Experimental and Applied Acarology 32: 301312, 2004. # 2004 Kluwer Academic Publishers. Printed in the Netherlands.

On molecular taxonomy: what is in a name?

GERRIT UILENBERG1,2,4,*, FRANCOIS THIAUCOURT1 and FRANS JONGEJAN2,3
1 CIRAD-EMVT, Campus International de Baillarguet, 34398 Montpellier Cedex 5, France; 2Department of Parasitology and Tropical Veterinary Medicine, Faculty of Veterinary Medicine, Utrecht University, P.O. Box 80.165, 3508 TD Utrecht, The Netherlands; 3Department of Veterinary Tropical Diseases, Faculty of Veterinary Science, University of Pretoria, Private Bag X04, Onderstepoort 0110, South Africa; 4 ` Current address: A Surgente, route du Port, 20130 Cargese, France; *Author for correspondence (e-mail: uilenber@club-internet.fr)

Accepted 25 November 2003

Key words: Anaplasma, Boophilus, Ehrlichia, Eperythrozoon, Haemobartonella, Ixodoidea, Molecular taxonomy, Mycoplasma, Mycoplasmatales, Rickettsiales, Rhipicephalus Abstract. Gene sequences of small portions of the genome are often used for premature detailed taxonomic changes, neglecting polyphasic taxonomy, which should also consider phenotypical characteristics. Three examples are given: (i) Recently, members of the genera Eperythrozoon and Haemobartonella have been moved, correctly so, from the Rickettsiales to the Mycoplasmatales, but were assigned to the genus Mycoplasma, mostly on the basis of 16S rRNA sequence analysis. Not only is the 16S rRNA sequence similarity between classical Mycoplasma and these species of Eperythrozoon and Haemobartonella less than that between some other well-recognised bacterial genera, but their biological differences amply justify their classication in different genera of the Mycoplasmatales. Furthermore, the move creates considerable confusion, as it necessitates new names for some species, with more confusion likely to come when the 16S rRNA sequences of the type species of Eperythrozoon, a name which has priority over Mycoplasma, will be analysed. (ii) In the Rickettsiales, members of the genera Anaplasma, Ehrlichia, Cowdria, Neorickettsia and Wolbachia are so closely related phylogenetically on the basis of 16S rRNA sequences, and for some also of groESL operon sequences, that they have recently been fused, correctly so, into one family, the Anaplasmataceae, while the tribes Ehrlichieae and Wolbachieae have been abolished. Sequence diversity within the classical genus Ehrlichia has led to classifying E. phagocytophila (including E. equi and the agent of human granulocytic ehrlichiosis), E. platys and E. bovis in the genus Anaplasma, while others have been retained in Ehrlichia, which also includes Cowdria ruminantium. E. sennetsu and E. risticii have been transferred to the genus Neorickettsia. 16S rRNA and GroEL sequences of classical Anaplasma and some members of classical Ehrlichia do show a close relationship, but differences in citrate synthase gene sequences, the GC content of this gene, and sequences of the gene encoding the b-subunit of RNA polymerase, not to speak of the phenotypical differences, do not justify the fusion into one genus. Because of the phylogenetical diversity in Ehrlichia it is recommended that a new genus name be created for the E. phagocytophila genogroup (and E. platys and E. bovis). (iii) One of the conclusions of studies on the phylogeny of ticks of the subfamilies Rhipicephalinae and Hyalomminae, based on nucleotide sequences from 12S rRNA, cytochrome c oxidase I, the internal transcribed spacer 2, 18S rRNA, as well as morphological characters, is that Boophilus should be considered as a subgenus of Rhipicephalus. While Boophilus and Rhipicephalus are undoubtedly close, the obviously important morphological and biological differences between the genera Rhipicephalus and Boophilus are thus overruled by similarities in the sequences of a number of genes and this leads to considerable confusion. Polyphasic taxonomy amply justies maintaining Boophilus as a separate genus, phylogenetically near to Rhipicephalus. This note is a plea for a cautious and balanced approach to taxonomy, taking into account molecular genotypical information, as far as is possible from different genes, as well as phenotypical characteristics.

