- Research article
- Open Access
One nutritional symbiosis begat another: Phylogenetic evidence that the ant tribe Camponotini acquired Blochmanniaby tending sap-feeding insects
BMC Evolutionary Biologyvolume 9, Article number: 292 (2009)
Bacterial endosymbiosis has a recurring significance in the evolution of insects. An estimated 10-20% of insect species depend on bacterial associates for their nutrition and reproductive viability. Members of the ant tribe Camponotini, the focus of this study, possess a stable, intracellular bacterial mutualist. The bacterium, Blochmannia, was first discovered in Camponotus and has since been documented in a distinct subgenus of Camponotus, Colobopsis, and in the related genus Polyrhachis. However, the distribution of Blochmannia throughout the Camponotini remains in question. Documenting the true host range of this bacterial mutualist is an important first step toward understanding the various ecological contexts in which it has evolved, and toward identifying its closest bacterial relatives. In this study, we performed a molecular screen, based on PCR amplification of 16S rDNA, to identify bacterial associates of diverse Camponotini species.
Phylogenetic analyses of 16S rDNA gave four important insights: (i) Blochmannia occurs in a broad range of Camponotini genera including Calomyrmex, Echinopla, and Opisthopsis, and did not occur in outgroups related to this tribe (e.g., Notostigma). This suggests that the mutualism originated in the ancestor of the tribe Camponotini. (ii) The known bacteriocyte-associated symbionts of ants, in Formica, Plagiolepis, and the Camponotini, arose independently. (iii) Blochmannia is nestled within a diverse clade of endosymbionts of sap-feeding hemipteran insects, such as mealybugs, aphids, and psyllids. In our analyses, a group of secondary symbionts of mealybugs are the closest relatives of Blochmannia. (iv) Blochmannia has cospeciated with its known hosts, although deep divergences at the genus level remain uncertain.
The Blochmannia mutualism occurs in Calomyrmex, Echinopla, and Opisthopsis, in addition to Camponotus, and probably originated in the ancestral lineage leading to the Camponotini. This significant expansion of its known host range implies that the mutualism is more ancient and ecologically diverse than previously documented. Blochmannia is most closely related to endosymbionts of sap-feeding hemipterans, which ants tend for their carbohydrate-rich honeydew. Based on phylogenetic results, we propose Camponotini might have originally acquired this bacterial mutualist through a nutritional symbiosis with other insects.
Bacteria play important roles in the success and diversification of many animal groups, and insects are especially prone to establishing long-term, mutualistic endosymbioses. An estimated 10-20% of insect species, in several taxonomic orders, depend on intracellular bacterial mutualists for their viability and reproduction . These obligate associates, called 'primary' (P-) endosymbionts because they are required for host survival and fecundity, often synthesize key nutrients that are lacking in the hosts' unbalanced diet (e.g., plant sap or vertebrate blood) [2–4]. In all cases, the bacteria live within specialized host cells called bacteriocytes and undergo maternal transmission to developing eggs or embryos. Consistent with this stable transmission through host lineages, the phylogenies of P-endosymbionts match those of their insect hosts (e.g., [5–9]). This phylogenetic congruence points to host-symbiont cospeciation, which can be traced back to a single, often ancient, infection event in each host group.
Among Hymenoptera, only members of the Formicidae (ants) are known to possess bacteriocyte-associated endosymbionts. And within ants, despite the group's wide variety of symbioses with microbes , only three known examples of such intracellular mutualisms exist. These three cases occur in subfamily Formicinae: Formica, Plagiolepis, and all members of the tribe Camponotini screened to date [11–14]. Although the symbioses are similar in many respects, phylogenetic analysis of bacterial 16S rDNA implies the three bacteriocyte-associated symbioses in ants evolved independently .
The symbionts of Formica and Plagiolepis have not been studied in depth, but have been documented within bacteriocytes on either side of the midgut epithelium. In Formica, symbionts are found within queen ovarioles and in developing brood [12, 13, 16]. The bacteria are considered maternally transmitted, but their occurrence is erratic among Formica species and can vary within species depending on nutritional status .
Endosymbionts in the tribe Camponotini, the focus of this study, are the best-studied bacterial mutualists in ants. The bacterium, Blochmannia, was first discovered in Camponotus  the second largest ant genus with ~1,000 described species worldwide . The bacteria occur within ant bacteriocytes, which are intercalated among midgut epithelial cells, as well as queen and worker ovaries where symbionts infect the developing oocyte and are closely integrated with host development [13, 14, 18, 19]. Although Blochmannia densities apparently decline in older lab-reared workers , and the bacteria can be eliminated from workers treated with antibiotics , we have never failed to find Blochmannia in any Camponotus worker collected in the field (unpublished data), which total many hundred samples across diverse species. The retention of many nutrient biosynthetic functions within the Blochmannia genomes implies this symbiosis plays a nutritional role [21, 22]. Likewise, experimental diet treatments indicate that Blochmannia provides nutrients to the host, including amino acids .
Among insects possessing long-term intracellular mutualists, which include several sap- and blood- feeding species, many Camponotus spp. stand out as true omnivores that scavenge other insects as part of their complex diet. Benefits conferred by Blochmannia might be critical during specific periods in host individual and colony development [23–25] (unpublished data). In addition, perhaps the symbiosis originated in an ant lineage that fed on a nutritionally unbalanced diet, such as extant arboreal Camponotus and Polyrhachis that obtain most nitrogen from plant and insect exudates [26–28]. Like many other ants, these arboreal species tend sap-feeding hemipterans, such as mealybugs, aphids, and psyllids, for their carbohydrate-rich excrement, or 'honeydew' .
Like other P-endosymbionts, Blochmannia has codiverged with its known ant hosts, reflecting an evolutionarily stable association. Several phylogenetic analyses of symbiont and host genes have demonstrated cospeciation within Camponotus [19, 30, 31]. In addition, Sameshima et al.  showed that Blochmannia occurs in Polyrhachis and Colobopsis (currently a subgenus of Camponotus  but probably a separate lineage ), suggesting the symbiosis originated before the divergence of these three taxa, on the order of 40 MYA.
