Hidden diversity within Beauveria and Metarhizium – Comparing morphology, barcoding, multilocus phylogenies and whole-genome sequences

Accurately identifying entomopathogenic fungi is crucially important, but the current approach of analysing four genes might not provide sufficient resolution. In this study, we investigated the different resolution provided by multilocus phylogenies and approaches based on whole genome sequence data. Fungi were isolated from soil samples that were collected from five different vegetation types (dry sclerophyll forest, agricultural grassland, rainforest, suburban parkland, and sugar cane fields) and across four different suburban soil habitat types in southeast Queensland. Three different agricultural pests were used as live baits, cotton stainer bug, diamondback moth, and rust-red flour beetle. Whole-genome sequencing was conducted for all 83 isolates recovered, and the ITS2 region was extracted from the genome assemblies to make initial species identifications with the UNITE database. We also extracted tef1a 3′, tef1a 5', rpb1 and rpb2 genes from the Metarhizium genomes and the bloc, tef1a and rpb1 genes from Beauveria genomes to construct multilocus phylogenies and obtain species identification. To investigate the genetic relationships across 14 isolates of Beauveria bassiana and (independently) across 43 isolates of Metarhizium based on whole-genome data, we genotyped single nucleotide polymorphism (SNP) markers and conducted principal components analysis on the whole-genome SNP data. The multilocus methods identified isolates to species more precisely than ITS2, except in the one unresolved clade in the Metarhizium phylogeny. The whole-genome approach identified more genetic clusters than the multilocus phylogenies identified species among the isolates, and the morphological results correlated with some of the genetic clusters, so they likely represent distinct species not detected by the other methods. The genetic clusters were not associated with vegetation type or bait insect species. This is the first comparison of the resolution of multilocus phylogenetics with that of whole-genome SNP data for these genera. We suggest how the genetic clusters identified here may be investigated further to determine whether they represent unrecognised species within these groups.