302 Introduction Many of the taxonomic denitions concerning eukaryotes are not applicable to prokaryotes. The comparison of small well-chosen gene sequences has therefore become a particularly powerful tool for establishing evolutionary relationships (phylogeny) between prokaryotes. The criteria for choosing adequate genes have been dened by Woese (1987). They must be universally distributed, code for products of similar function, not be subjected to lateral transfer and undergo mutations in a clockwise manner. The 16S rRNA gene sequences are universally used to construct new classications based on phylogenetic results. This tool has proved particularly useful also for the classication of unculturable organisms, as 16S rRNA genes can be amplied by PCR using universal primers. However, as was already pointed out more than a decade ago (Fox et al. 1992), 16S rRNA sequence identity does not necessarily mean that DNADNA hybridisation will conrm (species) identity and Stackebrandt and Goebel (1994) also conclude that studies on the homology of sequences of the conservative 16S rRNA gene, although relatively easy to perform, have their limitations and do not replace DNA reassociation. Vandamme et al. (2000) also found that comparison of these sequences may prove highly misleading and moreover that different isolates of one bacterial species do not always cluster together in a phylogenetic tree based on 16S rRNA gene sequences. Gillis et al. (2001), discussing polyphasic taxonomy, state that at present no clear-cut genus denition is available. According to these authors, if there is less than 97% 16S rRNA gene sequence similarity it is often not straightforward to decide whether the organisms belong to the same genus; they also state that high levels of sequence variation have been observed even between strains of the same species. Polyphasic taxonomy is based not only on genotypical information, but also on phenotypical characteristics. Of course, phenotype is ultimately determined by genes, but the genome contains innitely more sequences than those usually compared, and the great majority of the genes are often not taken into account. Limited sequence comparisons are sometimes (mis)used for rushing into detailed changes in taxonomy, which are, at the least, questionable. In this note we illustrate our case with two recent examples in bacteriology and one in acarology.

Eperythrozoon=Haemobartonella 5 Mycoplasma? As for other bacteria, phylogeny based on 16S rRNA gene sequences is used to classify mycoplasmas (Maniloff 1992). Neimark et al. (2001, 2002a), on the basis of the 16S rRNA sequences, proposed to transfer four members of the genera Eperythrozoon and Haemobartonella to the genus Mycoplasma. In doing so they also created new species names for Haemobartonella felis, H. muris and E. suis, as Mycoplasma species with these specic names already existed. (This is true for M. felis and M. muris, but M. suis did not exist previously, and the name Mycoplasma haemosuis is therefore not justied. If this organism is transferred to the

303 genus Mycoplasma, it should be called M. suis [Anonymous 2001].) Foley and Pedersen (2001), in a publication preceding that of Neimark et al. (2001) in the same issue of the journal, assigned a small type haemobartonella-like organism to the genus Mycoplasma, calling it Candidatus Mycoplasma haemominutum. Messick et al. (2002), also on the basis of the 16S rRNA genes, assigned microorganisms on the red cells of the opossum, the alpaca and the dog to the genus Mycoplasma. (The latter had been given previously the name Haemobartonella canis; as Mycoplasma canis exists, the authors gave it a new name, Mycoplasma haemocanis, without explaining why they call this new name only a new combination. The micro-organisms of the opossum and the alpaca were designed as Candidatus new species.) The highest scoring sequences obtained were 16S rRNA genes of the Eperythrozoon and Haemobartonella spp. assigned to the genus Mycoplasma by Neimark et al. (2001) and of species belonging to the Mycoplasma pneumoniae group. Neimark et al. (2002b) also assigned a putative Haemobartonella of the South American squirrel monkey to the genus Mycoplasma, again solely on the basis of 16S rRNA gene sequences, and named it Candidatus Mycoplasma kahanei sp. nov. It has been shown convincingly that species of Eperythrozoon and Haemobartonella do not belong in the order of the Rickettsiales, but that their closest relatives are the mycoplasmas, and particularly those of the Mycoplasma pneumoniae group (Neimark and Kocan 1997; Rikihisha et al. 1997; Johansson et al. 1999). The inclusion in the order of the Mycoplasmatales (class Mollicutes), and even in the family Mycoplasmataceae, is therefore logical. We agree with considering Haemobartonella as a synonym of Eperythrozoon (Neimark et al. 2001) as indeed the distinction between them is rather tenuous. But is there really any justication for including Eperythrozoon (and Haemobartonella) spp. in the genus Mycoplasma? Citing earlier publications mentioned above, Neimark et al. (2001) wrote: . . . these studies demonstrate clearly that these bacteria are not phylogenetically related to rickettsiae but instead that their closest relatives are species in the genus Mycoplasma. These data also afrm that each of these haemotropic bacteria is a valid species. Thus, these representatives of the genera Haemobartonella and Eperythrozoon are in fact members of a single genus, Mycoplasma . . .. We argue that the last sentence, starting famously with Thus, is not justied. The authors base their proposal on a comparison of a small portion of the genome, the 16S rRNA gene. While the results are convincing for excluding these micro-organisms from the Rickettsiales and classifying them in the Mollicutes, the sequence similarity between Eperythrozoon wenyonii and one of its closest relatives in the mycoplasmas (M. fastidiosum) is only 77.3%. Rikihisha et al. (1997) found a 7983% similarity of the 16S rRNA gene sequences of Haemobartonella muris, H. felis and Eperythrozoon suis with Mycoplasma spp. Neimark and Kocan (1997) reported a similarity of E. wenyonii to mycoplasmas of the M. pneumoniae group of approximately 76%. Most of these similarities are below those between 16S rRNA sequences of such well-recognised different bacterial genera as Staphylococcus and Streptococcus, Clostridium and Bacillus or Corynebacterium and Mycobacterium (see Table 1). Furthermore, the considerable differences in biological characteristics

