‘Misidentifications’ in fungal DNA sequence databanks

Abstract

Bridge et al. (2003) draw attention to the problem of DNA sequences published in GenBank being attributed to incorrectly named organisms. The problem of wrongly attributed data is well-known in biology, and the need to deposit voucher material in public collections has been repeatedly advocated (Hawksworth, 1984; Agerer et al., 2000; Ward et al., 2001; Tehler et al., 2003). With the largest tree of fungal sequences so far published, comprising 1551 rRNA SSU sequences from across the whole kingdom Fungi, comes the salutary warning that ‘we have found numerous obvious misidentifications, misspellings and outdated names in the data set’ (Tehler et al., 2003). The point could have been well-made by compiling references to published reports where voucher cultures and specimens had been re-examined and found to be wrongly named. However, Bridge et al. (2003) took a different approach by comparing deposited sequences and seeking to find evidence of errors because of sequence mismatches. They did not examine voucher material, nor did they cite the published papers in which the sequences they picked out were discussed. I have undertaken studies on taxa referred to Phoma (Hawksworth, 1981; Hawksworth & Cole, 2004), and it is quickly apparent that the paper has been written with little knowledge of this genus. Some species are known to be anamorphs of Leptosphaeria, so is it likely to be wrong that P. herbarum (actually the type species of the genus) comes out close to that? However, the comment that this Phoma has its ‘closest matches to mixed Leptosphaeria/Paraphaeosphaeria species’ is also worrying in that Leptosphaeria and Paraphaeosphaeria are quite distinct and placed in different families (Eriksson et al., 2003). Further, I disagree that anything that comes out in Dothideomycetes is automatically viewed as wrong, when the teleomorph of many Phoma species is Didymella, a genus of Dothideomycetes (as is Leptosphaeria). Similarly, the hyphomycete genus Epicoccum is well-recognized as a synanamorph (i.e. an additional asexual stage) of Phoma. These points are clear in the seminal overview of the genus by Boerema (1997). It is common practice in mycology for authors to use phrases such as ‘Phoma-like’ for anamorphs where anamorphic fungi share some characters of that genus, particularly pycnidial thin-walled conidiomata with short to rounded conidiogenous cells, and simple hyaline conidia, but are recognized as probably not being congeneric with P. herbarum. The use of such a phrase is not a ‘misidentification’, but simply reflects the current state of knowledge and a responsible reluctance to introduce additional scientific names without appropriately detailed studies. I would not want to comment in detail on the other examples featured, but in the case of Hypoxylon numerous species have extensive lists of synonyms and it is not evident whether the names used for the deposited strains were checked against the ones adopted in the monograph of Ju & Rogers (1996). Further, I am aware that the situation in Amanita is complex, with many genotypes occurring within single species (Sawyer et al., 2003), and other names being found to have been applied to cryptic but evidently distinct taxa (Oda et al., 2002). I am also concerned that seeming mismatches of helotialen cultures are dismissed as ‘obtained from cultures and so could again be the result of faster-growing contaminants’ where no attempt has been made to check voucher cultures or examine pertinent publications. In molecular phylogenetic work populations traditionally treated under the same scientific names based on morphological criteria prove to belong to distinct molecular phylogenetic clades (Kroken & Taylor, 2001; Dunham et al., 2003). In many cases, such clades prove to be separable on hitherto largely overlooked morphological features. Earlier identifications were not simply ‘wrong’, but reflected the current knowledge. It would be of interest to know the level of real misidentifications in deposited sequences, and it may be that ‘up to 20%’ would prove to be unreliable as Bridge et al. (2003) claim. However, a critical assessment of the situation requires careful research and the re-examination of voucher material in the context of experience of the groups of fungi concerned. Unfortunately, we do not have this presented, but instead claims based on superficial comparisons of sequences which impugn the mycological community as making mistakes at a level that the authors have not unequivocally established. Despite the foregoing comments, the paper does point out a key problem area in molecular studies. There is a need for a heightened awareness among molecular biologists of the need to work closely and continuously with experienced systematists (Scotland et al., 2003), and for authors to verify identifications by specialists and to deposit voucher specimens or permanently preserved cultures in public reference collections.