| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Biological Sciences, Towson University, Towson, MD 21252, USA
2 US Department of Agriculture, Vegetable Laboratory, BARC, Beltsville, MD 20705, USA
3 Business Group Diagnostics, Bayer Corporation, Emeryville, CA 94608, USA
4 Department of Biology, Georgia Southern University, Statesboro, GA 30460, USA
5 Institute for Biological Energy Alternatives, 1901 Research Boulevard, Suite 600, Manassas, VA 20850, USA
6 Department of Anatomical Sciences, State University of New York, Stony Brook, NY 11794, USA
Correspondence
Gail E. Gasparich
ggasparich{at}towson.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
and Johansson and Pettersson 2002) is a group of mollicutes that includes the type species – Mycoplasma mycoides – of Mycoplasmatales, Mycoplasmataceae and Mycoplasma. This cluster, associated solely with ruminants, contains five other species and subspecies. Earlier phylogenetic reconstructions based on partial 16S rDNA sequences and a limited sample of Spiroplasma and Mycoplasma sequences suggested that the genus Mycoplasma was polyphyletic, as the M. mycoides cluster and the grouping that consisted of the hominis and pneumoniae groups of Mycoplasma species were widely separated phylogenetically and the M. mycoides cluster was allied with Spiroplasma. It is shown here that the M. mycoides cluster arose from Spiroplasma through an intermediate group of non-helical spiroplasmal descendants – the Entomoplasmataceae. As this conclusion has profound implications in the taxonomy of Mollicutes, a detailed phylogenetic study of Spiroplasma and its non-helical descendants was undertaken. These analyses, done with maximum-parsimony, provide cladistic status; a new nomenclature is introduced here, based on ‘bottom-up’ rather than ‘top-down’ clade classification. The order Entomoplasmatales consists of four major clades: (i) the Mycoides–Entomoplasmataceae clade, which contains M. mycoides and its allies and Entomoplasma and Mesoplasma species and is a sister lineage to (ii) the Apis clade of Spiroplasma. Spiroplasma and the Entomoplasmataceae are paraphyletic, but this status does not diminish their phylogenetic usefulness. Five species that were previously unclassified phylogenetically are basal to the Apis clade sensu strictu and to the Mycoides clade. One of these species, Spiroplasma sp. TIUS-1, has very poor helicity and a very small genome (840 kbp); this putative species can be envisioned as a ‘missing link’ in the evolution of the Mycoides–Entomoplasmataceae clade. The other two Spiroplasma clades are: (iii) the Citri–Chrysopicola–Mirum clade (serogroups I, II, V and VIII) and (iv) the ixodetis clade (serogroup VI). As Mesoplasma lactucae represents a basal divergence within the Mycoides–Entomoplasmataceae clade, and as Entomoplasma freundtii is basal to the Mycoides clade, M. mycoides and its allies must have arisen from an ancestor in the Entomoplasmataceae. The paraphyletic grouping that consists of the Hominis and Pneumoniae groups (sensu Johansson & Pettersson 2002) of Mycoplasma species contains the ancestral roots of Ureaplasma spp. and haemoplasmas. This clade is a sister lineage to the Entomoplasmatales clade. Serological classifications of spiroplasma are very highly supported by the trees presented. Genome size and G+C content of micro-organismal DNA were moderately conserved, but there have been frequent and polyphyletically distributed genome reductions. Sterol requirements were polyphyletic, as was the ability to grow in the presence of polyoxyethylene sorbitan-supplemented, but not serum-supplemented, media. As this character is not phylogenetically distributed, Mesoplasma and Entomoplasma should be combined into a single genus. The phylogenetic trees presented here confirm previous reports of polyphyly of the genus Mycoplasma. As both clades of Mycoplasma contain several species of great practical importance, a change of the genus name for species in either clade would have immense practical implications. In addition, a change of the genus name for M. mycoides would have to be approved by the Judicial Commission. For these reasons, the Linnaean and phylogenetic classifications of Mycoplasma must for now be discrepant.
Published online ahead of print on 23 January 2004 as DOI 10.1099/ijs.0.02688-0.
The GenBank/EMBL/DDBJ accession numbers for new Spiroplasma 16S rDNA sequences are: Spiroplasma sp. strain 277F, AY189312; Spiroplasma sp. strain LB-12, AY189313; S. insolitum, AY189133; S. floricola, AY189131; S. syrphidicola, AY189309; S. chrysopicola, AY189127; Spiroplasma sp. strain TAAS-1, AY189314; S. culicicola, AY189129; S. velocicrescens, AY189311; S. sabaudiense, AY189308; S. corruscae, AY189128; Spiroplasma sp. strain CB-1, AY189315; Spiroplasma sp. strain Ar 1357, AY189316; S. turonicum, AY189310; S. litorale, AY189306; S. lampyridicola, AY189134; S. leptinotarsae, AY189305; Spiroplasma sp. strain W115, AY189317; S. chinense, AY189126; S. diminutum, AY189130; S. alleghenense, AY189125; Spiroplasma sp. strain TIUS-1, AY189318; Spiroplasma sp. strain BIUS-1, AY189319; S. montanense, AY189307; S. helicoides, AY189132; Spiroplasma sp. BARC 1901, AY189320.
Tables summarizing the phylogenetic trees constructed for the genus Spiroplasma and their support are available as supplementary material in IJSEM Online.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
, 1998; Tully & Whitcomb, 1990). In addition to their unique cellular morphology, species of this genus have attracted recent attention as the root of a small clade that contains the type species of the genus Mycoplasma. The genus Spiroplasma was discovered relatively recently. The first species described in this genus, Spiroplasma citri (Saglio et al., 1973), is the aetiological agent of citrus stubborn disease (Calavan & Bové, 1989) and resides in an obligate cycle in the plant phloem and the insect vector. Another spiroplasma – thought at the time to be a spirochaete – was shown to cause a sex-ratio abnormality in Drosophila (Poulson & Sakaguchi, 1961; Williamson & Poulson, 1979). A third spiroplasma, eventually named Spiroplasma kunkelii, discovered shortly before S. citri was described (Davis et al., 1972; Whitcomb et al., 1986), is also an inhabitant of the plant phloem/insect habitat (Hackett & Clark, 1989). Spiroplasmas were then found to inhabit ticks (Burgdorfer et al., 1975; Tully et al., 1976, 1982, 1995). One of these, Spiroplasma mirum, can cause experimental disease in suckling rodents (Tully et al., 1977). It was not realized until 1977 (Clark, 1977, 1982) that the major reservoir of spiroplasmas was insects and that the surfaces, rather than the phloem, of flowers and other plant parts were the major site for spiroplasma acquisition and transmission (Davis, 1978; McCoy et al., 1979; Clark et al., 1987).