304
Table 1. Percentage of identities observed when comparing 16S RNA sequences of various microbial genera. The lowest similarity is found between Mycoplasma wenyonii and M. fastidiosum. By comparison, well established different genera have higher similarities. GenBank Acc. No. AY167864 AB104850 BS16086 AY169422 AY168882 M20940 AF016546 AF125878 Species name Staphylococcus epidermitis Streptococcus dysgalactiae Bacillus silvestris Clostridium clostridioforme Corynebacterium bovis Mycobacterium bovis Mycoplasma wenyonii Mycoplasma fastidiosum % identity 84.4 80 90 77.3

(such as habitat, in vitro culture, and probably transmission) between the classical Mycoplasma species and those in Eperythrozoon (and Haemobartonella), amply justify their classication in different genera, within the class of the Mollicutes. And it should not be forgotten that biological differences are certainly based on gene differences! Polyphasic taxonomy is possible even in micro-organisms which cannot (yet) be cultured. The cell type in which the organism lives and multiplies certainly is a phenotypical characteristic, even if comparing some other sequences than those of the 16S rRNA gene conrms the close relationship of (ex)-Haemobartonella spp. with mycoplasms of the M. pneumoniae group (Harasawa et al. 2002; Tasker et al. 2003b). Strictly speaking, the law of priority does not (yet) apply, because the genus name Eperythrozoon remains valid as long as the 16S rRNA sequences of the type species of the genus Eperythrozoon, E. coccoides, have not yet been compared to those of Mycoplasma and of the Eperythrozoon species transferred to that genus by Neimark et al. (2001). However, it would be highly surprising if those sequences of E. coccoides would justify a different taxonomic position than that of E. wenyonii and E. suis, and Neimark et al. (2001) also expected that the remaining species of Haemobartonella and Eperythrozoon will be found to be mycoplasmas. If and when that occurs, all Mycoplasma species should then be called Eperythrozoon, created in 1928 by Schilling, while the name Mycoplasma was proposed a year later (Nowak 1929). Confusion would certainly arise if Mycoplasma and Eperythrozoon were to be lumped into one genus. If the priority of the name Eperythrozoon would be preserved, the names of the family and the order would also have to be changed accordingly. But the International Code of Nomenclature of Bacteria allows waiving priority if this would create confusion. Of course, it all depends for whom the confusion would exist! Scientists familiar with blood parasites will be very confused when Eperythrozoon becomes Mycoplasma, on the other hand those having worked with the classical mycoplasms would certainly not recognise the name Eperythrozoon mycoides. In any case, the creation of the new names proposed for H. felis, H. muris and H. canis will create considerable confusion, which could

305 simply be avoided by maintaining Eperythrozoon (including Haemobartonella) as a separate genus in the family of the Mycoplasmataceae. If necessary, any possible confusion could be far more easily and logically avoided by waiving the changes in the names of the family and the order. In conclusion, we agree with the synonymy of Haemobartonella with Eperythrozoon and the transfer of the enlarged genus Eperythrozoon to the family of the Mycoplasmataceae, order of the Mycoplasmatales, class Mollicutes, but denitely not with the synonymy of Eperythrozoon and Mycoplasma. This means we believe that the changes of the species names of H. felis, H. muris and H. canis, which have been uncritically ratied (Garrity et al. 2002), are uncalled for. The situation in cats is confused. While we are unable to clarify the situation and will not even attempt to do so, it should be realised that there may well be at least two separate species. In 1942 Clark described Eperythrozoon felis in a cat in South Africa. The microorganisms he described were practically all ring forms, and stained a delicate pale violet with Giemsa; comma and rod forms were infrequent. Its presence was associated with anaemia, and the animal had been put down in a weak, emaciated and anaemic condition. In 1956 Flint and McKelvie described Haemobartonella felis, apparently without realising that a similar organism had been given the name of E. felis; H. felis was also responsible for disease. There are important differences in the descriptions by Clark in South Africa, and Flint and McKelvie in the USA; the latter do not mention ring forms, so that there may well be at least two different species in cats. Rikihisha et al. (1997) found only an 85% similarity in the 16S rRNA gene sequences of isolates of H. felis from cats in California on the one hand and isolates from Ohio and Florida on the other hand. Foley and Pedersen described in 2001 a species in cats, which they called Candidatus Mycoplasma haemominutum, and on the basis of 16S rDNA considered to be distinct from the classical Haemobartonella felis, called Mycoplasma haemofelis by Neimark et al. (2001, 2002a,b). Foleys team had previously already distinguished their isolate as smaller and less pathogenic than classical H. felis, and its 16S rRNA sequence was that of the Californian isolate of Rikihisha et al. Similarly, Jensen et al. (2001) and Westfall et al. (2001) distinguished two types of Haemobartonella felis, the California form and the larger, more pathogenic Ohio form. (The small variety might be more pathogenic in coinfection with feline leukaemia virus (George et al. 2002).) Both types have also been reported outside the USA (Tasker et al. 2003a, b). Ring forms are not mentioned for any of these isolates, so that Eperythrozoon felis, Mycoplasma haemofelis and Candidatus Mycoplasma haemominutum may in fact well be three different species. Euzeby (2003) gives the history of the taxonomic changes of Eperythrozoon and Haemobartonella spp. in detail.