However, due to limited sampling, the actual distribution of Blochmannia throughout the Camponotini is unknown. Currently, the formal description of Camponotini includes eight extant genera: Camponotus (including subgenus Colobopsis), Polyrhachis, Calomyrmex, Echinopla, Opisthopsis, Forelophilus, Overbeckia, and Phasmomyrmex . To date, Blochmannia has been detected in Camponotus (including Colobopsis) and Polyrhachis as noted above, and Stoll et al.  detected a Blochmannia-like bacterium in association with an Echinopla species, based on a 367-bp region of the bacterial 16S rDNA gene. Because several Camponotini genera remain unsampled, the occurrence of Blochmannia within the tribe is uncertain.
For context, Dasch  reported that eight genera of Camponotini possess intracellular bacterial symbionts, and this number (eight) has been cited in subsequent work. However, the genera cited by Dasch have since been reclassified as subgenera of Camponotus (i.e., Tanaemyrmex, Myrmentoma, Myrmosericus, Myrmobrachys, Myrmocladoecus, Myrmothrix, and Colobopsis) .
In this study, our goals are (i) to better understand the actual host range of Blochmannia, and (ii) to identify the closest bacterial relatives and likely source of this mutualism. To this end, we screened for Blochmannia in a broader representation of Camponotini genera, including Calomyrmex, Opisthopsis, and Echinopla. We also sampled a representative of Notostigma, a genus formerly considered part of the Camponotini but recently placed in its own tribe . To screen for symbionts, we amplified and sequenced a region of the 16S rDNA gene. Our results indicate that Blochmannia is more widespread within the tribe than previously documented and strongly suggest a single, ancient origin for this endosymbiosis. In addition, a close relationship with secondary endosymbionts of mealybugs suggests a potential route for the acquisition of Blochmannia. Specifically, the tribe Camponotini may have acquired its bacterial partner by tending honeydew-producing hemipterans.
Results and discussion
PCR screen for Blochmannia
We screened 53 representatives of the Camponotini for bacterial associates (Table 1). Specimens were identified minimally to subgenus (when applicable) and in most cases to species. The sample includes 42 Camponotus isolates, ten of which belong to the subgenus Colobopsis, as well as four Polyrhachis, two Calomyrmex, four Opisthopsis, and one Echinopla isolate. We also screened a representative of Notostigma, which was recently removed from the tribe .
Our screen involved PCR with primers specific to the Blochmannia 16S rDNA gene. Direct sequencing of the resulting PCR products yielded high quality data for nearly all isolates. (The few exceptions requiring alternative methods are noted as a footnote in Table 1 and detailed in the Methods section.) The Genbank accession numbers for the 52 new sequences obtained in this study are listed in boldface in Table 2. All 16S rDNA sequences from the Camponotini isolates proved to be a close match to known Blochmannia strains, as detailed in the database comparisons and phylogenetic analyses below. This indicates that our specific primers amplify a wide range of Blochmannia lineages across the tribe, but generally did not amplify other bacteria that may associate with ants in various ways (e.g., on the body surface or in the gut lumen).
The 16S rDNA amplified from an outgroup related to this tribe (Notostigma) did not match Blochmannia. In addition, previous work has screened other members of the subfamily Formicinae for bacterial symbionts and failed to detect Blochmannia. The bacterial symbiont in Formica and Plagiolepis is not Blochmannia, as we have verified in our phylogenetic analysis here. Moreover, we were unable to amplify Blochmannia from a specimen of Oecophylla smaradina (unpublished data).
We compared new sequences to Genbank and to the Ribosomal Database Project (RDP, release 10.5) . These searches are based on sequence similarity and provide a very rough idea of taxonomic affiliation. Most new bacterial 16S rDNA sequences from the Camponotini most closely matched published Blochmannia in both databases. (Please see additional file 1 for a list of top matches.) These include isolates of Calomyrmex and Opisthopsis, where Blochmannia has not been detected before, and Echinopla, where a short, Blochmannia-like sequence was noted once previously . For an additional four sequences, Blochmannia was the top match in one of the two databases, and nearly the top in the other database. Four members of the Colobopsis subgenus most closely match other members of the Enterobacteriaceae. Nonetheless, subsequent phylogenetic analysis (below) groups these Colobopsis associates with Blochmannia with very high confidence, illustrating the approximate nature of the similarity-based database searches. The bacterial associate of Notostigma carazzii did not match Blochmannia in either database.
Gamma-Proteobacteria analysis: Camponotini isolates group with known Blochmanniastrains
To test for the monophyly of presumed Blochmannia isolates and to position this group within the gamma-Proteobacteria, we estimated the phylogeny of 16S rDNA using Bayesian and maximum likelihood (ML) approaches. Taxa included candidates for the closest relatives of Blochmannia, based on matches in databases such as Genbank, the RDP , and ARB . Bayesian posterior probabilities were higher than ML bootstrap values, as expected based on previous studies . As a consequence, the Bayesian consensus tree (Figure 1) showed higher resolution than did the ML bootstrap consensus tree. (The ML tree of the gamma-Proteobacteria dataset is presented as additional file 2.) Both methods support the monophyly of known Blochmannia isolates and newly sampled Camponotini associates (Bayesian posterior probability of 1.00 and bootstrap support of 78%). The phylogenies also illustrate that Plagiolepis and Formica endosymbionts represent independent lineages within the gamma-Proteobacteria, as suggested in an earlier study based on a smaller taxon sample . Interestingly, the Notostigma isolate is closely related to Blochmannia but clearly falls outside of that clade. This absence of Blochmannia is consistent with the removal of Notostigma from the Camponotini and assignment to its own tribe [32, 33]. Whether the Notostigma isolate represents a symbiosis or casual bacterial associate is uncertain.
These results demonstrate for the first time that Calomyrmex, Echinopla, and Opisthopsis possess bacterial associates that are part of the same clade as known Blochmannia strains, consistent with a single origin of endosymbiosis in the Camponotini. Although additional Camponotini taxa remain to be screened from more rarely collected genera, we predict that Blochmannia will be pervasive throughout the tribe. The three Camponotini genera not included in our study (Forelophilus, Overbeckia, and Phasmomyrmex) together comprise only six living species and represent a comparatively small component of the overall diversity within the tribe. Indeed, current taxonomy suggests that Forelophilus and Overbeckia may be synonymous with Camponotus .
