In his paper entitled “Is it time to retire the genus Rymovirus from the family Potyviridae?” Colin Ward [12] questions whether rymoviruses should be a separate potyvirid genus. The most important of the facts he uses to support his suggestion is that the pairwise identities (PIs) of rymovirus and potyvirus gene sequences overlap, and hence do not distinguish unequivocally between them. He also notes that these two virus groups have different vectors; rymoviruses are mite-borne and potyviruses aphid-borne, and he argues that this is taxonomically irrelevant. His suggestion, however, raises some important and interesting questions.

Ward bases his case on data from Shukla and Ward [11], which was published a quarter century ago, and so the first question is whether his claim about pairwise identities is still correct when the calculations are based on the more abundant and longer gene sequences now available in Genbank. We have therefore calculated the pairwise identities of the main ORFs of 102 potyvirus and six rymovirus genome sequences using the SDT program [9]. The sequences were downloaded from Genbank, edited and assembled using BioEdit [6], aligned using the TranslatorX server [1] with its MAFFT option; they became 16206 nts long including alignment indels, and contained no recombinant sequences as tested by RDP4 [8]. Fig. 1 is a ‘histograph’ of the PIs; the mean PI of the ORFs of the most different rymoviruses (i.e. all pairwise comparisons of three ryegrass mosaic virus sequences with three hordeum mosaic virus sequences) is 56.36% +/- 0.25%, which is little different from the mean PI of 55.09% +/- 1.04% for the ORFs of the most different clade of potyviruses (Narcissus degeneration, Onion yellow dwarf, Shallot yellow stripe and Vallota speciosa viruses) versus all the others. Thus the rymoviruses cannot be distinguished from the potyviruses by their PIs, and the basis of Ward’s claim is still absolutely correct.

Fig. 1
figure 1

Pairwise identity ‘histograph’ of 102 potyvirus and six rymovirus main ORF sequences showing only the comparisons connected through the basal phylogenetic nodes. All pairwise comparisons are in green, potyviruses versus potyvirus comparisons in blue, rymovirus versus rymovirus in mauve and rymovirus versus potyvirus in red. The Y-axis is of 100 comparisons/division for all except rymovirus versus rymovirus comparisons, which are of four comparisons/division. The Accession Codes of the sequences used are given in Supplementary File 1

Full size image

The use of PI estimates as a surrogate for distinguishing the natural groups, or more specifically phylogenetic groups, of potyviruses by Shulka and Ward [11] was expanded by Adams et al. [2], but subsequently questioned by Duffy and Seah [3], who highlighted the problem that “When researchers use the standard technique of per cent nucleotide identity to determine that the new sequence is closely related to another sequence, potentially erroneous conclusions can be drawn from the results.”

PI estimates are only tangentially linked to phylogenetic relatedness, and to determine whether rymoviruses and potyviruses are distinct groups worth designating as genera requires phylogenetic analyses. Phylogenetic algorithms assume that related organisms have descended from shared ancestors by a process of independent genetic divergence and selection, whereas PI measures do not, and are analogous to the measures used by cosmologists when defining the motions of stars, which although derived from a single ‘big bang’, continue to influence one another at a distance by gravitation and ‘dark matter’ (https://en.wikipedia.org/wiki/Dark_matter). Therefore we checked whether two standard, but different, phylogenetic methods distinguish between rymoviruses and potyviruses using the same 108 ORF sequences checked in the SDT analysis above, together with 16 tritimovirus sequences as an outgroup. One phylogeny was calculated by PhyML 3.0 [5] using the GTR+I+Ƭ4 substitution model, and statistical support for individual nodes was assessed using the SH option [10]. Fig. 2 shows the branching pattern of the resulting tree; it was drawn using Figtree 1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/) and a commercial computer illustration package. It can be seen that the sequences fall into three monophyletic clusters, the largest contains only potyvirus sequences, the smallest only rymovirus sequences and the third only tritimovirus sequences, with the basal nodes of all three clusters having 1.0 SH statistical support. A tree calculated from the same sequences by the neighbor-joining method in ClustalX [7] was topologically the same as the ML tree with the sequences again forming monophyletic clusters with 100% bootstrap support (1000 replicates). Thus phylogenetic methods distinguish the three potyvirid genera, whereas PI analyses do not as they have problems with closely related clusters, just as they have with outliers [3].

Fig. 2
figure 2

Branching pattern of a ML phylogeny of 102 potyvirus, six rymovirus and 16 tritimovirus ORFs. Arrows indicate the basal nodes of the three lineages. The truncated branch to the tritimoviruses had a length of c. 5.2 substitutions/site. The Accession Codes of the sequences used are given in Supplementary File 1

Full size image

Recently the Code of the International Committee on Taxonomy of Viruses (ICTV; https://talk.ictvonline.org/information/w/ictv-information/383/ictv-code) updated its Rule 3.20, to define species as “a monophyletic group of viruses whose properties can be distinguished from those of other species by multiple criteria.”, but did not similarly update Rule 3.23 for genera, which merely states that “A genus is a group of species sharing certain common characters.”. Thus the ICTV has now adopted phylogenetics as the basis for defining species, but not yet genera despite the widespread reported use of phylogenetic methods at both taxonomic levels. The ICTV also leaves the sorts of “multiple criteria” or “certain common characters” to be used for distinguishing species and genera to the imagination! The ICTV should clarify the situation by declaring that both species and genera to be monophyletic groupings. It could also aid the choice of characters for defining such groupings by explicitly stating that “virus species include strains/isolates, that are so similar there is no value in giving them separate names”, whereas virus genera are groups of viruses that “it is especially useful to define by their shared properties because the groupings thus formed help with such problems as identifying newly found viruses and predicting their properties” [4].

In summary, the members of the Potyvirus and Rymovirus genera are phylogenetically distinct, they also have different vectors, and they should continue to be recognised as members of distinct genera within the Potyviridae. Furthermore we suggest that, for publication, the monophyly of species and also genera should be established using phylogenetic methods, not sequence identities, and we note that several such methods (e.g. [5] and [7]) require no more computational skill than calculating sequence identities [9].