Anaplasmataceae Another recent example of detailed changes in taxonomy based mainly on comparing small portions of the genome of bacteria is the publication by Dumler et al. (2001).

306 The authors have rearranged much of the classication of the order Rickettsiales, by convincingly demonstrating, on the basis of 16S rRNA gene sequences, and for some organisms also of groESL operon sequences, that the genera Anaplasma, Ehrlichia, Cowdria, Neorickettsia and Wolbachia are so closely related phylogenetically that they may be fused into one and the same family, the Anaplasmataceae (which would provisionally include also Aegyptianella, on phenotypical grounds). The tribes Ehrlichieae and Wolbachieae are abolished. They have also shown that certain gene sequences of some of the Ehrlichia spp. are more similar to the equivalent sequences of Anaplasma than of some other Ehrlichia spp. We certainly agree with these ndings, but think that the authors should have shown more reserve following that considerable achievement, as did Inokuma et al. (2001b). The authors identify four genogroups: The E. phagocytophila=Anaplasma group, in which E. phagocytophila, E. equi and the agent of human granulocytic ehrlichiosis (HGE) are so closely related, including at least 99.1% 16S rRNA sequence similarity and identical GroEL amino acid sequences, that they are considered to be one species, probably rightly so. This conrms several earlier ndings. Furthermore, the members of this group are all classied in the genus Anaplasma, on the basis of the similarity of their 16S rRNA sequences; they found a minimum of 96.1% similarity (which is less than the minimum considered by Gillis et al. (2001) as the requirement for a straightforward conclusion). This genus would include Anaplasma marginale (with A. centrale and A. ovis perhaps as mere variants), Anaplasma sp. (Japan) (which is phenotypically like A. centrale, but differs in 16S rRNA sequences), A. (ex Ehrlichia) phagocytophilum (including the agents of equine ehrlichiosis and of HGE), A. (ex Ehrlichia) platys, and A. (ex Ehrlichia) bovis. The E. canis=Cowdria group. This second genetic cluster is classied as the genus Ehrlichia and includes Ehrlichia canis, E. chaffeensis, E. ewingii, E. muris and (ex) Cowdria ruminantium. They have at least 97.7% similarity in 16S rRNA gene sequences. These species remain in the genus Ehrlichia. The E. sennetsu=Neorickettsia group. The members of this group are classied in the genus Neorickettsia, including Neorickettsia helminthoeca, N. (ex Ehrlichia) sennetsu, N. (ex Ehrlichia) risticii and an ehrlichia-like bacterium of a uke (the SF agent), with 16S rRNA sequence similarity between only 94.9 and 100%. In spite of the differences in their 16 rRNA sequences the authors assign all of them to the genus Neorickettsia, on the grounds of a high degree of antigenic similarity. The Wolbachia group (which will not be discussed in the context of this paper). Dumler et al. have thus split up the members of the genus Ehrlichia (at least those that they have examined) between the genera Anaplasma and Ehrlichia, without taking into account the considerable biological differences between members of the classical genus Anaplasma and of classical Ehrlichia. Ehrlichia phagocytophila (including E. equi and the agent of human granulocytic ehrlichiosis [HGE]), E. platys and E. bovis become species of the enlarged genus Anaplasma,