The general phylogenetic position of the genus Spiroplasma was determined as a result of the pioneering work of Carl Woese and his associates [summarized by Woese (1987)], which utilized 16S rRNA for the study of prokaryote phylogeny. As the class Mollicutes was of special interest (Woese et al., 1985), these studies led to an in-depth study of its evolution [reviewed by Johansson & Pettersson (2002)]. The class Mollicutes was shown to be a terminus in the evolution of Gram-positive bacteria. Early studies (Woese et al., 1980) indicated that the class Mollicutes was divided into four major phylogenetic groups – a clade that contained Anaeroplasmatales and Acholeplasmatales, a clade that consisted of the hominis and pneumoniae groups of Mycoplasma, a clade that contained Spiroplasma and Mycoplasma mycoides and a monospecific clade that contained Asteroleplasma anaerobium. Although early studies were non-committal on the monophyly of the Acholeplasmatales–Anaeroplasma and Spiroplasma–Entomoplasmataceae–Mycoides clades (Woese et al., 1980), subsequent phylogenetic analyses have led workers to hypothesize that they are monophyletic. However, the studies of Weisburg et al. (1989) indicated that one of the three major groups of Mollicutes – the asteroleplasma group – may be allied more closely with certain genera of low-G+C content, Gram-positive bacteria than with other mollicutes. Thus, walllessness probably evolved at least twice in the evolution of Gram-positive bacteria. In addition to the asteroleplasma group, Weisburg and colleagues defined the pneumoniae, hominis, anaeroplasma and spiroplasma groups.
The M. mycoides cluster of Weisburg et al. (1989) is a group that, with all recent additions, contains 23 species (J. G. Tully, personal communication). Five of these form a monophyletic grouping that is restricted to ruminant animals. One of these, M. mycoides, is the type species of Mollicutes, Mycoplasmataceae and Mycoplasma. M. mycoides is divided into several subtaxa (some of which have been recognized at the subspecies level) that differ in their host range, pathogenicity and serological and genomic characters (Heldtander et al., 1998; Persson, 2002). M. mycoides subsp. mycoides SC is the causative agent of contagious bovine pleuropneumonia [reviewed by Persson (2002)]. This disease, which was recognized as long ago as the middle ages (reviewed by Provost et al., 1987), is the only bacterial disease on the A-list of communicable animal diseases (FAO EMPRES, 2000). This classification places it as one of the 15 most serious animal diseases in the world. The agent was cultivated in 1898 by Nocard and Roux. Dujardin-Beaumetz, a co-worker of Nocard and Roux, introduced filters to separate the organism from contaminants and was the first to cultivate the organisms on agar, where they formed fried-egg colonies [summarized by Bové et al. (1994)]. The organism was described as Asterococcus mycoides in 1910 and later reclassified in a novel genus, Mycoplasma (Nowak, 1929). In 1956, the current classification was established (Edward & Freundt, 1956). A second subspecies of the M. mycoides cluster, Mycoplasma capricolum subsp. capripneumoniae, is a serious pathogen of caprine animals (Johansson & Pettersson, 2002). A genome-sequencing project for M. mycoides has shown that the genome is 1140 kbp in size, with a very low G+C content of 24 mol%. The M. mycoides genome has 1067 fully annotated ORFs and many have been correlated with metabolic functions (Persson, 2002).
Earlier phylogenetic studies (Woese et al., 1980; Weisburg et al., 1989; Maniloff, 1992) produced a major surprise – that some members of the genus Mycoplasma, including the type species, M. mycoides, appeared to belong to the spiroplasma group and that the genus Mycoplasma, like Mollicutes, might therefore be polyphyletic. These studies suggested that a small group of species, including M. mycoides and its cohorts – all from bovine and caprine sources – were placed properly in the spiroplasma group of mollicutes.
The discovery of non-helical mollicutes in the insect gut/plant surface habitat (Clark, 1977; Davis, 1978; Clark et al., 1986; Tully et al., 1994) revealed a large assemblage of previously unrecognized mollicutes. These organisms appeared at first to be Acholeplasma or Mycoplasma species, but some were later found to be phylogenetically distinct from those genera (Weisburg et al., 1989). This newly discovered group of organisms was later referred to a novel family, Entomoplasmataceae (Tully et al., 1993), which had two genera. Entomoplasma was proposed for organisms that did not require sterol and Mesoplasma was proposed for certain species that could grow in the absence of sterol if, and only if, cultures of the organism were grown in media that were supplemented with 0·04 % polyoxyethelene sorbitan (PES).
The apparent polyphyly of Mycoplasma has profound implications for mollicute taxonomy. M. mycoides is the type species of the order Mycoplasmatales, the family Mycoplasmataceae and the genus Mycoplasma. If this species was shown conclusively to be phylogenetically remote from other Mycoplasma species, in an era that features strong attempts to reconcile binomial and phylogenetic classification, a major taxonomic quandary would be presented. Earlier phylogenies (Weisburg et al., 1989; Maniloff, 1992) were derived from a database of largely incomplete sequences and were analysed by phenetic analytical methods (distance and the largely phenetic neighbour-joining method). However, now that a much more complete dataset exists and cladistics is becoming one of the cornerstones of microbial systematics (Ludwig & Schleifer, 1999), it is important that this profound conclusion be revisited and confirmed.
An appendix of the taxonomic terms used in the text is given in Table 1.
|
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
. Most strains of Spiroplasma, Entomoplasma and Mesoplasma were isolated, cloned and preserved in laboratories at USDA, Beltsville, MD, USA; SUNY, Stony Brook, NY, USA; or at the NIAID laboratory at Frederick, MD, USA. Type or representative strains were submitted to, and are available from, the American Type Culture Collection (ATCC). Representative cultures of these strains and of other non-type strains have been deposited at Purdue University (West Layfayette, IN,USA). Most Spiroplasma, Mesoplasma and Entomoplasma strains were grown in M1D broth medium (Whitcomb, 1983), but some were cultured in SP-4 broth medium (Whitcomb, 1983). Cultures of Spiroplasma leptinotarsae were grown in DCCM broth medium (Hackett et al., 1996a). All cultures were maintained at their optimum temperatures (Konai et al., 1996).
|
) or the SDS lysis protocol, as described previously (Gasparich et al., 1993).
In vitro amplification and DNA sequencing of the 16S rRNA gene.
