Fungal phylogenomics with mitochondrial and nuclear genes: are fission yeasts Archiascomycota?
Yu Liu (1), Naiara Rodriguez-Ezpeleta (2), Henner Brinkmann (2), Herve Philippe (2), B. Franz Lang (2)
(1) Department de Biochemie, Universite de Montreal, Montreal, H3T1J4, Canada; (2) Department de Biochemie, Universite de Montreal, Edouard Montpetit, Montreal, H3T1J4, Canada
Molecular taxonomies based on small subunit rRNA (SSU-rRNA) and RBP2 sequences divide Ascomycota into three major lineages: Hemiascomycota (yeasts), Euascomycota (including various 'filamentous' ascomycetes), and Archiascomycota (fission yeasts and some other deeply diverging ascomycetes). However, solid statistical support for this view is lacking, and is incongruent with results from several multi-protein phylogenies using mitochondrial and nuclear protein sequences. We posit that the controversial interpretations are based on (i) the use of insufficient sequence data, (ii) poor species sampling, (iii) long-branch-attraction artifacts (LBA), or a combination of these factors. By employing methods that minimize the effect of LBA on a large data set (29 fungal species, 98 nuclear-encoded proteins, 18,196 amino acid positions), we obtain unequivocal support (100% bootstrap) for the divergence of Schizosaccharomyces pombe at the base of the Ascomycota, consistent with earlier SSU-rRNA based analyses. For the first time, we obtain the same result in multi-gene phylogenies of mitochondrial proteins (31 species, 12 proteins, 3,030 positions), by using only the more reliable, slowly evolving amino acid positions in this dataset. Together, these results confirm that phylogenetic artifacts cause the placement of S. pombe at the base of Hemiascomycota, and that phylogenomics approaches, broad taxon sampling and the use of sophisticated methodology is required to resolve this question.