307 while E. ovina (we presume that E. ovis in Figure 1 of Dumler et al. (2001), is a mistake for E. ovina), E. canis, E. chaffeensis, E. ewingii and E. muris remain in the genus Ehrlichia, which also includes Cowdria ruminantium. E. sennetsu and E. risticii are transferred to Neorickettsia. While we can agree to the bare results of these comparisons, we contend that transferring the causal agent of tick-borne fever (and of human and equine ehrlichiosis), E. phagocytophila, as well as E. bovis and E. platys, to the genus Anaplasma, clearly does not take into account phenotypical characteristics and is also based only on small sequences of the genome. Even if this very limited portion indicates a close relationship to Anaplasma spp., parts of the rest of the genome undoubtedly take care of the fact that classical Anaplasma spp. live and multiply in red cells, Cowdria ruminantium predominantly in endothelial cells, and classical Ehrlichia spp. in leukocytes (except for E. platys, in thrombocytes; in cidentally, it is to be noted that this is not a validly published name (J.P. Euzeby, correspondence), and is not in the List of Bacterial Names with Standing in No menclature, Euzeby (1997)). The confusion created is such that we have seen Anaplasma phagocytophilum (unfortunately still spelt sometimes as A. phagocytophila, an initial mistake corrected later [Anonymous 2002]) presented as the agent of human granulocytic ehrlichiosis! Anyhow, human granulocytic anaplasmosis is also extremely confusing. Organisms known to live and multiply in white cells are put in the same genus as those living and multiplying in red cells, which instead of clarication unnecessarily or at least prematurely (until other parts of the genome have been compared) achieves confusion. Furthermore, the results of Lew et al. (2003), while on the basis of 16 rDNA and GroEL (HSP60) sequences conrming the close relationship of Anaplasma and the E. phagocytophila group, show a considerably greater difference between the GroEL amino acid sequences and the 16S rDNA sequences of Ehrlichia phagocytophila and all species of classical Anaplasma tested, than between these sequences of the various classical Anaplasma species. On the basis of a comparison of other gene sequences and other criteria, Inokuma et al. (2001a) and Taillardat-Bisch et al. (2003) to some extent come to different conclusions. While sequence analyses of the citrate synthase gene and rpoB (the gene encoding the b-subunit of RNA polymerase) of various species conrm much of the phylogeny established on the basis of 16S rRNA and groESL sequences by Dumler et al. (2001), it certainly does not conrm all of them. On the basis of the citrate synthase gene the species in the Neorickettsia genogroup are even farther removed from other ehrlichial species than these are from Rickettsia prowazekii and Bartonella henselae. On the basis of the GC content of this gene, the Cowdria ruminantium genogroup [C. ruminantium, Ehrlichia canis, E. chaffeensis, E. muris, and an Ehrlichia sp. recently isolated in Japan from the tick Ixodes ovatus (Shibata et al. 2000)] is different from the E. phagocytophila genogroup, but the latter is again different from Anaplasma marginale and A. centrale (Japanese isolate). The phylogenetic tree constructed from the citrate synthase gene shows more diversity than that from the 16S rRNA gene. The authors state in the discussion In the present study, both A. marginale and A. centrale show low levels of similarity with

308 the E. phagocytophila genogroup . . .. And These two groups were distant in the phylogenetic tree. The authors also mention the biological differences between Anaplasma and the other species. The paper therefore supports the point of view that there is much more to taxonomy than a comparison of small portions of the genome. In a phylogenetic tree based on sequences of the rpoB gene, TaillardatBisch et al. (2003) conrm that A. marginale and the HGE agent are related, but their similarity is nevertheless relatively low (74.5%). It would in fact be more logical to create a new genus name for the Ehrlichia phagocytophila genogroup than lump it in the same genus as erythrocytic Ana plasma, and Euzeby (2003) (who gives an interesting review of the nomenclature of these organisms), suggests a similar approach. In conclusion, we agree that there is considerable diversity among the microorganisms that were grouped together in the classical genus Ehrlichia. We agree with moving E. sennetsu and E. risticii to the genus Neorickettsia and possibly with moving Cowdria ruminantium to Ehrlichia. We do not however consider that it is justied to move E. phagocytophila, E. platys and E. bovis to the genus Anaplasma and recommend that a new genus should be created for these bacteria.

Phylogeny of ticks The third example concerns ticks. Beati and Keirans (2001) studied the phylogeny of hard ticks, using mitochondrial 12S rDNA sequences and morphological characters. Their conclusion is that the genus Boophilus is monophyletic and arose within the genus Rhipicephalus. The conclusions of Murrell et al. (2001) are similar, and these authors relegate Boophilus to a subgenus of Rhipicephalus, as do Barker and Murrell (2003), Murrell and Barker (2003), as well as Horak et al. (2003) in their most recent world list of valid tick names. Murrell et al. base their conclusion on sequences from 12S rRNA, cytochrome c oxidase I, the internal transcribed spacer 2 of the nuclear rRNA, 18S rRNA, as well as on morphology. While they conclusively demonstrate the close relationship between Boophilus and Rhipicephalus, not even a dissecting microscope is needed to see the obvious morphological differences between them, not to mention the differences in life cycle. Our duty as taxonomists is to clarify rather than confuse. Barker and Murrell (2003) state that the name Boophilus could still be used for the ve species concerned if they are in a subgenus of Rhipicephalus, and that it is desirable that the name Boophilus be available. It seems to us much more simple to apply polyphasic taxonomy and retain Boophilus as a separate genus, which does not mean that its close phylogenetic relationship to Rhipicephalus should be denied. There are many examples of species in the animal world which differ less in morphological and molecular characters than Rhipicephalus and Boophilus, but are classied in separate genera. Just think of the family Canidae, foxes, jackals, coyotes, wolves (Vulpes, Alopex, Canis, etc.). Also, not everyone accepts the classication of the chimpanzee and the bonobo in the genus Homo, proposed by Wildman et al. (2003) on the basis of a similarity of 98.499.4%, depending on which genes are compared.