16S rRNA genes of the species investigated were PCR-amplified from genomic DNA. PCR sequence mixtures contained 2 µl extracted DNA, 50 µl MicroSeq 16S rDNA PCR kit (PE/Applied Biosystems) PCR mastermix and 48 µl water. PCR conditions were those recommended by the manufacturer (PE/Applied Biosystems). The PCR product generated was 1540 bp in length. PCR products were purified by using a Microcon-100 column (Amicon), following the protocol designated by the manufacturer. Reaction products were cycle-sequenced by using a MicroSeq 16S rRNA Gene Sequencing kit (PE/Applied Biosystems). Excess dye terminators were removed by using Centri-Sep spin columns (Princeton Separations) as recommended by the manufacturer. Reaction products were then dried in a Speed Vac (Savant) and resuspended in 4 µl sequencing gel loading buffer (25 mM EDTA, 50 mg Blue Dextran ml–1 and 1 : 5 deionized formamide). A 2 µl aliquot was then loaded onto an ABI 377 DNA sequencer (PE/Applied Biosystems) and electrophoresed at 1650 V and 52 °C for 7 h.
Assembly of sequences.
Sequences were analysed by using the software programs Sequencing Analysis (PE/Applied Biosystems) and Factura (PE/Applied Biosystems). Analysed sequences were then assembled and edited by using Auto Assembler (PE/Applied Biosystems) and a consensus sequence was generated.
Nucleotide sequence accession numbers.
The new 16S rDNA sequences that were used in this study have been deposited in GenBank under the accession numbers listed in Table 2
.
Phylogenetic analyses.
Sequences were aligned by using CLUSTALX (Thompson et al., 1994) and then aligned manually in MacClade (Maddison & Maddison, 1992). Maximum-parsimony, maximum-likelihood, distance and neighbour-joining analyses were performed by using PAUP (version 4.0b10; Swofford, 1998). Parameters for analyses are presented in Supplementary Table A (available in IJSEM Online).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
; Rogers et al., 1985; Weisburg et al., 1989; Maniloff, 1992; Johansson & Pettersson, 2002). The extensive work published by K. E. Johansson's group [reviewed in part by Johansson & Pettersson (2002)] has been done by using neighbour-joining, maximum-parsimony and maximum-likelihood. The large and globally inclusive trees generated by Gundersen et al. (1994) were constructed by using maximum-parsimony. Distance analyses (using p-distance) were performed in the initial work on mollicute phylogeny by Woese et al. (1980) and Weisburg et al. (1989), and neighbour-joining analyses were performed by Maniloff (1992). We are aware that each method has its detractors, especially distance (e.g. Maniloff, 1992; Ludwig & Schleifer, 1994; Farris et al., 1999). Despite intense debate (Ludwig & Schleifer, 1994), a single method for phylogenetic reconstruction that can be used with complete confidence has not yet been identified. Therefore, in the current study, we used all four major algorithms to analyse the dataset (see Supplementary Table B in IJSEM Online). Consensus for spiroplasma groupings was obtained by using maximum-parsimony analyses (See Figs 1–4).
|
|
|
|
). Some authors have discussed the influence of position choice on the topology of phylogenetic trees (Hansmann & Martin, 2000). These authors point out that there is no a priori rationale for inclusion or exclusion of variable regions of a molecule, which, while difficult to align, may contain valuable phylogenetic signal. It is not clear why variable regions, if they can be aligned with reasonable certainty, should be excluded from analyses. In fact, when workers turn to analysis of the 16S–23S spacer region, variability is considered valuable for the resolution of intraspecific variability or variability among candidate members of very closely related species. Given the small evolutionary distances among some members of our strain complexes (e.g. serogroups I and VIII), an inclusionary strategy was followed in selecting the positions for most of our trees. Under all these circumstances, results produced by trees generated under a wide range of assumptions offer important evidence. Support generated by the totality of the matrix of generated trees is important and where unanimous or nearly unanimous conclusions are reached, the phylogenetic inferences are surely robust.
Problems in nomenclature of groupings
Weisburg et al. (1989) named ‘groups' by choosing a well-known species that was represented in the clusters they obtained with their distance analyses. We use the term ‘clade’ rather than ‘group’ in this paper, as all of the groupings that we designate have been obtained or confirmed by maximum-parsimony analysis. Johansson & Pettersson (2002) have followed Weisburg's precedent with considerable success; however, in our study, we found it difficult to adapt the existing nomenclature to certain clade structures.
One problem is that the concept of ‘group’ was first introduced into spiroplasma taxonomy (Junca et al., 1980; Bové et al., 1983; Whitcomb et al., 1987; Williamson et al., 1989) to denote putative species that were awaiting full characterization, according to proposed minimal standards (International Committee on Systematic Bacteriology Subcommittee on the Taxonomy of Mollicutes, 1995). In this paper, to avoid confusion between this conflicting usage of ‘group’, we use the term ‘serogroup’ to denote spiroplasma strain assemblages that were classified under this system. However, in addition, problems arose with attempts to adapt the historical group–cluster–subcluster terminology (Weisburg et al., 1989; Johansson & Pettersson, 2002) to a rapidly growing array of mollicute taxa – the Apis clade.
All except one of the clades in the trees that we present here could be classified by Weisburg's group concept. Spiroplasma ixodetis could be easily designated as either a monospecific clade or group. Spiroplasmas in serogroups I and II or serogroup VIII, which each, respectively, form distinct clades, could be viewed as subclusters under the Weisburg nomenclature. Spiroplasma mirum, which we treat here as a monospecific clade, could be treated as a third subcluster of the spiroplasma group (sensu Weisburg et al., 1989). Also, the 18 species of Entomoplasmataceae and the five species of the M. mycoides cluster (which we show here to have clade status) could be classified as subclusters of Weisburg's ‘M. mycoides cluster’. However, the remaining 26 Spiroplasma species, which we define herein to comprise the ‘Apis clade sensu latu’, are difficult to classify without creating either a large number of subclusters or none at all. Weisburg et al. (1989), who studied only six of these species, termed this grouping the ‘Spiroplasma apis cluster’. However, as a result of discoveries of spiroplasmas in tabanids (Diptera: Tabanidae), mosquitoes (Diptera: Culicidae) and flower surfaces, the ‘S. apis cluster’ (which we show herein to have clade status) has now expanded to include 21 species that clearly fall within the topologic boundaries that were established by the Weisburg study. Five species sequenced in this study occupy topological positions and are moderately outside the original boundaries. Should the newly designated ‘Apis clade’ accrete these five species? In this paper, we have added these five species to the 21 species that can be assigned on the basis of their topological relationship to the Weisburg species set, to form an ‘Apis clade sensu latu’. The clade, which consists of the 21 species that fit within the boundaries of Weisburg's concept, we term the ‘Apis clade sensu strictu’. The designations ‘latu’ and ‘strictu’ are used commonly in taxonomy when group boundaries are redefined.