This discovery of Blochmannia in diverse Camponotini genera helps to refine the questions we ask about the evolution of this association. For example, it remains mysterious why the mutualism became established in this group, but apparently not in other ant clades. What is distinct about Camponotini? This remains an open question. Studying the full range of hosts will highlight conspicuous exceptions, where the life history or physiology of ants may impact the symbiosis. For example, while Blochmannia may be involved in claustral founding of new colonies , Polyrhachis shows instances of semi-claustral founding, the only known example in a formicine ant . In addition, in a tribe in which the metapleural gland is often missing , any role of symbionts in pathogen defense may be particularly important (D. E. Wheeler and J. F. A. Traniello, pers. comm.). This gland is a structure unique to ants whose external secretions have been hypothesized to serve antiseptic functions, although other functions have been proposed that involve chemical defense, recognition odor, and territorial marking (U. Mueller, pers. comm.). The metapleural gland is present in workers from the vast majority of ant genera, including Calomyrmex, Opisthopsis and Echinopla, but is absent from Polyrhachis and most species of Camponotus. It is known to be present in the workers of at least 13 Camponotus species including C. gigas , C. thadeus , C. sericeus , and ten species of Camponotus (Myrmonesites) from Madagascar (B. Fisher and U. Mueller, pers. comm.). These exceptions to the rule offer natural experiments to explore the influence of host biology on symbiont persistence and functions.
Our phylogenetic analysis of gamma-Proteobacteria position Blochmannia within a large, diverse group of insect endosymbionts (Figure 1). This group includes many secondary symbionts of sap-feeding mealybugs, psyllids, and aphids. In our analyses, a group of secondary symbionts of mealybugs are the closest relatives of Blochmannia. This group represents one of at least four distinct acquisitions of gamma-Proteobacterial symbionts by mealybugs . The most recent common ancestor (MRCA) of Blochmannia is reconstructed with very high likelihood to be endosymbiotic, with a proportional likelihood of 0.9999 under both Mk1 model and AsymmMk model (expressed as the proportion of the total probability of 1.0). Similarly, the ancestral node representing the MRCA of Blochmannia and the four closely-related mealybug endosymbionts is inferred to have been endosymbiotic (proportional likelihood 0.9999 under both models).
Camponotini is one of the major ant groups that commonly tend mealybugs, aphids, and other hemipterans for their carbohydrate-rich honeydew [29, 44] and may have acquired Blochmannia through this route. Ants and hemipterans have a long history of association, with fossil evidence for this relationship occurring in Baltic (~44 MYA) and Dominican (15-20 MYA) amber deposits [45, 46]. Ants often have been observed to scavenge or kill the hemipterans that they tend and transport the bodies to their nests, presumably to help feed their colonies [47–50]. Repeated consumption by ant larvae, the only stage that can ingest particulate matter, may have selected for sequestration of a hemipteran symbiont. Feeding interactions have been implicated in the horizontal transmission of other endosymbiotic bacteria in insects including Wolbachia  and Rickettsia  although definitive experimental evidence for the successful establishment of an endosymbiont via this pathway remains lacking .
We performed a more focused phylogenetic analysis that included new bacterial sequences obtained here, as well as 28 published Blochmannia sequences. We analyzed the data with and without mealybug endosymbionts as outgroups. Using Bayesian approaches, the unrooted (Figure 2) and rooted (Figure 3) phylogenies resolved four subgroups within Blochmannia that match genus-level distinctions among their ant hosts: Polyrhachis, Camponotus subgenus Colobopsis (called Colobopsis here, for brevity), Opisthopsis, and a fourth group composed of Camponotus, Calomyrmex, and Echinopla. In rooted Bayesian trees, each subgroup was resolved with high confidence (posterior probabilities of 0.97-1.00). The ML analyses gave similar results but with lower confidence. Unrooted and rooted ML trees of Blochmannia are presented in additional files 3 and 4, respectively.
In an analysis of ant relationships, Brady et al.  found strong support for the grouping of Polyrhachis with Camponotus + Calomyrmex, the relationship reflected in schematic tree A (Figure 4). While the rooting of the Camponotini was uncertain, it may occur along the Opisthopsis lineage . By contrast, our 16S rDNA dataset does not distinguish relationships among major subgroups within Blochmannia. For example, we considered the three possible unrooted trees for these four lineages (trees A-C; Figure 4). The Shimodaira-Hasegawa (SH) test indicated the data cannot distinguish among the three possible unrooted topologies (Table 3). To test whether our dataset can resolve the root position, we also used the SH test to evaluate the relative support for 15 trees that represent the alternative rootings of trees A-C. The data do not reject any of the 15 alternative trees, and thus cannot reject any hypotheses for relationships among major Blochmannia lineages (Table 4). Additional data from rDNA genes or other Blochmannia loci will be necessary to resolve ancient divergences within this mutualism.
Agreement of bacterial and ant relationships
Agreement of the Blochmannia and host phylogeny is expected, given that several previous studies demonstrate host-symbiont cospeciation in this mutualism [15, 19, 30, 31]. While a formal cospeciation analysis is beyond the scope of this study, the Blochmannia phylogeny suggests host-symbiont congruence at taxonomic levels previously unexamined. At the tribe level, we detected Blochmannia only in association with Camponotini, and not, for example, in Notostigma. At the genus level, the major groups within Blochmannia match distinctions among the ant hosts. In a comprehensive phylogenetic analysis of ants, Brady et al.  confirmed that Camponotus is a polyphyletic assemblage. Colobopsis, although still formally considered a subgenus of Camponotus , constitutes a distinct group that is separated from Camponotus by intervening genera. Here, we found an identical pattern in the 16S rDNA of associated Blochmannia. That is, Blochmannia from ants in the subgenus Colobopsis formed a group that is distinct from other Camponotus. In addition, the close relationship of bacteria from Calomyrmex and Camponotus matches the affiliation of these ant genera [32, 53].