309 Discussion and conclusions Comparing 16S RNA sequences is presently the tool of choice for the rapid identication of bacteria. It has also proved invaluable for correcting previous errors in classication. Such errors were mainly due to the fact that many bacteria were unculturable, hence difcult to characterise and compare. Molecular taxonomy is also much used in the phylogeny of higher organisms. However, modications in classication must be done with great caution, as repeated taxonomic changes will result in confusion rather than clarication. What we hope to get across is the importance of polyphasic taxonomy and that the study of small portions of the genome is an inadequate basis for detailed taxonomic changes, be it for bacteria or higher organisms. The results should always be confronted with biological characteristics before they are used to modify previous classications, and the rest of the genome should not be neglected. Fortunately, recent technical advantages in sequencing open the way to complete genome data and analysis, at least of bacteria, something that was considered a dream only a few years ago. At present, more than 100 complete bacterial genomes are available, with many more to come; this trend is likely to continue. It can be predicted that this will become standard in the near future, as 16S RNA sequencing is standard now. Stackebrandt et al. (2002) recommend that genomic methods other than 16S rRNA sequence analyses and DNADNA hybridisation should also be used in bacterial taxonomy, such as sequence analysis of other genes and G C content of DNA. New bioinformatic tools will decipher the relations between groups of bacteria, as it appears that gene transfer, gene duplication, gene deletion and functional replacement have played important roles in (bacterial) evolution (Bansal and Meyer 2002). It is safe to assume that classication will largely benet from these new tools to come. Until then, caution has to prevail and modications in the actual classication should be supported by unquestionable data. And it should be remembered that polyphasic taxonomy includes not only genotypical but also phenotypical information.

Acknowledgements We are very grateful to Prof. J.P. Euzeby for his valuable comments and suggestions. Writing of this article has been facilitated by the International Consortium on Ticks and Tick-borne Diseases (ICTTD-2) supported by the INCO-DEV program of the European Union under contract number ICA4-CT-2000-30006.

References
Anonymous 2001. Notication list. Notication that new names and new combinations have appeared in volume 51, part 2, of the IJSEM. Int. J. Syst. Evol. Microbiol. 51: 795796. Anonymous 2002. Notication list. Notication that new names and new combinations have appeared in volume 51, part 6, of the IJSEM. Int. J. Syst. Evol. Microbiol. 52: 56.