It was in the classification of the Apis clade that our ability to use the established ‘group classification’ vanished. We were unable to separate the 26 species of the Apis clade sensu latu into subclusters without considerable fragmentation. If this large clade had not been subdivided, however, a substantial number of clades with clear biological significance would have escaped naming. Subdivision into ‘subclusters' fails in the Apis clade, as it must in many cases of rapid taxic expansion, in that by starting at the top of the hierarchical level with ‘group’, defining only ‘cluster’ and ‘subcluster’ beneath, one is likely – in large groups – to exhaust the available nomina before all useful subdividing has been accomplished. To impose this kind of nomenclature onto a phylogenetic classification duplicates one of the most serious shortcomings of Linnaean nomenclature: the imposition of a template on clade structures that are unsuited to it (see discussion on Linnaean classification versus phylogenetic classification). These comments should not be construed in any way as criticism of the classification of Johansson & Pettersson (2002). Their useful classification represents the best possible attempt – in the interest of stability – to maintain the initial nomenclature of Weisburg et al. (1989) and has served well, even for the large and heterogeneous genus Mycoplasma. In summary, structural problems with clade topologies and the previous use of ‘group’ in Spiroplasma taxonomy in an entirely different sense make it difficult or impossible to use the ‘group’ concept for Spiroplasma clade nomenclature.
As noted in this paper, we approach nomenclature by defining clades. We use this term because all groupings to which we assign this term are derived from cladistic analyses (Swofford, 1998) – in most cases, maximum-parsimony. We also use the term ‘clade’ because the groups so defined will not be confused with any of the groups, clusters or subclusters of Johansson & Pettersson (2002). However, the difference in nomenclature does not imply that the underlying tree structures are substantially different; in fact, in all cases where the trees of Johansson & Pettersson (2002) and our trees describe the same species, the underlying topologies are quite similar and are sometimes identical.
Clade nomenclature
Clades mentioned in the text can be defined by referring to Tables 3 and 4 and Figs 1 and 2. Our nomenclatural approach differs from previous ones in that it starts not at a higher level, but at the level of the smallest clades, beginning with monospecific clades and clades that are composed of sister species. We present the data in this way as it is often clades of two to six species that are the most biologically and/or phenotypically significant. Our terminology uses: (i) lower-case species names (without italics) for clades with three or fewer species. For example, the sister species Spiroplasma lampyridicola and S. leptinotarsae form the lampyridicola–leptinotarsae clade. (ii) Clades with four or more species were named by using one or more of the component species, whose names served as nomina. These clades were named by using the single species name serving as the nomen, but in capital letters without italics. In Figs 1 and 2, we present phylogenetic trees that, with Tables 3 and 4, show the clade labelling. As the discussions in this paper involve not only Spiroplasma clades, but also clades of Mollicutes at higher hierarchical levels, we have adopted the standard clade terminology for higher-level taxa shown in Fig. 1. (iii) Where higher-level clades are identified, we retain the initial capital letter of the Linnaean name and the Linnaean spelling of the taxon, not italicized.
|
|
is a parsimony majority-rule consensus tree and Fig. 4 shows the same dataset with bootstrap values included (500 replicates).
Highly supported features of spiroplasma trees
The fundamental premise of cladistic taxonomy is that appropriate analyses will yield nested clusters (clades) of taxa that, having arisen from a single ancestor, share a common heritage. Twenty-four such clades are supported strongly by our studies, in that they occur in all, or almost all, trees constructed, regardless of the algorithm used in the analysis, and are usually afforded high support in bootstrap analyses. Supplementary Table B (available in IJSEM Online) lists the fraction of the 23 trees analysed that support each clade. In the following section, we list these highly supported clades and features and briefly discuss the biological characteristics of the clades that they define.
Monophyly of the Mycoplasmatales–Entomoplasmatales clade of Mollicutes.
The orders Mycoplasmatales and Entomoplasmatales form a clade that is a sister to the Acholeplasma–Anaeroplasma–Phytoplasma clade of Mollicutes. The order Mycoplasmatales is polyphyletic, in that Mycoplasma species are split into two phylogenetically separate sections that do not share a common ancestor. Nevertheless, the two orders, taken together, are shown herein to have derived from a single common ancestor. Unlike most other prokaryotes, members of the Mycoplasmatales–Entomoplasmatales clade – so far as is known – use UGA as a codon for tryptophan and not as a stop codon. This character, although it has not been determined for most mollicutes, has been hypothesized to be synapomorphic for the Mycoplasmatales–Entomoplasmatales clade. Although this clade has no formal recognition in Linnaean taxonomy, it represents a profound split in the mollicute lineage and may have diverged from an ancestral lineage hundreds of millions of years ago (Hackett, 1990; Hackett et al., 1990; Maniloff, 2002).
Position of the Spiroplasma–Entomoplasmataceae–Mycoides grouping (SEM) clade as a sister lineage to the Pneumoniae–Hominis–Ureaplasma–Haemoplasma (PHUH) clade.
As in other trees that represent mollicute phylogeny [reviewed by Johansson & Pettersson (2002)], our trees showed consistently that the paraphyletic SEM clade is a sister lineage to a large, also paraphyletic, clade (PHUH) that contains over 100 species of Mycoplasma, including the (paraphyletic) Pneumoniae clade, the Hominis clade, ureaplasmas and haemoplasmas. Mycoplasma species of this clade form a paraphyletic unit that has no Linnaean designation, but given its origin from a common ancestor, should be treated as monophyletic for purposes of binomial nomenclature. The Pneumoniae clade (e.g. Mycoplasma pneumoniae, Mycoplasma genitalium, Mycoplasma iowae and Mycoplasma pirum) has many species in which cytoskeletal modifications, including terminal blebs, are present. This technically paraphyletic clade contains the ancestral root of the genus Ureaplasma, an apomorphic taxon that presumably arose by transfer of a lineage of the Pneumoniae clade into the urogenital tract of vertebrates. The Pneumoniae clade also is the ancestral root of ‘haemoplasmas' (Neimark & Kocan, 1997), a trivial name that was proposed by Neimark et al. (2001) for a cluster of haemophilic mycoplasmas, which had been named before the general concept of Candidatus status had been proposed (Murray & Stackebrandt, 1995). Haemoplasma species have recently been transferred to the genus Mycoplasma (Neimark et al., 2001).