At shallower taxonomic levels, many well-resolved species-level relationships among Blochmannia agree with known or predicted host relationships. Congruence has already been documented in previous cospeciation studies of many species included here [15, 19, 30, 31]. In addition, our new data illustrate further examples of recently-diverged Blochmannia strains from ant hosts that are close relatives . Examples include C. quercicola and C. castaneus, C. rufipes and C. floridanus, C. hyatti and C. sayi, C. balzani and C. silvicola, and the group containing C. sansabeanus, C. maritimus, C. semitestaceus and C. vicinus. Although Camponotus subgenera are often polyphyletic and not always good predictors of host relationships [31, 54], notably the Blochmannia of C. claviscapus and C. occultus, both members of the subgenus Pseudocolobopsis, are sister taxa. Furthermore, Blochmannia sampled from the same Camponotus species but from different geographic areas are always closely related, usually as a strongly supported monophyletic unit.
Caveats: Strengths and limitations of molecular data
The molecular approach presented here has clear advantages. This approach let us screen for symbionts using one or few host specimens, infer relationships among the bacteria detected, and place newly-discovered isolates within a broader phylogenetic context. However, based on DNA sequence data alone, it is difficult to make strong conclusions about the type of association between Blochmannia and the additional genera sampled here. For instance, it is impossible to say for certain that the bacteria form stable, obligate, intracellular relationships in each host genus - i.e., that the type of mutualism well-characterized in Camponotus also occurs in Opisthopsis, Calomyrmex, and Echinopla. A complete characterization of Blochmannia in these genera is beyond the scope of the current study, but ideally will include additional representatives of each genus to test for symbiont persistence and stability, and will be coupled with ultrastructural work to document the location of bacteria within host tissues or cells. Such characterizations will be a fruitful area for future research.
Despite these caveats, the demonstrated specificity of Blochmannia makes its detection in additional host genera compelling. When a symbiont group is already well described, the presence of that symbiont is frequently based on molecular data alone, even when it is detected in new hosts. In addition, our results support the hypothesis that Blochmannia is monophyletic and evolved more than 40 MYA from an ancient, diverse clade consisting of intracellular endosymbionts. This pattern lends further support to the view that newly-discovered Blochmannia lineages within the Camponotini also live within cells. Moreover, the general agreement of host and symbiont phylogenies bolsters the view that these bacterial associates form meaningful, stable associations with their hosts.
Our molecular screen and phylogenetic analysis of ant-associated bacteria revealed four important results.
(i) We demonstrated for the first time that Calomyrmex and Opisthopsis possess bacterial associates that are members of the same clade as known Blochmannia strains. Furthermore, we confirmed a similar finding for Echinopla. These results significantly expand the range of known hosts of this symbiont and suggest the mutualism is more ancient and diverse than previously documented. Consistent with the recent removal of Notostigma from the Camponotini , we were unable to detect Blochmannia in a specimen of this genus. Although additional Camponotini genera remain to be screened (Forelophilus, Overbeckia, and Phasmomyrmex), we predict that Blochmannia will be pervasive throughout the Camponotini.
(ii) The three known bacteriocyte-associated symbionts in ants evolved independently in Formica, Plagiolepis, and the Camponotini. The three symbionts constitute distinct lineages within the gamma-Proteobacteria.
(iii) Blochmannia is positioned within a diverse, strictly endosymbiotic bacterial group and is reconstructed with very high likelihood to have originated from an endosymbiotic ancestor. This larger bacterial group includes endosymbionts of mealybugs, psyllids, lice, weevils, Plagiolepis, and tsetse flies. Our analysis suggests that a group of secondary symbionts of mealybugs constitute the closest relatives to Blochmannia, based on available sequence data, and suggest a possible origin for the ant mutualism. Unlike primary symbionts such as Blochmannia, secondary symbionts form dynamic associations and are known to transfer among distinct insect species and, occasionally, among insect superfamilies . Camponotini might have acquired the bacteria by tending mealybugs or other sap-feeding hemipterans for their carbohydrate-rich honeydew.
(iv) We found that the phylogeny of Blochmannia agrees with known relationships among Camponotini hosts at taxonomic levels previously unexamined. We detected four robust groups within Blochmannia: isolates from Polyrhachis, Colobopsis, Opisthopsis, and a fourth group composed of Camponotus (excluding Colobopsis), Calomyrmex, and Echinopla. Our data further support Colobopsis as a lineage distinct from remaining Camponotus.
Exciting areas for future research include testing the prediction that Blochmannia are pervasive throughout the Camponotini, by characterizing microbial associates of unsampled genera and by documenting the tissue- and cellular location of bacteria detected. Additional phylogenetic analysis, ideally based on expanded molecular and morphological datasets, will shed light on the deep divergences among Camponotini genera and whether hosts and symbionts codiverged in the earliest stages of this ancient mutualism.
For each ant sample, genomic DNA was prepared from whole ants or from only the gaster (QIAGEN DNeasy kit). In nearly all cases, single worker ants were used. Rarely, two or more ant individuals from the same colony were combined for a given DNA prep, particularly when ants were small. For samples that gave very low yields of DNA, we amplified the DNA using GenomiPhi (GE Life sciences). As a positive control for DNA quality and quantity, we PCR-amplified a 1.3-kb region of cytochrome oxidase I and II (COI/II). Nearly all samples gave a visible COI/II product. These PCR products generally were not sequenced for this study, but rather served as an indicator of high-quality DNA.
We used primers designed to match Blochmannia 16S rDNA to screen for this endosymbiont across ant specimens. These specific primers (Bloch16S-462F 5-AAACCCTGATGCAGCTATACCGTGTGTG-3', and Bloch16S-1299R 5'-CCATTGTAGCACGTTTGTAGCCCTACTCA-3') produce a PCR product of ~840 bp . 16S rDNA PCR reactions were repeated in large-scale (50 uL) format and the products purified (Promega Wizard PCR purification kit). PCR products were sequenced directly in both directions on an ABI 3730×l automated sequencer using Big Dye v3.0 (Applied Biosystems). This approach gave high quality DNA sequences for nearly all samples. All sequences were assembled and edited using PHRED, PHRAP and CONSED. DNA assemblies were checked by eye and any ambiguous base calls were re-sequenced or changed to N. Most sequences obtained were ~750-800 bp, slightly shorter than the PCR product itself.
