curtis’ distances, using as ‘associates’: (a, b) MOTHUR‐OTUs/‐ASVs with an abundance of ≥ 4 (a: OTUs; b: ASVs); (c–e) DADA2 ASVs (c: ‘Pooled’; d: ‘Pseudo‐pooled’; e: ‘Nonpooled’ options). Edge bundles of corresponding and expected neighbourhoods coloured accordingly, thick lines indicate neighbourhoods in line with known phylogenetic relationships. Grey line (b, c) refers to the edge bundle reflecting a split into Western Eurasian and East Asian species. All graphs have the same scale, terminal edge leading to sample #06, F. japonica (stippled), has been reduced by a factor of 2.

FIGURE 5

Map of the investigated beech samples with pie charts showing the relative abundance of main 5S‐IGS sequence variants; phylogenetic classification of MOTHUR‐OTUs (ab ≥ 4) and nonpooled ASVs based on the sample‐wise trees and EPA (cf. Cardoni et al. 2022).