Position of the Mycoides clade within Entomoplasmatales.
Here, we confirm and extend, by addition of 26 new complete Spiroplasma sequences (designated by GenBank accession numbers starting with AY in Table 2), the reports of others (Weisburg et al., 1989; Maniloff, 1992; Gundersen et al., 1994; Johansson & Pettersson, 2002) that M. mycoides and four allied Mycoplasma species, together with the Entomoplasmataceae clade, with which it has a sister relationship, comprise an evolutionary terminus that is derived from a spiroplasmal ancestor. Together, these groupings comprise the Mycoides–Entomoplasmataceae clade. Our trees demonstrate that the Mycoides clade is apomorphic in this clade of 23 species. Derivation of the Mycoides clade from the Entomoplasmataceae renders this family technically paraphyletic. We use the word ‘technically’ because we regard an emphasis on paraphyly as detrimental to the quest for groupings of species with common ancestors. The fact that a lineage has split off from an otherwise cohesive clade, in our view, does nothing to alter the evolutionary history of the clade itself or to change its status as having arisen from a common ancestor. Hence, realizing that not all systematists agree with us, we have chosen to note paraphyly without proposing it to be a nomenclatural determinant in Linnaean systems. It is polyphyly, not paraphyly, that threatens the utility of Linnaean classifications. The Mycoides–Entomoplasmataceae clade is a sister lineage to the Apis clade sensu latu, which is shown here to contain the 21 species (or putative species) of the Apis clade sensu strictu [the six species analysed by Weisburg et al. (1989) plus 15 spiroplasmas sequenced herein, whose phylogenetic position falls within the boundaries established by that study] plus two novel clades (the lampyridicola–leptinotarsae and sabaudiense–alleghenense–TIUS-1 clades), which together have a total of five species, all topologically placed outside the original group boundaries; thus, there are 26 spiroplasmas in the Apis clade sensu latu. Both novel clades are basal to the Apis clade sensu strictu and are thus related more closely to the Entomoplasmataceae and M. mycoides than members of the Apis clade sensu strictu. The discovery of other spiroplasmas that are related more closely to M. mycoides than those previously known enables us to better evaluate the relationship of the Mycoides clade (and its sister, the Entomoplasmataceae clade) to spiroplasmas.
Essential monophyly of Spiroplasma and subdivision of the SEM clade.
All currently classified Spiroplasma species form a single evolutionary unit that is derived from a common ancestor. As explained above, we do not believe that the status of Spiroplasma as being technically paraphyletic should be considered in the construction of Linnaean classifications. All trees and bootstrap values for the major nodes strongly support the subdivision of this clade into the ixodetis clade, the Citri–Chrysopicola–Mirum clade, the Apis clade sensu latu and the Mycoides–Entomoplasmataceae clade.
The ixodetis clade is basal in the spiroplasma lineage.
The ixodetis clade diverges at the base of the Spiroplasma evolutionary tree. Originally, this clade was thought to consist of a single species, S. ixodetis (serogroup VI) and, technically, it still does. However, there have been several recent claims that certain micro-organisms that are associated with insects cluster with S. ixodetis. In each case, evidence consists of the 16S rDNA sequence similarity of the candidate organisms to spiroplasmal DNA of S. ixodetis, as revealed by BLAST searches. It should be noted that BLAST searches may not reveal the nearest neighbour and are by no means final arbiters of phylogenetic position (Koski & Golding, 2001). Also, in no case were micro-organisms envisioned or cultured. In some cases (Hurst et al., 1999; Hurst & Jiggins, 2000; Jiggins et al., 2000), organisms were reported to be associated with sex-ratio abnormalities in coccinellid beetles (Adalia bipunctata) or a butterfly (Danaus chrysippus) (Jiggins et al., 2000). A publication from Russia (Sokolova et al., 2002) reported electron microscopic visualization of spiroplasmas in ovarioles of A. bipunctata. The structures depicted appear to be filamentous, but do not display clear helicity. There is no way of knowing whether the envisioned organisms are actually spiroplasmas and, if so, whether they represent the Adalia sex-ratio agent. In any event, phylogenetic placement of the Adalia agent is problematical, especially when one considers that the taxic sampling used in phylogenetic analyses concerning them was extremely limited. Other isolates with an apparent topological relationship were from the bamboo pseudococcid (Fukatsu & Nikoh, 2000) and green pea aphid (Fukatsu et al., 2001). The affinity of these hypothetical ‘group members' to S. ixodetis must be examined in the context of long-branch attraction (Carmean & Crespi, 1995; Siddall & Whiting, 1999; Stiller & Hall, 1999). S. ixodetis exhibits an unusual, tightly coiled helicity, whereas other spiroplasmas display a less tightly coiled helicity. Whether this morphology is apomorphic or plesiomorphic for spiroplasmas is at present unknown. Detailed study of the cytoskeletal structure of spiroplasmas suggests that helicity itself is a fundamental condition that can be modified without substantial genomic change (Trachtenberg & Gilad, 2001; S. Trachtenberg, personal communication). The S. ixodetis genome is 2220 kbp in size, which is the largest in the genus Spiroplasma and just 500 kbp smaller than the genome of Acholeplasma laidlawii. However, the complete genome sequence of S. ixodetis has not been determined, so we do not know how much of this genome represents ORFs or, alternatively, repeat sequences or integrated viral or plasmid DNA, which represent a substantial portion of some spiroplasmal genomes (U. Melcher, personal communication).
Monophyly of the Citri–Chrysopicola–Mirum clade.
High support is accorded to the monophyly of the clade Citri–Chrysopicola–Mirum, which contains serogroup I and II spiroplasmas (Citri–Poulsonii clade), serogroup VIII spiroplasmas (chrysopicola–syrphidicola–TAAS-1 clade) and the mirum clade [S. mirum (serogroup V)], despite the substantial phenotypic differences that characterize each of these three component clades. Given the conspicuous phenotypic and genotypic differences between the Chrysopicola–Syrphidicola–TAAS-1 and Citri–Poulsonii clades, their sister status is surprising.
Monophyly of the Citri–Poulsonii clade.