Three isolates required the use of alternative primer pairs and/or cloning, in order to obtain high quality data. These few exceptions are noted in the footnote in Table 1 and detailed here. First, for Opisthopsis PG01, the PCR product generated from Blochmannia-specific primers was cloned (Invitrogen TOPO TA kit). Six clones were selected for sequencing with M13F and M13R primers and showed considerable variation. Two sequences had a highest BLASTn hits to known Blochmannia isolates, and were included in the phylogenetic analyses presented here; the remaining four sequences showed closest BLASTn hits to bacteria from soils. While the latter four sequences might represent meaningful bacterial associates, we favor the conservative interpretation that they reflect environmental contaminants. Notably, we found significant variation between the two Blochmannia clones. Because two Opisthopsis PG01 individuals were pooled for the gDNA preparation (one worker and one pupa), it is possible that this reflects standing genetic variation between host individuals, a subject for future investigation.
Two samples (Notostigma carazzii 226 and Opisthopsis haddoni 244) did not give detectable PCR products with the Blochmannia-specific primers. For Notostigma carazzii 226, we used the eubacterial 16S primers SL and SR  to amplify a 1,202 bp fragment that was subsequently cloned and sequenced with M13F and M13R primers. Among the eight clones sequenced, only minor variation was observed, likely reflecting cloning artifacts. The single sequence used in our phylogenetic analyses was selected for its length and high quality. Finally, for Opisthopsis haddoni 244, we generated two overlapping PCR products using the primers 16S_10F (5'-AGTTTGATCATGGCTCAGATTG-3') + 23S_480R (5'-CACGGTACTGGTTCACTATCGGTC-3'), and 16S_777F (5'-AGCAAACAGGATTAGATACCC-3') + SR; these were sequenced directly to generate a 1,432-bp fragment.
Among the bacterial 16S rDNA sequences from Camponotini, all matched known Blochmannia, based on database comparisons and/or phylogeny reconstructions (see Results). Only three specimens, Camponotus latangulus 236, Campontous sp. 241, and Camponotus dimorphus 234, failed to work in our molecular screen. These isolates did not generate PCR for host COI/II, our positive control for DNA quality. In addition, a C. vitreus worker that we initially sampled apparently possessed two bacterial associates: Blochmannia, as well as an isolate that matched the soil bacterium Chromohalobacter. To test whether the latter was a persistent bacterial associate or contaminant, we screened a second C. vitreus worker collected from the same colony. The second worker (C. vitreus 231ii) possessed only Blochmannia and is included in the phylogenetic analyses here.
New bacterial 16S rDNA sequences were compared to all sequences in two comprehensive databases. First, we identified closest matches in the Ribosomal Database Project (RDP, release 10.5; http://rdp.cme.msu.edu/) using the Sequence Match utility . This allowed us to compare our new sequences to 671,510 high quality Bacterial SSU sequences. In addition, we used BLASTn for a more general comparison of our new sequences to all sequences in Genbank. These searches are based on DNA sequence similarity and provide a very rough prediction of taxonomic affiliation. The advantage of this approach is the ability to search numerous 16S rDNA sequences very rapidly. The disadvantage is that the similarity results are not always good indicators of phylogenetic relationships. Nonetheless, these broad comparisons helped us to identify the closest non-Blochmannia outgroups to include in phylogenetic analysis.
Taxa were selected according to the goals of two phylogenetic analyses. (i) Gamma-Proteobacteria analysis. First, we explored the broad placement of Blochmannia 16S rDNA sequences, including new sequences obtained here, within the gamma-Proteobacteria. This analysis (a) tested whether newly discovered bacterial associates of Camponotini form a monophyletic group that includes known Blochmannia, (b) identified the closest relative to Blochmannia, and (c) tested whether Plagiolepis and Formica endosymbionts group with Blochmannia or, alternatively, represent distinct symbiont acquisitions in ants. The full, 99-taxon dataset for this gamma-Proteobacteria analysis included representatives of new and published Blochmannia sequences, endosymbionts of other insects, non-endosymbiotic Enterobacteriaceae, and divergent outgroup taxa. To identify the immediate relatives of Blochmannia, we included its closest matches in databases such as Genbank, the Ribosomal Database Project , and ARB . We deliberately excluded the exceptionally AT-rich primary endosymbionts (e.g., Wigglesworthia, Buchnera, Carsonella) to reduce biases resulting from extreme base compositions, especially the artefactual grouping of AT-rich sequences [56, 57]. In addition, initial analyses showed their phylogenetic positions were unstable across the tree.
(ii) Blochmannia-focused analysis. Second, we explored relationships among Blochmannia in a targeted analysis of 78 bacterial isolates from Camponotini, including the previously unsampled host genera, Calomyrmex, Echinopla, and Opisthopsis. We analyzed Blochmannia sequences with and without sequences of four mealybug endosymbionts that proved to be the closest outgroups in the gamma-Proteobacteria analysis above.
16S rDNA sequences were aligned using SILVA INcremental Aligner, or SINA http://www.arb-silva.de/aligner/. The alignment was examined carefully and any regions considered ambiguous were excluded from analysis. This 16S rDNA alignment, with ambiguous regions marked, is provided in nexus format as additional file 5. After excluding ambiguous regions, alignment lengths were 1,510 sites (599 of which were variable) for the gamma-Proteobacteria analysis and 1,547 sites (429 variable) for the Blochmannia-focused analysis. At 700-800 bp, many Blochmannia sequences were considerably shorter than the full alignment length and thus included stretches of missing data. However, we chose to analyze a longer region to retain as much information as possible.
For both the gamma-Proteobacteria and Blochmannia-focused analyses, datasets were analyzed using Modeltest 3.7 to determine the appropriate model of sequence evolution . Based on the hierarchical likelihood ratio test (hLRT) results, the most appropriate model for both datasets is the General Time Reversible model with invariant sites and rate variation among sites (GTR+I+G).