310
Bansal A.K. and Meyer T.E. 2002. Evolutionary analysis by whole-genome comparisons. J. Bacteriol. 184: 22602272. Barker S.C. and Murrell A. 2003. Phylogeny, evolution and historical zoogeography of ticks: a review of recent progress. Exp. Appl. Acarol. 28: 5568. Beati L. and Keirans J.E. 2001. Analysis of the systematic relationships among ticks of the genera Rhipicephalus and Boophilus (Acari: Ixodidae) based on mitochondrial 12S ribosomal DNA gene sequences and morphological characters. J. Parasitol. 87: 3248. Clark R. 1942. Eperythrozoon felis (sp. nov.) in a cat. J. S. Afr. Vet. Med. Assoc. 13(1): 1516. Dumler J.S., Barbet A.F., Bekker C.P.J., Dasch G.A., Palmer G.H., Ray S.C., Rikihisa Y. and Rurangirwa F.R. 2001. Reorganization of genera in the families Rickettsiaceae and Anaplasmataceae in the order Rickettsiales: unication of some species of Ehrlichia with Anaplasma, Cowdria with Ehrlichia and Ehrlichia with Neorickettsia, description of six new species combinations and designation of Ehrlichia equi and HGE agent as subjective synonyms of Ehrlichia phagocytophila. Int. J. Syst. Evol. Microbiol. 51: 21452165. Euzeby J.P. 1997. List of bacterial names with standing in nomenclature: a folder available on the Internet (URL: http://www.bacterio.cict.fr). Int. J. Syst. Bacteriol. 47: 590592. Euzeby J.P. 2003. Dictionnaire de bacteriologie veterinaire. Available online: http://www.bacterio.cict.fr/ bacdico/garde.html and http://www.bacdico.net. Flint J.C. and McKelvie D.H. 1956. Feline infectious anemia diagnosis and treatment. Proc. 92nd Ann. Meet. Amer. Vet. Med. Assoc. 1955: 240242. Foley J.E. and Pedersen N.C. 2001. Candidatus Mycoplasma haemominutum, a low-virulence epierythrocytic parasite of cats. Int. J. Syst. Evol. Microbiol. 51: 815817. Fox G.E., Wisotzkey J.D. and Jurtshuk Jr. P. 1992. How close is close: 16S rRNA sequence identity may not be sufcient to guarantee species identity. Int. J. Syst. Bacteriol. 42: 166170. Garrity G.M., Johnson K.L., Bell J. and Searles D.B. 2002. Taxonomic outline of the procaryotes. Bergeys Manual1 of Systematic Bacteriology. 2nd edn. Release 3.0, July 2002. p. 150. (Available online: http://www.springer-ny.com/bergeysoutline/main.htm) George J.W., Rideout B.A., Griffey S.M. and Pedersen N.C. 2002. Effect of preexisting FeLV infection or FeLV and feline immunodeciency virus coinfection on pathogenicity of the small variant of Haemobartonella felis in cats. Am. J. Vet. Res. 63: 11721178. Gillis M., Vandamme P., De Vos P., Swings J. and Kersters K. 2001. Polyphasic taxonomy. In: Boone D.R., Castenholz R.W. and Garrity G.M. (eds) Bergeys Manual of Systematic Bacteriology, 2nd edn.. Springer, New York, pp. 4348. Harasawa R., Kawahara M. and Rikihisa Y. 2002. Characteristics of the 16S23S rRNA intergenic spacer region of Mycoplasma haemomuris, previously classied as Haemobartonella muris. J. Vet. Med. Sci. 64: 11611164. Horak I.G., Camicas J.-L. and Keirans J.E. 2003. The Argasidae, Ixodidae and Nuttalliellidae (Acari: Ixodida): a world list of valid tick names. Exp. Appl. Acarol. 28: 2754. Inokuma H., Brouqui P., Drancourt M. and Raoult D. 2001a. Citrate synthase gene sequence: a new tool for phylogenetic analysis and identication of Ehrlichia. J. Clin. Microbiol. 39: 30313039. Inokuma H., Terada Y., Kamio T., Raoult D. and Brouqui P. 2001b. Analysis of the 16S rRNA gene sequence of Anaplasma centrale and its phylogenetic relatedness to other ehrlichiae. Clin. Diagn. Lab. Immunol. 8: 241244. Jensen W.A., Lappin M.R., Kamkar S. and Reagan W.J. 2001. Use of a polymerase chain reaction assay to detect and differentiate two strains of Haemobartonella felis in naturally infected cats. Am. J. Vet. Res. 62: 604608. Johansson K.E., Tully J.G., Bolske G. and Petterson B. 1999. Mycoplasma cavipharyngis and Mycoplasma fastidiosum, the closest relatives to Eperythrozoon spp. and Haemobartonella spp. FEMS Microbiol. Lett. 174: 321326. Lew A.E., Gale K.R., Minchin C.M., Shkap V. and de Waal D.T. 2003. Phylogenetic analysis of the erythrocytic Anaplasma species based on 16S rDNA and GroEL (HSP60) sequences of A. marginale, A. centrale, and A. ovis and the specic detection of A. centrale vaccine strain. Vet. Microbiol. 92: 145160.

