Wolbachia pipientis should not be split into multiple species: A response to Ramírez-Puebla et al., “Species in Wolbachia? Proposal for the designation of ‘Candidatus Wolbachia bourtzisii’, ‘Candidatus Wolbachia onchocercicola’, ‘Candidatus Wolbachia blaxteri’, ‘Candidatus Wolbachia brugii’, ‘Candidatus Wolbachia taylori’, ‘Candidatus Wolbachia collembolicola’ and ‘Candidatus Wolbachia multihospitum’ for the different species within Wolbachia supergroups”

In their recent paper, Ramírez-Puebla et al. [26] propose the designation of species for a selection of Wolbachia lineages. While this study considers the largest number of Wolbachia strains in genomic analyses to date, there are several shortcomings and oversights, which render the conclusions problematic. The authors state that Wolbachia taxonomy has been understudied, and they disregard the caution implemented in other phylogenetic and phylogenomic studies to not define species [5,10,13,22]. As an obligate intracellular symbiont, Wolbachia is intimately connected to its host. For this reason, and because Wolbachia cannot be cultured axenically, host phenotypes and phylogenetic relationships of Wolbachia strains are the most well studied aspects of Wolbachia biology rather than Wolbachia themselves. The strain designations used across the Wolbachia literature provide information on the host animal (ex: strain wMel infects Drosophila melanogaster), allowing results to be interpreted in the appropriate biological context. Below we outline three major concerns with the framework proposed by Ramirez-Puebla et al., and argue that applying species status to Wolbachia lineages is premature.

First, the datasets used by the authors are unsound. While the authors have gathered a large amount of the publically available Wolbachia genomes, sequence data known to be unreliable was used in their analyses. For example, the sequences available for the wAna strain (from Drosophila ananassae) represent a combination of host genomic inserts of Wolbachia from lateral gene transfer events and sequence from the Wolbachia infection [18]. Additionally, for genome strains wDacA and wDacB (from Dactylopius coccus), the authors do not provide primary sequence data, instead stating that their analysis will be published at a later time. Defining species status should be held to higher standards and publicly available datasets that the community can evaluate. Ramirez-Puebla et al. present low in silico DNA–DNA hybridization (DDH) and average nucleotide identity (ANI) values comparing wDacA to other A supergroup strains. Because the authors sequenced wDacA co-infecting with wDacB (a B supergroup Wolbachia strain), the data produced may be a combination of both strains or potential lateral gene transfer events, as often happens with Wolbachia [6,10,18,24]. The evidence for phage and chromosomal genetic exchange across supergroups between co-infecting strains further complicates species designations [2,8,23], though we appreciate that some Wolbachia supergroups evolve cohesively to a degree despite frequent ecological overlap in coinfected hosts [10].

Second, and as previous studies recognized [5,13], the stability of any systematic framework is fundamentally dependent on data completeness for each candidate taxonomic grouping. As is evident by several recent studies that have identified novel Wolbachia supergroups, the field is still discovering new Wolbachia diversity [1,4,14,19]. Currently, there are 16 reported supergroups, denoted A–Q [1,4,14,15,21,27]. Supergroup G was de-commissioned as it was based on a recombinant from supergroups A and B [3]. Ramírez-Puebla et al. acknowledge the large number of Wolbachia supergroups, but only examine genomic data from strains representing six of these supergroups (A–F). They suggest that the name “Wolbachia pipientis” applies only to supergroup B, and propose new species names for seven other clades. This system of nomenclature would leave a total of ten supergroups nameless (supergroups H–Q), and it is unclear how these groups would be incorporated into their framework in future. Importantly, for two supergroups (C and D), there is a severely limited quantity of genomic data; two of the proposed species (Candidatus Wolbachia blaxteri and Candidatus Wolbachia taylori) are each based on data from a single Wolbachia strain. This should deter biologists from splitting each supergroup into two species (Candidatus Wolbachia onchocercicola and Candidatus Wolbachia blaxteri for supergroup C, and Candidatus Wolbachia brugii and Candidatus Wolbachia taylori for supergroup D). It is not clear how remaining members of the C and D supergroups not analyzed by Ramírez-Puebla et al. are to be classified. This could be especially problematic for strains that are found to cluster closely with, for example, supergroup D, but are sister to the clade of Candidatus Wolbachia brugii + Candidatus Wolbachia taylori.

Third, the utility of new species designations is not evident to us and presumably other researchers in the Wolbachia community. A combination of both genetic distance and functional divergence should be taken into consideration when devising taxonomic frameworks. For example, there are many closely related Wolbachia strains in supergroup A (referred to as Candidatus Wolbachia bourtzisii in Ramírez-Puebla et al.) that infect Drosophila spp. and have vastly different consequences for their hosts [16]. Also, the proposed nomenclature could easily cause confusion. While a transitional process is expected when adopting new nomenclature, several of the suggested species names provide confusing descriptors for the clades. The name suggested for supergroup E is “Candidatus Wolbachia collembolicola” due to the fact that this clade is found infecting Collembolan hosts. However, there is evidence that this clade is not specific to Collembola, and can infect Dipteran species as well (J. Rasgon, unpublished data). The clade that would make up Candidatus Wolbachia onchocercicola (a subset of the C supergroup) includes Wolbachia strains that do not infect the genus Onchocerca [11]. However, all hosts of both the C and D supergroup strains are members of the Onchocercidae family.

The authors reference Pfarr et al. [25] as a source of support for splitting Wolbachia into species. Given the sequencing of additional genomes, discovery of many new supergroups, and discovery of novel phenotypes since then, the arguments made in Pfarr et al. are no longer as strong. Pfarr et al. argue that the nematode-infecting Wolbachia should be classified differently because they are obligate, infect a different taxon of hosts, and have genomes with different features. However, we now know of obligate Wolbachia strains infecting various arthropods [9,17,20,28,29], the F supergroup that infects both arthropods and nematodes [7], and that the Wolbachia infecting Trichogramma wasps share some similarities with the nematode-infecting strains, such as the absence of phages [12]. It is clear that Wolbachia do not “bin” nicely according to the criteria in Pfarr et al.

It is well understood that the Wolbachia clade is diverse and contains a number of distinct lineages identifiable to the community in the current supergroup classification system. Therefore, for these reasons and those above, we recommend to the community that they continue to use the previously designated strain names and supergroups for clarity in the field. Should there be further evidence and community support for species designations within this unique clade of bacteria, a comprehensive and voluntary Wolbachia consortium should work together to propose nomenclature, as has been previously done in the field (International Wolbachia Conference, Heron Island, Australia, August 2004).