We performed phylogenetic analysis with Bayesian and maximum likelihood (ML) methods. Bayesian analysis. Using the parallel version of MrBayes v3.1.2, we implemented a GTR model with unequal base frequencies, portion of invariant sites estimated from the data, rate variation among sites according to the gamma distribution, and noninformative priors. We simultaneously performed two independent runs, each with four incrementally heated Markov Chain Monte Carlo (MCMC) chains starting from a random tree. Analyses were run for 10 million generations with trees sampled every 100 generations. Stationarity of log likelihood (-lnL) was confirmed by plotting the -lnL scores versus the number of generations. Discarding the first 90,000 trees provided a very conservative 90% burn-in, as -lnL was stationary well before this point. Posterior probabilities were determined by constructing a 50% majority-rule tree of the 10,000 trees sampled after the burn-in. To assess whether Bayesian runs had adequate convergence and mixing for a given data set, we confirmed that the convergence diagnostic PSRF (pscale reduction factor) approached or reached one in all cases, indicating convergence of the two independent runs. In addition, we used Compare and other functions available in AWTY (Are We There Yet; http://ceb.scs.fsu.edu/awty) to confirm convergence .
Maximum likelihood analysis
We performed ML analysis using GARLI Version 0.96 (Beta) https://www.nescent.org/wg_garli/. We used the default GTR model, in light of Modeltest results supporting a relatively complex model. Base frequencies and the portion of invariant sites were estimated from the data, and rates among sites were allowed to vary according to a discrete gamma distribution (4 categories). The runs were terminated when no topology with an lnL increase of 0.01 or greater had been found in 20,000 generations. We used these parameters to find the trees with the highest likelihood score and to perform bootstrap analyses of 100 replications.
We used likelihood tests to compare alternative hypotheses regarding: (i) possible relationships among the four deep Blochmannia lineages, represented by three unrooted trees, and (ii) the 15 various attachments of outgroups to the unrooted Blochmannia trees (three unrooted trees × five possible root attachments = 15). We first estimated the most likely tree (using GARLI as described above) under the appropriate topological constraints (e.g., the unrooted trees A-C shown in Figure 3, and the 15 possible attachments of outgroups to those unrooted trees). We then compared these ML trees using the Shimodaira-Hasegawa (SH) test  as implemented by PAUP version 4.0b10 for Unix . These SH tests involved optimizing each data set across the topologies being compared, using the GTR model described above, then evaluating whether the -lnL of a given dataset differed significantly between the "best" versus alternative topologies. The SH test was performed with all taxa in the Blochmannia-focused analysis.
Ancestral state reconstruction
We used ancestral state reconstruction to infer the history of endosymbiosis in this group, using the Bayesian majority consensus tree with branch lengths and employing the maximum likelihood methods provided by Mesquite v.2.6 . Taxa were coded as either endosymbiotic or non-endosymbiotic. We employed two different models of character evolution, the Mk1 model which uses a single parameter for the rate of change and the AsymmMk model which uses two parameters and thus allows for a bias in gains versus losses.
JJW explores the evolution of insect endosymbionts, studying the effects of lifestyle transitions (e.g., from free-living to host-dependent) on bacterial genomes and the specific roles of symbionts within their insect hosts. SNK is currently a graduate student at the University of California, Berkeley, where he studies the role of symbiotic associations in the generation and maintenance of tropical plant diversity. SGB is a Curator of Hymenoptera and Research Entomologist at the National Museum of Natural History whose research focuses on systematics, molecular evolution, coevolution, and social evolution, especially in ants and bees. PSW is an ant systematist whose research is concerned with delimiting species, inferring phylogenetic relationships, and seeking a better understanding of ant evolution.
Douglas AE: Mycetocyte symbiosis in insects. Biol Rev Camb Philos Soc. 1989, 64 (4): 409-434. 10.1111/j.1469-185X.1989.tb00682.x.
Baumann P, Moran N, Baumann L: Bacteriocyte-associated endosymbionts of insects. The Prokaryotes, A Handbook on the Biology of Bacteria; Ecophysiology, Isolation, Identification, Applications. Edited by: Dworkin M. 2000, New York: Springer-Verlag, 403-438.
Moran NA, Baumann P: Bacterial endosymbionts in animals. Curr Opin Microbiol. 2000, 3 (3): 270-275. 10.1016/S1369-5274(00)00088-6.
Feldhaar H, Gross R: Insects as hosts for mutualistic bacteria. Int J Med Microbiol. 2009, 299 (1): 1-8. 10.1016/j.ijmm.2008.05.010.
Munson MA, Baumann P, Clark MA, Baumann L, Moran NA, Voegtlin DJ, Campbell BC: Evidence for the establishment of aphid-eubacterium endosymbiosis in an ancestor of four aphid families. J Bacteriol. 1991, 173 (20): 6321-6324.
Moran NA, Munson MA, Baumann P, Ishikawa H: A molecular clock in endosymbiotic bacteria is calibrated using the insect hosts. Proc Roy Soc Lond B Bio. 1993, 253 (1337): 167-171. 10.1098/rspb.1993.0098.
Spaulding AW, von Dohlen CD: Psyllid endosymbionts exhibit patterns of co-speciation with hosts and destabilizing substitutions in ribosomal RNA. Insect Mol Biol. 2001, 10 (1): 57-67. 10.1046/j.1365-2583.2001.00231.x.
Baumann L, Baumann P: Cospeciation between the primary endosymbionts of mealybugs and their hosts. Curr Microbiol. 2005, 50 (2): 84-87. 10.1007/s00284-004-4437-x.
Thao ML, Moran NA, Abbot P, Brennan EB, Burckhardt DH, Baumann P: Cospeciation of psyllids and their primary prokaryotic endosymbionts. Appl Environ Microbiol. 2000, 66 (7): 2898-2905. 10.1128/AEM.66.7.2898-2905.2000.
Zientz E, Feldhaar H, Stoll S, Gross R: Insights into the microbial world associated with ants. Arch Microbiol. 2005, 184 (4): 199-206. 10.1007/s00203-005-0041-0.
Dasch G, Weiss E, Chang K: Endosymbionts of insects. Bergy's Manual of Systematic Bacteriology. Edited by: Holt J, Krieg N. 1984, Baltimore: Williams & Williams, 1: 811-833.
Dasch GA: Morphological and molecular studies on intracellular bacterial symbiotes of insects. PhD thesis. 1975, Yale University
Buchner P: Endosymbiosis of animals with plant microorganisms. 1965, New York: Interscience Publishers, Inc
Blochman F: Über das vorkommen von bakterienähnlichen gebilden in den geweben und eiern verschiedener insekten. Zentbl Bakteriol. 1882, 11: 234-240.