High support is obtained for the clade (Citri–Poulsonii) that contains the seven sequenced serogroup I spiroplasmas and the serogroup II species Spiroplasma poulsonii. Serogroup I spiroplasmas form an ecologically diverse cluster that has been divided into eight subgroups on the basis of serology and DNA–DNA reassociation (Junca et al., 1980, Bové et al., 1983). In this paper, we include in our trees only five of the eight sequenced members of the Citri–Poulsonii clade; S. poulsonii, Spiroplasma phoeniceum and S. kunkelii, which have been sequenced, are not represented in Figs 3 or 4, and strain N525, the representative strain for subgroup I-7, has not been sequenced. Trees for the Citri–Poulsonii clade that contained all eight available sequences showed some minor inconsistencies in their topologies. We have chosen to omit the three sequenced species from the trees we present here, as topologies for this group have not yet been resolved completely. Of course, the small inconsistencies observed in internal topology do not negate the integrity of the clade itself. It is to be expected that the nodes of the serogroup I cluster should be difficult to resolve (Stackebrandt & Goebel, 1994), as the similarity coefficients of their 16S rDNA sequences ranged from 0·986 to 0·991.The Citri–Poulsonii clade is of great interest, in that it has a wide diversity of spiroplasma–host associations. Members of this cluster occur in ticks, honeybees, leafhoppers (plant phloem-sucking insects), plant/flower surfaces and Drosophila, in which they cause sex-ratio abnormalities (Williamson & Poulson, 1979; Williamson et al., 1989). Given these host relationships, each node in the Citri–Poulsonii clade is of critical importance.
Basal position of subgroup I-5 spiroplasmas in the Citri–Poulsonii clade.
The basal position of subgroup I-5 strain LB-12 and its allies is supported by all topologies we obtained for the Citri–Poulsonii clade. This spiroplasma was isolated from the green leaf bug in Taiwan (Lei et al., 1979). As noted in the section on Evolution, this position is critical, as it implies that the plesiomorphic habitat in the Citri–Poulsonii clade may have been the plant phloem/sucking insect habitat.
Position of subgroups I-4 and I-6 as sister species in the Citri–Poulsonii clade.
Strain 277F, the sole member of subgroup I-4, and subgroup I-6, represented by Spiroplasma insolitum, were classified as sister species in all topologies. Although strain 277F was isolated from a tick, only a single strain is available, so the true habitat relationship of this spiroplasma is uncertain. S. insolitum has been isolated frequently from flowers and insects, including butterflies, in Maryland, USA (Hackett et al., 1984).
Monophyly of the chrysopicola–syrphidicola–TAAS-1 clade.
Spiroplasma chrysopicola, Spiroplasma syrphidicola and Spiroplasma sp. TAAS-1 (serogroup VIII) always formed a monophyletic clade. In addition, strains BARC 1357 and BARC 2649, candidate subgroups of serogroup VIII, also belong to this clade. However, the evolutionary distances among these five serogroup VIII strains are very small (16S rDNA similarity coefficients of 0·992–0·999) so these sequences failed to resolve the relationships among them (G. E. Gasparich, unpublished data), as would be expected (Stackebrandt & Goebel, 1994). Members of the chrysopicola–syrphidicola–TAAS-1 clade have been isolated almost exclusively from tabanid flies. Members of this group are, physically, the smallest spiroplasmas – passing quantitatively through 220 nm filters. This group also has a DNA G+C content (28–31 mol%) at the high end of the spiroplasma range and all members utilize arginine. More than 100 serogroup VIII isolates have been obtained from eastern Canada, south to Georgia, west to Idaho and Texas. The group is also known from Europe (Le Goff et al., 1991). Recently, more than a dozen isolates have been obtained, ten from Costa Rica and three from Australia (F. E. French, unpublished data). Three formal subgroups have been proposed for serogroup VIII strains on the basis of DNA–DNA hybridization, DNA G+C content and serological analyses (Gasparich et al., 1993). Type or representative strains are: EA-1 (S. syrphidicola, subgroup VIII-1); DF-1 (S. chrysopicola, subgroup VIII-2) and TAAS-1 (subgroup VIII-3). In addition, strain BARC 2649 was classified by Williamson et al. (1998) as an undesignated subgroup of serogroup VIII. However, as new isolates continue to accumulate, it has become more and more difficult to assign them to a subgroup (Stewart, 2001; F. E. French and R. F. Whitcomb, unpublished data).
Position of S. mirum as basal in the Citri–Chrysopicola–Mirum clade.
The phylogenetic position of the tick spiroplasma, S. mirum, as basal in this clade of insect specialists was highly supported. S. mirum is the most basal species of Spiroplasma that shows the modal spiroplasma helicity and motility.
Position of the Mycoides and Entomoplasmataceae clades as sister lineages.
High support is obtained for the apomorphic status of the Mycoides clade as a derivative of the Entomoplasmataceae clade. Hence, the taxon Entomoplasmataceae is technically paraphyletic. Mesoplasma lactucae was basal to the Entomoplasmataceae clade and Entomoplasma freundtii was basal to the Mycoides clade. As discussed below, divergence of the Mycoides and Entomoplasmataceae clades is in complete accord with the host affinities of the organisms. All species of the Mycoides clade are from ruminant animals and all entomoplasmas and mesoplasmas derive from the insect/plant surface habitat.
Polyphyly of Entomoplasma and Mesoplasma species.
All trees constructed to date, starting with the original distance trees of Weisburg et al. (1989) and including later trees of Johansson [summarized by Johansson & Pettersson (2002)] and Maniloff (2002) have indicated that species of the genera Mesoplasma and Entomoplasma do not form clades, but are intermixed, with small interspecies evolutionary distances. Our trees also fail to show distinct clusters for these genera. Furthermore, the similarity coefficients for 16S rDNA sequences are extremely high in some intergeneric comparisons [Entomoplasma melaleucae versus Mesoplasma entomophilum (0·992)].
The two serogroup XVI species studied are sisters.
Strains CB-1 (subgroup XVI-2) and Ar 1357 (subgroup XVI-3) were shown invariably to be sisters in the cladograms. Members of this serogroup are either isolates from flower surfaces or from cantharid beetles, which are frequent visitors of flowers. Several isolates in this clade have also been isolated from mosquitoes. The third subgroup, Spiroplasma cantharicola (XVI-1), was not sequenced in this study, but DNA–DNA hybridization data (Abalain-Colloc et al., 1993) indicate that this species is related closely to subgroups XVI-2 and XVI-3.
Sister taxa status of serogroup XVI spiroplasmas and Spiroplasma diminutum.
All trees showed S. diminutum and the CB-1–Ar 1357 (serogroup XVI) clade as sister lineages. However, S. diminutum is not related serologically to serogroup XVI spiroplasmas. S. diminutum and certain serogroup XVI strains have been isolated from mosquitoes.
