Consistent Species Richness Patterns but Not Richness Estimates Based on Both ASV and OTU Inference Methods on ITS2-Based Soil Fungal Communities

Abstract

Fungi play essential roles in key ecosystem functions and processes, yet they often occur in inconspicuous, species-rich, and complex communities that remain difficult to study. Studies of fungal communities based on DNA extracted from environmental samples commonly rely on clustering sequence reads into units of diversity, followed by taxonomic identification and, in some cases, linkage to ecological traits. In this study, we evaluated how two clustering approaches—amplicon sequence variants (ASVs) and operational taxonomic units (OTUs)—affect the characterization of fungal communities. Despite minor differences, both approaches recovered consistent taxonomic patterns and community structure. Although both methods produced a similar total number of sequence clusters, they differed in representation of fungal community composition. All ASV representative sequences matched OTU representative sequences with at least 92.2% similarity, whereas several rare OTUs showed low similarity to ASV reads, suggesting differences in the detection of low-abundance taxa. However, only a small fraction of OTU reads (< 0.1%) lacked a corresponding ASV, indicating that ASVs captured nearly all OTU-defined taxa. In contrast, 14% of ASV reads assigned to species hypotheses (SHs) did not match any OTU reads assigned to SHs, whereas only 1.3% of OTU SH-assigned reads lacked a corresponding ASV match. ASVs generally provided higher resolution than OTUs, as abundant SHs were often represented by multiple ASVs, suggesting that ASVs capture intraspecific diversity. Consequently, ASVs should not be used as direct species proxies but instead require post hoc grouping to reflect species-level diversity. OTUs-based community composition aligned more clearly with soil properties, particularly the N:C ratio. Overall, both approaches provided a similar overview of broad-scale species richness. The choice between two clustering methods depends on the research question and the desired level of taxonomic resolution, and our results provide little support for the claim that ASVs should categorically replace taxonomic units in marker-gene data analysis.