Sameshima S, Hasegawa E, Kitade O, Minaka N, Matsumoto T: Phylogenetic comparison of endosymbionts with their host ants based on molecular evidence. Zoological Science. 1999, 16 (6): 993-1000. 10.2108/zsj.16.993.
Jungen H: Endosymbionten bei ameisen. Insect Soc. 1968, 15: 227-232. 10.1007/BF02225845.
Bolton B, Alpert G, Ward PS, Nasrecki P: Bolton's Catalogue of Ants of the World. 2006, Cambridge, MA: Harvard University Press
Sauer C, Dudaczek D, Hölldobler B, Gross R: Tissue localization of the endosymbiotic bacterium " Candidatus Blochmannia floridanus" in adults and larvae of the carpenter ant Camponotus floridanus. Appl Environ Microbiol. 2002, 68 (9): 4187-4193. 10.1128/AEM.68.9.4187-4193.2002.
Schroder D, Deppisch H, Obermayer M, Krohne G, Stackebrandt E, Hölldobler B, Goebel W, Gross R: Intracellular endosymbiotic bacteria of Camponotus species (carpenter ants): Systematics, evolution and ultrastructural characterization. Mol Microbiol. 1996, 21 (3): 479-489. 10.1111/j.1365-2958.1996.tb02557.x.
Wolschin F, Hölldobler B, Gross R, Zientz E: Replication of the endosymbiotic bacterium Blochmannia floridanus is correlated with the developmental and reproductive stages of its ant host. Appl Environ Microbiol. 2004, 70 (7): 4096-4102. 10.1128/AEM.70.7.4096-4102.2004.
Gil R, Silva FJ, Zientz E, Delmotte F, Gonzalez-Candelas F, Latorre A, Rausell C, Kamerbeek J, Gadau J, Hölldobler B, van Ham RC, Gross R, Moya A: The genome sequence of Blochmannia floridanus: Comparative analysis of reduced genomes. Proc Natl Acad Sci USA. 2003, 100 (16): 9388-9393. 10.1073/pnas.1533499100.
Degnan PH, Lazarus AB, Wernegreen JJ: Genome sequence of Blochmannia pennsylvanicus indicates parallel evolutionary trends among bacterial mutualists of insects. Genome Res. 2005, 15 (8): 1023-1033. 10.1101/gr.3771305.
Feldhaar H, Straka J, Krischke M, Berthold K, Stoll S, Mueller MJ, Gross R: Nutritional upgrading for omnivorous carpenter ants by the endosymbiont Blochmannia. BMC Biol. 2007, 5: 48-10.1186/1741-7007-5-48.
Zientz E, Beyaert I, Gross R, Feldhaar H: Relevance of the endosymbiosis of Blochmannia floridanus and carpenter ants at different stages of the life cycle of the host. Appl Environ Microbiol. 2006, 72 (9): 6027-6033. 10.1128/AEM.00933-06.
Stoll S, Feldhaar H, Gross R: Transcriptional profiling of the endosymbiont Blochmannia floridanus during different developmental stages of its holometabolous ant host. Environ Microbiol. 2009, 11 (4): 877-888. 10.1111/j.1462-2920.2008.01808.x.
Davidson DW, Cook SC, Snelling RR, Chua TH: Explaining the abundance of ants in lowland tropical rainforest canopies. Science. 2003, 300 (5621): 969-972. 10.1126/science.1082074.
Cook S, Davidson D: Nutritional and functional biology of exudate-feeding ants. Entomologia Experimentalis et Applicata. 2006, 118: 1-10. 10.1111/j.1570-7458.2006.00374.x.
Blüthgen N, Gebauer G, Fiedler K: Disentangling a rainforest food web using stable isotopes: Dietary diversity in a species-rich ant community. Oecologia. 2003, 137: 426-435. 10.1007/s00442-003-1347-8.
Styrsky JD, Eubanks MD: Ecological consequences of interactions between ants and honeydew-producing insects. Proc Biol Sci. 2007, 274 (1607): 151-164. 10.1098/rspb.2006.3701.
Sauer C, Stackebrandt E, Gadau J, Hölldobler B, Gross R: Systematic relationships and cospeciation of bacterial endosymbionts and their carpenter ant host species: Proposal of the new taxon Candidatus Blochmannia gen. nov. Int J Syst Evol Microbiol. 2000, 50 (Pt 5): 1877-1886.
Degnan P, Lazarus A, Brock C, Wernegreen J: Host-symbiont stability and fast evolutionary rates in an ant-bacterium association: Cospeciation of Camponotus species and their endosymbionts, Candidatus Blochmannia. Syst Biol. 2004, 53: 95-110. 10.1080/10635150490264842.
Brady SG, Schultz TR, Fisher BL, Ward PS: Evaluating alternative hypotheses for the early evolution and diversification of ants. Proc Natl Acad Sci USA. 2006, 103 (48): 18172-18177. 10.1073/pnas.0605858103.
Bolton B: Synopsis and classification of Formicidae. Memoirs of the American Entomological Institute. 2003, 71: 1-370.
Stoll S, Gadau J, Gross R, Feldhaar H: Bacterial microbiota associated with ants of the genus Tetraponera. Biological Journal of the Linnean Society. 2007, 90 (3): 399-412. 10.1111/j.1095-8312.2006.00730.x.
Cole JR, Chai B, Farris RJ, Wang Q, Kulam-Syed-Mohideen AS, McGarrell DM, Bandela AM, Cardenas E, Garrity GM, Tiedje JM: The ribosomal database project (RDP-II): Introducing myRDP space and quality controlled public data. Nucleic Acids Res. 2007, D169-172. 10.1093/nar/gkl889. 35 Database
Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig W, Peplies J, Glockner FO: SILVA: A comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 2007, 35 (21): 7188-7196. 10.1093/nar/gkm864.
Cummings MP, Handley SA, Myers DS, Reed DL, Rokas A, Winka K: Comparing bootstrap and posterior probability values in the four-taxon case. Syst Biol. 2003, 52 (4): 477-487. 10.1080/10635150390218213.
de Souza DJ, Bezier A, Depoix D, Drezen JM, Lenoir A: Blochmannia endosymbionts improve colony growth and immune defence in the ant Camponotus fellah. BMC Microbiol. 2009, 9 (1): 29-10.1186/1471-2180-9-29.