Monophyly of the helicoides–gladiatoris–BARC 1901 clade.
This clade of three species (or putative species), all isolated from tabanid flies, was highly supported in trees generated by all four phylogenetic methods.
Monophyly of the corruscae–turonicum–litorale clade.
The monophyly of these three tabanid Spiroplasma species was also supported strongly. Spiroplasma corruscae was originally isolated from an overwintered lampyrid beetle, which has raised the issue of transmission and maintenance of the spiroplasma during winter (Hackett et al., 1996b). It is possible that beetles provide an overwintering reservoir for S. corruscae and that tabanids acquire the spiroplasma in the spring at feeding sites that are shared with the beetles (Gasparich et al., 1998).
Status of S. apis and Spiroplasma montanense as sister species.
These sister species formed a clade (apis–montanense clade) that was separated from other tabanid spiroplasma clades by relatively small evolutionary distances. Both species have been isolated from tabanids, but S. apis causes May Disease of honey bees in France and has been isolated from other insect host species.
Sister lineage status of serogroups XXI (strain W115) and XXX (strain BIUS-1) flower spiroplasmas.
These two spiroplasmas were sister species in most trees. Strains W115 and BIUS-1 were respectively isolated from flower surfaces in Oklahoma and Maryland, USA, so the actual (presumed) insect hosts of these species is unknown.
Position of S. lampyridicola and S. leptinotarsae as sisters.
These sister species are both beetle spiroplasmas. S. leptinotarsae is a specialist that is associated with the Colorado potato beetle (Clark, 1982) and S. lampyridicola is a specialist that is associated with firefly beetles (Coleoptera: Lampyridae), respectively (Clark et al., 1987; Hackett et al., 1992; Stevens et al., 1997). S. leptinotarsae is transmitted from beetle to beetle on leaf surfaces (Clark, 1984). The unusual morphology of this species is discussed in the Morphology section. S. lampyridicola has been isolated from both larvae and adult fireflies, but the transmission mechanism is unknown. Cells of this spiroplasma are unusual in that, when tested as an antigen, they react serologically in both deformation (DF) (Williamson et al., 1978) and metabolism inhibition (MI) (Williamson et al., 1979a) tests with the majority of spiroplasma antisera. Cells of S. leptinotarsae, when tested as antigens, also react with multiple spiroplasma antisera. Whether this property is synapomorphic for the leptinotarsae–lampyridicola clade is unknown.
Position of Spiroplasma sabaudiense (serogroup XIII), Spiroplasma alleghenense (serogroup XXVI) and Spiroplasma sp. strain TIUS-1 (serogroup XXVIII) as a monophyletic clade.
The sabaudiense–alleghenense–TIUS-1 clade, which appeared in most trees, makes less ecological sense than most of the other spiroplasma clades. It consists of serogroups XIII, XXVI and XXIX, which have been isolated from a diverse range of insects, including mosquitoes (postulated to feed on flowers) (Chastel & Humphery-Smith, 1991) and flower-visiting tiphiid wasps, as well as scorpionflies.
Phylogenetic classification and phenotype
Phylogenetic reconstructions have at least three practical uses. The first is to predict or anticipate the properties of organisms that have yet to be determined. The second is to shed light on the evolutionary process itself, which must be understood in order to make sense of biology. The third is guidance in construction of Linnaean classifications, so that the latter may reflect, as closely as possible, the natural relationships among taxa.
Previous authors, after examination of the relationship between mollicute phylogeny and phenotype, painted a rather bleak picture. Weisburg et al. (1989), whilst acknowledging that some correlations between phylogeny and phenotype were apparent ex post facto, felt that phenotype was not well predicted by phylogeny. Also, Dodge et al. (1998) felt that their phenetic tree showed that serological analysis of spiroplasmas was unreliable. Our trees and phenotypic data support neither of these views.
The study of Weisburg et al. (1989), which still stands as a defining paper on mollicute phylogeny, dealt with a relatively small sample of available organisms, which were chosen because they were perceived to represent a wide spectrum of mollicute biodiversity. Thus, they were chosen because they were likely to be members of diverse clades that were phylogenetically remote from one another and, hence, less likely to be phenotypically similar than species that were more closely related. Many microbial phenotypic properties are apomorphic traits that are only shared with closely related species and often converge despite phylogenetic disparity.
In the analysis of Dodge et al. (1998), weight was given to a sequence whose placement was suspiciously anomalous. The value of serology was discounted in large part, as one unsubstantiated sequence for strain TAUS-1 (Spiroplasma tabanidicola) appeared to be related closely to sequences of spiroplasmas in the chrysopicola–syrphidicola–TAAS-1 clade. If this relationship were genuine, it would not only fly in the face of strong serological evidence, but also would contradict the seemingly unerring congruence of spiroplasma serology with DNA–DNA reassociation data (Junca et al., 1980; Bové et al., 1983). Further, it would imply that extremely short and narrow organisms with a relatively high DNA G+C content and the ability to catabolize arginine aggressively were related extremely closely to spiroplasmas that were relatively long and wide with a considerably lower G+C content and with no capability to catabolize arginine. Thus, the rejection of serological evidence in the systematics of these organisms on the basis of an aberrant placement of a single sequence would require one to ignore a suite of robust characters that separate the chrysopicola–syrphidicola–TAAS-1 clade from the Apis clade sensu latu.
Morphology.
Morphology is a primary descriptor in most classifications of animals and plants; it is also important among microbes. Spiroplasma morphologies were not analysed herein as characters, because some distinctions are somewhat arbitrary. However, the consistency of morphology with cladistic topology is striking and must be discussed, even if it is not analysed rigorously. It is common for spiroplasma clades to share a common apomorphic – and in many cases, autapomorphic – set of morphological attributes. The following discussion refers to morphology as viewed microscopically during growth of the organisms in culture. The following morphotypes can be recognized.
(i) In the Citri–Poulsonii (serogroup I–serogroup II) clade, organisms are seen, after elongation of short forms during very early exponential phase, as long helices (10 or more turns) (Garnier et al., 1984). This morphology remains stable throughout the exponential and early stationary growth phases.
(ii) In the chrysopicola–syrphidicola–TAAS-1 (serogroup VIII) clade, organisms grow very fast to titres as high as 1011. Cells are short and narrow and can pass quantitatively through 220 nm filters. This morphology and the high titre to which cultures grow are diagnostic and, in routine screening, it is often possible to test candidate serogroup VIII strains against the bank of serogroup VIII antisera without preliminary screening against antiserum to organisms of other spiroplasma groups (Whitcomb & Hackett, 1996). Also, the vast majority of the serogroup VIII isolates (>100) have short cells, although in a few strains, longer cells have been observed (F. E. French, unpublished data).
