Identifying the ‘unidentified’ fungi: a global-scale long-read third-generation sequencing approach

Molecular identification methods, in particular high-throughput sequencing tools, have greatly improved our knowledge about fungal diversity and biogeography, but many of the recovered taxa from natural environments cannot be identified to species or even higher taxonomic levels. This study addresses the phylogenetic placement of previously unrecognized fungal groups by using two complementary approaches: (i) third-generation amplicon sequencing analysis of DNA from global soil samples, screening out ITS reads of < 90% similarity to other available Sanger sequences, and (ii) analysis of common fungal taxa that were previously indicated to be enigmatic in terms of taxonomic placement based on the ITS sequences alone (so-called top50 sequences). For the global soil samples, we chose to amplify the full rRNA gene operon using four partly overlapping amplicons and multiple newly developed primers or primer combinations that cover nearly all fungi and a vast majority of non-fungal eukaryotes. We extracted the rRNA 18S (SSU) and 28S (LSU) genes and performed phylogenetic analyses against carefully selected reference material. Both SSU and LSU analyses placed most soil sequences and top50 sequences to known orders and classes, but tens of monophyletic groups and single sequences remained outside described taxa. Furthermore, the LSU analyses recovered a few small groups of sequences that may potentially represent novel phyla. We conclude that rRNA genes-based phylogenetic analyses are efficient tools for determining phylogenetic relationships of fungal taxa that cannot be placed to any order or class using ITS sequences alone. However, in many instances, longer rRNA gene sequences and availability of both SSU and LSU reads are needed to improve taxonomic resolution. By leveraging third-generation sequencing from global soil samples, we successfully provided phylogenetic placement for many previously unidentified sequences and broadened our view on the fungal tree of life, with 10–20% new order-level taxa. In addition, the PacBio sequence data greatly extends fungal class-level information in reference databases.