Lenoir A, Dejean A: Semi-claustral colony foundation in the formicine ants of the genus Polyrhachis (Hymenoptera: Formicidae). Insect Soc. 1994, 41 (3): 225-234. 10.1007/BF01242293.
Hölldobler B, Engel-Siegel H: On the metapleural gland of ants. Psyche. 1984, 91: 201-224. 10.1155/1984/70141.
Shattuck SO: Review of the Camponotus aureopilus species group (Hymenoptera, Formicidae) including a second Camponotus with a metapleural gland. Zootaxa. 1984, 19: 75-85.
Dey D, Coumar A: Report of a third Camponotus species with metapleural gland from the world and first from India (Hymenoptera: Formicidae: Formicinae: Camponotini). Acta Entomologica Sinica. 2008, 51: 234-236.
Thao ML, Gullan PJ, Baumann P: Secondary (gamma-Proteobacteria) endosymbionts infect the primary (beta-Proteobacteria) endosymbionts of mealybugs multiple times and coevolve with their hosts. Appl Environ Microbiol. 2002, 68 (7): 3190-3197. 10.1128/AEM.68.7.3190-3197.2002.
Hölldobler B, Wilson EO: The Ants. 1990, Cambridge, Mass: Belknap Press of Harvard University Press
LaPolla J: Ancient trophophoresy: a fossil Acropyga (Hymenoptera: Formicidae) from Dominican amber. Trans Amer Entomol Soc. 2005, 131: 21-28.
Wheeler W: The ants of Baltic amber. Schriften der Physikalisch-ökonomischen Gesellschaft zu Königsberg. 1914, 55: 1-142.
Sakata H: How an ant decides to prey on or to attend aphids. Res Popul Ecol. 1994, 36: 45-51. 10.1007/BF02515084.
Rosengren R, Sundström L: The interaction between red wood ants, Cinara aphids, and pines. A ghost of mutualism past?. Ant-Plant Interactions. Edited by: Huxley C, Cutler D. 1991, Oxford: Oxford University Press, 80-91.
Offenberg J: Balancing between mutualism and exploitation: the symbiotic interaction between Lasius ants and aphids. Behav Ecol Sociobiol. 2001, 49: 304-310. 10.1007/s002650000303.
Way M: Mutualism between ants and honeydew-producing Homoptera. Annu Rev Entomol. 1963, 8: 307-344. 10.1146/annurev.en.08.010163.001515.
Kittayapong P, Jamnongluk W, Thipaksorn A, Milne JR, Sindhusake C: Wolbachia infection complexity among insects in the tropical rice-field community. Mol Ecol. 2003, 12 (4): 1049-1060. 10.1046/j.1365-294X.2003.01793.x.
Chiel E, Zchori-Fein E, Inbar M, Gottlieb Y, Adachi-Hagimori T, Kelly SE, Asplen MK, Hunter MS: Almost there: Transmission routes of bacterial symbionts between trophic levels. PLoS ONE. 2009, 4 (3): e4767-10.1371/journal.pone.0004767.
Johnson RN, Agapow PM, Crozier RH: A tree island approach to inferring phylogeny in the ant subfamily Formicinae, with especial reference to the evolution of weaving. Mol Phylogenet Evol. 2003, 29 (2): 317-330. 10.1016/S1055-7903(03)00114-3.
Brady SG, Gadau J, Ward PS: Systematics of the ant genus Camponotus (Hymenoptera: Formicidae): A preliminary analysis using data from the mitochondrial gene cytochrome oxidase I. Hymenoptera Evolution, Biodiversity and Biological Control. Edited by: Austin AD, Dowton M. 2000, Collingwood, Victoria: CSIRO Publishing, 131-139.
Russell JA, Latorre A, Sabater-Munoz B, Moya A, Moran NA: Side-stepping secondary symbionts: Widespread horizontal transfer across and beyond the Aphidoidea. Mol Ecol. 2003, 12 (4): 1061-1075. 10.1046/j.1365-294X.2003.01780.x.
Chang BS, Campbell DL: Bias in phylogenetic reconstruction of vertebrate rhodopsin sequences. Mol Biol Evol. 2000, 17 (8): 1220-1231.
Herbeck JT, Degnan PH, Wernegreen JJ: Nonhomogeneous model of sequence evolution indicates independent origins of primary endosymbionts within the Enterobacteriales (gamma-Proteobacteria). Mol Biol Evol. 2005, 22 (3): 520-532. 10.1093/molbev/msi036.
Posada D, Crandall KA: MODELTEST: Testing the model of DNA substitution. Bioinformatics. 1998, 14 (9): 817-818. 10.1093/bioinformatics/14.9.817.
Nylander JA, Wilgenbusch JC, Warren DL, Swofford DL: AWTY (are we there yet?): A system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics. 2008, 24 (4): 581-583. 10.1093/bioinformatics/btm388.
Shimodaira H, Hasegawa M: Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol Biol Evol. 1999, 16 (8): 1114-1116.
Swofford DL: PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. 2002, Sunderland, Mass.: Sinauer Associates
Maddison W, Maddison D: Mesquite: a modular system for evolutionary analysis. Version 2.6. 2009, [http://mesquiteproject.org]
We are grateful to Alan Andersen, Don Windsor, and Dinah Davidson for contributing insect specimens used for this study. We also thank Adam Lazarus for assisting with molecular lab work, Ulrich Mueller for access to his unpublished review of the metapleural gland, and three anonymous reviewers for helpful comments. Funding for this research was provided by grants from the NSF (MCB-0604177) and NIH (R01GM062626) to JJW, and from the NSF-supported Ant AToL project (EF-0431330) to PSW and SGB.
JJW performed the DNA sequence alignment, phylogenetic analysis, and database comparisons. SNK, as a research assistant at the MBL, performed the molecular biology lab work and assisted with data analysis. SGB and PSW provided ant specimens, identifications, and phylogenetic and myrmecological expertise. SGB performed ancestral character state reconstruction. All authors contributed to the design of the study, commented on the manuscript, and approved the final version.