311
Maniloff J. 1992. Phylogeny of mycoplasmas. In: Maniloff J., McElhaney R.N., Finch L.R. and Baseman J. (eds) Mycoplasmas Molecular Biology and Pathogenesis. ASM, Washington, USA, pp. 549559. Messick J.B., Walker P.G., Raphael W., Berent L. and Shi X. 2002. Candidatus Mycoplasma haemodidelphidis sp. nov., Candidatus Mycoplasma haemolamae sp. nov. and Mycoplasma haemocanis comb. nov., haemotrophic parasites from a naturally infected opossum (Didelphis virginiana), alpaca (Lama pacos) and dog (Canis familiaris): phylogenetic and secondary structural relatedness of their 16S rRNA genes to other mycoplasmas. Int. J. Syst. Evol. Microbiol. 52: 693698. Murrell A. and Barker S.C. 2003. Synonymy of Boophilus Curtice, 1891 with Rhipicephalus Koch, 1844 (Acari: Ixodidae). Syst. Parasitol. 56: 169172. Murrel A., Campbell N.J.H. and Barker S.C. 2001. A total-evidence phylogeny of ticks provides insights into the evolution of life cycles and biogeography. Mol. Phylogen. Evol. 21: 244258. Neimark H. and Kocan K.M. 1997. The cell wall-less rickettsia Eperythrozoon wenyonii is a Mycoplasma. FEMS Microbiol. Lett. 156: 287291. Neimark H., Johansson K.-E., Rikihisa Y. and Tully J.G. 2001. Proposal to transfer some members of the genera Haemobartonella and Eperythrozoon to the genus Mycoplasma with descriptions of Candidatus Mycoplasma haemofelis, Candidatus Mycoplasma haemomuris, Candidatus Mycoplasma haemosuis and Candidatus Mycoplasma wenyonii. Int. J. Syst. Evol. Microbiol. 51: 891899. Neimark H., Johansson K.-E., Rikihisa Y. and Tully J.G. 2002a. Revision of haemotrophic Mycoplasma species names. Int. J. Syst. Evol. Microbiol. 52: 683. Neimark H., Barnaud A., Gounon P., Michel J.-C. and Contamin H. 2002b. The putative haemobartonella that inuences Plasmodium falciparum parasitaemia in squirrel monkeys is a haemotrophic mycoplasma. Microbes Infection 4: 693698. Nowak J. 1929. Morphologie, nature et cycle evolutif du microbe de la peripneumonie des bovides. Ann. Inst. Pasteur 43: 13301352. Rikihisha Y., Kawahara M., Wen B., Kociba G., Fuerst P., Kawamori F., Suto C., Shibata S. and Futohashi M. 1997. Western immunoblot analysis of Haemobartonella muris and comparison of 16S rRNA gene sequences of H. muris, H. felis, and Eperythrozoon suis. J. Clin. Microbiol. 35: 823829. Schilling V. 1928. Eperythrozoon coccoides, eine neue durch Splenektomie aktivierbare Dauerinfektion der weissen Maus. Klin. Wochschr. 7: 18531855. Shibata S., Kawahara M., Rikihisa Y., Fujita H., Watanabe Y., Suto C. and Ito T. 2000. New Ehrlichia species closely related to Ehrlichia chaffeensis isolated from Ixodes ovatus ticks in Japan. J. Clin. Microbiol. 38: 13311338. Stackebrandt E. and Goebel B.M. 1994. Taxonomic note: a place for DNA-DNA reassociation and 16S rRNA sequence analysis in the present species denition in bacteriology. Int. J. Syst. Bacteriol. 44: 846849. Stackebrandt E., Frederiksen W., Garrity G.M., Grimont P.A.D., Kampfer P., Maiden M.C.J., Nesme X., Rossello-Mora R., Swings J., Truper H.G., Vauterin L., Ward A.C. and Whitman W.B. 2002. Report of the ad hoc committee for the re-evaluation of the species denition in bacteriology. Int. J. Syst. Evol. Microbiol. 52: 10431047. Taillardat-Bisch A.V., Raoult D. and Drancourt M. 2003. RNA polymerase b-subunit-based phylogeny of Ehrlichia spp., Anaplasma spp., Neorickettsia spp. and Wolbachia pipientis. Int. J. Syst. Evol. Microbiol. 53: 455458. Tasker S., Binns S.H., Day M.J., Gruffydd-Jones T.J., Harbour D.A., Helps C.R., Jensen W.A., Oliver C.S. and Lappin M.R. 2003a. Use of a PCR assay to assess the prevalence and risk factors for Mycoplasma haemofelis and Candidatus Mycoplasma haemominutum in cats in the United Kingdom. Vet. Rec. 152: 193198. Tasker S., Helps C.R., Day M.J., Harbour D.A., Shaw S.E., Harrus S., Baneth G., Lobetti R.G., Malik R., Beauls J.P., Belford C.R. and Gruffydd-Jones T.J. 2003b. Phylogenetic analysis of hemoplasma species: an international study. J. Clin. Microbiol. 41: 38773880. Vandamme P., Harrington C.S., Jalava K. and On S.L.W. 2000. Misidentifying helicobacters: the Helicobacter cinaedi example. J. Clin. Microbiol. 38: 22612266.

312
Westfall D.S., Jensen W.A., Reagan W.J., Radecki S.V. and Lappin M.R. 2001. Inoculation of two genotypes of Hemobartonella felis (California and Ohio variants) to induce infection in cats and the response to treatment with azithromycin. Am. J. Vet. Res. 62: 687691. Wildman D.E., Uddin M., Liu G., Grossman L.I. and Goodman M. 2003. Implications of natural selection in shaping 99.4% nonsynonymous DNA identity between humans and chimpanzees: enlarging genus Homo. Proc. Nat. Acad. Sci. 100: 71817188. Woese C.R. 1987. Bacterial evolution. Microbial Rev. 51: 221271.