(iii) Most spiroplasmas of the Apis clade sensu strictu tend to resemble those of the Citri–Poulsonii clade in their helicity, but are usually shorter. This morphology is presumably plesiomorphic in the clade and perhaps even in the genus (see Evolution). Some of the members of the clade (e.g. Spiroplasma helicoides) exhibit robust helical morphology that remains uncompromised throughout the exponential and early stationary growth phases.
(iv) Some species of the Apis clade sensu latu show autapomorphic modifications of simple helicity. In S. leptinotarsae, the cells are funnel-shaped, in that they are broad at one end but narrow to a tip at the other, with increasing amplitude from the tip to the posterior end of the cell when observed in early passages. In vitro, these cells exhibit extremely rapid translational motility. In vivo, in the gut of the host (the Colorado potato beetle), the helical cells flatten into a single plane, so that transverse sections show coiled tubules. Under dark-field microscopy, these structures appear to be coin-shaped. When these flattened coils are released into fresh culture medium, the cells regain their funnel (or spring-like) morphology and are again capable of rapid translational motility. As spectacular as this morphological adaptation is, it is ephemeral. In primary culture, S. leptinotarsae grows very poorly in artificial media, but cultures can be established in co-cultures with insect cells. After a relatively short number of passages in broth media, the ability to form helical cells with continuously varying helical amplitudes disappears and the cells then revert to simple helicity [see (iii) above].
(v) Some members of the BIUS-1–Floricola–Diabroticae–Monobiae clade of the Apis clade sensu strictu have short cells that may be helical at only certain phases of the growth cycle. Spiroplasma floricola (serogroup III) grows very rapidly in the exponential phase as short helices, but as the medium acidifies and is deprived of nutrients by the rapidly growing organisms, the cells become non-helical. For a short period, these non-helical cells, if they are reinoculated into fresh medium, are able to grow and helical morphology reappears (Whitcomb & Coan, 1980). In serogroup XXX (strain BIUS-1), the morphological cycle in vitro is somewhat similar, although helices appear for a short time only in mid-exponential phase. Cells of either S. floricola or strain BIUS-1, if examined superficially and viewed only in stationary phase, might appear to be non-helical and could conceivably be mistaken for entomoplasmas. In fact, when strain BNR1 of S. floricola was first isolated and passed in the laboratory, it was mistaken for a non-helical organism, but when examined in exponential phase, it was found to exhibit swarms of short helical cells (D. L. Williamson and R. F. Whitcomb, unpublished data). S. diminutum, although allied closely to the CB-1–Ar 1357 clade, in which cellular morphology is unremarkable, grows prolifically as extremely short cells, with perhaps no more than two or three turns per cell. Spiroplasma culicicola, a species of the Apis clade sensu strictu that has no close sister species, also grows as very short cells with few turns. As all spiroplasmas that have very short cells have rapid growth rates, it is possible that the tendency for short cells may be explained by the rapidity of cellular growth.
(vi) Strain TIUS-1 (serogroup XXIX) exhibits such poor helicity in culture that it could easily be mistaken for an entomoplasma. Helical forms are rarely seen in culture. This species is basal to the apis clade sensu latu and, evolutionarily, may represent a transitional morphological stage in the loss of helicity.
Three spiroplasma species show autapomorphic adaptations that involve a departure from simple generic modal helicity: (vii) S. ixodetis has cells that are extremely variable in culture. Some of them are not helical and some have extremely tightly coiled helices – so much so that under dark-field microscopy, they may not appear helical at all. (viii) Serogroup XV spiroplasmas, the 16S rDNA of which was not sequenced, lose helicity and motility and become stiff and rod-like as the growth medium acidifies in late exponential phase. However, if the pH of the medium is raised to neutral, helicity and motility return. (ix) Spiroplasma platyhelix, the 16S rDNA of which was also not sequenced, displays a low-pitch helical morphology and, along with serogroup XV spiroplasmas, exhibits a unique motility, in which tightly coiled segments move in waves along the length of the filament.
In summary, there are many modifications of the simple helicity that is exhibited by species of the Apis clade sensu strictu. In many cases, these modifications are synapomorphic for clades of closely related species. However, several are autapomorphies and thus do not provide phylogenetic information.
Serology.
Careful examination of the totality of spiroplasma data indicates a strong correlation between serology and molecular phylogeny. Serology of spiroplasmas has been studied in great detail [summarized by Williamson et al. (1998)] by the spiroplasma DF (Williamson et al., 1978) and MI (Williamson et al., 1979a) tests, and by growth inhibition (Whitcomb et al., 1982) for serogroups I–XI. Serogroup classification of spiroplasmas (Junca et al., 1980; Whitcomb et al., 1987) is based to a considerable degree on serological data. Minimum standards have been established for the serological evidence that is required to establish spiroplasma serogroups (Whitcomb et al., 1987). In these criteria, it was recognized that reciprocal cross-reactivity was required to establish serogroups, as one-way crosses, which can be of considerable magnitude in some cases, are not unusual in mollicute serology. Every combination of Spiroplasma antigen/antibody reaction has been tested by DF and MI serology in the course of defining the 34 serogroups and 14 subgroups of spiroplasmas (Williamson et al., 1998). Every one of the >5000 reciprocal cross-reactions that are observed in these tests is consistent with the topologies in our phylogenetic reconstructions. Invariably, reciprocal serological cross-reactions are predictive of close phylogenetic affinity. Pairs of species that shared reciprocal serological cross-reactivities were invariably separated by very small evolutionary distances and were members of the same clade. For example, sequences that represented the seven sequenced subgroups of serogroup I (Citri clade) had sequence similarities of no less than 0·986. The three subgroups of the chrysopicola–syrphidicola–TAAS-1 clade (serogroup VIII) reported here, and an assemblage of other serogroup VIII strains that were not included in our reconstruction, had much higher similarity values (>0·99) than serogroup I spiroplasmas, and all members of both serogroups I and VIII shared a high degree of intragroup serological reactivity (Williamson et al., 1998). When the 16S rDNA similarity between species is 97 % or more, Stackebrandt & Goebel (1994) recommend that DNA–DNA reassociation studies should be performed to assist in the decision concerning possible species identity of the paired strains. In the case of spiroplasma subgroups, DNA–DNA reassociation studies have already been correlated with serological re
